Journal of Indian Business Research

Emerald Article: Does financial outreach engender economic growth? Evidence from Indian states Saibal Ghosh

Article information:
To cite this document: Saibal Ghosh, (2011),"Does financial outreach engender economic growth? Evidence from Indian states", Journal of Indian Business Research, Vol. 3 Iss: 2 pp. 74 - 99 Permanent link to this document: http://dx.doi.org/10.1108/17554191111132206 Downloaded on: 30-03-2012 References: This document contains references to 71 other documents To copy this document: permissions@emeraldinsight.com This document has been downloaded 339 times.

Access to this document was granted through an Emerald subscription provided by UNIVERSIDADE FEDERAL DA BAHIA For Authors: If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service. Information about how to choose which publication to write for and submission guidelines are available for all. Additional help for authors is available for Emerald subscribers. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com With over forty years' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in business, society, public policy and education. In total, Emerald publishes over 275 journals and more than 130 book series, as well as an extensive range of online products and services. Emerald is both COUNTER 3 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1755-4195.htm

JIBR 3,2

Does nancial outreach engender economic growth? Evidence from Indian states
Saibal Ghosh
Reserve Bank of India, Mumbai, India
Abstract
Purpose Employing data on 14 major Indian states during 1973-2004, this paper aims to investigate the hypothesis that economic growth is affected by nancial outreach. Design/methodology/approach The paper employs univariate tests as well as advanced panel regression techniques to examine whether nancial outreach matters for state-level economic growth. Findings The analysis suggests that improvements in nancial outreach led to a perceptible rise in per capita growth. In terms of magnitudes, a rise in demographic outreach by 10 percent raises state per capita growth by 0.3 percent; in case of geographic outreach, the increase is lower. Finally, the analysis supports the hypotheses that states with higher manufacturing share tend to grow faster and the quality of state-level institutions and infrastructure exert a signicant bearing on growth. Research limitations/implications Although the denitions of nancial outreach are based on international best practice, they focus only on banks and are driven by the availability of data on relevant variables. Practical implications The article belongs to the broad strand of literature which examines the nance-growth nexus. Social implications Financial outreach is presently an avowed objective of policymakers, both in India and elsewhere. The article examines which sets of economic/policy variables impact nancial outreach. The analysis can provide policymakers with feedback as regards the feasibility of the strategies pursued to improve nancial outreach and thereby, how best to redesign and ne-tune them. Originality/value To the authors knowledge, this is presumably the rst study in India to examine the nancial outreach-growth nexus in a systematic manner at the sub-national level. Keywords Economic growth, Banking, Financial economics, India Paper type Research paper

74

1. Introduction The extent of nancial outreach is often regarded as a critical factor in making nancial products and services available to a wider segment of the population. This is all the more relevant in emerging economies where such facilities typically tend to exclude vast segments of the population, especially the underprivileged sections of society. Cross-country data suggest that in several African economies, there is less than one bank branch per 100,000 people, while in developed economies, these numbers are quite high (Beck et al., 2007).
Journal of Indian Business Research Vol. 3 No. 2, 2011 pp. 74-97 q Emerald Group Publishing Limited 1755-4195 DOI 10.1108/17554191111132206

The author would like to thank, without implicating, the Area Editor of the journal and, especially, an anonymous referee for the comments and insights on an earlier draft that improved the exposition and analysis. Needless to state, the views expressed and the approach pursued in the paper are strictly personal.

In large federal structures, an additional dimension is introduced by the existence of component federal states with their democratically elected governments. The latter, in effect, provides a convenient anchor for studying sub-national dimensions of policy actions. Since the nation comprises of several states with not only differential growth patterns, but also differential abilities to respond to macro policies, it would therefore be of interest to understand the extent of such reactions at the sub-national level[1]. Towards this end, the paper chooses India as a case study and explores the impact of nancial outreach on state-level economic growth. Borrowing from the recent literature in this area (Beck et al., 2007), it utilizes a consistent set of indicators of nancial outreach and explores their empirical association with state-level economic growth, after controlling for other state-level determinants of growth. The results indicate that improvements in nancial outreach, as measured in terms of expansion and use of nancial services, have a salutary impact on per capita economic growth. In fact, it is the high income, coastal states that exhibit relatively higher growth, which suggests a gradual divergence in growth patterns across regions. Besides, growth is higher in states that have bigger (in terms of size) state governments. As well, the quality of state-level institutions and infrastructure are found to be important factors which inuence state-level economic growth. Although preliminary, the analysis also suggests the growing role of technology as an important factor in promoting greater nancial outreach. The paper makes several contributions to the literature on nancial outreach and economic growth. First, the paper is linked to a large body of literature that emphasizes the importance of nance in inuencing economic growth. Cross-country studies (Levine and Zervos, 1998; Aghion et al., 2005a, b) and also evidence at the industry (Rajan and c-Kunt and Maksimovic, 1998) offer persuasive Zingales, 1998) and rm level (Demirgu evidence that various measures of nancial development have a positive effect on economic growth (see Levine (2005) for a survey). Even within a country, local nancial development matters for growth because it engenders increased entry of rms and greater growth of existing ones (Guiso et al., 2004). Judged thus, the paper complements cross-country studies with an in-depth country study. Second, the paper augments the evolving literature on nancial outreach. In emerging markets, owing to a lack of adequate institutional and informational infrastructure, imperfections in nancial markets are quite pervasive. The net effect of such asymmetries is felt particularly by small entrepreneurs with limited collateral backing. As a consequence, the importance of well-developed nancial systems for economic growth and poverty alleviation can hardly be over-emphasized. Access to nance can also have important effects on technological progress and the creation of ideas; with an absence of nancing ideas, the incentive for innovation is much lower (King and Levine, 1993). Finally, outreach is also intricately linked to human development. Financial literacy is a critical ingredient of nancial outreach, since it enables individuals to make informed decisions about their nancial choices. The sub-prime crisis, in large part, is believed to have been the result of inadequate awareness on the part of consumers to understand the nuances of complex nancial products (Bernanke, 2008). To the extent that better education feeds into better health (Grossman, 1972) and vice versa, the net effect of higher outreach is reected in better human development[2]. The index of nancial outreach developed by Honohan (2007) appears to corroborate this point. Third and more generally, the paper belongs to the class of literature that explores the issue of nancing constraints. Most studies exploit the World Business Environment

Financial outreach

75

JIBR 3,2

76

Survey a cross-sectional rm-level survey data covering 80 developed and developing countries to examine this issue. Clarke et al. (2001), for example, investigate the impact of foreign bank penetration on SME lending. They nd that foreign bank penetration increases the share of nancing from banks and lowers nancing obstacles as perceived by rms, particularly in case of large rms. Love and Mylenko (2003) explore how credit-reporting institutions affect nancing constraints. Their research appears to suggest that private credit registries relax nancing constraints and increase bank nancing, particularly for SMEs. Beck et al. (2003), study the impact of bank concentration on rms nancing obstacles and access to credit. They nd that in countries with low levels of institutional development, bank concentration leads to higher nancing obstacles and a lower share of bank nancing, particularly for SMEs. Finally, Galindo and Micco (2005) study the effect of several measures of creditor protection on the share of nancing from banks and nd that creditor rights improve access to nancing for SMEs. Finally, the paper belongs to the literature that explores the sub-national effects of greater nancial outreach and to a wider literature which examines the impact of economic policies on sub-national economies, both for India (Sachs et al., 2002; Besley and Burgess, 2004; Aghion et al., 2003, 2005a, b; Mitra and Ural, 2007; Hasan et al., 2007; Bhandari, 2009) and elsewhere (Carlino and Dena, 1998; Arnold and Vrugt, 2002; Wachtel et al., 2006). Many of these studies have identied important effects of varied legislations on sub-national growth performance. Although certain studies focus on the nance-growth interface (Burgess and Pande, 2005; Topalova, 2008), few studies assess the effects of nancial outreach on sub-national economic growth. Even more importantly, none have examined the effects of different measures of nancial outreach on sub-national growth and this is one of the major concerns of the paper. The choice of sub-national governments is premised on three major considerations. First, like the USA and other emerging economies, such as Argentina and Brazil, India is a federal polity comprising of states with their own governments and a measure of policy autonomy. Over time, these states have developed distinct characteristics, driven by a combination of their geographical location and economic policies pursued. In this process, the feasibility of the strategies pursued to increase nancial outreach can provide practitioners with useful leads to redesign and ne-tune their strategies. Second, the data comparability issues are less serious within a country than across countries. While the comparison of institutional and nancial characteristics across countries can be difcult owing to wide divergences in their historical experiences, institutional characteristics and the legal environment, sub-national data can control for such contexts and thereby reduce the unobserved heterogeneity at the cross-section level. Third, India is one of the few important emerging economies for which a comprehensive and reliable state-level database is available. The cross-sectional and time-varying nature of the data makes it amenable to rigorous statistical analysis. To the extent that such studies bypass the limitations inherent in cross-country studies (Rodrik, 2005), it provides an ideal laboratory to explore the effects of nancial outreach on sub-national economic growth and could be representative of such association in other emerging markets. The paper comprises of six sections after this introduction. Section 2 presents an overview of the literature with reference to India. This is followed by a description of the stylized facts for the Indian states with emphasis on nancial outreach.

