Versione Inglese PDG Sabbatini - Volume Completo - 03!10!2007

The euro cash changeover, ination and consumers perceptions.
Lessons from the case of Italy.
Paolo Del Giovane and Roberto Sabbatini
To Lorenza, Luciano and Marco To Pierluigi and Riccardo
VI
VII
Foreward [TO BE INCLUDED]
VIII
IX
Acknowledgments
This book is the result of a collective research effort conducted in the Research Area of the Bank of Italy between 2003 and 2007. We would like to thank all the authors of the chapters collected in this volume and the other colleagues of the Bank who provided stimulating comments at various stages of the project, in particular..... Special thanks go to Francesco Lippi, who, as the co-editor of a previous publication on the same topic, shared with us the challenge of describing and interpreting the complex phenomena treated in this volume. We have also beneted greatly from the comments received by the participants in various seminars and conferences where the material reported in the book was presented, in particular two seminars at the European Central Bank and the conferences organized by the OECD (June 2005), the Western Economic Association International (June 2006), the National Euro Changeover Committee of Malta (May 2007), the Government of Malta in cooperation with the European Central Bank and the European Commission (October 2007). Finally, we would like to thank Daniel Dichter, Carla Lucidi, Roger Meservey and Christine Stone for their excellent linguistic assistance and the Bank of Italys printers for their help in producing the book to their usual high standards. We acknowledge Il Mulino publishing house for the authorization to use part of the material published in Italian in the volume Leuro e linazione edited by Paolo Del Giovane, Francesco Lippi and Roberto Sabbatini (Bologna, 2005). The opinions expressed in the volume are those of the editors and the other authors, and they do not necessarily reect the views of the institutions they belong to.
XI
Contents
Foreword [TO BE INCLUDED] Acknowledgments Contributors List of abbreviations Introduction Paolo Del Giovane and Roberto Sabbatini 1 Perceived and measured ination after the launch of the euro: explaining the gap in Italy Paolo Del Giovane and Roberto Sabbatini 2 Rounding and anomalous changes in Italian consumer prices in 2002 Franco Mostacci and Roberto Sabbatini 3 Pricing behaviour and the introduction of the euro: evidence from a panel of restaurants Eugenio Gaiotti and Francesco Lippi 4 Did prices really soar after the euro cash changeover? Evidence from ATM withdrawals Paolo Angelini and Francesco Lippi 5 Memory for prices and the euro cash changeover: an analysis for cinema prices in Italy Vincenzo Cestari, Paolo Del Giovane and Clelia Rossi-Arnaud 6 What are ination perceptions about? A survey-based analysis on Italian consumers [TO BE INCLUDED] Paolo Del Giovane, Silvia Fabiani and Roberto Sabbatini 7 Impact of the cash changeover on prices and ination perceptions in the euro area: a survey of the empirical evidence Paolo Del Giovane and Roberto Sabbatini List of gures List of tables Authors index [TO BE INCLUDED] Subject index [TO BE INCLUDED]
XII
XIII
Contributors
Paolo Angelini Vincenzo Cestari
Director in staff, Department of Economic Outlook and Monetary Policy Studies, Bank of Italy. Professor of Psychology, Lumsa University, Faculty of Educational Science, and National Research Council (CNR), Institute of Neuroscience. Head of Monetary Analysis Division, Department of Economic Outlook and Monetary Policy Studies, Bank of Italy. Senior Economist, Department of Economic Outlook and Monetary Policy Studies, Bank of Italy. Deputy Head of the Department of Economic Outlook and Monetary Policy Studies, Bank of Italy. Professor of Economics, University of Sassari, and CEPR. Senior Researcher, Price Statistics Department, Italian National Statistical Institute (ISTAT). Professor of Psychology, University of Rome La Sapienza, Psychology Department. Deputy Head of Economic Outlook Division, Department of Economic Outlook and Monetary Policy Studies, Bank of Italy.
Paolo Del Giovane
Silvia Fabiani Eugenio Gaiotti
Francesco Lippi Franco Mostacci Clelia Rossi-Arnaud Roberto Sabbatini
XIV
XV
List of abbreviations
ATM CNR CPI EC ECB EU GR HICP INSEE ISAE ISTAT NSI OLS SD SHIW UIC ULC
Automated Teller Machine Consiglio Nazionale delle Ricerche (National Research Council) Consumer Price Index European Commission European Central Bank European Union Il Gambero Rosso Harmonized Index of Consumer Prices Institut National de la Statistique et des tudes conomiques Institute for Economic Research and Analysis Italian National Institute of Statistics National Statistical Institute Ordinary Least Squares Standard Deviation Survey of Households Income and Wealth Ufcio Italiano Cambi Unit Labour Costs
XVI
1. Perceived and measured ination after the launch of the euro: explaining the gap in Italy
Paolo Del Giovane and Roberto Sabbatini Bank of Italy, Department of Economic Outlook and Monetary Policy Studies
1.1 Introduction1
Following the euro cash changeover, residents in the area perceived a much larger rise in prices than was recorded in ofcial statistics. The gap between perceived and measured ination reached an unprecedented size in the course of 2002, then narrowed or closed in most countries. In Italy, it was extremely large and persistent, only beginning to diminish after two years. The perceived surge in prices was popularly associated with the introduction of the euro, even though national statistical institutes and central banks estimated the impact to be generally modest. In the ensuing debate between experts, consumer organizations and members of the public conducted under the spotlight of the media, the credibility of ofcial data on prices was called sharply into question. On the basis of their everyday experience, households became convinced that ination rates were much higher than ofcial gures. In Italy there was a widespread notion that retailers had converted the prices of many goods and services at a rate of 1,000 lire to the euro (rather than at the ofcial rate of 1,936.27 lire), which would have implied an almost 100 per cent price hike. Private research centres published alternative estimates of ination that were much higher than those of ISTAT, the National Institute of Statistics, and tended to bear out, if not actually reinforce, the publics impressions. The gap between perceived and measured ination is an important phenomenon: it impairs rms and consumers ability to judge individual
1
The authors thank Luigi Cannari, Vincenzo Cestari, Lorenzo Forni, Luigi Guiso, Francesco Lippi, Marco Magnani, Stefano Siviero, Daniele Terlizzese, and participants in ECB, OECD and WEA workshops, for useful comments. They acknowledge Giornale degli Economisti e Annali di Economia for the authorization to reprint this article, with small changes, from Volume 65, No. 2 (November 2006; copyright Universit Commerciale Luigi Bocconi).
P. Del Giovane and R. Sabbatini
prices, eroding the allocative efciency of the price system; by altering expectations of future price trends, it can distort decisions on prices and wages; the credibility of monetary policy may be undermined if the quality of the underlying price indices is called into question; and, in the case of the euro area, the publics acceptance of the common currency and its institutional framework may be shaken. In this chapter we rst dene the subject of inquiry and then examine whether the observations based on personal experience are truly irreconcilable with the ofcial statistics. We systematically review the various arguments offered for the sudden worsening of ination perceptions, starting from the hypothesis that a combination of factors was at work. In the case of Italy, we explore the possibility that a major cause was the interaction between how individual perceptions are formed (e.g. attributing more weight to price increases than to price cuts) and the observed behaviour of prices and other economic variables during and after the cash changeover (e.g. larger increases in the prices of frequently purchased products). We also investigate the collective dimension of the phenomenon by considering the role of the media. The chapter is organized as follows. Section 1.2 describes the indicator commonly used to measure ination perceptions. Section 1.3 examines whether there are grounds for doubting the accuracy of the ISTAT index and highlights a few factors that might be of some importance in explaining the discrepancy with respect to perceptions. On the hypothesis that the gap had numerous causes, section 1.4 analyzes and assesses other explanations, including the interaction between the psychology of individuals ination perceptions and price developments following the changeover. Section 1.5 provides an overview of the correlations between ination perceptions and the factors we identied. Section 1.6 summarizes and concludes.
1.2 Ination perceptions

Public debate on the perception of a much sharper increase in consumer prices than that measured by ofcial statistics has generally relied on anecdotal evidence, personal impressions very often limited to specic products or classes of goods and services. Yet a survey is available that offers information on the perceptions of a representative sample of the entire population, based on questions regarding ination in general: it is part of the harmonized monthly consumer surveys conducted in all the countries of the European Union, and in Italy by the Institute for Economic Research and Analysis (ISAE). The European Commission aggregates the results of the national surveys and publishes indicators for the whole
Perceived and measured ination after the launch of the euro
European Union and for the euro area in the monthly press release Business and Consumer Survey Results. On the subject of prices, respondents are asked to state their views regarding price developments in the last and the next twelve months. Regarding the past trend, consumers are asked: How do you think that consumer prices have developed over the last 12 months?. Answers must be chosen from the following list: risen a lot (N1), risen moderately (N2), risen slightly (N3), stayed about the same (N4), fallen (N5), and dont know. A composite measure of respondents opinions is obtained as the balance between the weighted frequencies of the different responses. The extreme responses (risen a lot, fallen) are assigned twice the weight of the intermediate responses, while the central response (stayed about the same) is not counted. The balance statistic (S) is therefore obtained as S = N1+ 0.5 N2 0.5 N4 N5, where Ni denotes the relative frequency of the i-th response. Fig. 1.1 traces the ination rate calculated by the statistical institutes and the balance of perceptions calculated as just described for Italy and for the euro area. The indicator of perceived ination is purely qualitative the respondents are not asked to make a numerical estimate and can be translated into a quantitative measure only with further assumptions (see below); in interpreting the gure, we can therefore compare the behaviour of the two curves but not their level. The behaviour of ination measured by the statistical institutes and that of consumers perceptions were closely correlated between 1991 and 2001 both in Italy and in the euro area, although variations in perceptions were more marked, especially in Italy. Between the end of 2001 and the beginning of 2003, while ofcial ination remained broadly stable, perceived ination increased sharply:2 the percentage of Italian consumers who thought prices had risen a lot jumped from 10 to 49 per cent; also counting those who answered risen moderately, the proportion almost doubled, from 48 to 89 per cent. Ination perceptions increased in 2002 in all the euro-area countries. In most countries, however, the trend reversed in the course of 2003; the turnaround was clear-cut in particular in Germany, where by the end of the year the balance was lower than before the changeover,3 less marked in other countries, where perceptions settled at a fairly high level. In Italy and Greece, the deterioration was exceptionally large and persistent; in Italy
2
See European Central Bank (2002a, 2002b, 2003b, 2007) and Aucremanne and Collin (2005) for a description of the phenomenon for the euro area as a whole, and chapter 7 of this book for a survey of the studies available for individual countries. For an analysis of the evolution of prices and perceptions in Germany, see Deutsche Bundesbank (2004).
(a) Italy
7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Perceptions (qualitative indicator) (1) 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 04 Inflation as measured (2) 70 60 50 40 30 20 10 0
(b) Euro area

8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 70
Perceptions (qualitative indicator) (1)
60 50 40 30 20 10
Inflation as measured
0 -10
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 04
Fig. 1.1. Ination measured by the statistical institutes and ination perceptions (twelve-month percentage change in the harmonized consumer price index; percentage balances of response). Source: European Commission and Eurostat. Notes: (1) Qualitative indicator obtained as the percentage balance of responses to the monthly consumer survey (right-hand scale). (2) Twelve-month rate of increase in the harmonized index of consumer prices (left-hand scale).
peak values were recorded throughout 2003, only beginning to decline in the rst half of 2004. The populations perception of pronounced price ination went hand in hand with the belief that the introduction of the euro was the direct cause. Fig. 1.2, which is based on the European Commission survey The uro, Two Years Later, carried out between October and November 2003, shows that the change in euro-area residents perceptions was positively related to the conviction that prices were converted into euros to the disadvantage of consumers (panel a). This conviction was most widespread in Italy, being held by 96 per cent of respondents, against 89 per cent for the euro area as a whole. Fig. 1.2 also shows that higher ination perceptions were usually accompanied by greater dissatisfaction with the common currency (panel b). The Italians and Greeks, who had originally been very much in favour of European unication, were the most dissatised after the Germans. By the end of 2003, 44 per cent of Italians declared they were quite/very unhappy that the euro had become their currency, 13 percentage points more than in the survey conducted a year earlier. A comparable increase was recorded only in Greece (14 percentage points), against an area average of 7 points. It is worth noting that the statistical signicance of the relation is diminished by Germanys outlier position; in fact, unhappiness about the euro was greater there than elsewhere in the euro area, even though ination perceptions diminished during 2003. In all likelihood, German dissatisfaction was due more to the performance of other macroeconomic variables, notably economic growth and employment, than to the perceived trend of ination. It cannot be ruled out that the same factors affected consumer sentiment towards the euro in other countries as well. The survey responses, which blame the euro for most of the perceived rise in ination, contrast sharply with the fairly low inationary impact of the cash changeover estimated by the central banks and statistical institutes. 4 Although, as we pointed out earlier, there are no systematic surveys asking respondents to put a number on the ination rate they perceive,5 it is possible to transform the European Commissions qualitative indicator into a quantitative measure for comparison with the ofcial statistics. A number of different methods are available. One is to regress measured
4
See European Central Bank (2003a) and Mostacci and Sabbatini in chapter 2 of this book, respectively, for an analysis of the impact in the euro area and in Italy. Two new questions aimed at obtaining quantitative estimates of ination perceptions and expectations were introduced into the harmonized consumer surveys, on an experimental basis, in 2003 (Lindn 2006). The resulting dataset is not public. Quantitative ination perceptions are investigated in depth in chapter 6, for different time horizons and different consumption bundles.
(a) Relation between change in ination perceptions and conviction that price conversion into euros was to the detriment of consumers (1)
4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 70 75 80 85 90 95
FI BE PT IE FR ES DE NL AT EL IT
4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 100
(b) Relation between change in ination perceptions and unhappiness about the euro (2)
4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 10 20 30 40 50 60 70
IE BE PT ES FR NL DE AT EL IT
4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0

FI
0.5 0.0
Fig. 1.2. Ination perceptions and sentiment about the euro. Source: European Commission (2003). Notes: (1) The y-axis plots the change in the qualitative indicator of ination perceptions (ratio of the average balance in 2002-03 to the gure for December 2001). The x-axis plots the percentage of those who responded to the detriment of consumers to the question Did you personally notice that in your country, when converted into euros, prices have been: rather in favour of consumers; rather to the detriment of consumers; or one way or another the rises and falls balanced out?. (2) As in panel (a), the y-axis shows the change in ination perceptions. The x-axis plots the percentage of those who answered quite/very unhappy to the question Are you personally: very happy, quite happy, quite unhappy, or very unhappy that the euro has become your currency?.
ination (the twelve-month increase in the general consumer price index) on the qualitative balance of perceptions. This method, in which the tted value of the regression is interpreted as a quantitative measure of perceived ination, is based on the assumption that in the long run perceived ination is equal to ofcially measured ination, i.e. that consumers do not make systematic errors over a sufciently long time span.6 By estimating the equation for Italy in the period 1992-2001, the perception was quantied at around 6 per cent at the end of 2003, with a 95 per cent condence interval of 1.5 percentage points. The results of the quantication proved to be very sensitive to the choice of the estimation period; for example, referring to a shorter time horizon (since 1997), estimated perceived ination amounted to around 3.5 per cent. Given the uncertainty surrounding these estimates, in the rest of this paper we prefer to use the qualitative balance of perceptions without further elaboration.
1.3 The ofcial index

In the debate on the gap between perceived and measured ination, the ofcial consumer price indices have often been criticized for using inappropriate methods of calculation and underestimating the ination actually experienced by households. In Italy these assertions have been upheld, among others, by a number of consumer associations and appear to be borne out by alternative estimates of ination that are much higher than those of ISTAT. In many cases, though, the methods underlying these estimates are not well-documented, particularly the size and nature of the sample, the criteria for recording prices, and the weights assigned to the single products for the calculation of ination (see below). The rest of this section examines whether there is cause to doubt the accuracy of the ofcial index, and highlights some elements that might help to explain the difference with respect to perceived ination. In the course of the 1990s the methods used in the EU countries to calculate the consumer price indices were harmonized in several essential respects. The denition of the index, the items in the basket, the classication, the formulae for aggregating elementary data, the frequency of re-basing, and the treatment of some specic items were all standardized (Eurostat 2001). As a
ISAE (2002) adopts an alternative approach (the latent variables method) to quantify perceptions; the results are comparable to ISTATs measure of ination.
10
result of this process, the methods employed today by the national statistical institutes, including ISTAT, to calculate the harmonized index of consumer prices (HICP) can be said to correspond to the best practices available internationally.7 However, there is undoubtedly room for improvement. In Europe, efforts are being made to improve quality adjustment and to assess how to treat owner-occupied housing, currently not included in the basket (see European Central Bank 2005). As regards Italy in particular, the treatment of specic items may need to be revised. An example is rent, for which the corresponding deator in national accounts data (see Banca dItalia 2005) and the data collected in the Banca dItalias Survey of Household Income and Wealth (Banca dItalia 2004, 2006) suggest higher increases in recent years than measured by the consumer price index. 8 As to the operating procedures, in Italy, as in other countries, the index is calculated from a sample survey of the prices of the goods and services that account for the largest share of household consumption; in Italy there are 1,043 such items. The survey is carried out centrally by ISTAT for just over a fth of the goods and services in the basket, while for all the rest it is conducted locally by the municipal statistical ofces in 87 provincial capitals and 39,000-odd survey points. The duties of these ofces therefore include the most delicate part of the calculation of a price index, namely the acquisition of the basic data. In carrying out these tasks they have considerable leeway, albeit within the general guidelines set by ISTAT. We have no direct evidence of the quality of the local price surveys, and in particular of how the list of goods and services to be surveyed is drawn up or the sample of shops selected. We can only rely on the indirect indications obtained by comparing price changes recorded in the different cities. There is limited dispersion of the distribution of the month-on-month and the twelvemonth ination rates, measured by the coefcient of variation, and this dispersion has diminished in recent years for both indicators (Del Giovane and Sabbatini 2005). The absence of extreme and systematic differences between cities suggests that no deep-rooted anomalies exist. The public and the press have repeatedly asserted that the ISTAT index does not capture exceptionally large movements in the prices of some products. This argument is usually raised after particularly sharp price increases for very specic
7
Del Giovane and Sabbatini (2005) take a closer look at the methods used to compute the price indices, considering specic aspects that have been the subject of some debate the reference population, and the effects of substituting and introducing new products in the basket and conclude that there are no features suggesting that they systematically generate an underestimation of ination; if anything, the substitution effects and the impact of introducing new products could lead to an overestimation. Concerning this particular issue, see also Sect. 1.4.1(c).
11
goods or services, such as a particular vegetable sold in a given market, and it fails to take into consideration the average nature of price indices in general, as well as specic methodological aspects of the treatment of some items. The breakdown into 207 product items that ISTAT publishes every month on its Internet site only occasionally refers to the prices of single goods or services. Most are aggregations of similar goods or services, obtained by weighting the single products in proportion to their shares in consumption. The prices of these products can move in different directions in the short term: for example, the price of zucchini, which are included in the item fresh vegetables, may diverge from that of other components of the same item. As the weight of single products tends to be small, their impact on changes in the price of the aggregate item and the general index is correspondingly limited. Moreover, even for single products the published index is an average of prices recorded throughout Italy. Averaging both products and prices results in the smoothing of larger movements. For the items fresh fruit and fresh vegetables, consisting of a variety of products, the comparison between individual experience and ISTATs observations can also be affected by the method of calculation employed to compute the price indices; this excludes varieties whose high prices mean they are likely to be discarded in favour of similar products a plausible hypothesis, given the range of alternatives.9 In practice, this procedure lessens the impact of supply shocks on the prices of specic varieties, but it does not tally with the everyday experience of individuals, who observe the extreme prices even if they decide not to pay them. Another much-discussed item is motor-vehicle insurance; here again, the comparison between individual impressions and the ofcial statistics is affected by the methodological approach. In Italy, as in the rest of Europe, the cost of insurance is included in the calculation of ination net of claims paid and therefore has a small weight in the average basket (just over 1 per cent).10 The rationale is that the household sector as a whole pays out premiums to insurance companies and receives payments for claims; but from the point of view of the individual consumer who has not had any accidents and not received any payments, the incidence of this expenditure item is well above the average. Although some individual price changes are smoothed as we describe above, a careful examination of the ISTAT data shows that large movements
10
See Del Giovane and Sabbatini (2005) for details. The answers to a specic question in the survey carried out by Del Giovane, Fabiani and Sabbatini (Chap. 6) suggest that the consumers behaviour is in fact consistent with this hypothesis. The weight refers to the 2003 basket. It was lower in previous surveys, as shown in Table 1.1.
12
are present, especially when looking at the monthly peaks instead of the annual average. Table 1.1 shows the ten items of the 207-item breakdown that recorded the largest twelve-month price increases in 2002 and 2003, regardless of the month in which they occurred, and the ten with the largest decreases. In 2002 there were extremely large increases in the prices of some food products (30 per cent for potatoes and 20 per cent for fresh vegetables) and services (between 14 and 22 per cent for air transport, insurance, bathing establishments and maritime transport). Some of the same items also registered substantial price reductions in other months, conrming their pronounced variability.11 In section 1.4.1 we discuss the possibility that increases and decreases may be perceived asymmetrically and that the weight of individual items in the formation of perceptions may not be the same as their actual weight in the basket. All told, the information available does not suggest that the indices calculated by ISTAT systematically miscalculate and underestimate ination,12 although the methodological treatment of specic items is open to improvement. Moreover, we cannot rule out the possibility of aws in the data collection procedures, especially regarding the choice of retail outlets and the sample of goods and services whose prices are recorded, for which the municipal statistical ofces are responsible.13 As to the alternative estimates with respect to ISTATs, the impression is that they are partial, based on a much more limited basket of goods and services and a much smaller sample of sales points than those monitored by ISTAT. The estimates released by Eurispes, a private research institute, are a case in point owing to the attention they received in the media. In the summer of 2002, the director of Eurispes announced that ination in Italy was running at a rate of 8-9 per cent, more than three times the gure calculated by ISTAT. The following year, Eurispes reported that food prices had risen by 29 per cent between December 2001 and December 2002, compared with ISTATs
The presence of exceptionally large price movements becomes even more evident if we look at more disaggregated data. Del Giovane and Sabbatini (2005) show that for a 550-item breakdown of ISTAT data price changes range between +58 and -18 per cent. Moreover, these changes were not unusual with respect to previous years. In every year between 1998 and 2001 the largest variations (both upwards and downwards) were comparable to those in the two years under review. 12 This conclusion is supported by the results of Angelini and Lippi (see Chap. 4), who look at trends in ATM cash withdrawals. They nd no substantial breaks between the period preceding and that following the cash changeover; this is the opposite of what should have been observed (with at consumption) if, as suggested by perceptions, actual ination had been much higher than that measured by the ofcial statistics. 13 See Buzzigoli, Di Iorio and Marliani (2002) for a detailed examination of the process used to produce ISTATs price indices, particularly the gathering of basic data.
11
13
Table 1.1. Extreme price changes: 207-item breakdown

2002 Weight % 12-month change % Annual average % month change 2003 Weight % 12-month change % Annual average % month change
Product item
Product item
Upward Potatoes Air transport Fresh vegetables Insurance services Bathing establishments Newspapers Maritime transport Fresh shellsh Fresh fruit Amusement parks Total Downward Air transport Potatoes Liquid fuels Electricity Other meats Gas Petrol Poultry Other fuels IT equipment Total 0.67 0.28 0.68 1.22 0.18 1.87 1.97 0.65 0.28 -3.9 Dec. -4.4 Nov. -4.5 Jan. -5.2 Mar. -6.7 Apr. -7.3 Apr. -7.5 Jan. -7.5 Mar. -9.6 Jan. 6.1 10.4 -0.2 -1.5 -2.4 4.8 -2.7 -2.4 -2.7 -11.5 0.28 0.67 1.22 0.31 0.17 0.28 0.14 0.28 0.96 0.09 4.40 30.5 Feb. 22.1 Aug. 19.6 Apr. 18.7 Mar. 15.9 July 14.9 July 13.8 June 12.6 Aug. 11.0 May 9.2 Sept. 10.4 6.1 13.6 11.6 6.6 13.0 8.9 11.0 8.9 6.0
Upward Other lodging services Postal bank services Potatoes Inland water transport Italian cigarettes Lubricants Other fuels Banking services Liquid fuels Fresh shellsh Total Downward Other fuels Audiovisual equipment Amusement parks Petrol Skilift Medical Products Maritime transport Telephones and accessories Potatoes IT equipment Total 0.28 0.44 0.09 1.97 0.09 2.92 0.13 -2.0 Oct. -2.3 Nov. -2.7 Sept. -2.8 Jan. -3.5 Dec. -6.5 Mar. -8.1 Sept. 3.4 -1.3 0.3 1.5 2.5 -3.8 -0.5 -3.9 2.6 -14.5 0.31 0.15 0.25 0.00 0.49 0.11 0.28 0.64 0.73 0.34 3.29 32.5 Aug. 26.7 Jan. 23.6 Nov. 20.8 June 15.6 Nov. 13.6 Mar. 12.6 Mar. 11.9 Jan. 11.7 Mar. 11.1 Jan. 15.6 26.7 2.6 12.8 12.1 10.6 3.4 8.9 2.9 5.9
0.90 -13.6 Dec. 0.25 -13.7 Apr. 0.21 -17.0 May 7.28
0.24 -14.7 Jan. 8.05
Source: Based on ISTAT data (2003 basket).
14
estimate of 3.8 per cent. When examining the estimates made by Eurispes we must bear in mind that they were largely based on an ex post reconstruction of prices in December 2001, the reference month for calculating ination, which the institute had not recorded at the time. As Eurispes itself explained (2003):
The prices in 2001 had to be derived from: printed price lists and those still stored on computers, shopping receipts of the previous year, newspaper advertisements and advertising circulars listing products and prices, discussion groups (composed of customers and sellers) that met in front of stalls in open-air markets or within shops (p. 2). Very often, for both shops and market stalls, the only way to obtain information on past prices was to involve the customers present at the time and the seller in a discussion (p 8).14
Apart from the methodological issues raised by the construction of the sample and the indices and the denition of the set of products, Eurispess use of respondents (households and retailers) retrospective evaluations to derive price changes makes this estimate better suited to measuring perceived than actual ination, since actual data recorded in 2002 are compared with data based partly on individuals recollections.
1.4 Why the gap between ofcial and perceived ination?

We have seen how some aspects of the methods used to compute the ofcial indices may have helped to create a gap with respect to perceptions of ination, particularly for specic items. However, these aspects cannot explain either the widespread nature or the magnitude of the divergence. In this section we look at other possible explanations to see whether they can reconcile individual ination perceptions with the ofcial statistics. Section 1.4.1 examines how some factors attributable to asymmetries in the way individuals ination perceptions are formed and in the actual behaviour of prices might have become accentuated during or after the euro cash changeover.15 It can be hypothesized that individuals ination perceptions are (a) more sensitive to increases in the prices of frequently purchased products; (b) more sensitive to price increases than to price decreases or over-inuenced by extreme changes; (c) formed or expressed differently depending on the
14 15
Translations of quotations from Italian are ours; emphasis added. Examples of asymmetry in subjective evaluation can be found in Kahneman and Tversky (1979). Although their analysis relates to a different issue (decision-making under uncertainty) it provides some interesting results for an understanding of the phenomenon considered in this paper. See also Tversky and Kahneman (1974) for an analysis of the heuristics employed in subjective assessments.
15
stratum of the population. These hypotheses regarding the psychology of ination perceptions imply that perceptions are sharper when the following price developments occur: (a) larger price rises for frequently purchased products than for other goods; (b) a change in the distribution of price changes, with a larger proportion of upward and downward movements and a greater number of extreme changes; (c) greater differences in ination rates for the consumption baskets of individuals or specic categories of persons. Such interactions may have been amplied if there was a link between individuals expectations concerning the impact of the changeover to the euro and their ex-post perceptions.16 Research in experimental psychology (Traut-Mattausch et al. 2004) has found that the more pessimistic their expectations, the more individuals tended to overestimate the changeovers inationary effects. This relation is attributed to a mechanism called selective output correction, the propensity to check data conrming ones expectations less accurately than data disproving them. In section 1.4.2 we investigate the possibility that consumers blamed ination for losses of purchasing power not directly attributable to it, such as variations in income or price rises for goods not included in the ofcial basket. Section 1.4.3 considers the degree to which the medias exceptional coverage of ination following the cash changeover may have amplied individual perceptions. Lastly, section 1.4.4 examines whether perceptions may have reected consumers propensity to gauge the new prices in euros by mentally converting them into lire at approximately 2,000 lire per euro (as against the ofcial locked-in rate of 1,936.27 lire) or on the basis of an imprecise recollection of the old prices. 1.4.1 Asymmetries in individual perceptions and actual price movements
(a) Frequency of purchases
After the cash changeover consumers had to memorize a large number of prices re-denominated in euros. A learning process of this scale can take a long time and its duration may vary from one product to another.
16
The publics concerns on the eve of the changeover and in the rst days of 2002 emerge from the accounts of daily newspapers (to cite one article among many, see The impact of rounding, in Il Sole 24 Ore of 28 December 2001) and from the survey conducted by the European Commission in November 2001, according to which 78 per cent of the Italian respondents were afraid of abuses and cheating on prices, against 70 per cent in the euro area.
16
It is plausible that the process will be swifter and easier for the prices of goods and services that are purchased frequently, in some cases even several times a day.17 We can assume, therefore, that perceptions after the changeover reected movements in the prices of frequently purchased goods and services to a greater degree than in the past and disproportionately with respect to their weight in the basket. However, simply paying more attention to these products is not sufcient to induce a higher perception of ination than the ofcial rate. For this to happen, an additional condition must be satised, i.e. the prices of frequently purchased goods and services must rise faster than those of the other products in the basket. We assess whether this condition was met in the data by classing the 207 items published by ISTAT for the consumer price index (CPI) into two groups: those purchased at least once a month (high frequency) and the remainder (low frequency). In the rst group (representing around 40 per cent of the ISTAT basket) we placed food products, tobacco, everyday household products (detergents, soaps, etc.), newspapers, some services (local transport and railways, postal and banking services, restaurants and coffee shops, recreational and cultural services) and fuels.18 There were marked divergences in the twelve-month changes in the two sub-indices as early as 2001 (Fig. 1.3) and this gap remained wide for the next two years. The prices of frequently purchased products rose sharply in the second half of 2002 and again in mid-2003, while in the other set of products ination was broadly stable or decreasing. According to these results part of the divergence between perceived and ofcially measured ination can be attributed to the gap between the ination rate for frequently purchased goods and services, to which consumers plausibly attached more weight in formulating their opinions of average
17
18
Marques and Dehaene (2004) examine the psychological mechanisms through which individuals form their estimates of prices and the possible relation between frequency of purchase and rapidity of the learning process. An experiment presented in the work, conducted on a sample of Austrian and Portuguese individuals, confirms the hypothesis that learning is faster for frequently consumed goods. On the importance of this aspect, see also Guiso (2003), Dziuda and Mastrobuoni (2005) and Ehrmann (2006), as well as Brachinger (2006) and the related discussion by Hoffmann, Leifer and Lorenz (2006). An alternative approach to a priori selection of the items is to identify them with a correlation analysis. This method has been used for Italy by ISAE (2002) and for France by INSEE (2003) and Crdit Agricole (2004).
17
ination, and the ination rate for less frequently purchased items.19 However, it should be noted that the ination rate for the high-frequency products was less than 4 per cent, well below many consumers personal estimates of general ination. It may be that the effect on ination perceptions of price rises for frequently purchased products was amplied by the consumers tendency to seek conrmation for their pessimistic expectations regarding the inationary effect of the changeover (Traut-Mattausch et al. 2004).
4.0 3.5 3.0 2.5 2.0 1.5 1.0 1997 1998 1999 2000 2001 2002 2003 4.0
high frequency
3.5 3.0
low frequency
2.5 2.0 1.5 1.0
Fig. 1.3. Ination rates for goods and services in Italy according to frequency of purchase (quarterly data; twelve-month percentage changes in the sub-indices) Source: Based on ISTAT data. (b) Distribution of price changes
The ofcial consumer price indices are calculated by weighting the individual items according to their shares in total household consumption. However, the individual consumers perceptions may be more responsive to the upward movement of one price than to the downward movement of another, even if the two changes are simultaneous and of the same order of magnitude; likewise, an exceptionally large price change for a product whose weight in the index basket is modest may have a stronger effect on perceptions
19
The relationship between perceptions and price changes for specic categories of goods and services, including the most frequently purchased, has been analyzed by European Central Bank (2003b) for the euro area as a whole and by various studies for single countries of the area (see De Nederlandsche Bank 2002, Santos et al. 2002, Walschots 2002, Banco de Espaa 2003, Buiten 2003, lvarez Gonzlez et al. 2004, Fluch and Stix 2005, Aalto-Setl 2006, Kurri 2006, in addition to the two French studies mentioned in the preceding footnote).
18
than a small price change for a product whose weight is considerable.20 This asymmetry of perceptions could grow in periods of heightened public debate and media coverage of ination; typically, large price rises receive more attention than small increases or reductions. At a given ination rate, perceptions formed in the ways described are affected by the dispersion of the individual price changes. Suppose, for example, that there was an increase in the number of goods and services recording large upward or downward movements but that the changes cancelled each other out. If households assign more weight to increases than to decreases and to large changes than to small ones, at a given ination rate there would be an increase in perceived ination. To assess the importance of these effects, we calculated the dispersion of the ination rates of the individual goods and services included in the CPI, measured as the standard deviation of the cross-section distribution of their percentage changes with respect to the previous quarter, using data at a different level of disaggregation. Fig. 1.4 shows the dispersion of the quarter-on-quarter percentage changes of consumer prices on the basis of the 207- and 550-item breakdowns
3.0 2.5 2.0 1.5 1.0 0.5 0.0 1997 1998 1999 2000 2001 2002 2003
standard deviation - 550 items
standard deviation - 207 items
3.0 2.5 2.0 1.5 1.0
CPI
0.5 0.0
Fig. 1.4. Average consumer price ination in Italy and its dispersion (percentage points) (1) Source: Based on ISTAT data Notes: (1) Standard deviation calculated for the cross-section destribution of the quarteron-quarter percentage changes of the series with the level of disaggregation indicated.
20
The survey-based results reported in chapter 7 are consistent with the asymmetry hypotesis
19
provided by ISTAT.21 The dispersion of ination rates for the individual items was similar at both levels of disaggregation between 1999 and 2003, with the notable exception of the rst quarter of 2002, when it was markedly higher for the more disaggregated series (the spike recorded a year earlier was of comparable magnitude for the two breakdowns). This supports the view that the changeover was accompanied by rather large movements in relative prices that do not show up in the aggregate index. This initial nding is conrmed by an examination of the dispersion of the quarter-on-quarter changes in the elementary prices of a subset of 48 goods and services.22 Fig. 1.5 shows that the dispersion increased rapidly in the rst quarter of 2002 and, though less fast, also in the second half of 2003, when average ination was broadly stable. In 2002-03 the standard deviation was 4.4 percentage points, compared with 3.3 points in the previous ve years; the average of the price changes was in line with past levels, except for the rst half of 2002. In short, our ndings validate the hypothesis that relative prices changed to an unusually large degree during the cash changeover and that, given the asymmetrical perceptions of consumers, this may have appreciably inuenced their assessment of ination. The disaggregated data on individual prices point to other signicant changes in the distribution of price movements during and after the changeover. Fig. 1.6 shows that in the rst quarter of 2002 a much higher percentage of prices changed than in the same period of the preceding years (around 17 per cent, against an average of around 9 per cent). This difference is also evident in subsequent quarters, although to a smaller degree. The gure shows, in addition, that the proportion of large price increases (i.e. more than 5 per cent quarter on quarter) in 2002-03 was similar to that of the ve previous years and only
21
22
The rst breakdown corresponds to the level of disaggregation of the data published on ISTATs website, while the second is provided by ISTAT on request. Since the number of items at this level of disaggregation changes from year to year and it was around 550 in the period we are looking at, we refer to this set as the 550 items. Here we refer to the elementary price quotes (e.g. 250 g of a specic brand of coffee sold in a certain shop in a given city) recorded each month for the CPI. Although the sample is partial and not representative of the composition of the basket (the total weight of the 48 products corresponds to around 20 per cent of the total basket; see Veronese et al. 2005 for a detailed description of the dataset), it does allow us to examine price behaviour at the highest level of disaggregation; this makes it particularly useful for ascertaining whether broadly stable average ination in Italy at the beginning of 2002 co-existed with large movements in relative prices that may have been perceived asymmetrically by consumers.
20
1.0

7.0 standard deviation (right-hand scale)
0.5
average (left-hand scale)
5.0
0.0
3.0
-0.5
1997
1998
1999
2000
2001
2002
2003
1.0
Fig. 1.5. Average consumer price ination in Italy and its dispersion: subset of 48 goods and services (percentage points) (1) Source: Based on ISTAT data Notes: (1) Average and standard deviation calculated for the cross-section distribution of the percentage changes on the previous quarter of the elementary series of a subset of 48 goods and services included in the CPI basket.
slightly more pronounced in the rst quarter. By contrast, there was a greater increase in the share of prices that rose by less than 5 per cent or decreased. These ndings are consistent with the evidence presented by Gaiotti and Lippi (see Chap. 3) for the Italian catering sector. They showed that, compared with previous years, in 2002 there was a sharp rise in the percentage of restaurants that changed prices rather than a peak in the size of the changes. Prices may have been changed more often in 2002, in catering and other sectors, as a result of the decision by businesses to bring forward planned price changes to coincide with re-pricing in euros and so avoid incurring the same costs twice (in the case of restaurants, for example, the cost of re-printing menus). The gures reported in Table 1.2 show that in 2002-03 the increase in the proportion of prices raised or lowered involved products in every category. They also highlight a specic development in the service sector, where there was a substantial rise in both the share of quarterly price increases between 5 and 10 per cent and that of increases exceeding 10 per cent. This nding is also consistent with the results presented in chapter 3 for Italian restaurants, according to which in 2002-03 price increases were larger in the provinces where competition was low, and with studies of other euro-area countries showing large price rises for some services, particularly those with less competition (e.g. see Deutsche Bundesbank 2004).

