Ethnic Diversity, Local Redistribution and Income Inequality: An Empirical Analysis Applied to Mexican Municipalities

Abstract

This article examines the relationship between ethnic diversity, redistributive policies and income inequality in the context of Mexican municipalities. Although we primarily focus on income inequality within municipalities, we are also interested in how ethnic diversity and redistribution affect income disparities between municipalities. This study uses multiple data sets to construct original measures of mean household income, income inequality, ethnolinguistic diversity and redistributive policies at the municipality level. Based on these variables, we carry out several econometric analyses accounting for endogeneity and spatial autocorrelation issues. First, while there is a clear negative impact of ethnic diversity on municipal mean income, the relevance of the redistribution channel in the mediation of this relationship seems to be partially relevant among Mexican municipalities. Second, our empirical investigations emphasise that the impact of ethnic diversity on intra-municipal income inequality depends on the extent of redistributive policies. More precisely, we show that ethnic diversity has a negative effect on inequality among municipalities with low levels of redistribution and that this inequality-reducing effect weakens along with increasing levels of redistribution until reaching a turning point. Our findings call for further research about the main channels through which ethnic diversity may affect redistribution and income inequality.

Acknowledgements

We would like to thank the two anonymous referees for their informed and helpful comments and suggestions. We are also grateful to Minh Nguyen from the World Bank for providing us information on the implementation of small area estimation with Stata. The data and code used in the study are available upon request.

Disclosure statement

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

Supplementary materials

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2020.1786060

Notes

1. There are three main indices from the class of generalised entropy (GE) indices: the mean log deviation GE(0), the Theil index GE(1) and half the squared coefficient of variation GE(2).

2. The ethnolinguistic fractionalisation index or the generalised variance (GV) index, in its normalised form (normalised generalised variance, NGV), can be expressed as follows (D.V. Budescu & Budescu, 2012): $NGV=\frac{C}{\left(C-1\right)}(1-\sum { }_{i=1}^C{P}_i^2)$ NGV measures ‘the probability that two randomly selected individuals from a particular population belong to different subgroups (…). A high value (probability) reflects a higher degree of diversity’ (D.V. Budescu & Budescu, 2012, p. 217).

3. For instance, in 1997, the creation of Ramo 33 (federal grants for federal entities and municipalities) led to increasing responsibilities for social policy for states and municipalities. Within Ramo 33, the Social Infrastructure Fund (FAIS) is specifically dedicated to address social infrastructure gaps across municipalities.

4. According to municipal public finance data, in average, these transfers only represent 6 per cent of total municipal expenditures in 2015. However, the 90^th percentile is equal to 12 per cent and the maximum value to 51 per cent.

5. This spatial two-step least-squares approach is implemented through the ‘spivreg’ Stata command (Drukker, Prucha, & Raciborski, 2013).