Agricultural Credit in India in the 2000s: Growth,Distribution and Linkages with Productivity

Abstract

The period of the 2000s witnessed a sharp revival in agricultural credit in India that was largely policy induced. There were emerging shifts in institutional, functional and regional distributions of agricultural credit during the decade. This study attempts to explore the relationship between agricultural credit and agricultural production/productivity. The state-level panel model attempted in this article suggests a positive impact of the intensity of agricultural credit on total factor productivity in agriculture. The impact was relatively stronger with respect to direct agricultural credit. A case study of the (combined) state of Andhra Pradesh also suggests a positive association between agricultural credit and agricultural production. The study lends credence to the policy approach of including agriculture, the largest employer in the Indian economy, as a sector for priority credit in India. It also highlights the point that the sector deserves continued policy support in order to move onto a sustainable and higher growth path.

JEL Classification: Q10, Q14, Q18, N55

Keywords

Agricultural Credit Agricultural Productivity Priority Sector Lending Capital Formation Dynamic Panel

1. Introduction

Agriculture remains the mainstay of the Indian economy despite a decline in its share in the country’s domestic product, on account of its central role in employment generation. The growth in agriculture is, thus, expected to augur well for poverty reduction in the country. Furthermore, given its demand and supply-related linkages with other sectors, sustained growth in agriculture not only supports a higher growth of the economy but also lends stability to the growth pattern. Finally, agricultural growth is key to the sustenance of domestic food security and also to improve the overall trade balance by strengthening exports and reducing the reliance on food imports.

Given that it impacts almost every aspect of the Indian economy, agriculture has remained the focus of all major developmental policy initiatives in the country. Among these, the banking sector policy can be considered the most noteworthy. The provision of formal finance to agriculture and its allied activities was considered instrumental in meeting various production-related needs in agriculture. In the late 1960s, India adopted a multi-agency approach towards agricultural credit, comprising credit co-operatives (the oldest rural credit institution), commercial banks (introduced to the field of agricultural credit after bank nationalisation) and Regional Rural Banks (RRBs) (a hybrid of commercial banks and co-operatives specially created to serve agricultural and rural credit needs).

Commercial banks were brought into the field of agricultural credit with the help of the priority sector lending programme. This programme was aimed at allocating credit to certain sectors of social and economic priority, which included a sub-target for agriculture and its allied activities. Notwithstanding the liberalisation of various sectors including the banking sector during the period of economic reforms since the early 1990s, the priority sector lending programme has remained an integral part of banking sector policy in India even today. Furthermore, agriculture and its allied activities are the single-largest sector with a targeted share of 45 per cent in the overall credit given to all priority sectors.

During the 2000s, particularly after 2004–05, a number of new policy measures relating to agriculture were adopted with a view to step up agricultural growth and also to address concerns relating to agrarian distress being raised since the second half of the 1990s from several parts of India.¹ The deceleration in the growth of formal credit to agriculture during the 1990s was regarded as one of the major reasons for the prevalence of distress among agriculturists during this period (GoI, 2007).² Hence, the revival in formal credit to agriculture was considered a key component of the new policy measures adopted in the 2000s. Consequently, a Comprehensive Credit Policy was adopted in 2004–05 with the objective of reviving (doubling) the amount of agricultural credit over a span of three years (see details in Ministry of Agriculture, 2007).

Studies have shown that agricultural credit growth picked up significantly in the 2000s, partly in response to the policy measures. In fact, studies even show that the revival in agricultural credit had set in from the beginning of the 2000s but was accentuated after 2004–05 (Ramakumar & Chavan, 2007). The 2000s, particularly after 2004–05, also witnessed a remarkable increase in the growth of domestic product from agriculture and allied activities. Contrary to the general feature in agriculture, the period after 2004–05 did not witness a single year of negative growth in agriculture (Deokar & Shetty, 2014). Prima facie, this period also witnessed an improvement in the productivity in agriculture measured in terms of yield per hectare for all major food and non-food crops (Kumar, Mittal & Hossain, 2008; Chaudhary, 2012).

There is, however, a need to probe deeper in order to quantify the productivity in agriculture, expressed in terms of total factor productivity, and to empirically capture its linkages with agricultural credit, which is attempted in this article. Here, we focus on the decade of the 2000s because, as discussed already, this decade offers an interesting case study of a period when (formal) agricultural credit, production and crop yields posted increased growth.

This article first attempts to estimate total factor productivity (productivity from now on) in agriculture using a state-level panel data model. The literature on productivity estimation in the context of Indian agriculture is rather limited, particularly for the sector as a whole. Most studies have focused on crop-wise productivity estimation and few attempts have been made to estimate productivity at the state level.² Furthermore, the attempts made to analyse trends in productivity growth in Indian agriculture during the 2000s are even fewer (Chaudhary, 2012; Kumar et al., 2008).

This article then analyses the relationship of productivity with agricultural credit, an exercise that has not been attempted in the Indian context so far.

It also focuses on the state of (combined) Andhra Pradesh as a case study and, using a district-level panel data model for the state, brings out the relation between agricultural credit and production.³ Andhra Pradesh offers an interesting case study because it has generally enjoyed a high share in overall agricultural credit in the country and has also witnessed an increase in this share during the 2000s.

The rest of the article is as follows. Section 2 provides a theoretical motivation for a study on the relation between agricultural credit and production/productivity, and certain empirical concerns in estimating this relation. Section 3 provides a review of the existing studies on this subject. Section 4 discusses major policy developments relating to agricultural credit. Section 5 brings out certain stylised facts about the trends in the growth and distribution of agricultural credit in the 2000s, compared and contrasted with the earlier decade. Section 6 provides a discussion on data sources and methodology used in the article. The discussion on methodology mainly relates to the estimation of productivity, in general, and productivity in agriculture, in particular. Section 7 provides a discussion on two empirical exercises, as discussed in the foregoing paragraphs, to analyse the relation between agricultural credit and production/productivity. Finally, Section 8 provides concluding observations from the article.

2. Agricultural Credit and Production/Productivity—Theoretical Motivation and Empirical Concerns

There is a vast body of economic literature that establishes the association of finance with economic growth.⁴ A pioneering work by Raymond Goldsmith in the late 1960s established the positive correlation between financial development and economic activity. Studies in more recent decades have carried forward the earlier work but on more robust empirical grounds in strengthening the understanding of how finance influences economic growth (Demirgüç-Kunt & Levine, 2004; Levine, Loayza & Beck, 2000).