8.0
21
8.0
6.0
decreases increases between 0 and 5% increases exceeding 5%
6.0
4.0
4.0
2.0
2.0
0.0
1998
1999
2000
2001
2002
2003
0.0
Fig. 1.6. Sign and size of price changes in Italy: subset of 48 goods and services (percentages) (1). Source: Based on ISTAT data. Notes: (1) The percentages shown are calculated on changes with respect to the previous quarter.
To summarize, the disaggregated data support the hypothesis that the price movements accompanying the changeover diverged somewhat from the past a higher percentage of price changes in general and of very large increases in the service sector and may have altered households ination perceptions, but are less visible from the aggregate indices alone. The differences with respect to the past do not appear to have been large enough on their own to justify such a surge in perceived ination, and may have been amplied by the link between pre-changeover expectations and post-changeover perceptions.
(c) Average basket and individual baskets
While the consumer price index refers to the average consumption of all resident households, the perceptions of an individual or category of individuals mainly reect movements in the prices of the goods and services composing a personal basket of consumption. Typically, a personal basket differs from the average basket according to the individuals social and economic situation (income, age, household status, education, etc.). For example, both essential and non-essential items have a signicant weight in the basket of wealthy households, while essentials predominate among the less well-off. Given these differences, an individuals personal ination rate and the average rate for the whole
22
Table 1.2. Sign and size of quarterly price changes in Italy by product categories: subset of 48 goods and services (percentages) (1) 1996
Negative 0 Between 0 and 5% Between 5 and 10% Exceeding 10% Total Negative 0 Between 0 and 5% Between 5 and 10% Exceeding 10% Total Negative 0 Between 0 and 5% Between 5 and 10% Exceeding 10% Total Negative 0 Between 0 and 5% Between 5 and 10% Exceeding 10% Total Negative 0 Between 0 and 5% Between 5 and 10% Exceeding 10% Total Negative 0 Between 0 and 5% Between 5 and 10% Exceeding 10% Total 23.3 44.0 30.1 2.5 0.1 100.0 11.5 78.5 3.3 3.0 3.7 100.0 4.9 90.1 1.8 1.7 1.6 100.0 1.1 95.4 1.2 1.4 0.9 100.0 0.4 97.2 0.6 0.7 1.1 100.0 4.0 89.7 3.0 1.7 1.6 100.0
1997
20.0 61.8 18.2 0.1 0.0 100.0 9.1 80.8 2.8 2.9 4.4 100.0 4.2 90.4 2.3 1.7 1.5 100.0 1.1 96.2 1.3 0.9 0.5 100.0 0.8 96.8 0.7 0.9 0.8 100.0 3.7 91.4 2.3 1.3 1.3 100.0
1998
36.9 56.2 6.7 0.2 0.0 100.0 8.4 81.6 3.0 2.5 4.5 100.0 3.5 92.2 2.3 1.2 0.8 100.0
1999
2000
2001
49.3 23.5 26.2 0.9 0.2 100.0 9.6 76.1 5.0 4.3 5.0 100.0 2.9 91.0 3.4 1.7 1.0 100.0
2002 2003
40.1 14.7 44.0 1.0 0.1 100.0 11.8 73.2 6.2 3.5 5.3 100.0 4.7 89.2 4.0 1.3 0.8 100.0 2.4 93.0 2.8 1.0 0.8 100.0 1.5 92.4 3.2 1.3 1.6 100.0 5.7 85.8 5.5 1.5 1.6 100.0 41.1 8.0 48.8 1.9 0.2 100.0 11.4 74.2 5.7 3.2 5.4 100.0 3.9 89.9 3.4 1.8 1.0 100.0 1.7 94.4 2.1 1.0 0.8 100.0 1.0 95.6 1.5 1.0 0.9 100.0 5.2 86.9 4.8 1.6 1.5 100.0
Energy products 13.0 23.4 30.3 23.0 51.0 47.5 5.6 4.9 0.1 1.1 100.0 100.0 Fresh foods 9.3 8.3 82.3 79.0 2.5 4.2 2.4 3.6 3.4 4.9 100.0 100.0 Processed foods 3.4 2.4 92.7 92.9 2.3 2.7 1.0 1.3 0.5 0.7 100.0 100.0
Non-food and non-energy products 1.1 1.5 1.2 1.2 96.4 96.0 96.0 95.5 1.0 1.1 1.3 1.5 1.0 0.9 0.9 1.2 0.5 0.5 0.7 0.7 100.0 100.0 100.0 100.0 0.7 97.1 0.8 0.7 0.7 100.0 4.1 91.8 1.8 1.1 1.1 100.0 Services 0.4 0.5 97.5 97.1 0.6 0.6 0.7 0.7 0.8 1.1 100.0 100.0 Total 3.2 3.2 91.0 90.1 3.6 3.9 1.3 1.5 0.9 1.3 100.0 100.0 0.5 97.0 0.7 0.8 1.0 100.0 4.7 89.1 3.3 1.6 1.4 100.0
Source: Based on ISTAT data. Notes: (1) The percentages shown are derived by calculating: (i) the quarteron-quarter percentage changes in the elementary prices considered; (ii) for each quarter, the percentages for the different intervals; and (iii) the annual average of the percentages obtained in step (ii).
23
population may hardly correspond at all. Moreover, the relation between the ination an individual experiences and the ination he or she perceives may vary with the consumer category to which that person belongs.23 The difference between the average and the individual basket may be considerable in the case of items with a large weight for their purchasers but that are purchased by a fairly small percentage of households and so have a modest impact on average ination. A very clear example of this is rent, which is a major expense for tenants but has no impact on the monthly expenditure of home-owners. In Italy, where only around 20 per cent of households live in rented accommodation, rent is a small item in the average basket (3.1 per cent), making a substantial difference between the average basket and the personal basket of anyone who pays rent. Further divergence at the individual level may reect the type of lease, geographical area, or city neighbourhood.24 Another reason for the gap between average and individually perceived ination is that the former is measured annually, while some products are usually purchased less than once a year. The weight of the price of durable goods, such as cars, in the CPI depends on the average frequency of purchase by Italian consumers. However, the items importance relative to total expenditure will be greater than average for people who actually made such a purchase in the last year, and will be nil for those who did not. To gauge the impact of the discrepancy between the individual basket of a category of consumers and the average basket, we made some estimates of ination by class of expenditure. These estimates are rough measures of the phenomenon as the available data allow us to capture the weight effect (i.e. the distribution by type of consumption for the segment of population in the group) but not the price effect (i.e. the prices paid by the households in the group, which depend on their shopping habits, including their preferences regarding retail outlets and types of products).25
23
24
25
The survey-based results reported in chapter 6 indicate in fact that individual ination perceptions vary depending on the consumption bundle they refer to and that a large majority of the consumers are not aware of the composition of ofcial basket. Avaible evidence suggests in fact that in recent years rents have increased much more in big cities than in smaller towns and for new leases more than for existing ones (Banca dItalia 2004, 2005, respectively). Capturing this second effect would require surveys of the different prices paid by consumers in each group, and these are not available. In this regard, it is to be noted that ISTAT observes the prices of several varieties of the same product at large distributors, but only the price of the best selling variety at traditional shops (ISTAT 2005). This is because by recording only the prices of best-selling varieties the institute can consider the consumer goods purchased by the majority of the population while minimizing the quantity of information collected every month. On this issue, see Campiglio (2004) and Del Giovane and Sabbatini (2005).
24
The estimates of the ination rate for each decile of equivalent expenditure show that the differences were small in 2002 and virtually nil in 2003 (Table 1.3).26 In 2002 average ination was slightly higher for the households with the highest consumption. This is because the households that consumed less were
Table 1.3. Average ination rate by decile of equivalent expenditure (year-on-year percentage change in the consumer price index; annual averages) (1) Decile of equivalent expenditure First Second Third Fourth Fifth Sixth Seventh Eight Ninth Tenth Total 2002 2.1 2.2 2.2 2.3 2.3 2.4 2.4 2.5 2.5 2.6 2.5 2003 2.5 2.6 2.6 2.6 2.6 2.6 2.5 2.6 2.6 2.4 2.6
Source: Based on ISTAT data. Notes: (1) Variations calculated on the basis of price indices rounded to 2 decimal places; for the annual average for 2003, this entails a slight difference with respect to the ISTAT gure (2.7 per cent), which is calculated on the index rounded to 1 decimal place.
hardest hit by the rise in food prices, which far outpaced the general index (3.6 as against 2.5 per cent), but beneted from the more favourable price dynamics for two items with a large weight in their basket: energy products (-1.9 per cent) and goods and services subject to price controls (2.2 per cent). In addition, the 3.9 per cent rise in the prices of unregulated services had much less impact for the least-well-off households (around 15 per cent of their shopping basket) than for the most afuent ones (around 35 per cent).
26
Baldini (2004a) and ISTAT (2003, 2007) obtain similar results using analogous data and methods.
25
These results, which, we repeat, do not take account of the price effect and are therefore only approximate, contrast with the widely held belief that the negative effects of the price movements of 2002 and 2003 were greater for low and middle-income households than for well-off ones.27 1.4.2 The individuals economic situation Perception of a larger rise in ination than shown by the ofcial rate and differences of perception between one type of household and another could be the result of assessments based not only on experienced ination but also on other factors affecting the individuals economic situation, such as increases in the prices of items not included in the index basket or income developments affecting various segments of the population. In other words, consumers might confuse the loss of purchasing power caused by ination with a more general impoverishment due to other factors. We shall consider two specic factors. The rst is the behaviour of house prices, which are not included in the consumer price index basket because the purchase of a dwelling involves both investment and consumption choices and it is difcult to distinguish between the two motives.28 If this exclusion were not clear to consumer survey respondents (who are unlikely to be familiar with the characteristics of the basket and the methodology of the ISTAT index), their responses would refer not so much to their ination perceptions as to their perceptions of purchasing power, which they would gauge taking into account the potential purchase of a house. The steep rise in house prices in 2002-03 by more than 28 per cent between the second half of 2001 and the second half of 2003 might have had an impact on perceptions of price trends in general and not just property prices. Moreover, depending on the distribution of home ownership among different segments of the population and the differences in the rate of increase in house prices in areas inhabited by different social classes, increases in house prices could have different effects on different categories of consumers.
27
28
For an attempt to evaluate the possible direction of the effect of different price changes in different varieties of a single product see Mostacci, Natale and Pugliese (2004). In most EU countries, including Italy, the consumer price index basket includes rent and, for both owner-occupiers and tenants, some housing-related expenditures, such as maintenance; however, it does not include expenditure by the home-owner for the purchase of the dwelling, nor other user costs, such as interest payments on mortgages. This approach was adopted by Eurostat for the harmonized consumer price index. Eurostat, in cooperation with national statistical institutes, is currently studying possible treatments of owner-occupied housing in the harmonized index. For more on the subject, see European Central Bank (2005).
26
As far as income developments are concerned, according to the Bank of Italys Survey of Household Income and Wealth in 2002 (Banca dItalia 2004) the changes in households annual disposable income with respect to 2000 varied considerably according to the work status of the head of the household (Table 1.4). Compared with a sample-wide increase of 6.8 per cent in nominal terms and 1.1 per cent in real terms, the income of households headed by selfemployed workers increased more (10.1 per cent in nominal terms and 4.4 per cent in real terms) than that of households headed by payroll employees (5.7 per cent in nominal terms, nil in real terms). The nominal income of households headed by blue and white-collar payroll workers rose by only 3.9 per cent, declining by 1.8 per cent in real terms. Households headed by pensioners recorded a gain of 6.4 per cent (0.7 per cent in real terms). Boeri and Brandolini (2004) show that differences in the trends in household income in recent years have affected the incidence of relative poverty the
Table 1.4. Italian household income by work status of the head of household, 2000-2002 Annual net household income Work status of the head of household (1) Euros 2000 Payroll employee of which: blue and whitecollar workers Percentage change 2002/2000 Annual net per capita income (2) Euros 2000 9,005 Percentage change 2002/2000
2002 nominal real (3) 5.7 0.0
2002 nominal real (3) 9,696 7.7 2.0
28,651 30,293
26,477 27,505
3.9 10.1 6.2 6.4 6.8
-1.8
8,306
8,798
5.9 9.7 7.4 7.2 8.0
0.2 4.0 1.7 1.5 2.3
Self-employed 36,568 40,245 Not in work 19,761 20,981 of which: pensioners 20,596 21,911 Total 26,098 27,868 Source: Banca dItalia (2004).
4.4 11,095 12,168 0.5 9,738 10,460
0.7 10,341 11,081 1.1 9,597 10,363
Notes: (1) Member with the highest income. (2) Annual net household income divided by the number of members. (3) Incomes were deated using the household consumption deator of the national accounts, which showed an increase in prices between 2000 and 2002 of 5.7 per cent.
27
share of low-income households among self-employed workers has diminished but it has increased among production and clerical workers although the indicators for the population as a whole have remained broadly stable. All told, although there appear to be no appreciable differences in 200203 in the ination experienced by the different classes of expenditure, the greater ination perceptions of some categories of households may have sprung from increased poverty due to factors other than ination.29 1.4.3 The media Price movements and, above all, the contrast between consumers perceptions and the ofcial statistics received enormous coverage in the media in the period following the cash changeover. This close coverage, especially when it gave greater resonance to declarations and studies disputing the ofcial estimates, may in turn have fuelled the perception of a generalized surge in prices.30 To obtain an idea of the extent of media coverage at the time, we selected articles published in two newspapers (Il Sole 24 Ore, Italys leading economic and nancial daily, and La Stampa, a newspaper with nationwide circulation) whose headlines included the expression inazione or caro-vita (a popular synonym for ination that combines the words expensive and life) and with the words associazioni or consumatori in the text.31 The annual data are reported in Table 1.5 and the monthly data in Fig. 1.7, which also show the gures on ination perceptions in the same period. For Il Sole 24 Ore, the annual data show a substantial increase in such articles in 2003 (94 articles, compared with 71 in 2002 and 65 in 2001) but not in the preceding year. However, Fig. 1.7 shows that in both 2002 and 2003 there were signicant peaks in that newspapers coverage of the subject: there are
29
30
31
The results presented in chapter 6 are consistent with this hypothesis. On these issues, see also Baldini (2004b), Golinelli and Parigi (2005) and Saraceno (2004). Boeri (2004) stressed the exceptional media coverage and suggested that it may have contributed to the gap between perceived and ofcial ination. He also noted that many newspapers lent credence to alternative, non-ofcial estimates of ination without checking their rigorousness. We chose these two newspapers because they were available in the same online database, allowing us to use the same search procedure to select the articles. It should be kept in mind that our articles came only from these two newspapers, one targeting a specialist readership, the other aimed at readers with aboveaverage levels of education and income, and that we did not include radio or television. Similar results to those presented below were obtained with an alternative search, in which we selected articles with headlines including at least one of the following three expressions for ination: caro-vita, caro-prezzi, costo della vita (see Del Giovane and Sabbatini 2005).
28
Table 1.5. Number of articles selected from Il Sole 24 Ore and La Stampa (Selection criteria: Inazione or caro-vita in the headline, associazioni or consumatori in the text) Il Sole 24 Ore La Stampa 1995 1996 1997 1998 1999 2000 2001 2002 2003 30 41 32 22 33 60 65 71 94 15 38 13 5 19 26 21 48 52 Total 45 79 45 27 52 86 86 119 146
ve months in which at least 10 articles appeared, compared with just a single instance in the seven previous years.32 In the case of La Stampa, the pronounced increase in coverage of the subject is very evident from the annual gures. The number of articles more than doubled between 2001 and 2002 (from 21 to 48) and rose again slightly in 2003. Moreover, Fig. 1.7 shows that the monthly peaks in the number of articles were easily the highest in the ten years considered, especially those recorded in the summer of 2002, when there was also an upsurge in ination perceptions. Many Italian newspapers published fairly sensationalist articles in that period, particularly those that gave ample space to criticisms of ISTATs methods and to alternative estimates of ination. There was also heavy media coverage between the end of 2002 and the beginning of 2003, when ination perceptions hit a post-changeover peak, and again in the autumn of 2003, when, after a brief decline, they rose back to the previous high.33
32
33
In March 2000, the attention of the press was justied by a rise in ination to a rate that was not particularly high in absolute terms (2.6 per cent) but that was roughly double the rate of a year earlier and the highest gure in more than three years. Among the topics covered in these two periods much space was given to the argument between Conndustria and Confcommercio (the confederations of Italian Industry and Italian Commerce, respectively) on the responsibility for the rise in prices and to the dispute between ISTAT and Eurispes on the accuracy of ofcial statistics. Of special interest is the storm over ISTATs admission of a measurement error in February 2002. The error was very small (a reduction in the prices of pharmaceuticals was attributed to January 2003 instead of to February, leading to a 0.1 percentage point underestimation of ination in January) and would have passed unnoticed in a less highly charged environment. On the episode, see various contributions on the website www.lavoce.info.
29
(a) Il Sole 24 Ore

70 60 50 40 30 20 10 0
Inflation perceptions (left-hand scale)
14 12 10 8 6 4 2
Number of articles (right-hand scale)
1995
1996
1997
1998
1999
2000
2001
2002
2003
(b) La Stampa
70 60 50 40 30 20 10 0
Inflation perceptions (balance, left-hand scale)
14 12 10 8 6 4 2 0
Number of articles (right-hand scale)
1995
1996
1997
1998
1999
2000
2001
2002
2003
Fig. 1.7. Ination perceptions and articles selected in Il Sole 24 Ore and La Stampa (Selection criteria: Inazione or caro-vita in the headline, associazioni or consumatori in the text)
Thus, in the period under review, we see that some abrupt rises in perceptions coincided with peaks in the number of articles containing key words bearing closely on the issue of the cash changeover and ination. Over the medium-term, a comparison of Fig. 1.6 and Fig. 1.1 shows that the previous phases of sharply deteriorating ination perceptions (1995-96 and 1999-2000), which were phases of rising ofcial ination, were indeed accompanied by an increase in media coverage, but not on the scale of the most recent period, when ofcial ination remained instead broadly stable. Our ndings suggest that there may be a relationship between the trend of ination perceptions and the attention the press pays to the phenomenon.
30
However, a priori, this causal link may work in both directions: a sharp deterioration in ination perceptions is newsworthy, and extensive media coverage may in turn validate and reinforce individual perceptions. This two-way link is conrmed by the Granger-causality tests we performed on the data presented in this section. 1.4.4 Rounding and recollection of lira prices According to ISAE (2003), consumers recourse to rounding may have inuenced their ination perceptions in 2002. Italian consumers supposedly converted prices mentally into euros at 2,000 lire to the euro instead of the correct rate of 1,936.27 lire. This meant an upward rounding of 3.2 per cent, which added to ofcial ination would have implied perceived ination of almost double the ofcial rate in 2002. This argument received considerable attention in the press when it was put forward by the president of ISTAT during Senate hearings in October 2003. Extending its study to the approximate conversion rates that were probably used in the other euro-area countries, ISAE calculated that the potential effect of rounding by consumers was positive by 2 to 3 per cent in Germany and Greece and 1 to 2 per cent in Austria and Ireland, negative by around 1 per cent in Belgium, and negligible in the remaining countries. The study also showed that in 2002 the gap between perceived and ofcial ination was generally higher in the countries where the potential effect of rounding was greatest.34 As ISAE itself points out, this argument may help to explain the gap between perceived and ofcial ination in 2002, when consumers were comparing prices denominated in euros (and mentally converted into lire) with lira prices in the previous year, but not the gap in 2003, when estimating the twelve-month change in prices involved comparing two prices expressed in euros. However, some consumers may have been mislead because in their mind they continued to refer to the last lira prices they had observed before the introduction of the euro. If this were so, then two types of approximation would have been at work. The rough and ready mental conversion of euro prices into lira would have caused a rst overestimation, followed by a second, equally important one caused by making a comparison, say, at the end of 2003 on the basis of a price from two years earlier. If these consumers did not realize that their
34
See also Ehrmann (2006) for a cross-country analysis of the relationship between the complexity of the currency conversion rates and the changeover effects on measured and perceived ination.
31
perceptions referred to a change over 24 months and not over 12, which is the standard lag for computing ination, this effect alone would have produced a perceived ination rate roughly twice as high as the actual rate (since annual ination in Italy was very similar in 2002 and 2003). The overestimation would be even greater if the last purchase to which the consumer mentally referred occurred well before December 2001 or if he had in mind not the last lira price paid for a good or service but an average of the prices observed over a certain span of time, such as the two or three years preceding the cash changeover. In both cases, the consumer would attribute to the post-changeover period a change that had actually accumulated over a much longer period of time. This type of effect can be particularly relevant in the case of the prices of rarely purchased goods and services (such as durable consumer goods) and of those that increased sharply over several consecutive years (restaurant meals). Evidence on the accuracy of consumers memory for prices is provided in chapters 5 and 6 of this book. In the former, Cestari, Del Giovane and RossiArnaud investigate Italian movie-goers recollection of pre-euro cinema prices, nding that only a small percentage of respondents recalled the correct price, and that the average prices recalled were much lower than the actual pre-euro prices and dated back to years before the changeover. In the latter, Del Giovane, Fabiani and Sabbatini obtain similar results for a different item (the newspaper) and a representative sample of italian consumers. Concerning the difculties consumers may have encountered in mentally converting prices, the European Commission survey, The uro, Two Years Later, conducted at the end of 2003, offers some interesting ndings. Italy was the euro-area country with the highest percentage of respondents experiencing a lot of difculty with the euro (29 per cent, against an area average of 14 per cent) and, together with Greece, the only country in which this percentage was higher than at the end of 2002 (when it had stood at 14 per cent). Italy also had the highest percentage of respondents who reckoned mentally in the old national currency for everyday purchases (46 per cent, compared with an area average of 30 per cent) and was the only country in which they outnumbered the respondents who thought in euros (33 per cent, compared with an area average of 46 per cent). 1.5 An overview of the correlations Our analysis so far suggests that a multiplicity of factors probably contributed to the large discrepancy between perceptions and ofcial measurements of ination. In this section we present the results of a descriptive analysis of the correlations between perceptions and the factors considered. We use a linear regression in which the dependent variable is the perception of ination
32
(balance of the responses in the European Commission survey; see Sect. 1.2) and the regressors include the factors in the preceding section for which we can construct a monthly indicator. The exercise is designed to measure the correlations between perceptions and the variables considered, and not necessarily to attribute a causal link to them, especially in the case of the variable referring to the media (see below). Table 1.6 contains the value of the estimated coefcients and a summary statistic of the correlation (the adjusted R2) for each regression.35 The term of comparison is the regression in which the only explanatory variable is ination measured by the consumer price index (CPI), whose R2 is equal to 0.68 for the estimation period 1997-2001 (regression 5 in the table) and drops to 0.56 when the period is extended to include the two years 2002-03 (regressions 1), exhibiting a discontinuity at the time of the cash changeover. The rst exercise (regression 2 in the Table) considers variables referring to the factors we discussed in section 1.4.1, attributable to the presence of asymmetries in the formation of perceptions and in the behaviour of prices:36 the twelve-month rate of increase in the prices of the most frequently purchased goods and services (CPI_high) and in the remaining ones (CPI_low); the standard deviation calculated on the cross-section distribution of the twelve-month percentage changes in the prices of a subset of 48 products in the ISTAT basket (STD).37 The coefcients have the expected positive sign and are statistically signicant for CPI_high and (marginally) for STD.38 The explanatory power is signicantly greater than when CPI alone is considered.
35
36
37
38
The estimation period we consider begins in 1997 in view of the availability of the indicator for the dispersion of price changes. Excluding this regressor, we also made the estimations starting from 1992, obtaining similar results. Coefcients for lagged variables were reported when statistically signicant. Indicators of the different composition of the shopping baskets by category of consumer were not included in the regression, as our analysis of ination by class of expenditure did not nd signicant differences. The variable that captures the importance of extreme price increases did not prove signicant, consistently with the evidence of Sect. 1.4.1(b) that at the time of the changeover their share in total price changes only increased in some sectors and marginally for the index as a whole. The coefcient of STD is probably of scant statistical signicance because some of the phenomena captured by the variable are similar to those captured with the breakdown of ination by frequency of purchase (the available evidence suggests that the dispersion was higher for frequently purchased goods). The variable has greater statistical signicance in the regression (not reported in the table) in which it is considered separately, together with the variable CPI. Only the two-period lagged value of STD proved to be signicant in regression 2.
33
Table 1.6. Summary of the correlations - dependent variable: balance of ination perceptions (1) Regressors
1 CPIt Coefcients(2):
Estimation period: 1997.4 2003.12
0.56
5 6
CPI 36.27 (10.06) Asymmetries CPI_hight, CPI_lowt, STDt-2 CPI_high 19.26 (10.95) CPI_low 2.30 (0.80) STDt-2 3.90 (1.86) Media CPIt, NEWSP t-1, NEWSP t-2, NEWSP t-3, NEWSP t-4 CPIt 14.65 (4.00) NEWSP t-1 1.34 (4.83) NEWSP t-2 0.78 (2.59) NEWSP t-3 0.68 (2.21) NEWSP t-4 0.73 (2.51) Asymmetries and media CPI_hight, CPI_lowt, STDt-2, NEWSP t-1, NEWSP t-2, NEWSP t-3, NEWSP t-4 CPI_high 9.12 (4.64) CPI_low 1.68 (0.57) STDt-2 3.75 (2.26) NEWSP t-1 1.14 (4.29) NEWSP t-2 0.71 (2.53) NEWSP t-3 0.50 (1.73) NEWSP t-4 0.61 (2.27) For comparison: estimation period 1997,4 2001,12 CPIt CPI_hight, CPI_lowt, STDt-2, NEWSP t-1, NEWSP t-2, NEWSP t-3, NEWSP t-4
0.68
0.77
0.80
0.68 0.79
Legend: R2 adjusted R-squared, CPI 12-month changes in the general index of consumer prices; CPI_high 12-month changes in the price index for frequently purchased goods and services; CPI_low 12-month changes in the price index for less frequently purchased goods and services; STD standard deviation calculated on the cross-section distribution of the 12-month percentage changes in the prices of a subset of 48 goods in the Istat basket; NEWSP number of articles published in Il Sole 24 Ore or La Stampa with the key words inazione or caro-vita in the headline and associazioni or consumatori in the text. Notes: (1) Regression of monthly ination perceptions (balance of responses to the ISAEs consumer survey; see section 1.2) on the variables indicated. (2) T-test in brackets.
34
In the second exercise (regression 3) we include, along with CPI, the variable that measures the media coverage of the debate on ination (see Sect. 1.4.3), given by the number of articles published in Il Sole 24 Ore or La Stampa containing the key words inazione or caro-vita in the headline and associazioni or consumatori in the text (NEWSP).39 Again, the coefcient has the expected sign and is signicant; R2 rises to 0.77. Given the two-way causal link between perceptions and the variable in question (see Sect. 1.4.3), this increase should be read as an indication of the possible importance of the reciprocal link between the collective formation of perceptions and the medias treatment of the subject. Lastly, when all the preceding factors are considered together (regression 4), the correlation rises to 0.80, coinciding with the gure we obtain with the same variables when we limit the estimation to the period preceding the changeover (regression 6). When the alternative regressors are included, the break found for the regression with only ofcially measured ination disappears. The same evidence is offered by Fig. 1.8, which shows the behaviour of perceptions and of the tted values obtained, respectively, with the regression of perceptions on ofcially measured ination alone and with the regression including all the alternative regressors.
80 70 60 50 40 30 20 10 0 -10
Inflation perceptions (balance)
80 70 60 50 40 30 20 10 0
Estimate with CPI alone (regression 1)
Estimate with alternative regressors (regression 4)
1997
1998
1999
2000
2001
2002
2003
-10
Fig. 1.8. Ination perceptions in Italy: balance resulting from the consumer survey and estimates Source: European Commission and authors estimates.
39
We excluded the contemporary value of the variable NEWSP, even though it proved signicant when included, because some of the articles surveyed each month could have appeared after that months consumer survey.
35
1.6 Conclusions
The paper shows that Italian households perception of a surge in ination following the euro cash changeover can be reconciled with the much more modest increase recorded by the ofcial statistics. The rst part of the paper shows that the information available does not suggest that the ofcial indices give a systematically distorted measure of ination, although the methodological treatment of specic items is open to improvement and problems in some phases of the data collection cannot be ruled out. Contrary to what is often publicly asserted, ISTATs statistics do capture the exceptionally high price rises for specic products, even if this is not fully discernable from the data released to the public as these are generally averages of single prices. Moreover, although the treatment of some specic items (e.g. vegetables and fruit, or motor-vehicle insurance) may have helped to create a gap with respect to perceptions, these elements are not enough in themselves to justify the general nature of the phenomenon or its magnitude. The evidence examined in the rest of the paper suggests that a combination of other factors contributed to the gap. First, individual consumers purchase the goods and services included in the ISTAT basket with differing frequencies, which range from every day to less than once a year. In the period following the changeover, frequently purchased products recorded larger price rises than less frequently purchased ones. If the former have a stronger impact on perceptions a hypothesis that is conrmed by psychological studies and is particularly plausible at a time when a large number of prices had to be memorized in a new currency these developments are consistent with a larger increase in perceived ination than in average ination measured by ofcial statistics. Second, during and following the changeover there was an increase in the proportion of prices that changed, up or down, and a higher percentage of very large increases in the service sector. Assuming consumers perceptions are inuenced more by a price increase than by an equivalent decrease or that they over-react to extreme price changes these developments are consistent with a gap between perceived and measured ination. In addition, the two effects above may have been amplied by the link that psychological studies have found between individuals anxiety about the possible inationary effects of the changeover and the way they perceived these effects afterwards. Third, the ination experienced by individual consumers reects the behaviour of the prices of the goods and services in their personal consumption basket. Since this differs from the average basket for the entire population, individual ination can diverge from the rate
36
calculated for the country as a whole. This may explain why perceived ination is higher for certain categories of consumers, particularly less well-off households. However, estimates of the ination experienced by households according to class of expenditure do not exhibit major differences in the two years 2002-03. This nding does not necessarily imply that the perception of a process of impoverishment on the part of some categories of consumers is unfounded. It may be the result not only of an evaluation based on experienced ination, but also of other factors affecting the individuals economic situation. In the period under review, a prominent role may have been played by the sharp rise in the price of items not included in the consumer price index basket (house prices) and the very modest increase in real household incomes: in 2002 these were barely higher than in 2000 for the population as a whole and lower in the case of households headed by production or clerical workers. Fourth, the media gave wide coverage to price movements and to the contrast between consumers perceptions and the ofcial statistics, even though ofcially measured ination was basically stable. Abrupt increases in perceptions coincided with unprecedented peaks in the number of news articles featuring words closely associated with the debate on ination. This suggests that the reciprocal inuence of ination perceptions and media coverage was a signicant factor. Last, an inaccurate recall of lira prices may have altered ination perceptions. In the case of a rarely purchased durable good, the consumer, thinking back to the last time he or she made such a purchase, may well recollect a time before December 2001. For other goods and services, the consumer might remember not a single price but an average of the prices observed over a period of time, such as the two or three years before the changeover to the euro. In either case, the consumer would attribute to the post-changeover period a change that had actually accumulated over a much longer time span.
37
References
Aalto-Setl (2006) Why do consumers believe that the euro raised prices?, International Journal of Consumer Studies, 30 (2), 202-206. lvarez Gonzlez, L.J., Cuadrado Salinas, P., Jareo Morago, J., Snchez Garca, I. (2004) El impacto de la puesta en circulatin del euro sobre los precios de consumo, Banco de Espaa, Servicio de Estudios, Documentos ocasionales, 0404. Aucremanne, L., Collin, M. (2005) Is there a discrepancy between measured and perceived ination in the euro area since the euro cash changeover?, paper presented at the OECD Seminar: Ination Measures: Too High Too Low Internationally Comparable?, Paris, June. Baldini, M. (2004a) Uninazione per ricchi. E una per poveri, www.lavoce.info. Baldini, M. (2004b) Prezzi, redditi e impoverimento delle famiglie, Il Mulino, 53 (412), 290-98. Banca dItalia (2004) Italian household budgets in 2002, Supplements to the Statistical Bulletin, XIV new series, 12. Banca dItalia (2005) Annual Report for 2004 (Italian unabridged version). Banca dItalia (2006) Household income and wealth in 2004, Supplements to the Statistical Bulletin, XVI new series, 7. Banco de Espaa (2003) The effect of the euro cash changeover on ination, Annual Report 2002, 95. Boeri, T. (2004) La psicosi del caro-euro, La Stampa, 2 February. Boeri, T., Brandolini, A. (2004) The age of discontent: Italian households at the beginning of the decade, Giornale degli Economisti e Annali di Economia, 63, 449-487. Brachinger, H.W. (2006) Euro or teuro? The euro-induced perceived ination in Germany, paper presented at the 2006 IAREP/SABE Conference, Paris, July. Buiten, G. (2003) Why ination still feels so high, CBS Webmagazine, www.cbs.nl. Buzzigoli, L., Di Iorio, F., Marliani, G. (2000) Il usso informativo dellindagine sui prezzi al consumo. Una rilettura critica, in C. Filuppicci (ed.) Tecnologie informatiche e fonti amministrative nella formazione dei dati economici, Franco Angeli, Bologna. Campiglio, L. (2004) Ination measurement, product diversity and rms price setting: new perspectives and empirical evidence, Rivista internazionale di scienze economiche e commerciali, 51, 19-34. Cestari, V., Del Giovane, P., Rossi-Arnaud, C. (2007) Memory for prices and the euro cash changeover: an analysis for cinema prices in Italy, Banca dItalia, Temi di Discussione, 619. Crdit Agricole (2004) Pourquoi les mnages franais surestiment-ils lination? Flash co 26, March. Del Giovane, P., Lippi, F., Sabbatini, R. (2005) (eds.) Leuro e linazione. Percezioni, fatti e analisi, Il Mulino, Bologna. Del Giovane, P., Sabbatini, R. (2005) The introduction of the euro and the divergence between perceived and measured ination: the case of Italy, paper presented at the OECD Seminar: Ination Measures: Too High Too Low Internationally Comparable?, Paris, June.
38
De Nederlandsche Bank (2002) Getting used to the euro, Quarterly Bulletin, September, 51-56. Deutsche Bundesbank (2004) The euro and prices two years on, Monthly Report 56 (1), 15-28. Dziuda, W., Mastrobuoni, G. (2005) The euro cash changeover and its effects on price tansparency and ination. Misson euro, mission accomplished!, Princeton University. Ehrmann, M. (2006) Rational inattention, ination developments and perceptions after the euro cash changeover, European Central Bank Working Paper Series, 588. Eurispes (2003) Caro cibo. Indagine statistica dellEurispes sui prezzi dei prodotti alimentari in collaborazione con la coalizione dei consumatori, www. eurispes.it. European Central Bank (2002a) Recent developments in consumers ination perceptions, Monthly Bulletin, July, 18-19. European Central Bank (2002b) Recent developments in perceived and actual ination, Monthly Bulletin, October, 20-23. European Central Bank (2003a) Effects of the introduction of the euro banknotes and coins on consumer prices, Annual Report 2002, 40-42. European Central Bank (2003b) Recent developments in euro area ination perceptions, Monthly Bulletin, October, 24-25. European Central Bank (2005) The harmonised index of consumer prices: concept, properties and experience to date, Monthly Bulletin, July, 55-69. European Central Bank (2007) Measured ination and ination perceptions in the euro area, Monthly Bulletin, May, 63-72. European Commission (2001) Euro attitudes (wave 6) euro zone, Flash Eurobarometer, 115. European Commission (2002) The euro, one year later, Flash Eurobarometer, 139. European Commission (2003) The euro, two years later, Flash Eurobarometer, 153. Eurostat (2001) Compendium of HICP reference documents, European Commission, Luxembourg. Fluch, M., Stix, H. (2005) Perceived ination in Austria extent, explanations, effects, Oesterreichische Nationalbank, Monetary policy and the economy Q3/05, 22-47. Golinelli, R., Parigi, G. (2005) Le famiglie italiane e lintroduzione delleuro: storia di uno shock annunciato, Politica Economica, 2, 201-225. Guiso, L. (2003) Inazione percepita e rilevata, www.lavoce.info. Hoffmann, J.H., Leifer, H., Lorenz, A. (2006) Index of perceived ination or EU consumer surveys? An assessment of Professor H.W. Brachingers approach, Intereconomics - Review of European Economic Policy, 41. INSEE (2003) Opinion des mnages sur les prix passs et lination, Note de conjoncture, December. ISAE (2002), Linazione effettiva e quella percepita dai consumatori, Quarterly Report, July, 114-120. ISTAT (2003) Caratteristiche del processo inazionistico nellanno dellintroduzione delleuro, Annual Report 2002, 64-79. ISTAT (2005) Come si rilevano i prezzi al consumo, www.istat.it.
39
ISTAT (2007) Indicatori della dinamica dei prezzi al consumo per alcune tipologie di famiglie Anni 2001-2006, February, www.istat.it Kahneman, D., Tversky, A. (1979) Prospect theory: an analysis of decision under risk, Econometrica, 47, 263-291. Kurri, S. (2006) Why does consumers perceived ination differ so much from actual ination?, Bank of Finland Bulletin, 3, 75-82. Lindn, S. (2006) 400,000 observations on ination perceptions and expectations in the EU. What will they tell us?, paper presented at the National Bank of Poland Workshop on the role of ination expectations in modelling and monetary policy making, February. Marques, J.F., Dehaene, S. (2004) Developing intuition for prices in euros: rescaling or relearning prices?, Journal of Experimental Psychology Applied, 10 (3), 148-155. Mostacci, F., Natale, G., Pugliese, E. (2004) Gli indici dei prezzi per sub-popolazioni, Contributi ISTAT, 3. Santos, D., Evangelista, R., Nascimento, T., Coimbra, C. (2002) Analysis of the impact of the conversion of escudos into euros, Banco de Portugal, Economic Bulletin, September, 1-14. Saraceno, C. (2004) Sentirsi poveri, www.lavoce.info. Traut-Mattausch, E., Schulz-Hardt, S., Greitemeyer, T., Frey, D. (2004) Expectancy conrmation in spite of disconrming evidence: The case of price increases due to the introduction of the Euro, European Journal of Social Psychology, 34 (6), 739-760. Tversky, A., Kahneman, D. (1974) Judgment under uncertainty: heuristics and biases, Science, 185, 1124-1131. Veronese, G., Fabiani, S., Gattulli, A., Sabbatini, R. (2005) Consumer price behaviour in Italy: evidence from micro CPI data, European Central Bank, Working Paper Series, 449. Walschots, J. (2002) Why does ination feel so high?, CBS Webmagazine, October, www.cbs.nl.
2. Rounding and anomalous changes in Italian consumer prices in 2002

Franco Mostacci* and Roberto Sabbatini** * ISTAT, Price Statistics Department ** Bank of Italy, Department of Economic Outlook and Monetary Policy Studies
2.1 Introduction1
One of the main concerns before the euro cash changeover was that the rounding of prices in the new currency would be mainly upwards, increasing consumer price ination (European Central Bank 2002). Community rules called on economic agents to convert and round prices to the nearest cent,2 but this could not rule out the possibility that the changeover would have an adverse impact on prices, in connection with the likelihood that rounding to what vendors considered attractive prices in euros would be mainly upward. In the run-up to the changeover there was a concern that the switch might give merchants the chance to raise their prices, though it was believed that the neutrality of the redenomination and the operation of market forces would ward off such effects. Empirical studies by central banks and statistical institutes during 2001 showed that the proportion of attractive prices in the national currencies was quite high, conrming the risk that from January 2002 there might be a signicant inationary impact from euro price rounding. Here we use a sample of monthly elementary price quotes that go into ISTATs calculation of the general consumer price index (CPI) to estimate the inationary impact of rounding with the changeover in Italy. Section 2.2 reports the prior estimates made in 2001, section 2.3 presents the ex post estimates for
1 2
The authors thank Moneta e Credito for permission to use material published (in Italian) in issue No. 221, March 2003. Council Regulation 1103/97, 17 June 1997, xed the rules for rounding prices converted into euros (European Commission, The introduction of the euro and the rounding of currency amounts on the Commissions website; see also Banca dItalia 2001).
F. Mostacci and R. Sabbatini
the period following the changeover, and section 2.4 supplements this analysis with an inquiry into the consumer price impact of the anomalous variations registered in 2002, whether in connection with rounding or not. The objective is to provide a range of estimates permitting an approximate evaluation of the order of magnitude of the inationary effects that can be reasonably attributed to the introduction of the new currency. The results are summarised in section 2.5.
2.2 The ex ante estimate

Empirical studies in various euro-area countries in 2001 were directed to two points (see the appendix to Chap. 7). Was the incidence of attractive prices in national currency such as to warrant fears of a signicant inationary impact due to the rounding of euro prices to produce equally attractive gures? And if so, assuming that merchants who set attractive prices in national currency would continue to do so in euros, what inationary impact would the changeover have through this channel? All these works postulated three types of attractive prices: psychological prices (e.g. 1.99 instead of 2.00), used in the belief that buyers unconsciously undervalue the last digit or two; fractional prices (e.g. 1.70 instead of 1.67), to simplify making change and make payments faster and easier; exact prices (e.g. 50.00), generally for large amounts, avoiding the use of coins and small notes in making change. The quantitative estimate of the incidence of attractive prices on the price observations for the CPI, hence the estimate of the inationary impact of rounding, depended on correct identication of attractive price levels in the two currencies and on the rounding behaviour of individual agents, which was hard to predict.3 For Italy, the analysis was based on some 90,000 elementary price quotes, or more than 30 per cent of ISTATs monthly observations for the Italian CPI, taken in September 2001 in the regional capitals.4 In all, these prices were representative of more than 65 per cent of the products included
3 4
See Mostacci and Sabbatini (2001) for the assumptions adopted in the empirical analysis. The data are collected in each municipality; quotes are for precisely specied goods and services (e.g. 1 kg of a given brand of spaghetti in a given store). For more detail, including the weighting used to aggregate the elementary quotes, see Mostacci and Sabbatini (2003).
Rounding and anomalous changes in Italian consumer prices in 2002
in the general consumer price basket at the time of the study (for a more thorough discussion of the choice of items, see Sect. 2.3.3). Attractive prices in lire were very common in Italy, accounting for some 90 per cent of our elementary quotes (Fig. 2.1), warranting fears of an inationary impact on the occasion of the changeover.
1,6 13,1 9,5 46,7
29,2 exact fractional psychological in euros (2) non-attractive
Fig. 2.1. Types of prices in Italy, autumn 2001 (percentage shares) (1) Source: Authors calculations based on ISTAT data. Notes: (1) Percentage share of each type in total price quotes in the sample (weighted quotes). (2) Considering the possibility that prices may already have been implicitly set in euros before January 2002; these are lira prices ending with a digit other than 0 or 5 (e.g. 1,936 lire = 1.00).
To estimate this impact, in addition to the rules dening attractive prices we had to make assumptions on how vendors would behave. We assumed that vendors with attractive prices in lire would continue to have them in euros. Three rounding scenarios were designed: rounding to an attractive level systematically upward (worst case scenario), systematically downward (virtuous scenario), or to the nearest attractive level (symmetrical scenario). We calculated the total impact on consumer prices, i.e. the cumulative variation in the CPI between the month of the survey and the hypothetical month in which all the adjustments are assumed to have been completed. The resulting estimates were for a 1 percentage point increase in the price level in the worst case scenario, a very slight decline (-0.1 point) in the symmetrical scenario, and a reduction of nearly 1 percentage point in the virtuous scenario (Mostacci and Sabbatini 2001). The studies for the other countries reached similar conclusions (see the appendix to Chap.
7). Attractive prices (exact, psychological or fractional) were common everywhere, and the worst-case impact averaged about 1 percentage point for the euro area.
2.3. The ex post estimate

