EXECUTIVE SUMMARY - of a study by NABARD in 2015
Amongst recent policy interventions implemented to revive the languishing agricultural sector in India, those pertaining to agricultural credit have been very much in the forefront. In particular, three major policy initiatives have shaped the past decade in institutional credit to agriculture. The policy of doubling of institutional credit to agriculture between 2004-05 and 2006-07 (over the 2004-05 base year) marked the first attempt to alleviate the financial constraints of farmers. In 2008-09, the Agricultural Debt Waiver and Debt Relief Scheme (ADWDRS) was introduced to waive specific outstanding debts for a large number of small farmers; this was followed by the interest subvention scheme, that sought to remedy the perceived negative impact of the waiver on loan repayment culture by rewarding timely repayment with loans carrying lower interest rates. These three schemes combined have, implicitly and explicitly, resulted in an increasing volume of institutional credit to agriculture. Whereas credit accounted for only 16% of the total value of paid out inputs in the triennium ending (TE) 1998-99, and 26.3% in TE 2003-04, by the end of the decade, in TE 2011-12, it had risen to as high as 80.3% of the estimated total paid out costs of inputs. Despite its importance, little is known about the effectiveness of credit in supporting agricultural growth as represented by the GDP and indeed the very nature of the relationship between formal agricultural credit and agricultural GDP. This research project is a modest effort in this direction. While acknowledging that questions of the impact of credit on agricultural output or value addition or productivity are best addressed through a textured understanding of household behaviour and micro studies, this study is based on the premise that aggregate secondary data too can reveal some of these important relationships. This study uses state-level data to examine the relationship between institutional credit to the agricultural sector and agricultural GDP at the national level.
Despite serious limitations of aggregation, which typically disregards distributional issues and often masks more than it reveals, it is also true that any systematic or pervasive relationship should reflect in aggregate data and offers a level of generalization not available in small scale surveys. The goal of this study is four-fold: How productive is institutional credit to the agricultural sector? What has been the trend since mid-1990s? What are the pathways through which credit impacts agriculture? How, if at all, have these pathways changed over the years? The analysis covers the period 1995-96 to 2011-12 using data that includes all major states within India. The study also conducts, data permitting, a disaggregate analysis of two sub-periods – the first phase denoting the Pre-doubling period (1995-96 to 2003-04) and the second representing the Post-doubling period (2004-05 to 2011-12). Where feasible, the study replicates the analysis at the state level. In this study, credit is conceptualized as an enabling input that influences agricultural GDP primarily via use of variable inputs and through investments in fixed capital that support agricultural production. To the extent that credit can also contribute to consumption smoothing of borrowers or better their capacity for risk bearing, credit could have a non-specific influence on agricultural GDP via variables that are typically unobserved by the researchers. To parse this complex relationship given the limitations of data, a combination of three approaches are used. The first is a simple model that regresses agricultural GDP on current credit flow using state level data. The second method estimates a hybrid profit-production function that regresses agricultural GDP on a vector of relevant inputs, prices and agricultural credit flow during that year. This is a direct approach to estimating the relationship between credit and agricultural GDP in reduced form. The possible indigeneity of credit is addressed by the use of a control function that “controls” for the estimated endogenous component of observed credit flow. The third method represents the `pathways approach’ which estimates input demand as a function of credit, among other things and controlling for indigeneity. The coefficients representing the responsiveness of input use to institutional credit are then used as components to construct the total impact of agricultural credit on agricultural GDP. The impact of credit on agricultural output is thus derived as the sum of the contribution of credit to the use of specific inputs, capital or the cropping pattern, weighted by the contribution of these to the total value of agricultural production.
The range of estimates obtained from the various methods suggest that the credit elasticity of agricultural GDP for the entire period 1995-96 to 2011-12 is 0..21, i.e. a 10% increase in institutional credit flow to agriculture in current prices is associated with a 2.1% increase in agricultural GDP the following year expressed in current prices. When controlling for prices represented by the wholesale price index, a 10% increase in nominal credit associated with a 0.97% increase in real GDP, indicating that inflation might be eroding some of gains made in nominal terms. Compared with these results in the simple one period lag model (method 1), the estimated credit elasticity is 0.04 when the model controls for the use of inputs and a vector of input and output prices and for the possible indigeneity of credit through a control function approach (method 2). The structural model incorporating the pathways through which credit influences agricultural GDP (method 3) yields estimates of credit elasticity of 0.21. These results however have weak statistical significance. The results from a period-wise disaggregate analysis is less conclusive. While the first model suggests that the elasticity continues to be statistically significant but has weakened in the post-doubling period, the other two approaches, one that controls for prices and input and the other the captures the pathways suggest that the relationship between credit and agricultural GDP may have declined, but none of the estimated credit elasticity coefficients are statistically significant implying that the hypothesis that the credit has no association with agricultural GDP cannot be rejected.