In the case of agriculture, credit is expected to influence production growth through two major channels. First, the increased supply of formal credit can lead to greater availability of working capital and investment in fixed capital. This can essentially relax the liquidity and investment constraints in agriculture. Second, credit also enables consumption smoothing for cultivators. This assumes considerable importance in a developing economy like India, given the uncertainties associated with agricultural production. In the literature, this is often referred to as the consumption-smoothing effect of credit (Das, Senapati & John, 2009).

However, empirically, any study analysing the relation between credit and output is beset with several methodological concerns.⁵ First, the relation between credit and output, perhaps much more than with other economic relations, shows considerable spatial variation resulting from agro-ecological factors, land use, cropping pattern and holding sizes, among others.

Second, any attempt to model the relation between credit and output can be subject to an endogeneity bias, resulting primarily from simultaneity or co-determinance of the dependent (output) and independent (credit) variables. This often renders the estimation of causality between credit and output difficult. In simpler terms, higher agricultural credit can have an impact on agricultural output, but higher output in itself can lead to a greater demand for agricultural credit, making it difficult to disentangle and exactly quantify the effect of one variable on the other.

Third, as already discussed, credit can influence output through not just an increased use of purchased inputs but also through consumption smoothing and increasing the ability to take on greater risks. Hence, the specification of variables is key to appropriately estimate the complete impact of credit on output. However, such specification is often difficult in the absence of variables that can capture this impact; the impact of credit in terms of consumption smoothing for farmers is particularly difficult to capture.

As regards credit and productivity, theoretically, the relation is expected to be positive. This is because the supply of credit enables the usage of more advanced productivity-inducing variable inputs as well as fixed investments in physical infrastructure in the form of irrigation, road connectivity and electrification.

However, empirically, it is an arduous task to establish this relation. In an excellent review of the issues involved in the role played by credit in improving productivity, Sriram (2007) notes that credit appears as part of the total investments into agriculture that include self-financing, borrowing from variety of informal sources along with the usage of labour and other inputs. An increase in the supply of private credit from formal sources may not be conducive to productivity increases without being accompanied by supporting public investments (Sriram, 2007). Hence, investments in irrigation, research and development related to improved varieties of seeds and other inputs and farming techniques may all contribute towards increased growth in productivity. An increase in formal credit may not be conducive for productivity gains also because this increase may merely substitute informal credit instead of actually raising the supply of agricultural credit as a whole (Sriram, 2007).

There are also certain data limitations in analysing the relation between credit and production/productivity in the Indian context.

The first relates to the inability to factor in the role played by informal sources of credit in this relation between credit and output due to paucity of data. National- and state-level data on informal sources are available only on a decennial basis through the All-India Debt and Investment Survey (AIDIS).⁶ Hence, Sriram’s conjecture (2007), about whether the increase in formal credit substitutes informal credit and hence may not lead to an increase in productivity, is difficult to test empirically except perhaps with a field-level survey.

Second, the impact of credit on output can be captured fully if the information on all three agencies of agricultural credit is available. However, the data on credit co-operatives come with a considerable lag and, unlike the Basic Statistical Returns of Scheduled Commercial Banks in India (BSR), updated district-level information on agricultural credit from credit co-operatives is not available from a single source. In a sense, this may render the analysis of agricultural credit partial in nature. Again, this limitation can be overcome if we undertake a field-level survey of debt and production profiles of cultivators.

From the foregoing discussion, it appears that a field-level survey may be a more appropriate method of analysing the linkage between credit and production/productivity. However, any field-level survey will be indicative and will not facilitate any generalised conclusions about this linkage. Hence, a better approach could be to analyse this linkage using a comprehensive district-level panel data. However, even this exercise is rendered difficult, given the non-availability of data at the required frequency or accuracy at the district level for all districts in India. While the data on bank credit to agriculture, with annual frequency and considerable accuracy up to the district level, are available in the BSR from the Reserve Bank of India (RBI), the information on cropping patterns, irrigated area, capital formation, labour inputs and other inputs is not uniformly available at the district level. Furthermore, the data on domestic product at the district level are not fully standardised and published by all state Economics and Statistics Directorates. This seriously limits the construction of a comprehensive district-level data panel for estimating the relation between credit and production/productivity.

Another concern relates to the estimation of productivity. Crop yields can be taken as a rough indicator of productivity. However, for understanding the linkage of credit and productivity, an aggregate estimate of farm-level productivity has to be worked out in the form of total factor productivity. In the absence of field-level or comprehensive district-level data on major factors of production, however, this estimation is also rendered difficult.

To sum up, notwithstanding a seemingly straightforward linkage between credit and production/productivity, the empirical estimation of this linkage suffers from various methodological and data concerns. Data concerns are particularly acute in the case of India.

3. Agricultural Credit and Production/Productivity—Review of earlier Studies

Notwithstanding the methodological and data concerns, the literature is replete with studies, conducted both for India as well as other developing economies, of the relation between credit and production but not productivity. This research interest in developing economies with regard to agricultural credit could have possibly stemmed from a policy-driven approach towards enhancing formal credit to agriculture across many of these economies since the 1960s.

Illustratively, Armas, Osorio and Moreno-Dodson (2010) have brought out the role played by public spending on agriculture, including the provision of subsidised credit to agriculture for stepping up agricultural growth in Indonesia. Khandker and Faruquee (2003) have noted a statistically significant impact of formal credit on agricultural output and household consumption in Pakistan. Ammani (2012) brought out the positive relation between crop credit and crop production in Nigeria. Carter and Weibe (1990) studied the relation between agricultural credit and production in Kenya. Shrestha (1992), however, observed a somewhat weak relation between credit and output growth in agriculture in Nepal.

In the Indian context, Binswanger and Khandker (1992) estimated the impact of formal credit on output in the farm sector based on district-level data and found a positive impact of the same, although the impact was observed to be greater in the non-farm sector, compared to the farm sector. Studies also showed a positive elasticity at the all-India level regarding the use of mechanised and other inputs with respect to direct agricultural credit (Bhalla & Singh, 2010). However, the elasticity varied significantly across regions, bringing out the role of region-specific factors in measuring the influence of credit on output and input usage; in technologically backward regions, the elasticity was found to be much higher than other regions (Bhalla & Singh, 2010).