2.3.1 Denition and estimation of the rounding effect After the cash changeover, empirical studies in Italy and elsewhere estimated the impact of rounding on price observations for the CPI. For Italy, Mostacci and Sabbatini (2003) studied the data gathered by ISTAT between December 2001 and October 2002 (the last month for which data were available at the time). Here we supplement the main ndings of that study with new evidence on the performance of some specic prices not considered earlier (in particular, unprocessed foods). In assessing the results, some limitations of the exercise should be borne in mind. First, the aim was only to quantify the effects of rounding, albeit applying both more and less restrictive interpretations. Second, rounding on the occasion of the cash changeover could have reected unrelated phenomena (say, demand and supply shocks). For comparability with the ex ante estimates, the analysis used the same sample of quotes. In this regard, some specications are needed. It was decided to exclude some items in both exercises: consumer durables and some services, including rents, for which the high level of prices made the effect of rounding likely to be negligible; and energy products, whose prices are not generally rounded. A supplement to the analysis was necessary for some fresh foods (fruit, vegetables, sh) that were excluded from the sample for methodological rather than economic reasons (see below). Finally, public service charges and regulated prices were excluded in view of the pledge by national governments that rounding would be neutral (or, if anything, in the consumers favour). Ex post, this decision no longer seemed fully justied, and we found it necessary to supplement the analysis by looking at the performance of these prices as well. In what follows we describe the ex post exercise. Section 2.3.2 summarizes the results and section 2.3.3 describes the supplements on specic items. The rules used to distinguish attractive from non-attractive prices in euros are the same as those of the ex ante study (Table 2.1).
Table 2.1. Rules for dening attractive prices in euros (1) Price in euros Exact or fractional Below 5.00 Over 5.00 Type of price Psychological Non-attractive *,XX *,XX
*,00; *,X0; *,X9; *,90; *,95; *,X5 *,99 *,00; *,50 *,X9; *,90; *,95; *,99
Notes: (1) Table gives the two decimals of the price in euros. X stands for a digit other than 0 or 9.
The timing of individual vendors shift to attractive prices in euros was not known in advance, and after the changeover, and even after the period of dual circulation (February 2002), some outlets continued to set prices in lire, naturally posting prices in both currencies. This might have been justied assuming, for instance, that consumers were still oriented to attractive prices in lire, which were obviously not necessarily attractive in euros. Both of the methods described below trace the course of each elementary price quote from December 2001 to October 2002 to measure the impact of rounding only in the month in which it rst becomes attractive in euros. Tracing individual prices restricts the methods that can be used to estimate the effect of rounding. Since with the change of base in December of every year the quotes observed over the next year may change (for instance, due to the rotation within each city of the sample of stores surveyed), the time series at this level of detail are generally short. We use two methods to measure the effects of rounding. The rst refers to the smallest variation necessary to produce an attractive price in euros. The ndings of this method should thus be read as an estimate of the effects of rounding in the strict sense. The second denes rounding in the broad sense as all changes in connection with the move to an attractive price are taken, though they are not necessarily entirely due to this motive. In Mostacci and Sabbatini (2003) and in comparable studies in other countries, the estimates of the two methods have been interpreted as the lower and upper bounds of the rounding effect.
Method 1: rounding in the strict sense (to nearest attractive price level)
This method assumes that in month t, when a given quote rst becomes attractive, the part of the change attributed to rounding is only the difference between the price at t-1 and the attractive price nearest to that quote, under the rules set out in Table 2.1. Rounding is either upward or downward
depending on how the price itself moved between the two observations. For example, if a quote goes from 1.36 to 1.45 (an attractive price by our rules), the portion of the rise attributed to rounding is just the difference between 1.36 and 1.39 (the nearest attractive price). If at t the price had gone down to 1.25, the rounding effect would be negative, equal to the difference between 1.36 and 1.35.
Method 2: rounding in the broad sense (difference between changes in the prices that become attractive each month and changes in the other quotes)
The measurements according to the second denition are based on a broader criterion; this is a two-phase method, which can be summarized as follows. Phase 1. In month t, under the rules of Table 2.1 for dening attractive prices, the elementary quotes are classed into four groups:
Prices in month t Prices in month t-1 Attractive Non-attractive Attractive S1A S2 Non-attractive S1C S1B
This classication enables us to isolate the subset of prices that were not attractive in t-1 and become so for the rst time in month t (S2). To estimate the impact of rounding, for each month these subsets are regrouped into two sets: S1=S1AS1BS1C: the set of prices that were attractive in month t-1 and remained so in month t (S1A), plus those that remained or became non-attractive in month t (S1B and S1C respectively).5 For all the quotes in this set (whose weight is denoted by W1), we calculate the percentage change on the previous month (VC1). By construction, this variation is not affected by rounding to attractive prices in euros but reects the other causal variables of price changes. S2: the set of prices that were not attractive in month t-1 and become attractive for the rst time in month t. Their weight is denoted by W2, and
The subset S1C serves to allow for the possibility of a previously attractive price reverting to non-attractive. Actually, these cases are in fact quite rare. And in any event, since their rounding impact was already taken into account in some month prior to t, it is not reconsidered if in a series of changes it reverts to non-attractive and then becomes attractive once more.
their overall change on the previous period by VC2; by construction, this represents both the effect of rounding and the effect of the other causal variables of price changes. Phase 2. The impact of rounding on the general price index in month t, IM2, is given by: IM2=(VC2-VC1)*W2 (2.1) This estimate reects a series of considerations. Part of the change in the prices becoming attractive in month t may be due to other factors that would have acted regardless of the cash changeover. To take this into account, the impact of rounding is estimated with reference to the difference between the average change in the quotes that become attractive for the rst time (set S2) and that in those that do not (set S1). We distinguished between goods groceries and non-groceries6 and services, in order to calculate the difference between VC2 and VC1 for relatively homogeneous product categories for which the causal factors in price trends should be similar. The overall impact in month t is thus given by: IM2=(VC2grocery-VC1grocery)*W2grocery+(VC2non-grocery-VC1non-grocery) *W2non-grocery+(VC2services-VC1services)* W2services. (2.2) For each month we repeat phases 1 and 2. Note that in the entire period from January through October 2002, for each elementary quote the impact of rounding is considered only once, in the month in which the price rst becomes attractive. All subsequent movements are attributed to other factors. In short, the set of prices rst becoming attractive in each month is obviously the same in both methods. The essential difference is that under Method 1 the rounding effect is as small as possible; for instance, if a quote goes from 0.73 to 0.80 the part imputed to rounding is just the difference between 0.73 and 0.75, the nearest attractive level. This is rounding in the strict sense, the lower limit of the rounding impact. The assumption underlying Method 2, i.e. that non-rounding factors have an equal impact each month on the prices that become attractive and on the others, may be extreme in that there is a risk of ascribing to rounding too large a part of the actual price change. This is thus termed rounding in the broad sense and can be considered as an upper limit to the rounding impact.
In addition to food, groceries includes home cleaning and personal hygiene products.
2.3.2 The main results With the cash changeover the Italian price system underwent a substantial adjustment. The share of attractive prices fell immediately from about 90 per cent to 20 per cent in January 2002, then recovering gradually to just over 50 per cent in October (the other countries showed similar patterns; see Chap. 7). A rst indication of the inationary impact of rounding is derived from a comparison between the changes in prices that became attractive and those that did not. These data are given in Tables 2.2-2.4. Table 2.2 shows that the prices becoming attractive rose more than the others, especially in September and October, the months when rms traditionally revise prices.
Table 2.2. Price increases by type of price (percentage change on previous month) (1) 2002 attractive (1) January February March April May June July August September October Cumulative ination, Jan.-Oct. 2.37 2.74 2.25 2.30 2.71 2.10 2.37 2.53 4.45 3.37 Type of price non-attractive 0.14 0.26 0.07 0.09 0.23 0.07 0.06 0.10 0.21 0.16 CPI 0.25 0.47 0.16 0.18 0.42 0.12 0.11 0.18 0.40 0.30 2.40
Notes: (1) The percentage change in attractive prices is calculated, for each month, with reference to the set of quotes designated S2 (i.e. those that become attractive for the rst time in that month).
Tables 2.3 and 2.4 highlight two additional characteristics of these developments. First, the rises in attractive prices in the service sector are perceptibly greater than for goods (Table 2.3). Second, there is a marked difference according to distribution channel: attractive prices are raised much more in traditional than in modern distribution channels (Table 2.4).
Table 2.3. Price increases by type of price and type of product (percentage change on previous month) 2002 Total Groceries (1) Other goods Services attractive nonattractive nonattractive non(2) attractive total (2) attractive total (2) attractive total January February March April May June July August September October 0.25 0.47 0.16 0.18 0.42 0.12 0.11 0.18 0.40 0.30 1.84 0.83 0.95 1.40 1.75 1.95 2.26 1.67 2.60 2.55 0.19 0.07 0.05 0.14 0.14 0.16 0.10 0.11 0.27 0.11 0.26 0.10 0.09 0.18 0.20 0.22 0.17 0.15 0.38 0.20 0.41 1.91 1.88 1.73 1.85 2.36 2.14 0.80 4.23 2.99 0.09 0.21 0.08 0.08 0.11 0.02 0.05 0.07 0.18 0.25 0.10 0.32 0.19 0.16 0.28 0.06 0.08 0.10 0.32 0.45 3.34 3.60 4.87 3.98 3.98 2.02 2.81 4.81 6.01 5.63 0.15 0.48 0.07 0.08 0.42 0.07 0.05 0.12 0.20 0.10 0.40 0.91 0.20 0.21 0.74 0.11 0.09 0.29 0.49 0.22
Cumulative ination, Jan.-Oct. 2.40 Notes: (1) Foods, home cleaning and personal hygiene products. (2) Identied by the rules given in Table 2.1. The percentage change in attractive prices is calculated, for each month, for the S2 set of prices (those becoming attractive for the rst time in that month).
The impact of rounding on the CPI in 2002 under the two methods is estimated in Tables 2.5-2.6. Table 2.5, on rounding in the strict sense, shows an impact of about 0.2 percentage points on the cumulative rise in the CPI between January and October 2002. That is, without this effect ination over the period would have been 2.2 per cent rather than the 2.4 per cent actually recorded. For rounding in the broad sense, Table 2.6 shows the impact over the same period at 0.8 percentage points, which is to say that in the absence of this effect cumulative ination would have been 1.6 per cent. Both methods conrm the relevance suggested by Tables 2.3-2.4 of the type of product and the distribution channel. The impact of rounding on service prices is twice as great as on goods prices. And the effect is signicant only for products sold through traditional channels; for mass retailers it is negligible (Tables 2.5-2.6). The difference between the two channels only partly reects differences in composition (in particular the fact that services are always classed as traditional distribution). Removing the estimated impact in the services sector from the total estimate for the traditional distribution to compare goods only under Method 2, which produces the greatest rounding
10
impact, the estimated impact for goods sold in traditional outlets is around 0.3 percentage points, compared with practically nil in the modern outlets.
Table 2.4. Price increases by type of price and distribution channel (percentage change on previous month) 2002 Total Modern retail outlets attractive non(1) attractive January February March April May June July August September October Cumulative ination, Jan.-Oct. 0.25 0.47 0.16 0.18 0.42 0.12 0.11 0.18 0.40 0.30 2.40 1.82 0.15 0.53 0.82 1.76 1.65 1.51 0.71 2.17 1.30 0.14 0.01 0.01 0.15 0.19 0.17 0.07 0.10 0.19 0.11 total 0.20 0.02 0.04 0.17 0.28 0.22 0.11 0.13 0.30 0.16 Traditional outlets attractive non(1) attractive 2.47 3.02 2.62 2.59 2.85 2.30 2.77 3.01 5.10 3.77 0.14 0.32 0.08 0.08 0.23 0.05 0.06 0.10 0.22 0.17 total 0.26 0.57 0.19 0.19 0.46 0.10 0.11 0.19 0.42 0.34
Notes: (1) Identied by the rules given in Table 2.1. The percentage change in attractive prices calculated, for each month, for the S2 set of prices (those becoming attractive for the rst time in that month).
2.3.3 Some extensions As noted, some items are excluded from the foregoing empirical analysis: regulated prices,7 unprocessed food (fruit, vegetables and sh), unregulated energy products (petrol), rents, certain durable goods (e.g. automobiles, motorcycles and personal computers), and nancial and insurance services. For some of them items the impact of rounding is probably very modest, so their exclusion does not alter the results. The contribution of upward rounding
7
The items involved are: tobacco goods, electricity, gas, drinking water, rail transport, maritime transport, urban transport, inter-city and suburban transport, taxis and delivery services, motorway tolls, the state television subscription fee, postal services, telephone services, medical services, betting pools, urban solid waste removal, certicates and licence duties, secondary education, university education, funeral transport, and medical products.
11
of durable goods, insurance and rents, in particular, is marginal, since these items have very high unit prices and the impact of any rounding up to attractive thresholds in euros is accordingly limited. Petrol prices are generally not set at attractive levels anyway, and they change frequently, often by as little as a few cents, following world crude oil price movements.
Table 2.5. Inationary effect of rounding in the strict sense (Method 1) by type of product and distribution channel (1) 2002 Ination Rounding in month effect (2) Type of product Groceries Other (3) goods January February March April May June July August September October Total 0.43 0.34 0.25 0.25 0.25 0.08 0.17 0.17 0.17 0.25 2.40 0.01 0.02 0.02 0.01 0.02 0.01 0.01 0.01 0.02 0.02 0.16 0.00 0.00 0.00 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.04 0.00 0.00 0.01 0.00 0.01 0.00 0.00 0.00 0.00 0.01 0.04 Distribution channel
Services Modern Traditional 0.01 0.02 0.01 0.01 0.00 0.00 0.00 0.01 0.01 0.01 0.08 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.01 0.02 0.02 0.01 0.02 0.01 0.01 0.01 0.02 0.02 0.14
Notes: (1) The impact is in percentage points. It is estimated by multiplying the percentage change in each price becoming attractive during the month to the nearest attractive level by its weight in the basket, assuming that the items not included underwent no rounding effect in that month. (2) CPI (percentage change on previous month). (3) Foods, home cleaning and personal hygiene products.
The case is different for regulated prices and unprocessed food. The former were excluded for consistency with the ex ante exercise, but if this exclusion was justied ex ante by governments commitment to ensure that conversion would be neutral (or actually advantageous to consumers), in the ex post exercise it is proper to evaluate the actual performance of these items.8 For unprocessed food, the exclusion from both exercises is motivated methodologically (these prices are surveyed twice a month, which complicates
8
From a methodological viewpoint, the complexity of the pricing structure makes it very hard to verify how regulated prices were rounded on the basis of the elementary data (consider, for example, the structure of the telephone bill).
12
treatment of the elementary data with respect to the two methods utilized in the empirical analysis of rounding). Accordingly, for both regulated prices and food products we considered it useful to develop a separate estimation of the price rises recorded on the occasion of the changeover. For the other items not included in the two exercises, Table 2.7 compares the price changes in the two years 2002-03 with those of the preceding years.
Table 2.6. Inationary effect of rounding in the broad sense (Method 2) by type of product and distribution channel (1) 2002 Ination Rounding in month effect (2) Type of product Groceries Other (3) goods January February March April May June July August September October Total 0.43 0.34 0.25 0.25 0.25 0.08 0.17 0.17 0.17 0.25 2.40 0.07 0.13 0.06 0.06 0.12 0.03 0.03 0.05 0.12 0.09 0.75 0.01 0.00 0.01 0.01 0.01 0.01 0.01 0.00 0.02 0.01 0.09 0.00 0.03 0.02 0.02 0.04 0.01 0.01 0.01 0.03 0.05 0.22 Distribution channel
Services Modern Traditional 0.06 0.10 0.03 0.03 0.07 0.01 0.01 0.04 0.07 0.03 0.44 0.01 0.00 0.00 0.00 0.01 0.01 0.01 0.00 0.01 0.00 0.05 0.06 0.13 0.06 0.06 0.11 0.02 0.02 0.05 0.11 0.09 0.70
Notes: (1) The impact is in percentage points. It is estimated by multiplying the percentage change in each price becoming attractive during the month by its weight in the basket, assuming that the items not included underwent no rounding effect in that month. (2) CPI (percentage change on previous month). (3) Foods, home cleaning and personal hygiene products.
For regulated prices, in general the commitment to make rounding neutral or advantageous for consumers appears to have been honoured in Italy; the increase in the prices of regulated goods and services in 2002 was close to the lows of recent years. Betting pools are an exception: in January 2002 the minimum bet for the Totocalcio, Totogol, Totosei and Totip pools was raised from 1,600 lire to 1.00 (an increase of 21 per cent) and that for Superenalotto from 1,900 lire to 1.00 (1.9 per cent). The impact of these increases on average monthly ination in January is estimated at 0.05 percentage points.9
9
Note that the calculation of the index does not take into account the fact that the higher cost of betting was accompanied by an increase in the total prize payout.
13
Table 2.7. Changes in the consumer prices of the items not included in the estimations of the changeover effect (quarterly data; percentages) Unprocessed Unregulated food energy goods Durable goods Financial services (1) Rents Regulated goods and services (2)
1 12 1 12 1 12 1 12 1 12 1 12 month months month months month months month months month months month months
1997 - I II III IV 1998 - I II III IV 1999 - I II III IV 2000 - I II III IV 2001 - I II III IV 2002 - I II III IV 2003 - I II III IV
-0.2 1.2 0.4 3.9 -0.6 0.0 -1.6 -1.2 1.0 0.4 -1.0 -2.1 0.1 1.4 0.1 0.8 -0.5 0.7 -0.1 0.6 1.3 0.4 -0.5 0.7 1.4 0.4 -2.0 0.2 1.4 1.8 -0.5 1.2 3.0 2.5 0.4 0.6 3.1 1.5 -1.3 1.3 1.3 1.5 1.1 1.9 1.0 1.5 2.0 1.9 -1.6 -1.3 -0.3 -0.9 -2.1 0.5 -2.2 0.4 -2.5 -0.2 -4.1 -0.4 0.3 0.4 0.3 0.5 0.1 0.4 0.4 0.5 0.3 0.4 0.3 0.2 0.7 0.6 0.2 0.3
0.9 0.4 0.1 0.5 1.6 1.7 1.4 0.4 0.2 0.2 0.7 1.6
1.2 1.2 2.4 0.8 2.8 1.0 3.0 0.9
5.1 5.7 5.8 5.7 7.4 7.2 7.8 8.0
1.3 1.3 1.1 1.8 0.9 1.8 0.6 1.1 0.5 1.1 0.4 0.6 0.6 0.5 0.6 0.5 0.6 0.4 0.7 0.5 0.5 0.7 0.4 0.9 0.7 0.7 0.5 0.7
7.9 6.7 6.2 5.7
2.6 1.7 0.3 0.2
2.2 3.6 5.2 4.8 3.1 2.2 2.1 1.8 0.6 0.0 1.0 1.7 2.4 3.2 3.2 3.2 5.2 4.5 2.8 2.3
5.3 0.9 5.8 0.8 5.2 0.2 4.5 -0.1 4.0 -0.3 3.4 0.1 3.2 1.2 2.7 0.6 2.8 2.2 2.5 2.4 0.4 0.9 1.2 0.7
2.0 -0.2 -2.8 2.0 4.4 2.9 0.5 3.4 6.7 0.0 2.3 10.2 0.0 1.4 2.9 3.8 3.6 2.7 4.2 1.6 14.4 12.5 13.3 12.6
2.8 8.0 3.1 10.1 5.0 12.2 2.7 14.2
1.4 2.4 13.8 1.4 -1.8 8.4 1.6 0.6 3.9 1.5 0.7 1.8 1.7 1.6 1.5 1.2 1.6 1.8 1.7 1.8 1.5 1.2 1.2 0.8 3.0 2.4 5.3 9.8 1.8 11.1 1.5 12.0 1.9 10.8 3.0 8.4 1.9 8.6 1.1 8.2 3.6 10.0 0.5 7.3 0.7 6.0 1.3 6.2
5.5 -5.2 3.0 6.2 2.2 2.5 7.3 -2.3 -3.8 6.6 -4.1 -9.2 6.8 -1.3 -5.5 5.8 4.4 -3.4 4.0 -0.7 -1.8 4.7 0.9 3.4
2.4 2.3 2.2 0.2 2.2 -0.3 2.2 0.2
2.1 0.4 0.4 2.4 -0.5 -0.4 2.1 0.6 0.6 2.5 0.3 0.7 2.7 2.8 2.9 2.7 0.7 1.3 0.0 0.1 1.1 2.9 2.3 2.2
2.8 3.5 8.4 0.4 2.8 -3.2 0.5 0.3 5.3 -0.2 1.0 0.2 5.9 -0.4 -0.4 -0.1
Source: Based on ISTAT data. Notes: (1) Insurance and banking services. (2) Excluding rents.
As regards unprocessed food, bad weather was blamed for the historically high twelve-month price rises recorded in January 2002 in nearly all the euroarea countries (7 per cent in Italy and 8 per cent on average for the area). It took considerably longer for these strains to abate in Italy than elsewhere:
14
between the rst and third quarters of 2002 the year-on-year rate of increase in the prices of unprocessed food fell by almost 6 percentage points in the area as a whole (harmonized index of consumer prices) and by just 2.5 points in Italy. This suggests that the increases in Italy may have reected not only the weather but also pricing decisions connected with the introduction of the euro.10 We ran a simple econometric exercise to distinguish the latter effect from the former, regressing the monthly changes in the prices of unprocessed food on seasonal dummies and monthly mean temperatures, nding that weather conditions account for only 0.7 percentage points of the 2.2 per cent price rise recorded in January 2002. Accordingly, the changeover effect could have accounted for up to 1.5 points of the January rise in the prices of unprocessed food, with an impact on the consumer price index of approximately 0.1 points.11 2.3.4 Summary of the results The results are summarized in Table 2.8, which reports the estimates of rounding obtained on the basis of elementary quotes and those for the items not included in the exercise, obtained as described. Rounding up to attractive levels of goods and services for which the elementary quotes were analyzed had an overall inationary impact of between 0.16 and 0.75 percentage points in the period from January through October 2002 (Table 2.8; rst line, columns A and B). These estimates do not take account of the fact that in October 2002 (the last month for which elementary quotes were available) the proportion of attractive prices was around 30 percentage points lower than it had been estimated for the price distribution in lire. Assuming that the two distributions, in lire and in euros, resemble each other once all price adjustments have been completed, and that the effect of rounding on consumer prices is similar to that estimated using the available micro-data, the total impact of rounding works out to between 0.2 and 1 percentage point (rst line, columns C and D). Adding the contribution of unprocessed food and regulated prices, the ultimate impact of rounding, all adjustments made, works out to between 0.4 and 1.2 percentage points of additional cumulative ination. Average ination in 2002, as measured by the year-on-year increase in the CPI, was 2.5 per cent; considering only the share of rounding actually recorded
10
11
It should also be borne in mind that in Italy, unlike the other European countries, the method used for calculating the price index for fresh fruit and vegetables attenuates the impact of supply shocks on consumer prices (see Chap. 1). Similar results are obtained by restricting the analysis to fruit and vegetables.
15
from January to October 2002 and the estimated monthly impact on prices, the contribution of the currency changeover through the upward rounding to attractive thresholds in euros on the average annual ination in 2002 is estimated to lie between 0.2 and 0.5 percentage points of extra ination.
Table 2.8. Summary of the estimates of the changeovers inationary impact in the rst ten months of 2002 (percentage points) Estimates Estimates without assuming a assuming a proportion of proportion of attractive attractive prices equal to prices equal to that that observed before the observed before the changeover (1) changeover Method 1 (A) Estimates for the items covered by the empirical analysis of micro-data Estimates for specic items based on ad hoc assumptions Unprocessed food Regulated prices Total (2) Method 2 (B) Method 1 (C) Method 2 (D)
Items
0.16
0.75
0.20
1.00
0.10 0.05 0.31
0.10 0.05 0.90
0.10 0.05 0.35
0.10 0.05 1.15
Notes: (1) The proportion of attractive prices was around 30 percentage points lower in October 2002 than that calculated on the distribution of lira prices observed in September 2001. In the calculation it is assumed that the impact due to rounding of euro prices that must still become attractive is the same as that estimated for those that already have become so. (2) Cumulated impact between January through October2002.
2.4 Estimates based on anomalous price movements

This section presents the results of two exercises based on an examination of the consumer price indices published each month by ISTAT. The analysis is similar to those conducted for the euro area as a whole by Eurostat (2002), for Portugal by Santo et al. (2002) and for Spain by lvarez Gonzlez et al. (2004). The basic idea is to identify anomalous price movements observed in the course of 2002, on the hypothesis that they can be attributed to the introduction of the euro.
16
2.4.1 Comparison with the average movements of the previous years The rst test for anomalies in the behaviour of consumer prices in 2002 is based on the comparison of the price dynamics that year with the average for the years 1999-2001 for each of the approximately 200 categories of expenditure released by ISTAT. The contribution of each item to consumer price ination in 2002 is approximated by assigning the CPI basket weight to the difference between its increase in 2002 and its average increase in the 1999-2001 period. Table 2.9 shows a practically negligible overall difference for the items considered, suggesting that the price movements in 2002 were not very different from the averages recorded in the years preceding; however, excluding energy products whose prices fell sharply in 2002 mirroring the fall in oil prices, the difference comes to 0.6 percentage points.
Table 2.9. Difference between price dynamic in 2002 and its average in 1999-2001 (1) Contribution Percentage weights (percentage points) (2002=100) Non-regulated prices Unprocessed food Energy products Processed food Non-food and non-energy products Services Regulated prices Processed food Energy products Non-food and non-energy products Services Total Total net of energy products 0.45 0.16 -0.22 0.03 0.19 0.29 -0.32 0.03 -0.24 -0.12 0.05 0.13 0.59 82.0 7.2 3.1 9.9 32.6 29.2 18.0 1.8 3.1 3.3 9.8 100.0 93.8
Source: Based on ISTAT data. Notes: (1) For each of the 207 product items posted by ISTAT on its website we calculate the following: (i) the average percentage change in each year; (ii) the average change in the years 1999-2001; (iii) the difference between average ination in 2002 and the average change in the years 1999-2001 (point ii). Lastly, the difference obtained at point (iii) is weighted using the items respective weights. The Table shows the sum of the contributions estimated as just described for each component indicated.
17
2.4.2 A regression analysis A regression was performed for each of the roughly 200 items taken into the CPI calculated by ISTAT, using as regressors the lags of the dependent variable, a linear trend, four seasonal dummies and four changeover dummies (one for each quarter of 2002).12 The basic hypothesis is that the latter will capture any anomalous behaviour of the indices during 2002. The statistically signicant coefcients of these variables, obtained by regressions run individually for each time series, were summed, using the weights of the respective items in the ISTAT basket. The results, reported in Table 2.10, indicate that the contribution of anomalous changes in the behaviour of the CPI amounted to approximately 0.5 percentage points in 2002 as a whole. Half of this impact is in the rst quarter; the effect is concentrated in unprocessed food in the rst quarter and in services in all four quarters.
Table 2.10. Regression analysis of anomalous contributions to ination in 2002 (percentage points) Q1 General index Processed food Unprocessed food Energy products Non-food and non-energy products Services 0.25 0.00 0.13 0.00 0.03 0.09 Q2 0.05 0.00 0.01 0.00 0.00 0.04 Q3 0.10 0.00 0.00 0.00 0.01 0.09 Q4 0.07 0.00 0.00 0.00 0.01 0.06 Average 2002 0.46 0.01 0.13 0.00 0.04 0.27
The results for Spain and Portugal, derived by a similar method, are in line with those for Italy (see Chap. 7). The study on Portugal (Santos et al. 2002), covering only the rst quarter of 2002, nds that the introduction of the euro caused around 0.2 percentage points of added ination. That on Spain (lvarez Gonzlez et al. 2004) attributes about 0.4 percentage points of consumer price ination in the whole 2002 to the changeover. For the euro area as a whole, Eurostat (2002) estimates that the changeover added no more than 0.16 points to consumer price ination in the rst quarter of 2002.
12
This specication is similar to that of Santos et al. (2002).
18
2.5 Conclusion
Before the introduction of euro notes and coins, fears of the potentially inationary impact centred on the possibility that the propensity of economic agents to set prices at attractive threshold gures would result in prevalently upward rounding. In the rst part of the chapter we estimated this impact in Italy for the period January-October 2002 on the basis of a set of elementary price quotes observed by ISTAT representing more than 60 per cent of the CPI basket. We found that between 0.2 and 0.5 percentage points of the average consumer price ination rate of 2.5 per cent in 2002 can be attributed more or less directly to rounding to attractive prices. The effect was considerably greater for services than for goods; for goods, it was larger for traditional sales channels, and very limited for mass retailers. The second part of the chapter seeks to identify any anomalous movement in consumer prices in 2002, whether associated with rounding or not. The results indicate an impact of between 0.1 and 0.6 percentage points of additional ination on average for the year, comparable in magnitude to that found in the exercises on the rounding effects and in similar studies of other euro-area countries (see Chap. 7).
19
References
lvarez Gonzlez, L.J., Cuadrado Salinas, P., Jareo Morago, J., Snchez Garca, I. (2004) El impacto de la puesta en circulatin del euro sobre los precios de consumo, Banco de Espaa, Servicio de Estudios, Documentos ocasionales 0404. Banca dItalia (2001) Lintroduzione delle banconote e delle monete in euro, Bollettino Economico (Italian unabridged version), October, 90-91. European Central Bank (2002) Recent developments in perceived and actual ination, Monthly Bulletin, October, 21-23. Eurostat (2002), Euro changeover effects, Euro-indicators news release, 58. Mostacci, F., Sabbatini, R. (2001) Una stima ex ante dellimpatto del changeover sui prezzi al consumo in Italia, Contributi Istat, 11. Mostacci, F., Sabbatini, R. (2003) Leuro ha creato inazione? Changeover e arrotondamenti dei prezzi al consumo in Italia nel 2002, Moneta e Credito, 221, 45-95. Santos, D., Evangelista, R. Nascimento T., Coimbra, C. (2002) Analysis of the impact of the conversion of escudos into euros, Banco de Portugal, Economic Bulletin, September, 1-14.
3. Pricing behaviour and the introduction of the euro: evidence from a panel of restaurants
Eugenio Gaiotti* and Francesco Lippi** * Bank of Italy, Department of Economic Outlook and Monetary Policy Studies ** University of Sassari and CEPR
3.1 Introduction 1
In the heated debate on the alleged inationary effects of the introduction of the euro, the Italian media and consumers associations frequently reported about extraordinary price increases. Among those, the ones recorded by restaurants gave rise to great controversies. According to a survey conducted in an Italian region, one third of citizens blamed restaurants for excessive increases, a percentage second only to the share of respondents who blamed food prices.2 Various newspaper articles reported anecdotal evidence allegedly showing that the cost a meal in a restaurant or a pizzeria had increased by 80-100 per cent after the introduction of the euro.3 It became almost common wisdom to argue that price setters took advantage of the psychological conversion of 1,000 lire into 1 euro, which would basically amount to a doubling of the price:4 everybody knows that one euro is now worth 1,000, not 2,000, lire,5 an opinion sometimes also put forward by members of the administration at the time. After the introduction of the euro, in Italy and in most euro area countries, restaurant prices were indeed among the fastest growing items in the
1
2 3 4 5
The authors thank Enrico Giovannini, Michele Polo, Massimo Roccas, Stefano Siviero, Giovanni Veronese and Guglielmo Weber for useful comments on a previous draft of this paper. They acknowledge Giornale degli Economisti e Annali di Economia for the authorisation to reprint this article, with small changes, from Volume 63, No. 3/4 (December 2004; copyright Universit Commerciale Luigi Bocconi). Francesco Lippi worked at the Bank of Italy Economic Research Department at the time of the study. EU.R.E.S., Rapporto 2002 sullo stato delle province del Lazio. See, for instance, La Repubblica, 23 and 24 August 2003. The ofcial conversion rate was 1936.27 lire per 1 euro. La Repubblica, 21 August 2003.
E. Gaiotti and F. Lippi
Harmonized Index of Consumer Prices (HICP); however, the size of the increase is in contrast with the perceptions voiced in the public discussions. Ination in the restaurant and cafes sector showed a blip in January 2002 (a month-on-month increase of 0.8 per cent) and reached a 12-month growth rate of 5 per cent at the end of the same year, compared to previous gures ranging between 2 and 3 per cent (see Fig. 3.1).
6.0 5.0 4.0 3.0 2.0 1.0 0.0 1.2
12-months (left axis) 1 month (right axis)
1.0 0.8 0.6 0.4 0.2 0.0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Fig. 3.1. Italy: HICP, category restaurants and cafes (percentage changes) Source: Eurostat.
Such facts raise two distinct questions. First, it is of interest to understand what happened at a more disaggregated level, which may shed light on why consumers perceptions about ination diverged so markedly from the ofcial measures. Other studies have conjectured that consumers ination expectations may have been particularly affected by rise in the price of frequently purchased goods, and/or by the detection of a greater-than-usual fraction of rising prices, whose impact on the general price level would however be limited.6 The second question concerns the identication of the economic mechanism which may have favoured increases of some prices in the face of an event, such as the change in the unit of account, which was widely expected to be neutral. Two candidate explanations have been advanced so far. The rst one emphasizes the presence of collusion and/or lack of competition. This is the prevailing interpretation in the popular debate and in the press; for restaurant prices, a more formal version of this view, based on collusive behaviour and multiple equilibria is advanced by Adriani et al. (2003). The second set of explanations stresses the existence of sticky prices and small menu costs: the
6
See Chap. 1 for an analysis of the Italian case.
Pricing behaviour and the introduction of the euro: evidence from a panel of restaurants
changeover is viewed as inducing more rms than usual to review their prices simultaneously, which may have short-run inationary consequences. Hobjin et al. (2004) propose a model along these lines and use it to study the dynamics of restaurant prices after the changeover, based on HICP gures. Our strategy is to address these issues by making extensive recourse to micro-data. We begin by assembling an original panel of data from 2,500 restaurants, pizzerie and the like in Italy over 1998-2004 using data from a leading guide to Italian restaurants. The sample includes well-known establishments and is not designed to be representative of the whole sector. Our interest stems from the possibility of gaining deeper insights from the individual data. In particular, the individual data allow us to study whether there are features of the distribution of individual price changes in the year 2002 (e.g. the frequency of price revisions or the number of exceptionally large increases) which may contribute to explain the widespread perception of a large effect of the introduction of the euro on prices.7 The descriptive evidence drawn from these data opens the way to the analysis of the two above-mentioned mechanisms capable of explaining a changeoverrelated increase in prices. First, we study whether the observed price setting behaviour provides any support to the menu cost hypothesis, which suggests that the year 2002 is a special one as the change in the unit of account invites all agents to revise their prices. We analyze the distribution of individual price increases, asking whether the average price increase and the frequency of price revisions in the year of the cash changeover were larger than normal. Next, we analyze whether the price increases recorded during the changeover relate to the local degree of competition, on the basis of a pricing equation that controls for several determinants of restaurant prices. We argue that if the changeover made prices less transparent to consumers, producers might have exploited such confusion to raise prices, at least temporarily. Such an incentive is greater the less elastic the demand faced by the consumer, as this ensures the producer a smaller fall in sales in case the consumer realizes the true value of the new price. Both exercises provide a rst quantication of the magnitude of, respectively, the menu-cost and the market power channel. The analysis
7
Note that disaggregated individual data are essential to analyze these questions. A related study based on disaggregated data is offered by Adriani et al. (2003), who use data from six European countries for the years 2002 and 2003. These authors nd a greater than average increase in the prices of the euro-area countries, interpreting such effect as a consequence of the changeover. Their analysis, however, only measures the increase in prices of a single year, for a limited number of restaurants (155 for Italy). We believe that our analysis goes a step further by signicantly extending the number of restaurants (2500) and years (seven) included in the sample, which allows us to measure the effect of the changeover in comparison to a historical benchmark.
is completed by the presentation of a formal model in which both effects appear simultaneously. The model is calibrated on historical data and then used to simulate the effect of the changeover on prices. This chapter is organized as follows. The next section describes the database: an open panel of about 2500 establishments (including restaurants, pizzerie and trattorie) for the years 1998-2004, a total of about 17,000 observations which were purposely assembled for this study. The main facts affecting demand and costs in this industry over the period and the key descriptive statistics on the distribution of individual price changes are presented in section 3.3. Section 3.4 discusses the frequency and size of price revisions, a decomposition which is helpful in identifying the source of the price increase and the relevance of the menu cost assumption. An econometric analysis of the role of market power in the setting of prices is in section 3.5. Section 3.6 formally analyzes the role of the menu cost and the market power hypotheses by means of a simple price setting model, attempting to quantify the relative importance of these two channels. A concluding section summarises the main ndings.
3.2 A description of the database

Restaurant prices are collected from the yearly publications Guida dei ristoranti dItalia and Roma; both are leading guides to Italian restaurants, wine-bars, trattorie and pizzerie.8 The data report the price of a full meal.9 The price records are expressed in lire until 2001 (rounded to the nearest 5,000 lire, approximately 2.6 euro), in euro since 2002 (rounded to the nearest euro). For the sole restaurant category, the guide also assigns a score, on a 0-100 scale, concerning its quality; it includes information about each establishment, such as location or category. Data are annual, collected each year during the rst semester; the guide is published towards the end of September. The data span the 1998-2004 period. The panel is unbalanced, since the list of surveyed establishments is revised each year. On average, 85 per cent of the establishments surveyed in a given year also appear in the guide the following year. This subset allows us to compute price increases. We converted the lira prices in euro, rounding them to the nearest euro. In order to avoid the results to be affected by the
The guides are edited by the company Il Gambero Rosso. The second guide is specically about the Rome area and contains more detailed information, such as the location of the various restaurants in the city (e.g. Vatican, city centre versus suburbs, etc). This includes appetizer, rst and second course and a dessert (beverages are not included).
rounding conventions used by the guide in each year, in computing price increases we set to zero all price changes smaller than 1 euro in absolute value.10 Tables 3.A1-3.A5 in appendix 3.A provide a detailed description of the sample composition. The number of establishments surveyed in each year hovers between 1,800 and 2,500 observations, for a total of 17,000 observations (Table 3.A1). Due to sample turnover, the number of price changes computed in each year is smaller, between 1,600 and 2,300 per year, for a total of 12,300 observations. The data cover the entire national territory. The distribution by geographical area (Table 3.A2) reveals a slight under-representation of the establishments located in the South (23 per cent of the total, which compares with a 32 per cent computed from the registry of the Chamber of Commerce).11 As to the type of establishments, the majority is represented by restaurants (about 70 per cent; see Table 3.A3); the remaining part consists of trattorie, pizzerie and wine bars. Overall, the surveyed establishments cover approximately 3.5 per cent of the total number of businesses recorded by the Chamber of Commerce (Table 3.A4). Assuming that the surveyed establishments are scarcely substitutable with lower-quality and less visible dining places, we constructed an indicator of the degree of market competition at the local (province) level. A rst measure is given by the ratio between the number of quality dining places that are present in a given province, proxied by the number of establishments reported in the guide, and resident population. This indicator, whose averages are reported in the rst column of Table 3.A5 for the Italian Regioni,12 suggests a greater degree of competition in the North-east regions, a smaller one in the South. By overlooking the presence of tourists, however, the indicator might overestimate competition in the provinces where tourism is highly relevant. To this end, a second measure is constructed as the ratio between the number of quality dining places and a weighted sum of local population and foreign and domestic tourist
10
11
12
This adjustment operates only on 2002, 2003 and 2004 data and it has negligible effects (a few basis points) on most results reported in the paper. The only exception is the percentage of unchanged prices in 2002, 2003 and 2004 which is discussed at greater length in section 3.5. The regional distribution of establishments, for the year 1999, is drawn from Rapporto sul turismo italiano 2001, p 47, Table 7. In Italy there are 20 regions and there were 103 provinces at the time of the study.
presence.13 The indicator, reported in the second column of Table 3.A5, broadly conrms the previous picture, with the notable exception of a few small regions which attract considerable tourism ows. Our sample was obviously not designed with the aim of being representative of the whole restoration industry and should not be considered as such. As discussed in the next section, the rise in the price index in the whole sector is smaller than in our high-quality establishments. One possible explanation for the divergence is due to the fact that these belong to a segment facing a less elastic demand (the implications of this hypothesis is discussed in detail in Sects. 3.5 and 3.6). More generally, the demand for quality food increased signicantly in Italy in the last few years, as witnessed by the growing diffusion of publications and guides on the subject, possibly commanding an increasing premium on quality restaurants. Table 3.A6 reports several descriptive statistics on the price of a meal from 1998 to 2004, sorted by year and establishment type. The average price of a meal in a restaurant varies from 33 euro in 1998 to 49 euro in 2004; in wine bars, pizzerie and trattorie it increases from 20 to 27 euro. The price dispersion (standard deviation) increases in time.
3.3 Demand and costs conditions and the dynamics of meal prices over 1999-2004
In the period under examination, both demand and supply conditions in the industry were subject to shocks that should be considered when examining the data. Between the second half of 2000 and the rst half of 2001, a strong rise in the demand for meal services took place, related to the spike of tourism sparked by the Jubilaeum of the Catholic church in the year 2000. Expenses by foreign tourists (Fig. 3.2) then decreased sharply following the terrorist attacks of September 11, 2001.14 The growth of households
13
14
Statistics on tourist presence in hotels and rented apartments in each province are from Chambers of commerce data; we used data referring to the year 1998. Population and tourist presence are not directly comparable: the former is measured in heads, the latter in days of presence; also, tourists are likely to go to the restaurants more often than local resident. Assuming local population eat out once a month (an assumption broadly consistent with indications from the Istat survey on household consumption), we multiplied population by 12 to make the two magnitudes comparable. See a discussion in Rapporto sul Turismo Italiano 2002, p 3.
expenditure (at constant prices) in the restaurant and hotel services sector also reached a peak in 2000; employment and value added in this industry grew at around 8 per cent.
14 12 10 8 6 4 2 0 -2 -4 1998 1999 2000 2001 2002 2003 14
household expenditure in h&r employment, h&r value added, h&r expenditure by tourists
12 10 8 6 4 2 0 -2 -4
Fig. 3.2. Activity and demand in the sector hotels and restaurants (annual percentage changes) Source: ISTAT, UIC.
On the costs side, substantial increases in the price of unprocessed food took place in 2001-03, due to exceptional events, as the spreading of rst BSE (mad cow) and then foot-and-mouth diseases in 2000-01, or adverse climatic conditions at the beginning of 2002. According to Eurostat data (Fig. 3.3), the agricultural production prices increased rapidly in 2001, at a rate above 5 per cent; they slowed down, but still kept increasing, in the following years. Fresh food consumer prices followed a similar, although more persistent, pattern; in this case, growth was above 4 per cent until 2003. Unit labour costs in the hotels and restaurants sector increased substantially in 2001 and, notably, 2002 and 2003. All in all, the annual growth in costs can be roughly estimated to have been between 4 and 6 per cent in 2001-03. The main statistics on price increases in our sample are reported in Table 3.1. Two facts emerge from the data: in the year of the euro cash changeover the average increase is sizable (9.3 per cent), but it does not represent a peak (which is located in the year 2001, at 10.5 per cent). This holds irrespective of the type of establishments under investigation (Table 3.A7 in appendix A). A sample split by geographical areas (North, Centre and South) shows that the year of the changeover records a peak increase in prices only in the South (Table 3.A8).
8
8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0