The data and variables and the regression specication are detailed in the subsequent two sections. Section 6 discusses the results along with some robustness tests, followed by the concluding remarks. 2. Financial inclusion: overview and Indian experience This paper is related to an emerging literature on access to nancial services. Extant research analyzes access to nancial services at the rm (Beck et al., 2006b) and bank level (Beck et al., 2006a). More recently, Beck et al. (2007), present aggregate cross-country data on banking sector outreach (such as branch and ATM penetration, deposits per capita, and loans per capita) and show that these indicators closely track more difcult and costly to collect micro-level statistics of household and rm use of banking services. A recent study by the European Commission (2008) classies the 25 EU economies in terms of their level of nancial exclusion and nds several transition economies (Hungary, Poland, Lithuania, and Latvia) experiencing maximum nancial exclusion. The reported gures for the USA are much higher, with anywhere between 9.5 and 20 percent of households having no access to bank accounts (His Majestys Treasury, 2004). In the Indian case, the foundation for broad-basing the institutional credit structure and promoting greater nancial access can be traced to the ndings of the All-India Rural Credit Survey (RBI, 1954). The ndings of the survey indicated that, out of the total borrowings of farmers in 1951-1952 estimated at Rs 7,500 million, commercial banks provided less than 1 percent, while moneylenders provided 70 percent. The distribution of bank branches was also highly skewed, with nearly 38 percent of the bank branches being located in urban and metropolitan/port town locales. The distribution of bank credit was also highly skewed with an overwhelming proportion of credit being cornered by private corporate business. These and several other disquieting features in the allocation of bank credit eventually culminated in the process of bank nationalization in July 1969. In essence, the state took control of the banking sector and made it a tool for promoting social objectives. Salient elements of the process included control over interest rates and dovetailing of lending towards priority sectors. A critical ingredient of this strategy entailed the imposition of the 1:4 license rule in 1977, wherein, banks could open a branch in a location with one or more branches only if it had opened four in a location with no branches (unbanked location). Thus, over the period 1969-1991, over 50,000 new bank branches were built, predominantly in rural locales (Table I). As Burgess and Pande (2005) demonstrate, by improving access to cheap formal credit for the rural poor, this redistributive nature of branch expansion strategy made a signicant dent on rural poverty. The second phase of public policy towards promoting greater nancial access can be traced to the inception of nancial sector reforms. Salient features of this period included a higher allocation of credit to the private sector, moving away from administered to market-determined interest rates both for commercial and government borrowing, increased competitiveness, and liberal entry of foreign banks (Chairlone and Ghosh, 2008). In a sense, the period demonstrated that policies for inclusive banking have to exist concurrently with encouraging strong and efcient nancial institutions. Three important features of the strategy towards promoting inclusive banking during this period deserve mention. The rst was the initiation of the Self Help Group Bank linkage program in 1992. After a tepid beginning, the program gathered momentum thereafter with increasing outreach both in coverage and bank loan disbursement.

Financial outreach

77

JIBR 3,2

Indicators Number of commercial banks Of which: regional rural banks (RRBs) Number of bank ofces

June 1973 83 15,362 11,282 36 92 167 64 117 24 110

December 1980 154 107 34,594 (4,471) 23,227 15.1 16 404.4 738 250.8 457 36 710.8

March 1991 272 196 60,570 (14,519) 46,115 8.9 14 2,011.9 2,368 1,218.7 1,434 48.1 3,275.2

March 1998 300 196 66,408 (14,471) 47,130 0.31 15 6,054 6,270 3,241 3,356 46.4 5,215.4

March 2004 290 196 69,170 (14,446) 47,766 0.22 16 15,044 14,089 8,408 8,273 60 11,516.2

78

Table I. Commercial banking in India: progress since 1973

Of which: rural/semi-urban branches Annual growth rate (%) of rural/semi-urban branches Population per bank ofce (000s) Deposits of commercial banks (Rs billion) Per capita deposit (Rs) Credit of commercial banks (Rs billion) Per capita credit (Rs) Deposits/national income (%) Total Assets (Rs billion)

Note: Figures in parentheses are branches of RRBs Source: RBI (various years (a)(b))

The second was the establishment of the Rural Infrastructure Development Fund during 1995-1996 towards completion of on-going projects of rural infrastructure. This was complemented by strengthening the capital base of the National Bank for Agriculture and Rural Development (NABARD)[3]. Furthermore, a system of Kisan Credit Cards (Kisan, meaning farmer) was initiated in 1999 to provide adequate and timely nancial support in a exible and cost-effective manner from the banking system to farmers for cultivation needs and purchases. On a broader plane, the Reserve Bank has adopted a two-pronged strategy to expand the reach of banking services. The rst strategy, termed empowerment, entails inculcating awareness among the masses about nancial products through nancial education along with supporting advisory mechanisms (e.g. credit counseling). The second one, termed protection, envisages a comprehensive code of conduct for minimum standards of banking services to be offered by banks with closer and continuous regulatory oversight. 3. Financial outreach in India: data and summary statistics We begin the analysis by providing an overview of nancial outreach across states. Following from Beck et al. (2007), we utilize the following indicators of banking sector outreach at the state level: (1) Geographic outreach: number of bank branches per 1,000 km2. (2) Demographic outreach: number of bank branches per 100,000 people. (3) Loan accounts per capita: number of loan accounts per 1,000 people. (4) Deposit accounts per capita: number of deposit (aggregate of savings, term, and current) accounts per 1,000 people. (5) Loan-income ratio: average size of loans to per capita net state domestic product (PCNSDP). (6) Deposit-income ratio: average size of deposits to PCNSDP.

In particular, we focus exclusively on banking outreach for two major reasons. First, in a majority of countries including India, the banking sector intermediates most of the funds in the economy. Second, the statistical information for this sector is easier to obtain as compared to other non-bank service providers. Indicators (1) and (2) measure the outreach of the nancial sector in terms of access to banks physical outlets. These measures, however, have limitations as indicators of access to physical banking outlets. More importantly, these measures implicitly assume a uniform distribution of bank outlets within a countrys area and across its population. In reality, bank branches and ATMs could be concentrated across population groups, delimiting its utility in certain cases. To overcome this drawback, indicators (3) through (6) measure the use of banking services. We focus on bank deposits and loans because these are the main services offered by banks for which information is available. Table II reports the correlations of growth in PCNSDP with these various measures described above. What is striking is the fact that the correlation of the growth indicator with measures of nancial outreach has, in fact, weakened over the reform period. By way of example, the correlation of PCNSDP growth with bank ofce per 100,000 people was 54 percent for the entire period and 56 percent in the pre-reform regime. In contract, the correlation was insignicant in the post-reform era. What this indicates is a possible weakening of the growth-nancial outreach nexus, especially in the post-reforms era. We also examine the extent of nancial outreach across states. For expositional simplicity, we classify the states by three-fold criteria: income, region and geography. In our subsequent analysis, we employ dummies to examine differences in nancial outreach across these classications. Specically, high-income states are as dened by the The World Bank (2005) and corroborate the earlier classication to this effect by Sachs et al. (2002) and Ahluwalia (2002). Likewise, states have also been classied according to regions, following RBI (2008) and nally, as coastal or land locked (Government of India, 2008a)[4]. The evidence indicates signicant differences in both geographic and demographic outreach across high- and low-income states; similar evidence is also manifest in the case of deposit and loan accounts. There is also evidence to suggest limited use of deposit services in the low-income states (Beck et al., 2007). By way of example, the mean deposit/income ratio in the low-income state is 3.78 as compared to 2.54 in the high-income states. The difference is statistically signicant at the 0.05 level. In terms of regional divergence, the differences in demographic outreach and deposit accounts per capita are most substantial. Illustratively, the number of bank ofces
1972-1973 to 2003-2004 0.321 * 0.535 * 0.568 * 0.588 * 20.554 * 20.641 * 1972-1973 to 1991-1992 0.368 * 0.562 * 0.661 * 0.618 * 2 0.663 * 2 0.544 * 1992-1993 to 2003-2004 0.037 0.242 0.276 * * * 0.189 0.032 2 0.026

