An Assessment of the Causal Relationship between Growth and Inequality in Spanish Regions

Abstract

This paper analyses the causal relationship between economic growth and income inequality in Spanish regions from 1970 to 2000. We examine such a relationship using a panel of data with four time observations on the level variables for each region. Thus, we use a modified form of traditional Granger causality tests to suit the short times series that are available. Applying a sum–difference test, we conclude that the empirical evidence supports the hypothesis that gross domestic product (GDP) per capita growth in Spanish regions leads to less income inequality, rather than any other possible causal relationship.

Acknowledgements

The author thanks Dr Diana Weinhold for her helpful comments during his stay at the London School of Economics and Political Science as Visiting Research Associate. Likewise, the author is also indebted to two anonymous referees for their valuable suggestions on the paper.

Notes

This hypothesis claims that growth and inequality are related in an inverted U-shaped curve: inequality would increase over the initial stages of development as an economy transforms from rural to urban and from agricultural to industrial and, subsequently, inequality would decrease as the labour force in the industrial sector expands and that of the agricultural sector falls.

Although both surveys have different sample sizes, Ayala et al. (2005) show that their results are perfectly comparable, arguing that the data obtained for 2000 through the Continuous Household Budget Survey are fully coherent with those obtained via the Household Budget Surveys for 1973–1974, 1980–1981 and 1990–1991.

There is an overall consensus in the literature on this decreasing trend, though some authors highlight the existence of a certain rise in inequality during some years of the 1990s (Goerlich & Mas, 2004; Prieto & García, 2007).

It would result in a fall in the number of observations from 34 to 17 and the number of degrees of freedom would be very low, especially for models (1) and (3). Therefore, the results with such low number of observations/degrees of freedom would not be very solid.