8.0 6.0 4.0 2.0 0.0
CPI, fresh food agricultural production prices ULC, h&r sector

1998 1999 2000 2001 2002 2003
-2.0 -4.0 -6.0
Fig. 3.3. Fresh food prices and ULC in the sector hotels and restaurants (annual percentage changes) Source: Eurostat, ISTAT.
A similar picture emerges from other descriptive statistics, such as the median, or an indicator of the largest increases (the 95th percentile or the maximum increase). In all these cases, the increases recorded in 2002 and 2003 are smaller than in the previous year. The year of the changeover is characterized by price increases that, although sizable, are not out of line with those recorded in previous years.
Table 3.1. The rise in the price of a meal (12-month change) Year Sample average 5th percentile Median 95th percentile Standard deviation 1999 4.3 0.0 0.0 25.0 10.4 2000 6.7 0.0 0.0 30.0 12.3 2001 10.5 0.0 10.0 33.3 12.5 2002 9.3 0.0 8.7 29.0 11.3 2003 5.8 0.0 4.9 20.0 8.6 2004 3.8 -0.1 0.0 20.0 9.2
A comparison of the average annual price increase in our dataset with those from other studies is shown in Table 3.A9. Our data are basically identical to those obtained from a similar sample of quality restaurants by Adriani et al. (2003), although their sole observation concerns 2003. In contrast, notable differences emerge between our data and the increase of the average
price level for the whole sector produced by the National statistical institute. Although the time prole of the changes is similar, the rise in the price index in the whole sector is much smaller than in our high-quality establishments. This holds for all years and it is not limited to 2002. According to the facts presented above, price increases in 2000 and 2001 may be related to increases in demand and costs, which are likely to have affected prices with a lag. Such a lag may just reect the way our data are constructed: the prices reported in the Guide are surveyed in the rst half of the year. The trends recorded in the city of Rome, most directly connected to the Jubilaeum, conrm the conjecture that tourism played a role in the evolution of prices. In Rome (Table 3.2) the largest increase occurred in the year 2000 (12.7 per cent); in that year, the average price growth in the establishments located in the Vatican area and the close-by tourist district of Trastevere are, respectively, of 23 and 29 per cent, much greater than in the rest of the city.
Table 3.2. The rise in the price of a meal - selected Rome districts (12-month change) 1999 Average City Centre Vatican Trastevere 6.1 5.4 10.6 6.4 2000 12.7 12.0 23.4 29.3 2001 10.7 12.9 8.9 11.6 2002 8.6 7.1 3.0 1.3 2003 3.8 4.8 0.0 4.8 2004 3.8 3.8 5.1 2.1
Considering the tails of the distribution, there is simply no evidence of any doubling of prices in 2002. Table 3.3 shows the number of price increases larger than a given threshold (respectively, 50, 75 or 90 per cent) in each year. As in the previous case, the years 2002 and 2003 stand out for the smaller number (virtually nil) of large increases. A few increases greater than 50 per cent can be found in 2000-02, but not later. The conjecture that high perceived ination was caused by large increases which hit the imagination of the consumers nds little support in the data (such increases represent less than 2 per cent of the total sample observations). A major factor affecting the public perception, however, may have been the substantial cumulate price increase which took place over the whole 19982002: around 40 per cent on average, or even 75 per cent for the 10 per cent of establishments which recorded the largest price rises. These observations help to form a conjecture on the factors that may have favoured the perception of an exceptional rise in prices among consumers; such an increase, while it actually took place, occurred more gradually during the years than perceived.
10
Table 3.3. Number of extreme price increases Year Greater than 50 per cent Greater than 75 per cent Greater than 90 per cent Total of observations 1999 11 4 2 2,239 2000 20 4 1 2,292 2001 30 4 2 2,144 2002 17 4 1 2,023 2003 5 0 0 2,028 2004 6 0 0 1,552
3.4 The frequency of price revisions

The histograms in Fig. 3.4 show the distribution of price changes in each year. The distributions are strongly asymmetric, with few negative records. The mode is at zero. This indicates that in most years a large mass of rms leave their price unchanged, or change it by a small amount. In the years 1999, 2000 and 2004 this mass amounts to about 60 per cent of the whole population (Table 3.4); on average, the percentage is around 40 per cent. These data correspond to an average duration of prices of respectively 23 and 13 months.15 In both 2001 and 2002 the percentage of unchanged prices is much smaller, reducing to below 30 per cent.16
15
16
Under the assumption that the decision whether to revise prices is taken with a monthly frequency, duration is computed as [1-(1-a)1/12]-1, where a is the annual share of revised prices; a similar result is obtained by assuming continuous decisions and applying the formula reported by Bils and Klenow (2002), [-1/ log(1-a)]*12. According to Bils and Klenow, the average price duration for items lunch or dinner is 11 months. As mentioned, in constructing the sample we set to zero all price variations smaller than one euro in absolute value. This was done for two reasons. First, the gures reported in Table 3.4 would otherwise not be comparable through time: before 2002, any price change smaller than 1 euro in absolute value would not have been recorded by the Guide due to the rounding convention described in section 3.2. Second, in 2002 many small variations in prices (equal to 1 euro) would be observed, even if prices were unchanged. This is due to the new rounding convention adopted by the Guide since 2002. For example, in 2001 a price of 17,500 lire would have been rounded to 20,000 lire, corresponding to 10 euros; in 2002 the same price would be rounded to 9 euros. Had we not made these adjustments, the gures in the fourth row of Table 3.4 would have been equal to 5, 30 and 50 per cent in 2002, 2003 and 2004, respectively (there would have been no change in 1999-2001).
11
.7
.7
Fraction
0 -.500161
Fraction 0 Percent change 1999 1.2 -.5 Percentage change 2000 1.20062 .7 .7 0 -.5 Percentage change 2001 1.2 Fraction 0
Fraction
-.5
Percentage change 2002
1.2
.7
.7
Fraction
Fraction
-.5
Percentage change
1.2
-.5
2003
Percentage change
1.2
2004
Fig. 3.4. Frequency distribution of price increases (1). Notes: (1) Frequency distribution of price increases in each year. Range of percentage increases from -0.5 (-50 per cent) to 1.2 (+120 per cent) (horizontal axis). The tick marks 0. Each bar includes a range of price increases from x to x+3.5 percentage points.
12
Table 3.4. Number of constant and falling prices Year 1999 2000 1,309 72 2,292 0.57 2001 600 74 2,144 0.28 2002 542 90 2,023 0.27 2003 861 45 2,028 0.42 2004 874 109 1,552 0.56
(a) number of unchanged prices 1,432 (b) number of price reductions (c) number of observations (a)/(c) 108 2,239 0.64
The data are consistent with the menu cost hypothesis: small price revisions are prevented, in normal years, by an adjustment cost.17 From this standpoint, the changeover, and the associated redenomination of all prices in the new unit of account, can be though of as an aggregate shock which requires each agent to repost its price, creating an opportunity to revise it, consequently determining an increase in average ination.18 The size of the average (as well as median) price revision (Table 3.5) reinforces the impression that pricing polices in years 2002 and 2003, rather than representing a discontinuity, were quite moderate in relative terms. Average increases were among the smallest since 1999 (13 and 10 per cent, respectively, compared to 15 and 16 per cent in the previous 2 years).
Table 3.5. Distribution of price revisions, unchanged prices excluded (12-month change) Year Sample average 5th percentile Median 95 percentile
th
1999 12.0 -12.5 11.1 33.3 14.5
2000 15.7 -8.3 13.3 42.8 14.5
2001 14.6 3.7 12.5 36.3 12.6
2002 12.7 -3.8 11.1 32.2 11.5
2003 10.1 3.1 8.3 25.0 9.2
2004 8.7 -12.4 8.3 27.3 12.4
Standard deviation
17
18
Fabiani, Gattulli and Sabbatini (2007) study a sample of Italian rms and nd that in the services sector the main reason given by the rms for not adjusting their price is fear of competitors not following. In this sector, according to their study, about 30 per cent of rms kept prices unchanged in 2002. The fact that, according to mainstream menu costs models, at the moment of the changeover the revision of most prices should produce a blip in ination is discussed in section 3.6. Also see Hobjin et al. (2004).
13
Altogether, it may be conjectured that the dynamics of prices reects several shocks: an increase in demand and costs affecting both 2000 and 2001; and a mandatory rewriting of menus in 2002, coinciding with the cash changeover, which prompted most agents to post new prices. To compare the features of the pricing policies during these two episodes, we decomposed the average price change as follows: we rst computed the differential between the average price change in each year (i) and in a base period (j), and then decomposed it into two contributions: the change in the share of prices which were reset () and the change in the average increase of those prices which were reset ():19
( )
+ ( )
A model of infrequent state-dependent price revisions has implications for the decomposition (3.1). Assuming that the main effect of the introduction of the euro is the simultaneous revision of all menus, the rst component in (3.1) should explain most of the change in ination; in contrast, in case of a shock to demand, costs or prot margins both components should contribute to the increase in ination.20 The results are reported in Table 3.6. The year 1999 is used as the benchmark period j.
Table 3.6. Decomposition of price increases Year Price increase: differential with respect to 1999 due to larger share of price revisions due to larger average increase of revised prices 2000 2001 2002 2003 2004
2.4 1.0 1.4
6.2 4.8 1.4
5.0 4.6 0.4
1.5 2.4 -0.9
19
20
The rst term is weighted with the average increase in prices, the second with the average share . Eq. (3.1) is obtained by manipulating the identity: i=i i,, where i is the average price increase, i is the share of price revisions in year i and i is the average increase of those prices that are revised. See Sect. 3.6 below.
+ 2
+ 2
(3.1)
-0.5 0.8 -1.3
14
In 2001, the acceleration of prices with respect to the benchmark year (more than 6 percentage points) is due to both factors: almost ve percentage points are due to a larger number of rms that revised their prices; about 1.5 percentage points are due to larger increases made by those restaurants which changed their price. This evidence is consistent with a demand and costs shock, which may have prompted more agents than usual to revise their price, while also increasing them substantially (e.g. because marginal costs increased more rapidly). In 2002, the share of unchanged prices is also particularly low. However, the considerable number of price revisions determines 4.6 percentage points of higher ination, the bulk of the overall increase. Unlike 2001, the effect of the size of the individual revisions on ination is negligible (0.4 per cent). This evidence is consistent with the conjecture that in 2002 a price shock was due to a mandatory price revision, due to the cash changeover. Many prices changed, but the amount of individual increases was not particularly large. This followed two years of already sustained price growth, related to exogenous demand and cost factors. Perceptions of a large price increase may have been the cumulated effect of all these events
3.5 Market structure and the effect of the changeover

The previous section suggests that the revision of the menus was the main factor explaining the impact of the euro on prices. Still, market structure could also have contributed to the decision about how much to revise prices in the face of the changeover. A possible explanation of such an effect is articulated in section 3.6 below: in industries characterized by low competition, producers may, at least temporarily, take advantage of the consumers confusion to raise their price. Such an incentive is greater the less elastic the demand faced by the consumer, as this ensures the producer a smaller fall in sales in case the consumer realizes the true value of the new price. An implication of this conjecture is that only in year 2002 - and possibly in 2003, since increases taking place in the second half of 2002 are recorded in the following years data - the magnitude of price increases should be negatively correlated to the degree of local market competition, proxied by the indicator discussed in section 3.2 (and shown in Table 3.A5). Some descriptive evidence, consistent with this conjecture, is presented in Table 3.7. Firms operating in more competitive local markets recorded a smaller increase in prices in 2002, and vice-versa. The Table compares the average price increase with those of the upper and lower tails of the distribution, ordered according to the competition indicator; the second measure discussed in section 3.2 is employed. The difference is notable:
15
about two and a half percentage points between the top and the bottom fourth of rms.
Table 3.7. Distribution of price increases in 2002 by market competition Market Market competition competition < 1st quartile < median Mean 5 percentile
th
Whole sample
Market Market competition competition > median > 3rd quartile 8.8 -3.2 7.7 28.6 7.3 -13.0 7.4 28.2
9.7 0.0 8.7 29.0
9.8 0.0 8.7 29.6
9.3 0.0 8.7 29.0
Median 95 percentile
th
To address the issue more formally, we estimate a model of the determination of price changes over the period 1999-2003. Since price increases are truncated at zero (with a negligible number of exceptions), we model the price change as a case of censored data. To control for selectivity bias, we implement the estimates with the maximum likelihood equivalent of Heckmans two step selection model.21 The main results are quite robust to the use of alternative estimation methods, as Tobit, OLS, random effects or xed effect estimators. We estimate the equation:
(3.2)
where pi is the individual price of rm i and Xi is the vector of explanatory variables.22 Among the variables in Xi (affecting both the likelihood to change the price and the size of the price increase) we include the lagged log-price level of the rm (pt-1); a dummy taking value one for restaurants, zero otherwise (Drest); and, for those establishments classied as restaurants, their lagged quality judgement (i.e. the score),23 and the (lagged) age of the price (age), expressed as the number of years since the last price revision. We assume that the actual price revision is an increasing function of the distance of the lagged price from its equilibrium level, a function of the score variable. The
21 22
23
See Johnston and DiNardo (1997), Chap.13. We assume that the same explanatory variables enter in the selection equation underlying Heckman estimates. Since the score variable is only available for restaurants, the variable is interacted with dummy Drest in the regression.
16
(lagged) age of the price captures a similar effect: under positive ination the time elapsed since the last change increases the probability that the price is revised again, and increases the expected size of the revision. As a proxy for cyclical demand pressure, we include the annual inows of foreign tourists in each province, normalised by the population in the province (tourism). As for domestic demand, we include the (lagged) per capita value added in each province (vadd). Lacking other measures of real expenditure by households which have sufcient cross-sectional variability, any remaining demand effect is captured by the time dummies included in the regression. The same applies to cost variables (like fresh food prices) who are not available at the disaggregate level. Among the explanatory variables we include our measure of local market competition (comp) dened in Section 3.2, based on the ratio of the number of restaurants in the guide to a weighted sum of local population and tourist presence, interacted with the time dummies (T1999 to T2003). Our conjecture is that the degree of competition affects price dynamics only after the changeover; hence, the variable is entered separately for each year. The estimation results are reported in the rst column of Table 3.8. Most variables are strongly signicant and have the expected sign. In 2002 and in 2003, the degree of local market competition signicantly enters the equation for price determination. In contrast, it does not signicantly enter the model in the previous years.
Table 3.8. Determinants of price changes (1) Coef. p(t-1) Score*Drest Drest Age Tourism v.add. comp*T1999 comp*T2000 comp*T2001 comp*T2002 comp*T2003 T2000 T2001 T2002 T2003 Input cost Observations Censored -0.13 0.44 -1.81 0.01 0.00 0.04 -0.30 -0.06 -0.29 -0.76 -0.91 0.04 0.09 0.08 0.04 10,525 4,689 P>|z| 0.00 0.00 0.00 0.00 0.05 0.00 0.14 0.77 0.08 0.00 0.00 0.00 0.00 0.00 0.00 () () () () () () Coef. -0.13 0.44 -1.81 0.01 0.00 0.04 -0.30 -0.06 -0.29 -0.76 -0.90 0.04 0.06 0.04 0.01 1.00 10,525 4,689 P>|z| 0.00 0.00 0.00 0.00 0.05 0.00 0.14 0.78 0.08 0.00 0.00 0.00 0.00 0.00 0.15 restr. () () () () () ()
() () () () () ()
() () () () ()
Notes: (1) Heckman selection model. Intercept (not shown) included. () indicates 1 per cent rejection condence level.
17
Market structure thus seems to have contributed to the impact of the changeover, therefore giving a possible rationale for its non-neutrality. The size of the effect is relevant: differences in market concentration can be shown to explain a few points differences in the price increases across provinces. Considering that in the equation for pt the estimated coefcient on comp (when interacted with either T2002 or T2003) is around -0.8, a decrease in comp (i.e. less competition in the local market) equal to its standard deviation (which is 0.03) would determine a price increase of about 2.5 percentage points. The coefcients of the time dummies T2000-T2003 provide us with an estimate of the time effects. In each year there is an unexplained increase relative to the 1999 benchmark, which captures the effect of price determinants not included in our regression, e.g. input costs. An attempt to control for the latter is presented in the second column of Table 3.8, where a proxy for the growth rate of input costs is added to the price equation (with a coefcient restricted to one). Once the model dynamics are accounted for,24 the unexplained growth in prices in the year 2000 and 2001 is, respectively, 4 and 6 percentage points. In the year of the changeover unexplained ination is still positive but smaller, around 3 percentage points; it is slightly negative in 2003.
3.6 The euro, competition and sticky prices: a simple model

This section proposes an analytical framework to provide a rigorous formulation of the menu cost and the market power hypotheses and to quantify the effect implied by each hypothesis. We begin by showing how the cash changeover may cause a larger price increase where the market is less competitive. Next, we show how this effect may interact with the presence of menu costs and the consequent clustering of adjustments at the moment of introduction of the euro. We use the model presented above to t some characteristics of the micro data before the changeover and to study the implications of the changeover under three alternative set of assumptions, namely that the introduction of the euro determined i) a simultaneous revision of all prices; ii) a temporary increase in market power; iii) both. A key assumption is that the introduction of a new currency increases the difculty faced by the consumer in evaluating the relative price of the goods he or she is purchasing. Such a lack of price visibility, similar to what happens when engaging in transactions in a foreign currency, has the effect of enabling producers to increase their selling prices the more, the greater is their market power.
24
Dening the time-effect on prices in each period j as Ej, the model dynamics imply Ej=Tj + [(1+1)Ej-1 - T j-1] , where Tj is the coefcient of the dummy Tj pertaining to the same period and 1 is the coefcient on the lagged price level p(t-1).
18
We consider a rm i producing a good with a constant unit cost of production, dened as Wi, under monopolistic competition, facing a demand curve Di = D(Pi,Pk), where Pi is rm is price and Pk is the price set by the rms competitors (which rm i takes as given); price elasticity is constant, >1. We omit the rm sufx in what follows. The rst order condition from prot maximization (mP ax DPWD) yields the price: P = W (1+1/(1)) =P. We dene the quantity demanded at this equilibrium price; we assume that before the cash changeover the market clears, with the rm producing and sellingD at priceP. The consumer is confused by the cash changeover. This means that the rm can increase the price without the consumer noticing. More precisely, we assume that the consumer is presented with a new price P, not necessarily corresponding to the previous one. We assume that the probability that the consumer notices that the price has been changed is . If the consumer realizes that the price has changed, he just follows its demand curve. If he does not realize it, he keeps demanding the previous quantityD. In this framework, the problem of the rm is:
max (
The rst order conditions are:
) + (1 )(
+ ( )
(3.3) (3.4)
+ (1 )
=0,
which yield the new equilibrium price, where the new (lower) demand elasticity is h; h is a positive term not larger than 1, increasing in .25 The percent price change with respect to the old level is:
1 . 1
(3.5)
Equation (3.5) shows that after the changeover the price is unchanged if =1 (trivially, in this case there is no incentive to move the price since the consumer is on the original demand curve with probability 1)26 or if (perfect competition). The ination induced by the euro changeover is larger, the larger is the rm's market power (small ) and, obviously, the larger the probability that the consumer does not perceive the price change (small ).
25
It is
= + (1 )
. Using a linear approximation of the demand function,
it may be shown that in equilibrium it is
( 1) (1 ) . From second ( 1) (1 )
order conditions one gets (1) (1) >0. Hence 0<h<1. The condition h>1/ is necessary to have an internal solution.
19
The interpretation of the result is intuitive. The consumer's confusion caused by the cash changeover, captured by the variable , decreases the demand elasticity expected by the rm; the decrease is proportional to the original demand elasticity, . In the problem (3.3) faced by the rm, the marginal benet of a higher price is the increase in the unit value of the old quantity sold,D; the marginal disadvantage depends on the probability that the consumer realizes that the price has changed and on the consequent decrease in sales, which is function of the elasticity (in this case the consumer follows the demand curve D(P)) . To capture the high observed degree of price stickiness, the model needs to be cast in a dynamic setting that accounts for costly price adjustment. We follow Dotsey, King and Wolman (1999) and assume that rms incur a random menu cost t if their price is revised. The decision that in period t the rm will not revise its price thus depends on the comparison between the adjustment cost and the increase in prots that would be obtained by changing the price. Marginal costs and the demand curve D(Pt) are assumed to be the same for all rms, so we omit the rm sufx in our notation, but introduce time sufxes. We dene t|t-j as the probability that in period t a price originally set in period t-j is left unchanged; as in Dotsey, King and Wolman (1999), the probability depends on the distribution in the menu cost: rms balance the gains from adjusting the price with the menu cost (see Appendix 3.B). Firms must take into account the effect of the current price on future periods prots, in case they will not revise the price again. The solution of the optimisation problem is derived in Appendix 3.B. From the rst order conditions, the optimally reset price in each period is equal to the present discounted value of future marginal costs divided by the present discounted value of the (inverse of) future mark-ups:
*=
=0
+ | +
1
+
] [
1 =0
+ |
+ |
+ +
(3.6)
where is the discount factor,
+|

=0
, is the (possibly time
varying) price derivative of the demand function (normalised, without loss of generality, as a ratio to a steady state value), t is the (possibly time varying)
26
Note that if =1, then h=1.
20
demand elasticity and Xt is the general (whole economy) price level.27 The sectors aggregate price level Pt is obtained as a weighted average of vintages of past prices P*t s, where the share of vintage h prices in period t, is a function of the t|t-js. As assumed above, in t0 with probability (1-) consumers keep demanding the benchmark quantity D (Pt0), wherePt0 is the price that would have prevailed without the changeover. Considering expected demand Dt0=D(Pt0)+(1)D (Pt0), t0 and t0 in (3.6) can be derived.28 We calibrate (3.6) assuming the annual discount factor is =0.96 and, based on the features of our sample in 1999, that the steady state marginal cost Wt grows at a constant annual rate of 4 per cent. Using a relatively standard assumption, demand elasticity is set at t|tt0=11 and t|tt0=1, which corresponds to an equilibrium markup of 10 per cent. The path of t|t-j is endogenously determined by positing a uniform distribution for adjustment costs, calibrated to reproduce the fact than on average about 60 per cent of rms do not revise their price, as from our panel data. The probability that the consumer does not notice that the price has changed is set at (1-)=0.3. This assumption matches the evidence of the European Commissions Eurobarometer; according to this survey, at end of 2003 36 per cent of Italian consumers reckoned, in retrospective, that their attitude was to buy more, because they did not realize how much they are spending after the introduction of the euro.29 In the rst scenario the changeover involves a simultaneous forced revision of most prices (the menu cost is nil for all rms). In the second we posit that the consumer is confused with probability 1- , which increases the rms degree of monopolistic power. In the third scenario both shocks occur. The results of the simulations are shown in the four panels of Figure 3.5 which include the rate of ination, the size of the average price revision, the share of unrevised prices and the price level. In each panel, the solid line shows the responses to a menu cost shock, the dashed line the responses to a market power shock and the dotted line the responses to both shocks occurring simultaneously. All shocks are assumed to take place in t0.
27
If t= , (3.6) can be shown to be consistent with the result of Dotsey, King and Wolman (1999). If t|k=, t=1, t= , the standard result of Calvo (1983) obtains. 1 It is t0=. In a neighbourhood of equilibrium, 0 = 1 + (1 )
28
29
This expression can be solved simultaneously with (3.6) to derive t0 and Pt0. European Commission Eurobarometer, The euro, two years later, January 2004.
Inflation
21
15
16
14
menu cost market power both
Average price revision

12
10
8
10
6
4
0 15
20
25
period
30
35
40
2 15
20
25
period
30
35
40
Share of unrevised prices

380
Price level
menu cost market power steady state
0.8
360
340
0.6
320
300
280
260
0.4
0.2
240
220 20
-0.2 15
20
25
period
30
35
40
22
24
period
26
28
30
32
Fig. 3.5. Model based effect of the changeover (1). Notes: (1) Based on the model in Sect. 3.6.
The menu-cost shock scenario causes an increase in ination, reaching 12.9 per cent in t0 (top-left panel). The increase mostly reects the share of unrevised prices dropping to zero (bottom-left panel), thereby sustaining average ination; in contrast, the average size of individual price revisions is only marginally larger than in steady state (top-right panel). Before t0, ination falls slightly: the agents incorporate less future expected marginal cost increases in their prices, knowing they will be soon resetting prices. Afterwards, the ination rate gradually returns to its steady state value (4 per cent) with small oscillations.30
30
As discussed by Hobjin et al. (2004), these oscillation are typical of models which include a maximum lag beyond which all prices are adjusted.
22
In the second scenario, the market power shock has an impact on periodt0 ination, which increases to 6.3 per cent.31 The increase in ination is the result both of a temporary increase in margins and of an (endogenously) larger number of price revisions: the share of prices which stay xed drops to 53 per cent, as more rms nd it convenient to sustain the adjustment cost to exploit the increase in market power. The third scenario involving both shocks yields an higher ination in t0, around 15 per cent. In all three cases, the deviation of the price level from its long-run trend is relatively short-lived (bottom-right panel). After temporarily rising above their steady state path, prices slow down and are back to baseline on the second year after the shock.
Table 3.9. Model-based decomposition of price increases (percentage points; in brackets percentages) Menu cost shock Price increase: differential with respect to steady state due to larger share of price revisions due to larger average increase of revised prices Market power shock 2.3 (100%) 1.7 (74%) 0.6 (26%) Both
8.9 (100%) 8.1 (92%) 0.8 (8%)
11.0 (100%) 8.8 (80%) 2.2 (20%)
The decomposition of the price increase, along the lines described in section 3.4, is quite different in the three cases (Table 3.9). After the menu cost shock, the increase in ination is almost entirely (92 per cent) due to the large number of price revisions. After the market power shock, the contribution to the increase in ination is more evenly split among the two components (74 and 26 per cent, respectively). When both shocks hit the economy, the decomposition of ination lies somewhere in between.32 How does the model compare to the facts described in the previous sections? The prole of ination and its decomposition are consistent with the idea that the menu cost shock may have determined the bulk of the impact of the introduction of the euro, with the market power shock
31
32
The size of this effect, however, is strongly dependent on the exact shape assumed for the menu cost distribution, and it would be greater under parametrisations which increase the degree of state-dependency of the adjustment. These percentages are rather robust to different assumptions and calibrations for the menu cost distribution.
23
possibly playing a complementary role. The overall size of the increase in ination observed in the sample is fully consistent with what would have been produced by a mere revision of all menus, given the relatively high degree of price stickiness on this market. It is possible, however, that an increase in market power reinforced the effect of the menu cost shock. According to the model the price level will gradually return to its steadystate path. Such a convergence takes the form of small price revisions and a relatively high share of sticky prices. This implication is qualitatively albeit maybe not quantitatively - matched by the features observed in our sample in 2004; in that year, the average price revision and the number of rms adjusting prices are below their medium-run levels.
3.7 Conclusions
This chapter analysed a panel of restaurant prices to study price setting behaviour at the rm level during the euro cash changeover. The unique database allows us to shed light on two controversial issues concerning the effects of the changeover: rst, to document what happened to prices at the micro level, in the year of the changeover and in the neighbouring ones. This evidence is a rst key step in understanding what caused the public perceptions to deviate so markedly from ofcial ination measures. Second, to begin a systematic analysis of the mechanisms that may have affected price setting during the changeover. On the rst question, our results show that a sizeable average price increase took place in 2002 (about 9 per cent). This increase is slightly smaller than the one recorded by the industry in the previous year (about 10 per cent). The evidence suggests that such increases are related to rising demand and costs conditions; this should induce caution in attributing a large inationary effect exclusively to the introduction of the euro banknotes. The evidence clearly dismisses the hypothesis that the euro cash changeover favoured a doubling of prices. Such a widespread perception might be ascribed to the substantial price increase which took place in this sector over a longer time period (between 1998 and 2003 the average price of a meal increases by 40 per cent; in the 10 per cent of restaurants recording the largest increases the rise is 75 per cent). The changeover might have focussed the public attention on the price level, prompting the attribution of the whole increase to the introduction of the euro. On the second question, we found that both the menu cost and the market power hypotheses nd support in the data. As to the former, the evidence shows that much of the aggregate price increase during the changeover results from a greater number of prices being revised, rather than from large individual price increases. This appears consistent with the idea that the new
24
currency denomination made it mandatory for almost all rms to revise their prices in 2002. In normal years, a great proportion of restaurants do not adjust their price, probably due to the presence of small menu costs. We also found that market structure affected price dynamics after the changeover. In 2002 and 2003 price increases were larger in the provinces which featured a smaller degree of competitiveness. Both the menu cost and the market power hypothesis imply only a temporary effect on the price level. As a consequence, one may conjecture that price increases will be gradually reabsorbed. Indeed, in 2003 and in 2004 both the average price increase and the number of price changes were relatively moderate.
25
Appendix 3.A. Descriptive statistics
Table 3.A1. Sample size (number of observations) Price level 1998 1999 2000 2001 2002 2003 2004 Total 2,632 2,739 2,650 2,569 2,356 2,321 1,769 17,036 Price change
2,239 2,292 2,144 2,023 2,028 1,552 12,278
Source: Authors computation based on data from Il Gambero Rosso.
26
Table 3.A2. Sample composition, by region (number of observations and percentages) Region Observations Frequency (percentages) Italy: distribution of all restaurants, 1999 (percentages) (1) 2.9 0.7 3.4 9.7 6.5 3.1 9.5 4.2 13.6 3.0 0.7 7.3 6.0 3.2 5.4 7.5 2.6 1.7 0.3 8.9 46.4 21.2 31.9 100.0
Abruzzo Basilicata Calabria Campania Emilia Romagna Friuli Venezia Giulia Lazio Liguria Lombardia Marche Molise Piemonte Puglia Sardegna Sicilia Toscana Trentino Alto Adige Umbria Val dAosta Veneto North Centre South & islands Total
336 151 269 574 1,022 426 1,471 621 1,608 396 73 984 530 307 600 1,240 445 271 113 841 6,060 3,378 2,840 12,278
2.7 1.2 2.2 4.7 8.3 3.5 12.0 5.1 13.1 3.2 0.6 8.0 4.3 2.5 4.9 10.1 3.6 2.2 0.9 6.8 49.4 27.5 23.1 100.0
Source: Authors computation based on data from Il Gambero Rosso and Chambers of Commerce. Notes: (1) Restaurants registered at the Chambers of Commerce in 1999 (source: Rapporto sul turismo italiano - X edizione).
27
Table 3.A3. Sample composition, by type (number of observations and percentages) Region Abruzzo Basilicata Calabria Campania Emilia Romagna Friuli Venezia Giulia Lazio Liguria Lombardia Marche Molise Piemonte Puglia Sardegna Sicilia Toscana Trentino Alto Adige Umbria Val dAosta Veneto North Centre South and islands Total (a) Restaurants 218 113 170 323 703 325 846 442 1,208 230 27 861 328 218 404 837 366 198 76 644 4,625 2,111 1,801 8,537 (b) Others 118 38 99 251 319 101 625 179 400 166 46 123 202 89 196 403 79 73 37 197 1,435 1,267 1,039 3,741 (a) / (a+b) (percentages) 64.9 74.8 63.2 56.3 68.8 76.3 57.5 71.2 75.1 58.1 37.0 87.5 61.9 71.0 67.3 67.5 82.2 73.1 67.3 76.6 76.3 62.5 63.4 69.5
28
Table 3.A4. Sample representativeness (number of observations and percentages) Region (a) Sample (b) Total observations in establishments in 1999 Italy in 1999 52 27 48 101 209 70 247 103 302 72 14 154 92 57 129 252 84 54 21 151 1,094 625 520 2,239 1,914 476 2,196 6,272 4,203 1,985 6,157 2,701 8,834 1,933 437 4,764 3,889 2,053 3,483 4,851 1,672 1,100 206 5,758 30,123 14,041 20,720 64,884 (a)/(b) (percentages) 2.7 5.7 2.2 1.6 5.0 3.5 4.0 3.8 3.4 3.7 3.2 3.2 2.4 2.8 3.7 5.2 5.0 4.9 10.2 2.6 3.6 4.5 2.5 3.5
Abruzzo Basilicata Calabria Campania Emilia Romagna Friuli Venezia Giulia Lazio Liguria Lombardia Marche Molise Piemonte Puglia Sardegna Sicilia Toscana Trentino Alto Adige Umbria Val dAosta Veneto North Centre South and islands Italy
Source: Authors computation based on data from Il Gambero Rosso and Chambers of Commerce.
29
Table 3.A5. Competitive pressure (number of restaurants included in the guide per capita) (1) Region Abruzzo Basilicata Calabria Campania Emilia Romagna Friuli Venezia Giulia Lazio Liguria Lombardia Marche Molise Piemonte Puglia Sardegna Sicilia Toscana Trentino Alto Adige Umbria Val dAosta Veneto (a) 0.56 0.58 0.27 0.22 0.63 0.81 0.70 0.79 0.40 0.64 0.51 0.67 0.29 0.46 0.35 0.93 1.01 0.71 2.07 0.44 (b) 0.39 0.50 0.22 0.17 0.44 0.55 0.51 0.44 0.33 0.40 0.46 0.59 0.26 0.32 0.28 0.51 0.26 0.55 0.62 0.23
Source: Chambers of Commerce, ISTAT and authors computation based on data from Il Gambero Rosso. Notes: (1) Averages of ratios computed at the province level. (a) Number of establishments included in the guide over total population in the province (tens of thousands); (b) Number of establishments included in the guide over (days of tourist presence/12 + local population) - see text.
30
Table 3.A6. The price of a meal (euros) Restaurants mean min 1st median 4th quartile quartile 25.8 25.8 28.4 31.0 32.0 35.0 36.0 31.0 31.0 33.6 36.2 40.0 40.0 45.0 36.2 38.7 41.3 43.9 48.0 50.0 55.0 max std. dev. coeff. var. 11.6 12.4 13.2 14.8 16.7 17.6 19.8 0.35 0.36 0.36 0.38 0.39 0.39 0.41
1998 1999 2000 2001 2002 2003 2004 Others
33.2 34.3 36.3 39.3 42.8 45.5 48.6
12.9 12.9 12.9 15.5 16.0 20.0 21.0
113.6 129.1 129.1 147.2 220.0 220.0 250.0
mean
min
1st median 4th quartile quartile 15.5 15.5 18.1 18.1 20.0 23.0 18.1 20.7 20.7 23.2 25.0 27.0 23.2 23.2 23.2 25.8 28.0 30.0
max
std. dev. coeff. var. 5.6 4.9 4.6 5.3 6.1 9.1 0.29 0.25 0.22 0.23 0.25 0.32
1998 1999 2000 2001 2002 2003
19.5 19.8 20.6 22.5 24.7 28.0
7.8 7.8 7.8 8.8 9.0 11.0
72.3 56.8 49.1 46.5 75.0 105.0
Source: Authors computation based on data from Il Gambero Rosso
31
Table 3.A7. Price changes, by type (percentage changes) Restaurants mean min 5th median 95th percenpercentile tile -6.7 0.0 0.0 -3.0 0.0 -7.5 0.0 0.0 9.1 7.7 4.3 0.0 25.0 30.8 33.3 29.0 20.0 21.2 max std. dev.
1999 2000 2001 2002 2003 2004 Others
4.8 7.1 10.3 9.1 5.5 3.9
-50.0 -33.3 -36.8 -27.1 -27.0 -36.4
100.0 120.1 100.1 83.3 71.9 75.0
10.8 12.3 12.7 11.2 8.5 9.6
mean
min
5th median 95th percenpercentile tile 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 11.1 8.7 6.7 0.0 19.9 28.5 33.4 29.0 20.0 18.5
max
std. dev.
1999 2000 2001 2002 2003 2004
3.3 5.8 10.8 9.8 6.4 3.6
-42.9 -33.3 -33.3 -27.8 -29.4 -21.4
75.0 80.0 80.0 94.4 66.7 60.0
9.5 12.2 12.1 11.6 8.7 8.2
Source: Authors computation based on data from Il Gambero Rosso
32
Table 3.A8. Price changes, by area (percentage changes) North mean 1999 2000 2001 2002 2003 2004 Centre mean 1999 2000 2001 2002 2003 2004 5.9 8.6 11.2 9.2 5.3 4.1 min -23.1 -27.3 -36.8 -26.9 -29.4 -36.2 5th median 95th percentile percentile 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 10.0 7.7 0.0 0.0 25.0 37.5 37.5 29.0 20.7 20.5 max 99.9 120.1 100.0 80.6 50.0 57.1 std. dev. 10.6 14.3 13.4 11.1 8.8 9.3 3.9 6.1 10.9 8.9 5.9 3.9 min -27.3 -33.3 -30.0 -26.8 -24.3 -31.8 5th median 95th percentile percentile -7.1 0.0 0.0 -3.8 0.0 -6.7 0.0 0.0 10.0 7.7 5.0 0.0 23.1 27.8 33.3 30.4 20.0 20.0 max 100.0 75.0 87.5 84.6 66.7 75.0 std. dev. 10.2 11.1 12.0 11.8 8.1 9.5
South and islands mean 1999 2000 2001 2002 2003 2004 3.5 5.8 8.8 10.2 5.9 3.1 min -50.0 -33.3 -33.3 -27.8 -12.7 -36.4 5th median 95th percentile percentile 0.0 0.0 0.0 0.0 0.0 -7.1 0.0 0.0 9.1 9.5 4.7 0.0 21.1 28.5 25.0 26.9 20.7 19.4 max 75.0 75.0 100.1 94.4 71.9 46.9 std. dev. 10.6 11.8 12.2 10.4 9.0 8.3
33
Table 3.A9. Restaurant prices in Italy: a comparison among datasets (annual percentage changes) 1999 GR guide (our panel) Michelin Red Guide (Adriani et al, 2003) HICP restaurant and cafes (Hobjin et al., 2004) ISTAT Ristoranti, pizzerie, pubblici esercizi 4.3 n.a. 1.9 1.7 2000 6.7 n.a. 3.1 3.2 2001 10.5 n.a. 3.6 3.8 2002 9.3 n.a. 5.0 4.8 2003 5.8 5.6 3.7 4.6 2004 3.8 n.a. 3.5 3.6
Source: Authors computation based on data from Il Gambero Rosso, ISTAT, Eurostat, Adriani et al. (2003).
34
Appendix 3.B - Mathematical details on the model of Section 3.6

The problem analyzed in this appendix is a small variation of Dotsey, King and Wolman (1999), which allows for a time-varying elasticity of demand.33 The value of the rm in period t if it adjusts the price to the new optimal level is dened as vt|j and it is:
|
= max
*
( *
) + +1|
+1|
+ (1 +1| )
+1| +1
+1|
(3.B1)
where Xt is the general price level, Wt denotes the marginal cost, t|j the period t (endogenous) probability of not revising a price which was set in period j, and t+1|t is the present expected value of the adjustment costs to be sustained next period. The value of the rm in period t if it maintains its price at the current level, P*t-j (where j is the price vintage) is given by:
|
( *
) + +1|
+1|
+ (1 +1| )
+1| +1
+1|
} (3.B2)
There is a xed distribution of menu costs, which we assume for simplicity to be uniform over the interval (0, 1/k). The marginal rm (i. e., the one indifferent about resetting the price) is the one for which the realized menu cost equates the gain in its value resulting from price adjustment (vt|t - vt|t-j); hence, the share of rms adjusting their price is given by the c.d.f. of the menu cost, evaluated at the point (vt| t- vt|t-j).
(1
+1| =
+1|
)/
+1| +1
+1|
)2
(3.B3)
The expected adjustment cost next period is:

(1
+1|
+1|
)/
+1| +1
+1|
)2
(3.B4)
33
We also assume consumers have access to complete markets so that the ratio of marginal utility affecting the stochastic discount factor in Dotsey, King and Wolman is constant at 1.
35
Taking the derivative of (3.B3) and (3.B4):
(
+1| +1
+1|
+1|
(3.B5)
+1| (
+1| +1
+1|
= (
+1| +1
+1|
(3.B6)
The dynamic program (3.B1) implies the condition:

0=( * ) 1 + *
+ +1|
+1|
+1|
*
+1| +1
+1|
(
+1|
+1| +1
+1|
+1|
*
+1| +1
(
+1| +1
+1| +1
*
1 1 + *
(3.B7)
= ( * )
+ +1|
+1|
Taking the derivative of (3.B2) and using (3.B5) and (3.B6):
* +
=( * (
+1| +1 +1| +1
+
+1| +1
+1|
+1|
* ) (
+1| +1
(
+1|
+1| +1
+1|
* ) =
+1|
* ) 1
+1|
+1|
(3.B8)
=( *
+ +1|
+1|
Substituting (3.B8) into (3.B7) and iterating yields:

0 =
=0 +
=0
+ |
( *
+ + *
1
+ +
(3.B9)
Dening +
+ *
and +
+ *
, yields the expression for

+
the optimal reset price Pt* (equation 3.7 in the main text).
36
Based on (3.B8), vt|t can be linearized in a neighbourhood of [Pt-j*, vt|t-j] to obtain an expression for the impact of a price revision on rms value:
|
| *
) =
=0
+ |
(1 + (1
+ *
))
1
+ +
(3.B10)
where
+|

=0
+1+ |
. Finally, the current price Pt is a weighted average
of past P*t s:
= |
=0
(3.B11)
where the share of vintage h prices in the total of period t prices, t|t-h , is recursively given by:
t|t-j = t|t-j t-1|t-j
(3.B12)
The simulation follows these steps: i) a matrix of probabilities {i|j} is assumed as an initial guess; ii) the path for Pt* is derived using (3.6); iii) the composition of rms by price vintage i,j is derived using (3.B12), given beginning-of-period values; iv) the path for Pt is derived, as a weighted average of past Pt*s, using (3.B11); v) the increase in prots resulting from price adjustment is derived using (3.B10); vi) a new matrix {i|j} is derived using (3.B3); the process is iterated until convergence. To avoid the curse of dimensionality, a steady state solution is rst derived, assuming marginal cost grow at a constant rate and keeping constant t and t through time (without loss of generality, t can be normalised to 1). Subsequently, we introduce in period t0 the shocks discussed in the text. In the rst simulation (menu cost) we set t0|t0-j=0 (for all js); in the second simulation (market power) we set t0= and determine t0 according to the expression in footnote 28 above. In each simulation, we re-estimate the elements t|t-js in a sufciently large neighbourhood of t0, i.e. for t(t0c,t0+c),34 while keeping the remaining elements in {i|j} at their steady state value.
34
We rst try with c=1 and then increase c until the estimates of t|t-j are no longer affected.
37
References
Adriani, F., Marini, G., Scaramozzino, P. (2003) The inationary consequences of a currency changeover: evidence form the Michelin Red Guide, CEIS Tor Vergata Research Paper Series, 27, July. Bils, M., Klenow, P. J. (2002) Some evidence on the importance of sticky prices, NBER working paper 9069. Calvo, G. A. (1983) Staggered prices in a utility-maximizing framework, Journal of Monetary Economics, 121, 383-398. Deutsche Bundesbank (2004) The euro and prices two years on, Monthly Report 56 (1), 15-28. Dotsey, M., King, R. G., Wolman, A. L. (1999) State-dependent pricing and the general equilibrium dynamics of money and output, Quarterly Journal of Economics, 114, 655-690 European Central Bank (2003a) Effects of the introduction of the euro banknotes and coins on consumer prices, Annual Report on 2002, 40-42. European Central Bank (2003b) Recent developments in euro area ination perceptions, Monthly Bulletin, October, 24-25. Fabiani, S., Gattulli, A., Sabbatini, R. (2007) The pricing behaviour of Italian rms: new survey evidence on price stickiness, in Fabiani, S., Loupias, C., Martins, F., Sabbatini, R. (eds.), Prices decisions in the euro area, Oxford University Press, New York. Hobijn, B., Ravenna, F., Tambalotti, A. (2004) Menu costs at work: restaurant prices and the introduction of the euro, Federal Reserve Bank of New York, Staff Reports, 195. Johnston, J., Di Nardo, J. (1997) Econometric methods, Mc-Graw Hill. Rapporto sul turismo italiano - X edizione, 2001, Touring Club Italiano. Rapporto sul turismo italiano - XI edizione, 2002, Touring Club Italiano. Ristoranti dItalia, Gambero Rosso Editore, years 1999-2004. Roma del Gambero Rosso, Gambero Rosso Editore, years 1999-2004.
4. Did prices really soar after the euro cash changeover? Evidence from ATM withdrawals
Paolo Angelini* and Francesco Lippi** * Bank of Italy, Department of Economic Outlook and Monetary Policy Studies ** University of Sassari and CEPR
4.1 Introduction1
As described in chapter 1 of this book, there is a widespread perception among the citizens of the Euro area that the introduction of the euro notes and coins in the rst months of 2002 spurred a rise in prices that was much sharper than measured by the national statistical ofces (see European Central Bank 2002b, 2003a,b). This phenomenon (see Fig. 1.1 in Chap. 1) has been the subject of countless newspaper articles and of several ofcial speeches by policymakers and politicians and the perception that the euro brought about price increases is still vivid.2 It is somewhat puzzling that a change in the unit of account might have an impact on ination. Indeed, a number of conjectures have been formulated to explain the discrepancy between ination perceptions and the ofcial statistics, emphasizing the role of psychological factors (e.g. Traut-Mattausch et al. 2004) and/or the disproportionate inuence of a few industry prices on individual
1
The study was initially drafted while Francesco Lippi was working at the Bank of Italy. The authors are indebted to Guerino Ardizzi and Eugenio Gaiotti for comments on an earlier draft. All remaining errors are their own. In May 2002 Prof. O. Issing gave a speech in Mainz on Der Euro - eine stabile Whrung. After the speech, the rst question from the audience was about the teuro phenomenon. Seeing the look of disbelief with which his explanation was met, Issing replied: You seem not to believe me. And even my wife doesnt believe me. This sentence found wide coverage in the German press. Two years later, President Trichet still deemed it necessary to reassure European customers on this issue: European citizens who still perceive that ination is higher than measured by ofcial indices should be assured that the ofcial measures are accurate and that we will continue to maintain price stability in the future (introductory statement after the Governing council meeting of April 2004).
P. Angelini and F. Lippi
perceptions. Hobijn, Ravenna and Tambalotti (2006) and Gaiotti and Lippi (2004 and chapter 3 of this book) analyze the dynamics of restaurant prices and nd evidence consistent with a price hike (mainly driven by a lumping of price revisions in an industry where price revisions are normally infrequent). Deutsche Bundesbank (2004) provides comparable evidence for some German services (restaurants, cinemas, dry-cleaning and hairdressers). Other papers argue that ination perceptions are mainly affected by the prices of goods that are cheaper and more frequently purchased (Del Giovane and Sabbatini 2006 and chapter 1 of this book; Ehrmann 2006). Dziuda and Mastrobuoni (2005) and Mastrobuoni (2004) present a model that rationalizes why such goods are the ones that actually record greater price increases. While useful, these studies do not provide a direct answer on whether the general price level was measured with error during the changeover. Rather, they maintain the assumption that ofcial statistics are correct. The main obstacle faced by researchers interested in verifying that assumption is the absence of reliable alternative ination measures. The thesis that price increases were much larger than measured by the national statistical ofces, suggested by the indicators of perceived ination, remains mostly based on anecdotal evidence. This chapter investigates the dynamics of the general price level in Italy after the introduction of the euro notes and coins, at the beginning of January 2002, by using data on currency withdrawn from the ATM network. We believe that this inference is useful because it relies on data collected and assembled by central banks, with methodologies that are completely independent from those used by the national statistical ofces to measure prices. The basic steps of our investigation can be summarized as follows. We setup a simple model of ATM withdrawals and estimate it prior to the changeover, when ofcial statistics were arguably correct. We then present a set of assumption under which the estimated demand equation for ATM withdrawals can be used to back out price level dynamics from the observed nominal time series for ATM withdrawals and consumer expenditures. Specically, we show that if a bias materialized in the ofcial data on prices after the changeover, but not in those on cash withdrawals and consumption expenditures, then extending the estimation sample to the changeover period (2002-03) should cause a specic form of instability in the estimated coefcients, which can be captured econometrically. Formally, we test the null hypothesis that the increase in consumer prices is correctly measured by ofcial statistics after the changeover. Both a price-level bias and an ination-bias hypothesis are formulated and tested. The analysis fails to nd evidence consistent with the occurrence of a price hike after the changeover. This result cannot be attributed to lack of power of the test: a counterfactual exercise suggests that our methodology is sufciently powerful to identify an ination bias greater than 0.5 percentage points. Several reasons motivate our focus on Italy. First, the country is broadly representative of the euro area in terms of the discrepancy between ofcial and perceived ination (see Fig. 1.1 in Chap. 1). Also, quarterly data on
Did prices really soar after the euro cash changeover? Evidence from ATM withdrawals
cash withdrawals are available, whereas comparable data are available only at an annual frequency for other euro area countries (to our knowledge). It is important to explain why we focus on the ow of currency withdrawn from the ATM circuit, rather than on more traditional monetary aggregates. The stock of currency experienced a strong decline from the beginning of 2001 (apparently reecting weak demand of banknotes as a store of wealth due to the approaching currency changeover) and an equally strong rebound thereafter.3 Among the traditional monetary aggregates, M1 is strongly affected by the erratic behavior of currency. M2 and M3 are comparatively less affected, but they are typically less related to transactions; in addition, over the recent past the dynamics of these aggregates has been inuenced by portfolio reasons, as repeatedly stressed by the European Central Bank. By contrast, there is no obvious reason why ATM withdrawals mainly driven by transactions demand should have been affected by these same factors. The data in Fig. 4.1 broadly conrm this view: neither the average number nor the unit value of ATM withdrawals made in each quarter by a typical cardholder show the discontinuity which clearly stands out for the traditional monetary aggregates around the changeover.
8 6 4 2 0 -2 Jan- Nov- Sep- Jul- May- Mar- Jan- Nov- Aug- Jun- Apr- Feb- Dec- Oct93 93 94 95 96 97 98 98 99 00 01 02 02 03 Average number of ATM withdrawals per card per quarter Real consumption of non-durable goods (y/y growth rate) Real consumption of non-durable goods and services (y/y growth rate) Average unit value of ATM withdrawals (right-hand scale)
2002 2003
170 166 162 158 154 150
Fig. 4.1. ATM usage and consumption growth (quarterly data; left-hand scale if not differently indicated) Source: Bank of Italy and ISTAT.
3
See e.g. European Central Bank (2002a). As we argue in what follows, this decline is likely due to the attempt by currency holders to run down their cash inventories. These are mainly held in large denomination notes, and are to a signicant (although not easily quantiable) extent held outside the national borders, and/or for grey/black economy purposes.
The chapter is organized as follows. The next section presents a simple model of the demand for ATM withdrawals, which is used as a guideline in the empirical analysis of section 4.3. Section 4.4 concludes.
4.2 A simple model of aggregate ATM withdrawals

This section presents a model aimed at interpreting the evolution of aggregate ATM withdrawals. To match the aggregate data we rst focus on the choice concerning the withdrawals by a representative ATM cardholder and, next, present an aggregation to account for the growing number of cardholders over our estimation period.4 Let i index an agent who possesses an ATM card and Ei denote her nominal consumption expenditure over a given time span. To pay for Ei the agent can use cash Ci withdrawn from an ATM, bearing a cost rc, or some other means of payment Q (e.g. POS, credit card or cheques), whose cost is denoted by rq. The demand schedule for the agent ATM withdrawals is of the following form: Ci = (rq / rc) Ei, where the function () is increasing and concave.5 This demand function stipulates that the proportion of cash withdrawals over total nominal expenditure is decreasing in the relative cost between ATM cash and that of alternative means of payments. To give empirical content to these costs, in the analysis that follows we assume that the cost of ATM cash, rc, is increasing in the nominal interest rate R (whose value determines the amount of forgone interest on deposits) and decreasing in the size of the ATM network, dATM. Moreover, we proxy the cost of alternative payment means, rq, using a measure of development of the POS network, denoted by dQ.6 The individual proportion of cash expenditures is thus hypothesized to depend on the diffusion of the ATM and POS network and the nominal interest rate: Ci = (dQ/rc(dATM, R))Ei (4.1)
Let us now consider the aggregation problem. In order to bring equation (4.1) to the data, we need to relate the nation-wide demand for ATM cash
4
Based on the Bank of Italys Survey of Households Income and Wealth, between 1989 and 2004 the proportion of households owning an ATM card rose from 15 to 66 per cent. This cash/credit choice can be thought of along the lines of Lucas and Stockey (1987). In Italy, our data source for the tests presented in the next section, it is not possible to get cash back at POS. This option, available to customers in several industrialized countries, would have made this proxy questionable.
to aggregate consumption, as data on total expenditure of ATM cardholders are not available. Let n and E/P denote the number of ATM cardholders and their aggregate expenditure (deated by the price level), respectively. Analogously, let N and E/P denote the population size and real aggregate expenditures. We postulate the following: E/P = n/N (E/P) (4.2)
This equation states that the growth in the aggregate expenditure of cardholders is proportional to the growth of the population fraction of cardholders and to the growth rate of aggregate expenditures, with a constant of proportionality that may differ from one. Aggregating equation (4.1) over all cardholders and replacing E using (4.2) yields an expression relating the real aggregate ow of ATM withdrawals to real aggregate expenditure: C/P = n/N (dQ/rc(dATM, R)) (E/P) (4.3)
Equation (4.3) summarizes the determinants of aggregate ATM cash withdrawals discussed thus far. C is increasing in ATM diffusion (dATM) and decreasing in the nominal interest rate (R) and in the ease of resorting to non-cash payments (dQ). The elasticity of aggregate cash withdrawals with respect to aggregate expenditure is given by . It is immediate that if is equal to one then the price level drops from equation (4.3) and no information about it can be retrieved from that equation. For all other values, instead, one can invert equation (4.3) and use the information on C and E to back out the price level. In Section 4.3.2 we discuss the functional form adopted in equation (4.2) and review the available empirical evidence on , which is key for our identication of price level dynamics.
4.3 Empirical evidence

This section begins by presenting some descriptive evidence on ATM withdrawals during the period 1993-2003. Using equation (4.3) as a guide for the empirical specication, we estimate a currency expenditure equation and formally test for the presence of measurement error in the aggregate price level. We conclude the section by exploring the power of the statistical test and analyzing the robustness of the estimates (considering e.g. parameter instability and potential endogeneity of the regressors).
4.3.1 A preliminary look at the data Our dataset comprises quarterly time series over the 1993.II-2003.IV period.7 Fig. 4.1 shows that during the rst three quarters of 2002 the average ATM withdrawal per card records a sharp increase (from 157 to 165 euro), but only after an equally sharp fall in 2001.8 Overall, the withdrawal per-card after the changeover remains close to the values recorded in the previous ve years.9 The same conclusion holds for the frequency of ATM usage (solid line), roughly constant at 6.3 withdrawals per card per quarter since 1995. Fig. 4.1 also displays the behavior of households real consumption of non durable goods and services in the period. Growth begins to slow down in the second half of 2000, bottoming in the fourth quarter of 2001. At least prima facie, none of these time series displays a behavior that might signal a sharp increase in the price level after the introduction of the euro banknotes and coins. However, this descriptive evidence can be potentially misleading, for at least two reasons. The rst relates to our measure of consumption. Assume that ination perceptions are correct, and that the ofcial nondurable goods deator underestimates ination in 2002-03; then, real consumption growth in Fig. 4.1 would be correspondingly overestimated. Second, the descriptive evidence does not take into account several structural changes which have occurred over the last 15 years: most notably, the diffusion of ATM and POS terminals, and the associated increase in aggregate cash withdrawals from ATMs and in the use of non-cash payments via debit cards. Such developments typically follow lowfrequency trends, so in our view it is unlikely that they may have obscured the effect of the hypothesized price jump on the demand for cash and alternative payment instruments in 2002-03. However, the model of section 4.2 suggests that they affect the demand for ATM withdrawals and should thus be taken into account. This is done in the next subsection.
7
We refer to two sources. The ow of cash withdrawn from ATMs in Italy, the number of ATM cards, POS and ATM terminals, and the interest rate on checking accounts are provided to the Bank of Italy by the banking system for supervisory reasons. Data on consumption of services and non-durable goods, and the related deators, are released by the National Institute of Statistics (ISTAT). All the series used in the chapter refer to Italy. This swing is likely due to the need by currency holders to run down their cash inventories. As mentioned in the introduction, a massive drop in currency outstanding was recorded in 2001. As the cash changeover was largely anticipated by the public, it seems plausible that, in the nal months of the year, currency holders avoided withdrawing cash and used the inventory to face day-to-day purchases. After the changeover the average use of ATM cash returned back to normal. Such stationarity, in a period of moderate but positive ination, might reect the increasing use of cash substitutes; the regression analysis below supports this hypothesis.
4.3.2 Inference from an estimated currency expenditure equation A log-linear version of equation (4.3) yields: ct pt = + dtATM+ dtQ + (et pt) + rt + nt + t (4.4)
where t is an error term with variance 2 . We measure ct by the (log) nominal value of nationwide quarterly withdrawals from ATM terminals, nt by the (log) number of outstanding ATM cards (at the end of the quarter). rt is (log) interest rate on checking accounts. The diffusion of the ATM network, dATM, is proxied by the ratio between the number of ATM terminals nationwide and the number of ATM cards. Similarly, the diffusion of alternative payment instruments, dQ, is measured by the ratio between the number of POS terminals and nt. Finally, et and pt are proxied by the (log) aggregate nominal consumption of services and non durable goods, and by its deator, in the order.10 Can equation (4.4) help us shed light on the issue at the core of this paper? A rst assumption needed for the answer to be positive is that all of the time series appearing in (4.4) be measured with no error until the fourth quarter of 2001.11 This assumption seems reasonable, since until the end of 2001 there was no argument about data quality. It implies that the coefcients of (4.4) will not be affected by measurement problems if estimated over the pre-changeover period. Concerning the post-changeover period, we assume that all variables appearing in (4.4) are measured correctly, with the possible exception of the price level. Specically, we allow for the possibility that pt, the true (log) price level, may suffer from measurement error and deviate from its observed counterpart, po. t As anticipated in section 4.3, a condition on the elasticity of cash withdrawals with respect to real consumption () must also be satised. It is easy to see that if =1 the price level cancels from both sides of (4.4), and the equation becomes uninformative on the price level dynamics. Thus, an estimated 1 is necessary (and sufcient) to back out the price level after 2001 from (4.4).
10
11
Based on (4.3) the population size, N, should also appear among the regressors. We omit it because it was roughly constant over the estimation period. More precisely, we are assuming that the measurement error present in ofcial statistics (e.g. the well-known price-index bias) is not affected by the cash changeover. Under our alternative hypothesis we allow the changeover to bring about a source of measurement error in price measures additional to those that are ordinarily present.
We begin by estimating (4.4) using data until 2001.IV. The results are reported in Table 4.1, column (a).12 The coefcients are in line with the model suggestions: the diffusion of non-cash forms of payment reduces cash withdrawals, while the diffusion of ATM terminals increases them. The coefcient of the interest rate on checking accounts is negative but not signicant. The coefcient on the number of ATM cards is 0.89, not signicantly different from one (t statistic of 0.7). The point estimate of the parameter is 2.82, statistically greater than one at the 5 per cent condence level (t statistic of 2.2). Thus, the requirement spelled out above is satised; the functional form hypothesized in equation (4.2) nds support in the data.13 Altogether, while simple, equation (4.4) seems to capture some essential features of the demand for ATM withdrawals. Considering that it does not feature a lagged dependent variable on the right-hand side (we experimented with specications incorporating one, but the related coefcient turned out to be non signicant), it tracks the data quite well. We can now test the null hypothesis that after the changeover pt= po, t against the alternative pt > po. Since pt is not observable after 2001.IV, we t consider two hypotheses about its behaviour. The rst is the following: pt= po, t2001.IV, t pt= po + + t , t2002.I t (4.5)
where is a positive constant and t a white noise term independent of t, with variance 2 . Expressions (4.5) could be an appropriate description of a one-off increase in the true price level after the changeover. An alternative hypothesis is: pt= po, t2001.IV, t pt= po + gT + t , t2002.I t (4.5)
where g is a positive constant and T is a linear trend (T=1 in 2002. I, T=2 in 2002.II, ...). This formulation would entail a widening gap between the observed (ofcial) and the true price deator, implying a permanent ination bias. It would be unrealistic for large T, but could
12
13
Three quarterly dummies (not shown) are included among the explanatory variables to account for seasonal effects. The micro data from Bank of Italys Survey of Households Income and Wealth also suggest that is greater than one. Introducing time subscripts in (4.2), taking logs of both sides and differentiating, can be shown to equal the ratio between the growth rate of consumption expenditure for the average ATM cardholder and the corresponding growth rate for the average consumer. Between 1998 and 2004 this ratio averaged about 1.4, as consumption among cardholders consistently outgrew average per capita consumption at the nationwide level.
Table 4.1. Estimates of equations 4.4 and 4.6 (1)

Equation (4.4) estimated over 1993.II2001.IV Equation (4.6) estimated over 1993.II-2003.IV test of hypothesis (4.5) test of hypothesis (4.5)
estimaestimaestimaestimating only ting only ting all ting all dummy dummy coefcoefcoefcoefcients cients cient cient (a) ATM terminals diffusion (dtATM) POS terminals diffusion (dtQ) 2.38** 5.9 -17.05** -5.0 2.82** 3.6 0.89** 5.6 -7.1e-4 0.0 -8.82 -0.8 35 0.98 1.47 (b) 2.38 Constr. (c) 2.34** 7.0 (d) 2.38 Constr. (e) 2.43** 6.8
-17.05 -17.00** Constr. -6.9 2.82 Constr. 0.89 Constr. -7.1e-4 Constr. -8.82 Constr. -6.2e-3 -0.6 0.90 43 1.59 2.75** 4.0 0.91** 6.5 3.4e-3 0.2 -8.57 -1.1 -6.4e-3 -0.4 0.96 43 0.99 1.58
-17.05 -15.75* Constr. -5.7 2.82 Constr. 0.89 Constr. -7.1e-4 Constr. -8.82 Constr. -2.0e-3 -1.0 0.95 43 1.60 2.57** 3.8 0.89** 6.2 6.9e-3 0.3 -5.42 -0.6 -3.6e-3 -1.1 0.99 43 0.99 1.60
Real consumption (log; et- pt) Number of ATM cards (log; nt) Interest rate on deposits (log; rt) Constant Dummy 2002.I-03.IV Dummy 2002.I-03.IV*linear trend F test for 2 = 2 vs. 2 > 2 (p-value) N observations R DW
2
Notes: (1) The dependent variable is the (log of) aggregate cash withdrawals from ATM in real terms. 2 and 2 denote, in order, the variance of the error term in equation (4.6) before and after 2001.IV. OLS estimates. Heteroskedasticityrobust t statistics are reported below each coefcient. One or two asterisks denote, respectively, 5 and 1 per cent signicance levels. The regressions also include three seasonal dummies (coefcients not reported). The linear trend takes integer values between 1 and 8 over the 2002.I 2003.IV period.
10
be appropriate over our sample period, which only covers eight quarters after the changeover. Substituting (4.5) into (4.4) yields: t2001.IV (4.6) o ATM Q o ct p t = + 0 + dt + dt + (et p t ) + rt + nt + t, t2002.I where 0 = (1), and t = t + (1) t, with variance 2 = 2 + (1)2 2 . A way to test the null hypothesis of no distortion in the price level after the changeover against the alternative hypothesis (4.5) would then entail estimating equation (4.6) over the entire sample period 1993.II-2003.IV, introducing a dummy variable to allow the constant to change over the last two years, and checking for heteroskedasticity. However, it is easy to check that under the alternative hypothesis (4.5), equation (4.6) is affected by a classic errors in variables problem; if 2 >0, OLS coefcients would be inconsistent and biased towards zero. To circumvent this difculty, we restrict the parameters in (4.5) to take the values estimated over the 1993.II-2001.IV period, and estimate only the coefcient of the 2002-03 dummy, 0, which is an unbiased estimator of (1). The results of this exercise are presented in column (b) of Table 4.1. The estimated 0 has a value of 0.0062; using this gure and the estimated value for (2.82), it can be reckoned that the average ination rate in 2002 was 0.3 percentage points higher than that computed using the ofcial deator. However, a onetail t test cannot reject the null that 0 is zero against the alternative that it is negative. Column (b) also reports the result of an F tests of the null hypothesis 2 = 2 against the alternative 2 < 2 , which should hold based on (4.5). Again, the null is not rejected. As mentioned above, if (4.5) were true and the error-in-variables problem were serious after 2001.IV, OLS coefcients should be biased towards zero. Therefore, as a further check we estimate all the parameters of (4.6) over the entire sample period. The estimated coefcients show that the parameters remain stable (see column c). Next, we replicate the exercise for our second alternative hypothesis. Substituting (4.5) in (4.4) yields an equation identical to (4.6), except that now 0 = (1)gT. Thus, beginning with 2002.I a linear trend with coefcient g(1) should enter the equation. Also, the error term should display the same form of heteroskedasticity as under hypothesis (4.5). Specications in columns (d)-(e) of Table 4.1 show no evidence consistent with the hypothesis of an increase in the price level after the changeover. In both cases the coefcient of the 2002.I-2003.IV dummy interacted with the time trend is negative. A calculation based on the larger value, -3.6e-3, together with the estimated value of 2.82 for , suggests that the ct po = + dtATM + dtQ + (et po ) + rt + nt + t, t t
11
average ination rate in 2002 was 0.5 percentage points higher than based on the ofcial deator. However, even in this case the coefcient is not statistically different from zero, and the null of homoskedasticity cannot be rejected.14 4.3.3 Exploring the power of the statistical test Our econometric procedure amounts to a t test on the coefcient of a dummy in a linear regression. Therefore, the properties of our tests and their statistical power are well grounded in standard asymptotic and small sample theory. It could be argued, however, that the precision of our estimates is insufcient to generate adequate power, e.g. because of the short sample period. To investigate this hypothesis, we perform a counterfactual exercise; we assume that beginning in 2002.I the true price deator is higher than the ofcial one. Using (4.5), we set po = pt after the changeover. We then t assign numeric values to and re-estimate specications (b) and (c). If our tests have sufcient power, the coefcient of the 2002-03 dummy should become negative and signicant for relatively small values of . We report the results of this exercise for values of ranging between 0.005 and 0.1, implying that in 2002 ination was between 0.5 and 10 percentage points higher than recorded by ofcial statistics. Fig. 4.2 plots the true , measured on the horizontal axis, against its ^ estimated value, , obtained as the ratio between the coefcient of the 200203 dummy and (1-). The curves obtained with specications (b) and (c) virtually overlap, so that they can hardly be distinguished in the gure. They are very close to the 45 lines, indicating that the size of the distortion is captured quite well in fact, it is systematically slightly overestimated. The gure also shows the precision of the estimates, measured by the t statistic of the 2002.I-2003.IV dummy. Both specications fail to detect the presence of a 0.5 percentage point distortion. However, specication (b) manages to correctly signal as statistically signicant (with a 5 per cent condence level) a value of as little as one per cent; in this case specication (c) yields a p-value of 0.09. Both specications capture values of greater than one per cent at least at the 5 per cent condence level.
14
While we focused on hypotheses (4.5) and (4.5) in order to maximize the power of the test, we also tested hypothesis pt = pt0 + + gT + t, t2002.I, which nests (4.5) and (4.5). The estimated coefcients for the 2002-03 dummy and for the time trend results were not statistically different from zero (individually as well as jointly).
12

0.12 0.10 24 20 16 12 8 4
5% significance level
estimated
0.08 0.06 0.04 0.02 0.00 0.005 0.01 0.02 0.03 0.04
0 0.08 0.09 0.1
0.05 true
0.06
0.07
Estimate of parameter from specification (c) (left-hand scale) Estimate of parameter from specification (b) (left-hand scale) t statistic of 2002.I-03.IV dummy from Table 4.1, specification (c) (right-hand scale) t statistic of 2002.I-03.IV dummy from Table 4.1, specification (b) (right-hand scale)
Fig. 4.2. The power of the statistical test: results from a counterfactual exercise Notes: (1) Specications (b) and (c) in Table 4.1 were re-estimated using a counterfactual price deator, which jumps by a in 2002.I. See the text for a detailed description of the exercise. The horizontal axis measures , the shift in the price level occurred after the changeover, based on hypothesis (4.6) in the text. The vertical left axis measures the estimated , computed as: = (estimated coefcient of 2002.I2003.IV dummy)/(1-). The t statistics, measured on the right axis, are reported in absolute value.
4.3.4 Robustness check The above results were subjected to a number of robustness checks. First, we checked the stability of the specication reported in Table 4.1, column (a). The exercises described in section 4.3.2 entail detecting a structural break in the equation after the fourth quarter of 2001. Thus, it is important that the coefcients in column (a) of Table 4.1 be stable. An obvious candidate for a structural break is the beginning of the single euro area monetary policy regime, in January 1999. Therefore, the ve coefcients through of specication (a) were allowed to change over the 1999.I-2001.IV period. The F test of the null hypothesis that the changes in the coefcients are jointly equal to zero yields F(5,20) =2.66, which does not allow to reject the null of parameter stability at the 5 per cent condence level. However, since this value is close to signicance, the tests in Table 4.1 were replicated using the equation which allows for parameter change over the 1999.I-2001.IV
13
period. The results were qualitatively analogous to those in Table 4.1, and therefore are not reported. Second, the estimates of equation (4.4) reported in Table 4.1 are subject to a potential endogeneity bias problem, as some right hand-side variables (e.g. expenditure) may be simultaneously determined with the dependent variable. Thus, we re-estimated specication (a) with two-stage least squares, instrumenting dtATM , dtQ , et pt and nt with their lagged values. The results (not reported) are virtually unchanged. Third, the analysis in section 4.3.2 relies on the hypothesis that nominal expenditure et is measured without error prior and after the changeover. However, nominal components of consumption expenditure are computed using both value data (i.e. data measured in nominal terms) and data built from price and quantity indices. Possible mis-measurements in the prices of these components after 2001.IV will in principle bias et as well. Since a detailed breakdown of the data on household consumption by construction method is not available, it is not possible to address this concern in a precise way (see ISTAT 2000). However, ISTAT does publish a breakdown of consumption expenditure used in section 4.3.2 into two categories: non durable goods and services. The former is virtually entirely built from value data, and therefore it is not affected by possible mis-measurement in pt after the changeover. Thus, we re-run the regressions in Table 4.1 proxying et with consumption of nondurable goods, i.e. excluding expenditure on services.15 Table 4.A1 in the appendix reports the instrumental variables estimates of this last specication. The elasticity of money demand to real consumption is now 2.3, slightly lower than in Table 4.1, but still signicantly different from one at the 5 per cent condence level. No appreciable changes in the other results emerge. Similar results were obtained with OLS. Finally, we re-run our exercises using specications featuring several alternative combinations of regressors: the number of ATM terminals, the number of POS terminals, and the number of ATM cards separately, and/or total households consumption (as opposed to consumption of nondurables and services). In this case as well, the main results of the analysis remain unchanged.
4.4 Conclusions
Did the euro cash changeover trigger a sudden, substantial increase of the price level in the euro area, largely undetected by the national statistical
15
Angelini, Ardizzi and Lippi (2005) also use consumption of non durables as a proxy for expenditure, in a specication featuring ination and a time trend among the regressors. The results of the tests are analogous to those reported here.
14
ofces? This chapter presented a simple indirect method to address this question for Italy. The basic idea underlying the testing strategy entails searching for the effects that the hypothesized large increase in the price level should have induced on the dynamics of payment instruments, notably cash withdrawals from ATM terminals. A quarterly dataset on aggregate ATM withdrawals and nominal consumption expenditures in Italy was used to test the hypothesis. The estimation of a demand equation for ATM cash withdrawals, conducted along the lines suggested by a simple theoretical model, allows us to set up econometric tests of the null hypothesis that after the currency changeover, in the rst months of 2002, the ofcial price index continued to measure the general price level correctly, against the alternative that it underestimated it. The main result of the analysis is that none of the various tests performed provides evidence against the null. Specically, our point estimates imply that the average ination rate in 2002 was about a half percentage point higher than computed using the ofcial deator; however, this effect is not statistically signicant. To assess the possibility that failure to reject the null is due to lack of power, we perform a counterfactual exercise: we introduce an articial increase in the deator time series beginning in 2002, and re-run our tests. The equation accurately captures the magnitude of the ination distortion, correctly signaling it as statistically signicant as soon as it grows greater than or equal to one percent on an annual basis. We conclude that the determinants of the well-documented disconnect between ofcial and perceived measures of ination cannot be ascribed to a sizeable mis-measurement by the national statistical ofces of the euro area countries.
15
Appendix 4.A
Table 4.A1. Alternative estimates of equations 4.4 and 4.6 (1) (et - pt measured as consumption of non-durable goods; I.V. estimates)
Equation (4.4) estimated over 1993. II-2001. IV Equation (4.6) estimated over 1993.II2003.IV test of hypothesis (4.5) estimating only dummy coefcient ATM terminals diffusion (dtATM) POS terminals diffusion (dtQ) Real consumption (log; et- pt) Number of ATM cards (log; nt) Interest rate on deposits (log; rt) Constant Dummy 2002.I-03.IV Dummy 2002.I-03.IV*linear trend F test for 2 = 2 vs. 2 > 2 (p-value N observations R2 DW (a) 2.47** 4.1 -12.49** -5.2 2.30** 3.9 1.11** 7.7 0.01 0.4 -4.40 -0.6 34 0.99 1.85 estimating all coefcients test of hypothesis (4.5) estimating only dummy coefcient estimating all coefcients
(b) (c) 2.47 2.49** Constr. 6.6 -12.49 -12.96** Constr. -6.6 2.30 2.52** Constr. 4.3 1.11 1.10** Constr. 8.7 0.01 0.02 Constr. 0.8 -4.40 -6.56 Constr. -1.1 -8.5e-3 -3.7e-3 -0.9 -0.3 0.89 43 1.69 0.92 42 0.99 1.88
(d) (e) 2.47 2.54** Constr. 4.9 -12.49 -12.13** Constr. -5.4 2.30 2.30** Constr. 3.7 1.11 1.08** Constr. 8.7 0.01 7.2e-3 Constr. 0.3 -4.40 -3.54 Constr. -0.5 -1.8e-3 -2.2e-3 -1.0 -0.7 0.90 43 1.69 0.96 42 0.99 1.88
Notes: (1) The dependent variable is the (log of) aggregate cash withdrawals from ATM in real terms. 2 and 2 denote, in the order, the variance of the error term in equation (4.6) before and after 2001.IV. Instrumental variables estimates. dtATM, dtQ, nt and et - pt are instrumented using their own lags. et - pt is proxied by real consumption of nondurable goods. Heteroskedasticity-robust t-statistics are reported below each coefcient. One or two asterisks denote, respectively, 5 and 1 per cent signicance levels. The regressions also include three seasonal dummies (coefcients not reported). The linear trend takes integer values between 1 and 8 over the 2002.I 2003.IV period.
16
References
Angelini, P., Ardizzi, G., Lippi, F. (2005) Indicazioni indirette sulla dinamica dei prezzi: landamento del circolante e degli strumenti di pagamento in Italia e nellarea delleuro, in Del Giovane, P., Lippi, F., Sabbatini R. (eds.) Leuro e linazione. Percezioni, fatti e analisi, Il Mulino, Bologna. Del Giovane, P., Sabbatini, R. (2005) Perceived and measured ination after the launch of the euro: explaining the gap in Italy, Giornale degli Economisti, 65, 155-192. Deutsche Bundesbank (2004) The euro and prices two years on, Monthly Report, 56 (1), 15-28. Dziuda, W., Mastrobuoni, G. (2005) The Euro Changeover and Its Effects on Price Transparency, and Ination. Mission Euro, Mission Accomplished!, Princeton University, mimeo. Ehrmann, M. (2006) Rational inattention, ination developments and perceptions after the euro cash changeover, European Central Bank Working Paper Series, 588, February. European Central Bank (2002a) Impact of the euro cash changeover on currency in circulation and M3, Annual Report 2001, 18-19. European Central Bank (2002b) Recent developments in consumers ination perceptions, Monthly Bulletin, July, 18-19. European Central Bank (2003a) Effects of the introduction of the euro banknotes and coins on consumer prices, Annual Report on 2002, 40-42. European Central Bank (2003b) Recent developments in euro area ination perceptions, Monthly Bulletin, October, 24-25. Gaiotti, E., Lippi, F. (2004) Pricing behavior and the introduction of the euro: evidence from a panel of restaurants, Giornale degli Economisti, 63, 491-526. Hobijn, B., Ravenna, F., Tambalotti, A. (2006) Menu costs at work: restaurant prices and the introduction of the euro, The Quarterly Journal of Economics, 121, 3, 1103-1131. ISTAT (2000) Le nuove stime dei consumi nali delle famiglie, Metodi e Norme, 7. Lucas, R.E. Jr., Stockey, N. (1987) Money and interest in a cash in advance economy, Econometrica, 55, 491-513. Mastrobuoni, G. (2004) The Effects of the Euro-Conversion on Prices and Price Perceptions, CEPS Working Paper 101. Traut-Mattausch E., Schultz-Hardt, S., Greitemeyer, T., Frey, D. (2004) Expectancy conrmation in spite of disconrming evidence: the case of price increases due to the introduction of the euro, European Journal of Social Psychology, 34, 739-760.
5. Memory for prices and the euro cash changeover: an analysis for cinema prices in Italy
Vincenzo Cestari*, Paolo Del Giovane** and Clelia Rossi-Arnaud*** * Lumsa University, Faculty of Educational Science, and CNR, Institute of Neuroscience ** Bank of Italy, Department of Economic Outlook and Monetary Policy Studies *** University of Rome La Sapienza, Psychology Department
5.1 Introduction 1
The lively debate in the post-changeover period shows that consumers have strong opinions on ination. In principle, this implies that they should be able to calculate, at least roughly, the rate of change in the prices of the wide range of goods and services included in the ofcial index, given that ination is the rate of change of these prices between two moments in time. To do so, however, they need to remember the past prices of these products. Even considering the individuals personal experience of ination, rather than the average ination in a country, consumers should at least be able to remember the past prices of the products that make up their personal consumption basket. The question addressed by this study is whether they actually can. The analysis focuses on the price of a specic service and on the case of the euro cash changeover, but the issue has a broader relevance. The price of cinema tickets was chosen because it can be reconstructed for the past with extreme precision and is less subject to differences (due to quality or brand) than many other goods and services. And as the survey was conducted by
The authors thank Paolo Angelini, Luigi Cannari, Giovanni DAlessio, Andrea Generale, Luigi Guiso, Francesco Lippi, Domenico J. Marchetti, Stefano Siviero, and Daniele Terlizzese for useful comments and Pietro Spataro for help with data processing. They also gratefully acknowledge the co-operation of Cinema Nuovo Sacher and Cinema Quattro Fontane in Rome.
V. Cestari, P. Del Giovane and C. Rossi-Arnaud
submitting a questionnaire to movie-goers we can be sure that the sample consists of people who actually consume that particular service at least occasionally. The changeover period is chosen for the study as a particularly interesting case per se and because it bears on the key question of whether ination perceptions in the post-changeover period are related to the presence of a bias in price recall. As argued in chapters 1 and 3, individual perceptions of rising ination may have been related to an inaccurate memory of pre-euro prices and a difculty in distinguishing the period of time over which the price change has taken place. This, in turn, may have induced consumers to blame the euro also for price rises that came before its introduction and had nothing to do with the new currency. Moreover, although ination perceptions have receded in Italy since 2004 (as shown by the European Commission surveys), we are still confronted with a widespread belief among consumers that the present prices of many products are twice as high, or in any case much higher, than they were before the introduction of the euro. As mentioned in previous chapters, the old prices of many goods and services are commonly believed to have been converted at a rate of 1,000 lire to the euro, rather than at the ofcial rate of almost 2,000. The fact that this widespread belief is based on what consumers claim to be precise recall of the pre-euro prices is a good reason to investigate whether their condence in their memory is well-founded. But we also consider the possibility of a reversed causal link, from the conviction that many prices have doubled to a memory that simply conrms that belief. Little is known about the mechanisms underlying the memory for prices. Although this can be considered as part of the question of everyday memory (Neisser 1982; Cohen 1996), evidence from cognitive studies on this issue is seldom found in psychological research. Rather, it is covered in a broader social sciences literature that includes marketing and consumer behaviour (Dickson and Sawyer 1990; Monroe and Lee 1999). Our analysis ts in with the recent behavioural economic literature, which postulates deviations from the assumption of rational, computationally unconstrained agents. This literature has explored the implications of consumers having limited information-processing capacity and being rationally inattentive (see, among others, Sims 2003, Ameriks et al. 2004, Levy et al. 2006, Reis 2006) or having limited memory (Mullainathan 2002). Dziuda and Mastrobuoni (2005) and Ehrmann (2006) investigate whether imperfect information models can explain the evolution of prices and the mismatch between perceived and actual ination in the aftermath of the euro cash changeover. The study was conducted by giving questionnaires to members of the audience in two well-know cinemas in Rome, on four evenings in May 2006. The rst two questions regarded memory for pre-euro cinema prices, for the
Memory for prices and the euro cash changeover: an analysis for cinema prices in Italy
full and the reduced price respectively. Italian cinemas typically offer reduced prices on weekday afternoons or on one weekday evening. In Rome, where the questionnaire was to be given, a Wednesday reduction was introduced in February 1998 and maintained thereafter, including the changeover period. To control that the answers for the full price were not inuenced by memory of the reduced price and also to test respondents memory using a second price we included separate questions for the two prices. The answers on recalled prices are analysed in the light of those to further questions aimed at elucidating the mechanisms of price recall; measuring the relationship between price memories and such other factors as habits in using the new currency (mental conversion into lire vs. thinking in euro), frequency of movie-going and attitudes toward the euro; and assessing the relationship between the memory of the specic price considered and the perceptions on the average change in the general price level. Lastly, the results on price recall are also related to personal characteristics (age, gender, education and occupation). To investigate the extent to which consumers memory for prices is inuenced by the change of currency, alternative questionnaires were given to two different sub-groups of respondents, asking, respectively, the pre-euro price expressed in euro (rather than in lire as in the standard questionnaire) or the price in 2003.
5.2 The method

