Credit off-take from formal financial institutions in rural India: Quantile regression results

Pal, Debdatta; Laha, Arnab Article Credit off-take from formal financial institutions in rural India: Quantile regression results Agricultural and Food Economics Provided in Cooperation with: Italian Society of Agricultural Economics (SIDEA) Suggested Citation: Pal, Debdatta; Laha, Arnab (2014) : Credit off-take from formal financial institutions in rural India: Quantile regression results, Agricultural and Food Economics, ISSN 2193-7532, Springer, Heidelberg, Vol. 2, pp. 1-20, https://doi.org/10.1186/s40100-014-0009-y This Version is available at: https://hdl.handle.net/10419/108920 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/2.0/ RESEARCH Open Access Credit off-take from formal financial institutions in rural India: quantile regression results Debdatta Pal 1* and Arnab Kumar Laha 2 * Correspondence: [email protected] 1 Economics and Business Environment, Indian Institute of Management Raipur, GEC Campus, Raipur 352015, India Full list of author information is available at the end of the article Abstract: Extending financial services to unbanked population in India has remained a central part of the policy thrust of the Indian government for decades. To that effect, a widespread formal credit delivery mechanism has been established to meet the credit requirements of rural communities. However, the Indian governmentbacked formal financial sector has had limited success in providing resources to poor rural households, which has led to strong criticism of the policy and its implementation. In this study, we use data from 600 rural households spread across six Indian states to examine the changing distribution of credit off-take among borrowers of formal financial institutions. By using quantile regression, we find that even among rural households that could access loans from the formal banking sector, the distribution of credit off-take is skewed towards resource-rich households. We also find that even among borrowers in the upper quantiles of the conditional loan distribution, marginal farmers received substantially less loan amounts than those belonging to the category of medium and large farmers. JEL classifications: C31, Q14 Keywords: Rural credit; Formal finance; Quantile regression; India Background The policy thrust to make formal credit inclusive has been a central focus of the Indian government. This may be attributed to two widely accepted facts. First, access to credit is a precursor for growth, and second, alternative credit sources are costly. Policy-makers in India have long pushed the formal financial sector to open more outlets and bankers to offer credit below the market clearing price (Burgess and Pande 2005). However, since credit is an inter-temporal contract, lenders face the dual problem of adverse selection (hidden information) and moral hazard (hidden action). If lenders, through screening, cannot identify borrowers based on their likelihood of default, willful default may rise. In response to this, informal lenders may add a default premium on top of the nominal interest rate, an option that is unavailable to formal financial sources, where interest rates are controlled by the government (Gonzalez-Vega 1984). Hence, formal lenders, namely banks and cooperatives, may ration credit, request additional documents from and frequent visits by the borrower (leading to higher borrower-side transaction costs), demand marketable collateral, and favor production credit over consumption loan. The void created out of the unmet credit demand from formal financial sources has led to the widespread dependence of Indian rural households on informal credit sources, namely moneylenders and traders (see Basu 2006 for empirical evidence). © 2014 Pal and Laha; licensee Springer. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Pal and Laha Agricultural and Food Economics 2014, 2:9 http://www.agrifoodecon.com/content/2/1/9 Academic inquiries to identify the causes of such dismal performance on the part of the formal financial sector in agrarian economies can be classified under three broad groups: first, those that attempt to identify factors affecting the choice of credit sectors in rural economies (see, for example, Kochar 1997; Pal 2002; Sahu et al. 2004; Boucher and Guirkinger 2007; Datta and Ghosh, 2013; Pal and Laha 2013), second, those that explore the determinants of credit amounts received by rural households from the formal sector (see, for example, Datta 2003; Sahu et al. 2004) and third, those that estimate the effects of the lack of access to formal-sector credit on agricultural production (Carter 1989; Guirkinger and Boucher 2008). The first two groups of empirical investigations indicate that a large agricultural land size, good land quality, the presence of irrigation and regarding the loan applicant, membership in a higher caste and higher levels