Taking district-level data for the four most populous states in India, Das et al. (2009) observed a positive and immediate impact of direct agricultural credit (credit provided directly to farmers) on agricultural output. However, they found variations across states in the degree of impact of agricultural credit on output. Subbarao (2012) noted a unidirectional causality from credit to output at the all-India level. In a recent study, Narayanan (2014), based on her state-level study of credit elasticity of agricultural output, asserted that the conception that credit is ‘ineffective’ in influencing output is ‘misplaced’ (p. 19). Her study suggested a positive impact of credit on purchased non-labour inputs in agriculture.

In sum, barring a few exceptions, the existing literature on India has shown a positive impact of formal credit on agricultural production. However, each of these studies brings out the time-specific and region-specific factors in determining this impact. Importantly, none of these studies contain an analysis of the linkage between credit and productivity, as attempted in this article.

4. Major Policy Developments Relating to Agricultural Credit with special Reference to policies in the 2000s

Historically, rural credit, in general, and agricultural credit, in particular, were almost entirely under the control of informal sources. In 1951, as much as 92.7 per cent of the total debt of cultivator households was owed to informal sources (RBI, 1956). Various official committee reports, including the All-India Rural Credit Survey (AIRCS) of 1951–52, noted the usurious and coercive lending and recovery practices of informal sources, particularly moneylenders.

With systematic and continued policy-driven efforts towards building the formal institutions of rural and agricultural credit, the situation has changed considerably over the last six decades. Among the various measures for the development of the institutional set-up for agricultural credit, the most noteworthy was the nationalisation of commercial banks in two phases in 1969 and 1980. The explicit objective of nationalisation was to increase the commitment of banks to agricultural production and rural development. With bank nationalisation, a multi-agency approach to rural and agricultural credit evolved, comprising co-operatives and scheduled commercial banks (SCBs) along with RRBs that were created in 1974 to provide credit exclusively to the poor and under-privileged sections from rural areas.

Furthermore, to increase credit flow to certain priority sectors, namely agriculture, the concept of priority sector lending was introduced in 1969 and formal guidelines were issued in 1972 by the RBI. In 1974, banks were advised to raise the share of these sectors in their aggregate advances to 33.3 per cent by March 1979 and to 40 per cent by 1985. Further, a target for agricultural lending was fixed at 15 per cent to be achieved by March 1985 and 18 per cent by March 1990. Since then, there have been several changes in the scope and sub-targets of priority sector lending.

Moreover, contrary to recommendations in the Report of the Committee on the Financial System (RBI, 1991) of reducing and then phasing out priority sector lending norms, these norms were retained, although changes were made in the definitions of priority sectors to suit changing economic needs. Since 2015–16, the priority sector lending target for RRBs has been increased to 75 per cent of their outstanding advances, compared to 40 per cent earlier.

In June 2004, the government announced a Comprehensive Credit Policy, which envisaged the doubling of agriculture credit over the next three years, that is, by 2006–07. SCBs, co-operatives and RRBs were mandated to step up formal credit to agriculture by 30 per cent every year. In subsequent annual budgets, the government announced targets for credit to agriculture to ensure adequate credit flow to the sector. Further, the Interest Subvention Scheme, introduced by the central government in 2006–07, was aimed at providing direct credit to farmers at a concessional rate of 7 per cent for crop loans up to ₹ 0.3 million. To incentivise prompt repayment, an additional interest subvention of 3 per cent was introduced in 2011–12. In 2013–14, the Interest Subvention Scheme was extended to private sector banks as well.

Finally, the Agricultural Debt Waiver and Debt Relief Scheme of 2008, which can be broadly regarded as a part of the comprehensive credit policy, provided debt relief of about ₹ 500 billion and a one-time settlement relief on overdue loans of ₹ 100 billion through SCBs, RRBs and credit co-operatives.

In sum, there have been several institution-building initiatives for agricultural credit following bank nationalisation. There have also been several new policy initiatives for increasing the supply and reducing the cost of agricultural credit in the 2000s.

5. Key Stylised Facts about Growth and distribution of Agricultural Credit in the 2000s

In this section, we present the changing character of agricultural credit and its trends in terms of growth and distribution by type and regions (states/districts) during the 2000s.⁷

5.1 Growth in Agricultural Credit

During the 2000s, there was a distinct change in the nature of agricultural credit. As noted earlier, agricultural credit is supplied by three formal agencies, namely, SCBs, credit co-operatives and RRBs. Among these three, SCBs showed a continued rise in their share in the 2000s (Figure 1). They came to account for about 67 per cent of the total agricultural credit in India by the end of the decade. This was the highest level ever achieved by the SCBs since the early 1980s. The SCBs along with the RRBs accounted for 73 per cent of total agricultural credit.

The changing nature of agricultural credit, with commercial banks emerging as the dominant agency in the 2000s, was partly on account of a high growth in agricultural credit supplied by commercial banks during this decade. Total credit to agriculture and allied activities by commercial banks (including RRBs) (henceforth, referred to as agricultural credit) at the all-India level grew at the rate of 25.5 per cent per annum (in nominal terms) during this decade as against 11.3 per cent per annum during the 1990s.⁸ As a result of this growth, the bank credit-to-GDP ratio picked up significantly in agriculture in the 2000s (Figure 2).

Figure 1

Source: Handbook of Statistics on the Indian Economy, RBI, various issues.

Figure 2

Source: Database on the Indian economy, RBI.

The credit-to-GDP ratio is commonly used as an indicator of the depth and intensity of credit in a given sector/economy. This decade also witnessed a considerable increase in this overall bank credit-to-GDP ratio. Notwithstanding the increased penetration, however, it is noteworthy that the credit-to-GDP ratio in agriculture remained less than half of the overall credit-to-GDP ratio.

5.2 Some Features of the Distribution of Agricultural Credit

5.2.1 Distribution by type

The two major components of agricultural credit have been direct and indirect agricultural credits.⁹ Direct credit to agriculture referred to loans to individual farmers (including self-help groups and joint liability groups) directly engaged in agriculture and allied activities. Its definition was widened in recent years to also include part of the loans given to corporates including farmers’ producer companies of individual farmers, partnership firms and co-operatives of farmers directly engaged in agriculture and allied activities. Indirect credit traditionally referred to loans given to institutions and organisations that supported agricultural production. As per the changes introduced in the priority sector guidelines over the years, indirect credit also came to include credit given to warehouse operators, agricultural input dealers, loans to RRBs and micro-finance institutions for on-lending to farmers as well as the portion of loans to corporates and other entities that does not get included under direct credit.¹⁰

The period of the 2000s, which was a period of high growth in agricultural credit, witnessed a sharp spurt in both direct and indirect agricultural credit, but the growth in indirect credit was clearly higher than that in direct credit (Table 1).¹¹ Given the rise in the growth of indirect credit, the 2000s witnessed an increase in the share of indirect agricultural credit in total agricultural credit (Figure 3). Although the growth as well as share of indirect credit seemed to have waned by the beginning of the next decade, its rise during the 2000s cannot be overlooked (Table 1 and Figure 3).

