India has grown at an average rate of 5.7 % for the last 20 years. In per-capita terms, income growth has accelerated from the Hindu rate of less than 1 % prior to the eighties, to 3.7 % per annum since 1980-81. These National Account estimates have been provided by the Central Statistical Organization and are accepted as the approximate 1 truth by all government bodies, domestic and international, non-government bodies and researchers and analysts around the world.
However, the acceptance is considerably different on the flip side of growth – poverty levels, and changes, in levels of poverty. The definition is itself a complicated issue, and not the subject of this paper. The accepted, and official, definition of poverty in India is consumption of goods worth Rs. 49 (per person per month) in the rural areas and Rs. 57 in the urban areas, 1973-74 prices. The government of India conducts, via the National Sample Survey (NSS), annual surveys of the consumption expenditure of Indian households. Every five-years, a five times larger sample size (approximately 125,000 households) is used. In addition to the NSS, the National Council of Applied Economic Research runs occasional surveys of consumption and income on Indian households.
Thus, there is no dearth of data on the poverty and income levels of Indian households. Yet there is a raging controversy about whether poverty levels in India have increased in the nineties, a period co-incident with the ushering in of major economic reforms. India, containing the largest amount of poor in the world, is an obvious test case for whether economic liberalization works, whether reforms have a human face, etc. Hence, the debate on Indian reforms, and Indian poverty levels, is not just academic.
The ideological stakes are high, and India is the centerpiece, the jewel, in the assertion that poverty in the world has gone up in the nineties. (World Bank (1999). This major conclusion is examined in detail in this paper, while a companion paper, Bhalla(2000b), explores the relationship between research and ideology. This nexus seems to be particularly strong in the study of data on income distribution and poverty.
The primary goal of major lending institutions like the World Bank (and now it appears, even the International Monetary Fund 2) is poverty reduction. Research results, and policy prescriptions, therefore, may be endogenous to the goals of international organizations. The same may be equally true about the different states in India, who receive subsidies on the basis of how many poor they have – and therefore have an interest in stating higher than actual poverty figures.
The debate is about what has happened to poverty during a period of unprecedented growth in the Indian economy. The highly influential NSS reports, with the unofficial official stamp of government of India (the NSS surveys are conducted under the direction of the Planning Commission), state that absolute poverty in India, declined from 54 percent in 1973-74 to 43 percent in 1983-84, and declined further to 36 percent in 1993-94. (Table 1) However, for 1998, the last year for which data are available, poverty is reported (estimated) to have increased by about 6 percentage points to around 42 percent. This in spite of per-capita consumption and income having grown by about 4.5 percent per annum or about 20 percent during the 4 year intervening period. Thus, the NSS data for the last decade suggests that there is an inverse relationship between income growth and poverty reduction – as growth accelerates upward, poverty reduction hurtles downward.
The NSS “findings” assume significance because post reforms (initiated in 1991), the country has witnessed accelerated economic growth. From 3% growth in the seventies, the record on growth improved significantly in the eighties when the economy grew at an average rate of 5.5%. Further momentum was provided by the reforms and the economy grew at more than 6% in the post-reform nineties. What is more relevant for the measurement of poverty point is the growth in per-capita income (and/or consumption). Per capita income growth has gone up from 3.6% in the eighties to 4.2% in the nineties. Thus, collating the national accounts (NA) and NSS data, one is confronted with a bizarre situation of accelerated growth, higher per-capita income and an increase in absolute poverty to levels observed in the early eighties. These results call for an explanation. Datt(1999), who along with Martin Ravallion has studied extensively the phenomenon of NSS poverty in India (World Bank (1997) is mostly based on their joint work) suggests two possibilities – first, an increase in the
household saving rate, and second an under-estimation of the consumption of households at the upper-end of the income distribution. The first explanation does not (empirically) apply – both consumption and income show an increase of about 20 percent during the period 1993-94 to 1998 (and 41 percent increase from 1983-84 when poverty levels were close to those reportedly observed by the NSS in 1998). The veracity of the second explanation is explored in Section 3.
Can the NSS data be believed? Is the NSS messenger wrong ? This is not just an academic issue, since policies, both domestic and international, are based on what happens in India, and what is communicated by the NSS surveys. The NSS started nationwide surveys before most developing countries achieved independence; these surveys have, till date, a monopolistic brand of authority.
From both a research and policy point of view, it is imperative that the NSS data be scrutinized. But what alternative is there than to accept the NSS results? There are the National Accounts data on income and consumption. But these data, can, at best, give indications about trends in the level of consumption; absolute poverty, measured by the Head Count Ratio (HCR) is about the magnitude of the population consuming below a certain absolute level of income. The NA data cannot yield information on HCR’s; only surveys can.
National accounts data yield estimates of mean consumption; the NSS data yield estimates of the distribution of consumption. Together, the two sets of data can yield “reliable” estimates of trends in poverty. Indeed, until 1993, this was precisely the approach taken by the government of India. For reasons that are not entirely clear 3, the government of India set up an Expert Group under the chairmanship of the late eminent scholar, Mr. Lakdavala. In 1993, the Group came out with its report entitled “Report of the Expert Group on Estimation of Proportion and Number of Poor”, (EG-GOI (1993)). One of the most significant policy conclusions contained in this report was the recommendation that national accounts data no longer be used for adjusting mean survey estimates i.e. the National Sample Survey be used, in to, to generate results on poverty trends. One of the major conclusions of this paper is that this single decision has been responsible for the considerable amount of noise, and ideological misinformation, that now exists about poverty in India.
The plan of the paper is as follows. Section 2 documents the trends in poverty in India since 1973. The next section outlines a methodology to assess the relative accuracy, in terms of mean consumption, of the two major sets of data – surveys and national accounts. Section 4 contains alternative estimates of the poor in India. These estimates are based on other surveys (NCAER) and data on important elements of food consumption. Section 5 contains a discussion of ideology and how it impinges on
research. Section 6 concludes.