of educational attainment increase the applicant’s chance of gaining access to formal credit sources as well as the quantum of credit disbursed by the formal sector. The third set consists of a growing body of empirical literature that suggest that credit constraints adversely affect farm production, farm profit, and farm investment. This study contributes additional empirical evidence to support the findings of the second set of inquiries and attempts to answer the following question: among the rural households that have received loans from the formal sector, what is the changing distribution of credit off-take? Finding an answer to this question is pertinent from the following policy perspectives. First, it would help central bankers as well as policy makers understand variations in outreach of the formal banking sector within its own client group. Secondly, it would elucidate the extent of the gap between the credit off-take from the formal financial sector by the rich and poor households in the context of a developing country like India. Moreover, apart from including those variables mentioned in earlier studies, namely Datta (2003) and Sahu et al. (2004), this study also attempts to capture the effects of village-level attributes, such as the distance of households from commercial, financial, and educational institutions, on the loan amount. These findings would complement the theoretical model advanced by Nakamura and Nakajima (2011) which shows that improvements in the credit market during the early stages of economic development is essential to escape poverty. We used quantile regression, which was introduced by Koenker and Bassett (1978), to address this issue of improving the credit off-take from formal financial sources in rural India. Quantile regression is used when conditional quantile functions and even extremes are of interest, and it has been used in various areas of applied economics, such as health economics (see Winkelmann 2006, for example), financial economics (see Bassett and Chen 2001, for example), labor economics (see Buchinsky 1998, for a detailed survey), and education economics (see Eide and Showalter 1998). To the best of our knowledge, quantile regression has not been employed thus far to study changes in the distribution of credit off-take by rural households. Consistent with the theory of credit rationing, we observed that between resource-rich and resource-poor rural households, there is considerable variation in the quantum of loan off-take from formal financial institutions. By using quantile regression, we found that among borrowers in the 90 th quantile of the conditional loan distribution, marginal farmers a received only approximately 80 percent of the loan amount of their counterparts belonging to the category of medium and large farmers. Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 2 of 20 http://www.agrifoodecon.com/content/2/1/9 Briefly, the structure of this paper is as follows. In Survey and data description, we describe the survey and dataset. Method presents the methodology used for analysis. Results and discussion contains the results and discussion, and in Conclusions, we indicate the policy implications. Survey and data description The survey was conducted between May and December 2010 as part of the research project of the Centre for Management in Agriculture, Indian Institute of Management Ahmedabad, titled “Assessing Policy Interventions in Agri-Business and Allied Sector Credit versus Credit Plus Approach for Livelihood Promotion,”sponsored by the Ministry of Agriculture, Government of India. Six Indian states—Chhattisgarh, Maharashtra, West Bengal, Andhra Pradesh, Tamil Nadu, and Gujarat—were selected based on two state-level indicator criteria: first, the average population per branch of formal financial institution as of the end of June 2009, and second, incidences of indebtedness of rural households. Table 1 shows these two indicators against the national average. Against the national average of 6554 people per rural branch of formal financial institution attempt has been made to choose states with even spread on this criterion. While a bank branch caters to 10410 people on average in Chhattisgarh, the number of people served by a bank branch in Maharashtra is as low as 2688. The average population per bank branch in West Bengal is the closest to national average. Therefore, variation on the level of branch penetration is well maintained in the sample. Similarly, the sample states are distributed evenly above, below, and at the average for the criterion of incidences of indebtedness of rural households. Through this process, we attempted to capture variations among the sample states. To ensure that various lending institutions are represented, we selected a village or cluster of villages within a single agroclimatic region of a state where all credit sources, namely, formal, semi-formal, and informal, are