Table 1
Average Rates of Growth of Agricultural Credit (% per annum)

Period Total Agricultural Credit Direct Agricultural Credit Indirect Agricultural Credit

1990–91 to 1999–2000 11.3 8.6 30.1

1999–2000 to 2009–10 25.5 24.8 27.3

2009–10 to 2013–14 13.5 18.3 –2.3

Period	Total Agricultural Credit	Direct Agricultural Credit	Indirect Agricultural Credit
1990–91 to 1999–2000	11.3	8.6	30.1
1999–2000 to 2009–10	25.5	24.8	27.3
2009–10 to 2013–14	13.5	18.3	–2.3

Source: Handbook of Statistics on the Indian Economy, RBI and Basic Statistical Returns of Scheduled Commercial Banks in India, RBI.

Figure 3

Source: Handbook of Statistics on Indian Economy and Basic Statistical Returns of Scheduled Commercial Banks in India, RBI.

Table 2

Region-wise Outstanding Agricultural Credit and Total Credit of SCBs (% share)

Region	2001	2005	2010	2012
Outstanding credit
Northern region	21.2	23.9	22.6	20.6
Northeastern region	0.9	0.8	0.9	1.0
Eastern region	8.0	8.3	8.5	8.5
Central region	18.9	13.9	18.2	16.2
Western region	14.8	15.7	13.7	12.7
Southern region	36.3	37.4	36.2	41.0
All-India	100.0	100.0	100.0	100.0
Total credit
Northern region	19.7	22.8	22.6	23.9
Northeastern region	1.0	1.1	0.9	0.8
Eastern region	8.1	9.4	8.4	8.0
Central region	9.4	6.4	8.0	7.6
Western region	33.9	31.5	32.4	31.5
Southern region	28.0	28.9	27.8	28.1
All-India	100.0	100.0	100.0	100.0

Source: Database on the Indian Economy (DBIE), RBI.

5.2.2 Regional/State/District-wise distribution

Agricultural credit, like overall bank credit in India, has traditionally been concentrated in the southern and northern regions of the country. Over the 2000s—the period of high growth in agricultural credit—this pattern of concentration remained broadly unchanged. In 2012, the southern and northern regions together accounted for about 62 per cent, with the southern region alone accounting for 41 per cent of total agricultural credit in the country (Table 2). In fact, the share of the southern region was on a rise during the 2000s, particularly over the second half of the decade (see Table A.1).

Probing further into the regions, we first identified the states that had witnessed an increase in their shares in agricultural credit during the 2000s. These were Haryana, Punjab, Rajasthan, West Bengal, Maharashtra, Andhra Pradesh, Kerala, Tamil Nadu and Delhi, which were some of the most populous states in the country, and mainly states in the northern and southern regions except for Maharashtra and West Bengal.

We then identified 15 districts with the largest share in agricultural credit in 2012 and traced changes in the shares of these districts between 2005 and 2012 (Figure 4). The major findings from this exercise are:

Figure 4

Source: Compiled from BSR, RBI.

There were large disparities in the spatial distribution of agricultural credit across districts.

In 2012, the top 15 districts had a share of about 21 per cent.

In 2012, Delhi accounted for about 4 per cent of total agricultural credit in the country, followed by Greater Mumbai with a share of 3 per cent. These two districts were followed by 13 districts, with shares ranging between 0.5 per cent and 2 per cent. Of these, six districts were from Andhra Pradesh, including Hyderabad, Guntur, West Godavari, East Godavari, Krishna and Prakasam.

Importantly, between 2005 and 2012, there were signs of increasing dispersion of agricultural credit. The shares of Delhi and Greater Mumbai, which together accounted for about 11 per cent of the agricultural credit in 2005, witnessed a decline to 7 per cent by 2012. The gainers in this period were the southern districts of Guntur, West Godavari, East Godavari, Prakasam, Ernakulam and Krishna—nearly all from Andhra Pradesh.

The following stylised facts were discussed in this section: (a) there was an increase in the growth of agricultural credit in the 2000s. A key driver of this growth was indirect credit, although direct credit also posted an increase during this period; (b) the decade of the 2000s witnessed an increased concentration of agricultural credit in the southern region, particularly in the state of Andhra Pradesh.

The analysis offers a background to the empirical exercises undertaken in this article to bring out how the high growth in agricultural credit impacted agricultural productivity across major Indian states and, more so, how it influenced agricultural production in Andhra Pradesh, a state showing a growing concentration of agricultural credit.

6. Data Sources and Methodology

In this article, we attempt two exercises: first, we estimate productivity using state-level panel data and then study the linkages between agricultural credit (from SCBs and RRBs) and productivity. The panel included the 14 most populous states in the country,¹² and these included the eight states identified in the earlier section, which had shown an increase in their shares in agricultural credit in the 2000s. As discussed already, either a farm-level or district-level data panel may be more suited for estimating the relation between credit and production/productivity. However, in the absence of such detailed disaggregated information, a preliminary attempt has been made in this article to estimate the long-term relation between agricultural credit and productivity using state-level panel data.

Our second exercise is a case study of the (combined) state of Andhra Pradesh using panel data at the district level to estimate the linkage between agricultural credit and agricultural production for the state.

Any study of the relationship between credit and production/productivity is beset with certain methodological concerns, as discussed earlier. To address these concerns, apart from credit, we have included a variety of inputs that are likely to influence productivity, which capture (a) farmer-specific inputs, namely, fertilisers; (b) inputs that depend on farmer-specific investments as well as related public investments, namely, electricity consumption and irrigation; and (c) inputs that depend entirely on public investments, namely, infrastructure of roads.

For the state-level panel data estimation, we work out an aggregated estimate for total factor productivity using the method suggested by Levinsohn and Petrin (the LP method) (2003). Generally, the estimation of the production function using ordinary least squares (OLS) gives inconsistent and biased estimates of the explanatory variables. This could be on account of a host of spatial-, time- and farm-specific unobservable factors. These unobservable factors might influence the usage of production inputs and usage of inputs is thus determined endogenously. Since the OLS framework assumes that production inputs are uncorrelated with omitted unobservable variables, it fails to address the endogeneity issue.

