Syndicate content

Add new comment

Are Banks Responsive to Credit Demand Shocks in Rural Economies?

Sankar De's picture

How effectively does the commercial banking system respond to idiosyncratic shocks to the income and consumption of the households in the rural sector of an economy, and does it make extra credit available at a reasonable cost? This is an enormously important question. Farming operations in emerging economies are still heavily dependent on rainfall. Intermittent failure of monsoons and other weather-related vicissitudes often upset the normal income and consumption patterns of many rural households. However, a survey of the literature on rural financial markets finds few studies on the responsiveness of the financial intermediation system to credit demand shocks.

By contrast, local bilateral credit and insurance arrangements with landlords, moneylenders, family and friends, or group-based mutual savings and insurance arrangements such as rotating savings and credit associations (ROSCAs) have received much attention in the literature (see, for example, Coate and Ravallion 1993; La Ferrara 2003; Townsend 1995; Genicot and Ray 2002). However, the risks to income and consumption that rural households face are typically correlated, as they arise from common external shocks such as floods and famines, and the pool of savings is usually limited. As a result, local markets fail to offer adequate diversification opportunities and funds at a reasonable cost. Consequently, individuals and households in the rural economy are left facing considerable residual risk, with no option but to adopt costly and inefficient strategies to smooth income or consumption. A number of such strategies have been discussed in the existing literature, including scattering plots of cultivable land (McCloskey 1976; Townsend 1993) and  opting for a more diversified mix of crops and nonfarm production activities at the price of a lower average return, adjustment of inter-temporal labor supply in response to shocks (Kochar 1999), labor bonding (Srinivasan 1989; Genicot 2002), selling investment assets to smooth consumption (Rosenzweig and Wolpin 1993) and several other options. Not surprisingly, the welfare implications of the strategies are typically very negative.


While the discussions in the existing literature have been insightful and advanced our understanding of the strategies, they rest on the implicit premise that risk diversification opportunities offered by the existing system of financial intermediation, namely the commercial banking system, are either very limited or altogether missing in the rural economy. However, the premise itself has remained largely unexamined. Conning and Udry observe in the introduction to the Handbook of Agricultural Economics (2007, p. 6-7): “while these studies have advanced our understanding of local bilateral financial contracting and mutual insurance within poor communities, the study of financial intermediation has remained relatively neglected.”

In an ongoing research project Siddharth Vij, currently a student in the finance PhD program at Stern School, New York University, and I attempt to redress this imbalance in the existing literature. We examine the responsiveness of the commercial banking system in the rural economy of India to exogenous shocks to credit demand following a drought, and conduct multiple tests with extensive data of bank credit and rainfall at the district level in India. The results that we have obtained so far are revealing.1

Why did we choose this particular setting? Agriculture remains a major sector of the economy in India and other emerging countries. It accounts for about 19 percent of the GDP in India. The importance of the sector to India is actually much higher because of its role in job creation and poverty alleviation in the countryside. About two-thirds of the Indian population depends on the sector for their livelihood. Rainfall and supply of credit are two key determinants of agricultural output in India and other emerging economies. There is substantial evidence in the existing academic literature as well as professional reports and government policy papers in India that rainfall is an important determinant of Indian farm output. For example, using rainfall and crop yield data for a panel of 272 districts over 32 years, Cole et al. (2009) report that, on average, a one-standard deviation increase in rainfall results in a 3% - 4% increase in the value of output in their sample. Correspondingly, there is wide recognition of the importance of rainfall to agriculture and national income in different agencies of the government, policy forums, think tanks, and the popular press.2 

The importance of agricultural credit, the other key determinant of farm output in India, is inherently tied to the heavy dependence of agriculture on rainfall. Undoubtedly, poor rainfall causes idiosyncratic shocks to income and consumption of the households in the rural sector of India and in other emerging economies, which in turn creates shocks to credit demand. It results in a high cost of credit and pervasive rural indebtedness, notes Rakesh Mohan, an erstwhile Deputy Governor of the Reserve Bank of India (RBI), in an RBI paper in 2006. His observation suggests lack of responsiveness of the commercial banking system to credit demand shocks. However, it is a commonsense observation based perhaps on his experience, but not on an empirical analysis of extensive rainfall and farm credit data. From time to time, the financial press and other media outlets in India have also reported isolated instances of the effect of poor rainfall on bank credit availability. But so far there has been no systematic study.