Financial outreach

79

Financial outreach indicator Bank ofce/1,000 km2 Bank ofce/100,000 people Number of loan accounts per capita Number of deposits accounts per capita Loan income ratio Deposit income ratio

Note: Signicance at: *1, * *5 and * * *10 percent levels

Table II. Correlations of nancial outreach measures with PCNSDP growth

JIBR 3,2

80

per 100,000 people is the highest at 7.67 in the southern region as compared with 4.73 in the eastern region. This difference is statistically signicant at the 0.01 level. Differences in loan accounts per capita are signicantly higher in the southern as compared to other regions; both deposit-income ratio and loan-income ratio exhibits limited divergence across regions. The evidence is consistent with recent ndings which reports signicant regional variation in the provision of nancial services (Basu, 2006; Government of India, 2008b). Finally, across most indicators, coastal states exhibit higher levels of nancial outreach; only in the case of loan accounts per capita was the difference statistically signicant (Table III). The differences in the nancial outreach for the pre- and post-reform period are particularly striking. The evidence indicates a signicant improvement in both geographic and demographic outreach; as well, the values of both deposit- and loan-income ratios have declined, signifying the greater outreach of banking services (Beck et al., 2007). 4. Data and variables For the analysis, we use state-level data for the 30-year period, 1973-2004. We selected this time period for several reasons. First, it seems sufciently long to allow the long-run inuences to play out. Second, the period coincides with the availability of consistent data on the concerned variables at the state level to clearly discern the impact of nancial outreach on state level growth. And nally, the period of analysis is the post-bank nationalization period, which helps to clearly isolate the impact of economic policies uncontaminated by other confounding inuences. The data include 14 states in India, in line with the standard practice of comparing the economic performance of states that treats smaller or north-eastern states differently (Ahluwalia, 2002; Sachs et al., 2002; Nachane et al., 2002)[5]. Therefore, we exclude from the study the special category states that receive exceptionally generous grants from the Federal Government on account of their specic institutional characteristics (Rao and Singh, 2007). In the nancial year 2003-2004, these states accounted for roughly 80 percent of Indias land area, over 70 percent of her population and nearly 75 percent of the domestic product. Many of the variables in the model vary less over time. Therefore, rather than using annual data, following earlier analysis in this area (Grier and Tullock, 1989; Barro, 1997), we grouped the data into eight, non-overlapping, four-year time intervals. Therefore, we have eight sets of observations on each of the 14 states, yielding a total of 112 observations. The dependent variable is change in logarithm of real PCNSDP, which represents value added originating in each state. A similar variable was employed by Rodrik and Subramanium (2004) in their analysis of Indias growth trajectory during the last ve decades. The level of initial per capita income is included to test the convergence hypothesis: if there is convergence, states with higher income levels will tend to grow at a slower rate. The vector of state-level controls includes the ratio of manufacturing to NSDP (to examine the existence of interest rate channel). Rodrik and Subramanium (2004), for instance, nd a signicant role for registered manufacturing in explaining inter-state growth rates. Furthermore, the state-level literacy rate is included to control for the quality of educational attainment in the state.

Variables 7.15 (2.28) 5.39 (1.91) 4.36 * 7.48 7.67 5.92 4.73 6.06 2 0.31 2.22 * * 4.68 * 2.07 * * 2.81 * 5.97 * 2.66 * 1.92 * * * 2 0.21 2 3.21 * 6.57 (2.25) 6.46 (2.41) 0.24 6.04 (2.45) 7.32 (1.79) 2 3.15 * 4.15 (2.83) 1.06 (0.48) 8.93 * 24.42 (26.39) 13.54 (11.57) 6.59 * 257.54 (16.81) 442.52 (15.89) 2 5.83 * 2.75 (2.13) 14.83 (20.93) 3.30 (3.32) 18.94 (25.89) 21.01 20.89 349.82 (16.53) 296.35 (21.04) 1.46 64.64 (37.24) 38.08 (25.05) 4.51 * 43.57 (33.14) 69.40 (32.45) 24.05 * 64 48 70 42 0.06 21.09 22.14 * * 21.54 21.21 22.24 * * 21.64 21.21 20.51 0.70 22.17 * * 20.93 20.36 0.05 22.69 * 21.83 * * * 21.57 0.44 0.87 0.37 2 0.11 0.44 3.12 * 2.96 * 0.75 4.57 * 4.25 * 3.68 * 3.39 * 2 0.14 25.85 * 0.89 20.41 0.61 6.86 * 4.27 * 4.99 * 21.09 0.07 0.85 (2.35) (2.02) (2.06) (1.67) (1.96) 2.29 2.27 3.04 4.26 3.49 (1.84) (1.62) (2.27) (4.11) (2.77) 17.79 18.03 24.90 20.78 17.42 (21.39) (16.56) (24.77) (33.78) (23.58) 390.77 397.69 364.26 214.49 220.21 (24.56) (17.42) (12.63) (12.49) (11.30) 41.83 84.63 36.39 45.37 36.33 (22.12) (32.51) (16.22) (35.16) (31.14) 24 32 16 24 16 2.54 (1.99) 16.45 (22.21) 3.78 (3.55) 16.85 (25.10) 22.04 * * 20.08 399.92 (18.36) 195.48 (10.29) 7.55 * 61.75 (36.10) 37.97 (27.33) 3.92 * 72 40

Bank ofce/ 1,000 km2 Loan/income No. obs

Bank ofce/100,000 people

Deposit/ income

Deposit account/1,000 people

Loan account/1,000 people

28.77 (19.56) 14.70 (9.81) 5.06 * (15.63) (23.94) (5.25) (15.71) (9.87)

23.75 32.69 23.29 13.30 16.95

1.69 * * * 3.04 * 0.09 1.69 * * * 4.38 * 1.77 * * * 3.21 * 2 2.89 * 2 1.31 1.57

Panel A: income High income Low income t-test for difference Panel B: region Northern Southern Western Eastern Central t-test for difference Northern vs southern Northern vs western Northern vs eastern Northern vs central Southern vs western Southern vs eastern Southern vs central Western vs eastern Western vs central Eastern vs central Panel C: location Coastal Land locked t-test for difference Panel D: reforms Pre-reforms Post-reforms t-test for difference

25.63 (20.71) 21.23 (13.41) 1.36

19.41 (15.64) 30.97 (19.51) 2 3.26 *

Notes: Signicance at: *1, * *5 and * * *10 percent levels; standard deviation within parentheses

Financial outreach

81

Table III. Financial outreach across state characteristics: mean and standard deviation

