Financing Welfare States and the Structure of Taxation

This dissertation examines how countries fund welfare states in a context of increasing globalization, structural unemployment and changing demographic and economic structures. The dissertation focuses on two taxes that have seen significant growth over the period of welfare state expansion, and now finance large fractions of the social welfare system in continental Europe and elsewhere: income taxes and social security contributions. As both of these taxes are levied largely on wage income, I show evidence that these two taxes are substitutes. I explain variation in these two taxes using a formal model of insider-outsider politics based on a model of tax and transfer by Moene and Wallerstein (2001). In my argument, labor market insiders, facing little risk of unemployment, prefer social security contributions, and while outsiders with irregular employment patterns prefer income taxes. Insiders prefer social security contributions because they allow them to ring-fence benefits for themselves. Outsiders prefer income taxes because the progressivity and broad base of income taxes allows outsiders' high levels of unemployment to be cross-subsidized by the wealthy and the regularly employed. More broadly, I argue that the tax mix countries choose is part of a broader equilibrium including labor market structures as well as employment outcomes. Echoing the arguments made by Calmfors and Driffill (1988) concerning non-encompassing unions, I argue that rigid labor markets and high social security contribution levels lead to a dualized labor market with high levels of outsider unemployment, which in turn leads powerful insiders to prefer social security contributions that help dualize the labor market in the first place. The argument proceeds as follows: negative employment shocks, combined with nonencompassing unions, leads to these unions to demand job security, which results in unemployment being visited largely on outsiders, who then become part of a `structural unemployment problem'. This chronic underemployment leads to a renewed focus on job security on the part of insiders. This in turn leads to insiders being reluctant to cross-subsidize welfare for the chronically unemployed, and to demands for increased 'contributoryness' of the tax system. The resulting increasing reliance on social security contributions (instead of broad-based income taxes or indirect taxes) raises the marginal cost of labor and contributes to the unemployment problem. This combination of a tax code that reinforces unemployment, and a high level of unemployment that leads to demands for a contribution-financed insurance system, results in a policy equilibrium that can be very hard to reform. By contrast, where unions are encompassing, consequent labor market flexiblity and active labor market policies means that insider-outsider employment cleavages do not result. This means that the pressure on social insurance systems to become more `contribution-financed' remains low, and thus the tax burden on labor remains consistent with high employment. I evaluate the theory in two ways: first, I test the theory qualitatively, by examining the preferences of unions and politicians in Ireland and France since the 1970s. I examine the preferences of French unions in response to reforms to pensions and other benefits in the 1970s and 1980s, as well as the Jospin reforms in the 1990s, focusing particularly on a cleavage between the largest French union confederations. In Ireland, I examine the votes of the constituent unions in the ICTU in response to the five national pay/policy agreements in the from 1987 to 2003, as well as the 13 agreements between 1948 and 1979. I also examine the Irish budgets introduced in those years. Secondly, I evaluate the theory at a cross national-level. The specific hypothesis I test is that trade-union density as well as employment outcomes should impact on tax mixes across country years. I employ panel data from the OECD and IMF for this. Measures of trade union density are taken from the OECD. I run a seemingly unrelated regression analysis correcting for the compositional structure of tax mix data. There are five dependent variables per country-year; the share of revenue raised from Income Taxes, Social Security Contributions, Taxes on Goods and Services, Property Taxes, and Other Taxes. I demonstrate evidence that trade union density is positively associated with increases in social security contributions, and decreases in income taxes in the tax mix