Benchmarking selected European Airports by their Profitability Envelope a Break-Even Analysis

Branko Bubalo1 German Airport Performance (GAP) Research Project (Berlin, June 13th 2012) Abstract: In this article a simplified benchmarking methodology is presented. This new approach is based on the creation of a stepwise discrete maximum profitability envelope over distributed data points. Financial and operational data of 210 European airports from 17 countries have been analyzed over a period of nine years (2002 to 2010). For reasons of comparability the cross-country time-series data are deflated to a reference location, currency and time. The data requirements have been reduced to the core variables of the airport production process - level of passengers and profits or deficits (before interests and taxes) in a particular year. These data are used to isolate industry best practices, to estimate its benchmarks and to aid decision-making about feasible maximum profits, given the volumes of passengers, and the expected break-even point. The benchmarks can be used to calculate potential efficiency gains for underperforming airports. Keywords: Airport benchmarking, profit maximization, break-even analysis

I. Introduction To this day large-scale comparisons of airports, either financial or operational, across countries are still rare and far from sufficient in guiding decision makers in a simplified manner. Given the importance of airports in a globalized economy regarding the linkage of national and international destinations and the magnitude of daily travelers and cargo, I feel that there exists an overemphasis on theoretical modeling in favor of careful empirical and numerical descriptions, which would aid our understanding of the real processes. Therefore, I am developing a new partial factor productivity (PFP) measurement approach in airport benchmarking (see Vogel and Graham, 2010, p. 22) driven by extensive amounts of collected data2 by strictly focusing on the core variables of the production process and by pinpointing the best-practices. In the discussion about maximizing social welfare, I take the view that airports, independent of type of ownership - public, partially private or fully privatized -, should strive for (at least) a financial break-even by maximizing profitability and performance, given the managerial feasibility limited by existing demand. Similar to most (private) businesses economic losses need to be minimized to keep subsidies or other compensations, e.g. debt from credits, down. It
1

Doctoral student at Hamburg University under supervision of Prof. Dr. Stefan Vo; branko.bubalo@googlemail.com 2 The data for this paper was gathered while working on a benchmarking study commissioned by the Norwegian Ministry of Transport and Communications. I appreciate the participation in our data collection survey by airport staff, for example by the main airport operator in Norway, Avinor, and by the Highland and Island airport operator in the United Kingdom, HIAL. The help in the data collection by our GAP project members Ivana Strycekova, Keith Lukwago, Sascha Michalski, Eric Njoya and Tolga lk is greatly acknowledged.

should be a common goal for airports to maximize profits and to reach break-even, so taxes, benefitting the whole society, may be paid. There is no clear reason why the society should constantly pay for subsidies benefitting few passengers travelling to or from loss-making airports frequently in remote locations. Hence, in this article I take a general view which leads to reducing the data requirements to an absolute minimum, where demand as a given input (and output) is measured in number of arriving (and departing) passengers and unit profits (or losses) as an output is measured in earnings before interests and taxes (EBIT) per passenger (Figure 1). In many cases3 these data can be directly extracted from income statements or other profit-and-loss accounts and airport operational statistics (Doganis and Thompson, 1975, chapter II; Koopmans, 2008, p. 23; Vogel and Graham, 2010, p. 23). It is obvious that loss-making airports would not be able to survive in a competitive and fully private market (Gillen and Lall 1997, p.6). Airports rarely or never achieving a break-even must therefore receive subsidies to some degree4, often times by cross-subsidization inside a pool of a multi-airport operator between profitable and non-profitable airports, or by direct subsidies in form of public funds (taxpayers money), (low interest) grants, national capital expenditure programs or other non-operating sources of income. Adler, Liebert and Yazhemsky (2012) define an airport as a private [or public] production system in which society maximizes social welfare by encouraging airport management to maximise profits. I will look at the trend and shift of the frontier of maximum profitability the profitability envelope at 210 European airports over a time-frame of nine years, answering general questions like: Which benchmark or range of profits could a priori be expected for any airport given the local level of demand and, how does my airport perform relative to the bestpractice? Where is the break-even point for the airport-industry located in a particular year, and in which direction does the break-even point and the profitability envelope shift over the years? The paper is structured in five parts, starting with this introduction. I continue in part two with the theoretical background including the review of literature, overview of different methodologies and description of data. The third part will show the application of the profitability envelope over the collected dataset. The fourth part will look at some possible gains in profitability, if the identified benchmarks would be reached. The final part will draw some conclusions and will give an outlook on further research.

II. Theoretical Background, Methodology and Data Description In this short discussion on measuring the profitability and cost efficiency of airports, I am aiming to draw parallel conclusions to more sophisticated multiple input and output linear programming5 and optimization techniques, such as Data Envelopment Analysis (DEA), but with a straightforward and simpler approach, where the analyst preserves full comprehension and control over his or her data6. It was found by Adler et al. (2012) and in supplement regression analyses for a recent benchmarking study7 that there exists a sufficient correlation between DEA relative efficiency scores8 and the PFP measure of unit profits, measured in
3

There exist Ressentiments against the publication of confidential financial data in the airport industry, which requires personal appeasements, non-disclosure agreements or the usage of carefully coordinated questionnaires. 4 Subsidies lower social welfare by requiring public funds, which could be used elsewhere on e.g. schools, hospitals or universities. 5 Linear Programming is an optimization technique invented by George Dantzig in 1947. See an application of the method in Dorfman (1951). 6 For example by reducing the analysis to two dimensions, x and f(x), for explanatory convenience compared to ndimensions used in the DEA. 7 The second-stage regressions were conducted by Tolga lk of the GAP research project for the benchmarking study of Avinor airports (which is currently under review) in comparison to other European airports. 8 The DEA model in Adler et al. (2012) makes use of four inputs (staff costs, other operating costs, runway capacity and terminal capacity [largely avoiding the complications involved in the definition of capacity, especially in