At the state level, estimates of credit elasticity of agricultural GDP from a simple one-period lag model, the only feasible option given the data, vary mostly between 0.05 and 0.7. For only a few exceptions, the credit elasticity turned out to be statistically insignificant. Further, at the state level, the time trend of elasticity estimates varies across states. In some states the relationship appears to have strengthened post doubling whereas for others it has weakened. Further clarity and insight can only be obtained through detailed case studies or primary surveys, owing to the paucity of state level data that precludes modelling efforts at the state level. This study goes beyond to understand the precise role of credit, in other words, the pathways through which it influences or is associated with agricultural GDP. The findings from the analysis suggest that all the inputs, are highly responsive to an increase in institutional credit to agriculture, after controlling for input prices, output prices, sectoral composition of agriculture, area sown and so on. A 10 % increase in credit flow in nominal terms leads to an increase by 1.7% in fertilizers (N, P, K) consumption in physical quantities, 5.1% increase in the tonnes of pesticides, 10.8% increase in tractor purchases. The credit elasticity of new pump sets energized is however not statistically significant. A disaggregate analysis, for the pre-doubling and post-doubling phases, suggest that the relative importance of the inputs have changed. Whereas in the pre-doubling phase, fertilizers were statistically significantly responsive, in the post-doubling phase credit appears to have a strong relationship with tractors. Overall, it seems quite clear that input use is sensitive to credit flow, whereas GDP of agriculture is not. This seems to indicate that the ability of credit to engineer growth in agricultural GDP is impeded by a problem of productivity and efficiency where the increase in input use and adjustments in the pattern of input use are not (yet) translating into higher agricultural GDP. Credit seems therefore to be an enabling input, but one whose effectiveness is undermined by low technical efficiency and productivity
Amongst recent policy interventions implemented to revive the languishing agricultural sector in India, those pertaining to agricultural credit have been very much in the forefront. In particular, three major policy initiatives have shaped the past decade in institutional credit to agriculture. The policy of doubling of institutional credit to agriculture between 2004-05 and 2006-07 (over the 2004-05 base year) marked the first attempt to alleviate the financial constraints of farmers. In 2008-09, the Agricultural Debt Waiver and Debt Relief Scheme (ADWDRS) was introduced to waive specific outstanding debts for a large number of small farmers; this was followed by the interest subvention scheme, that sought to remedy the perceived negative impact of the waiver on loan repayment culture by rewarding timely repayment with loans carrying lower interest rates. These three schemes combined have, implicitly and explicitly, resulted in an increasing volume of institutional credit to agriculture. Whereas credit accounted for only 16% of the total value of paid out inputs in the triennium ending (TE) 1998-99, and 26.3% in TE 2003-04, by the end of the decade, in TE 2011-12, it had risen to as high as 80.3% of the estimated total paid out costs of inputs. Despite its importance, little is known about the effectiveness of credit in supporting agricultural growth as represented by the GDP and indeed the very nature of the relationship between formal agricultural credit and agricultural GDP. This research project is a modest effort in this direction. While acknowledging that questions of the impact of credit on agricultural output or value addition or productivity are best addressed through a textured understanding of household behaviour and micro studies, this study is based on the premise that aggregate secondary data too can reveal some of these important relationships. This study uses state-level data to examine the relationship between institutional credit to the agricultural sector and agricultural GDP at the national level.