present (see Additional file 1: Appendix 1). Formal lending institutions are scheduled public or private sector commercial banks c (SCB), regional rural banks d (RRB), and cooperative banks, e namely, primary agricultural cooperative societies (PACS), multipurpose PACS (MPACS), District Central Cooperative Bank (DCCB), and State Cooperative Banks. Semi-formal credit sources include self-help groups f (SHG) and microfinance institutions g (MFI). The informal sector includes moneylenders, input dealers, output traders, shopkeepers, and friends and relatives. A complete enumeration of all Table 1 Indicators of branch penetration and rural indebtedness State Average population per rural financial institution branch as of March 2009 # Incidence of indebtedness of rural households (in %) as of June 2002 $ Institutional Non-institutional All Chhattisgarh 10410 14.4 6.7 19.8 Maharashtra 2688 22.8 7.2 27.5 West Bengal 6095 12.1 11.0 21.8 Andhra Pradesh 11784 14.9 32.9 42.3 Tamil Nadu 5976 13.9 21.3 31.3 Gujarat 3759 14.7 15.8 28.1 All of India 6554 13.4 15.5 26.5 # Authors' own calculation b . $ Household indebtedness in India as of 06.30.02 (National Sample Survey Organisation 2005). Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 3 of 20 http://www.agrifoodecon.com/content/2/1/9 households h in a village was undertaken to form a sampling frame that was used to draw a representative sample of village households. The survey was conducted in five villages in West Bengal, three villages each in Maharashtra and Chhattisgarh, and one village each in Andhra Pradesh, Gujarat, and Tamil Nadu. From each village in West Bengal, 30 households were chosen through random sampling, whereas in the other states, 50 households were randomly chosen from each village. Thus, the total sample size in this study consists of 600 households. Each sample household was administered a structured questionnaire to collect general information on household demographics, social characteristics, occupation (both primary and secondary), resource endowment, loan-related information (i.e., access to different credit sources between April 2009 and March 2010), and the terms and conditions of the loan facilities experienced by the borrowers. Incidentally, besides respondents that borrowed from mutually exclusive sources (i.e., formal, semi-formal, or informal), this survey also included those that borrowed from multiple sources. Of the 600 sampled households, 231 met their full credit demand entirely from formal sources, 129 met part of their credit demand from formal sources, 91 respondents were excluded from formal sources but have used credit facilities from semi-formal and/or informal sources, 71 were dependent solely on informal credit sources, and 75 respondents did not avail any credit facilities during the corresponding period (see Table 2). Of the 600 households, 525 availed themselves of at least one credit facility during April 2009 to March 2010. Out of these 525 borrower households, 293 households have used only one loan during this period. During the corresponding period, the numbers of households with two, three, and four loans are 140, 53, and 39, respectively. Therefore, the total number of loan cases gathered from this survey is 888 (see Table 3). Of the 888 loan cases considered, 399 loans were offered by formal credit sources, 109 loans by semi-formal sources, and the remaining 370 loans by the informal sector. We followed the coding pattern suggested by Walter et al. (1987), which is given in Additional file 2: Appendix 2. To elucidate the characteristics of the villages and households covered in this study, we report summary statistics in Table 4. Method The analysis is based on 399 loan contracts extended by the formal financial sector among the 600 sample households. Earlier analyses (see, Datta 2003; Sahu et al. 2004) Table 2 Categories of households based on loan sources Source of loan Category of household Formal Semi-formal Informal Yes No No Households met their full credit demand entirely from the formal sector Yes Yes No Households met part of their credit demand from the formal sector Yes No Yes Yes Yes Yes No Yes No Households excluded from the formal sector but which used credit facilities from semi-formal and/or informal sectors No Yes Yes No No Yes Households solely dependent on the informal sector No No No Households without any credit facilities Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 4 of 20 http://www.agrifoodecon.com/content/2/1/9 of the determinants of the loan amount (Y) from the formal sector used ordinary least squares (OLS) regression methods. However, these OLS estimates of the various effects on the conditional mean of the loan amount, E(Y | covariates), may not truly represent the nature and size of those effects on the lower tail of the loan amount distribution. Here, we deviate from the earlier approach and use quantile regression (see, Koenker and