Poverty and Inequality
POVERTY AND INEQUALITY
POVERTY AND INEQUALITY An attempt to examine the intertwined issues of poverty, inequality, and growth in India in the 1990s involves an exploration of all three topics. The conventional view of what happened in the Indian economy in the 1980s and the 1990s is that economic growth averaged around 5.5 percent per annum; population growth was around 2 percent, so per capita growth averaged 3.5 percent. This rate of growth was a major acceleration from the past, and both the agricultural and the nonagricultural sectors of the economy shared in it. Agriculture grew at a robust 3.7 percent per annum, and nonagriculture grew at 6.6 percent. Rural India, where most of the poor reside, benefited enormously, and absolute poverty declined at a fast pace in the 1980s. The head-count ratio of poverty (the proportion of people below the poverty line, compared to the total population) declined from 45 percent in 1983 to 37 percent in 1993–1994. This decline is remarkable because both in the 1950s and in the 1960s, the poverty ratio had, in some year or another, hovered around 45 percent. (For the same poverty line, poverty in 1951 was 45.3, and in 1961, it was 46.5 percent.)
In the early 1990s, the Indian economy ran into a balance of payments crisis, a situation that forced India to return for help to the World Bank and the International Monetary Fund. The government of India then instituted major economic reforms from 1991 to 1993. The reforms were primarily oriented toward industry and external trade (the rupee was devalued by 20 percent in two quick steps, over two days, and tariffs were reduced from outright bans and astronomical levels to open imports, and lower levels).
These reforms led to an acceleration of growth for a few years (particularly from 1994 to 1997, when growth of the gross domestic product [GDP] averaged over 7 percent). However, except for these years, GDP growth for the eleven years following the reforms (1992 to 2002) was exactly the same as ten years prior (5.7% per annum) to the reforms. Some have argued that the reforms were oriented toward the rich, urban sectors of the economy, that these sectors benefited disproportionately, and income inequality worsened.
According to this interpretation, four important pieces of evidence were cited. First, the rate of agricultural growth collapsed in the 1990s—from 3.7 percent per annum to only 2.6 percent. Second, and perhaps even reflecting this slowdown, the rate of growth of real wages in agriculture collapsed to less than half the robust rate of the 1980s—from about 5.0 to 5.5 percent per annum to only 2.5 percent. For considerations of poverty, this is significant, since agricultural households constitute the poorest of the poor. Third, the gap between urban per capita expenditure and rural per capita expenditure increased significantly. Fourth, the poorest states lagged behind the growth of richer states. The surveys of the National Sample Surveys Organization (NSSO) also showed that the rate of poverty decline was definitely slower in the 1990s. The critics of reforms further argue that in the 1980s the head-count ratio of poverty declined by eight percentage points, while in the 1990s poverty declined by only five percentage points. This slower decline in poverty is all the more shocking, according to the critics, because it occurred with considerably faster economic growth.
The critics of the economic reforms further argued that the four factors mentioned above conclusively demonstrate that prior to 1990 India performed better than China, where inequality worsened by more than 30 percent in the short span of six years. India, in contrast, at least until 1993–1994, had witnessed generally declining inequality. Even more importantly, the skewed nature of growth in the 1990s represented a period of (relatively) jobless growth. Organized sector employment, after registering an annual growth of 1.6 percent in the 1980s, barely grew in the 1990s (0.4 percent per annum).
Some of the above conclusions are controversial. For instance, the estimates of poverty and inequality by Angus Deaton and Jean Drèze (2002) are biased downward if a more acceptable real measure of equality is used. Changes in inequality are more accurately measured by changes in inequality measured at constant prices. Furthermore, the trends in real consumption cited above are contradicted by the trends in growth in real wages, which almost doubled between the 1980s and 1990s. Whether taken separately or together, the evidence provided by the increase in real wages of the poor and/or the change in real inequality, is indicative of a very minor change in inequality, rather than the "pervasive increase" postulated by Deaton and Drèze.
Growth and Inequality
Poverty and inequality are generally ideological and emotive issues. Often, the discussion assumes that changes in the level of poverty are independent from economic growth and inequality. As for the latter, it is not overall inequality, but rather inequality around the poverty line and changes therein that are most important for the determination of poverty decline.