JIBR 3,2

82

Several studies in the Indian context highlight the importance of infrastructure in impacting state output (Nagaraj et al., 2000; Ahluwalia, 2002; Kochhar et al., 2006). Kochhar et al. (2006), point out that the transmission and distribution (T&D) losses of state electricity boards, to the extent it is the outcome of state-level decisions, reect on both the quality of its institutions (lack of viability of state electricity boards) as also its infrastructure (high T&D losses reect low-power quality which directly impinges on manufacturing) and could end up dampening economic growth. Earlier studies on India do not control for state-level scal policy. Studies have documented that large government sectors tend to impede growth, ceteris paribus (Barro, 1991; Easterly and Levine, 2002). We control for this factor by incorporating the ratio of state budget decit to NSDP. State-level income data are derived from the Economic and Political Weekly States Database (EPW Research Foundation, 2003). Using these data, we construct an annual series on real net per capita income and the shares of agriculture, manufacturing and services, by appropriately splicing the three base-year series. Data on state-level scal variables are hand coded from various issues of annual reports on state nance, supplemented by the Handbook of Statistics on State Government Finances (RBI, 2004). Infrastructural variables are obtained from the Statistical Abstract, supplemented with annual reports of state electricity boards as available on the web site of the Indian Planning Commission. The data on the banking variables are culled from the Basic Statistical Returns, a yearly central bank publication which provides detailed state-level information on deposits and credit of commercial banks. The Appendix provides a description of the variables, data sources and summary statistics. 5. Regression analysis The univariate tests do not control for factors that might systematically impact state economic growth. For one, we do not account for state-level controls. The pace of economic activity could also be an important consideration. Another major concern is the possibility of endogeneity: one must determine that correlations between output and nance are due to output responding to nance and not the other way around. Taking these aspects on board, we estimate the effect of nancial outreach on state-level economic growth, employing regressions of the following form: Gr_PCNSDP s;t f Initial PCNSDP; Inclusion Control variables; Year dummies error 1

where s indexes state and t denotes year. In equation (1), the dependent variable (PCNSDP) is assumed to be a function of state-level controls (Control variables) including measures of its institutional structure, size, educational attainment and infrastructure. Year dummies are included to control for shocks to the state economy. The variable of interest is Inclusion. Under the hypothesis, that greater nancial outreach improves growth, we would expect the coefcient of this variable to be positive. The outreach of lagged dependent variable renders static panel estimation of equation (1) inconsistent. To address this concern, and also to overcome the potential endogeneity of other regressors, we employ panel GMM estimations. More specically, we employ the system GMM estimator due to Blundell and Bond (1998). We employ

the system GMM as compared to its competing rival, the difference GMM estimator, since the latter may suffer from weak instruments problems (Blundell and Bond, 1998). The reliability of the GMM estimation procedure depends on the validity of the instruments. We consider the validity of the instruments by presenting the Sargan test on over-identifying restrictions. It is asymptotically distributed as x 2 and tests the null hypothesis of validity of the (over-identifying) instruments. The p-values report the probability of incorrectly rejecting the null hypothesis, so that a p-value above 0.05 implies that the probability of incorrectly rejecting the null is above 0.05. As a result, a higher p-value makes it more likely that the instruments are valid. The consistency of the estimates also depends on the absence of serial correlation in the error terms. This will be the case if the differenced residuals display signicant negative rst-order serial correlation and no second-order serial correlation. We present tests for rst- and second-order serial correlations related to the estimated residuals in rst differences. The test statistics are asymptotically distributed as standard normal variables. The null hypothesis relates to insignicance, so that a low p-value for the test on rst-order serial correlation and a high p-value for the test on second-order serial correlation suggest that the disturbances are not serially correlated. 6. Results and discussion 6.1 Baseline results The results, reported in Table IV, show that we are not able to reject the Sargan test. Moreover, we are not able to reject the null hypothesis of no second-order serial correlation. In other words, the GMM model is well specied. In all cases, we run the model with and without the control variables. In specication (1), the coefcient on initial per capita income is 0.63, which suggests that initially poor states grow faster than rich ones. In other words, absent controls for differences in state policies and economic structure, the speed of absolute convergence occurs slowly. The rate of convergence is 1.4 percent per annum, which implies that it takes roughly 50 years to close half the gap (also called as half life)[6] between a states initial PCNSDP and its steady-state level of income. This convergence rate is comparable to those obtained in cross-country regressions (Barro and Sala-i-Martin, 1995). In the Indian case, the convergence rate has been estimated at anywhere between 7 and 48 percent (Nosbusch, 1999; Nagaraj et al., 2000) and more recently, at around 5.3 percent (Trivedi, 2006). Accounting for controls, the half life is smaller across all models, ranging from 24 to 46 years. We rst discuss the control variables. States with greater dependence on manufacturing are observed to grow faster. Earlier evidence for India (Nagaraj et al., 2000) and the USA (Carlino and Dena, 1998) also suggests a positive coefcient on manufacturing, which can be interpreted as evidence in favor of an interest rate channel of monetary policy. The coefcient on literacy exhibits mixed signs. Although the positive sign is consistent with the growth-enhancing effects of human capital as popularized in the endogenous growth literature (Lucas, 1988), the negative sign could be picking up standard conditional convergence effects, whereby states with lower initial human capital grow faster, as observed by Barro and Sala-i-Martin (1999) in their cross-country studies. Contextually, studies for India (Ahluwalia, 2002) as also cross-country regressions (Islam, 1995) conrm a negative coefcient on the literacy variable[7]. The coefcient on size is negative, suggesting that the higher this ratio, the greater is the states reliance on decit nancing of outlays, which tends

Financial outreach

83

84

JIBR 3,2

Ln (initial pc income) 20.63 (0.35) * * Geographic 0.004 (0.003) Demographic 20.36 (0.15) * * 20.39 (0.21) * * * 20.27 (0.15) * * * 20.59 (0.26) * * 20.44 20.41 (0.24) * * * (0.20) * * 0.029 (0.014) * * 2 0.003 (0.003) 0.001 (0.0007) 2 0.033 (0.023) 2 0.009 (0.004) * 2 0.061 (0.027) * * 20.009 (0.019) 0.001 (0.0008) 0.002 (0.0008) * *

Deposit a/c pc

Loan a/c pc

Deposit/income

Loan/income

Control variables Ln (Mfg.)

Ln (Literacy)

Ln (Size)

Ln (T&D losses)

dy_Merger

Constant

0.504 0.581 Initial pc income, literacy,T&D loss Lags of endogenous UR variables used in instrumentation

Year controls States (no. obs) x 2: Hansen over-id test ( p-value) AR (2): p-value Endogenous variables used as instruments 0.197 0.698 Initial pc income, literacy,T&D loss UR 0.190 0.587 Initial pc income, literacy,T&D loss UR 0.205 0.443 Initial pc income, literacy,T&D loss UR 0.288 0.460 Initial pc income, literacy,T&D loss UR 0.253 0.858 Initial pc income, literacy,T&D loss UR 0.122 0.302 Initial pc income, literacy,T&D loss UR

Notes: Signicance at: *1, * *5 and * * *10 percent levels; UR, unrestricted; robust standard errors are reported in parentheses; AR(2) is a test of second-order serial correlation and follows N(0, 1); details of the GMM procedure, with all the choices, reported in the last rows of the table

Table IV. Regression estimation: system GMM results

(2) 2 0.41 (0.24) 2 0.33 (0.14) * * 2 0.67 (0.19) * (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 20.058 (0.073) 0.035 (0.115) 20.073 (0.044) * * * 20.010 (0.029) 20.065 (0.040) * * * 20.365 (0.358) Yes 14, 97 2 0.007 (0.023) 2 1.147 (1.022) Yes 14, 97 0.023 (0.017) 2 0.791 (0.202) * Yes 14, 97 0.198 0.063 (0.148) (0.038) * * * 0.373 0.029 (0.167) * * (0.056) 20.131 20.084 (0.098) (0.044) * * 20.049 0.003 *** (0.026) (0.032) 20.049 2 0.059 20.053 (0.067) (0.037) * * * (0.021) * 21.867 2 0.164 20.319 (1.004) * * * (0.170) (0.249) Yes Yes Yes 14, 97 14, 97 14, 97 0.064 (0.026) * 20.606 (0.267) * * 20.173 (0.048) * 20.376 (0.162) * * 20.351 (0.137) * 20.717 (0.302) * Yes 14, 97 20.092 (0.069) 20.030 (0.015) * * 20.186 (0.105) * * * 20.029 (0.096) 20.027 20.117 (0.022) (0.033) * 20.411 20.578 (0.183) * * (0.395) Yes Yes 14, 97 14, 97 0.334 0.770 Initial pc income, literacy,T&D loss UR 0.685 0.618 Initial pc income, literacy,T&D loss UR 0.590 0.767 Initial pc income, literacy,T&D loss UR 0.329 0.717 Initial pc income, literacy,T&D loss UR