Bassett 1978; Koenker 2005), which models the dependence of the conditional quantiles of Ygiven the covariates. This is helpful in models wherein extremes are vital. Here, quantile regression is expected to yield a comprehensive picture of the conditional distribution of loan off-take from formal creditors by rural customers given different village, household, and loan attributes for both the lower and upper quantiles. Let (y i ,x i ), i=1,…,n, be a sample from a population, where x i is a (K× 1) vector of regressors. Assuming that the τ th quantile of the conditional distribution of y i is linear in x i , let us denote ξxi;βτ ðÞas the τ th quantile of the distribution of y i |x i .Thatis, ξxi;βτ ðÞ≡inf y:FiyxÞ≥τ j ð½ Then, we assume that ξxi;βτ ðÞ¼x 0 iβτ, where, β τ is the unknown vector of parameters to be estimated for varying values of τwhere 0 < τ<1. By changing the value of τ, the entire distribution of yconditional on xcan be traced. The estimator for β τ comes from: minX n i ρτyi−x 0 iβτ  where ρ τ (u) denotes the function defined as ρτuðÞ¼ τuifu≥0 τ−1ðÞuifu<0  For further details, see Koenker (2005), pp. 28-67. In this study, values of τare taken as 1 10 ;1 2, and 9 10, which results in the 10 th quantile, median, and 90 th quantile, respectively. Here, y i is the amount of loan received by the sampled rural households from formal credit sources during 2009-10, and x i is a vector of independent variables describing the village, household, and loan characteristics. Results and discussion As an outcome of the exchange mechanism, a credit transaction is governed not only by the demand for credit but also by the lender’s willingness to extend credit. Furthermore, being an inter-temporal transaction, credit exposes the lender to postcontractual default risks. Therefore, the lender would take utmost care to ensure timely repayment of his or her dues. Hence, it is likely that the formal lender will consider a Table 3 Distribution of loans across households Number of loans received during 2009-10 Number of households Total loans 1 293 293 2 140 280 3 53 159 4 39 156 Total 525 888 Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 5 of 20 http://www.agrifoodecon.com/content/2/1/9 Table 4 Summary statistics of exogenous variables used in the empirical analysis A Variable Definition of variable Mean Minimum Maximum n PIRRI Percentage of agricultural land under assured irrigation in a particular village 58.60 0 100 14 PELECTRIC Percentage of households having access to electricity in a particular village 72.73 3 98 14 AGRI Average distance (in km) from agricultural infrastructures (e.g., Agriculture Produce Marketing Committee sub-yard, office of extension service providers, agricultural input retailers, and farm machinery dealers) 5.15 0.83 10.33 14 EDU Average distance (in km) from educational institutions 2.23 0 6.5 14 FININS Average distance (in km) from formal financial institutions (e.g., commercial banks and cooperatives) 3.76 0 8.33 14 FAMTOT Number of key social leaders (e.g., Panchayat Pradhan i , government extension officers, local school head masters, and financial institution officials) whom the household is familiar with 3.95 0 7 600 LOANAMT Amount (in Indian Rupee - INR j ) of loan made available 20656.19 200 600000 888 TCOST Borrower’s total transaction cost in INR 246.89 0 2960 888 RAINT Annualized interest rate in reducing balance 18.47 0 61 888 B Variable Definition of variable Frequency Cumulative 10 CASTRDUM Caste and religion:1 = Upper caste Hindu, 0 = Otherwise (Lower caste Hindu and minorities, such as Muslims) 186 (31.00) 414 (69.00) 600 (100) LANDD1 Household is minimum from marginal farmer category (Yes = 1, No = 0) 536 (89.33) 64 (10.67) 600 (100) LANDD2 Household is minimum from small farmer category (Yes = 1, No = 0) 199 (33.17) 401 (66.83) 600 (100) LANDD3 Household is minimum from medium/large farmer category (Yes = 1, No = 0) 69 (11.50) 531 (88.50) 600 (100) IRRISTATD1 At least a part of the agricultural land of the household has assured irrigation (Yes = 1, No = 0) 343 (57.17) 257 (42.83) 600 (100) IRRISTATD2 Agricultural land of the household is covered under assured irrigation (Yes = 1, No = 0) 188 (31.33) 412 (68.67) 600 (100) ALACDUM Household engaged in allied activity as fisheries/orchard/farm forestry (Yes = 1, No = 0) 93 (15.50) 507 (84.50) 600 (100) DWELLD1 Household possesses at least a mud built house (Yes = 1, No = 0) 580 (96.67) 20 (3.33) 600 (100) DWELLD2 Household possesses at least a mud and brick built house (Yes = 1, No = 0) 285 (47.50) 315 (52.50) 600 (100) DWELLD3 Household possesses a brick built house (Yes = 1, No = 0) 74 (12.33) 426 (87.67) 600 (100) SANITD1 Household having at least an open toilet (Yes = 1, No = 0) 438 (73.00) 162 (27.00) 600 (100) Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 6 of 20 http://www.agrifoodecon.com/content/2/1/9 variety of factors before deciding to approve a loan