Earnings before Interests and Taxes (EBIT) per passenger (PAX). However, it is evident that airports need the critical mass (Heymann, 2005, p. 2) in number of PAX and consequential revenues in order to operate self-sufficient. Therefore, deeming loss-making airports to be inefficient and vice versa profit-making airports to be efficient would not allow a fair comparison among different sized entities. Similar to the DEA we aim at revealing and computing the relative efficiency compared to other similar sized peers (Adler et al., 2012). Accordingly, I relate ratios of outputs over inputs to the size of airports, thus relative to comparable peers. An airport is deemed the more inefficient the further the calculated ratio is located from the according profitability benchmark. One reason of not using EBIT or other profit measures in the DEA is the inability of the method to deal with negative numbers, although this drawback could be circumvented in all linear programs by adding a sufficiently large constant9 to the data in order to make them all positive, thereby not changing the optimum strategies (Dorfman, 1951, p. 349). A DEA study using EBIT or another measure of profitability has yet to be conducted. Most existing studies focus on the input-output balance of costs and revenues and make use of additional traffic or capacity data.
Total Revenues
Total Costs 900

EBIT 10,000
Power (Total Revenues) 800

Power (Total Costs)

Revenue and Costs in NOK (2010 prices) (log. Scale)
Poly. (EBIT)

Revenues = 108.7 * PAX1.03142 R = 0.96 700

1,000

500 100

Costs = 7841.136*PAX 0.72307 R = 0.92

400

300
10

200

100

Break-Even Line
0
EBIT = 0.0000006*PAX 2 + 34.0532*PAX - 25,972,498 R = 0.84

0 1,000 10,000 100,000 1,000,000 10,000,000

-100 100,000,000

Passengers (log Scale)

Figure 1: Revenues, Costs and EBIT for sample airports in the years 2002 to 2010 (PPPadjusted base Norway in 2010 prices) (Source: Own survey data) I observed, when plotting the adjusted data10 of the 210 European airports11 that especially small and regional airports lie below some threshold of (unit) profits (the break-even line) and

connection to level-of-service]), four intermediate goods (international and domestic PAX, tons of cargo, and number of flights) and two outputs (non-aeronautical and aeronautical revenues). 9 Similar to the conversion from Celsius to Kelvin temperature scales. 10 I made the required financial data adjustments for price-level differences and inflation to the German Airport Performance (GAP) database. I used for Purchasing Power Parity (PPP) adjustments and currency conversion, based on Norwegian Kroners, the following table from OECD: http://stats.oecd.org/Index.aspx?datasetcode=SNA_TABLE4. Furthermore, I used, for Consumer Price Index (CPI) adjustments, based on base year 2010, from another table from OECD: http://stats.oecd.org/Index.aspx?DataSetCode=MEI_PRICES.

EBIT in NOK (2-010 prices)

600

Millions

Millions

100,000

1,000

usually do not report any profits in subsequent years (Figure 1). There are very few airports which are volatile with regard to their profitability, showing deficits as well as profits in following years. Another main observation are decreasing returns-to-scale on a unit profit basis (i.e. EBIT per PAX), but constant (or even increasing) returns-to-scale on absolute profits (i.e. EBIT). With increasing output, which can be observed in figure 1 and subsequent figure 4 and 5 in Section 3, these returns-to-scale pick up below the break-even line at volumes of 300.000 to 400.000 PAX per year. With the approximation functions in figure 1 the average industry breakeven point is calculated at 1.06 million PAX when total operating costs equal total operating revenues, and is about 750.000 PAX, when EBIT is equal to zero12. In the literature one can find different assumptions regarding the airport industry break-even point. For example in the 1970s Doganis and Thompson (1975, p. 338) have calculated the break-even point for British airports, where costs are fully covered by revenues, to be around 3 million passengers. Heymann (2005, p. 3) (citing a study by the European Commission) estimates the break-even point for European airports to be in the broad range of 500.000 to 2 million passengers13. Koopmans (2009) has empirically tested the correlations between the closely related profitability measure earnings before interests, taxes, depreciation and amortization (EBITDA) per workload unit (WLU)14 and found it a good descriptor of operational performance (Koopmans, 2009, p.60). Hence, this strongly implies that EBIT per PAX could be a sufficiently good descriptor of airport managerial efficiency. This is a similar result as shown by Vogel and Graham (2010, p. 37) who found that profit per WLU correlates significantly and highly significantly to ten of 15 of the studied airport performance measures, such as return-oninvestments, EBITDA margin and asset turnover. Doganis and Thompson (1975, p. 334-337) use absolute surplus and deficits (including interests15), costs and revenues, relative to the output level as a descriptor of managerial effectiveness and for determining the break-even point. Gritta, Adams and Adrangi (2006, p.1) have chosen EBIT and its variation over time as a measure to quantify the business risk of a firm and to construct financial performance measures in their study of the effects of operating and financial leverage on [] Major U.S. Air carriers rate of return. Therefore, I am convinced that annual EBIT per passenger (PAX) serves as an adequate approximation of airport operational productivity and relative efficiency. However, Gillen and Lall (1997, p. 6) dispute the usefulness of profitability measures and state that these are totally misleading, given the unique position of airports. In light of our graphical and numerical analysis of data from 213 airports over a period of nine years this judgement may not be accurate, since we observe strong industry-wide (Doganis and Thompson, 1975, p. 336) trends in profit generating ability with regard to the output level, largely independent of type of ownership and/or capacity16. It should be noted that our results only account for European airports. The situation may look differently at North-American airports and other continents.
11