Despite serious limitations of aggregation, which typically disregards distributional issues and often masks more than it reveals, it is also true that any systematic or pervasive relationship should reflect in aggregate data and offers a level of generalization not available in small scale surveys. The goal of this study is four-fold: How productive is institutional credit to the agricultural sector? What has been the trend since mid-1990s? What are the pathways through which credit impacts agriculture? How, if at all, have these pathways changed over the years? The analysis covers the period 1995-96 to 2011-12 using data that includes all major states within India. The study also conducts, data permitting, a disaggregate analysis of two sub-periods – the first phase denoting the Pre-doubling period (1995-96 to 2003-04) and the second representing the Post-doubling period (2004-05 to 2011-12). Where feasible, the study replicates the analysis at the state level. In this study, credit is conceptualized as an enabling input that influences agricultural GDP primarily via use of variable inputs and through investments in fixed capital that support agricultural production. To the extent that credit can also contribute to consumption smoothing of borrowers or better their capacity for risk bearing, credit could have a non-specific influence on agricultural GDP via variables that are typically unobserved by the researchers. To parse this complex relationship given the limitations of data, a combination of three approaches are used. The first is a simple model that regresses agricultural GDP on current credit flow using state level data. The second method estimates a hybrid profit-production function that regresses agricultural GDP on a vector of relevant inputs, prices and agricultural credit flow during that year. This is a direct approach to estimating the relationship between credit and agricultural GDP in reduced form. The possible indigeneity of credit is addressed by the use of a control function that “controls” for the estimated endogenous component of observed credit flow. The third method represents the `pathways approach’ which estimates input demand as a function of credit, among other things and controlling for indigeneity. The coefficients representing the responsiveness of input use to institutional credit are then used as components to construct the total impact of agricultural credit on agricultural GDP. The impact of credit on agricultural output is thus derived as the sum of the contribution of credit to the use of specific inputs, capital or the cropping pattern, weighted by the contribution of these to the total value of agricultural production.
The range of estimates obtained from the various methods suggest that the credit elasticity of agricultural GDP for the entire period 1995-96 to 2011-12 is 0..21, i.e. a 10% increase in institutional credit flow to agriculture in current prices is associated with a 2.1% increase in agricultural GDP the following year expressed in current prices. When controlling for prices represented by the wholesale price index, a 10% increase in nominal credit associated with a 0.97% increase in real GDP, indicating that inflation might be eroding some of gains made in nominal terms. Compared with these results in the simple one period lag model (method 1), the estimated credit elasticity is 0.04 when the model controls for the use of inputs and a vector of input and output prices and for the possible indigeneity of credit through a control function approach (method 2). The structural model incorporating the pathways through which credit influences agricultural GDP (method 3) yields estimates of credit elasticity of 0.21. These results however have weak statistical significance. The results from a period-wise disaggregate analysis is less conclusive. While the first model suggests that the elasticity continues to be statistically significant but has weakened in the post-doubling period, the other two approaches, one that controls for prices and input and the other the captures the pathways suggest that the relationship between credit and agricultural GDP may have declined, but none of the estimated credit elasticity coefficients are statistically significant implying that the hypothesis that the credit has no association with agricultural GDP cannot be rejected.
At the state level, estimates of credit elasticity of agricultural GDP from a simple one-period lag model, the only feasible option given the data, vary mostly between 0.05 and 0.7. For only a few exceptions, the credit elasticity turned out to be statistically insignificant. Further, at the state level, the time trend of elasticity estimates varies across states. In some states the relationship appears to have strengthened post doubling whereas for others it has weakened. Further clarity and insight can only be obtained through detailed case studies or primary surveys, owing to the paucity of state level data that precludes modelling efforts at the state level. This study goes beyond to understand the precise role of credit, in other words, the pathways through which it influences or is associated with agricultural GDP. The findings from the analysis suggest that all the inputs, are highly responsive to an increase in institutional credit to agriculture, after controlling for input prices, output prices, sectoral composition of agriculture, area sown and so on. A 10 % increase in credit flow in nominal terms leads to an increase by 1.7% in fertilizers (N, P, K) consumption in physical quantities, 5.1% increase in the tonnes of pesticides, 10.8% increase in tractor purchases. The credit elasticity of new pump sets energized is however not statistically significant. A disaggregate analysis, for the pre-doubling and post-doubling phases, suggest that the relative importance of the inputs have changed. Whereas in the pre-doubling phase, fertilizers were statistically significantly responsive, in the post-doubling phase credit appears to have a strong relationship with tractors. Overall, it seems quite clear that input use is sensitive to credit flow, whereas GDP of agriculture is not. This seems to indicate that the ability of credit to engineer growth in agricultural GDP is impeded by a problem of productivity and efficiency where the increase in input use and adjustments in the pattern of input use are not (yet) translating into higher agricultural GDP. Credit seems therefore to be an enabling input, but one whose effectiveness is undermined by low technical efficiency and productivity
