Three richest states in India are three times as rich as three poorest, which is why we can’t ignore spatial inequality.
In a recent review of James Crabtree’s new book, The Billionaire Raj, also reviewed by me, columnist and part-time member of PM Modi’s Economic Advisory Council, Surjit Bhalla, pays my co-author, political economist and presently data guru for the Indian National Congress, Praveen Chakravarty, the following compliment: “In a much cited paper by Praveen Chakravarty and Vivek Dehejia (‘India’s Curious Case of Economic Divergence’) — Crabtree cites it approvingly as well — the authors contend that unlike other countries, India displays a divergence over time.” Unfortunately, much of the rest of what he says about our argument is tendentious or plain wrong.
Our 2016 briefing paper that Crabtree and he cited, along with our subsequent writings in Economic & Political Weekly, and elsewhere, argue that India has been witnessing a skyrocketing spatial divergence of income per person — in particular, both between major states, and within them. Our empirical work uses standard and publicly available data on state-level income, as well as a novel data set that we developed at the IDFC Institute using nightlights luminosity as the first-order proxy for economic activity. Our headline finding is that the three richest states are three times as rich as the three poorest, clearly making India an outliner amongst major federal economic zones in the world.
Our peer-reviewed research clearly demonstrates the fact of increasing spatial inequality in India, an observation which may or not coincide with increasing inter-personal inequality, in the sense of Thomas Piketty, which seems to be Bhalla’s main bugbear. Unfortunately, in attempting to rebut our research, Bhalla seems to demonstrate a fundamental lack of understanding of our claim, as he tediously invokes the low R-squared (or “goodness of fit”) of our “beta convergence” regressions.
To restate our claim: we argue that the “null hypothesis” of income convergence among major states is not supported by the data. And, paradoxically, the low R-squared in these regressions for India works in our favour and against Bhalla. He needs to prove that convergence is occurring; we just need to show that it is not proven — so states are diverging, at worst, or not converging, at best, neither of which fits the conventional narrative.
What is more, Bhalla more or less entirely ignores our results on “sigma convergence” — which measures the variation in income per person among states at a point in time — which is a population-level statistic that is simply computed from data, to which no regression is attached, and hence no corresponding R-squared is required nor is offered. Given his statistical acumen, this is a puzzling misunderstanding by Bhalla.
That data clearly show state-level inequality taking off after the economic liberalisation of 1991. You do not have to have a degree from an Ivy League university (although Bhalla, Chakravarty and I all do) to see this pattern of rising regional inequality, coinciding with the year of reforms. Bhalla also fails to note that we are careful not to make a causal claim linking liberalisation with rising regional inequality, and merely posit this as one hypothesis which is consistent with the data. The object, in any case, is not to denigrate economic reforms, but to point to their likely downside, which is widening inequality, measured in several different ways.
More generally, the burden of Bhalla’s argument is to pooh-pooh concerns about rising income inequality in India. While there are valid theoretical questions that may be asked about the use of consumption rather than income data to impute Gini coefficients, that critique does not apply to our research on state-level inequality, which uses official state income (not consumption) data and standard statistical techniques (in particular, beta convergence) to document rising inequality. Our results on intra-state inequality use nightlights luminosity data, as mentioned, which is by now accepted in the academic literature as a reasonable proxy for economic activity in the absence of reliable direct observation of the level of regional GDP.
What is more, our finding is broadly in line with previous research, except we get sharper and more precise findings as we focus on the 12 largest (undivided) states of independent India, which account for some three-quarters of population, GDP, and parliamentary seats. It is our view that including all of the states and union territories introduces outliers and noise into the analysis which detracts from the undeniable fact of rising regional inequality.
In terms of political economy, this is as, if not more important, than rising inter-personal inequality. After all, states are political units in our federal system; individuals are not. And ignoring the spatial dimension of inequality and treating India as flat is missing, in our judgement, a crucial dimension, which we believe will come to dominate the political economy of Centre-state relations in the years to come.
Unfortunately, with an apparent predisposition to discount the problem of inequality, Bhalla blithely misses all of this in our research and elsewhere. Perhaps that is why his column series is titled “No proof required”.
Vivek Dehejia is resident senior fellow at the IDFC Institute, Mumbai. Views are personal.