To motivate our approach, we use a simple model that links commercial banks’ responsiveness to exogenous credit demand shocks to the bankers’ incentive structure. The model incorporates a standard feature of the rural credit cycle and a few typical features of bankers’ incentives that have been documented by other researchers (Banerjee, Cole, and Duflo 2005; Banerjee and Duflo 2008). The farmers seek bank credit for their operating expenses (seeds, fertilizers, etc.) during the crop planting season and, in a year of normal rainfall, pay off their debt from the proceeds of the harvest. In a year of poor rainfall, their ability to pay off their current debt is diminished, and some of them default. But they still need a fresh loan for the next planting season. The bankers face a penalty if they recognize a bad loan, and prefer to bail out the defaulting farmers and give them fresh loans. In many cases, bailouts substitute the probability of a larger future default for the certainty of a smaller current default. But this is not so in the case of drought-driven defaults, because a year of drought is typically followed by a year of normal rainfall. In our sample, using one measure of drought, a district experiences a drought in two consecutive years in 15.3% of the cases, compared with an average occurrence of drought in 11.3% of the cases. By a second measure of drought, the corresponding numbers are 15.3% and 17.5%, respectively.

The model offers several testable predictions. First, the volume of outstanding agricultural credit extended by banks increases following a year of poor rainfall, driven by those farmers who are unable to pay off their current loans but still get fresh loans. Second, the credit increase occurs in the intensive margin (the average size of the existing loans) rather than in the extensive margin (the number of loan accounts). The banks typically have more information about the types of their current borrowers than they have about new borrowers. Following a difficult year their information set is more refined, and they are better able to target better farmers within their current pool of borrowers for more credit. Third, while bank managers with both public sector and private sector banks have an incentive to bail out a farmer facing default and not to recognize a non-performing loan, public sector bank managers typically lack incentives of the other kind, namely to provide additional loans to borrowers who are proven to be good and do not default on their current loans (Banerjee, Cole, and Duflo 2005; Banerjee and Duflo 2008). As a result, private banks are observed to respond more positively than public banks to combined credit demand from farmers of all types.

To test the hypotheses, we use extensive panel data of droughts and agricultural credit at the district level. Our data on rainfall and drought come from the Indian Meteorological Department (IMD). We use the Standardized Precipitation Index (SPI) as our primary measure of drought conditions.3 We obtained the SPI data for 458 Indian districts for the period 1993-2003 from a study by researchers associated with the IMD.4 We also construct an alternative measure of drought, using the percentage of normal (PN) rainfall method. In this method, the rainfall in a particular year is compared with the district’s long period average (LPA) rainfall. If the rainfall is less than 75% of the LPA, it indicates a drought in the district. We were able to calculate the PN measures for 334 districts for the period 1993-2010. The SPI measure, available for a larger number of districts, and the PN measure, available for a longer period, yield more or less the same number of district-year observations for our tests (about 3,300). Our data on bank credit come from the Reserve Bank of India’s annual publication Basic Statistical Returns of Scheduled Commercial Banks (BSR). The publication provides data on the amount of credit outstanding, occupation-wise, at the end of the fiscal year (March 31 in the case of India) in each district, as well as the number of accounts for which credit is outstanding. We have these data from 1993-94 to 2009-10.

Our data offer two types of variation at the district level which are important for our purpose: considerable cross-sectional variation between credit observations and time-series variation between occurrences of droughts at the district level. We exploit the variation in the data to conduct clean difference-in-difference tests to identify the causal impact of unanticipated changes in the demand for farm credit due to exogenous changes in rainfall on the supply of farm credit. The panel setting enables us to include district fixed-effects in our regression models, ruling out spurious correlations due to time-invariant cross-sectional variations. Similarly, year-fixed effects control for yearly variation in macroeconomic and other factors that may affect the supply of agricultural credit. We also use region-year fixed effects to control for time trends in credit supply so that the regression coefficients reflect the relationships between droughts and agricultural credit net of trends unrelated to droughts.

The test results confirm our hypotheses. The results indicate that banks increase agricultural credit following drought-affected years compared with normal years. The increase is of the order of 4–5 percent by the SPI measure of drought and 3-4 percent by the PN measure. The results are significant at the 1% level in each case. Further examination suggests that the observed increase in outstanding credit consists primarily of fresh loans, and not addition of overdue interest and other charges added to old loans. The increase in agricultural credit represents an increase in the intensive margin, rather than an increase in the extensive margin. We also find that bank ownership makes a difference. The percentage increase in agricultural lending appears to be significantly higher in the case of private banks than public sector banks, although both types increase their lending. Interestingly, there is no significant difference in additional credit origination following a drought between districts that are drought-prone and districts that are not. On the whole, we find positive evidence on the role of the commercial banking system in the rural economy.

Our work contributes to several other existing literatures as well. In standard economic theory of financial intermediation, the primary role of financial institutions is to channel capital from depositors and other savers to uses with high marginal returns. Substantial empirical evidence from emerging capital markets suggests that this role is performed poorly. Using a sample of loan data from a public sector bank in India, Banerjee and Duflo (2008) use a government-mandated directed lending program as a natural experiment to establish that many of the firms in their sample were severely credit constrained, and that the marginal rate of return to capital was very high for them. To explain their findings, the authors cite aversion to managerial risk-taking in public sector banks. If a loan performs poorly, the managers face a penalty but, on the other hand, a good loan decision does not bring them rewards. It should be noted that the above studies focus on credit for small businesses, not rural credit. Conceivably, agricultural loans to farmers in drought-affected areas are likely to have high marginal returns, or at least higher marginal returns than during normal times. A drought typically depletes their savings, causing serious capital scarcity.5 Our finding that banks in rural India increase agricultural credit following a drought compared with non-drought years suggests that allocation of bank credit is not always sub-optimal. However, we note that our evidence is not inconsistent with inertia and lack of managerial risk-taking in public sector banks, as suggested by Banerjee and Duflo (2008), since we do find evidence that private banks respond more positively than public banks to drought-driven credit demand.