The set consists mainly of 49 French, 49 Norwegian, 39 British, 32 Italian and 18 German airports. See full sample list in the Appendix A. 12 There are exceptional airports in France and the United Kingdom (and in the United States), which make profits at lower throughput levels than the break-even point. At some cases these data have been removed from the analysis, since data inconsistencies have been suspected. These exceptional cases will be looked at in preceding analyses and publications. 13 Certainly these figures vary with our definition of profits and which costs and/or revenues are included in it. Our data is cleaned from outside revenue sources such as state grants, government transfers and so on. 14 One workload unit (WLU) equals one passenger or 100 kg of cargo. In the literature the usage of this combined measure and the equivalence of passengers to cargo is disputed, especially in terms of different handling costs, revenues and infrastructure requirements. 15 The measure in Doganis and Thompson (1975) is in financial terms called earnings before taxes (EBT). 16 An independence of capacity and/or ownership has not been proven, but these influences may be rather small across the range, especially for airports unable to breaking-even. Logical reasoning may relate operational efficiency closer to capacity utilization and private ownership to profit makin g entities.

One could argue, that first, using EBIT as a single aggregate output measure has the advantage that this figure accounts for the net amount and balance of all required operating resources (costs), all generated operating income (revenues), and all investments in airport development and capacity (depreciation). In a simplified sense annual (positive or negative) EBIT are defined as operating revenues minus operating costs minus depreciation of assets (capital). For many (privatized) airports EBIT is generally published in annual reports, financial statements or other statistics, as it represents a common figure in finance. For reasons of comparability the collected nominal financial data are converted to Norwegian Kroners (NOK), then PPP-adjusted relative to Norway17 to account for different price levels and wages between countries, and furthermore inflation adjusted to 2010 prices to account for loss of currency value over time18.
Inputs Depreciation of Assets Costs Facility Capacity Outputs EBIT

Production Process
financial operational

Revenues

arriving PAX originating PAX

departing PAX terminating PAX externalities

environmental

Figure 2: Core Inputs and Output of the Production Process of Airports (Source: Own illustration) Second, I limit the operational variables only to the input number of passengers (PAX), as this figure represents the original (origin, transfer and destination) demand, and is usually easily available. Divisions by passenger characteristics, such as international, domestic, business or leisure, have not been made, since these ultimately reflect groups of people with different spending behaviour, service requirements and associated costs, which need to be targeted commercially by the airport business and charges model. In Doganis and Thompson (1975, p. 331) it was emphasised that airport management has only limited control over externally defined factors, such as demand, i.e. level of passengers departing from or arriving at a particular location, and fixed costs. Therefore, the airport managements prime function is to balance (expected) revenues and cost, and to maximize output (EBIT) given the level of input (PAX). To generate demand for certain destinations is a major function of airline marketing and route development. Figure 2 exemplarily shows the main inputs and outputs of the airport production process, yet, it is not so clear which side needs to be maximized or minimized in an optimization program. Typically, the inputs are minimized and the outputs are maximized, for example in an output-oriented DEA. In figure 2 one may define all costs, including depreciation, as a financial input, and all revenues as a financial output. On the operational side of the production process arriving or originating PAX, who request to be served by the airport, may be viewed as an operational input, and all departing and terminating PAX, who have been served by the airports may be viewed as an operational output. Therefore, total PAX show input and output characteristics, in our case these are defined as inputs. Externalities, such as noise and delay,

17

Norwegian Kroners have been chosen, since this study is partly based on the results of a Benchmarking study for the Norwegian Ministry of Transport and Communications. 18 See footnote 4, which tables have been used to derive the real monetary values.

may be viewed as undesirable output (Adler et. al., 2012) and its reciprocal (i.e. inverse) may be maximized19. The data points can have five characteristic locations in relation to the profitability envelope, which displays the benchmark profits. Figure 3 shows theoretically the different stages of development of an airport with regard to increase in size and profitability20. The first group would represent loss-making airport, which data points lay below the reference profitability benchmark(s). The second group would include loss-making airports which define the profitability envelope and represent a benchmark. The third group consists of one (or more) airports, which break-even exactly. More realistically it is this airport in the data set, which is first making profits in EBIT per PAX, with minimum demand in terms of number of PAX. The fourth group consist of profitable airports, which are relatively inefficient with regard to their position below the (maximum) profitability envelope. Similar to the first group, it is possible to calculate productivity gains, if the benchmarks would be reached (please see Chapter 4). The fifth group represents superior airports, which are profitable and define the profitability envelope. These airports show maximum profitability with regard to their output level and comparable peers, therefore these truly represent best-practices. It is this group of airport, where one could expect the largest interest from private investors, as these promise the highest dividends and return-on-investments (ROI).
EBIT per PAX

Group 1

Group 2

Group 3

Group 4

Group 5

Profits

+
Break-Even line

(maximum) Profitablity Envelope

Losses

: Data Point : Benchmark

not profitable, benchmark not reached not profitable, benchmark break-even point profitable, benchmark not reached profitable, benchmark

PAX

Figure 3: Data points in relation to Break-Even line, (Industry) Profitability Envelope and Break-Even Point (Source: Own illustration) Frequencies of flights and available seats per route are under the control of carriers which try to match capacity to demand on each flight on a particular route by maximizing load factors with the exception of subsidized PSO traffic. Certainly, the aircraft sizes in each carriers fleet put a significant limitation to matching airport demand exactly, since the ability of a flight breaking even and economies of scale must be considered by airline management. Therefore the strategies behind the required number of total flights measured in annual air transport movements (ATM) have not been analysed. We find that the applicability of ATM as a holistic operational variable is not given, since this measure is typically only related to airside input and output.

