Political Economyof Statistics
A mass of economic statistics is being collected in surveys carried out by a number of official organisations. However, these statistics are often misinterpreted. A case in point is the data on indebtedness.
VIDYA PITRE
“The Indian farmer is born in debt,
lives in debt and dies in debt” was
the focal point of textbooks in economics in the 1960s. After decades of fiveyear plans, the situation, far from improving, has worsened so much that it led to the fall of the Bharatiya Janata Party-led government in the last national election. Even now not a day passes without a report on suicides by farmers.
The policy response to this tragedy has been to make credit available at a concessional interest rate. Increased access to institutional finance is perceived to be the solution. But whether this will solve the problem or push farmers further into a debt trap is the question that remains unanswered. The point was raised by several scholars including V M Dandekar in a hard hitting article titled ‘Limits of Credit and Not Credit Limits’ (EPW, September 25, 1993). He argued that given the prevalent pattern of land distribution, agricultural activity is unviable especially for marginal farmers. The real issue is an agrarian crisis and the debt crisis is only its manifestation. A similar argument is made by Ratna Reddy (EPW, May 13, 2006). Further, the phenomenon seems to be confined to a few regions and crops. So the response has to be differentiated and not uniform.
This is specially so in a country of continental size and diversity. It makes measurement of variables at aggregate levels or assessment of trends not only difficult, but meaningless. This apart even the same event or trend at different places needs to be interpreted differently. So any event has to be viewed in its context. This note seeks to address these issues.
A trend fall in the share of agriculture in the country’s national income, re-presents a certain pattern of development. But this will not be so if this fall is a result of the failure of the monsoon or a natural calamity, or without a commensurate decline in population that depends on agriculture. Similarly, a decline in per capita consumption of cereals in a prosperous state like Punjab and backward state like Bihar will convey different meanings. In the former it may not indicate a rise in undernourishment while in the case of Bihar it would suggest that. The point that emerges is that the relationship between economic variables is valid only within a certain range and is not universal. The same is true for indebtedness. Increases in credit offtake by large farmers and poor farmers per se convey very little unless they are placed in an overall framework of economic analysis.
Against this background, we take up the issue of data collection through sample surveys with special reference to rural credit. Currently, the discussions about the magnitude and trend in rural indebtedness is based on the decennial all-India surveys of rural debt since 1951-52 for the first two decades and all-India debt and investment surveys thereafter. These are sample surveys – stratified multistage sample surveys, the All-India Rural Credit Survey 1951-52 was the landmark being the first comprehensive work in this area. It is not merely a survey but a presentation and analysis of the survey results in the light of working of the various credit agencies including the moneylender in rural India. Tremendous efforts had gone into constructing a systematic analytical framework. Unfortunately it was also the last of its kind. Thereafter there are only decennial surveys on rural debt and investment containing presentation of tables. Up to 1971-72, this was done by RBI, later it was shed in favour of the National Sample Survey Organisation (NSSO).
The data churned out from these surveys provided ready source material for policymakers and researchers. The advent of computers made the processing and trend
Economic and Political Weekly December 30, 2006 fitting possible at a click of the finger. An inter-temporal comparison was resorted to indiscriminately. After release of the NSSO data on rural debt and investment in 2003, a time series analysis of credit disbursement by various agencies, viz, commercial banks, RRBs, cooperative banks and moneylenders from 1951-52 to 2003 became a favourite tool of researchers to explain the process of indebtedness.
In this anxiety and rush to push credit offtake, there was little realisation that these are estimates based on sample surveys. The sample is based on households accounting for less than 0.5 per cent of total households, which is only a few thousands. However, the pattern of development over the past few decades has accentuated differences among subgroups of population making any aggregate analysis meaningless.
The NSSO 2003 survey for example is based on 51,137 households (two stage sampling) data which as pointed above accounts for less than 0.5 per cent of rural households. Such a small sample for India as a whole cannot capture all the diversities and the changes in the structure. As a result all exercises of the time series comparison throw erroneous results. A very sharp fall in the share of moneylenders up to 1991 particularly in the 1971-81 period, is observed, the moneylenders’ share being as an indicator of a return to feudal system. The fall should be examined against the backdrop of the behaviour of other economic variables. The trend of declining importance of the moneylender suggests the onset of capitalist relations entailing fundamental changes in the mode of production. But other indicators such as the growth rate of agriculture which should have registered a spectacular increase remained almost the same during the decade 1971-81. The dependence of the population on agriculture continued to be almost the same except in a few pockets. So any inference based on a set of figures without an analytical framework becomes irrelevant and a futile exercise.