5.2.1 The respondents Respondents were randomly sampled among members of the audience in movie theatres; they participated on a voluntary basis. Respondents older than 65 years of age were excluded as they enjoyed a discounted ticket in 2001 and/or at the moment of the study. A total of 225 people lled in one of the questionnaires. The number of respondents for each questionnaire, with gender, age and education, is shown in Table 5.1. The questionnaires were given in two well-know cinemas in Rome2 whose price in 2001 coincided with the mode for all rst-run movie theatres in Rome. We chose these two cinemas also because they ensured good control on the survey conditions (full co-operation by the employees and a suitable area where people could ll out the questionnaires without interference). It is worth noting that these cinemas are attended by an audience characterized by a high level of
2
Cinema Nuovo Sacher and Cinema Quattro Fontane.
Table 5.1. Respondents Questionnaire on Questionnaire on Questionnaire on pre-euro pre-euro 2003 price price in lire price in euro Number of respondents Male Female Gender not indicated Age range Mean age (SD) Lower than high-school degree High-school degree University degree Education not indicated 150 53 88 9 19-65 42 (11.8) 3 36 107 4 40 20 18 2 18-62 40 (12.9) 1 14 24 1 35 10 24 1 18-65 44 (15.7) 1 10 24 0
Notes: Information is provided separately for each of the three questionnaires used (see Sect. 5.2.2).
education (and presumably income) and rather high frequency of moviegoing. The experiment followed a between-subject design, i.e. each group of respondents lled in only one questionnaire. 5.2.2 The questionnaires Our standard questionnaire is reproduced in Appendix 5.A. It comprised 14 questions plus 4 items on personal data (age, gender, education, occupation). The rst two questions regarded memory for cinema prices (for the full and the reduced price respectively) with reference to the period immediately preceding the cash changeover, when prices were still in lire. The other questions were intended rst, to capture the mechanisms through which prices were recovered (e.g. memory vs. calculation); second, to elicit frequency of movie-going, habits in using the new currency (mental conversion into lire vs. thinking in euro) and attitudes towards the euro both before its introduction and currently; third, to record the perceptions on the average change in the general price level. To investigate how the memory of prices is inuenced by the currency changeover, two alternative questionnaires were administered to separate groups of respondents. One was asked to indicate the pre-euro price expressed in euro rather than in lire as in the standard questionnaire, the other the price in 2003. Where necessary the other questions were slightly adjusted. The rationale
was to check whether expressing the old price in euros encouraged consumers to better reect on the difference between the recalled and the present price and on the implicit rate of change, thereby improving the accuracy of the recall. The question on the 2003 price allowed us to verify the correctness of memory with respect to a more recent period that did not coincide with a currency changeover. 5.2.3 The procedure The questionnaires were administered by doctoral students in psychology, under the supervision of the authors, on four evenings in May 2006. Participants were asked to answer the questions in order and without reading ahead. They were given no time limit and asked to complete the questionnaire as accurately as possible without communicating with other participants. The answers on the pre-euro price expressed in euro and on the price in 2003 were compared with those on the pre-euro price expressed in lire by considering, for the latter, only the questionnaires administered in the same movie theatre and on the same evenings. Price memory was compared with the actual prices in effect before the launch of the euro. Since the questionnaire asked for the price in lire in the period immediately preceding the cash changeover we could have simply considered the price at the very end of 2001. However, since several movie theatres in Rome had raised their prices in mid-December 2001, we considered the actual prices both in the rst half and in the second half of December, taking into account that: a) the price increase in mid-December could be interpreted as an anticipated effect of the euro and b) the last price in lire might have not been memorized by the public, as it lasted only two weeks. On 8 December the average cinema price was 12,630 lire, while both the median and the mode were 13,000 lire. On 22 December the mean was slightly higher, the median unchanged and the mode had become 14,000. As to the possibility that the prices in the rst half of December had also been affected by the imminent changeover, analysis of the time series in the previous years shows that that this was not the case, since prices had been at about the same levels in the four years preceding the changeover (see below for more evidence). On this basis we considered as a correct memory for the pre-euro price of the full ticket both 13,000 lire (the mode prevailing until a few weeks before the changeover and in the four previous years except 2000) and 14,000 lire, which was the mode of the very last price in lire. These were the prices charged in the rst and second half of December, respectively, in our two cinemas and in other movie theatres with the same quality standard and so correspond closely to the questionnaire (which asked the price in a rst-run movie theatre similar to this one). The answer 12,000 lire was considered
as close to correct, since this was the price still charged by about 20 per cent of the cinemas at the end of 2001. For the reduced price, the correct answer was 8,000 lire, the price charged by a large majority of the cinemas that offered a discount on Wednesday. At the time of the changeover that reduced price (agreed and advertised by the association of cinemas in Rome) had been unchanged for almost four years, since its introduction in February 1998.
5.3. The results

5.3.1 Documenting the discrepancy between recalled and actual pre-euro prices The main nding is that our samples recalled pre-euro prices are much lower than the actual prices at the end of 2001. Fig. 5.1 compares the percentage
45.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 Distribution of actual pre-euro prices (full ticket) Distribution of prices recalled by respondents (full ticket) 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 45.0 40.0
5,000 6,000 7,000
8,000 9,000 10,000 11,000 12,000 13,000 14,000 14,500
Fig. 5.1. Recalled vs. actual cinema price (full ticket) (1) Source: La Repubblica daily newspaper. Notes: (1) On the horizontal axis, the prices of full cinema tickets (in lire); on the vertical axis, the percentage of respondents indicating the price as their recall (dark histogram) and the percentage of cinemas that charged that price in the rst half of December 2001 (light histogram). The cinemas considered are those located in Rome showing newly released lms.
of respondents recalling each price (dark histogram) with the percentage of cinemas that actually charged that price in the rst half of December 2001 (light histogram). Only 8 per cent of the respondents said either 13,000 or 14,000, which we consider to be the correct answers (see Sect. 5.2.3). A sixth (17 per cent) indicated 12,000, which is close to correct. The majority (56 per cent) indicated prices lower than 10,000 lire, which were non-existent in the months preceding the launch of the euro. Interestingly, the mode of the recalled price (7,000 lire) corresponds to the mode of the prices at the time of the study (7 euro) converted at a rate of 1,000 lire to 1 euro. The ndings obtained for the reduced price are similar (Fig. 5.2). Only a very small fraction of the respondents (8 per cent, as in the case of the
80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0

Distribution of actual pre-euro prices (reduced ticket) Distribution of prices recalled by respondents (reduced ticket)
80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
3,000 3,500 4,000 4,500 5,000 5,500 6,000 6,500 7,000 7,500 8,000 8,500 9,000 9,500 10,000
Fig. 5.2. Recalled vs. actual cinema prices (reduced ticket) (1) Source: La Repubblica daily newspaper. Notes: (1) On the horizontal axis, the prices of reduced cinema tickets (in lire); on the vertical axis, the percentage of respondents indicating the price as their recall (dark histogram) and the percentage of cinemas that charged that price in December 2001 (light histogram). The few cinemas that did not offer a reduced price on Wednesday were excluded from the computation.
full price) recall the actual price (8,000 lire). Around a fth recall a price (7,000 or 7,500) that is not too far off. Most respondents (61 per cent) recall a price (6,000 lire or less) that is much lower than the actual price. The mean of the recalled price is 5,500 lire. As in the case of the full price, the mode of the recalled price (5,000 lire) corresponds to the mode of the actual prices at the time of the study (5 euro) converted at a rate of 1,000 lire to 1 euro. Finally, the prices recalled date back to a period long before the changeover: as shown in Fig. 5.3, the mean and the mode of the recalled prices (full ticket) correspond to the actual prices observed, respectively, in 1990 and 1987, more than ten years before the launch of the euro.
Lire 15,000
14,000 13,000 12,000 11,000 10,000 9,000 8,000 7,000 6,000
7.75
Actual prices
Mode Mean
7.23 6.71 6.20 5.68 5.16 4.65 4.13 3.62 3.10
Actual price corresponding to the mean of the pre-euro prices recalled by respondents
Actual price corresponding to the mode of the pre-euro prices recalled by respondents 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Fig. 5.3. Recalled pre-euro prices compared to actual prices in historical perspective (1) Source: La Repubblica daily newspaper for actual prices. Notes: (1) Actual prices for a full ticket in rst-run movie theatres in Rome (left-hand scale: lire; right-hand scale: euros); the two circles indicate the actual prices corresponding to the mean and the mode of the pre-euro prices recalled by respondents. End-of-year gures (rst half of December for 2001, see Sect. 5.2.3) with the exception of 2006, for which they refer to May 2006, when we carried out the study.
5.3.2 Memory inaccuracy: some insights from the respondents answers Complete descriptive evidence on the relationship between the price recalled by the respondents and their answers to the other questions is given in Table 5.B1 in Appendix 5.B. The results of the statistical analyses for the relation between the recall of full prices and the answers relating to individual memory (reported precision of recall and memory of the changeover year), individual habits concerning mental conversion into lire and frequency of movie-going are given in Table 5.2. The main ndings are:3 no signicant difference between the prices indicated by respondents claiming precise recall and by participants having a vague memory;4 participants who correctly indicate 2002 as the year of the euro cash changeover have more accurate memory than those who get the year wrong;5 the price reported by participants who say that they always think in euro or convert rarely is signicantly higher (and thus more accurate) than that recalled by those who say they convert sometimes, often or always;6 the price recalled by participants who went to the movies twice a month or more is signicantly higher than for participants who attended the cinema once a month or less.
The same tests were conducted for the reduced price, yielding a similar pattern of results. Participants were requested rst to say whether their price assessment was a recall of the old lira price or a mental calculation starting from the present euro price. Only respondents who reported the former (95 per cent) were asked to indicate the precision of their own recall. Almost half the respondents (47 per cent; Table 5.B1) report the correct year. A third (33 per cent) indicate 2001, which is not surprising, considering that in 2001 the media and the public already paid a lot of attention to the impending introduction of euro banknotes and coins. Note that the respondents who have an incorrect memory of the year are implicitly recalling a pre-euro price that goes back to an earlier or (in a few cases) a later period than the end of 2001. We kept this in mind in the analyses below. This relationship, however, is no longer signicant once we consider it in the context of a multivariate analysis (see Sect. 5.3.7).
10
Table 5.2. Price recalled and respondents answers relating to individual memory, habits of mental conversion and frequency of movie-going (1) Reported precision of recall Precise Number of observation Mean (SD) of price recalled T-test Vague
52 80 9,173 (2849) 9,125 (2340) t(130) = 0.10, not signicant Reported year of the euro cash changeover Correct year Mistaken year
Number of observation Mean (SD) of price recalled T-test
68 78 9,419 (2541) 8,564 (2552) t(144) = 1.81, p < 0.10 Habits of mental conversion in lire Answers always think in euro or convert rarely Answers convert sometimes, often or always
72 77 9,417 (2659) 8,591 (2399) t(147) = 1.99, p < 0.05 Frequency of movie-going Twice a month or more Once a month or less
85 65 9,406 (2527) 8,462 (2487) t(148) = 2.29, p < 0.05
Notes: (1) For each group, number of respondents and mean and standard deviation of the price recalled; for each comparison, t-test of difference between means. Number of subjects and degrees of freedom vary depending on the number of missing answers.
5.3.3 Recall of prices and ination perceptions The descriptive statistics (Table 5.B1) suggest a link between the recalled price and individual ination perceptions (measured by the opinion on the rate of change of the general price level), both in the two years following
11
the changeover and in the two years preceding the study.7 However, no statistical test could be conducted on the rst period due to the extremely asymmetrical distribution of the answers (only 12 respondents answered fallen, stayed about the same or risen slightly). The statistical analysis was thus conducted by grouping subjects according to their opinion on ination both in the two years after the changeover and in the two years preceding the study, according to two criteria. a) Responses (fallen, stayed about the same, risen slightly, risen moderately, risen a lot) were assigned scores from 1 to 5 and the scores for the two different periods were summed (e.g. a score of 10 corresponds to risen a lot for both periods). The sum of the two periods thus corresponds to the time horizon for which we tested memory for prices. b) Responses were assigned scores from 1 to 5 and the score for the more recent period was subtracted from that for the two years following the changeover, to capture the persistence of individuals beliefs concerning ination. The scores thus obtained ranged, respectively, from 5 to 10 and from -4 to 3. They were distributed as reported in Table 5.3.
Table 5.3. Perceived increase in the general price level synthetic measures of the respondents opinion (1). Combined scores Sum Number of observations 2 _ -4 1 3 _ -3 1 4 _ -2 4 5 6 7 2 8 13 Difference -1 11 0 67 1 30 8 53 2 29 9 32 3 5 10 40 4 _
Number of observations
Notes: (1) Responses about the past behaviour of prices (fallen, stayed about the same, risen slightly, risen moderately, risen a lot), respectively for the two years following the euro cash changeover and for the two years before the study, were assigned scores from 1 to 5. The sum of the scores for the two periods was calculated and also the difference between the post-changeover period and the more recent two-year period. The table shows the number of observations for each combined score.
Although referred to longer time horizons, the questions were phrased as in the EC survey, with the same list of possible answers. As in the EC survey, respondents were asked to state their views regarding past price developments in qualitative terms, rather than to make a numerical estimate.
12
The statistical analysis (Table 5.4) shows that movie prices recalled by participants who indicate low ination over the entire period are signicantly higher than those reported by participants with a high level of perceived ination, especially when high perceived ination is dened as a combined score of 10 (i.e. risen a lot both in the two years following the changeover and the last two years). And the prices recalled by participants who perceived the same degree of price increase for both periods are signicantly lower (and thus further from the actual price) than for participants who perceived lower ination in the more recent period than in the two years following the changeover. Opinions on ination were also investigated by asking for the degree of agreement with the statement What cost 1,000 lire before the introduction of the euro now costs 1 euro, which is apparently what many Italian consumers believe (implying a price rise of practically 100 per cent). The results do not provide clear indications. They do conrm that this opinion is widely spread, almost half of the respondents (45 per cent) strongly agreeing. And these respondents recall a price that is lower than that recalled, on average, by all the others (Table 5.4). On the other hand, we obtained the unexpected result that the median and the mode of the prices recalled are the same for those who strongly agree with the statement and for the few respondents who strongly disagree (Table 5.B1). Moreover, in multivariate analysis the relationship between the price recalled and the answers to this specic question is not signicant (see Sect. 5.3.7). As to the expected effects of the euro introduction on prices, while the participants who said that they had expected a very strong increase recalled a much lower price, this could not be tested statistically since only 10 subjects belong to this category (Table 5.B1). The vast majority (125 out of 138) report that they had expected no change or a moderate increase and recall very similar prices. The prices recalled are slightly lower for subjects whose opinion on the overall effect of the euro for Italy ranges from very disadvantageous to neutral than for those who believe it to have been rather or very advantageous (Table 5.B1). But the difference is not statistically signicant: t (140) = 0.84. 5.3.4 Price recall and age We divided participants into 3 age groups: age 19-34, 35-50 and 51-65 (Table 5.5). The analysis of variance (ANOVA) showed a signicant effect of age on the accuracy of price recall (F (2,144) = 4.23, p < 0.05). A post hoc test (Scheffe multiple comparison) indicated that respondents aged 51 to 65 recall prices that are signicantly lower (and thus more inaccurate) than the 19-34
13
Table 5.4. Price recalled and ination perceptions (1) Ination perceptions (sum of scores for the entire post-changeover period) Low perceived ination High perceived ination (combined scores from 5 to 8) (combined scores from 9 to 10) Number of observations Mean (SD) of price recalled T-test 76 72 9388 (2524) 8597 (2544) t(146) = 1.90, p < 0.10 Low perceived ination (combined scores from 5 to 9) Number of observations Mean (SD) of price recalled T-test High perceived ination (combined score of 10)
108 40 9310 (2455) 8175 (2671) t(146) = 2.35, p < 0.05 Persistence of ination perceptions (difference between scores for the two periods considered) Unchanged perceived ination (0 rating) Lower perceived ination in the more recent period (ratings from of 1 to 3)
Number of observations Mean (SD) of price recalled T-test
67 64 8493 (2683) 9461 (2493) t(129) = 2.14, p < 0.05 Agreement with the statement What cost 1,000 lire before the introduction of the euro now costs 1 euro Strong agreement All others
Number of observations Mean (SD) of price recalled T-test
68 82 8602 (2622) 9223 (2449) t(148) = 1.73, p < 0.10
Notes: (1) Relationship between price recall and respondents ination perceptions (based on the combined scores reported in Table 5.3) and agreement with the statement 1 euro = 1,000 lire For each group, number of respondents and mean and standard deviation of the price recalled. For each comparison, t-test of difference between means.
14
Table 5.5. Price recalled and age (1) Age 19-34 Number of observations 25 Age mean (SD) 28.84 (4.55) Price recalled mean (SD) 10,080 (2,344) Age 35-50 86 41.50 (4.65) 9,006 (2,504) Age 51-65 34 57.65 (3.16) 8,176 (2,528)
Notes: (1) For each age group, number of respondents, mean and standard deviation of age, mean and standard deviation of price recalled.
age group (p < 0.05), while the difference between the price recall of the intermediate age group and that of the other two groups was not signicant. No signicant differences were found on price recalled according to gender, occupational status and level of education. On education, however, the sample is strongly concentrated at a high level (see Sect. 2.1). 5.3.5 Expressing the recalled price in euro The ndings obtained by giving a questionnaire in which respondents were asked to indicate the pre-euro price in euro rather than in lire are summarised in Table 5.6, where answers expressed in euros are also presented converted in lire to make them comparable with the answers to the standard questionnaire. In doing this we used both the ofcial conversion rate of 1,936.27 lire to the euro and the approximate rate of 2,000 to 1 that Italians use in mental conversions. The answers were compared to the answers expressed in lire by considering, for the latter, only the 46 questionnaires administered in the same movie theatre and on the same evenings. The respondents who were asked to express the pre-euro price in euros had a slightly better recall: the mean recalled price is 4.73 euros, which corresponds to about 9,200 or 9,500 lire depending on the conversion rate, against about 8,900 lire when the price is expressed in lire. But the difference is not statistically signicant (t(84) = 0.53, and t(84) = 1.09, respectively). 5.3.6 Recalling the 2003 price The prices recalled for 2003 are reported in Table 5.6. The analysis of variance (ANOVA) showed that the type of price requested (pre-euro price in lire, pre-euro price in euro, or price in 2003) had a signicant effect on accuracy (measured by the percentage difference between the mean recalled price and the corresponding actual price): F(2,118) = 4.97, p < 0.01. A post hoc test
15
Table 5.6. Types of price recall: pre-euro price expressed in lire or euros; 2003 price (1) Pre-euro price recalled in lire (obs. = 46) mean full median cinema mode ticket SD mean reduced median cinema mode ticket SD 8,870 8,000 7,000 2,475 (obs. = 45) 5,622 5,000 5,000 1,819 3.10 3.00 2.50 1.04 4.73 5.00 5.00 1.23 Pre-euro price recalled in euro Converted in lire 1= 1 = 2,000 1,936.27 (obs. = 41) 9,159 9,681 9,681 2,382 (obs. = 37) 6,002 5,809 4,841 2,014 6,200 6,000 5,000 2,080 3.71 3.50 3.00 0.92 9,460 10,000 10,000 2,460 5.67 6.00 5.00 1.09 Recalled 2003 price Converted in lire 1= 1 = 2,000 1,936.27 (obs. = 35) 10,979 11,618 9,681 2,111 (obs. = 35) 7,184 6,777 5,809 1,781 7,420 7,000 6,000 1,840 11,340 12,000 10,000 2,180
Notes: (1) Answers obtained by giving to three different groups of respondents one of three questionnaires in which respondents were asked to indicate, respectively, the pre-euro price expressed in lire (standard questionnaire), the pre-euro price expressed in euros, or the 2003 price. For comparison, answers expressed in euros are also presented converted in lire, both at the ofcial conversion rate and at the approximate mental conversion rate of 2,000 lire to 1 euro. The answers expressed in euros are compared with those expressed in lire by considering, for the latter, only the 46 questionnaires administered in the same movie theatre and on the same evenings.
(Scheffe multiple comparison) indicated that the memory error for the 2003 price was smaller than for the pre-euro price in lire (p < 0.05) and, though less signicantly, for the pre-euro price in euro (p < 0.10). Although closer to the actual price than for the pre-euro period, the prices recalled for 2003 are still signicantly lower than the actual prices (t (33) = 5.93, p < 0.01). The mode of the prices recalled is 27 per cent lower than the actual price. The respondents memory for 2003 implies a perceived rate of increase in cinema prices of about 8 per cent on an annual basis in the 2003-06 period, while in reality prices were virtually unchanged. This distortion was somewhat less than the corresponding bias for the previous part of the sample period (from the changeover to 2003), for which respondents
16
recall of pre-euro prices implied an upward bias of about 11 per cent a year (the difference between a perceived implicit rate of increase of 13 per cent and an actual rate of about 2 per cent between December 2001 and mid-2003).8 5.3.7 Regression analysis As a synthesis of the relationships between the price recalled and the other variables, and to test which variables remain signicant when they are considered simultaneously, Table 5.7 reports the results of linear regressions of the pre-euro price recalled for the full cinema ticket against a set of variables derived from the other questions. Given that the pre-euro price recalled did not differ signicantly with the currency in which it was expressed, we considered both the answers to the standard questionnaire (pre-euro price recalled in lire) and those to the alternative questionnaire (pre-euro price recalled in euros). In any case, a dummy variable to control for the possible inuence of the currency was also included.9 All the coefcients have the expected signs. Among the variables that were
We divided the period between the changeover and May 2006, when we carried out the study, into two sub-periods: from the launch to mid-2003 and from mid-2003 to May 2006. We considered mid-2003 rather the end of the year because respondents were requested to indicate generically the price in 2003. In estimating the implicit rates of price increase we kept in mind that some respondents did not correctly remember the year of the euro cash changeover (see Sect. 5.3.2), so in annualizing we considered time horizons of varying length, according to the year recalled. The estimate of the implicit rate in the latter period is based on a comparison between the price recalled for 2003 and the actual price observed in May 2006; if respondents had an imperfect knowledge of the current price the rate of increase implicit in their perception would differ correspondingly from our estimate (respondents were not asked to indicate the current price, since it was posted at the box ofce and easily readable). Each variable was assigned a value from 1 and 5 (or 6) according to the sequence of possible answers to the respective question, as reported in Table 5.B1. The exceptions are the regressors age, included as collected in the questionnaire, and recall for the year of the euro launch, which takes value 0 or 1 depending on whether the answer is wrong or right (assigning to this variable a score that depends on the difference between the answer and 2002 yields similar results). A dummy variable is included for the currency in which prices are recalled (lire or euros).
17
Table 5.7. Summary regressions - dependent variable: recalled price for cinema tickets before the euro cash changeover (1) Recall for the year of the euro launch Frequency of cinema attendance Mental conversion (2) (3) (4) (5) (6)
0.11 0.11 0.10 0.10 0.11 t = 1.38 t = 1.40 t = 1.25 t = 1.26 t = 1.36 p = 0.17 p = 0.16 p = 0.21 p = 0.21 p = 0.18 0.14* 0.14 * 0.14 * 0.16 ** 0.15 * 0.17** t = 1.74 t = 1.82 t = 1.83 t = 1.98 t = 1.90 t = 2.29 p = 0.08 p = 0.07 p = 0.07 p = 0.05 p = 0.06 p = 0.02 -0.11 -0.11 -0.09 -0.09 -0.06 t = -1.44 t = -1.42 t = -1.22 t = -1.18 t = -0.71 p = 0.15 p = 0.16 p = 0.23 p = 0.24 p = 0.50 -0.00 -0.02 -0.02 -0.02 -0.03 t = -0.01 t = -0.21 t = -0.22 t = -0.31 t = -0.41 p = 0.99 p = 0.84 p = 0.83 p = 0.76 p = 0.68 -0.14 * t = -1.81 p = 0.07 -0.15 ** -0.14 * -0.14 * -0.17** t = -2.04 t = -1.71 t = -1.79 t = -2.38 p = 0.04 p = 0.09 p = 0.08 p = 0.02
Currency of response (lire/ euro) Ination perception (2 years following the changeover) Ination perception (2 years following the changeover + last 2 years) Conversion 1000 lire = 1 euro Expectations
-0.003 -0.03 t = -0.04 t = -0.33 p = 0.97 p = 0.75 -0.09 -0.08 t = -1.16 t = -0.99 p = 0.25 p = 0.32 0.06 0.10 t = 0.81 t = 1.22 p = 0.42 p = 0.22 -0.20 ** -0.19*** t = -2.51 t = -2.60 p = 0.01 p = 0.01
General opinion on euro adoption Age
18
contd Table 5.7 R-squared Adjusted R-squared 0.05 0.03 0.07 0.04 0.08 0.05 0.09 0.04 0.12 0.07 0.10 0.08
Notes: (1) Regression of recalled pre-euro cinema prices on other variables derived from the answers to the questionnaire, each variable being assigned a value from 1 to 5 (or 6) according to the sequence of possible answers, as reported in Table 5.B1. The only exceptions are the regressors age, included as collected in the questionnaire, and recall of the year of the euro launch, which takes 0 for a wrong answer and 1 for a right answer (results assigning a score that depends on the difference between the answer and 2002 are similar). A dummy variable is included for the currency in which prices are recalled (lire or euros). *, ** and *** indicate statistical signicance at respectively 0.10, 0.05 and 0.01 levels.
signicantly related to the correctness of recall according to the bivariate analysis (Sect. 5.3.2), the coefcients of frequency of movie-going and of perceived ination considering either the two years following the changeover or the combined score variable that sums the responses for the two periods are both signicantly different from zero. The coefcient of age is also signicantly different from zero, while the coefcient of agreement with the statement 1 euro = 1,000 lire and of the habits on conversion are not signicantly different from zero. For the latter variable, this probably reects a correlation with age. Older respondents, whose recall is less accurate, are more inclined to mental conversion than younger participants (a signicant correlation, r (145) = 0.20, p < 0.05, was found between age and reported habits on mental conversion). The coefcient for the euro/lire dummy variable is not signicantly different from zero, conrming the results reported in section 5.3.5.10
10
As a robustness test, and to capture possible nonlinear relationships, we also ran the regression with an alternative coding of the answers, i.e. 5 (or 6) dummy variables for each question, corresponding to the alternative answers to that question. The results were in line with those reported in Table 5.7: the coefcients were signicantly different from zero only for some of the answers relating to frequency of movie-going, ination perceptions and age.
19
5.4. Discussion
The main results described in section 5.3 show that most participants had highly inaccurate memory of pre-euro prices in lire. An overwhelming majority indicated prices for the full ticket that were non-existent or in effect in just a few cinemas in the months preceding the launch of the euro, while very few respondents reported prices that can be deemed correct. The prices recalled date to a period long before the changeover as far back as 1987, in fact. The results for the reduced price were similar, which is particularly interesting considering that this price had been unvaried for almost four years at the moment of the changeover. Overall these results suggest that the consumers common claim that they can properly compare present and pre-euro prices is likely to be less than well-founded. This conclusion is strengthened by our nding that the accuracy of price recall is no better among respondents who claim a precise recall than among those who claim only a vague memory. Many consumers, one could conclude, cannot judge the actual precision of their price recall and are overcondent about its quality. The distorted memory of the pre-euro price cannot apparently be ascribed simply to the length of time that has passed, since memory for prices in 2003 is also quite poor. This suggests that the difculty in retrieving prices from the past is not specic to the changeover period, although in this case the difference between memory and reality is somewhat smaller than for the pre-euro price. The results from the alternative questionnaire indicate that asking respondents to express the pre-euro price in euros rather than in lire was not enough to stimulate them to compare their answer to the present price and reect better on the variation implied by the recalled price. The great discrepancy between the cinema prices recalled and the prices actually in effect before the launch of the euro suggests, prima facie, that poor memory could be a relevant factor in explaining individual ination perceptions. In our study, the recall for the pre-euro period would imply a 50 to 90 per cent rise in cinema prices (depending on whether we consider the mean or the mode) over four and a half years, as against the actual increase of about 5 per cent. There is a relationship between the price recalled and the general ination perception (the individuals opinion on the rate of change of the aggregate price level): respondents who perceive higher and more persistent ination had on average a more inaccurate recall of pre-euro cinema prices. However, these results must be interpreted with caution. In the rst place, the study was conducted in 2006, while the leap in ination perceptions, in Italy as in other euro-area countries, came in the two years immediately following the changeover. Although we asked separate questions for the
20
post-changeover period and for the more recent period, we are aware that retrospective answers require very careful interpretation. Moreover, it would be hazardous to draw general conclusions on the relation between the individual perception about ination (i.e. the rate of change in the aggregate price level) and individual price memory from the recalled price of a single item. More evidence is needed, and we plan to conduct further analysis on a range of products and different time spans. An interesting nding is that the mode of the price recalled 7,000 lire for the full price and 5,000 lire for the reduced price corresponds to the mode of the prices prevailing when we conducted the study (7 and 5 euros, respectively) converted at a rate of 1,000 lire. This result suggests the possibility of a back-causation from the fairly widespread conviction that most prices have been converted at a rate of 1,000 lire to 1 euro to the recall of a price in lire that corresponds to that belief. Yet, the answers to the question that investigates this belief directly asking whether the respondent agrees with the statement What cost 1,000 lire before the introduction of the euro now costs 1 euro do not provide a clear indication in this regard. They do show that this opinion is widely shared, and respondents who strongly agree with the statement recall a price that is lower than that recalled, on average, by the other participants. But in multivariate analysis the correlation between the price recalled and the answer to this specic question is not signicant. Moreover, unexpectedly, the mode of the price recalled is the same (7,000 lire) for those who strongly agree with the statement and for the few respondents who strongly disagree. While this recall of the pre-euro price though completely wrong is perfectly consistent with the belief of the former group, it is difcult indeed to reconcile with the belief of the latter. This inconsistency might be explained to some extent by the fact that we asked participants to express their agreement with a general statement while the memory for prices is measured for the price of cinema tickets only. In principle, we cannot exclude that some respondents considered the statement to be false in general, but with some exceptions including the price of cinema. However, this result might also suggest difculty for some respondents in relating the memory for single prices and the assessment on the rate of change of these prices, or more generally of the rate of change of the overall price level. Another relevant result is that the accuracy of price recall is signicantly affected by the frequency of cinema attendance. This suggests that the more frequently they buy items, the better is consumers memory for prices. This in turn is consistent with the empirical evidence that ination perceptions are affected disproportionately by the price rises of the goods that are purchased most frequently (see Chap. 1). The nding of no signicant relationship between prices recalled and attitudes toward the euro (considering both the expectations for prices and
21
the present opinion on the overall effect of the euro for Italy) suggests that these factors did not count much in orienting the memory of prices.11 The relationship between age and price recall is consistent with the existing literature, showing a progressive age-related decline in various systems of memory (Craik et al. 1990, Cockburn and Smith 1991, Baddeley 1999). In this regard, an interesting study is Aalto-Setala and Raijas (2003), showing a decline in the ability to estimate prices after the euro changeover among Finnish consumers aged above fty years. Finally, while explaining such very bad price memory is beyond the scope of this paper, some possible explanations, in addition to the specic factors considered so far, can be usefully discussed on the basis of the cognitive literature on memory.12 One possible explanation of incorrect recall is that the price was not encoded (i.e. stored in memory for later retrieval) in the rst place. Price knowledge surveys conducted in stores show a surprisingly low level of price knowledge. Dickson and Sawyer (1990) found that only 56 per cent of the respondents knew the price of an item they had just placed in their shopping cart. Alternatively, the price could have been encoded but no longer present in the subjects long-term memory. The trace could have been removed by proactive inhibition (earlier memories supplanting more recent ones; Underwood 1957) or retroactive interference (forgetting old information due to new information; McGeoch and McDonald 1931, Slamecka 1960). Due to price variation both in time and across locations and sales points, interference is clearly at work on most prices. In our case of cinema tickets, this effect should not be large, as prices were quite stable both in the years preceding the changeover and in the following years and also quite uniform (on 8 December 2001 only four different prices were charged by at least 10 per cent of movie theatres; Fig. 5.1). However, as noted above, the mode of the recalled prices in lire and that of the prices in euro prevailing at the
11
12
Naturally, participants accounts of the expectations they had had before the launch of the euro is subject to the standard objections to all retrospective answers. However, the answers in this study are as a whole consistent with the results of the consumer surveys conducted before the changeover (European Commission 2001), according to which the expectation of strong price increases was less common in Italy than in other euro-area countries, in particular Germany (the country considered by Traut-Mattausch et al. 2004). This discussion takes into account the interdisciplinary character of the study and the wide interest in the questions it treats. We have therefore tried to minimize or explain specialized language, so as to make this section accessible to both economists and psychologists and, as far as possible, to laymen.
22
time of the study have a feature in common, namely the prominence of the number 7 for the full ticket and of the number 5 for the reduced one. Therefore, we cannot rule out retroactive interference of the price of the cinema ticket in euro at the time of the survey (7 and 5 euros) with the price recalled (7,000 and 5,000 lire). A third possibility is that the price is still stored in long-term memory but for some reason cannot be retrieved. A lack of contextual information might explain the difculty in retrieving the correct price (Tulving 1983), although, following Vanhuele and Drze (2002), the test was conducted inside movie theatres in order to maximise contextual cues. Lastly, given that prices have different types of representation in memory (Dehaene 1992 classies them as auditory verbal code, visual Arabic code, and approximate quantities on a dimension termed number line), one could argue that price recall was inuenced by the way in which we have measured price knowledge (see Monroe, Powell and Choudhury 1986). However, most studies investigating memory for price information have focused on recall and this is the main reason why we chose this for the present study. Recall was used instead of recognition also to avoid providing respondents with any information regarding the magnitude of prices in the period considered. Further research, possibly including experimental work, will help elucidate which of the above memory mechanisms may best explain our results.
5.5. Summary and conclusions

This study investigated the memory for pre-euro cinema prices, submitting questionnaires to members of the public in movie theatres in Rome. The results show that the prices recalled are much lower than the actual prices that were in effect before the euro cash changeover. In the case of the full ticket, the mode and the mean are 7,000 and 9,000 lire, respectively, compared to the actual price in December 2001 of 13,000 lire at most theatres in Rome. The prices recalled date to a period long before the changeover, indeed as early as 1987. Similar results are obtained for the price of the reduced ticket offered on Wednesdays by most theatres: the mode of the price recalled is 5,000 lire, compared to the actual price of 8,000 lire, which is particularly striking considering that at the time of the changeover this price had been unchanged for almost four years. Overall, the results suggest that people have highly distorted memories for prices and that the claim of many consumers that they can properly compare present prices with those in being before the introduction of the euro is likely to be less well-founded than they may believe. This conclusion is strengthened by our nding that the quality of price recall is not better for respondents claiming precise recall than for those admitting to having
23
a vague memory, which suggests that many respondents are unaware of the real degree of accuracy of their price recall and overcondent about its quality. It appears that the distorted memory for the price in 2001 cannot be ascribed simply to amount of time that has elapsed, since memory for the 2003 prices also appears to be quite poor. The results on the 2003 price suggest that the respondents difculty in retrieving prices is not specic to the changeover period, although the difference between the price recalled and the actual price is smaller than for the pre-euro price. A signicant relationship is found between the price recalled and subjects individual ination perceptions, measured by the assessments of the rate of change of the general price level in the two years following the changeover and in the two years preceding the study. Price recall is less accurate for those who perceive higher and more persistent ination. An interesting result is that the mode of the pre-euro price recalled for both full and reduced tickets corresponds to the actual prices at the time of the study (7 and 5 euros respectively) converted at a rate of 1,000 lire to 1 euro (compared to the ofcial conversion rate of almost 2,000 lire). While this suggests the possibility of a back-causation from the belief that most prices were converted at that rate to a consequent recall, the answers to the question specically investigating that belief do not provide clear indications in this regard. Finally, price recall is more inaccurate for older respondents, and improves with the frequency of movie-going. The latter result is consistent with the hypothesis that consumers have a better memory for prices the more frequently they buy an item and with the empirical evidence that ination perceptions are disproportionately affected by the prices of the products that are bought most frequently. All in all, this chapter provides a neat case study. A caveat, however, is essential. While the choice of the item and the size and composition of the sample are appropriate for the objective of the study, excessive generalization in interpreting the results is unwarranted. In particular, although we nd a relationship between the inaccuracy of price recall and individual opinions on ination, we do not claim that this alone explains the Italian consumers perception of a surge in ination following the cash changeover, because of the long time between that event and the study and the fact that only a single price is considered. Nor do we believe that our results imply that the perception of a process of impoverishment on the part of some categories of consumers is due only to inaccurate memory for past prices or other psychological factors, as the perception may also reect real factors affecting ones economic situation, not necessarily related to consumer price ination but hard to distinguish from it (see chapters 1 and 6 and Boeri and Brandolini 2004).
24
Appendix 5.A: Standard Questionnaire (English translation)

1) Could you tell us what was the price in Lire for the full cinema ticket (in a movie theatre showing newly released movies similar to this one) in the period immediately preceding the euro cash changeover? Price in Lire:.. 2) What was the price for the reduced ticket in that period (Wednesday evening)? Price in Lire:.. 3) Your answers are yielded from: A recall of the old price in lire A mental calculation based upon the present price in euro Other (please specify).
4) If your answers were based upon a recall of the price in lire, please specify which type of recall: A precise recall of the price A precise recall linked to an action (for instance taking the money from your wallet) A precise recall linked to a specic event A vague memory Other (please specify) .
5) Roughly how frequently did you go to the movies in the period preceding the euro cash changeover? Less than once in 2 months Once in 2 months Once a month Twice or three times a month Once a week Twice or more a week 6) How frequently do you go to the movies now? Less than once in 2 months
25
Once in 2 months Once a month Twice or three times a month Once a week Twice or more a week
7) If the frequency with which you go to the movies has increased/ decreased between the two periods, what is this change due to? (do not answer unless frequency has changed) Changes in my personal/family situation Changes in my nancial position Changes in the cinema ticket price Alternatives to going to the movies (DVD, Pay-TV) Other (please specify) .
8) Do you remember at the beginning of which year the euro cash changeover took place? . 9) In your opinion, in the 2 years that followed the introduction of the euro banknotes and coins, prices of goods consumed by Italian families have on average: Fallen Stayed about the same Risen slightly Risen moderately Risen a lot Dont know
10) In your opinion, in the last 2 years, prices of goods consumed by Italian families have on average: Fallen Stayed about the same Risen slightly Risen moderately Risen a lot Dont know
26
11) In your daily shopping how do you do to estimate how much you spend? I always think in euros I rarely convert into lire I sometimes convert into lire I often convert into lire I always convert into lire Dont know
12) Before the euro cash changeover, you expected that the euro would cause: A sharp decrease in prices A moderate decrease in prices Neither an increase nor a decrease in prices A moderate increase in prices A sharp increase in prices Dont know
13) In your opinion, overall for Italy the adoption of the euro has been: Very disadvantageous Rather disadvantageous Neither disadvantageous nor advantageous Rather advantageous Very advantageous Dont know
14) What is your degree of agreement with the statement What cost 1,000 lire before the introduction of the euro now costs 1 euro : Strongly disagree Partially disagree Neither agree nor disagree Partially agree Strongly agree Dont know
Questions on gender, age, education and occupation (not included here) completed the questionnaire
27
Appendix 5.B: Descriptive statistics