19 20

I appreciate this suggestion by Nicole Adler. See for example Reinhard (1973, p. 830) for an application of a similar function to the break-even analysis of the Tri-star aircraft development program of Lockheed.

Airport Leipzig Cologne East-Midlands Bergamo Benbecula Brussels Rennes Billund

IATA LEJ CGN EMA BGY BEB BRU RNS BLL

% Cargo of Total WLU >60% 41% 38% 30% 28% 27% 23% 21%

Table 1: Airports with relevant average share of cargo >20% of total Workload Units (WLU) across the years 2002 to 2010 (Own survey data) In order to account for airports which make a majority of their turnover with the handling of cargo the use of the measure Workload Unit (WLU = 1 PAX or 1/10 tonnes of cargo) may be considered as an alternative operational measure, i.e. EBIT per WLU. Sensitivity analyses across the sample only showed minor differences between the trends of average EBIT per PAX and average EBIT per WLU. For the full data sample the average amount of cargo accounts for about 5% of the total WLU, with eight airports having shares of more than 20% of the total WLU (Table 1). The Norwegian airports Hasvik (HAA), Rost (RET) and Svalbard (LYR) are the only ones having significant shares of cargo of up to 10% cargo of total WLU. For reasons of clarity we do not discuss the alternative measure WLU any further in this article.

III. Application of the Profitability Envelope on the Dataset In our heterogeneous sample of airports across different countries, years and sizes we relate the surplus or losses in EBIT to the number of passengers, which gives us the profitability ratio of average unit of output per average unit of input - EBIT per PAX. It is state-of-the-art in current research to divide the airport sample by classes or clusters (see Adler et al. 2012), such as above 2 million or below 2 million passengers, and present averages for each class. Since this kind of division is always arbitrary we chose to relate the ratio to the according continuous volume of passengers, ordered from the lowest to the highest observation in the sample. This made the choice of displaying a logarithmic scale for passenger numbers in a majority of our figures necessary. We observe in Figure 4 that growth in number of passengers is a strong driver of more than proportional increases in EBIT per passenger. However, the chosen profitability ratio stagnates at a certain level of saturation, after which it appears the average EBIT per passenger can only marginally be increased any further. The maximum benchmark has been observed at LondonCity (LCY) airport with 144 NOK per passenger at a level of 3.3 million passengers in 2008. LCY defined the maximum benchmark for the years 2007 and 2009 as well, with passenger levels of 2.9 or 2.8 million PAX and an according EBIT per PAX of 133 and 99 NOK, respectively. LCY airport is operating a short (STOL) runway and has very limited terminal and/or aircraft parking stand capacity, which makes this achievement of high average profitability even more special. This example shows that the airport management at LCY airport has reached a high value-added productivity by fully exploiting its available resources.

Passengers (log. Scale)

1,000 10,000 100,000 1,000,000 10,000,000 100,000,000

500

Break-Even Line

-1,000

-1,500

-2,000

-2,500

-3,000

Figure 4: Profitability and Profitability Envelope by Airport Size for the Years 2002 to 2010 (All Airports in the DEA Sample, PPP-adjusted base Norway in 2010 prices; Italy and France only until 2009) (Source: Own survey data) It can be concluded that airports with a critical level around the break-even point of more than, say 300.000 to 1 million passengers, are becoming increasingly interesting for investors, but also request less regulation for achieving maximum profitability. A marginal increase in annual number of passengers directly leads to increasing absolute profits and returns to scale at this high level of output (see Figure 1). However, from a welfare state point of view, the application of price-cap regulation to profitable airports (group four and five) may be an advantageous instrument. This regulation limits monopoly power and decelerates growth. Therefore it could balance airport revenues from charges and from non-aeronautical sources against true (societal) costs. For loss-making airports (in groups one to three) a laissez-faire approach may be appropriate, if social obligations for transportation services are not of public interest. Otherwise, there is a lack of incentives and degrees of freedom for airport management to change the revenue and cost structure in order to become profitable. Similar to private entrepreneurs, the state may provide start-up funds or low-interest grants in initial stages of airport development. Even airport closures should remain as a last option, if an airport has no financially sustainable strategy. To derive the profitability envelope exemplarily shown in figure 4, I made use of a fairly simple algorithm, which plots the envelope over the data points. The result of the algorithm provides profitability benchmarks for each level of passengers, thus gives a feasible strategic target for the airport management. The algorithm and required steps to create the profitability envelope are outlined as follows: 1. 2. i = 1 to n PAXi < PAXi+1 < < PAXn # n = Number of airports in the sample. # Sort PAX column in ascending order.
8

EBIT per Passenger in NOK (2010 prices)

-500

3. 4.