Although it is not the main focus of our research, the paper also provides evidence on the political economy of credit allocation following a drought. Public sector banks are known to be vulnerable to political capture, and loans can be targeted in ways that many other government expenditures, such as public works projects, cannot. In our paper, we investigate the joint effects of elections and droughts on agricultural credit and test whether firm credit increases in election years following a drought after controlling for the effect of the drought itself. For all bank groups together, our difference-in-difference-in-difference tests do not find evidence of an increase in agricultural credit beyond what is typically observed following a drought-affected year. However, when we examine credit extended by different bank groups, public sector banks appear to increase agricultural credit in an economically significant manner, while other bank groups do not.

A final contribution of the paper is that it presents the first systematic study of the relationship between the two primary determinants of farm output in India: rainfall and supply of credit.

References
Banerjee, A.V., S.A. Cole, and E. Duflo. 2005. “Bank Financing in India.” In Wanda Tseng and David Cowen, eds., India’s and China’s Recent Experience with Reform and Growth, Palgrave Macmillan.
Banerjee, A.V., and E. Duflo. 2008. “Do Firms Want to Borrow More? Testing Credit Constraints Using a Targeted Lending Program.” Working Paper, MIT.
Cole, S.A. 2009. “Financial Development, Bank Ownership, and Growth. Or, Does Quantity Imply Quality?" Review of Economics and Statistics 91(1): 33-51.
Conning, J., and C. Udry. 2007. “Rural Financial Markets in Developing Countries.” In Robert Evenson and Prabhu Pingali, eds., Handbook of Agricultural Economics, Elsevier.
De, Sankar, and Siddharth Vij. 2013. “Are Banks Responsive to Exogenous Shocks to Credit Demand in Rural Economies? District-level Evidence from India.” Working Paper, Indian School of Business.
Genicot, G. 2002. “Bonded Labor and Serfdom: A Paradox of Voluntary Choice.” Journal of Development Economics 67(1): 101-27.
Kochar, A. 1999. “Smoothing Consumption by Smoothing Income: Hours-of-Work Responses to Idiosyncratic Agricultural Shocks in Rural India.” Review of Economics and Statistics 81(1): 50-61.
McKee, T.B., N.J. Doesken, and J. Kleist. 1993. "The relationship of drought frequency and duration to time scales." 8th Conference on Applied Climatology, pp. 179-84.
Mohan, R. 2006. “Agricultural Credit in India: Status, Issues and Future Agenda.” Economic and Political Weekly 41(11): 1013-23.
Rosenzweig, M.R., and K.I. Wolpin. 1993. “Credit Market Constraints, Consumption Smoothing, and the Accumulation of Durable Production Assets in Low-income Countries: Investments in Bullocks in India.” Journal of Political Economy 101(2): 223-44.
Srinivasan, T.N. 1989. “On Choice among Creditors and Bonded Labour Contracts.” In Pranab Bardhan, ed., The Economic Theory of Agrarian Institutions, Oxford University Press.
Townsend, R.M. 1999. “The medieval village economy: A study of the Pareto mapping in general equilibrium models.” Frontiers of Economic Research series. Princeton University Press.
 


[1] A preliminary discussion of some of the findings is available in a working paper entitled
Are Banks Responsive to Exogenous Shocks to Credit Demand in Rural Economies? District-level Evidence from India.
[2] The Financial Express, a major financial newspaper in India, carried the following item on August 24, 2009: “Approximately 25% of the country is affected by drought and agricultural output is set to plummet this year. Lower income for rural workers will in turn be a huge drag on private consumption, an important driver of India's economic expansion.”  The drought in question affected 25% of the country and was associated with 29% below normal rainfall during the busy Kharif season (June - September) in 2009.
[3] The SPI is a drought index developed in McKee, Doesken, and Kleist (1993). The deviation from the median rainfall over a long period is standardized to arrive at the index value for a particular year. A value of less than minus one indicates drought.
[4] Pai, D.S., L. Sridhar, P. Guhathakurta and H.R. Hatwar (2010). “Districtwise Drought Climatology of the Southwest Monsoon Season over India Based on Standardized Precipitation Index (SPI).” National Climate Centre Research Report No. 2/2010, India Meteorological Department
[5] Rosenzweig and Wolpin (1993) have documented that farmers in India sell their main investment assets, such as bullocks, to smooth consumption during times of poor weather.