Hence, a semi-parametric method like the LP method is used in the literature for the estimation of the production function. Such a method uses some ‘intermediate’ inputs as proxies like electricity or other inputs besides labour and capital since intermediates may respond more smoothly to productivity shocks. The LP method is most suited for sectors that are prone to productivity shocks such as agriculture (see Levinsohn & Petrin, 2003).

Olley and Pakes (1996) also provide another type of non-parametric framework, comprising investment decisions to proxy for unobserved productivity. However, the data on variables capturing the investment decisions may not be readily available for the agricultural sector and hence the LP method is preferred.

More formally, LP method in brief is as follows:

Production function for a farm i in period t is

y_{i t} = β_{0} + β_{l} l_{i t} + β_{k} k_{i t} + v_{i t} + e_{i t}, t = 1, ..., T

(1)

where

y_it = natural logarithm of the farm’s output

l_it = 1 * J vector of variable inputs (labour)

k_it = 1 * k vector of observed state variables (capital)

The sequence {v_it : t = 1, ..., T} is unobserved productivity and {e_it : t = 1, ..., T} is a sequence of shocks.

Levinsohn and Petrin (2003) use intermediate inputs to proxy for unobserved productivity. The intermediate inputs m_it are expressed as a function of capital and productivity, that is,

m_{i t} = m_{t} (k_{i t}, v_{i t})

Provided the monotonicity condition is satisfied and materials inputs are strictly increasing in v_it, this function can be inverted, allowing us to express unobserved productivity as a function of observables, that is,

v_{i t} = g (k_{i t}, m_{i t}), t = 1, ...., T, w h e r e g (.) = m_{t}^{- 1} (.)

Using this expression, (1) can be written as

y_{i t} = β_{0} + β_{l} l_{i t} + β_{k} k_{i t} + β_{m} m_{i t} + g (k_{i t}, m_{i t}) + e_{i t}, t = 1, ..., T

(2)

Estimation of the production function is carried out in two stages using conditional moments. The final estimation is done in several steps (Levinsohn & Petrin, 2003) and bootstrap approximation is used to construct standard errors for the estimates of β_l and β_k. Using these estimates, the measure of productivity has been estimated.

For modelling the impact of agricultural credit and productivity growth, a panel model has been used. Here, agricultural credit is modelled as credit-to-GDP ratio in order to normalise credit penetration across states. A state-level panel model is expected to reasonably capture the spatial differences in productivity. Further, in order to address the problem of endogeneity in understanding the relation between credit and productivity, the dynamic panel methodology suggested by Arellano and Bond (1991) has been used. It uses lagged dependent variables, total factor productivity in agriculture in the case of this article, as instruments. The time period for this analysis is from 2000 to 2012.

The time period for the district-level study for Andhra Pradesh is also from 2000 to 2012. Here, agricultural domestic product per capita has been regressed on agricultural credit per capita along with a vector of other explanatory variables.

As already noted, there is a glaring absence of updated information on agricultural credit from informal sources and, also, on credit co-operatives in the Indian context. Hence, our focus in this article has been on commercial banks (including RRBs) as the key agency of agricultural credit. However, such an analysis can be considered as broadly representative, at least, of the formal credit situation, given the emerging dominance of commercial banks in the provision of agricultural credit in recent years, as shown earlier.

For the data on other economic variables including gross domestic product (GDP), capital formation and the consumption of agricultural inputs at the state and district levels, we have relied on a number of data sources as listed in Table A.2.

7. Findings from the Empirical Analysis

7.1 Agricultural Credit and Productivity—The State-level Panel

The descriptive statistics of the variables used in the exercise are presented in Table A.3 (Panel A). The production function estimated using the LP method shows capital with a positive and significant relation with agricultural production (Table 3).¹³

Table 3
Production Function Estimation

Explanatory Variable Coefficient p-value No. of Observations

LP method

Labour 0.042 0.808 147

Capital 0.379 0.001* 147

Robust OLS

Labour 0.182 0.000* 174

Capital 0.194 0.000* 174

Explanatory Variable	Coefficient	p-value	No. of Observations
LP method
Labour	0.042	0.808	147
Capital	0.379	0.001*	147
Robust OLS
Labour	0.182	0.000*	174
Capital	0.194	0.000*	174

Source: Authors’ calculations.

Notes: (1) * significant at the 1% level, (2) Wald test of constant returns to scale: Chi-square = 14.08 (p = 0.0002) and (3) R-square = 0.50.

Using the coefficient of capital and labour as given by the LP method, state-wise estimates for total factor productivity (TFP) were generated using residual method as follows:

TFP = NSDPAg – 0.042*Wagebill – 0.379*GFCFAg,

where

NSDPAg refers to net state domestic product in agriculture;

Wagebill refers to labour input; and

GFCFAg refers to gross fixed capital formation in agriculture.

The productivity growth, thus arrived at, was negative for 8 out of 14 states, a finding that was broadly similar to the finding of Chaudhary (2012).¹⁴

The TFP estimates were modelled in a dynamic panel framework with respect to a set of explanatory variables including credit intensity in agriculture as follows:

TFP = f(Diragrcratio, Totalagcratio, Rfall, Fertipergca, GCA, Roads, Electric),

where

Diragrcratio—ratio of direct agricultural credit to domestic product from agriculture;

Totalagcratio—ratio of total agricultural credit to domestic product from agriculture;

Rfall—deviations of rainfall from long period average;

Ferti—fertiliser consumption per unit of gross cropped area;

GCA—gross cropped area per capita;

Roads—log of road length per square km; and

Electric—electricity consumption for agricultural purposes.

We ran two alternate models, one involving total agricultural credit and another with direct agricultural credit. The two models are summarised in Table 4.

Table 4

Dependent Variable: TFP

Method: Arellano–Bond Dynamic Panel Data Estimation
Explanatory Variable	Model 1	Model 2
TFP(–1)	0.228 (0.009)	0.066 (0.392)
TFP(–2)	0.177 (0.012)	0.216 (0.002)
Totalagcratio (–1)	–	6.320 (0.014)
Diragrcratio (–1)	8.099 (0.000)	–
Rfall#	4.284 (0.001)	3.856 (0.003)
Ferti#	8.020 (0.007)	9.162 (0.002)
GCA#	4.185 (0.018)	5.194 (0.002)
Roads#	–1.698 (0.331)	1.048 (0.547)
Electric #	–0.346 (0.762)	0.231 (0.840)
Sargan test (p-value)$	0.22	0.22
Serial correlation test order 1 (p-value)	0.00	0.00
Serial correlation test order 2 (p-value)^	0.49	0.29
Number of observations	118	118
Number of instruments	64	65

Source: Authors’ calculations.