Envelopei = EBIT per PAXi For i to n 2, Envelopei+1 = If Then Else

# First data in EBIT per PAX column is equal to the # initial starting point for the Envelope; i = 1.

# From the second entry onwards, the new entry is # compared, if larger than last Benchmark. If yes, it # is set as new EBIT per PAX Benchmark. EBIT per PAXi+1 > Envelopei EBIT per PAXi+1 Envelopei .

IV. Break-Even Analysis We want to get clearer answers on where the best-practice break-even point lies in each year, particularly at which level of demand observations surpass the break-even line. Thus, at which critical level of demand full coverage of operating cost could possibly be managed. We have attempted no effort in calculating the break-even points exactly by interpolation between the last observation below break-even and first observation above break-even, because we recognize when working with real data, the quality is commonly mixed (also because of transmission issues) and certain limitations of accuracy exist. Therefore, all presented figures and thresholds are real and accurate, given the resources. For a more mathematical and theoretical treatment in the future, such as the derivation of the profitability envelope or an approximation of an exact function, a smoother construction of the curve by interpolation seems appropriate.
Passengers (log. Scale)
1,000 10,000 100,000 1,000,000 10,000,000 100,000,000

500

Break-Even Line

-1,000
Profitability Envelope 2002 2002

Profitability Envelope 2004 Profitability Envelope 2006 Profitability Envelope 2008 Profitability Envelope 2010

2004 2006 2008 2010 -2,000

-1,500

-2,500

-3,000

Figure 5: Sample Airports Profitability Trend by selected years (PPP-adjusted base Norway in 2010 prices; 2010 without Italy and France) (Source: Own survey data) By dividing the sample under study into years we are able to derive more precise benchmarks for particular years which can be related to the observations. Figure 5 gives an overview of the profitability envelope change for even years between 2002 and 2010. As it can

EBIT per Passenger in NOK (2010 prices)

-500

be seen the part of the envelope below the break even line shifts downwards, which could be interpreted, that, in general, the profitability worsened for the low-demand airports. The breakeven point moves right suggesting that a higher level of passengers is required to increase the chances of breaking even on the EBIT level, while the top part of the envelope stagnates at average EBIT levels of 60 to 100 NOK per PAX, with London-City (LCY) airport with 144 NOK per passenger being the exception. In 2002 airports on the lower tail were more profitable than in the following years. The break-even point is located between such low passenger volumes as 17,680 and 63,000 defined by French airports AUR and EGC. Already in 2003 the break-even point significantly shifts to the right towards larger passenger volumes and is then situated somewhere between the observations Pescara (PSR) airport with about 300,000 passengers and an average EBIT of -3 NOK per passenger and Forli (FRL) airport (both in Italy) with about 350,000 passengers and an average EBIT of 4 NOK per passenger. In the years 2004 to 2009 the break even point lies between top performing French airports in the range of between 180,000 and 290,000 passengers. In 2010, where Italian and French airport data is missing, the break even point is shifted even more significantly to the right and lies approximately between German airport Friedrichshafen (FDH) with 590,000 passengers and an average EBIT of -23 NOK per PAX and British airport Exeter (EXT) with 737,000 passengers and an average EBIT of 12 NOK per PAX. This leads to the conclusion that France is managing its low-demand airports quite well, while achieving break even with around 200,000 passengers.
Number of Times defining the Envelope 9 8 8 7 6 5 5 5 5 4 4 4 3 3 3 3 3 3 2 2 2 2 2 2 2 2 IATA Code Benchmark Airports

Country

2003

2002

2004

2005

2006

2007

2008

2009 1 1 1 1 1 1 1 1 1

Tiree Aurillac Barra Dinard-Pleurtuit-Saint-Malo London City Bournemouth Bergerac-Roumaniere Southampton Graz London Heathrow Stavanger Fagernes Caen-Carpiquet Berlevg Svolvr Rennes Exeter Hasvik Bristol Lorient-Lann-Bihoue Stokmarknes Rst rsta-Volda Calvi-Sainte-Catherine Vads La-Rochelle-Ile De Re

TRE AUR BRR DNR LCY BOH EGC SOU GRZ LHR SVG VDB CFR BVG SVJ RNS EXT HAA BRS LRT SKN RET HOV CLY VDS LRH

HIAL France HIAL France UK UK France UK Austria UK Norway Norway France Norway Norway France UK Norway UK France Norway Norway Norway France Norway France

1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1

1 1 1 1 1 1 1 1 1 1 1

1 1 1 1

1 1 1 1 1 1 1

1 1 1 1 1 1 1

1 1 1 1 1

1 1 1

1 1 1 1 1 1 1 1 1

1 1

1 1 1

1 1 1 1 1 1 1 1

Table 3: Airports frequently defining the profitability envelope (2010 without Italy and France)