application. Further, for the loan applicant, the quantum of loan requirement would also be dependent on various factors. Below, we describe these factors and their relationship to the chance of loan approval by a formal lender as well as to the quantum of loan requirement. Factors affecting the supply of credit Since credit is extended to agricultural households who borrow on the basis of their land holding, such land holding acts as collateral from the lender’s perspective. The lender is assured that borrowers with a larger acreage of land would produce sufficient output to generate income to repay loans without difficulty. An assured irrigation facility increases the reliability of higher crop production and is, in turn, associated with a higher profit potential of the farmer applicant. This would be perceived positively by the lender and would carry a higher probability that loan applications for higher amounts will be considered favorably by the formal lender. Lenders may also prefer households that engage in allied activities such as cattle rearing and fishery, which are alternative earning sources to farming, as such activities may lower the probability of default in the event of crop failure. Access to information from extension agents is expected to help farmers adopt improved farming technology, thereby increasing the probability of higher yields. Lenders may prefer farmers with a possibility of registering more income from higher yields. The possession of landed house property as well as in-house toilet facilities would signal a resource-rich household in rural India, thereby increasing the household’s chance of receiving a higher amount in loans from the formal lender. Familiarity with key social leaders is expected to improve the credentials of the loan applicant and, hence, may be considered favorably by the formal lender. Since large tracks of land are traditionally owned by resource-rich upper caste Hindu families in India, belonging to such a family may also create a favorable impression on the lender. Indeed, banks may prefer households from villages with a higher coverage of assured irrigation and access to electrical power, as farmers from those villages are expected to undertake multiple-cropping and thus carry better profit potential as compared to households that do not have access to such public infrastructure support. Villages closer to agricultural infrastructure carry higher business potential and, thus, may be favored by formal lenders. Furthermore, proximity to formal financial institutions is expected to lead to a wider outreach of financial services as a result of minimum transaction costs incurred by the lender to service those clients. Banks may also prefer to lend in villages closer to educational institutions, as those villages would have a higher Table 4 Summary statistics of exogenous variables used in the empirical analysis (Continued) SANITD2 Household with modern toilet facility (Yes = 1, No = 0) 271 (45.17) 329 (54.83) 600 (100) PURPDUM Loan is for production purpose (Yes = 1, No = 0) 525 (59.12) 363 (40.88) 888 (100) SECDUM Loan is secured by marketable collateral (Yes = 1, No = 0) 294 (33.11) 594 (66.89) 888 (100) Note: Figures in parentheses are percentages. Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 7 of 20 http://www.agrifoodecon.com/content/2/1/9 number of educated households, who may use loan funds more judiciously and carry better employability in non-farm sectors and, hence, have sufficient income to repay loans. Lastly, formal creditors would prefer to lend for the purpose of production over that of consumption, since the former is associated with the possibility of generating positive surplus required to meet debt obligations. Factors affecting the demand for credit A large track of land with a good irrigation facility not only extends a positive signal to the formal lender but it also boosts the demand for credit to meet higher production expenses incurred from the purchase of improved varieties of seeds and other inputs, wage payments to laborers, the rental of agricultural machinery such as tractors, and electricity charges. Moreover, households with allied activities such as cattle rearing and fishery could assume a higher debt exposure, as income from alternative sources would act as a contingency for avoiding any default risks in the event of production failure. Households with access to information from extension agents are expected to adopt improved farming technology, thus, they may require a higher quantum of loan to meet higher production costs. Furthermore, upper caste Hindu families with large land holdings may demand higher loan amounts to meet higher costs of cultivation. Peasant households belonging to villages with access to assured irrigation and electrical power are expected to undertake multiple-cropping and thus require a higher amount in loans compared to villagers farming under rain-fed mono-cropping