Relationships between household consumption and inequality in the Indian States.

I. INTRODUCTION

It is widely argued that economic growth plays a key role in enabling effective poverty reduction. But how effective growth is in delivering poverty reduction depends critically on one of the age-old development issues: the relationship between growth and distribution. Obviously the distributional pattern of growth has implications for the evolution of inequality, which has direct consequences for the extent of poverty reduction. Equally though there are a number of arguments as to why the extent of inequality in a society will be an important determinant of growth, with many recent studies suggesting that high inequality may have a retarding effect on subsequent growth [Aghion et al, 1999].

Current empirical evidence on the (potentially two-way) relationships between growth and inequality has been based almost entirely on cross-country regression analysis or, more recently, on cross-country panel data sets where the number of observations capturing the variation over time is very small. This is a very serious limitation in that the relationships being investigated are fundamentally dynamic in nature, and need to be considered over a much longer time dimension. In addition, cross-country regressions (even where using short panels) suffer from a number of other serious limitations, including the facts that the relationships may differ substantially across countries, and questions of data comparability may arise, especially inequality data, across countries (notwithstanding the valiant efforts of Deininger and Squire [1996] to try to create a comparable cross country data set).

In this article we circumvent many of these issues by looking at the relationships between inequality and average consumption levels and their growth rates based on Indian state level data collected by the National Sample Survey. Although not available for every year over this period, the currently available series (1960-94) represents probably the longest time series of inequality information relating to developing countries. This long time dimension means that it is possible to consider the time series relationships between growth and inequality in each state separately, and to give serious consideration to dynamic issues, including causality. But also because the same data are collected in each state, it is possible to make meaningful comparisons of the relationships across states. (1) In addition, a major advantage of this data set, in contrast to many cross country studies of growth inequality relationships, is that the measures of growth and inequality are both based on household level standard of living measures. Most cross-country studies (for example, Deininger and Squire [1998]), one of the few studies to consider the potentially two-way relationship between inequality and growth) of this issue instead focus on the relationship between the inequality of household income or consumption levels and growth in GDP per capita, rather than the growth of household income or consumption which is of more direct relevance, especially given a focus on poverty reduction.

The case of India is of particular interest given that the country accounts for a substantial number of the world's poor, and given that disparities in living standards between and within states, including between urban and rural areas, have been a matter of serious concern for policymakers for decades. This article uses this unique data set to examine the interrelationships between average consumption levels and inequality within the different states, distinguishing urban and rural areas and testing for causality (allowing for possible two-way relationships). Both time series and panel data techniques are used to examine the potential two-way relationship between average consumption and inequality (or their changes). The article is structured as follows. Section II briefly summarises the nature of the potential relationships between growth and inequality, including reviewing evidence in the Indian context. Section III then describes the data in more detail and provides a descriptive analysis of state-level trends in inequality and growth, while section IV describes the methodology and the results obtained from the time series analysis. This discussion is then complemented by a panel data analysis of the same issues in section V. Section VI concludes.

II. RELATIONSHIPS BETWEEN GROWTH AND INEQUALITY

There is in fact a relatively limited literature on the relationships, which are the specific focus of this article, that between the average level of household consumption and the inequality of its distribution. Of course, a vast literature has been devoted to the related question of the relationship between overall economic growth (GNP or GDP, often per capita) and inequality. In this literature there are arguments both for changes in average mean income or its growth affecting distribution and for inequality influencing future growth. Theoretical arguments can be advanced for-either positive or negative relationships in each case.