2010 1 1 1 1 1 1 1 1 1 1 1

10

The profitability benchmarks back the argument, that volatile airports, which achieve profitability in certain years, should receive special attention from the airport operator towards increasing profitability and/or demand (for example by pushing route development) in order for the airport to reach or remain profitable in the long-term. The observed trend of the profitability envelope is clearly downwards for low demand airports below one million passengers and upwards for the profitable airports above the break-even point. Obviously, profitability on the EBIT level could mainly be achieved either by reducing operating costs, such as staff costs, and/or by increasing charges in cases of high elasticity (> than 1) of demand (see RamseyPricing). Furthermore, profitability may be increased by a higher degree of commercialization, such as revenues from non-aeronautical sources, and/or by reducing investments, hence depreciation. The ideal balance of these factors should ideally be under the responsibility of the airport management, in order to set motivating incentives. In such cases where the airports lie below the profitability benchmarks (groups two and four), we do not expect the airports to have positive EBIT per se, but rather to achieve the according benchmark by analyzing the business model of its best-practice peers. Furthermore, I calculated the potential relative efficiency gains for each airport, if the according profitability benchmarks of other European airports would be reached. Reaching these gains means that increasingly less profit must be paid in cross-subsidies to loss-making airports in the system and increasingly more profits could be paid in dividends (or would be subject to taxes, hence would increase social welfare). We highly recommend a long-term sustainable efficiency policy, 1) by continuously benchmarking with comparable peer airports, 2) by analyzing business models of national and international best-practices and 3) by trying to adjust profits (hence, finding the ideal balance between revenues and costs) to the bestpractices.
Percentage Rank of Sample of 140 European Airports ordered by EBIT

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

Cum EBIT 2005 = 0.145x6 - 68.289x5 + 12,802.876x4 - 1,219,443.284x3 + 62,794,660.032x2 - 1,700,218,782.785x + 17,423,475,065.827 R = 0.985 Cum EBIT incl. eff. Gain 2005 = 0.183x 6 - 87.620x5 + 16,760.816x4 - 1,654,500.889x3 + 91,136,901.084x2 - 2,803,133,373.909x + 40,023,314,477.013 R = 0.996 Cum EBIT 2007 = 0.150x6 - 71.996x5 + 13,831.015x4 - 1,361,985.164x3 + 73,299,220.107x2 - 2,091,814,403.661x + 23,326,803,710.582 R = 0.997

60,000

Cum EBIT incl. eff. Gain 2007 = 0.232x6 - 113.289x5 + 22,391.200x4 - 2,319,496.593x3 + 136,198,580.808x2 - 4,474,370,714.490x + 66,612,887,900.123 R = 0.997 Cum EBIT 2009 = 0.117x6 - 55.216x5 + 10,383.057x4 - 998,314.971x3 + 52,302,272.173x2 - 1,445,033,037.298x + 13,612,692,500.415 R = 0.996 Cum EBIT incl. eff. Gain 2009 = 0.183x 6 - 88.170x5 + 17,043.047x4 - 1,713,834.180x3 + 97,505,968.200x2 - 3,152,336,965.074x + 47,980,934,444.730 R = 0.998
50,000

40,000

Cumulative EBIT in 2005 Cumulative EBIT in 2007 Cumulative EBIT in 2009

Cumulative EBIT incl. efficiency gains in 2005 Cumulative EBIT incl. efficiency gains in 2007

30,000

20,000

Cumulative EBIT incl. efficiency gains in 2009 10,000

-10,000

Figure 6: Cumulative EBIT (in NOK, 2010 prices) with and without potential efficiency gains for 140 European Airports 2005, 2007, 2009 (Source: Own survey data)

Earnings before Interests and Taxes (EBIT) in Norwegian Kroners (2010 prices)

70,000

Millions

80,000

11

As a consequence, cross-subsidizations will be minimized from profitable to loss-making airports, particularly in a national system or in a multi-airport organization. This effect is easiest shown by a Lorenz-like cumulative curve for a sample of 140 European airports (of the 210 airports in appendix A), showing cumulative EBIT (ordered from smallest to largest). In Figure 6 it can be observed that 80% to 90% of the European airports are not profitable. In 2009 only the top 13% of the airports made a profit after depreciation. Overall the 140 airports made a substantial system profit of 12 to 22 billion Kroners (1.5 to 2.75 billion Euro21). This picture changes significantly, if we account for the potential efficiency gains, which amount to about 28 to 38 billion Kroners (3.5 to 4.75 billion Euro) in absolute terms, or increases of between 233 to 317%. The order of magnitude of these calculations shows that efforts to increase efficiency are a worthwhile endeavor, especially for larger systems of airports. When including the feasible profitability gains in the cumulative distribution curves in Figure 6, the immediate relief on the loss-making airports underneath the break-even line can be seen. With the gains included in the 2009 figures, significantly less overall cross-subsidies would be required for the loss-making airports. In this case the proportions are much different and about 50% of the 140 airports report profits. In more detail the results for the 140 European airports are as follows: In 2005, 54% of the airports report losses, which are compensated by the profits of the next 36% higher ranked airports. The break-even point lies at the rank of the top 18% of the airports. With efficiency gains in place the proportions would change significantly and 33% of the airports would report losses. These losses are compensated by the next 22% higher ranked airports, therefore the break-even lies at 55%. The top 45% airports in this case contribute to the system profits of 41.3 billion Kroners. In the strong financial year of 2007 the efficiency gains have also the most significant impact on the feasible cumulative profits. In this year, 55% of the airports report losses, which are offset by 25% higher ranked airports which are reporting profits. We find the break-even point at around the top 20% of the sample, which generate the accounted profits of around 23.2 billion Kroners. If the efficiency gains are included in the 2007 data, we could expect the largest increases in system profitability. Again, 33% (100% 67%) of the lowest ranked airports would report just losses, which are compensated by profits of the next 13% of airports to reach the system equilibrium. In 2007, 50% of all airports, if managed efficiently, could contribute to the total system profit of 67.4 billion Kroners. The system profits would almost triple by these profitability improvements. In this sense, large airport operators could a priori postulate clear strategic management targets, whereas its achievements may a posteriori be evaluated, for example in cost-benefit analyses regarding policy changes such as changes in the structure of airport charges schemes. These efforts should be continuously conducted and closely monitored.

V. Conclusions and Outlook It is the aim of this article to present a method of comparison and to spot best-practice airports with regard to profitability. We wanted to give some numerical evidence of managerial effectiveness (Doganis and Thompson, 1975) and airport performance. It was recognized that a benchmarking approach needs the right choice of peers to give meaningful results, therefore a continuous scale rather than an arbitrary classification has been chosen, where we find the right peers along the scale. Thereby particular justice is done towards small airports, especially in the region below 100.000 PAX, which have very different characteristics compared to airports with, say, 100.000 to 1 million PAX and above. As it was shown with empirical data, airports with more than 1 million PAX can be expected to operate above break-even, thus making profits.
Average nominal exchange rate in 2010: 1 Norwegian Kroner 0.124921205 Euro (Source: OECD; see footnote 10)
21

12

These profit making airports show sharply increasing returns-to-scale on the absolute EBIT level, although the marginal returns-to-scale are decreasing. This means that the effort to exploit each additional passenger is much higher at large airports, than it is at small airports, where it may be much easier to increase the revenues per passenger. However, any small increase in revenues per PAX leads directly to increasing profits at large airports, because fixed costs are already covered. With the construction of a maximum possibility frontier, which here is named profitability envelope, it is relatively straightforward to generate appropriate benchmarks and to recognize trends and shifts of the curve over time. In this way, airport management can formulate realistic strategic targets towards profitability. To be able to replicate this analysis, large amounts of data are needed, which mostly do not lie in the public domain. This is a major drawback of the presented method, because the viewpoint may be only one-dimensional. However, from the airport management perspective just peers with comparable PAX figures are required, which lowers the data collecting effort drastically. Therefore, our method is more suitable for large airport system operators, regulators or other large scale decision-makers. If the fundamental effects, which are presented in this paper, are understood, it may seem reasonable for future research to aggregate the numerical evidence further (as was exemplarily shown with the cumulative curves in Figure 6) towards even fewer control variables, such as the Gini-Coefficient, to measure the changes in the distribution of airport profits in a multi-airport environment.

13

Literature Adler, N., Liebert, V., and Yazhemsky, E. (2012), Benchmarking airports from a managerial perspective, Omega, ISSN 0305-0483, doi: 10.1016/j.omega.2012.02.004 [accessed on June 4th 2012 at: http://userpage.fu-berlin.de/~jmueller/gapprojekt/downloads/gap_papers/bmarking.pdf] Doganis, R.S., and Thompson, G.F. (1975), Airport Profitability and Managerial Effectiveness, The Manchester School, 43: 331352. doi: 10.1111/j.1467-9957.1975.tb01236.x. Dorfman, R. (1951), Application of the Simplex Method to a Game Theory Problem, in: Activity Analysis of Production and Allocation - Proceedings of a Conference, Edited by T.C. Koopmans, John Wiley & Sons, Inc., New York, Chapman & Hall, Limited, London, pp. 348358 [accessed on June 3rd 2012 at: http://cowles.econ.yale.edu/P/cm/m13/m13-22.pdf] Gillen, D., and Lall, A. (1997), Developing measures of airport productivity and performance: an application of data envelopment analysis, Transportation Research Part E: Logistics and Transportation Review, Volume 33, Issue 4, December 1997, pp. 261-273, ISSN 1366-5545, doi: 10.1016/S1366-5545(97)00028-8 [accessed on June 3rd 2012 at: http://www.wilfridlaurier.ca/documents/4296/dea_airport_paper.pdf] Gritta, R., Adams, B., and Adrangi, B. (2006), An Analysis of the Effects of Operating and Financial Leverage on the Major U.S. Air Carriers' Rates of Return: 1990-2000, The 47th Annual Transportation Research Forum (TRF) on Mar. 23-25, 2006, New York University, New York, NY [accessed on June 3rd 2012 at: http://www.trforum.org/forum/downloads/2006_8A_FinLeverage_paper.pdf] Heymann, E., and Vollenkemper, J. (2005), Expansion of regional airports: Misallocation of resources, Deutsche Bank Research, Frankfurt am Main [accessed on June 4th 2012 at: http://www.dbresearch.com/PROD/DBR_INTERNET_ENPROD/PROD0000000000193311.PDF] Hyndman, L. D. (1992), Directions: The Final Report of the Royal Commission on National Passenger Transportation, Volume 2, pp. 92-108, Ottawa, Canada, ISBN 0-660-14545-6 [accessed on June 4th 2012 at: http://www.e-ajd.net/source-pdf/Directions-VOL2-eng.pdf] Koopmans, J.-P. (2008), The Impact of Commercial Activities on Airport Financial Performance (An Analysis of 35 Airports), Master Thesis submitted to the Faculty of Economics and Business Administration, Maastricht University, Netherlands [accessed on May 19th 2012 at: http://arno.unimaas.nl/show.cgi?fid=15656] Reinhardt, U. E. (1973), Break-Even Analysis for Lockheed's Tri Star: An Application of Financial Theory. The Journal of Finance, 28: 821838. doi: 10.1111/j.15406261.1973.tb01408.x [accessed on June 3rd 2012 at: http://www.estg.ipleiria.pt/files/350738_BREAKEVE_4628e59997e2a.pdf] Vogel, H.-A., and Graham, A. (2010), Airport valuation: an alternative driver-based approach. Journal of Air Transport Studies, 1 (1), pp. 20-47. ISSN 1791-6771 [accessed on June 4th 2012 at: http://www.airtransportnews.aero/content/jats/jats11.pdf]

