Re-examining Kuznets Hypothesis: Does Data Matter?

Kuznets Hypothesis has been in debate since Simon Kuznets published his seminal paper “Economic Growth and Income Inequality” hypothesizing that inequality follows an inverted U shaped curve. He suggested that inequality rises as an economy develops, due to urbanization and industrialization, which later is abated as leveling forces gradually reduces inequality. Extensive research has been done in this arena but consensus is yet to be reached. In this paper we test the robustness of Kuznets hypothesis by employing newly available EHII and UTIP manufacturing pay inequality dataset which has been developed by University of Texas Inequality Project. Panel unit root tests are undertaken to develop the parametric equation for testing the hypothesis. We also use various econometric methods (Fixed Effect, dynamic panel analysis, fixed effect with autoregressive term) to analyze the effect of economic model on existence or absence of Kuznets Curve. Data are also segregated to Global dataset and OECD dataset, one including all the countries in the world and the other only OECD countries respectively. The objective is to see if there exists a different inequality dynamics for highly developed economy. The paper found that income variables should be in log formed and not level form while testing the hypothesis, as otherwise they are not stationary. It was also found that gender segregated cohort size has an implication for inequality, with matured male cohort having negative and matured female cohort size having positive relation with inequality In relation to Kuznets hypothesis it was found that inverted U shaped curve appears in case of D&S and WIID2 data but U shape curve appears in case of EHII and UTIP dataset. In case of OECD countries the difference is enhanced. The primary reason rests on rising manufacturing pay inequality which EHII captures. It is hypothesized that inequality follows a zigzag pattern with inequality rising and falling as economy develops and moves from an agrarian to an urbanized industrialized economy. This is in line with Kuznets hypothesis. After that, the economy faces major technological innovations which on onset increase the inequality within the manufacturing sector first but given the industrialized nature of the economy, this translates quickly to overall rise in inequality. Hence it seems that absence or presence of Kuznets curve greatly depends on the usage of inequality dataset. It is suggested that existing EHII dataset may be augmented by basing it on WIID2 instead of D & S and by making estimates which are gross of individual income resulting in data harmonization.