systems. Villages closer to agricultural infrastructures are expected to carry a higher business potential and are therefore more likely to borrow money for expanding agricultural or other business activities. In addition, agricultural households from villages located near formal financial institutions are expected to perform more transactions and avail themselves of a higher quantum of loans due to the minimum transaction costs involved in dealing with formal lenders. Similarly, villages located near educational institutions are expected to have a higher number of educated households that are more informed about the lending process of formal financial institutions and thus have a better scope of obtaining higher loan amounts from the formal sector. Households may apply for loans for agricultural production or other purposes, namely consumption purpose. Such loans may or may not be backed by marketable collateral. These, as well as the total ex ante monetary costs involved in negotiating the loan and annual interest rates, represent loan attributes that may affect both access to credit from the formal lender and the quantum of loan. The variable definitions given in Table 4 include all of the variables mentioned above. Initially, we outlined the effects of these variables on the loan amount, by using OLS. Next, three quantile regression models, for the 10 th quantile, 50 th quantile, and 90 th quantile, were fitted with the loan amount as the dependent variable and the various village, household, and loan attributes as explanatory variables. The model parameters were estimated using the PROCREG and QUANTREG procedures of the SAS software. State-level variations in both OLS and quantile regression were captured by state dummies. Although data were collected from 3-5 villages for Chhattisgarh, Maharashtra, and West Bengal, these villages are from a single cluster of an agro-climatic region. Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 8 of 20 http://www.agrifoodecon.com/content/2/1/9 Table 7 Loan amount across quantiles under varying situations (Continued) 24 AP 90 4 2 2 90 Yes Yes Yes Yes 25 GUJ 50 10 6 8 50 Yes Yes No No 26 GUJ 50 10 6 8 50 Yes Yes No No 27 GUJ 50 10 6 8 50 Yes Yes No No 28 GUJ 90 4 2 2 90 Yes Yes Yes Yes 29 GUJ 90 4 2 2 90 Yes Yes Yes Yes 30 GUJ 90 4 2 2 90 Yes Yes Yes Yes Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 15 of 20 http://www.agrifoodecon.com/content/2/1/9 Table 7 Loan amount across quantiles under varying situations (Continued) Case no. DWELLD2 SANITD1 FAMTOT CASTRDUM PURPDUM SECDUM TCOST RAINT LOAN AMT 10% 50% 90% 1 No Yes 2 No Yes Yes 500 10 7089 30934 75653 2 No Yes 2 No Yes Yes 500 10 7804 44717 87152 3 No Yes 2 No Yes Yes 500 10 9405 44618 124442 4 Yes Yes 4 Yes Yes No 300 4 5579 22667 81223 5 Yes Yes 4 Yes Yes No 300 4 6294 36450 92723 6 Yes Yes 4 Yes Yes No 300 4 7894 36351 130012 7 No Yes 2 No Yes Yes 500 10 12198 36813 82157 8 No Yes 2 No Yes Yes 500 10 12913 50596 93656 9 No Yes 2 No Yes Yes 500 10 14513 50497 130946 10 Yes Yes 4 Yes Yes No 300 4 10687 28546 87727 11 Yes Yes 4 Yes Yes No 300 4 11402 42329 99227 12 Yes Yes 4 Yes Yes No 300 4 13003 42230 136516 13 No Yes 2 No Yes Yes 500 10 3238 23861 65198 14 No Yes 2 No Yes Yes 500 10 3952 37644 76697 15 No Yes 2 No Yes Yes 500 10 5553 37545 113987 16 Yes Yes 4 Yes Yes No 300 4 1727 15594 70768 17 Yes Yes 4 Yes Yes No 300 4 2442 29377 82268 18 Yes Yes 4 Yes Yes No 300 4 4042 29278 119557 19 No Yes 2 No Yes Yes 500 10 6284 9353 55307 20 No Yes 2 No Yes Yes 500 10 6999 23136 66806 21 No Yes 2 No Yes Yes 500 10 8600 23037 104096 22 Yes Yes 4 Yes Yes No 300 4 4774 1086 60877 23 Yes Yes 4 Yes Yes No 300 4 5489 14869 72377 Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 16 of 20 http://www.agrifoodecon.com/content/2/1/9 Table 7 Loan amount across quantiles under varying situations (Continued) 24 Yes Yes 4 Yes Yes No 300 4 7089 14770 109666 25 No Yes 2 No Yes Yes 500 10 7272 55445 230534 26 No Yes 2 No Yes Yes 500 10 7987 69228 242034 27 No Yes 2 No Yes Yes 500 10 9588 69129 279323 28 Yes Yes 4 Yes Yes No 300 4 5709 42497 206908 29 Yes Yes 4 Yes Yes No 300 4 6423 56280 218408 30 Yes Yes 4 Yes Yes No 300 4 8024 56181 255697 Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 17 of 20 http://www.agrifoodecon.com/content/2/1/9 perspective, since the financial inclusion policy set by the government of India aims to ensure adequate and affordable credit flow even to resource-poor households. We see from this comparison that the loan off-take is skewed towards resource-rich households, whereas it should ideally be in the other direction. Table 7 also shows that even among borrowers in the 90 th quantile of the conditional loan distribution, marginal farmers received only approximately 80 percent of the loan amount of their counterparts who belong to the category of medium and large farmers (compare case 25 with case 27). Conclusions This study was motivated by the growing