Financialization and income inequality: An extreme bounds analysis

Abstract

Several theoretical hypotheses predict conflicting effects of financialization on income inequality, and the lack of consensus is reinforced by the mixed findings of related empirical studies. In this paper we address the model uncertainty issue by applying extreme bounds analysis (EBA) to examine the robustness of financialization variables as determinants of income inequality. The results suggest that financialization is a leading driver of income inequality and that the effects are larger when transmitted through financial markets. In addition, the results show that law and order, labour union density, population, globalization, remittances, education, and the agriculture sector share are also robust determinants of income inequality. Policy recommendations are put forward to reduce the power of the financial sector.

KEYWORDS:

• Financialization
• financial development
• financial structure
• income inequality
• extreme bounds analysis

JEL Classifications:

• C52
• D3
• E44
• O1

Acknowledgement

We are grateful to two anonymous referees for useful comments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 There is no rule to determine the number of Z variables. Following Sala-I-Martin (1997), the number of Z variables is taken to be three.

2 In general, EBA produces (N!/[Z! (N-Z)!]) combinations for each variable of interest, where N is the number of potential explanatory variables available for selection, and Z is a fixed number of explanatory variables selected for each model.

3 Period averages are used because EBA is designed for cross-sectional data as demonstrated by Sala-I-Martin (1997) who examined some 60 determinants of economic growth in a large cross section of countries. For this purpose, he had to run some two million regressions (hence the title of his paper). For the original version of the paper, he actually ran four million regressions (hence the title of the working paper ‘I Just Ran Four Million Regressions’).

4 The Gini index is the percentage form of Gini coefficient, which is equal to the value of Gini coefficient multiplied by 100.

5 Initially created as ‘The World Top Incomes Database (WTID)’. Source: https://wid.world/wid-world/

6 The full results are not reported here but they are available from the corresponding author upon request.