Notes: (1)^ The null hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation, (2) # in the regression, these variables are taken as logarithmic transformations, (3) $ the null hypothesis is that the instruments used are not correlated with the residuals and (4) figures in parentheses refer to p-value.

In these models, the generalised method of moments (GMM) estimation is used for dynamic panel data estimation. The models are estimated in levels by using lagged values of TFP and other explanatory variables as instruments, following Arellano and Bond (1991).

In these models, productivity lagged for one and two years had a positive association with productivity in the current period. This is expected as the current value of TFP will be affected by its lagged values due to persistence in productivity shocks.

In these two models, the intensity of agricultural credit with a one-year lag was found to be a significant determinant of agricultural productivity. The higher the credit intensity, the higher the productivity, ceteris paribus. This implied that credit enabled the adoption of productivity inducing technology and other fixed and variable inputs. However, the coefficient size for intensity of direct agricultural credit was found to be higher than that for total agricultural credit, which underlined the significance of direct credit for enhancing agricultural productivity.

Rainfall and gross cropped area also were significant determinants of productivity with positive signs. Fertiliser consumption also determined agricultural productivity positively. In Indian agriculture, the role of rainfall as a determinant of agricultural production has been highlighted often, as nearly 50 per cent of the total area under cultivation is still rainfed (Dehadrai, 2008). The kharif crop is directly affected by variations in the monsoon (Prasanna, 2014).¹⁵

Irrigation also plays an important role in increasing cropping intensity. It also enhances crop yields due to its complementarity with improved seed varieties and fertiliser use (Sharma, 2011). In fact, in our model also, we observed a moderate-to-high degree of correlation between the irrigation index and fertiliser consumption. Hence, we used gross cropped area as a proxy for the intensity of irrigation, given that a higher irrigation index is expected to increase cropping intensity.¹⁶

Finally, road length (as a reflection of improved infrastructure and connectivity to markets for getting better returns from agriculture) and electricity consumption were found to be non-significant variables in determining productivity. However, their signs were also found to be counter intuitive in Model 1. In the Indian context, there is limited evidence of the association between market connectivity and agricultural productivity.¹⁷ However, studies from other emerging economies show that market access facilitates specialisation and exchange transactions in rural areas and also leads to intensification of input use (Kamara, 2004). Market access also leads to higher competition in input and output markets and increased access to information on agricultural technology and other market opportunities.¹⁸

As noted already, apart from credit and rainfall, our attempt was to include inputs that are likely to influence productivity capturing farmer-specific inputs, inputs that partly depend on farmer-specific investments as well as related public investments and inputs that almost entirely depend on public investments. Accordingly, we observed that fertiliser consumption, as a farmer-specific input, had a positive association with productivity. Similarly, the intensity of irrigation captured through gross cropped area also positively determined productivity. The variables capturing public investments in roads or electricity consumption did not show a significant association with productivity.

However, it is to be noted that in choosing these variables, we were constrained by the availability of time-series data at the state level on various other likely indicators of productivity, and hence our conclusions about public investments in determining productivity need to be interpreted with caution.

The consistency of the GMM estimator in the model depends on the validity of the instruments. A necessary condition for the validity of the instruments is that the error term should be serially uncorrelated. To address these issues, two specification tests as suggested by Arellano and Bond (1991) have been given in Table 4. First, the Sargan test of over-identifying restrictions gives a satisfactory statistic corroborating the overall validity of the instruments. Second, under the test of auto-correlation, we accept the null suggesting that the error term differenced regression is not serially correlated at the second order. The consistency tests underline the robustness of these two models.

7.2 Agricultural Credit and Output—A Case Study of Andhra Pradesh

For estimating the relation between agricultural credit and production for Andhra Pradesh (combined), two alternative models were used, one based on total agricultural credit and the other with direct agricultural credit.¹⁹ The relation could be expressed as follows:

APAggdp = f(APTotalagricredit; APIrrig; APRfall, APFerti, APWage),

where

APAggdp—log of per capita district domestic product from agriculture (and allied activities);

APTotalagricredit—log of total agricultural credit per capita at district level;

APDireccredit—log of direct district agricultural credit per capita;

APIrrig—log of the district-level irrigation index;

APRfall—log of actual district-level rainfall;

APFerti—log ratio of fertiliser consumption to gross cropped area for districts

APWage—log of wage bill in agriculture at the district level.

An underlying hypothesis of the district-level model is that agricultural GDP is an increasing function of land, labour, fertilisers, irrigation and rainfall.

The two models are summarised in Table 5:

Table 5
Dependent Variable: APAggdp

Method: Random Effects Model

Explanatory Variable Model 1 Model 2

C –10.641
(0.000) –9.175
(0.000)

APTotalcredit 0.105
(0.026) –

APDireccredit – 0.241
(0.000)

APIrrig 0.147
(0.018) 0.132
(0.026)

APRfall 0.320
(0.001) 0.295
(0.002)

APFerti 0.082
(0.112) 0.067
(0.179)

APWage 0.231
(0.000) 0.201
(0.001)

R-squared (overall) 0.62 0.68

Method: Random Effects Model
C	–10.641 (0.000)	–9.175 (0.000)
APTotalcredit	0.105 (0.026)	–
APDireccredit	–	0.241 (0.000)
APIrrig	0.147 (0.018)	0.132 (0.026)
APRfall	0.320 (0.001)	0.295 (0.002)
APFerti	0.082 (0.112)	0.067 (0.179)
APWage	0.231 (0.000)	0.201 (0.001)
R-squared (overall)	0.62	0.68

Source: Authors’ calculations.

Note: Figures in parentheses refer to p-value.

The model specification for Andhra Pradesh is different from the state-level panel data model as the determinants of the district domestic product from agriculture are different from agricultural productivity. However, the difficulty in the choice of variables on account of the lack of availability of data, noted while discussing the state-level findings, was even more acute at the district level.