(1)

20.55 2 0.39 (0.32) * * (0.14) * 0.002 (0.001) * * * 0.032 (0.015) * *

0.014 (0.033) 20.600 (0.267) * * Yes 14, 97

0.371 (0.125) * 0.204 (0.082) * 20.069 (0.026) * 20.061 (0.035) * * * 2 0.030 20.111 (0.027) (0.044) * 21.447 2 1.105 (0.568) * (0.389) * * * Yes Yes 14, 97 14, 97

0.545 0.256 Initial pc income, literacy,T&D loss UR

to dampen growth. For the US states, Razzolini and Shughart (1997) also uncovered evidence that government size had a signicant and negative effect on economic growth. Across several specications, the coefcient on T&D losses is negative, indicating that states with better power network, are typically more attractive investment destinations. All specications control for the impact of the merger of several of the states during the post-reform period. Wherever signicant, the coefcient is negative, suggesting that the net effect of mergers has been a lowering in economic growth in the concerned states. Our coefcient of interest is outreach. Across all models, the coefcient exhibits expected signs, although it is signicant in six (out of the 12) specications. The magnitude of the effect is however, economically small. To understand the economic signicance of this variable, take for instance, specication (4), where the coefcient on demographic outreach is 0.03. Consider a state with demographic outreach at 0.31, the minimum value for the sample; the corresponding growth rate in PCNSDP being 0.89 percent. An increase in demographic outreach to 1 (over 200 percent increase) would raise its per capita growth to nearly 0.95 percent, a rise of about 7 percent. Likewise, in specication (10), the coefcient on deposit-income ratio is 2 0.06. Therefore, greater outreach of banking services as reected in a decline in this ratio by roughly 86 percent from its median value of 1.92-1.03, the value obtaining at the 25th percentile, would improve the per capita growth by roughly 5 percentage points. In sum, nancial outreach seems to exert a salutary effect on economic growth. 6.2 Does nancial technological outreach matter?[8] Table V reports cross-sectional regressions of measures of nancial technological outreach on economic growth, after accounting for various controls. These results provide further insights into the role of outreach in inuencing state-level economic growth. We obtained state-level data on the number of ATMs of public sector banks for 2002[9]. Following from Beck et al. (2007), we computed the two measures: ATMs/100,000 people (demographic technological outreach) and ATMs/1,000 km2 (geographic technological outreach). We did not have sufcient time-series data available for the above explanatory variable. Therefore, we estimated a basic cross-sectional OLS model with robust standard errors for 2002. Given the limited number of observations, the model specication is fragile and many variables cannot be simultaneously employed in order to avoid multicollinearity problems. We run the regressions with and without the state-level variables, although we control for state area across all models[10]. We nd that demographic technological outreach has a signicant impact on PCNSDP. In contrast, however, the impact of geographic technological outreach on PCNSDP appears limited. The result is signicant not only statistically, but economically as well. In model (2), the estimates reveal that a rise in demographic technological outreach by one standard deviation results in 0.528 percentage point higher NSDP. Although not directly comparable, it appears that the potential of technological outreach to advance economic growth is much higher, since in absolute terms, the magnitude on the ATM-related variables (the measures of technological outreach) far outweighs the other indicators discussed earlier. To examine the differences in technological outreach across states, we interacted the outreach variables with a dummy, depending on whether a state is high income. A similar exercise was conducted to examine the differences in technological outreach of costal versus landlocked states. The results indicate signicant differences

Financial outreach

85

86

JIBR 3,2

ATMs/100,000 people ATMs/1,000 km2 Interaction terms ATMs/100,000 people*high income ATMs/100,000 people*coastal ATMs/1,000 km2*high income ATMs/ 1,000 km2*coastal Ln mfg. Ln unregistered mfg. Ln decit Ln crime Constant Control for state area Observations R2 2 0.023 (0.161) 0.039 (0.031) 0.190 (0.100) * * * 0.047 (0.080) 0.072 (0.005) * 0.061 (0.077) 0.022 (0.006) * 0.171 (0.111) 0.309 2 0.046 2 0.144 2 0.393 13.268 0.337 2 0.035 2 0.213 2 0.467 13.935 (0.191) 0.652 (0.112) * 0.461 (0.263) (0.221) 2 0.221 (0.137) 2 0.126 (0.279) *** * * * 2 0.104 (0.057) * * * (0.079) 2 0.239 (0.130) (0.149) * * 2 0.706 (0.131) * 2 0.309 (0.135) * * (1.202) * 8.982 (2.373) * 12.105 (0.635) * 12.597 (1.338) * Yes Yes Yes Yes 14 14 14 14 0.882 0.294 0.930 0.910 (0.196) 0.544 (0.059) * (0.203) 2 0.227 (0.057) * (0.131) * * * 2 0.249 (0.050) * (0.256) * * * 2 0.638 (0.048) * (1.636) * 12.022 (0.315) * Yes Yes 14 14 0.888 0.973

Notes: Signicance at: *1, * *5 and * * *10 percent levels; robust standard errors in parentheses

Table V. Financial technological outreach and economic growth: cross-sectional results for 2002
(2) 0.233 (0.078) * (3) (4) (5) (6) (7) (8) 0.039 (0.025) 0.018 (0.023) 0.639 (0.061) * 2 0.214 (0.089) * * 2 0.400 (0.332) 2 0.877 (0.311) * 13.994 (2.022) * Yes 14 0.935

(1)

0.328 (0.094) *

9.508 (1.255) * Yes 14 0.632

in technological outreach across these two types of categories. Consider for instance, model (7). A rise in the number of ATMs per 1,000 km2 by 10 percent improves PCNSDP by 0.2 percent. For high-income states, there is an increase of an additional 0.7 percent. The t of the model is fairly high: inclusive of state controls, all models explain over 85 percent of the variation in the dependent variable. Together, these results suggest that efforts at improving technological outreach have a perceptible effect on state per capita incomes. 6.3 Sensitivity tests Following from the earlier discussion, we consider several robustness tests of the baseline results pertaining to geographic outreach (Panel A), demographic outreach (Panel B), deposit accounts per capita (Panel C) and deposit-income ratio (Panel D). In particular, we consider the impact of these variables on state per capita growth on the basis of a priori classication such as income and geography as also the pre- and post-reform period. The results are highlighted in Table VI. The results suggest that high-income states exhibited high levels of geographic exclusion as compared to low-income ones. Illustratively, an increase in geographic outreach by 10 percent would entail a rise in per capita income by 0.1 percent. As compared to this, per capita growth in the high-income states is 80 percent higher for the same magnitude of improvement in banking outreach. It is a similar story in the case of coastal states. In terms of magnitudes, a 10 percent increase in geographic outreach raises the per capita income of coastal states by 33 percent more as compared to landlocked states. The nal two specications introduces three-way interaction variables, high income*coastal*geographic and high income*pre-1990s*geographic. In specication (4), the coefcient on the interaction term is positive and signicant at the 0.05 level, indicating that high income, coastal states with higher levels of geographical outreach typically exhibit higher per capita income growth. The evidence also indicates that high income, coastal states with higher levels of geographical outreach typically grow faster. To the extent that the analysis categorizes states based on certain inherent characteristics, ` it offers a rationale for the high growth of certain states/regions vis-a-vis others[11]. The remaining categories focus on the three nancial outreach measures which were observed to be signicant in the baseline analysis. Thus, the evidence suggests that high income states with higher demographic outreach exhibited higher per capita growth in the pre-1990s (specication (5)). Likewise, by extending the outreach of banking services, higher deposit accounts per capita in the pre-1990s regime was an important factor in raising per capita growth in high-income states (Panel C, specication (5)). Finally, the evidence emanating from Panel D appears to indicate that, as compared to low-income states, the use of banking services in high-income states has been lower by nearly 80 percent[12]. 7. Concluding remarks The paper makes a systematic attempt to ascertain the nexus between nance and growth at the sub-national level for an emerging economy. Borrowing from the literature, we employ measures of nancial outreach that capture both the outreach and also the use of banking services. We subsequently examine the impact of these measures of nancial outreach on per capita economic growth, using data on major states in India covering the period 1973-2004.