Table 5.B1. Descriptive statistics on price recall and related factors Reported precision of recall Precise Precise, Precise, linked to linked to specic event action (obs.=36) (obs.=13) price recalled (full) mean 9,500 7,692 median 9,000 8,000 mode 7,000 8,000 price recalled (reduced) mean 5,819 4,538 median 5,000 4,000 mode 5,000 3,000 1999 2000 (obs.=3) 11,667 11,000 10,000 5,000 5,000 5,000 2001 2002 Vague Dont know Total
(obs.=80) (obs.=3) (obs.=135) 9,125 8,000 7,000 5,663 5,000 5,000 2003 7,000 7,000 7,000 5,000 5,000 4,000 2004 9,096 8,000 7,000 5,567 5,000 5,000 Total
Reported year of the euro cash changeover (obs.=2) (obs.=20)(obs.=48)(obs.=68) (obs.=7) (obs.=1) (obs.=146) price recalled (full) mean 6,500 8,700 8,917 9,419 7,571 7,000 9,010 median 6,500 8,000 8,000 9,000 7,000 7,000 8,000 mode 6,000 7,000 7,000 7,000 7,000 7,000 7,000 price recalled (reduced) mean 4,000 5,225 5,365 5799 5,429 4,000 5,521 median 4,000 5,000 5,000 5,000 5,000 4,000 5,000 mode 3,000 5,000 5,000 5,000 5,000 4,000 5,000 Habits of mental conversion to lire Always Convert Convert Sometimes Convert Convert think in rarely often always euro (obs.=50)(obs.=22) price recalled (full) mean 9,360 9,545 median 8,500 10,000 mode 7,000 7,000 price recalled (reduced) mean 5,740 5,932 median 5,000 5,500 mode 5,000 5,000 (obs.=38) 8,526 8,000 7,000 5,184 5,000 5,000 Total
(obs.=32) (obs.=7) (obs.=149) 8,703 8,000 7,000 5,452 5,000 5,000 8,429 7,000 7,000 4,857 5,000 4,000 8,990 8,000 7,000 5,524 5,000 5,000
28
contd Table 5.B1 Frequency of movie-going before the euro launch Less Once in 2 Once a than months month once in 2 months 2-3 times a month Once a week Twice or more a week Total
(obs.=15) (obs.=15) (obs.=35)(obs.=54) (obs.=25) (obs.=6) (obs.=150) price recalled (full) mean 8,667 8,200 8,486 9,176 9,440 11,333 8,997 median 7,000 7,000 8,000 8,000 10,000 12,500 8,000 mode 7,000 7,000 7,000 7,000 10,000 14,000 7,000 price recalled (reduced) mean 5,267 5,214 5,329 5,620 5,700 6,667 5,534 median 5,000 5,000 5,000 5,000 5,000 7,000 5,000 mode 3,000 5,000 5,000 5,000 5,000 5,000 5,000 Perceived change in prices in the two years following the changeover Price Fallen Stayed about the same Risen slightly Risen Risen a moderately lot Total
(obs.=1) (obs.=3) price recalled (full) mean 12,000 10,667 median 12,000 10,000 mode 12,000 10,000 price recalled (reduced) mean 6,000 8,000 median 6,000 7,000 mode 6,000 7,000 Fallen Stayed about the same
(obs.=8) 9,500 10,000 10,000 5,688 5,500 5,000 Risen slightly
(obs.=46) (obs.=92)(obs.=150) 8,957 8,000 8,000 5,344 5,000 5,000 8,886 8,000 7,000 5,527 5,000 5,000 8,997 8,000 7,000 5,534 5,000 5,000 Total
Perceived change in prices in the two years before the study Risen Risen a moderately lot
(obs.=0) (obs.=11) price recalled (full) mean 9,091 median 8,000 mode 7,000 price recalled (reduced) mean 5,318 median 5,000 mode 5,000
(obs.=33) 9,682 10,000 12,000 5,939 7,000 7,000
(obs.=49) (obs.=55)(obs.=148) 9,143 8,000 7,000 5,646 5,000 5,000 8,455 7,500 7,000 5,164 5,000 5,000 9,003 8,000 7,000 5,507 5,000 5,000
29
contd Table 5.B1 Degree of agreement with the statement 1,000 lire equal to 1 euro Strongly disagree (obs.=22) price recalled (full) mean 8,841 median 8,000 mode 7,000 price recalled (reduced) mean 5,114 median 5,000 mode 5,000 Partially Neither agree Partially Strongly disagree nor disagree agree agree (obs.=8) 9,125 9,000 7,000 5,875 5,500 4,000 (obs.=5) 11,600 12,000 12,000 7,600 7,000 7,000 Total
(obs.=47) (obs.=68) (obs.=150) 9,340 10,000 7,000 5,745 5,000 5,000 8,603 8,000 7,000 5,328 5,000 5,000 8,997 8,000 7,000 5,534 5,000 5,000 Total
Expected effect on prices of the euro introduction Sharp Moderate Neither Moderate Sharp decrease decrease increase nor increase increase decrease (obs.=1) price recalled (full) mean 14,000 median 14,000 mode 8,000 price recalled (reduced) mean 8,000 median 8,000 mode 8,000 (obs.=2) 9,500 9,500 7,000 6,000 6,000 4,000 (obs.=80) 9,094 8,000 7,000 5,722 5,000 5,000
(obs.=45) (obs.=10) (obs.=138) 9,167 8,000 7,000 5,300 5,000 5,000 7,850 7,250 7,000 4,900 5,000 5,000 Very advant. 9,069 8,000 7,000 5,544 5,000 5,000 Total
Opinion on the overall effect of the euro adoption for Italy Very Rather Neither Rather disadvant. disadvant. disadvant. nor advant. advant. (obs.=18) (obs.=39) price recalled (full) mean 8,722 8,962 median 8,000 8,000 mode 7,000 7,000 price recalled (reduced) mean 5,639 5,303 median 5,000 5,000 mode 5,000 5,000 (obs.=17) 8,824 8,000 8,000 5,265 5,000 5,000
(obs.=46) (obs.=22) (obs.=142) 9,348 8,000 7,000 5,902 5,000 5,000 9,000 8,000 7,000 5,591 5,500 5,000 9,046 8,000 7,000 5,582 5,000 5,000
30
References
Aalto-Setala V., Raijas A. (2003) Consumer price knowledge before and after the euro changeover, International Journal of Consumer Studies, 27, 210-217. Ameriks J., Caplin A., Leahy J. (2004) The absent-minded consumer, NBER Working Papers, 10216, January. Aucremanne L., Collin M. (2005) Is there a discrepancy between measured and perceived ination in the euro area since the euro cash changeover?, paper presented at the OECD Seminar: Ination Measures: Too High Too Low Internationally Comparable?, Paris, 21-22 June. Baddeley A. D. (1999) Essentials of human memory, Psychology Press, Hove. Boeri T., Brandolini A. (2004) The age of discontent: Italian households at the beginning of the decade, Giornale degli Economisti e Annali di Economia, 63, 449-487. Cockburn J., Smith P.T. (1991) The relative inuence of intelligence and age on everyday memory, Journal of Gerontology: Psychological Sciences, 46, 3136. Cohen G. (1996) Memory in the Real World, Psychology Press, Hove. Craik F.I.M., Morris R.G., Gick M.L. (1990) Adult age differences in working memory, in Vallar G., Shallice T. (eds.) Neuropsychological impairments of short-term memory, Cambridge University Press, Cambridge. Dehaene S. (1992) Varieties of numerical abilities, Cognition, 44, 1-42. Dickson P.R., Sawyer A.G. (1990) The price knowledge and search of supermarket shoppers, Journal of Marketing, 54, 42-53. Dziuda, W., Mastrobuoni, G. (2005) The euro changeover and its effects on price transparency, and ination. Mission euro, mission accomplished!, Princeton University. Ehrmann M. (2006) Rational inattention, ination developments and perceptions after the euro cash changeover, European Central Bank Working Paper Series, 588, February. European Central Bank (2003a) Effects of the introduction of the euro banknotes and coins on consumer prices, Annual Report 2002, 40-42. European Central Bank (2003b) Recent developments in euro area ination perceptions, Monthly Bulletin, October, 24-25. European Commission (2001) Euro Attitudes (wave 6) Euro Zone, Flash Eurobarometer, 115. Kahneman D., Tversky A. (1979) Prospect theory: an analysis of decision under risk, Econometrica, 47, 263-291. Levy D., Chen H., Ray S., Bergen M. (2006) Asymmetric price adjustment in the small, Bar-Ilan University Economics Working Paper Series, November. Marques J.F., Dehaene S. (2004) Developing intuition for prices in euros: Rescaling or relearning prices?, Journal of Experimental Psychology: Applied, 10, 148155. Mastrobuoni G. (2004) The effects of the euro-conversion on prices and price perceptions, CEPS Working Paper, 101, September. McGeoch J.A., McDonald W.T. (1931) Meaningful relation and retroactive inhibition, American Journal of Psychology, 43, 579-588.
31
Monroe K.B., Lee A.Y. (1999) Remembering versus knowing: Issues in buyers processing of price information, Journal of the Academy of Marketing Science, 27, 207-255. Monroe, K.B., Powell P., Choudhury P.K. (1986) Recall versus recognition as a measure of price awareness, in Lutz R. (ed) Advances in Consumer Research, vol. 13. Mullainathan S. (2002) A memory-based model of bounded rationality, The Quarterly Journal of Economics, CXVII, 3, 735-774. Neisser U. (1982) (ed.) Memory observed: remembering in natural contexts, San Francisco, CA, Freeman. Reis R. (2006) Inattentive Consumers, Journal of Monetary Economics, 53, 17611800. Sims, C.A. (2003) Implications of rational inattention, Journal of Monetary Economics, 50, 665-690. Slamecka N.J. (1960) Retroactive inhibition of connected discourse as a function of practice level, Journal of Experimental Psychology, 59, 104-108. Traut-Mattausch E., Schulz-Hardt S., Greitemeyer T., Frey D. (2004) Expectancy conrmation in spite of disconrming evidence: The case of price increases due to the introduction of the Euro, European Journal of Social Psychology, 34, 739-760. Tulving E. (1983) Elements of episodic memory, Oxford: Oxford University Press. Tversky A., Kahneman D. (1974) Judgment under uncertainty: heuristics and biases, Science, 185, 1124-1131. Underwood B.J. (1957) Interference and forgetting, Psychological review, 64, 4960. Vanhuele M., Drze X. (2002) Measuring the price knowledge shoppers bring to the store, Journal of Marketing, 66, 72-85.
[CHAPTER 6 TO BE INCLUDED]
7. Impact of the cash changeover on prices and ination perceptions in the euro area: a survey of the empirical evidence
Paolo Del Giovane and Roberto Sabbatini Bank of Italy, Department of Economic Outlook and Monetary Policy Studies
7.1 Introduction 1
This chapter reviews the studies for other euro-area countries on the issues treated in the six preceding chapters for Italy. Like them, it focuses on two issues. The rst, extensively discussed in section 7.2, is the wide post-changeover disparity between consumers ination perceptions and ofcially measured ination. The contributions on this theme are differentiated and surveyed according to their method of inquiry. 2 One approach hypothesizes that perceptions were inuenced disproportionately by the price movements of specic subsets of goods and services included in the index, especially those that consumers purchase most frequently. A second approach posits that in concomitance with the changeover perceptions were affected by modications in the distribution of prices and price changes, above all by the proportion of products whose price changed, the number and the size of the increases compared with the decreases, the exceptionally large increases for certain products and the number of different prices for the same type of product. A third approach relies on experimental psychology and probes more deeply into the mechanisms by which individuals price estimates are formed and on the role of memory. Finally, more recent contributions consider a combination of factors and assess their relative contributions in a unied analytical framework; a few of them also refer to quantitative measures of ination perceptions, as in chapter 6 for Italy. The second issue, discussed in section 7.3, concerns the documentation of the changeovers effects on actually measured prices and the factors that could account for these price movements. The studies available for the other euro-area countries differ in their datasets and methodologies. Most of
1
The authors thank the members of the Eurosystem/ESCB Communications Committee and the other collegues of the Eurosystem national central banks who provided them with useful information on the studies available for their countries. An overview of the various factors that may have affected ination perceptions in the euro area is presented in European Central Bank (2007).
them use information gathered by the national statistical institutes either to examine the mechanics of the shift from price systems in national currencies to price systems in euros, focusing on the rounding of prices to attractive thresholds, or to identify the components of the price index that registered anomalous changes during or immediately after the changeover and test for a connection between these changes and the shift to the euro. A few studies use data on the prices of specic goods or services taken from non-ofcial sources or adopt indirect methodologies. For some of the aspects examined for Italy in the preceding chapters, we failed to turn up analogous studies for other euro-area countries (though we may have missed relevant works especially if they have not been translated into English). These aspects concern the possible relation between the rise in ination perceptions and media coverage (Chap. 1), the utilization of indirect methods of inquiry based on the behaviour of monetary variables linked to prices (Chap. 4), the role of memory for prices (Chap. 5), and that played by households income and nancial situation (Chap. 6). Nor did we nd, for other euro-area countries, studies tracing to a combination of factors that were based on consumer surveys as detailed and comprehensive as the one used in chapter 6.
7.2 The disparity between perceived and ofcially measured ination

As shown in chapter 1, perceived and ofcially measured ination diverged sharply in most of the euro area following the cash changeover, albeit with an intensity and persistence that varied from country to country. Economic analyses of the causes of the phenomenon have turned chiey on two hypotheses. One is that ination perceptions were more heavily inuenced by changes in the prices of a few specic items, in particular the most frequently purchased products (such as food), that recorded the largest price increases in the changeover period (Table 7.1, column a). The other is that ination perceptions were more strongly affected by price increases than decreases and reected exceptional increases to a disproportionate extent, in which case modications in the distribution of price changes may have contributed to the disparity (column b). The results of studies in experimental psychology on the mechanisms by which price estimates were made in connection with the introduction of euro (column c) appear consistent with both of the above hypotheses.3 Finally, a few works have used consumers surveys to investigate the relative role played by a combination of explanatory variables, in some cases with regard not only to qualitative but also to quantitative measures of the phenomenon (column d).
3
This review of the psychological literature does not claim to be exhaustive, nor does it intend to enter into specialist discussion of the issues studied in the individual works. We only report some ndings particularly relevant to the matters treated in this volume.
A survey of the empirical evidence for the euro area
Table 7.1. Studies on the disparity between perceived and ofcially measured ination Price Number, size Psychological General studies developments of and distribution and memory and analyses factors on quantitative specic prices of price changes perceptions (a) (b) (c) (d)
Euro area European Central Bank 2003b Aucremanne, Collin and Stragier 2007 Pollan 2002 Lindn 2006 Aucremanne, Collin and Stragier 2007 European Central Bank 2007
Austria
Kamleitner, Kirchler Fluch and Stix 2005 and Hofman 2004 Stix 2006 Marques and Dehaene 2004 -
Belgium
Banque Nationale de Belgique 2002 Cornille 2003 -
Finland France Germany
Aalto-Setala 2006 Kurri 2006 INSEE 2003 Crdit Agricole 2004
Aalto-Setala 2006 -
Brachinger 2006 Deutsche Traut-Mattausch et Hoffman et al. 2006 Bundesbank 2004, al. 2004 Brachinger 2006, Brachinger 2006 Hoffman et al. 2006 Del Giovane and Sabbatini 2006 (Chap.1 of this book) -
Greece Ireland Italy
Del Giovane, Fabiani and Sabbatini 2007 (Chap. 6 of this book) Malgarini 2007
Del Giovane and Del Giovane and Sabbatini 2006 Sabbatini 2006 (Chap.1 of this Cestari, Del Giovane book) and Rossi-Arnaud 2007 Del Giovane, Fabiani and Sabbatini 2007 (Chaps. 1, 5 and 6 of this book) Van Raaij and Van Rijen 2003 Kooreman, Faber and Hofmans 2004 Marques and Dehaene 2004 -
Luxembourg Netherlands
De Nederlandsche Bank 2002b Walschots 2002 Buiten 2003 Santos et al. 2002 Banco de Espaa 2003 Alvarez Gonzlez et al. 2004
Portugal Spain
7.2.1
Price developments of specic items
A rst hypothesis considered by studies of the disparity between perceived and measured ination is that consumers may have been disproportionately inuenced by changes in the prices of the goods and services they buy most frequently. As we saw in chapter 1, this hypothesis is plausible in general and even more so for a period like the changeover, when consumers had to become quickly familiar with a host of new prices. We also saw that in Italy these products recorded larger increases in the two years 2002-03 than those purchased less frequently, so that under the frequency-of-purchase hypothesis this differentiation in observed price behaviour help to account for the gap between perceived and measured ination in the case of Italy. Similar studies have been conducted for the euro area as a whole and for several countries in the area. In most of them the indicator of perceptions is the balance statistic obtained from the consumer surveys conducted on the basis of harmonized methodologies in all the EU countries; this is the same indicator used and analyzed in chapter 1. Starting from the analysis for the euro area, the European Central Bank (2003b) compares price developments of frequently purchased products, labelled outof-pocket expenditure items because they are generally paid for in cash, with those of more rarely purchased products. 4 Overall, for the period considered (1996-2003) the correlation between ination perceptions and changes in the out-of-pocket index was not much closer than that between perceptions and headline ination. However, at the start of 2002 there was an abrupt acceleration in the out-of-pocket index (whose year-on-year rate of increase jumped from 2.5 per cent in December 2001 to 4.0 per cent in January 2002) and a sharp deterioration in perceptions. The ECB remarks that the two developments may have been related, even though large increases in the out-of-pocket index had also been recorded in the past without their being accompanied by a marked worsening in ination perceptions. However, the latter observation risks underestimating the greater impact that a given increase in the out-of-pocket index may have had on consumers at a time when public opinion was disoriented and paying exceptional attention to price developments. Moreover, as the ECB itself suggests, analysis of the phenomenon using aggregated indices for the euro area as a whole could cloud the importance of relations that stand out when individual countries are investigated.
4
The indicators for both types of product were developed using the ofcial disaggregated data released for the harmonized index of consumer prices. More frequently purchased products include non-durable goods and daily consumer services (e.g. food, beverages, tobacco, non-durable household goods, transport services, fuel, postal services, hotels, restaurants and cafs, and hairdressing),which on a weighted basis represent 35-40 per cent of the all-items harmonized basket for the euro area.
Aucremanne, Collin and Stragier (2007) provide econometric evidence of a structural break in the relationship between perceived and ofcial ination in the euro area following the cash changeover. On the basis of their empirical ndings the authors also argue that that the appearance of the gap does not depend on individual socio-economic characteristics (cohort-specic differences in individual ination perceptions are found to be signicant but stationary over the entire period considered) and is explained neither by the incidence of frequent purchases nor by the fact that owner-occupied housing is not included in the HICP index. The latter nding indicates that the perception gap is a general phenomenon, rather than specic to the HICP. In the authors view this is a reassuring result, given the prominent role this index plays in the Eurosystem monetary policy strategy. Turning to studies on individual countries, for Spain, according to the work carried out by Banco de Espaa (2003) and lvarez Gonzlez et al. (2004), the ination rate for frequently purchased products was much more strongly correlated with ination perceptions than the rate calculated on the whole basket.5 The introduction of euro notes and coins coincided with a phase in which the prices of those products were accelerating sharply, in contrast with the stability of ination for the other goods and services included in the general index. As in Italy, in Spain too the difference between the ination rates for the two groups of products was not exceptionally wide (approximately 1 percentage point). For France, an analysis by the national statistical institute of the link between consumers ination perceptions and the different components of the consumer price index (INSEE 2003) shows that in the period preceding the changeover (1987-2001) perceptions prevalently reected the price behaviour of food and energy products, both frequently purchased items. Surprisingly, however, in 2002 the behaviour of the prices of these two items was not consistent with the jump in ination perceptions: food price ination actually decelerated to an average annual rate of 2.6 per cent, from 5.1 per cent in 2001, while the prices of energy products fell further by 1.6 per cent. INSEE concludes that the serious deterioration in ination perceptions during the changeover year can basically be explained by psychological factors, notably consumers propensity to sift the data selectively for conrmation of their prior beliefs.6 Support for this view is found in the results of the monthly survey of French households, which
5
The index of frequently purchased items includes food, beverages and tobacco products, energy products, local transport, restaurants, bars and cafs, communication and cultural services, newspapers and magazines. Overall, the weight of these items corresponds to about 50 per cent of the general index. This hypothesis is corroborated by the results of the psychological studies conducted for Germany and Austria (see Sect. 7.3).
show that the fear of an inationary impact of the shift to the euro grew steadily as the changeover drew near: the percentage of respondents who expected an increase in prices rose from 47 per cent in December 2000 to 78 per cent in September 2001. As regards ination perceptions in 2003, and particularly the surge recorded in the autumn, INSEE judges that the deterioration can be traced to the large price hikes for some services and for tobacco products. In a similar analysis, again for France, conducted by Crdit Agricole (2004), the balance of perceptions is regressed on all the components of the consumer price index for the period 1998-2004. The results conrm that households pay closest attention to some specic items. In fact, just ve items are statistically signicant, with a combined weight equal to 60 per cent of the consumer price index basket: processed food (including beverages),7 tobacco products, energy products, rent and other housing expenses, and other services (among them, such highly visible services for households as restaurant services, education and hairdressing). For rent and other housing expenses and other services, it is estimated that consumers implicitly assign a much higher weight to these items than their weighting in the ofcial basket: 80 per cent as against scarcely 30 per cent. By contrast, no manufactured goods (except energy products) are found to be signicant in the regression, even though they represent 30 per cent of the index basket. The study concludes that the opening of the gap between ination perceptions and ofcial statistics in 2002 was largely due to the combination of two factors: the weight households implicitly assign to the two above-mentioned items and the rise in these items prices, which was appreciably greater than that in the general index (an annual average of 2.4 per cent for rent and other housing expenses and 3.4 per cent for other services, compared with 1.9 per cent for the general index).8 The Dutch central bank (De Nederlandsche Bank 2002b) similarly holds that the gap between perceived and measured ination in 2002 in the Netherlands can be explained in part by the impact on expectations of sharp price rises for frequently purchased products. In 2002 the annual average increase in these prices was 4.5 per cent, against a rise of 3.6 per cent in the general index. Even so, the increase was well below that perceived by consumer for
7
Contrary to expectations, unprocessed food is not signicant. Crdit Agricoles interpretation is that consumers do not assign signicant weight to the item because they are aware of its pronounced volatility. Tobacco products are another item that consumers concentrate on and that recorded very large price rises in France in both 2002 and 2003 (8.4 and 14.3 per cent, respectively). However, the study indicates that the implicit weight assigned to this item in the formation of perceptions is in line with that of the ofcial index.
the same prices (7.3 per cent in the rst six months of 2002, according to a central bank survey). Again for the Netherlands, Walschots (2002) and Buiten (2003) argue that perceptions may have reected the price hikes for products whose weight in the basket is modest but that are frequently purchased, while consumers attached less importance to the smaller price increases, or reductions, for products whose weight in the basket is signicant but that are purchased only rarely. Furthermore, these studies make it clear that for the goods and services with the greatest impact on perceptions the acceleration in prices began in 2001, not in concomitance with the changeover. A possible implication, which the two studies do not pursue, is that the introduction of the euro focused attention on a phenomenon that was already under way but which households had largely ignored until then.9 Pollan (2002) offers analogous observations with regard to Austria: the fact that the period preceding the changeover saw a generalized acceleration of prices prompted consumers to link the pick-up in ination with the shift to the euro; this impression persisted in the course of 2002 despite the reduction in ofcially measured ination. According to the study on Portugal by Santos et al. (2002), the increase in ination perceptions is again at least partly explained by the fact that many of the products for which exceptional price increases were recorded in concomitance with the changeover were goods and services frequently purchased and paid for in cash. Lastly, for Finland Aalto-Setala (2006) suggests that ination perceptions reected consumers expectations of the changeovers inationary effects, the developments of food prices in the winter of 2002 and the use of rounded prices and conversion rates in consumers price comparison. The empirical evidence presented by Kurri (2006) is only partially consistent with the view that perceptions were driven by price developments of frequently purchased items: while the surge in unprocessed food prices at the beginning of 2002 may have played a role, the negative variations of the same prices in 2003 did not lower perceptions. His analysis provides some support for the view that consumers expectations of the changeovers inationary impact led them to concentrate on prices increases and neglect reductions. 7.2.2 Number, size and distribution of price changes
Another hypothesis considered in chapter 1 is that individual ination perceptions are heavily more inuenced by price increases than decreases, even if the two are simultaneous and of the same order of magnitude, or by exceptionally large changes for specic goods, even if the goods in question make up a modest portion of the basket.
9
See chapters 1 and 3 with reference to Italy.
10
In the case of Italy in concomitance with the changeover there were price movements at variance with the past: an exceptionally high proportion of prices changed and there were very large rises for some services (see Chaps. 1 and 2). These patterns may have inuenced households ination perceptions in the direction indicated above. Analogous evidence has been found for other countries of the euro area. Banque Nationale de Belgique (2002) notes that in Belgium the adoption of the euro brought a return to decimal prices and the utilization of coins in cents, accompanied by a sharp rise in the number of different prices for the same type of product.10 According to the study, both factors may have disoriented consumers and contributed to the divergence between perceived and ofcial ination. Another work on Belgium (Cornille 2003) demonstrates that the changeover brought an increase in the percentage of prices that changed both upwards and downwards. A study by the German central bank (Deutsche Bundesbank 2004), conducted on a sample of elementary quotes for 25 products, shows that the introduction of euro notes and coins had a considerable impact on the distribution of the changes in individual prices in Germany. Following the changeover two effects may have disoriented German consumers: a sharp reduction in the proportion of attractive prices (easy to recognize and memorize because they end in familiar two-decimal-point gures) and a further diversication of prices for the same product. These effects were reversed only gradually and were still observable to some extent in 2003. In addition, the changeover was accompanied by an exceptional number of price changes and, particularly for services, some very large hikes.11 Lastly, the study underscores that many of the products which recorded large price increases have a modest weight in the ofcial index but nonetheless may have strongly inuenced perceptions.12
10
11
12
The central bank explains that there are appreciably more possible prices in euros than there were in Belgian francs: there are 100 different prices between 1 euro cent and 1 euro, compared with 40 in the corresponding interval of value in francs (i.e. between 1 and 40 francs, excluding the rarely used 50-centime piece). This possibility was exploited: in the sample of approximately 100,000 individual price quotes considered in the study, the number of different prices rose from just over 4,000 at the start of 2001 to around 7,500 in 2002. These two effects are traced both to menu costs and to the possibility that the phase of price confusion created an opening for price rises not attributable to cost or demand factors (see Chaps. 2 and 3 for an analysis of these aspects with reference to Italy). This is in line with what was noted in chapter 1 with regard to the ISTAT data for Italy (see Tables 1.1 and 1.2). The study also considers other factors that may have contributed to the surge in ination perceptions, including the use of rough mental rounding (see Chap. 1, Sect. 1.4.4), the tendency of consumers to overweight exceptional changes and their propensity to seek conrmation of their pessimistic expectations regarding the inationary effects of the euro. We shall address these points in the following section.
11
The distribution of price changes plays a crucial role in the index of perceived ination for Germany proposed by Brachinger (2005) On the basis of psychological theory of decision-making, he assumes that: consumers perceive price changes as absolute differences between the observed price and an individual reference price and rate higher (lower) prices as losses (gains); price increases impact more strongly on perceptions than price reductions or unchanged prices; the more often a product is bought, the stronger is the perception of the price changes; the level of the individual reference price still refers to the old prices in D-mark, i.e. those prevailing before the introduction of euro cash. The computation of the index of perceived ination relies on the ofcial elementary price indices weighted using a set of weights derived by the re-scaling the ofcial ones on the basis of the above assumptions. The main conclusions of the paper are that the proposed index captures ination as perceived by the general public and that this perception in 2005 was still close to the levels reached at the time of the cash changeover, above 7 per cent. This approach has been criticised by Hoffman, Leifer and Lorenz (2006) on two main grounds. First, extending psychological assumptions and ndings derived from other contexts to prices may be arbitrary; this goes, for instance, for Brachinger assigning of a weight to price increases that is double that of price decreases on the basis of experiments on decision-making from lotteries and gambling. Moreover, the authors argue that the index cannot be considered as a measure of perceived ination, since it is based on disaggregated ofcial price series (though differently weighted), not on a direct survey of consumers opinions, as in the case of the EC survey. 7.2.3 Psychological and memory factors Some psychological studies have explored mechanisms that may have inuenced individuals ination perceptions. The main results, summarized below, lend support to the hypotheses adopted in chapter 1 and in the works surveyed in the previous two sections of this chapter. Marques and Dehaene (2004) look at the psychological mechanisms through which consumers estimate individual prices following a change in the unit of account and the possible relationship between how often products are purchased and how fast consumers become familiar with their respective prices. They present the results of a series of experiments conducted at monthly intervals on a sample of Portuguese and Austrian university students between November 2001 and June 2002. The participants in each experiment were asked to state the price of 40 commonly used goods and services and to judge the extent to which they based their responses on mnemonic devices rather than on monetary conversion. The results conrm
12
the hypothesis of faster learning for more frequently purchased items: the estimates of their prices were more accurate from the outset and improved more between the rst and last test than those of less frequently purchased products. In addition, in the case of the Portuguese sample, the estimates of prices in euros at mid-2002 were less accurate than similar estimates made in national currency in 2000, before the changeover. On the basis of an experimental study performed on a sample of German citizens, Traut-Mattausch et al. (2004) found that individuals tended to overestimate the inationary effect of the changeover (even when the subjects were given the opportunity to compare the new prices in euros with the old prices in marks and thus did not have to reconstruct the latter purely from potentially faulty memory). Given two menus of a hypothetical restaurant, one in marks, the other in euros, the subjects systematically overestimated the change in prices during the shift from marks to euros. The study also found a positive relationship between expectations of the inationary effects of the introduction of the euro and their ex post overestimation of these effects. The authors attribute this relation to consumers propensity to be less accurate in checking data that conrm their expectations than data that disprove them, according to a mechanism they call selective output correction.13 A similar survey was performed on a sample of Austrian citizens by Kamleitner, Kirchler and Hofmann (2004), with analogous results as regards the overestimation of the change in prices and the relation between this overestimation and expectations. In addition, the authors extended the survey to perceptions of income, nding a tendency to underestimate incomes expressed in euros compared with those denominated in Austrian schillings. In an experimental study on a sample of Dutch citizens, Van Raaij and Van Rijen (2003) test the hypothesis that consumers are subject to a money illusion effect when the euro price is lower in nominal terms than its equivalent in the old national currency (this holds for all the euro-area countries except Ireland). This effect would induce consumers to underestimate the euro prices of individual products and spend more than they would if they had an accurate perception of the price in the new unit of account. In addition, contrary to what might be supposed on the basis of a simple aggregation of the perceptions of individual prices, the authors assert that this could help to account for the overestimation of ination for the complete set of goods and services. The paradox is explained by
13
In a situation in which errors of numerical calculation are possible (as in the mental conversion of the prices from one currency to another), individuals tend to accept results that are consistent with their expectations without further verication and, conversely, are likely to subject results that are inconsistent with their expectations to systematic checks. Consequently, they will end up accepting gures that are erroneous but consistent with their expectations more often than erroneous gures that clash with their expectations.
13
the fact that at the end of the month consumers, realizing that they have spent more than they intended and could afford, ascribe this outcome to an acrossthe-board increase in prices.14 In all, the results of the experiments presented in the study, conducted for two widely consumed products, are consistent with the money illusion hypothesis: the participants were willing to accept a higher price for both products when the price was denominated in euros instead of guilders; the result is statistically signicant for one of the two products. The authors do not explore the possibility that the presence of money illusion, apart from its effects on perceptions, may have encouraged tradesmen to exploit the phenomenon by marking up prices more than they would otherwise have done and thus may have had some impact on actual as well as perceived ination. Cestari, Del Giovane and Rossi-Arnaud in chapter 5 of this book examine the accuracy of consumers memory of pre-euro prices and its possible relationship with ination perceptions. They investigate Italian movie-goers recollection of pre-euro cinema prices, nding that only a small percentage of respondents recalled the correct price and that the average prices recalled were much lower than the actual pre-euro prices and dated back to years before the changeover. They also nd that the respondents who perceived higher and more persistent ination had, on average, a less accurate price recall. These results are conrmed by those obtained in chapter 6 for a different item (newspaper) and a representative sample of Italian consumers. 7.2.4 General studies and analysis of quantitative perceptions The empirical studies reviewed so far focus on specic factors that may have affected ination perceptions as measured by qualitative indicators.
14
The possibility of a money illusion effect connected with the changeover is also discussed in Gamble et al. (2002), with experiments applied to a sample of Swedish and British citizens; these two nationalities were chosen in order to test the effects connected exclusively with the conversion from one denomination to the other, which in the authors view are easier to measure on persons who had no experience of using the euro. The results of the experiments accord with the money illusion hypothesis in the case of the comparison between prices in kronor and euros, conict with it in the case of sterling. The authors also argue that the money illusion effect can be reversed when the values expressed in a currency are so large or small that consumers are induced to reproportion them mentally; this hypothesis, labelled the lira effect (in the case of the Italian lira, prices are ostensibly translated into kilo lire by dividing them by one thousand) appears to be supported by the results of the experiments performed on the Swedish sample. Finally, evidence of money illusion is provided by Kooreman, Faber and Hofmans (2004) on the basis of data on charity donations in the Netherlands.
14
The partial approach they adopt does not allow an overall assessment of the relative importance of the various factors considered. A number of recent works have tried to overcome this drawback by investigating the relative role played by a combination of explanatory variables; some of them have also collected quantitative indicators of ination perceptions. The study by Del Giovane, Fabiani and Sabbatini presented in chapter 6 collects information on ination perceptions, both in qualitative and in quantitative terms, through a survey of a representative sample of Italian consumers, and relates them to various individual characteristics. It indicates that the quantitative assessment of ination reported by consumers is much higher than that measured by ofcial statistics. These assessments reect the contribution of various factors; in particular, individual perceptions are signicantly affected by the degree of knowledge about the concept of ination and the related statistics, asymmetric perceptions of price increases and decreases, inaccurate memory of past prices, socio-demographic characteristics such as gender and education, and the individuals economic and nancial conditions. The bottom line is that consumers assessment on what they report as ination is based on a complex combination of factors, including developments that inuence households economic conditions, though they are not necessarily related to ination. This interpretation is supported by the nding that the information collected through the consumers shopping activity and thus more directly related to price behaviour does not appear signicantly to shape individuals perceptions. Two other recent studies (Lindn 2006, Malgarini 2007) have investigated qualitative as well as quantitative ination perceptions on the basis of the experimental quantitative questions that were added in 2003 to the monthly Harmonized Consumer Survey carried out by national institutes in the EU countries and coordinated by the European Commission.15 Lindn (2006) shows that over the period considered (May 2003-January 2006) euro area consumers reported, on average, an annual ination rate of about 17 per cent (with a median of 10 per cent). Both qualitative and quantitative perceptions follow a downward trend over the period, indicating a slow adjustment after the surge that followed the launch of the euro. Perceived ination is higher in the case of women, poorer and less-educated respondents; it is lower among older ones. The study also shows that perceptions (as well as expectations) are lower in the case of respondents who reported that over the next 12 months they planned to buy a car, buy or build a home, or spend a large sum of money on home improvements or renovations, consistently with the authors hypothesis that planned expenditure of this kind represents an incentive to assess the current and future inationary outlook more attentively.
15
An experimental question on price expectations was also included in the surveys, but we concentrate here only on assessments about past price developments.
15
The results obtained by Malgarini (2007) for Italy are in line with the ndings reported in chapter 6. Perceived ination is much higher than ofcial measures (about 26 per cent, on average, over the period February 2003December 2006). Although the perception of high ination characterizes the entire sample, some differences are found according to household category. In particular, perceived ination is higher among women, less well-off and less educated respondents, and in southern regions of the country. Moreover, higher ination perceptions are associated with negative assessments on both personal and general economic conditions, suggesting the existence of a strong link between a perception of economic distress and ination overestimation. Finally, consumers planning major purchases report lower ination rates, consistently with the ndings by Linden (2006). Survey evidence is also used for Austria by Fluch and Stix (2005) and Stix (2006). On the basis of data collected in summer 2004 from a sample of 2,000 Austrian citizens, they show that the divergence between perceived and ofcial ination can be related to various factors. In particular, they nd that ination perceptions are higher among respondents more exposed to the price rises of frequently purchased goods, among those who had expected inationary effects from the launch of the euro, and among those who habitually evaluated prices in euros by converting them mentally in schilling (and thus tended to anchor their price assessments to outdated reference prices).
7.3 The changeover impact on prices

This section reviews the studies of the impact of the cash changeover on consumer prices for various euro-area countries (Table 7.2).16 Section 7.3.1 describes work based on the data collected by statistical institutes to calculate national consumer price indices. As for Italy, Belgium, Germany, Luxembourg, Portugal and Spain samples of elementary price quotes were made available by the national statistical institutes, making it possible to examine price variations on the occasion of the changeover and in particular the effects of rounding to attractive prices. Other works use only aggregate data, i.e. the ofcial consumer price indices and their components as released monthly by the statistical institutes. Their aim is to identify the basket items that registered exceptionally large changes in 2002
16
The European Central Bank (2003a), on the basis of the empirical evidence from individual countries, described the consumer price impact of the introduction of the euro as modest for the area as a whole, while recognizing the difculty of combining results based on different approaches. The overall impact was ascribed to a combination of components: rms passing on changeover-specic costs to consumers (e.g. adaptation of cash registers and staff training), rounding to attractive levels, and in some cases a widening of prot margins. The ECB also found that the effects were concentrated mainly in the service sector.
16
and subsequently to check for a possible link with the changeover. The studies reviewed in section 7.3.2 are based on non-ofcial statistical sources, such as prices observed in specic sectors and qualitative survey information, or indirect methodologies; a few of them consider imperfect information on the side of consumers and the role played by the national changeover regulations.
Table 7.2. Studies on the changeover impact on prices
Analyses based on ofcial price statistics Rounding to attractive prices Anomalous variations Analyses based on non-ofcial sources or indirect methodologies Sector studies Indirect methodologies Imperfect information and changeover regulation
Mastrobuoni 2004 Dziuda and Mastrobuoni 2005 Ehrmann 2006 Eife 2006 -
Euro area
European Central Bank 2003a
Eurostat 2002 European Central Bank 2003a
Austria Belgium
Finland France Germany Greece Ireland Italy
Banque Nationale de Belgique 2002, 2003 Deutsche Bundesbank 2002, 2004 Mostacci and Sabbatini 2005 (Chap. 2 in this book)
Pollan 2002 Banque Nationale de Belgique 2002 Deutsche Bundesbank 2002, 2004 Mostacci and Sabbatini 2005 (Chap. 2 in this book)
Adriani, Marini and Scaramozzino 2003 Hobijn, Ravenna and Tambalotti 2006 -
Luxembourg Netherlands Portugal Spain
Banque Centrale du Luxembourg 2002, 2003 -
Gaiotti and Marini, Piergallini Lippi, 2004 and Scaramozzino (Chap. 3 in this 2004b book) Angelini and Marini, Lippi 2005 Piergallini and (Chap. 4 in this Scaramozzino book) 2004a De Nederlandsche Bank 2002a, 2002b -
Santos et al. 2002Santos et al. 2002 lvarez lvarez Gonzlez et al. Gonzlez et al. 2004 2004
17
Table 7.3. Harmonized index of consumer prices (annual averages; percentage change on previous year)
Unprocessed Processed food food
Austria
Energy products
Non-food, Services non-energy products

