the best growth phase for India was during the period 2003–2008, when the average annual growth rate of real gross domestic product (GDP) was 8.9%. During the period 1991–2008, the corresponding figure was 6.5%. There is considerable debate on the reasons for accelerated growth and when Indian growth really took off, in particular whether it took off in the 1980s or after the reforms.3 There is also a recent controversy on the new estimates put forward by the Central Statistical Office (CSO) which has rendered the official growth figures problematic.4 In my opinion, these debates and controversies do not distract from the broad idea that Indian economy has displayed unprecedented growth by its own historical standards and has been one of the fastest growing economies in the world in recent decades. Also, what is relevant for us is the distributional changes associated with this growth process. We will first focus on changes in the interpersonal distribution.
Many scholars and policymakers have used the surveys on consumption expenditure conducted by the National Sample Survey Organization (NSSO) to understand the Indian interpersonal distribution.5 These are large, nationally representative surveys that are cross-sectional in nature (i.e. not panels) and usually conducted every five years, in various rounds. The relevant rounds for us are: 50th (1993–1994), 55th (1999–2000), 61st (2004–2005), 66th (2009–2010), and 68th (2011–2012).6 The limitations of these surveys are well known and have been discussed in the literature (e.g. Jayadev et al., 2007; Motiram and Vakulabharanam, 2012) viz., the rich are underrepresented and their consumption is undervalued. These limitations are likely to result in an underestimation of the level of inequality and its increase over time. The consumption expenditure survey in 1999–2000 (55th round) differed in its methodology from previous surveys, and therefore many recent studies ignore it.7
Most studies of inequality in India have used relative measures of inequality (e.g. the Relative Gini (RG) or Theil). These measures are unaffected if the consumption expenditure of every one is scaled up or down by the same factor (e.g. doubled or halved). This requirement, referred to as scale invariance, facilitates comparisons. An important insight provided by Kolm (1973a, 1973b) is that while relative measures have the virtue of convenience, they come with their own ethical baggage. He termed these as ‘rightist’ measures and compared them with ‘leftist’ (absolute) measures and illustrated his point by considering the case of strikes by workers in France in the late 1960s. An analogous hypothetical example can be given from the Indian context: suppose the daily wages of workers and managers, which stand at Rs. 100 and 1000, respectively, are doubled. Inequality, as measured by a relative measure remains unchanged, but since the managers are earning much more than the workers now, some could argue that inequality has actually increased.8 Although still sparse, a recent literature has emerged in the Indian context that draws upon these insights and goes beyond relative measures. I will draw upon Motiram (2013) to provide a brief and non-technical description of the relevant concepts and ideas. A more elaborate and technical description is presented in Kolm’s work cited above and Subramanian and Jayaraj (2013). Let ci denote the consumption expenditure of an individual i (= 1, 2, … , N) and μ denote the mean consumption. The RG is as follows:
Note that the RG is unit-less. Absolute inequality measures embody the principle of translation invariance, which requires that they are unaffected if all incomes increase or decrease by the same amount. Examples of absolute inequality measures are the Absolute Gini (AG) and Standard Deviation. The AG is calculated as follows:
For the above hypothetical example, it is quite easy to see that the AG increases when the incomes of the workers and managers double. However, the AG is not unit-less. Can we preserve the convenience of relative inequality measures while avoiding the ethical problem that they embody? Intermediate measures try to achieve this trade-off by incorporating the principle of unit consistency, which requires that inequality ranking between two distributions is unaffected if both are scaled by the same factor, i.e. inequality comparisons do not depend upon the units in which distributions are expressed. Examples of intermediate measures are the Intermediate Gini (IG) and the product of Standard Deviation and Coefficient of Variation. The IG is determined as follows:
Note that IG satisfies unit consistency since RG is unaffected by scaling, and if distributions are scaled by a factor, AG simply gets multiplied by this factor.
In my opinion, it is useful to examine inequality using different measures, and I therefore present various estimates in Tables 1 and 2. As we can observe (from column (1)), RG for nominal consumption expenditure has increased in both rural and urban areas, although the increase in urban areas is more pronounced. Column (2) presents the RG for real consumption expenditure, and the trends are roughly similar. Both AG and IG show increases in rural and urban areas. India is a large country with substantial variation in prices that different households face. Mishra and Ray (2011) take this into account and show that inequality has increased even after doing so.
Table 1:Inequality in Monthly Per-Capita Expenditure (MPCE), rural India.
Notes: RG: Relative Gini, AG: Absolute Gini, IG: Intermediate Gini, W: Wolfson Index.
Table 2:Inequality in Monthly Per-Capita Expenditure (MPCE), urban India.
Notes:
1.Estimates for RG (Nominal) are author’s computations from NSS data.
2.Estimates for RG (Real) are from Dubey and Thorat (2012).
3.Absolute and IG are from Subramanian and Jayaraj (2013). These are for real MPCE computed by using Consumer Price Index for Agricultural Laborers (CPIAL) in rural areas and Consumer Price Index for Industrial Workers (CPIIW) in urban areas.
4.Estimates for Wolfson Index are for nominal MPCE and from Motiram and Sarma (2014).
A growing body of research has emerged recently that argues that we should move beyond traditional notions and measures of inequality, particularly if we want to understand conflict. This literature on ‘polarization’ is discussed in greater detail in Chakravarty (2009) and Motiram and Sarma (2014). One important concept in this literature is ‘bipolarization,’ which is based upon the understanding that a decline in the share of the middle could have negative implications for stability and could accentuate the possibility of conflict. Measures of bipolarization are derived by conceptualizing the middle in terms of the median and replacing the Dalton–Pigou principle used in traditional measures of inequality (e.g. RG) with two other principles: Increasing Spread and Increased Bipolarity. The Dalton–Pigou principle holds that a regressive transfer (from a poorer to a richer person) should increase inequality and a progressive transfer should decrease inequality. The principle of Increasing Spread holds that bipolarization increases under the following circumstances: a rich person becomes richer or a poor person becomes poorer with the median being unaffected; or a transfer occurs from a poor person to a rich person across the middle (median). On the contrary, Increased Bipolarity holds that bipolarization increases if a richer and poorer person on the same side of the median are brought together due to a transfer from the former to the latter — this process is linked to the formation of poles on either side of the median. One popular index that has been proposed in the literature is the Wolfson Index. Let m and L(0.5) denote the median and the share of consumption held by the bottom half of the population, respectively.9 The Wolfson’s index is denoted as follows: