Economic growth and welfare state: a debate of econometrics

This study uses a two-way fixed effect model for panel data of all OECD nations, which includes most of the determinants of growth in previous empirical growth studies for either cross section or panel data as control variables and carefully checks possible endogeneity of the key variables of interest: welfare measures by Durbin-Wu-Hausman test. The empirical analysis shows a robust negative correlation between welfare spending rate, tax-to-GDP ratio and GDP growth. In particular, the estimates suggest that a 1% increase in welfare spending as percentage of GDP would increase the per capita GDP growth rate by 0.19%. Among three biggest components of welfare expenditure, pension spending is identified as the most important source of detrimental effect on growth while income support and public health expenditure are found to have no significant impact on growth. I also find that a 1% increase in tax revenue-to-GDP ratio would increase the per capita GDP growth rate by 0.18%. Since this estimate is close to that of welfare spending rate and welfare spending is only part of tax revenue used by government, it implies that decreasing welfare expenditure is more important and more effective for promoting growth than cutting tax. All results are subject to robustness checks including unit root test for panel data, slope poolability test, dependent variable persistence test, informal check of IV exogeneity and serial correlation test.