Studies on the relationship between inequality and redistributive spending often rely on cross-national data and find that countries with higher inequality spend more on redistribution (Milanovic 2000, Kenworthy & McCall 2008, Finseraas 2009, Lupu & Pontusson 2011). The median voter model is usually at the heart of the explanations of this positive effect: when the median voter income is lower than the mean income - as is often the case in countries with high inequality - the median voter can benefit from redistribution and signal these preferences to the government. As the median voter demand for redistribution grows, government spending begins to reflect it.

Alternatively, scholars who find a negative relationship between inequality and redistributive spending usually ground their explanations in a social insurance theory of redistributive spending (Moene and Wallerstein 2001, de Mello and Tiogson 2006). It suggests that people with lower incomes are willing to fund redistributive spending through taxes only as much as they are willing to fund purchases of other normal goods. When incomes go down, the consumption of normal goods also goes down, and so does the demand for the social ‘safety net.’ The implication of this theory is that higher inequality will be negatively associated with redistributive spending.

In this paper we offer a test of the median voter and social insurance hypotheses by studying redistributive spending at the subnational level – among fifty largest counties in Texas in years 2006-2012. One of the advantages of the regional sample is that counties are relatively more homogeneous than cross-national or state-level samples. Random and fixed effects models suggest that inequality is positively associated with spending in the current year and that the effect persists when we predict spending on a one-year lagged inequality index. Besides supporting the median voter model at the sub-national level, we find evidence that redistributive spending is negatively associated with inequality in a subsequent year. The robust negative effect suggests that redistributive spending does equalize public well-being. The study improves our understanding of the patterns of redistributive spending at the subnational level and highlights the importance of careful inter-temporal modelling of relationships between redistributive spending and inequality.