14

Appendix A:
IATA GRZ INN LNZ SZG VIE PRG AAL AAR BLL CPH FAE TLL AGF AJA AUR AVN BES BIA BIQ BOD BVA BVE BZR CCF CFE CFR CHR CLY DCM DNR DOL EBU EGC FNI FSC GNB LDE LEH LIG LIL LRH LRT LYS MLH MPL MRS NCE NCY NTE PGF PIS PTP PUF RDZ RNS SXB TLN TLS TUF UIP Airport Name Thalerhof Kranebitten Hoersching W. A. Mozart Vienna International Prague-Ruzyne Aalborg Tirstrup Billund Copenhagen Vagar Ulemiste La Garenne Campo Dell Oro Aurillac Avignon-Caum Guipavas Poretta Biarritz Parme Bordeaux Beauvais-Tille Laroche Beziers Vias Salvaza Aulnat Carpiquet Chateauroux Sainte Catherine Mazamet Pleurtuit Saint Gatien Boutheon Roumanieres Garons Sud Corse Grenoble-Isere Tarbes Ossun Lourdes Octeville Bellegarde Lesquin Laleu Lann Bihoue Lyon Saint-Exupery International EuroAirport French Mediterranee Marseille Cote D'Azur Annecy-Meythet Nantes Atlantique Llabanere Biard Le Raizet Uzein Marcillac St Jacques Entzheim Hyeres Blagnac St Symphorien Pluguffan Country Austria Austria Austria Austria Austria Czech Republic Denmark Denmark Denmark Denmark Denmark Estonia France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France France IATA HAJ HAM HHN LEJ MUC NUE PAD SCN STR ATH BUD KIR AHO AOI BGY BLQ BZO CAG CRV CTA CUF FLR FRL GOA GRS LCV LIDE NAP OLB PEG PMF PMO PSA PSR QSR REG RMI SAY SUF TPS TRN TRS VCE VRN MLA AES ALF ANX BDU BGO BJF BNN BOO BVG EVE FDE FRO HAA HAU HFT Airport Name Hanover Fuhlsbuettel Frankfurt - Hahn Leipzig/Halle Franz Josef Strauss Metropolitan Area Paderborn Ensheim Echterdingen Eleftherios Venizelos International Liszt Ferenc International Kerry County Fertilia Falconara Orio Al Serio Guglielmo Marconi Bolzano Elmas Crotone Fontanarossa Levaldigi Peretola Luigi Ridolfi Cristoforo Colombo Baccarini Lucca Reggio Emilia Naples Costa Smeralda Sant Egidio Parma Punta Raisi Gal Galilei Liberi Salerno Costa d'Amalfi Tito Menniti Miramare Siena S Eufemia Birgi Citta Di Torino Dei Legionari Marco Polo Verona Luqa Vigra Alta Andenes Bardufoss Flesland Batsfjord Bronnoy Bodo Berlevag Evenes Bringeland Flora Hasvik Haugesund Hammerfest Country Germany Germany Germany Germany Germany Germany Germany Germany Germany Greece Hungary Ireland Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Italy Malta Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway IATA MOL MQN NVK OSL OSY RET RRS RVK RYG SDN SKN SOG SOJ SSJ SVG SVJ TOS TRD TRF VAW VDB VDS VRY BTS LJU THN BLP BSL GVA LUG ZRH ABZ BEB BFS BHX BLK BOH BQH BRR BRS CAL CVT CWL DND DSA EDI EMA EXT GLA HUY ILY INV KOI LBA LCY LGW LHR LPL LSI LTN Airport Name Aro Mo I Rana Framnes Oslo Namsos Stolport Roros Ryumsjoen Moss-Rygge Sandane Skagen Haukasen Sorkjosen Stokka Sola Helle Tromso/Langnes Vaernes Sandefjord Vardoe Valdres Vadso Stolport Ivanka Brnik Trollhattan Bellavista EuroAirport Swiss Geneve-Cointrin Lugano Zurich Dyce Benbecula Belfast International Birmingham International Blackpool Bournemouth Biggin Hill North Bay Bristol Machrihanish Baginton Cardiff-Wales Dundee Doncaster Sheffield Turnhouse East Midlands Exeter Glasgow International Humberside Glenegedale Inverness Kirkwall Leeds/Bradford London City Gatwick Heathrow Liverpool John Lennon Sumburgh Luton Country Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway Slovakia Slovenia Sweden Switzerland Switzerland Switzerland Switzerland Switzerland United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom

URO BRE CGN DRS DTM DUS ERF FDH FMO FRA

Boos Bremen Koeln/Bonn Dresden Dortmund Dusseldorf Erfurt Friedrichshafen Muenster Frankfurt International

France Germany Germany Germany Germany Germany Germany Germany Germany Germany

HOV HVG KKN KRS KSU LKL LKN LYR MEH MJF

Hovden Valan Hoeybuktmoen Kjevik Kvernberget Banak Leknes Svalbard Mehamn Kjaerstad

Norway Norway Norway Norway Norway Norway Norway Norway Norway Norway

MAN MME NCL NWI SEN SOU STN SYY TRE WIC

Manchester International Durham Tees Valley Newcastle Norwich Southend Municipal Southampton Stansted Stornoway Tiree Wick

United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom

Appendix A: Airport sample list

16