debate concerning the large-scale exclusion of rural households from the ambit of formal credit delivery channels, despite the continuous thrust of the state towards inclusive growth. Unlike studies that cover only limited geographical areas, this study includes six Indian states with varying institutional settings. Several conclusions related to policy can be drawn from this analysis. First, the formal credit market in rural areas is sensitive to the characteristics of the village and the applicant’s household and production. Second, there is considerable variation in the quantum of loans received from formal creditors. Third, even among borrowers in the 90 th quantile of the conditional loan distribution, marginal farmers received only approximately 80 percent of the loan amount of their counterparts belonging to the category of medium and large farmers. In other words, even where poor clients managed to access formal-sector creditors, the quantum of credit off-take is lower than that of their resource-rich counterparts. Therefore, despite government pushes for financial inclusion, the distribution of credit from the formal financial sector remains skewed towards resource-rich rural households. Endnotes a Households with operational land holdings of up to 1 hectare, above 1 hectare and up to 2 hectare, above 2 hectare and up to 4 hectare, and more than 4 hectare are termed marginal, small, medium, and large farmer, respectively. Though medium farmer and large farmer are separate categories in government documents, we have placed them together in this study as medium and large farmers (i.e., households with operational landholdings of more than 2 hectares). b Mid-year rural population (source: Centre for Monitoring Indian Economy) divided by the total number of a commercial bank’s rural branches (source: Quarterly Statistics on Deposits and Credit of Scheduled Commercial Banks, Reserve Bank of India) and cooperative societies (source: Trend and Progress of Banking in India, Reserve Bank of India). c Scheduled commercial banks (SCB) comprise those banks that are registered under the second schedule of India’s central bank (i.e., Reserve Bank of India’s Act, 1934). They include both public and privately owned banks. Their operation is mostly spread over multiple districts and even across states. d Regional Rural Banks (RRBs) came into existence in 1975, with their primary goal being to cater to rural clients. The operation of RRBs is limited to few Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 18 of 20 http://www.agrifoodecon.com/content/2/1/9 districts. RRBs are jointly owned by the Government of India, the respective State Government, and the sponsor bank, which is either an SCB or State Cooperative Bank. e Cooperative banks are financial institutions wherein the customers are the owners. Primary agricultural cooperative societies (PACS) are localized units owned by local people who share a common interest and who are involved in deposit mobilization and lending activities. In cases where cooperatives are also engaged in non-financial activities such as the selling of agricultural input, the running of consumer stores, and the trading of agricultural output, these banks are known as Multi-purpose Cooperative Societies (MPACS). f Self Help Group (SHG) is a group of approximately 20 financially challenged people who belong to a common socioeconomic strata that have come together initially to save money and to use the corpus for lending among themselves. After a certain time period of savings, SHGs may approach the bank for a loan. This process obviates the need for marketable collateral, which banks usually ask to secure repayment, by instituting peer pressure and a group guarantee mechanism. g Microfinance Institutions (MFI) form groups of poor people and act as lenders to the groups after sourcing loan from other financial institutions such as commercial banks. h If a village is too large in terms of the number of households, some representative hamlets of that village (not exceeding 300 households) were used for complete enumeration. In such cases, only those households covered under complete enumeration constitute the population from which the sample was drawn. i The elected head of a local village-level governance structure. j During 2009-10: US$1 = Indian Rupee (INR) 47. Additional files Additional file 1: Appendix 1. Distribution of lenders across sample villages in certain states in India. Additional file 2: Appendix 2. Coding of ordered exogenous variables. Competing interests The authors declare that they have no competing interests. Acknowledgements The authors would like to thank the editor, two anonymous referees, Jeffrey B. Nugent and Vasant P. Gandhi for their valuable comments, which have led to substantial improvements to this paper. Authors have benefitted from conversations on this topic with Samar K. Datta and M.S. Sriram, as well as from the comments of the participants of the 88 th Annual Conference of the Western Economic Association held at Seattle. Authors are grateful to Samar K. Datta for making available the dataset used in this study. We also acknowledge doctoral fellowship and dissertation support grant extended to the first author by Indian Institute of Management Ahmedabad. Author details 1 Economics and Business Environment, Indian Institute of Management Raipur, GEC Campus, Raipur 352015, India. 