0.0 0.9 0.7 0.0 2.0 1.7 0.4 1.8 0.9 0.1 0.9 0.9 -0.5 0.2 0.4 0.6 2.9 2.5 0.5 1.6 1.0 1.7 1.8 (1) 2.4 1.5 1.7 1.3 0.9 3.7 2.7 1.5 2.5 3.1 2.0 2.4 2.6 0.5 0.9 1.5 1.9 3.2 3.0 2.3 2.0 2.6 3.5 3.8 3.5 0.6 1.5 2.8 0.7 1.7 2.1 2.6 3.9 4.5 5.7 6.0 7.6 2.3 2.9 3.4 2.5 3.6 3.3 1.5 4.5 4.9 4.0 4.7 5.9 3.6 3.9 4.6 1.5 2.5 3.1
Total
Belgium
2000 2001 2002 2000 2001 2002 2000 2001 2002
3.3 5.1 1.5 0.2 6.9 3.2 0.7 6.2 5.5 2.4 7.4 2.9 0.2 7.0 -0.1 1.7 6.7 6.9 2.3 9.2 3.0 1.8 5.8 4.9 2.2 6.6 4.7 0.3 9.8 3.6 2.5 8.9 0.2 3.8 7.2 4.7 1.8 7.0 3.1
0.5 2.5 1.9 1.3 2.2 1.5 1.8 2.7 1.6 2.3 3.4 3.3 0.0 2.3 2.5 2.7 5.1 5.5 7.1 3.5 4.7 1.3 2.5 2.2 7.1 3.5 4.7 1.7 5.6 3.8 1.4 3.1 3.8 0.9 2.7 4.9 1.2 2.9 3.1
11.6 0.3 -2.4 16.3 1.4 -3.6 11.2 -1.0 0.2 12.1 -1.5 -1.5 13.9 5.7 0.3 17.3 -1.7 -0.3 13.8 -2.7 3.4 11.6 1.6 -2.6 20.3 -2.6 -3.7 14.9 8.4 3.3 6.1 5.2 1.2 13.4 -0.7 -0.2 13.0 2.2 -0.6
2.0 2.3 1.7 2.7 2.4 1.6 3.0 2.7 2.0 1.8 1.8 1.9 1.4 1.9 1.3 2.9 3.7 3.9 5.3 4.0 4.7 2.6 2.7 (1) 2.6 3.8 2.4 2.1 2.3 5.1 3.9 2.8 4.4 3.7 3.5 3.7 3.6 2.1 2.3 2.3
Finland
France
2000 2001 2002 Germany 2000 2001 2002 Greece 2000 2001 2002 Ireland 2000 2001 2002 Italy 2000 2001 2002 Luxembourg 2000 2001 2002 Netherlands 2000 2001 2002 Portugal 2000 2001 2002 Spain (1) 2000 2001 2002 Euro area 2000 2001 2002
Source: Based on Eurostat data. Notes: (1) The percentage change is calculated with respect to homogeneous series for 2000 and 2001 that do not take account of the introduction in January 2001 of new methods of observing special promotional offers.
18
Before examining the individual studies, let us look briey at overall developments in the harmonized indices and their components released monthly by Eurostat for the various area countries (Table 7.3). We see that there is no sign of generalized anomalous changes in 2002. An exception is service prices, which accelerated sharply in nearly all countries. This gives an initial indication of a possible sectoral effect of the changeover to the new currency, but not of generalized price increases specically due to it. 7.3.1 Analyses based on ofcial price statistics
(a) Rounding to attractive prices
For some countries the adjustment of prices to the introduction of the euro has been studied using the subsets of elementary price quotes collected by national statistical institutes for the consumer price index.17 A number of studies have looked at the rounding of prices in the new currency to attractive levels; in some cases they have estimated the impact on consumer prices. For Italy, the case analyzed in chapter 2, the share of attractive prices fell with the changeover and then gradually rose back to approach its previous level. The impact of rounding on consumer prices between January and October 2002 was estimated at between 0.16 and 0.75 percentage points of extra ination, depending on the denition of rounding used Here we give the evidence available for the other countries of the euro area. In most cases the data available are only for variations during and following the changeover in the share of attractive prices. This information is interesting since it gives an idea of the distribution over time both of any price effects due to rounding and of the impact on consumers ination perceptions (Sect. 7.2.2) of repricing in euros. Aside from Italy, only Luxembourg has also produced an estimate of the consumer price effect of rounding.18
17
18
These data are not made public. The institutes release consumer price indices at a less disaggregated level that varies from country to country (at the most, some 200 disaggregated series). For the harmonized index of consumer prices, Eurostat releases about 80 series. For the changeover, however, some central banks, in cooperation with the statistical institutes, conducted empirical studies on a subset of elementary price quotes for the indices. By contrast, ex ante studies were conducted in all the countries to measure the possible inationary impact of the changeover. For an account of these results, see chapter 2 for Italy and the Appendix 7.A for the other countries.
19
For Germany, two joint studies by the central bank and the national statistical institute (Deutsche Bundesbank 2002, 2004) analyze elementary price quotes for a sample of widely consumed goods and services (35 products in the earlier study, comprising 18,000 quotes, and 25 in the later one). Whereas at the start of 2001 over 80 per cent of the prices (in marks) were at attractive levels, with the changeover in January 2002 this proportion fell by half. The study hypothesizes that this was in part because stores continued to display prices in marks and kept these, rather than euro prices, at attractive levels. The share of attractive prices gradually recovered, and by September 2003 was only slightly below the pre-changeover level (Table 7.4).
Table 7.4. Share of attractive prices and effect of rounding on consumer prices Country Share of attractive prices (percentages) (1) Before the changeover (in national currency) After the changeover (in euros) Impact of rounding (percentage points)
January 2002 Belgium Germany Italy Luxembourg (2) Spain 88 73 80 89 35 45 21 25 42
Other month 70 (September 2003) 75 (September 2003) 48 (October 2002) 35 (July 2002) 59 0 .16-0 .75 (Jan.-Oct. 2002) 0 .45 (Jan.-July 2002)
Notes: (1) For some countries, approximate values. (2) The study does not give the percentage of attractive prices in national currency before the changeover.
The Banque Centrale du Luxembourg (2002, 2003) conducted a study in cooperation with the national statistical institute, observing 7,000 monthly elementary quotes used in calculating the consumer price indices. It found that the proportion of attractive prices in euros fell to around 25 per cent with the changeover in January 2002 and went back up to 40 per cent by July (the last month covered). The cumulative impact on prices (national and harmonized indices) due exclusively to rounding to attractive levels was estimated at 0.45
20
percentage points between January and July 2002; nearly half of this impact was caused by the items hotels, cafs and restaurants and recreation, entertainment and culture. The study notes also that the adaptation of the price system actually got under way in 2001, and estimates the cumulative impact of rounding between January 2001 and July 2002 at 0.7 points. Methodologically, these estimates attribute the entire variation in each price that becomes attractive in a given month to the changeover.19 Accordingly, the study interprets this as the upper bound of the rounding effect. The study performed by the Banque Nationale de Belgique (2002) using elementary price quotes between January and September 2002 indicates that in Belgium the share of attractive prices, historically around 70 per cent of those expressed in Belgian francs, fell sharply to 35 per cent in 2002 in euros and recovered gradually to 61 per cent by September. A subsequent study (Banque Nationale de Belgique 2003) found that it was not until September 2003 that the share of attractive prices regained the pre-changeover level. The works on Spain (lvarez Gonzlez et al. 2004) and Portugal (Santos et al. 2002) indicate that the price systems in those countries underwent signicant adjustment following the introduction of the euro. In Spain, attractive prices in pesetas accounted for 88 per cent of all prices in November 2001. In January 2002, in euros, that share plummeted to 42 per cent and in January 2004 had still only recovered to 59 per cent. No precise data are given for Portugal, but one infers that in March 2002 the share of attractive prices was considerably lower than it had been at the end of 2001.
(b) Anomalous variations
Another way of estimating the inationary impact of the changeover is to identify anomalous changes (not necessarily due to rounding) in the disaggregated consumer price indices released by the national statistical institutes. Eurostat took this approach to provide an estimate for the entire area and, country by country, for Austria, Belgium, Germany, the Netherlands, Portugal and Spain. For Italy an estimate of this sort indicates an impact of between 0.1 and 0.6 percentage points of additional ination on average for 2002 (see Chap. 2). Eurostat (2002) estimated that the changeover had an impact of no more than 0.16 percentage points on average consumer price ination in the euro area in the rst quarter of 2002. This estimate was obtained from the 88 component series into which the basket of the harmonized index is broken down. To identify the items displaying anomalous variations in the rst quarter of 2002, the study used a method of statistical analysis that could determine whether developments in each disaggregated time series
19
On the methodological alternatives used in Italy, see chapter 2.
21
deviated signicantly from what was to be expected based on past trends. The deviations thus identied were adjusted for changes ascribed to specic factors independent of the introduction of the euro, such as rises in regulated prices, changes in indirect taxes and jumps in unprocessed food prices due to bad weather.20 The remaining differences were summed, using their weights in the harmonized index basket. The estimate so obtained was essentially conrmed by an alternative method, in which anomalous variations were detected based on the difference between the change observed in the rst quarter of 2002 and the largest rise observed in any single quarter since the harmonized index was instituted (in 1995). Turning to individual countries, for Germany the national central bank (Deutsche Bundesbank 2004) concluded that the overall inationary impact of the changeover in January 2002 was slight. True, there was a jump in twelve-month consumer price ination from 1.4 per cent in December to 2.1 per cent in January, but most of this was due to indirect taxes on energy and tobacco products and higher food prices caused by bad weather. Considering these factors, the upper limit of the consumer price impact of the changeover was estimated at 0.3 percentage points in January 2002 (Table 7.5). The study also examined the elementary price quotes of 25 widely used goods and services. For each item it ran a regression for the years from 1991 through 2000 of the consumer price against the main cost factors. This analysis shows that some service prices (hairdressing, laundries, restaurants, cinemas) deviated sharply in 2002 from what would have been expected extrapolating the regression analysis, but that the divergence was largely eliminated over the next two years.
Table 7.5. The inationary impact of anomalous variations on consumer prices (percentage points) Belgium Germany Portugal Spain Euro area Memorandum: Italy 0.5-0.6 (average 2002) 0.2 (Jan.-Apr. 2002) 0.3 (Jan. 2002) 0.2 (1st quarter 2002) 0.4 (average 2002) 0-0.2 (Jan.-Mar. 2002)
For Austria, Pollan (2002) reaches conclusions similar to those for Germany: the sharp acceleration in fresh food prices at the start of 2002 is
20
The three EU countries that did not adopt the euro (Denmark, Sweden and United Kingdom) had comparable rises in these prices.
22
traced to bad weather conditions, while exceptional rises in some service prices appear to be ascribable to the changeover (hairdressers, car repairs and home plumbing services). The main difference from Germany is the absence of exceptional increases for restaurants. The fact that some of the largest increases had already been recorded in 2001 is interpreted as evidence that the adaptation of the price system partly preceded the cash changeover. Overall, the study concludes that in Austria the introduction of the euro did not have a signicant impact on consumer price trends (but no quantication is given). For Belgium the national central bank (Banque Nationale de Belgique 2002) examined price movements in various product categories to identify anomalous changes with respect to trends in previous years, using data available up to April 2002. The study nds that the exceptional increases in fresh food prices observed at the start of 2002 were due to causes other than the introduction of the euro (bad weather). But the central bank argues that the changeover may have contributed to the exceptional rises in the prices of services, notably restaurants and cafs and some personal services. The overall impact in the rst four months of 2002 is estimated at 0.2 percentage points. For Portugal, Santos et al. (2002) estimated the inationary impact of the new currency by a regression analysis to detect anomalous rises in concomitance with the changeover. For the rst three months of 2002, a regression was run for each of the 189 published series making up the statistical institutes general consumer price index, with explanatory variables including seasonal dummies, a linear trend, some lags in the dependent variable and three changeover dummies set at 1 for January, February and March 2002 and 0 for all other months. The statistically signicant coefcients of these three dummies were summed for each regression, and then the results of the various regressions were aggregated using the weights of the series in the basket. The resulting values were adjusted for the contributions of anomalous changes clearly due to factors other than the changeover.21 As derived by this method, the estimated price impact in connection with the introduction of euro banknotes and coins was small, just 0.2 percentage points in the rst three months of 2002. About two thirds of this effect was due to service prices, notably hotels, restaurants and cafs.
21
Specically, the procedure excluded sharp rises in some energy prices (decided prior to the changeover as a function of international oil prices), in the prices of fresh foods (due to bad weather in early 2002) and rents. For rents, the rise in January 2002 was due not only to the standard adjustments for the previous years ination but also to a change in the law that extended indexation to past ination to a larger number of apartments.
23
The study for Spain (lvarez Gonzlez et al. 2004) examined 117 disaggregated series used by the statistical institute in compiling the CPI. Anomalous changes in the course of 2002 were identied by calculating, for each series and each month, the difference between the percentage change on the previous period and the average for a control period consisting of the years 1993-2001 plus 2003. For the sub-indices for which this difference was positive, the portion of the variation ascribed to the changeover was isolated by excluding variations due to other clearly identied factors (such as bad weather or indirect taxes) and changes in regulated prices. Finally, a regression similar to that for Portugal was run to determine the extent to which these variations were exceptional. The inationary impact of the anomalous changes that may be considered to be changeover-related came to 0.4 percentage points for 2002 as a whole. Most of the impact was concentrated in the rst few months and, as for Portugal, mainly involved services (restaurants and cafs accounted for about half of the total effect). To sum up, the study of the anomalous variations in the price series released by national statistical institutes, where available, suggests that the inationary effects of the changeover in 2002 were not substantial. 7.3.2 Analyses based on non-ofcial sources or indirect methodologies This section reviews works based on data other than from ofcial sources or on indirect evidence. There are essentially two reasons why some inquiries took these approaches. The rst is that in the debate over the effects of the changeover the reliability of the ofcial statistics was challenged in many countries. Alternative sources thus serve to check for results signicantly different from those obtained using ofcial data. A second reason is that the ofcial data are publicly available only at a relatively high level of aggregation, typically for groups of goods and services but not for individual items. For instance, the ofcial data on restaurants give the average of prices at a range of establishments of vastly differing types, qualities and locations. And Eurostats series for the harmonized index put restaurants together with other eating establishments and hotels. Alternative price sources (in this case, restaurant guides) provide a higher degree of disaggregation.
(a) Sector studies
Gaiotti and Lippi (2004 and Chap. 3) examine Italian restaurant prices taken from a leading guide in order to study pricing before and after the cash changeover. The results show that 2002 was marked by a larger number of establishments changing their prices than in previous years but not by
24
exceptionally large increases. The increase in prices was considerable in all the years surveyed (1998-2004); that in the twelve months ending in June 2002, while substantial (9 per cent), was less than the 10.5 per cent registered in the previous year. The study also found that increases in 2002 and 2003 were larger in provinces where there was less competition.22 Adriani, Marini and Scaramozzino (2003) extend the study of restaurants to six countries: France, Germany and Italy in the euro area and Denmark, Sweden and the United Kingdom outside it. The data are taken from the 2002 and 2003 editions of a leading international restaurant guide. The restaurants are divided into three categories (patronized respectively by tourists, by regular clients, and by occasional customers). The study indicates that price rises in the euro area countries were larger than in the non-area countries, owing mainly to sharper increases at tourist restaurants, while in the other types of restaurant the increases were substantial but did not differ signicantly by country. The authors cite this as evidence that the introduction of euro banknotes had an especially large impact in sectors where competition was weak (for a comment on these conclusions, see Chap. 3). Hobijn, Ravenna and Tambalotti (2006) offer a theory to explain the effect of the introduction of the euro on restaurant prices, citing the rigidity of price revisions due to menu costs. From this standpoint 2002 was a special year, because the change in currency required all restaurants to reprint their menus, even those that otherwise would not have done so, with potential inationary impact. The study shows that the sharp rises found in January 2002 are explained by this theoretical model, as are those observed in Italy (Chap. 3). Non-ofcial data, from a survey of 800 retailers and 750 consumers in June 2002, were used by the Dutch central bank (De Nederlandsche Bank 2002b) to get a rough estimate of the price impact of the changeover.23 The responses of the retailers who said they had raised prices with the changeover (about a third of the total) indicated an inationary impact of 0.6 percentage points for the rst half of 2002.24 The retailers who raised prices most signicantly were relatively smaller operators in sectors where competition was less intense.
22
23
24
The study also comments on the work of Marini, Piergallini and Scaramozzino (2004a), which covers not only restaurants but also wines and hotels in Italy. Obviously the reliability of the estimate is weakened by the fact that the data are drawn from survey responses rather than direct observation of prices. A survey of a sample of retailing rms in January 2002 (De Nederlandsche Bank 2002a) had produced a lower estimate of the impact of between 0.2 and 0.4 percentage points. Most of this effect was ascribed to rounding to attractive prices in euros, while about one third was attributed to the costs of preparing for the new currency.
25
(b) Indirect methodologies
Angelini and Lippi (2007 and Chap.4) gather indirect evidence on the behaviour of prices from the analysis of cash withdrawals from ATM. Since this analysis does not rely on ISTAT data but on gures collected by the national central bank, it provides an independent check for ination developments at the changeover based on ofcial statistics. Their conjecture is that in presence of a substantial rise in the price level, a discontinuity should be observed in the dynamics of payment instruments, notably cash withdrawals from the ATM network. The authors set up a simple model of ATM withdrawals and show how, under certain assumptions, it can be used to infer price dynamics from the observed nominal time series for ATM withdrawals and consumer expenditure; in particular, they want to assess whether consumer prices underwent a post-changeover increase not recorded by ofcial ination statistics. Formally, the study provides a test for the hypothesis that the coefcients of the model estimated prior to 2002 are subject to instability, which can be captured econometrically, once the time horizon is extended to years 2002-03. The econometric analysis fails to nd evidence supporting this instability, in other words that a price hike not captured by ofcial statistics occurred after the changeover. The paper by Marini et al. (2004b) also adopts an indirect methodology to investigate the existence of a measurement bias in the Harmonized Index of Consumer Prices (HICP) for United Kingdom and Italy. The analytical approach is based on the assumption that there should be a consistency between the dynamics of the households income deated by the HICP, measured by the statistical agencies, and that of their nancial situation as captured through qualitative household surveys. In particular, if more households reported a worsening in their nancial situation at a time when the deated median household income remains unchanged, one could infer that ination had been underestimated by the ofcial statistics. The empirical analysis reported in the paper fails to nd evidence of a signicant inationary bias for the United Kingdom over the period 1996-2003, and for Italy up to 2001; for the following period it indicates that ination in Italy had been underestimated signicantly (by around 6 percentage points in the two years 2002-03), with a structural break in the relationship between the two variables occurring in January 2002. A limit of this analysis is that the survey collects information on households self-assessment of their own nancial condition, not objective gures ; hence, it cannot be excluded that, in particular following the changeover, the measure used in the paper corresponded to a perceived rather than an actual development.
(c) Imperfect information and changeover regulation
Dziuda and Mastrobuoni (2005) present an analysis of the changeover effects on prices based on an imperfect information model; the assumption is that consumers
26
are rational but face difculty in assessing prices after the changeover.25 The model implies that retailers may take advantage of this situation by increasing prices, and that the incentive to do so is larger for cheaper goods and, contrary to the implications of standard competitive models, for less concentrated markets. The model implies, in fact, that retailers operating in more concentrated markets may have found it optimal to undertake transparency-enhancing investments, such as advertising and dual-pricing, to attract customers on the occasion of the changeover. Consistently with the model, the empirical analysis nds that after the changeover cheaper products recorded higher price increases in the euro area than in other EU countries and that this effect was less intense where retailer concentration was higher. Ehrmann (2006) tests whether price developments and perceived ination in the aftermath of the euro cash changeover are related to the complexity of the currency conversion rate. Based on a model of information-processing constraints of economic agents in which consumers need to gather costly information to evaluate prices and make decisions on their purchases the author argues that the cash changeover increased the required information processing more on the side of consumers than on that of sellers, and this mismatch created the opportunity for price increases, the larger the more complex the conversion rules in the various countries. The empirical analysis suggests that ination developments for food products after the changeover were related to the complexity of the conversion rate in a non-linear way: price increases were higher for the countries characterized by an intermediate degree of complexity of the conversion rate, and lower for both the simplest and the most complex conversions (the latter result suggesting that consumers faced with very complex rates relied less on simple rules of thumb). The paper also nds that legal requirements for dual display of prices helped to contain changeover effects. Eife (2006) argues that the impact of the euro cash changeover on prices, captured by a qualitative indicator he computes on the basis of individual HICP series released by Eurostat, was noticeable in a few countries and almost negligible in others, and that these differences can be related to changeover regulations and effectiveness of national agencies communication about the event.26 He notices that the introduction of a new currency is not necessarily neutral on prices, as a result of the opposite effects of changeover confusion
25 26
A similar analysis is conducted in a previous paper by Mastrobuoni (2004). For each series the ination rate is regressed on its own lags, seasonal dummies and a changeover dummy, equal to 1 in January 2002 and 0 otherwise, and the impact is measured as the percentage of sub-indices that recorded a signicant increases in this month (similar results are found by considering a window from 1 to up to 7 months). As Eife acknowledges, this qualitative and unweighted indicator is not necessarily related to the absolute size of the price increases, nor to the impact on overall ination.
27
(people are unfamiliar with the new currency and rms may take advantage of this) and changeover awareness (the publics concern about the risk of price increases encourages a more careful monitoring). This implies an important role played by both the regulatory framework and the communication strategy. Concerning the former, the author shows that the impact on prices was in general more moderate in countries where the dual pricing was mandatory (before and after the changeover). In particular, Germany and Austria are presented as opposite cases. Germany, which allowed dual pricing only in the two months following the changeover, recorded the highest estimated impact on prices. Austria, which according to the author recorded the lowest impact on prices and whose regulation could serve as a model for future changeovers, was very timely in establishing formal rules for dual pricing. In particular, dual pricing had to start three months before the changeover and continue as long as the Austrian schilling was accepted as legal tender (leaving the option for extending this period to the end of 2002); moreover, the law established detailed prescriptions on how schilling and euro prices had to be shown on price tags, punished violations with nes and provided for close monitoring by the public authorities. Austria was also the only country to prohibit economically unjustied price increases on the occasion of the changeover; this ban, though difcult to enforce, may have exerted some pressure on rms and discouraged them from proting from the changeover. Eife also discusses the shortcomings in the communication strategy in Germany, arguing that they have also contributed to the changeover impact on prices, and draws some policy implications.
7.4 Conclusions
The works on individual countries reviewed here bring out some remarkable common features of consumers ination perceptions and price developments with the introduction of euro banknotes and coins. With reference to the discrepancy between perceived and ofcially measured ination, they show that perceptions are disproportionately affected by the price variations of the most frequently purchased products, and that in some countries in 2002 and 2003 these prices increased more than those of less frequently purchased items. Second, they nd that in a number of countries the changeover brought an increase in the share of prices that changed (both upward and downward), an increase in price diversication for the same product, and a higher number of exceptionally large increases in service prices. These developments are consistent with a sharper rise in perceived than in measured ination, as long as perceptions are more heavily inuenced by price rises than by price cuts and disproportionately by increases for certain goods. Third, experimental studies in psychology nd that for individuals there was a correlation between prior expectations concerning the effects of the introduction of
28
the new currency and the way the effects were perceived after the event: the worse ones expectations, the stronger ones tendency to overestimate the actual price impact. This relationship is ascribed to the propensity to be less assiduous in checking data that validate ones expectations than those that disconrm them. It is also found that the consumers memory of pre-euro prices can be largely distorted and that this can contribute to individual ination perceptions. A few recent studies, some of which also investigate quantitative perceptions, assess the relative importance of the various above-mentioned factors within a unied analytical framework. All in all, they conclude that the individual assessments on what consumers report as ination reect a complex combination of factors, including demographic aspects and factors that, though not necessarily related to ination, nevertheless inuence households economic conditions. The actual evolution of the price system also proved to be similar between countries. Wherever evidence is available, it shows a sharp fall in the proportion of attractive prices at the moment of the changeover. The return to a proportion comparable to that in the old national currencies was slow and by the start of 2003 still incomplete everywhere. Estimates of the effects of rounding to attractive prices in euros, available only for Italy and Luxembourg, suggest that the inationary impact was limited, around half a percentage point. For the Netherlands, a survey of retailers found a comparable effect. In many countries the prices of certain goods and services rose very sharply in 2002. For some products fresh foods and, at the start of the year, energy products these rises were temporary and largely due to readily identied specic factors having nothing to do with the changeover (bad weather and indirect taxes). For some jumps in service prices, however, no such obvious cause is to be found (in particular, increases in production costs were not observed for all products), so we can hypothesize a connection with the introduction of euro cash. All in all, the estimates for the individual countries concur that the inationary impact of anomalous price variations in 2002 presumably connected with the changeover was no more than half a percentage point. A few papers consider the possible effects of limited information on the side of consumers and suggest that inationary impacts at the sector level were related to country- and marketspecic factors, such as the degree of complexity of conversion rules, the degree of concentration in the various market segments, and changeover regulations.
29
Appendix 7.A. Ex ante studies of the changeovers possible inationary impact

In the period preceding the changeover, studies estimating its possible effect on consumer prices were conducted in all the euro-area countries.27 They were largely of two types: (a) qualitative surveys of rms pricing intentions, changeover-related extraordinary costs, preparation for the changeover, and so forth; and (b) analyses based on data collected by the national statistical institutes for the calculation of consumer price indices, designed to arrive at an ex ante quantitative estimate of the changeovers inationary impact as a consequence of rounding to attractive prices (the concept is dened in Chap. 2, Sect. 2.3). The surveys of rms determined that businesses had a generally low estimate of the specic changeover-related costs (i.e. cash registers, IT equipment and staff training); furthermore, they also found that rms had already incurred part of these costs in the preceding years. All in all, rms did not expect the introduction of euro notes and coins to have a major impact on their selling prices. The empirical analyses of the changeovers potential impact were very similar in their objectives and methods; they differed mainly in the level of disaggregation of the data they used. Focusing on rounding to attractive prices, these exercises adopted the working hypothesis that the changeover would inuence consumer prices exclusively through the combination of two types of rounding effect: rst a conversion rounding effect, derived by applying the ofcial exchange rate to prices in national currency and rounding the result to the nearest euro cent, as recommended by the European Commission; and subsequent rounding to an attractive level for prices in euros (assuming that businesses that set attractive prices in national currency would continue to do so in euros). The estimated potential impact therefore did not include additional possible effects, such as wage adjustments and the risk that retailers might exploit the populations difculty in estimating the new prices to apply increases justied neither by cost nor by demand factors. The empirical studies took the possible behaviour of rms into account by formulating alternative scenarios rounding to attractive prices: all upwards (worst case), all downwards (best case), or upwards or downwards to the closest attractive level. For some countries a most likely scenario was constructed. The ultimate overall impact on prices, i.e. the impact
27
For a list of these works, see Appendix A in Mostacci and Sabbatini (2003).
30
corresponding to a complete adjustment of the system of prices to the new attractive levels in euros, was calculated for all the scenarios. The main conclusions reached by these studies can be summarized as follows: (a) the practice of setting attractive prices (rounded, psychological or fractional) would be very widespread in all the countries; (b) in the worst-case scenario, the impact on average euro-area consumer price ination would be approximately 1 percentage point; (c) in the most likely scenarios, available for some countries, the estimated impact was between 0.1 and 0.35 percentage points. Referring to these results, the European Central Bank (2002) remarked that: signicant risks of a strong upward impact on consumer prices did not emerge at aggregate level. In the short run, the cyclical slowdown in the euro area, the vigilance of consumers organizations and their monitoring of retailers, the period of dual circulation (ensuring transparency of price conversion) and the governments commitments regarding regulated prices would help to limit the impact; in 2001 the effects of the approach of the cash changeover had been conned to a few specic items and were negligible, in any event, in terms of their overall impact on the all-items harmonized index; in the longer run, the ination-curbing effects of the strengthening of competition in the euro area thanks to the greater transparency of a system of prices expressed in a single currency were expected to outweigh any temporary price pressures.
31
References
Aalto-Setala (2006) Why do consumers believe that the euro raised prices?, International Journal of Consumer Studies, 30, 202-206. Adriani, F., Marini, G., Scaramozzino, P. (2003) The inationary consequences of a currency changeover: evidence from the Michelin Red Guide, CEIS Tor Vergata Research Paper Series, 27, July. lvarez Gonzlez, L.J., Cuadrado Salinas, P., Jareo Morago, J., Snchez Garca, I. (2004) El impacto de la puesta en circulatin del euro sobre los precios de consumo, Banco de Espaa, Servicio de Estudios, Documentos ocasionales, 0404. Angelini P., Lippi, F. (2005) Did ination really soar after the Euro changeover? Indirect evidence from ATM withdrawals, CEPR discussion paper, 4950, March. Aucremanne, L., Collin, M., Stragier, T. (2007) Assessing the gap between observed and perceived ination in the euro area: is the credibility of the HICP at stake?, National Bank of Belgium Working papers series, No. 122, April. Banco de Espaa (2002) The effect of the euro cash changeover on ination, Annual Report 2002, 95. Banque Centrale du Luxembourg (2002) Les effets inationnistes du basculement vers leuro, Rapport annuel, 32-35. Banque Centrale du Luxembourg (2003), Limpact du passage leuro duciaire, Press release, March 5. Banque Nationale de Belgique (2002) Effets du passage a leuro duciaire sur lination, Revue Economique, 2, 41-55. Banque Nationale de Belgique (2003) Adaptation of prices to the euro: a survey, Press release, December 15. Brachinger, H.W. (2006) Euro or Teuro? The euro-induced perceived ination in Germany, paper presented at the 2006 IAREP/SABE Conference, Paris, July. Buiten G. (2003) Why ination still feels so high, CBS Webmagazine, www.cbs. nl. Cestari, V., Del Giovane, P., Rossi-Arnaud C. (2007) Memory for prices and the euro cash changeover: an analysis for cinema prices in Italy, Banca dItalia, Temi di Discussione, 619, February Cornille, D. (2003) Ladaptation des prix au passage leuro: une mise en perspective, Revue Economique, Banque Nationale de Belgique. Crdit Agricole (2004) Pourquoi les mnages franais surestiment-ils lination?, Flash co, March 26. Del Giovane P., Sabbatini, R. (2006) Perceived and measured ination after the launch of the euro: explaining the gap in Italy, Giornale degli Economisti e Annali di Economia, 65(2), 152-192. Del Giovane, P., Fabiani, S., Sabbatini, R. (2007) What are ination perceptions about? A survey-based analysis on Italian consumers, Banca dItalia, mimeo, May. De Nederlandsche Bank (2002a) Smooth euro, higher prices?, Quarterly Bulletin, 51-56, March.
32
De Nederlandsche Bank (2002b) Getting used to the euro, Quarterly Bulletin, September, 51-56. Deutsche Bundesbank (2002) Consumer prices and the changeover from Deutsche Mark to euro, Monthly Report, July, 15-24. Deutsche Bundesbank (2004) The euro and prices two years on, Monthly Report, 56 (1), 15-28. Dziuda, W., Mastrobuoni, G. (2005) The euro changeover and its effects on price transparency, and ination. Mission euro, mission accomplished!, Princeton University, 2005. Eife, T.A. (2006) Price setting behaviour and price setting regulations at the euro changeover, paper presented at a Seminar at the Eesti Pank, June, 1. Ehrmann, M. (2006) Rational inattention, ination developments and perceptions after the euro cash changeover, European Central bank, Working paper series, No. 588, February. European Central Bank (2002) Euro cash changeover not expected to have a signicant impact on consumer prices at the aggregate level, Monthly Bulletin, January, 24-25. European Central Bank (2003a) Effects of the introduction of the euro banknotes and coins on consumer prices, Annual Report 2002, 40-42. European Central Bank (2003b) Recent developments in euro area ination perceptions, Monthly Bulletin, October, 24-25. European Central Bank (2007) Measured ination and ination perceptions in the euro area, Monthly Bulletin, May, 63-72. Eurostat (2002) Euro changeover effects , Euro-indicators news release, 58, May. Fluch, M., Stix, H. (2005) Perceived ination in Austria Extent, explanations, effects, Monetary Policy the Economy, 22-47. Gaiotti, E., Lippi, F. (2004) Pricing behaviour and the introduction of the euro: evidence from a panel of restaurants, Giornale degli Economisti e Annali di Economia, 63, 491-526. Gamble, A., Grling, T., Charlton, J., Ranyard, R. (2002) Euro illusion: Psychological insights into price evaluations with a unitary currency, European Psychologist, 7, 302-311. Hobijn, B., Ravenna, F., Tambalotti, A. (2006) Menu costs at work: restaurant prices and the introduction of the euro, The Quarterly Journal of Economics, 121 (3), 1103-1131. Hoffmann, J.H., Leifer, H., Lorenz, A. (2006) Index of perceived ination or EU consumer surveys? An assessment of Professor H.W. Brachingers approach, Review of European Economic Policy, 41, 1-9. INSEE (2003) Opinion des mnages sur les prix passs et lination, Note de conjoncture, December. Kamleitner, B., Kirchler, E., Hofmann, E. (2004) The Euro : Perception of a Loss in Purchasing Power, paper presented at the SABE/IAREP Conference CrossFertilization between Economics and Psychology, Philadelphia, 15-18 July. Kooreman, P., Faber R., Hofmans, H. (2004), Charity donations and the euro introduction: some quasi-experimental evidence on money illusion, IZA Discussion Paper 1318, September.
33
Kurri, S. (2006) Why do consumers perceived ination differ so much from actual ination?, Bank of Finland, 3, 75-81. Lindn, S. (2006) 400 000 observations on ination perceptions and expectations in the EU, paper presented at the 28th CIRET Conference on Cyclical Indicators and Economic Policy Decisions, Rome, September 20-23 Malgarini, M. (2007), Quantitative ination perceptions and expectations of Italian consumers, ISAE, mimeo. Marini, G., Piergallini, A., Scaramozzino, P. (2004a) Euro e dinamica inazionistica, Economia Italiana, 1, 163-177. Marini, G., Piergallini, A., Scaramozzino, P. (2004b) The Bias in Measuring Ination: Evidence from the UK and Italy, University of London, Centre for Financial & Management Studies, Discussion Paper 50. Marques, J.F., Dehaene, S. (2004) Developing Intuition for Prices in Euros: Rescaling or Relearning Prices?, Journal of Experimental Psychology: Applied, 10 (3), 148-155. Mastrobuoni, G. (2004) The effects of the Euro-Conversion prices and price perceptions, CEPS Working Paper 101. Mostacci, F., Sabbatini, R. (2003) Has the euro caused ination? The changeover and the consumer price rounding in Italy in 2002, www.istat.it. Pollan, W. (2002) Price developments upon introduction of euro banknotes and coins, Austrian Economic Quarterly, 4, 167-175. Santos, D., Evangelista, R., Nascimento, T., Coimbra, C. (2002) Analysis of the impact of the conversion of escudos into euros, Banco de Portugal, Economic bulletin, September, 1-14. Stix, H. (2006) Perceived ination and the euro: why high? Why persistent?, paper presented at the European Economic Association and the Econometric Society European Meetings, Vienna, August 24-28. Traut-Mattausch, E., Schulz-Hardt, S., Greitemeyer, T., Frey, D. (2004) Expectancy conrmation in spite of disconrming evidence: The case of price increases due to the introduction of the Euro, European Journal of Social Psychology, 34 (6), 739-760. Van Raaij, W.F., Van Rijen, C.L.A. (2003) Money illusion and euro pricing, paper presented at the IAREP Conference EURO currency and symbol, Wien, 3-5 July. Walschots, J. (2002) Why does ination feel so high?, CBS Webmagazine, October, www.cbs.nl.
List of gures
Fig. 1.1 Fig. 1.2 Fig. 1.3 Fig. 1.4 Fig. 1.5 Fig. 1.6 Fig. 1.7 Fig. 1.8 Fig. 2.1 Fig. 3.1 Fig. 3.2 Fig. 3.3 Fig. 3.4 Fig. 3.5 Fig. 4.1 Fig. 4.2 Fig. 5.1 Fig. 5.2 Fig. 5.3 Ination measured by the statistical institutes and ination perceptions Ination perceptions and sentiment about the euro Ination rates for goods and services in Italy according to the frequency of purchase Average consumer price ination in Italy and its dispersion Average consumer price ination in Italy and its dispersion: subset of 48 goods and services Sign and size of price changes in Italy: subset of 48 goods and services Ination perceptions and articles selected in Il Sole 24 ore and La Stampa Ination perceptions in Italy: balance resulting from the consumer survey and estimates Types of prices in Italy, autumn 2001 Italy: HICP, category restaurant and cafes Activity and demand in the sector hotels and restaurants Fresh food prices and ULC in the sector hotels and restaurants Frequency distribution of price increases Model based effect of the changeover ATM usage and consumption growth (quarterly data) The power of the statistical test: results from a counterfactual exercise Recalled vs. actual cinema price (full ticket) Recalled vs. actual cinema price (reduced ticket) Recalled pre-euro prices compared to actual prices in historical perspective
List of tables
Extreme price changes: 207-item breakdown Sign and size of quarterly price changes in Italy by product categories: subset of 48 goods and services Table 1.3 Average ination rate by decile of equivalent expenditure Table 1.4 Italian household income by work status of the head of household, 2000-2002 Table 1.5 Number of articles selected from Il Sole 24 ore and La Stampa (Selection criteria: Inazione or caro-vita in the headline, associazioni or consumatori in the text) Table 1.6 Summary of the correlations - dependent variable: balance of perceptions Table 2.1 Rules for dening attractive prices in euros Table 2.2 Price increases by type of price Table 2.3 Price increases by type of price and type of product Table 2.4 Price increases by type of price and distribution channel Table 2.5 Inationary effect of rounding in the strict sense (Method 1) by type of price and distribution channel Table 2.6 Inationary effect of rounding in the broad sense (Method 2) by type of price and distribution channel Table 2.7 Changes in the consumer prices of the items not included in the estimations of the changeover effect Table 2.8 Summary of the estimates of the changeovers inationary impact in the rst ten months of 2002 Table 2.9 Difference between price dynamic in 2002 and its average in 1999-2001 Table 2.10 Regression analysis of anomalous contributions to ination in 2002 Table 3.1 The rise in the price of a meal Table 3.2 The rise in the price of a meal selected Rome districts Table 3.3 Number of extreme price increases Table 3.4 Number of constant and falling prices Table 3.5 Distribution of price revisions, unchanged prices excluded Table 3.6 Decomposition of price increases Table 3.7 Distribution of price increases in 2002 by market competition Table 3.8 Determinants of price changes Table 3.9 Model based decomposition of price increases Table 3.A1 Sample size Table 3.A2 Sample composition, by region Table 1.1 Table 1.2
Table 3.A3 Table 3.A4 Table 3.A5 Table 3.A6 Table 3.A7 Table 3.A8 Table 3.A9 Table 4.1 Table 4.A1 Table 5.1 Table 5.2
Sample composition, by type Sample representativeness Competitive pressure The price of a meal Price changes, by type Price changes, by area Restaurant price in Italy: a comparison among datasets Estimates of equations 4.5 and 4.6 Alternative estimates of equations 4.5 and 4.7 Respondents Price recalled and respondents answers relating to individual memory, habits of mental conversion and frequency of moviegoing Table 5.3 Perceived increase in the general price level synthetic measures of the respondents opinion Table 5.4 Price recalled and ination perceptions Table 5.5 Price recalled and age Table 5.6 Types of price recall: pre-euro price expressed in lire or euros; 2003 price Table 5.7 Summary regression dependent veriable: recalled price for cinema tickets before the euro cash changeover Table 5.B1 Descriptive statistics on price recall and related factors.
[SUBJECT INDEX TO BE INCLUDED]
[AUTHORS INDEX TO BE INCLUDED]

Versione Inglese PDG Sabbatini - Volume Completo - 03!10!2007

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Versione Inglese PDG Sabbatini - Volume Completo - 03!10!2007

Încărcat de

Drepturi de autor:

Formate disponibile

The euro cash changeover, ination and consumers perceptions.

Lessons from the case of Italy.

Paolo Del Giovane and Roberto Sabbatini

To Lorenza, Luciano and Marco To Pierluigi and Riccardo

Foreward [TO BE INCLUDED]

Paolo Angelini Vincenzo Cestari

Paolo Del Giovane

Silvia Fabiani Eugenio Gaiotti

Francesco Lippi Franco Mostacci Clelia Rossi-Arnaud Roberto Sabbatini

P. Del Giovane and R. Sabbatini

1.2 Ination perceptions

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

(b) Euro area

Perceptions (qualitative indicator) (1)

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0

Perceived and measured ination after the launch of the euro

1.3 The ofcial index

P. Del Giovane and R. Sabbatini

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

Perceived and measured ination after the launch of the euro

Table 1.1. Extreme price changes: 207-item breakdown

0.24 -14.7 Jan. 8.05

Source: Based on ISTAT data (2003 basket).

P. Del Giovane and R. Sabbatini

1.4 Why the gap between ofcial and perceived ination?

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

Perceived and measured ination after the launch of the euro

2.5 2.0 1.5 1.0

P. Del Giovane and R. Sabbatini

standard deviation - 550 items

standard deviation - 207 items

3.0 2.5 2.0 1.5 1.0

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

average (left-hand scale)

Perceived and measured ination after the launch of the euro

decreases increases between 0 and 5% increases exceeding 5%

P. Del Giovane and R. Sabbatini

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

2002 nominal real (3) 5.7 0.0

2002 nominal real (3) 9,696 7.7 2.0

3.9 10.1 6.2 6.4 6.8

5.9 9.7 7.4 7.2 8.0

0.2 4.0 1.7 1.5 2.3

4.4 11,095 12,168 0.5 9,738 10,460

0.7 10,341 11,081 1.1 9,597 10,363

Perceived and measured ination after the launch of the euro

P. Del Giovane and R. Sabbatini

Perceived and measured ination after the launch of the euro

(a) Il Sole 24 Ore

Inflation perceptions (left-hand scale)

Number of articles (right-hand scale)

Inflation perceptions (balance, left-hand scale)

Number of articles (right-hand scale)