Financial outreach

87

88

JIBR 3,2

Panel A Geographic High income*geographic Coastal*geographic Pre-1990s*geographic High income*coastal*geographic High income*pre-1990s*geographic Groups (no. obs) x 2 (Hansen over-id test) AR (2) test Endogenous variables used as instruments Lags of endogenous variables used in instrumentation Exogenous instrumenting variable Panel B Demographic High income*demographic Coastal*demographic Pre-1990s*demographic High income*coastal*demographic High income*pre-1990s*demographic 0.010 (0.003) * 0.008 (0.003) * * 20.0005 (0.001) 20.002 (0.001) 0.002 (0.001) * * 14, 83 0.370 0.246 0.002 (0.002) 14, 83 0.507 0.637 14, 83 0.832 0.191 14, 83 0.333 0.301 14, 83 0.433 0.401 20.0003 (0.001) 0.001 (0.009) * * 0.003 (0.001) * * 0.004 (0.001) * UR Region 0.029 (0.032) 0.013 (0.010) 0.003 (0.002) 0.031 (0.028) 0.0003 (0.002) 0.001 (0.0006) * * * (continued) 0.011 (0.006) * * * 0.005 (0.013) UR Region UR Region UR Region 0.002 (0.006) UR Region 0.004 (0.002) * * *

Table VI. Regression estimation: system GMM results: sensitivity tests (1) (2) (3) (4) (5)

(1) 14, 83 0.166 0.816 14, 83 0.413 0.780 14, 83 0.850 0.416 14, 83 0.648 0.805 14, 83 0.735 0.868

(2)

(3)

(4)

(5)

UR Region 0.0009 (0.0001) * 26.8 102 6 (0.00005) 0.0004 (0.0008) 6.6 102 6 (0.0003) 0.0003 (0.0001) * * * 2 0.0001 (0.0002) 14, 83 0.655 0.920 14, 83 0.796 0.549 14, 83 0.900 0.134 14, 83 0.941 0.998 0.0001 (0.00008) * * * 14, 83 0.923 0.165 20.0001 (0.0002) 0.0008 (0.0007) 2 0.0007 (0.0010)

Groups, N. Obs x 2 (Hansen over-id test) AR (2) test Endogenous variables used as instruments Lags of endogenous variables used in instrumentation Exogenous instrumenting variable Panel C Deposit a/c pc High income*deposit a/c pc Coastal*deposit a/c pc Pre-1990s*deposit a/c pc High income*coastal*deposit a/c pc High income*pre-1990s*deposit a/c pc Groups, N. Obs x 2 (Hansen over-id test) AR (2) test Endogenous variables used as instruments Lags of endogenous variables used in instrumentation Exogenous instrumenting variable UR Region UR Region UR Region UR Region UR Region UR Region UR Region UR Region UR Region (continued)

Financial outreach

89

Table VI.

90

JIBR 3,2

Panel D Deposit/income High income*deposit/income Coastal*deposit/income Pre-1990s*deposit/income High income*coastal*deposit/income High income*pre-1990s*deposit/ income Groups, N. Obs x 2 (Hansen over-id test) AR (2) test Endogenous variables used as instruments Lags of endogenous variables used in instrumentation Exogenous instrumenting variable 2 0.103 (0.051) * * 0.059 (0.032) * * * 2 0.030 (0.038) 0.004 (0.012) 0.015 (0.061) 2 0.006 (0.007) 14, 83 0.955 0.132 14, 83 0.247 0.831 14, 83 0.277 0.752 14, 83 0.632 0.732 20.008 (0.007) 14, 83 0.618 0.463 20.021 (0.027) 2 0.002 (0.009) 20.005 (0.009) UR Region UR Region UR Region UR Region UR Region

Notes: Signicance at: *1, * *5 and * * *10 percent levels; UR, unrestricted; in all specications, the dependent variable is growth in PCNSDP; all specications control for state-level variables, but they are not reported to save space; robust standard errors are reported in parentheses; AR(2) is a test of second-order serial correlation and follows N(0, 1); details of the GMM procedure along with the choices are reported in the last rows of the table

Table VI. (1) (2) (3) (4) (5)

The analysis indicates signicant regional divergences in nancial outreach across states, and also in terms of their income characteristics. As well, while indicators of nancial outreach have exhibited improvements, their relationship with growth appears to have become tenuous, especially over the reforms period. More importantly, the multivariate regressions that take on board the state-level controls indicate a signicant impact of nancial outreach on economic growth. In particular, efforts to improve outreach of the nancial sector appear to have led to a perceptible improvement in economic growth. In terms of magnitude, a 10 percent rise in demographic outreach raises per capita growth by roughly 0.3 percent, all other things equal. In the case of geographic outreach, the magnitude is distinctly lower with per capita growth rising by only 0.02 percent for a 10 percent branch expansion. In addition, the results also suggest that the efforts at raising economic growth via the use of banking services operate primarily through the deposit side. The study points to an alternative impact of the social banking experiment: the positive impact of nancial outreach on state economic growth. In a sense, we conrm the ndings of Burgess and Pande (2005). The evidence also supports the fact that, as compared to low-income states, high-income states with higher geographic outreach levels have higher economic growth. Furthermore, high-income coastal states with higher geographic outreach levels have higher per capita income growth. The analysis therefore offers a rationale for the high growth of certain regions and states in India based on their intrinsic characteristics. We, therefore, substantiate the ndings of the growth literature in India which nds unequal growth across regions and states (Bajpai and Sachs, 1999; Ahluwalia, 2002; Nagaraj et al., 2000; Kochhar et al., 2006). The ndings appear to indicate that nancial outreach as measured by demographic outreach has a signicant impact on PCNSDP. In addition, high-income and coastal states have higher technological outreach levels as compared to their counterparts. It seems that the innate advantages of high-income states are compounded by the fact that the nancial system has been redirecting nancial savings towards the better-performing states, as the analysis would testify. A reorientation of the policy framework in these states coupled with appropriate utilization of public investment to build economic and social infrastructure in the low-income states assumes relevance in this regard. In sum, the results are also a pointer to the fact that the social banking strategy pursued in India was inuential in raising state per capita growth, through the effects of expanding the outreach and use of nancial services. Given the potential for nancial technological outreach to promote economic growth, the analysis suggests a role for technology to reach out to larger segments of the populace. Analysis of such interlinkages between growth and technological outreach using comprehensive datasets constitutes elements for future research. Although the present analysis represents perhaps the rst systematic attempt to analyze nancial outreach at the state-level, it is not without its limitations either. First, the focus is primarily on banking exclusion; other dimensions of nancial exclusion, such as price exclusion, marketing exclusion, and self-exclusion have not been addressed (FSA, 2000). Additionally, the measures of nancial exclusion considered suffer from certain shortcomings. An individual or rm may receive more than one loan or have multiple deposit accounts, so that the number of loans and deposits accounts