2 Production and Quantitative Methods, Indian Institute of Management Ahmedabad, Vastrapur, Ahmedabad 380015, India. Received: 19 June 2013 Accepted: 12 May 2014 References Bassett G, Chen H (2001) Quantile style: return-based attribution using regression quantiles. Empir Econ 26(1):293–305 Basu P (2006) Improving Access to Finance for India’s Rural Poor. Manuscript No. 36448 The World Bank, Washington, D.C Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 19 of 20 http://www.agrifoodecon.com/content/2/1/9 Boucher SR, Guirkinger C (2007) Risk, wealth, and sectoral choice in rural credit markets. Am J Agric Econ 89(4):991–1004 Buchinsky M (1998) Recent advances in quantile regression models: a practical guideline for empirical research. J Hum Resour 33(1):88–126 Burgess R, Pande R (2005) Do rural banks matter? Evidence from the Indian social banking experiment. Am Econ Rev 95(3):780–795 Carter MR (1989) The impact of credit on peasant productivity and differentiation in Nicaragua. J Dev Econ 31(1):13–36 Datta SK (2003) An institutional economics approach to the problems of small farmer credit in India. Working Paper 2003-07-01, Indian Institute of Management, Ahmedabad Datta S, Ghosh A (2013) Explaining access to credit by rural households: results based on a study of several states in India. Working Paper 2013-02-03. Indian Institute of Management, Ahmedabad Eide E, Showalter MH (1998) The effect of school quality on student performance: a quantile regression approach. Econ Lett 58(3):345–350 Gonzalez-Vega C (1984) Credit-rationing behavior of agricultural lenders: the iron law of interest-rate restrictions. In: Dale W, Adams D, Graham H, Von-Pischke JD (ed) Undermining Rural Development with Cheap Credit. Westview Special Studies in Social, Political and Economic Development series (pp.78-95). Westview Press, Boulder, Colo, London Guirkinger C, Boucher SR (2008) Credit constraints and productivity in Peruvian agriculture. Agric Econ 39(2):295–308 Kochar A (1997) An empirical investigation of rationing constraints in rural credit markets in India. J Dev Econ 53(2):339–371 Koenker R, Bassett G (1978) Regression Quantiles. Econometrica 46(1):33–50 Koenker R (2005) Quantile Regression, 1st edition. Cambridge University Press, Cambridge Nakamura H, Nakajima T (2011) A credit market in early stages of economic development. Econ Lett 112(1):42–44 National Sample Survey Organisation (2005) Household Indebtedness in India as on 30-06-2002, All-India Debt and Investment Survey, NSS 59th Round (January-December 2003), Report No.501 (59/18.2/2). Ministry of Statistics and Programme Implementation, Government of India. http://mospi.nic.in/Mospi_New/site/inner.aspx? status=3&menu_id=31 Pal D, Laha AK (2013) Sectoral Choice of Credit in Rural India. IIMA Working Paper 2013-03-01 Indian Institute of Management, Ahmedabad Pal S (2002) Household sectoral choice and effective demand for rural credit in India. Appl Econ 34(14):1743–1755 Reserve Bank of India Quarterly Statistics on Deposits and Credit of Scheduled Commercial Banks, various issues. http:// www.rbi.org.in/scripts/QuarterlyPublications.aspx?head=Quarterly%20Statistics%20on%20Deposits%20and%20Credit %20of%20Scheduled%20Commercial%20Banks Reserve Bank of India Trend and Progress of Banking, various issues. http://www.rbi.org.in/scripts/AnnualPublications. aspx?head=Trend%20and%20Progress%20of%20Banking%20in%20India Sahu GB, Madheswaran S, Rajasekhar D (2004) Credit constraints and distress sales in rural India: evidence from Kalahandi District, Orissa. J Peasant Stud 31(2):210–241 Walter SD, Feinstein AR, Wells CK (1987) Coding ordinal independent variables in multiple regression analyses. Am J Epidemiol 125(2):319–323 Winkelmann R (2006) Reforming health care: evidence from quantile regressions for counts. J Health Econ 25(1):131–145 doi:10.1186/s40100-014-0009-y Cite this article as: Pal and Laha: Credit off-take from formal financial institutions in rural India: quantile regression results. Agricultural and Food Economics 2014 2:9. Submit your manuscript to a journal and benefi t from: 7 Convenient online submission 7 Rigorous peer review 7 Immediate publication on acceptance 7 Open access: articles freely available online 7 High visibility within the fi eld 7 Retaining the copyright to your article Submit your next manuscript at 7 springeropen.com Pal and Laha Agricultural and Food Economics 2014, 2:9 Page 20 of 20 http://www.agrifoodecon.com/content/2/1/9