The results of the panel regression using the random effects model (as upheld by the Hausman specification test) show that all the explanatory variables had positive signs, as expected. While total credit had a significant impact on agricultural production, ceteris paribus, rainfall turned out to be the most important determinant of agricultural production under Model 1. Source-wise irrigation in 2012–13 revealed that 68.8 per cent of the irrigation in Andhra Pradesh was through tanks and wells, while 27.7 per cent of irrigation was provided by canals (Ministry of Agriculture, 2014). The dominance of tanks and wells as major sources of irrigation underscored the importance of rainfall in the agriculture of Andhra Pradesh (Ramanamurthy & Misra, 2012). However, in the second model, direct agricultural credit was the strongest determinant of agricultural production in the state. Similar to the state-wise panel model, the coefficient size of direct agricultural credit was found to be higher than that of total agricultural credit, which highlighted the greater significance of direct credit in determining agricultural production.

The impact of wage bills and irrigation, as inputs into agricultural production, was also observed to be positive and significant. In the regression, the wage bill was taken as a proxy for the labour component and irrigated area for capital formation in agriculture. Ceteris paribus, a higher wage bill was expected to lead to higher agricultural production. A higher wage bill could have been either due to an increased number of labourers or due to higher labour productivity reflected in higher wages. Furthermore, fertiliser consumption, as another agricultural input, had a positive impact on agricultural production but did not have a statistically significant coefficient.

8. Concluding Observations

Since the late 1960s agriculture has been a priority sector for Indian banking. This continued policy support has catapulted commercial banks into the most important formal source of agricultural credit in the country. In recent decades, agriculture has witnessed a considerable fall in its share in India’s GDP. However, as the sector remains the single-largest employer, with several backward and forward linkages with the other two sectors of the economy, it has been retained as a priority sector.

The second half of the 2000s witnessed a number of affirmative measures to revive the falling growth in agricultural credit, to boost agricultural growth and reduce distress in the rural areas. The findings from this study suggest that this period indeed witnessed a favourable and significant impact of agricultural credit on agricultural growth.

First, using a state-level panel for the decade of the 2000s—a period of largely policy-induced and high growth in agricultural credit—the study showed a positive impact of the intensity of agricultural credit on total factor productivity with a lag of one year. The impact was relatively stronger with respect to direct agricultural credit. Apart from credit, fertiliser consumption, gross cropped area and rainfall also showed a positive association with agricultural productivity. Second, the case study of the (combined) state of Andhra Pradesh also suggested a positive association between agricultural credit and agricultural production, ceteris paribus.

This study may be treated as a preliminary attempt to understand the relation between credit and productivity, an issue that has not been studied in the Indian context. The study, undertaken with a state-level data panel and then a district-level data panel for Andhra Pradesh, enables certain generalised conclusions about the linkages between credit and productivity in a period that witnessed high growth in both formal credit to agriculture and increased crop yields. However, a comprehensive district-level data panel can be a way forward for capturing the spatial and regional variations in production/productivity and credit more accurately. Furthermore, given the importance of informal sources of agricultural credit, particularly for the marginalised sections, capturing these sources may provide a more complete picture of agricultural credit in India.

It is well acknowledged that Indian agriculture is a sector beset with several concerns including low productivity and vulnerability to weather changes, and a resulting volatility in earnings. However, the structural shift that is seen in economic activity has not translated into a similar shift in employment generation for various reasons that are mostly external to the sector. Furthermore, credit to agriculture also includes credit to allied activities, including dairying, poultry and fishing, which have been seen to have an increasing role in employment generation for India’s rural population, and hence, given the tardy shift in employment generation away from agriculture and its allied activities, it is necessary to continue the policy-based support to this sector with credit as an important component of such support.

The findings from this study lend credence to the approach taken by the RBI of including agricultural credit as part of priority sector credit in India. It also highlights that for India to move onto a sustainable and higher growth path in agriculture, this sector deserves a continued support. It indicates that direct agricultural credit has a much larger impact and hence needs to be encouraged.

Footnotes

Acknowledgements

Acknowledgements: The authors thank Dr Raghuram G. Rajan, Dr Urjit R. Patel and Dr M.D. Patra for their encouragement in undertaking this study. They also thank Dr Saibal Ghosh, Dr N. Prabhala, Professor Sudha Narayanan and an anonymous referee for insightful comments on earlier drafts of the article and Professor N. Ramanamurthy for providing access to district-level data for Andhra Pradesh. The views expressed in the article are personal views of the authors and do not represent the views of the organisation to which they are affiliated.

Appendix

Panel B: Variables for District-wise Analysis for (combined) Andhra Pradesh

Variables	No of Observations	Mean	Median	Std Dev.
GDDP from agriculture and allied activities (₹ million)	299	21963.7	18633.0	12197.4
Total agricultural credit from SCBs outstanding (₹ million)	299	9603.9	5105.0	12624.5
Direct agricultural credit from SCBs outstanding (₹ million)	299	8090.3	4493.0	10461.7
Rainfall (mm)	299	102.6	102.4	22.8
Irrigation Index	286	47.0	45.5	20.0
Fertiliser consumption (000 tonne)	286	112.4	103.3	63.8
Gross cropped area (lakh hectare)	286	6.0	5.7	2.1
Wagebill (₹ million)	286	7409.4	6060.5	4302.4

Source: Authors’ work and calculations.

Notes

References

Ammani

(2012). An investigation into the relationship between agricultural production and formal credit supply in Nigeria. International Journal of Agriculture and Forestry, 2(1), 46–52.

Arellano

, & Bond

(1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies, 58(2), 277–97.

Armas

E.B.

, Osorio

C.G.

, & Moreno-Dodson

(2010, April). Agriculture public spending and growth: The example of Indonesia. Economic Premise No. 9. World Bank, Washington, DC.

Bhalla

G.S.

, & Singh

(2010). Growth of Indian agriculture: A district-level study. New Delhi: Centre for the Study of Regional Development, Jawaharlal Nehru University.

Binswanger

, & Khandker

(1992). The impact of formal finance on the rural economy of India. Policy Research Working Paper, WPS 949. World Bank, Washington, DC.

Carter

M.R.

, & Weibe

K.D.

(1990). Access to capital and its impact on agrarian structure and productivity in Kenya. American Journal of Agricultural Economics, 72(5), 1146–50.

Chaudhary

(2012). Trends in total factor productivity in Indian agriculture: State-level evidence using non-parametric sequential Malmquist index. Working Paper No. 215. Centre for Development Economics, Delhi School of Economics, New Delhi.

Chavan

(2013). Credit and capital formation in agriculture: A growing disconnect. Social Scientist, 41(6), 59–68.

Das

, Senapati

, & John

(2009). Impact of agricultural credit on agricultural production: An empirical analysis in India. RBI Occasional Papers, 30(2), 75–107.

10.