The point we are making is neither new nor innovative. This is a reiteration of the basic premises of scientific research. Here we may refer to the great debate between the Indian Statistical Institute (ISI), Kolkata and Gokhale Institute of Politics and Economics (GIPE), Pune in the 1950s about the form and methods of data collection through a centralised agency like the National Sample Survey (NSS). The correspondence between P C Mahalanobis and D R Gadgil edited with comments from V M Dandekar (published as the
Report on the Poona Schedules of the National Sample Survey) is a classic and should serve as a primary text for research methods. It is not only revealing but also the issues raised are still valid.
The main debate was about the character of organisation as well as the methodology for data collection like the all-purpose National Sample Survey. Gokhale Institute took the stand that instead of setting up an all-purpose centralised institute like the NSS, it is advisable that the respective government agencies should be given the task of data collection pertaining to their field. Internally-generated information in the course of administration should be supplemented by in-depth special studies/ surveys taking village as the unit.
The government agencies had the advantage of possessing special knowledge and experience of the subject and local conditions. D R Gadgil warned against the attempt to subordinate them to, or replace them by, an entirely independent organisation, whose only expertise is in statistical sampling design. The elaborate regional and departmental machinery that a centralised agency would have to gradually build up over decades could be justified only if it was proved in open debate, at various levels of expertise and in various aspects, to have overwhelming advantages. The collection of statistical material, which was divorced from dayto-day activity in this manner, might therefore, not be as meaningful as the normal means of collecting continuing statistical data. This was apart from the fact that such an independent agency would not have the special knowledge of either the subject or local circumstance, which the ordinary administrative machinery would usually possess.
It was resolved through compromise. The NSS was set up. This was guided by the historical necessity of having a few national aggregates for a new-born country. But now after experience of five decades there is need for a thorough re-examination. This is because several limitations of the NSS have glaringly come to the fore, especially estimates of consumer expenditure and calorie consumption which have generated absurd poverty estimates.
Similarly, the estimates of rural indebtedness thrown up by the NSS are used recklessly by researchers including official committees. The Report of the Expert Group on Investment Credit (Table 3.10), ‘Agricultural Credit in India: Status, Issues and Future Agenda’, RBI Bulletin, November 2004 (Table 2) and draft report of the Internal Group to Examine Issues Relating to Rural Credit and Microfinance (p 1) make observations about the share of borrowing households from different sources. Inter-temporal comparison of the NSS estimates is the base. The methodology adopted thus is not only valid statistically but also meaningless for economic interpretation. This is because an inter-temporal comparison of sample estimates is not permissible. The samples are different. Standard error, which is so vital for tests of significance to examine difference of means, is ignored completely. It amounts to violation of the basic rules of statistical inference.
It is time the Reserve Bank of India (RBI) initiated in-depth village studies of the extent of indebtedness and other related economic variables to supplement the macro level survey by the NSS and data generated internally to present a comprehensive report a la the AIRCS report of 1951-52.
A new paradigm taking note of fungible character of finance capital and both the sides, namely, supply and demand, needs to be formulated. The mismatch between funds supplied and funds required indicates a need for adequate and proper understanding of the concept of a “rural credit market”. This should include all categories, viz, large and small farmers, agricultural labourers, etc, with their varied requirements for credit; for productive as well as non-productive purposes, and those who are served by different credit agencies. The role of credit and its linkages with other inputs of production needs to be studied. Also its role in the total expenses of households would indicate the dependence of households on credit agencies.
In the present situation of a grave agricultural crisis, the RBI with its record of rich experience should take up a study of this kind. A database for 100 villages through careful selection may be created on an annual basis for key economic variables to get a grasp of the dynamics of rural transformation. This is where the role of micro level studies becomes vital.
EPW[This article was written before the author, an occasional contributor to the EPW, passed away in September 2006.]
Economic and Political Weekly December 30, 2006