Part of the reason for varying estimates of poverty decline in India in the since the 1970s is the fact that there are several different estimates of what happened to growth. According to national accounts data, the estimate used by most analysts and policy makers, income (GDP) growth has averaged about 5.6 percent per annum for the last twenty-four years. Per capita growth, the one relevant for poverty calculations, has accelerated to 4.5 percent per annum in 2004 from 3 percent per annum in the early 1980s. India's growth performance for the twenty-two-year period from 1980 to 2002 was bettered by only a handful of countries, led by China and Korea.
In cumulative terms, per capita income growth, if translated into consumption growth (i.e., no increase in the rate of savings), would result in a per capita consumption level in 1999–2000 that would be about 80 percent higher than that which prevailed in the early 1980s. However, there has been an increase in the savings rate, so the appropriate measure of growth is that revealed by the growth in per capita consumption. This estimate of growth is an average of 2.4 percent per annum, or a per capita consumption level in 1999–2000 that is 48 percent higher than the 1983 level. With no inequality change, this would mean that the head-count ratio of poverty in India was in the low teens in 1999–2000, or about half the level indicated by official data.
Poverty, however, is not calculated on the basis of growth in national accounts data, but rather on the basis of growth as revealed by consumption as reported in household surveys. And the growth as revealed by household surveys has been considerably less than the growth revealed by national accounts. The most widely used forms of survey data, and indeed the official source for measurement of poverty, are the consumer expenditure (CE) surveys conducted by the NSSO, which conducted the first household surveys as far back as 1950–1951. The NSSO large-sample surveys indicate a ten-and-a-half-year growth rate from 1983 to 1993–1994 (hereafter the 1980s) of only 1.2 percent per annum, followed by the mildest of accelerations to 1.3 percent per annum growth rate in the six-year period from 1993–1994 to 1999–2000 (hereafter the 1990s). Accordingly, for the entire period from 1983 to 1999, the average growth rate stands at only 1.25 percent per annum. With this growth rate, the cumulative increase in per capita consumption is only 23 percent, and it is this estimate of growth that results in a poverty decline from 43 percent in 1983 to 26 percent in 1999–2000.
This is a gross relationship and one which incorporates all inequality changes that may have taken place. Over this sixteen-and-a-half-year period, inequality thus diminished in India, as indicated by both the Gini (marginal change) and the share in total expenditures of the poor (bottom 40 percent). Thus, what is causing a huge divergence in the poverty estimates for India are not measures of changes in inequality, but rather the widely diverging estimates of growth in per capita consumption, as indicated by national accounts and household survey data.
It is difficult to identify which of the two estimates of growth is more accurate, though this divergence is a common phenomenon, observed in most parts of the world. For India, however, there is another NSSO survey–based estimate of growth for the period 1983 to 2000. For 1983 and 1993–1994, the NSSO surveyed the same households in each year, as canvassed by the CE survey. In 1999–2000, the same sampling frame was used. These NSSO Employment and Unemployment (E&U) surveys also recorded, in great detail, the employment in different activities and the wages of each member of the household. Nonwage income was not recorded. Since the poor gain their incomes mostly from labor, the E&U survey data can be used to estimate the income growth of poor households. And this income growth is considerably higher than the consumption growth estimated for the same households (1983 and 1993–1994).
Rather than the 1.25 percent average for sixteen and a half years, per capita income growth averaged almost three times higher, at 3.2 percent per annum. But this is for all wage earners, including rich, educated urbanites. For rural India, where a large proportion of the poor reside, the average growth rate was only marginally lower, at 3.1 percent per annum. For the poorest of the poor, the agricultural labor households, the average growth rate was still 2.5 percent per annum. Thus, consumption growth of such households was at least twice the rate indicated by consumer expenditure surveys.
So we have three estimates of consumption and income growth in India: first, per capita consumption growth, national accounts, at 2.4 percent per annum; second, per capita consumption growth, NSSO CE survey, 1.25 percent per annum; and third, growth in incomes of the poorest of the poor, the agricultural labor households, 2.5 percent per annum. All other sources of income growth (e.g., surveys conducted by other survey organizations like the National Council of Applied Economic Research) are closer to the NSSO income growth estimate. There is no evidence, direct or indirect, that supports the NSSO CE low growth estimate of only 1.25 percent. A safe conclusion is that consumption growth of the poorest in India was at least 2.5 percent. Given the fact that average consumption growth was also at this level, consumption inequality could not have changed much for these years.
