بیمه بیکاری: نقش سیستم های انتخاباتی و بازارهای کار منطقه ای

Levels of insurance against unemployment vary considerably across countries. Replacement rates, the ratio relating income from unemployment to what people earned when they were employed, are higher in countries with proportional electoral systems than in countries with majoritarian systems. Also, replacement rates are positively correlated with per capita income and negatively correlated with the countries' unemployment rates. I develop an electoral competition model that replicates these stylized facts.

Unemployment insurance systems that provide financial support for people experiencing spells of unemployment have been identified as one of the key suspects underlying high unemployment in OECD countries. See for example Siebert (1997), Nickell (1997), Elmeskov et al. (1998), Nickell and Layard (1999) and Blanchard and Wolfers (2000). In the policy debate, it seems to have been forgotten that labour market institutions were created in order to protect workers and households from the risk of unemployment.1 This is a risk for which markets do not provide adequate insurance, possibly, but not only, for reasons of adverse selection.2 This causality running from market conditions to the creation of labour market institutions is explored in this paper. In particular, we are interested in how the interdependence of regional labour markets and countries' electoral systems shape unemployment insurance. In order to analyze the issue, I set up an electoral competition model in which two office-motivated parties that do not exactly know voters' preferences make proposals for unemployment compensation schemes. I compare the policy choices that emerge from majoritarian and proportional electoral systems and analyze the relationship between replacement rates and employment and income per capita, respectively. The model can replicate all three types of relations that we find in the data: countries with proportional electoral systems have higher replacement rates than countries with a majoritarian electoral system, replacement rates are negatively correlated with countries' unemployment rates, and positively correlated with income per capita. Contrary to the strand of literature that assumes labour market institutions as exogenous, reverse causality in the sense that unemployment benefit systems are a function of unemployment, or agents foreseeing the likelihood of unemployment spells, has rarely been discussed. However, evidence points in the direction of causation that I explore, namely, that unemployment insurance systems are a response to labour market conditions. Baicker et al. (1998) report on the U.S. unemployment compensation system, describing its origins and development from the mid 1930s. They present evidence that a greater degree of seasonality in the states' manufacturing employment shares in the two decades from 1909 onwards is related to higher unemployment insurance benefits in the years 1949 to 1969. Agell (2002) searches for explanatory variables for the cross-country variance in labour market institutions, finding some support for endogenous institutions. The political economy of unemployment insurance systems has been studied by Wright (1986), Saint-Paul (1996), DiTella and MacCulloch (2002) and Hassler et al. (2002), among others. Common to all these papers is a median voter mechanism, which is thought to capture a richer political sphere in which policies are made. One purpose of this paper is to propose a model that adds more structure to the policymaking process, following the claims by Myerson (1995) and recent evidence on the role of the political system on the size and scope of governments (Persson and Tabellini, 1999 and Persson, 2002). Although the evidence on labour market institutions being the product of market conditions cannot be neglected, we do not want to rule out the consequence that institutions, once in place, may result in malperforming markets. We are interested in why these institutions were created in the first place, and especially in the reasons for the variation across countries. The starting point of our argument is that the probability of unemployment varies across regions within countries. Thus, incentives for parties' policy proposals for unemployment insurance schemes may differ with the institutional setting of the electoral system. In a majoritarian system one will have a pivotal district that, if won, guarantees office. Thus, the competing parties will shape their proposals for an unemployment insurance system in such a way that they please the voters in the decisive district. The swing voter, however, may have a different individual risk of unemployment, had the country had a proportional electoral system. If the distribution of voters is such that the swing voter has a higher individual risk of becoming unemployed under a proportional election system, one will find higher replacement rates in countries with proportional electoral systems. The paper proceeds with some stylized facts on the relation among unemployment insurance, electoral systems, per capita income, and unemployment in a cross-country perspective. Section 3 develops the model. Section 4 discusses the results and presents data that supports the outcomes of the model. The last section provides a conclusion.

Using net replacement rates as a proxy for unemployment insurance, cross-country data show that insurance against unemployment is lower in countries with majoritarian electoral systems than in countries that have proportional elections. The data also suggest that replacement rates are positively correlated with per capita income and negatively correlated with unemployment rates. I have set out a model of electoral competition between two parties in which people vote according to their individual risks of becoming unemployed, which differ across regions. The model can explain the fact that countries with majoritarian electoral systems have lower replacement rates when the distribution of voters is skewed towards higher risks of becoming unemployed, for which there is supportive evidence. Furthermore, if higher per capita income is associated with a more unequal distribution of wages, the model also replicates the positive correlation of replacement rates with income per capita. For a sufficiently strong ‘tax effect’, as opposed to an effect that originates from the risk of the decisive voter becoming unemployed, the model also explains the negative correlation between replacement rates and unemployment found in the data.