Financial outreach

91

JIBR 3,2

is far from being a perfect proxy of the number of people using these services within a state. As well, the average size of loans and deposits to PCNSDP might not be representative of the value of services that a typical individual might receive. Accordingly, the results emanating from the analysis would need to be interpreted with care and caution.
Notes 1. The terms state and sub-national are used interchangeably. 2. Kremer and Miguel (2001) report that kids in schools treated with de-worming drugs (drugs against hookworm, roundworm, whipworm, and schistosomiasis) display reduced absenteeism by one quarter (with gains being especially large among the youngest children). A second effect of education on health operates through the Beckerian quality-quantity trade-off (of children). Parents know that their children are very likely to expire early will tend to have many kids in order to end up with some adult descendants. With a binding budget constraint, the amount of resources devoted to each child consequently declines, and as a result, each child ends up with lower education and human capital. Parents, therefore, substitute quality of children for quantity of children. A third effect of health on education operates through incentives: the rate of return to education is the present discounted value of all future wages that an educated person receives. Low life expectancy tends to reduce the rate of return and, as a result, the incentives to educate and accumulate human capital. 3. NABARD is the apex institution concerning policy, planning and operations in the eld of credit for agriculture and other economic activities. 4. High-income states are in alphabetical order: AP, GUJ, HAR, KARN, KER, MAH, PUNJ, TN and WB. Likewise, coastal states are AP, GUJ, KARN, KER, MAH, ORIS, TN and WB (See footnore [6]). 5. These states, in order, are regional location, are Andhra Pradesh (AP), Karnataka (KARN), Kerala (KER) and, Tamil Nadu (TN) in Southern region, Haryana (HAR), Punjab (PUNJ) and Rajasthan (RAJ) in the Northern region, Bihar (BIH), Orissa (ORIS), and West Bengal (WB) in the Eastern region, Gujarat (GUJ) Maharashtra (MAH) in the Western region and Madhya Pradesh (MP) and Uttar Pradesh (UP) in the central region. 6. The half-life means the half time for a state to converge to its steady state. It is computed as 2ln2=ln1 r, where r is the convergence coefcient. 7. Using the gross enrolment ratio (6-11 years) did not signicantly alter the results. 8. I am grateful to the referee for the observations on this point. 9. Obtained from the statistical web site, www.Indiastat.com. The nancial technological outreach numbers are likely to be under-estimates, since we do not have information of the number of ATMs for private and foreign banks. 10. We also ran the regressions with PCNSDP growth as dependent variable for all models. The results (not reported) suggest that neither of the outreach variables is signicant across the specications. 11. To see this, note that the common set of states that are both high income and coastal include AP, GUJ, KARN, KER, MAH, TN and WB. Without loss of generality, the high income, coastal states are those belonging to the Western and Southern regions. 12. To see this, note that, the coefcient for high income states 20.044 ( 20.103 0.059) is higher by 77 percent as compared to the coefcient on deposit/income.

92

References Aghion, P., Howitt, P. and Mayer-Foulkes, D. (2005a), The effect of nancial development on convergence: theory and evidence, Quarterly Journal of Economics, Vol. 120, pp. 173-222. Aghion, P., Burgess, R., Redding, S. and Zilibotti, F. (2003), Entry liberalization and inequality in industrial performance, Journal of the European Economic Association, Vols. 2/3, pp. 291-302. Aghion, P., Burgess, R., Redding, S. and Zilibotti, F. (2005b), The unequal effects of liberalization: evidence from dismantling the License Raj in India, available at: www. economics.harvard.edu/faculty/aghion Ahluwalia, M.S. (2002), State level performance under economic reforms in India, in Krueger, A.O. (Ed.), Economic Policy Reforms and the Indian Economy, Oxford University Press, New Delhi. Arnold, I. and Vrugt, E.B. (2002), Regional effects of monetary policy in The Netherlands, International Journal of Business and Economics, Vol. 1, pp. 122-34. Bajpai, N. and Sachs, J.D. (1999), The progress of policy reform and variations in performance at the sub-national level in India, Discussion Paper No. 730, Harvard Institute for International Development, Harvard University, Cambridge, MA. Barro, R.J. (1991), Economic growth in a cross section of countries, Quarterly Journal of Economics, Vol. 51, pp. 407-43. Barro, R.J. (1997), Determinants of Economic Growth: A Cross-country Empirical Study, MIT Press, Cambridge, MA. Barro, R.J. and Sala-i-Martin, X. (1995), Convergence, Journal of Political Economy, Vol. 100, pp. 223-51. Barro, R.J. and Sala-i-Martin, X. (1999), Economic Growth, MIT Press, Cambridge, MA. Basu, P. (2006), Improving Access to Finance for Indias Rural Poor, The World Bank, Washington, DC. Beck, T., Demirguc-Kunt, A. and Levine, R. (2007), Reaching out: access to and use of banking services across countries, Journal of Financial Economics, Vol. 85, pp. 234-66. Beck, T., Demirguc-Kunt, A. and Martinez Peria, M.S. (2006a), Banking services for everyone? Barriers to bank access and use around the world, World Bank Policy Research Working Paper No. 4079, The World Bank, Washington, DC. Beck, T., Demirguc-Kunt, A., Laeven, L. and Maksimovic, V. (2003), Bank competition, nancing obstacles and access to credit, Policy Research Working Paper No. 2996, The World Bank, Washington, DC. Beck, T., Demirguc-Kunt, A., Laeven, L. and Maksimovic, V. (2006b), The determinants of nancing obstacles, Journal of International Money and Finance, Vol. 25, pp. 932-52. Bernanke, B. (2008), The importance of nancial education and the National JumpStart Coalition Survey, address delivered at the JumpStart Coalition for Personal Financial Literacy, Washington, DC. Besley, T. and Burgess, R. (2004), Can labor regulation hinder economic performance? Evidence from India, Quarterly Journal of Economics, Vol. 119, pp. 91-134. Bhandari, A. (2009), Access to banking services and poverty reduction: a state-wise assessment in India, IZA Discussion Paper No. 4132, IZA, Bonn. Blundell, R. and Bond, S. (1998), Initial conditions and moment restrictions in dynamic panel data models, Journal of Econometrics, Vol. 87, pp. 115-43.

Financial outreach

93

JIBR 3,2

94

Burgess, R. and Pande, R. (2005), Do rural banks matter? Evidence from the Indian social banking experiment, American Economic Review, Vol. 95, pp. 780-95. Carlino, G. and Dena, R. (1998), The differential regional effects of monetary policy, Review of Economics and Statistics, Vol. 80, pp. 572-87. Chairlone, S. and Ghosh, S. (2008), Indian banking system, in Bongini, P., Chairlone, S. and Ferri, G. (Eds), Emerging Banking Systems, Palgrave McMillan, London, pp. 89-122. Clarke, G., Cull, R. and Martinez Peria, M.S. (2001), Does foreign bank penetration reduce access to credit in developing countries? Evidence from asking borrowers, Policy Research Working Paper No. 2716, The World Bank, Washington, DC. Demirguc-Kunt, A. and Maksimovic, V. (1998), Law, nance and rm growth, Journal of Finance, Vol. 53, pp. 2107-37. Easterly, W. and Levine, R. (2002), Its not factor accumulation: stylized facts and growth models, Central Bank of Chile Working Paper No. 164, Santiago de Chile, June. EPW Research Foundation (2003), Domestic Product of States of India: 1960-61 to 2000-01, EPW, Mumbai. European Commission (2008), Financial Services Provision and Prevention of Financial Exclusion, Directorate-General for Employment, Social Affairs and Equal Opportunities, Brussels. FSA (2000), In or Out? Financial Exclusion: A Literature and Research Review, Financial Services Authority, London. Galindo, A. and Micco, A. (2005), Bank credit to small and medium enterprises: the role of creditor protection, Working Paper No. 347, Central Bank of Chile, Santiago de Chile. Government of India (2008a), Legislations on Environment, Forests and Wildlife, Ministry of Environment and Forests, available at: http://envfor.nic.in Government of India (2008b), Report of the Committee on Financial Inclusion, Government of India, New Delhi (Chairman: Dr C. Rangarajan). Grier, K.B. and Tullock, G. (1989), An empirical analysis of cross-national economic growth, 1951-80, Journal of Monetary Economics, Vol. 24, pp. 259-76. Grossman, M. (1972), On the concept of health capital and the demand for health, Journal of Political Economy, Vol. 80, pp. 223-55. Guiso, L., Sapienza, P. and Zingales, L. (2004), Does local nancial development matter?, Quarterly Journal of Economics, Vol. 119, pp. 929-69. Hasan, R., Mitra, D. and Ramaswamy, K.V. (2007), Trade reforms, labor regulations and labor-demand elasticities: empirical evidence from India, Review of Economics and Statistics, Vol. 89, pp. 466-81. His Majestys Treasury (2004), Promoting Financial Inclusion, His Majestys Treasury, London, December. Honohan, P. (2007), Cross-country variation in household access to nancial services, paper presented at the World Bank Conference on Access to Finance, Washington, DC, available at: www.worldbank.org Islam, N. (1995), Growth empirics: a panel data approach, Quarterly Journal of Economics, Vol. 110, pp. 1127-70. King, R. and Levine, R. (1993), Finance, entrepreneurship and growth: theory and evidence, Journal of Monetary Economics, Vol. 32, pp. 513-42. Kochhar, K., Kumar, U., Rajan, R. and Subramanium, A. (2006), Indias pattern of development: what happened/what follows?, Journal of Monetary Economics, Vol. 53, pp. 981-1019.