The question of which consumption growth estimate (NSSO consumer expenditure data supported by NSSO wage income data, or non-NSSO survey data) is correct has an obvious and large effect on conclusions about poverty levels and poverty decline. An overwhelming amount of recent poverty literature in India has been devoted to establishing the percentage by which the NSSO CE survey of 1999–2000 overstated the cumulative six-year growth between 1993–1994 and 1999–2000. The "error" being talked about is only 1 to 2 percent, and this will have a very small effect on aggregate growth. However, the different cumulative NSSO survey estimates of growth (consumption and income based) diverge by about ten to twenty times this amount. Thus, the significant "growth gap" between per worker wage growth and per capita consumption growth during both the 1980s and 1990s dwarfs any calculations of the over-estimate or underestimate of mean per capita expenditures in 1999–2000.
It was generally believed that inequality worsened significantly in the 1990s, but to accurately assess trends in inequality, it is more appropriate to consider changes in real inequality (i.e., per capita expenditures deflated by an appropriate price index) rather than changes in nominal inequality (expenditures not deflated by any price index). Unfortunately, the latter has been used by those economists who contend that inequality worsened significantly.
If inequality change is estimated separately for the 1980s and the 1990s, the following results emerge: inequality either stayed flat, or its level improved slightly between 1983 and 1993–1994, both in absolute terms and in comparative terms (with other countries). Since overall inequality improved slightly between 1983 and the terminal year, 1999–2000, if inequality did worsen in the 1990s, it could not have worsened by a significant amount. But this result is contrary to the result of either a "pervasive" increase in inequality reached by Deaton and Drèze, or an increase comparable to that of China (Sen and Himanshu, 2004) or a trend increase in inequality (Ravallion, 2000). How is it possible that the official data yield such little change, yet experts document large changes?
This divergence is possible because in 1999–2000, the NSSO survey authorities changed the definition of per capita consumption from that which prevailed in 1993–1994 (and earlier) concerning different recall periods and different consumption items. Official data inequality levels were reported under the assumption that there was no bias concerning food expenditures and that the 1999–2000 consumption data are comparable to the 1993–1994 data. Given that the raw official estimate of inequality was biased upward, the measured Gini increase of 6 percent can be considered a likely upper bound to inequality change in the 1990s.
This increase was for nominal per capita expenditures. For real per capita expenditures (nominal data deflated by the Planning Commission poverty line deflator) the increase in the Gini was from .28 to 0.29, only a 3 percent increase.
Pooling all the results, a fair conclusion appears to be a mild V-shaped pattern of inequality change in India, meaning that inequality improved by about 2 to 8 percent in the 1980s, and deteriorated between 0 and 10 percent in the 1990s. For consistent definitions and estimation methods, a mild improvement was observed between 1983 and 1999–2000. Whatever worsening in inequality which occurred in the 1990s was small, mild, and less than the small improvement in the 1980s.
Poverty in India, 1983 to 1999–2000
The likely estimate of growth in the 1980s and 1990s was somewhere between 23 and 51 percent. The former is the estimate of per capita expenditures, NSSO surveys, and the latter is a lower-bound estimate of income growth of landless, agricultural laborers (also NSSO surveys). Inequality change, as documented in the previous section, is shown to have either stayed the same, or most likely improved slightly, since 1983. The juxtaposition of these two "facts" indicates that what matters for a realistic poverty decline estimate for India is a judgment of what happened to growth in India, and not what happened to inequality.
What level of poverty in 1999–2000 is suggested by growth in rural wages from 1983 to 1999–2000? If the distribution of consumption did not change between 1983 and 1999 (and all indications are that it did not worsen), then the 1983 distribution, along with growth in the wages of the poorest agricultural workers, can be used to estimate the poverty level in 1999–2000. This method allocates consumption growth to each household, the income growth of the poorest that is, a conservative, upper-bound estimate of poverty in 1999–2000. Thus, if real incomes are assumed to grow at 2.5 percent annum real for sixteen and a half years, the average real consumption in 1999–2000 would be 51 percent higher than that observed in 1983. Recall that 23 percent growth resulted in a decline of 17 percentage points in the head-count ratio (official data). So 51 percent growth would result in a reduction in poverty of approximately 38 percentage points, or a 5 percent level of the head-count ratio (HCR) in 1999–2000.
However, it is the case that often, and especially when the HCR is in the low teens or lower, more and more growth is needed to bring about the same decline in poverty. Calculations suggest that 51 percent growth would have reduced poverty by 30 and not 38 percentage points, that is, the level of poverty in India in 1999–2000 was close to 13 percent, rather than the 26 percent stated by the official data, or even higher estimates by some authors.
The 1990s was the beginning of several transitions in the Indian economy—a decline in fertility rates, population growth, labor force growth, and potential labor force growth. These transitions have not been emphasized by authors who emphasize that the 1990s were characterized by slow growth in jobs, or "jobless growth." The "shortfall" in jobs is mostly explained by examination of a little emphasized statistic: the potential labor force (people between the ages of 15 and 59). This grew by only 1.9 percent per annum in the 1990s, a steep decline from the 2.7 percent growth rate of the 1980s. What this simple statistic implies is that the much-hyped jobless growth was much more about the supply of labor bottlenecks in the 1990s rather than about jobless growth. This is the case because total output growth (GDP growth) stayed the same or was higher in the 1990s. With a declining rate of growth of the labor force, labor tightening was to be expected, even with the same rate of growth, let alone with sharply accelerating growth. Rural wages grew at 2.3 percent in the 1980s, somewhat ahead of urban wages, which increased at 1.4 percent. In the 1990s there was a significant across-the-board acceleration, a doubling in average growth rates. As a result, wage growth in rural areas jumped to a rate of 4.3 percent per annum, and in urban areas to 4.8 percent.
Bhalla, Surjit S. "Growth and Poverty in India: Myth and Reality." In Development, Poverty and Fiscal Policy: Decentralization of Institutions, edited by Govinda Rao. New Delhi and New York: Oxford University Press, 2002.
——. "Recounting the Poor: 1983–99." Economic and Political Weekly 38, no. 4 (25–31 January 2003): 338–349.
——. "Poor Results and Poorer Policy: A Comparative Analysis of Estimates of Global Inequality and Poverty." CESifo Economic Studies 50 (2004): 85–132.
Datt, Gaurav, and Martin Ravallion. "Is India's Economic Growth Leaving the Poor Behind?" Journal of Economic Perspectives 16, no. 3 (Summer 2002): 89–108.
Deaton, Angus, and Jean Drèze. "Poverty and Inequality in India: A Re-examination." Economic and Political Weekly 37, no. 36 (7 September 2002): 3729–3748.
Kakwani, Nanak. "Poverty and Economics Growth with Application to Côte d' Ivoire." Review of Income and Wealth 39 (1993): 121–139.
Ravallion, Martin. "Should Poverty Measures Be Anchored to the National Accounts." Economic and Political Weekly (26 August–2 September 2000): 3245–3252.
Sen, Abhijit, and Himanshu. "Poverty and Inequality in India: I." Economic and Political Weekly 39, no. 38 (18 September 2004): 4247–4263.
——. "Poverty and Inequality in India II: Widening Disparities during the 1990s." Economic and Political Weekly 39, no. 39 (25 September 2004): 4361–4375.
Sundaram, K. "Employment-Unemployment Situation in the Nineties: Some Results from NSS 55th Round Survey." Economic and Political Weekly (17 March 2001): 931–940.
——. "Employment and Poverty in the 1990s: Further Results from NSS 55th Round Employment-Unemployment Survey 1990–2000." Economic and Political Weekly (11–17 August 2001): 3039–3049.
Sundaram, K., and D. Tendulkar Suresh. "Poverty in India in the 1900s: Revised Results for All-India and Fifteen Major States for 1993–94." Economic and Political Weekly (15–22 January 2003).
——. "Poverty Has Declined in the 1990s: A Resolution of Comparability Problems in NSS Consumer Expenditure Data." Economic and Political Weekly (25–31 January 2003).
——. "Poverty in India in the 1990s: An Analysis of Changes in Fifteen Major States." Economic and Political Weekly (5–11 April 2003).