Dehadrai

P.V.

(2008). Irrigation in India. FAO Corporate Document Repository. Retrieved 24 September 2015, from http://www.fao.org/docrep/007/y5082e/y5082e08.htm

11.

Demirgüç-Kunt

, & Levine

(Eds). (2004). Financial structure and economic growth: A cross-country comparison of banks, markets and development. Cambridge, MA: MIT Press.

12.

Deokar

B.K.

, & Shetty

S.L.

(2014). Growth in Indian agriculture. Economic and Political Weekly, 49(26–27), 101–4.

13.

EPWRF. (2014, April). Agricultural credit in India: Trends, regional spreads and database issues. Mumbai: NABARD.

14.

Government of India. (2007). Committee on agricultural indebtedness (Chairman: R. Radhakrishna). New Delhi: Ministry of Finance.

15.

Johnston

R.B.

, & Pazarbasioglu

(1995). Linkages between financial variables, financial sector reform, economic growth and efficiency. IMF Working Paper, WP/95/103. International Monetary Fund, Washington, DC.

16.

Khandker

S.R.

& Faruquee

R.R.

(2003). The impact of farm credit in Pakistan. Agricultural Economics, 28(3), 197–213.

17.

Kamara

(2004). The impact of market access on input use and agricultural productivity: Evidence from Machakos district, Kenya. Agrekon, 43(2), 202–16.

18.

Kumar

, Mittal

, & Hossain

(2008). Agricultural growth accounting and total factor productivity in South Asia: A review and policy implications. Agricultural Economics Research Review, 21(July–December), 145–72.

19.

Levine

, Loayza

, & Beck

(2000). Financial intermediation and growth: Causality and causes. Journal of Monetary Economics, 46(1), 31–77.

20.

Levinsohn

, & Petrin

(2003). Estimating production functions using inputs to control for unobservables. Review of Economic Studies, 70(2), 317–341.

21.

Ministry of Agriculture. (2007). Agricultural credit (Annual Report: 2006–07). New Delhi: Ministry of Agriculture, Government of India.

22.

Ministry of Agriculture. (2014). Land-use statistics at a glance—state wise. New Delhi: Ministry of Agriculture, Government of India.

23.

Nagaraj

(2008). Farmers’ suicides in India: Magnitudes, trends and spatial patterns. Retrieved 28 September 2015, from http://www.macroscan.org/anl/mar08/pdf/farmers_suicides.pdf

24.

Narayanan

(2014). The productivity of agricultural credit-assessing the recent role of institutional credit to agriculture in India using state-level data. Mumbai: NABARD.

25.

Olley

& Pakes

(1996). The dynamics of productivity in the telecommunications equipment industry. Econometrica, 64(6), 1263–98.

26.

Prasanna

(2014). Impact of monsoon rainfall on the total foodgrains yield over India. Journal of Earth System Science, 123(5), 1129–45.

27.

Ramakumar

, & Chavan

(2007). Revival in agricultural credit in the 2000s: An explanation. Economic and Political Weekly, 42(52), 57–64.

28.

Ramakumar

, & Chavan

(2014). Agricultural credit in India in the 2000s: Dissecting the revival. Review of Agrarian Studies, 4(1), 50–79.

29.

Ramanamurthy

R.V.

, & Misra

(2012). Pricing of paddy: A case study of Andhra Pradesh. Development Research Group Study No. 38. Reserve Bank of India, Mumbai.

30.

Reserve Bank of India (RBI). (1956). All-India rural credit survey—The survey report (Part II: Credit agencies), Volume I. Mumbai: RBI.

31.

Reserve Bank of India (RBI). (1991, November). Report of the committee on the financial system (Chairman: M. Narasimham). Mumbai: RBI.

32.

Sharma

V.P.

(2011). India’s agricultural development under the new economic regime: Policy perspective and strategy for the 12th Five-Year Plan. Working Paper No. 2011-11-01. Indian Institute of Management, Ahmedabad.

33.

Shrestha

C.M.

(1992). Institutional credit as a catalyst for agricultural sector growth: Evidence from Nepal. Journal of Economic Development, 17(2), 137–44.

34.

Solow

(1957). Technical change and the aggregate production function. Review of Economics and Statistics, 39(3), 312–20.

35.

Sriram

M.S.

(2007). Productivity of rural credit: A review of issues and some recent literature. Working Paper No. 2007-06-01. Indian Institute of Management, Ahmedabad.

36.

Subbarao

(2012). Agricultural credit—Accomplishments and challenges. Speech at NABARD, Mumbai. Retrieved 28 September 2015, from http://www.bis.org/review/r120713a.pdf

Agricultural Credit in India in the 2000s: Growth,Distribution and Linkages with Productivity

Abstract

Keywords

1. Introduction

2. Agricultural Credit and Production/Productivity—Theoretical Motivation and Empirical Concerns

3. Agricultural Credit and Production/Productivity—Review of earlier Studies

4. Major Policy Developments Relating to Agricultural Credit with special Reference to policies in the 2000s

5. Key Stylised Facts about Growth and distribution of Agricultural Credit in the 2000s

5.1 Growth in Agricultural Credit

5.2.1 Distribution by type

Table 1 Average Rates of Growth of Agricultural Credit (% per annum) Period Total Agricultural Credit Direct Agricultural Credit Indirect Agricultural Credit 1990–91 to 1999–2000 11.3 8.6 30.1 1999–2000 to 2009–10 25.5 24.8 27.3 2009–10 to 2013–14 13.5 18.3 –2.3

7.1 Agricultural Credit and Productivity—The State-level Panel

Table 3 Production Function Estimation Explanatory Variable Coefficient p-value No. of Observations LP method Labour 0.042 0.808 147 Capital 0.379 0.001* 147 Robust OLS Labour 0.182 0.000* 174 Capital 0.194 0.000* 174

Footnotes

Acknowledgements

Appendix

Notes

References

Table 1
Average Rates of Growth of Agricultural Credit (% per annum)

Period Total Agricultural Credit Direct Agricultural Credit Indirect Agricultural Credit

1990–91 to 1999–2000 11.3 8.6 30.1

1999–2000 to 2009–10 25.5 24.8 27.3

2009–10 to 2013–14 13.5 18.3 –2.3

Table 3
Production Function Estimation

Explanatory Variable Coefficient p-value No. of Observations

LP method

Labour 0.042 0.808 147

Capital 0.379 0.001* 147

Robust OLS

Labour 0.182 0.000* 174

Capital 0.194 0.000* 174