Kremer, M. and Miguel, E. (2001), Worms: education and health externalities in Kenya, NBER Working Paper No. 8481, NBER, Cambridge, MA. Levine, R. (2005), Finance and growth: theory and evidence, in Aghion, P. and Durlauf, S. (Eds), Handbook of Economic Growth, Elsevier Science, Amsterdam. Levine, R. and Zervos, S. (1998), Stock markets, banks, and economic growth, American Economic Review, Vol. 88, pp. 537-58. Love, I. and Mylenko, N. (2003), Credit reporting and nancing constraints, Policy Research Working Paper No. 3142, The World Bank, Washington, DC. Lucas, R.E. (1988), On the mechanics of economic development, Journal of Monetary Economics, Vol. 22, pp. 3-42. Mitra, D. and Ural, B.P. (2007), Indian manufacturing: a slow sector in a rapidly growing economy, World Bank Working Paper No. 4233, The World Bank, Washington, DC. Nachane, D.M., Ray, P. and Ghosh, S. (2002), Does monetary policy have differential state-level effects? An empirical evaluation, Economic and Political Weekly, Vol. 47, pp. 4723-8. Nagaraj, R., Varoudakis, A. and Veganzones, M.-A. (2000), Long run growth trends and convergence across Indian states, Journal of International Development, Vol. 12, pp. 45-70. Nosbusch, Y. (1999), Convergence across regions: evidence from India, Quantitative Economics Project, London School of Economics, London. Rajan, R.G. and Zingales, L. (1998), Financial dependence and growth, American Economic Review, Vol. 88, pp. 559-86. Rao, G. and Singh, N. (2007), The political economy of Indias scal federal system and its reform, Publius: The Journal of Federalism, Vol. 37, pp. 26-44. Razzolini, L. and Shughart, F. (1997), On the (relative) unimportance of a balanced budget, Public Choice, Vol. 90, pp. 215-33. RBI (1954), Report of the Committee on All India Rural Credit Survey, Reserve Bank of India, Mumbai (Chairman: A.D. Gorwala). RBI (2004), Handbook of Statistics on State Government Finances, Reserve Bank of India, Mumbai. RBI (2006), Handbook of Statistics on State Government Finances, Reserve Bank of India, Mumbai. RBI (2007), Handbook of Statistics on Indian Economy, Reserve Bank of India, Mumbai. RBI (2008), Quarterly Statistics on Deposits and Credit of Scheduled Commercial Banks, Reserve Bank of India, Mumbai. RBI (various years (a)), Basic Statistical Returns of Scheduled Commercial Banks in India, Reserve Bank of India, Mumbai. RBI (various years (b)), Statistical Tables Relating to Banks in India, Reserve Bank of India, Mumbai. Rodrik, D. (2005), Why we learn nothing from regressing economic growth on policies, available at: http://ksghome.harvard.edu Rodrik, D. and Subramanium, A. (2004), From Hindu growth to productivity surge: the mystery of the Indian growth transition, CEPR Discussion Paper No. 4371, Center for Economic Policy Research, London. Sachs, J.D., Bajpai, N. and Ramaiah, A. (2002), Understanding regional economic growth in India, CID Working Paper No. 88, Center for International Development, Harvard University, Cambridge, MA.

Financial outreach

95

JIBR 3,2

96

Topalova, P. (2008), India: is the rising tide lifting all boats?, IMF Working Paper No. 54, IMF, Washington, DC. Trivedi, K. (2006), Educational human capital and levels of income: evidence from the states of India 1965-92, Journal of Development Studies, Vol. 42, pp. 1350-78. Wachtel, P., Hasan, I. and Zhou, M. (2006), Institutional development, nancial deepening and economic growth; evidence from China, Bank of Finland Discussion Paper No. 12, Bank of Finland, Helsinki. (The) World Bank (2005), State Fiscal Reforms in India: Progress and Prospects, Macmillan, New Delhi. Further reading Ghosh, S. (2010), Determinants of banking outreach: an empirical assessment of Indian states, Journal of Developing Areas (in press). Government of India (2002), National Human Development Report, Planning Commission, Government of India, New Delhi. Government of India (various years (a)), Economic Survey, Government of India, New Delhi. Government of India (various years (b)), Statistical Abstract of India, Government of India, New Delhi. Mohan, R. (2006), Economic growth, nancial deepening and nancial outreach, address delivered at the Annual Bankers Conference, Hyderabad, available at: www.rbi.org.in About the author Saibal Ghosh is working as an Ofcer in the Reserve Bank of India, the Indian Central Bank, at its headquarters in Mumbai, India. He has over 12 years of experience in central banking in diverse areas including banking, international nance, monetary policy, and nancial stability. Prior to the present job, Dr Saibal Ghosh worked with the Industrial Development Bank of India, an erstwhile development bank. Dr Saibal Ghosh has worked extensively in the areas of banking and nance and has published papers in leading international journals including Manchester School (2010), Journal of Economics and Business (2009), Economic Systems (2008), International Journal of Auditing (2007) and Review of Financial Economics (2006). Saibal Ghosh can be contacted at: saibalghosh@rbi.org.in

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints

Notation 112 0.005 0.559

Variable

Data source

N.Obs

Mean

SD

Appendix

Gr_PCNSDP

First difference of natural log of per capita net EPW states database state domestic product (PCNSDP) 112 112 112 112 112 112 112 112 112 97 112 112 112 112 112 112 112 112 0.493 0.050 0.23 0.027 0.571 0.643 0.214 0.286 0.214 0.143 0.143 0.155 0.029 0.175 0.023 0.06 0.162 0.497 0.481 0.412 0.454 0.412 0.351 0.351 2.99 2.70 326.91 16.59 18.70 23.17 6.52 53.26 2.31 35.03 23.74 18.01

Financial outreach indicators Geographic

bank branches per 1,000 km2

Demographic Loan a/c pc

bank branches per 1 lac people Loan accounts per 1,000 people

Deposit a/c pc Loan/income

Deposit accounts per 1,000 people Average size of loans/PCNSDP

Deposit/income State-level controls Manufacturing (mfg.) EPW States database and RBI (2007)

Average size of deposits/PCNSDP

Numerator is from Statistical Tables Relating to Banks in India (STB). Denominator is from Wikipedia web page www.en.wikipedia. org Numerator is obtained from STB Numerator is obtained from Basic Statistical Returns o f Scheduled Commercial Banks (BSR) Numerator is obtained from BSR Data on deposits is from BSR. PCNSDP is from EPW States database Data on credit is from BSR

Literacy Size

Share of manufacturing in NSDP (1980-1981 prices) Literacy rate Gross scal decit/NSDP

T&D losses

Statistical abstract RBI (2006) supplemented by RBI Bulletin, various years Transmission and distribution losses of State Planning Commission web site Electricity Boards

Dummy variables dy_Merger

dy_Coastal dy_High income dy_Northern dy_Southern dy_Eastern dy_Western dy_Central

Dummy 1 for the years 2001 onwards for states from which new states were carved out Dummy 1 for the coastal states, else 0 Dummy 1, for high-income states, else 0 Dummy 1 for the Northern region, else 0 Dummy 1 for the Northern region, else 0 Dummy 1 for the Northern region, else 0 Dummy 1 for the Northern region, else 0 Dummy 1 for the Northern region, else 0

Financial outreach

97

Table AI. Variables, data source, and summary statistics