Published on Development Impact

Kuznets Waves and the Great Epistemological Challenge to Inequality Analysis

This page in:
A couple weeks ago I was fortunate to serve as a discussant at one (of the many) launch events for Branko Milanovic’s latest book: Global Inequality: A new approach for the age of globalization. The book is hugely thought-provoking, and a pleasure to read. Along with many people in the audience, we had a great conversation. Over lunch afterwards, Branko urged me to put my thoughts into a blog – so here they are!

Branko is best known for his painstaking work on measuring global inequality, adding the component between nations to the inequality among citizens within the world’s many countries. And this book does contain the latest update on that, describing the trends up to 2011: global inequality continues to decline, driven primarily by the ongoing process of inter-country convergence, as the Asian giants (and occasionally some others) catch up to the developed countries. Inequality within countries is rising (on average), but too weakly to offset the falling between-country component. As this process unfolds, the composition of the global middle class is changing: some or the largest income gains of the last decades have accrued to workers and entrepreneurs in China, India and other parts of Asia, just as incomes have stagnated for many in the middle strata in Western Europe and the United States.

Yet, somewhat surprisingly, the most ambitious contribution of Branko’s new book does not lie in the international domain. Instead, it is his attempt to resolve the tension between the views of two of his intellectual heroes, regarding long-term inequality dynamics within countries. An attempt, in other words, to reconcile Simon Kuznets’s “traditional” inverted-U curve – which postulated that inequality tended to rise during the process of economic development, as resources flowed from a low-productivity, low-inequality sector to another where both of those things were higher, and then fell when the advanced sector became dominant and between-sector inequality fell - and Thomas Piketty’s recent empirical characterization of the 20th Century as a period where inequality in advanced economies basically described a U curve – falling through World War II and its aftermath, and rising since the 1980s, or thereabouts.  

Piketty does pick on Kuznets quite a bit in his own book, Capital in the Twenty-First Century, which Branko glowingly reviewed here. Yet, Branko is not quite willing to throw the baby out with the bathwater, and part with all that Kuznets had built into his grand theory of inequality dynamics. So he suggests that the two previously identified patterns, the inverted U and the U, can be reconciled through the notion of “Kuznets Waves”. In the long run, the story goes, most countries are characterized by cycles of rising and falling inequality that respond to three dominant forces: technology, openness and politics (TOP).

My job as a discussant at the book launch – and in this blog – is to expose this ambitious new “grand theory” to some critical scrutiny, i.e., to give Branko a little bit of a hard time… And I’d like to do this on two grounds: First, I was not convinced that the book provides a stable theory of Kuznets waves; that is: a convincing account of the mechanisms which would regularly drive inequality up and then inherently down, before some internal dynamics would push it up again. I emphasize the adverbs above because such regular, inherent mechanisms are what would make this a new theory of inequality dynamics. In their absence, one is left with an empirical description along the lines of “inequality responds to many forces, so sometimes it rises (when the divergent forces dominate), and at other times it falls (when the convergent forces are stronger)”.

Now, although this latter “empiricist description” approach may sound much less grandiose than some overarching, all-encompassing theory, capable of explaining broad vistas of human history in terms of some key “irreconcilable antagonisms”, or of a single little inequality (“r > g”),  it is not clear to me that it is less correct. Of course it is appealing to build a grand, complete theory of distributional dynamics, as Marx did more than a century ago, and as some feel that Piketty has done again. But the fact remains that inequality is necessarily a general equilibrium outcome, reflecting a plethora of very different factors: the quantity of schooling in the labor force, the changing nature of the demand for skills (affected in turn both by technical change and by trading patterns), changes in labor market institutions (such as unionization rates, or minimum wage policies), changes in tax and expenditure regimes, changes in marriage patterns, changes in returns to capital, changes in land prices, etc. The list is almost endless.

At times, one senses that tension in Branko’s writing: as a primarily empirical economist, he is alive to the multifaceted nature of the forces affecting domestic inequality, and often describes the myriad ways in which they can interact.  But against that grain is juxtaposed a budding “grand theorist”, seeking to tell an “over-arching” story of secular “Kuznets” cycles or waves. One senses the battle has not been resolved yet…

The second ground on which one can criticize the account of Kuznets waves in this book is perhaps (even) less fair – in the sense that it really isn’t a specific criticism of Branko, but more of a reflection on the epistemological challenge of causal identification in the domain of distributional change. Essentially, the point is that Branko’s account of how changes in technology, for example, are linked to changes in the labor income share; or how the entry of massive numbers of Chinese workers into manufacturing over the last two or three decades is associated with stagnating manufacturing wages in the West, consists largely of well-informed speculation. 

There is, of course, a vast scholarly literature in labor and public economics that examines each of these mechanisms in turn, with varying degrees of success in terms of causal identification. But whenever economists have tried to quantify the contribution of a set of different forces to an observed distributional outcome, we have more or less inevitably fallen back on various kinds of decomposition analysis, relying on counterfactual income distributions.  (For the truly tortured nerds, my own struggle with how much causality can be read into different versions of these exercises is summarized here.)

In this fascinating book, Branko has largely avoided the frustrations and complexities of this epistemological struggle with the attribution of cause and effect to different forces driving changes in inequality. But in doing so, he has inevitably left a big gap between the empirical description of what happened and the theoretical hypotheses of why it might have happened. From the perspective of understanding mechanisms, there are probably more questions than answers here.


Francisco Ferreira

Former Acting Director for Development Policy in the World Bank’s Development Research Group

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000