On the political economics of tax reforms: survey and empirical assessment

For a survey of the normative side of the literature, i.e., optimal taxation, see Castanheira et al. (2011). For contributions related to the political economy of optimal taxation, see Acemoglu et al. (2010, 2011) and the empirical analysis of Gruber and Saez (2002).

See Persson and Tabellini (2000, Chap., p. 2) for a review of some sufficiency conditions.

This partly biases the results: a rich individual who retires and benefits from a generous public pension will be perceived as receiving a large transfer. We thus focus on the results that leave pensions out.

Several authors worked on narrowing down the set of policies that could be proposed in a mixed strategy equilibrium (see, for instance, Davis et al. 1972; Schofield 1983; Laver and Shepsle 1996).

See also Calvert (1985), Wittman (1983) and Roemer (2001) for models in which the politicians’ ideology prevents platform convergence.

“Net” means the difference between the taxes paid and the subsidies received.

Interestingly, this means that the entry of a new party, say on the right-end side of the political spectrum, should induce the other parties to propose lower taxes for the rich than before the entry of this new party.

The typical equilibrium of such a game can only be represented by a Lorenz curve, which identifies how many percents p(x) of the population pays less taxes (or benefits from larger transfers) than some level x and how many percents 1−p(x) receives less (or pays more) than x. See Myerson (1993), Lizzeri and Persico (2001, 2005), and Crutzen and Sahuguet (2009) for more detail.

For a more in-depth discussion of these processes, see Castanheira et al. (2010) who accompany their review of the economic literature with an illustration of the political economy of flat tax reform in Estonia.

Calculated as (54−18)/54.

See Roland (2000) for an in-depth analysis that focuses on transition countries and Valenduc (2006) for an application to Belgium.

Note that efficiency gains are a necessary, but not sufficient, condition to satisfying both the ex ante and the ex post constraint.

Rodrik (1996) challenges the latter theory on the ground that it is “unfalsifiable”: if a reform takes place, it is normally meant to address some problem but is it necessarily a “crisis”? Alternatively, if no reform takes places, would it mean that the crisis was not serious enough?

Notice that there is no information for reforms in social security contributions in Bulgaria between 2000 and 2002.

Instead, the political economics of indirect and corporate taxes is likely driven by special interest politics (see, e.g., Grossman and Helpman 2001).

A question is whether these reforms are meant to increase economic efficiency or primarily to target voting groups. In an ideal world, one would solve for the optimal tax system conditional on the socioeconomic and political conditions in each country and check whether a given reform goes in that direction or not. But this is much beyond current feasibility constraints. To proxy for potential efficiency enhancements, we identify a subset of targeted reforms which we call “Lisbon.” These are the reforms targeted on older workers, female workers, job seekers, or low income taxpayers that decrease either their (tax or social contribution) rate or the base. The “Lisbon strategy” (now: Europe2020) specifically targets these categories to boost economic growth, employment, and social progress in Europe. There are 65 reforms of that type.

In an unreported regression, we jointly used time and country fixed effects. Including these effects does not affect the signs and magnitude of our coefficients but reduces their significance: they pick up a large part of the influence of political systems that we aim to study. Because of this and because of the relatively short timeframe that limits the variance of some variables, we chose not to include them.

To test further the hypothesis of political competition, we ran several additional unreported regressions. Replacing the Herfindahl indexes by a global Herfindahl index for all parties does not affect the results, nor does removing them altogether. We also checked for the effect of the mean district magnitude for the Parliament, a variable taken from the World Bank DPI database that represents the degree of proportionality in the electoral system: this variable takes a negative and significant (at 5 % level) sign, indicating that, ceteris paribus, proportional representation tends to hinder reform, which is consistent with Persson et al. (1997), Milesi-Ferretti et al. (2002), and Persson and Tabellini (2003). The effect that broader coalitions reform more is thus a feature of the reform tactics chosen by the government, and not of the electoral system itself. In this regression, all the other variables keep their sign but only execoalition maintains its significance.

Unfortunately, our data do not allow us to disentangle international information spillovers from the timing of elections (see Sect. 5).

To confirm this, we substitute the unemployment rate for the output gap (unreported regression). Its coefficient is negative and significant at 10 %, indicating that high unemployment hinders reforms. A one percentage point increase in the unemployment rate decreases the likelihood of tax reform by 2.1 %.

We can only identify 4 “Lisbon” reforms that are specifically targeted to women. This prevents us from testing if the effect of the gender employment gap is directly related to gender-specific measures. The effect of the gap on the reforms targeted to low income is not statistically significant.

The interactions between labor tax reforms and labor markets legislation reforms are potentially important, in particular whether they are substitute or complement. To the best of our knowledge, there is no empirical study on this particular issue, and the lack of significance in our regression leaves the question open.

As indicated above, most targeted reforms are of Lisbon type. Regressing only on Lisbon reforms provides similar results (unreported). This suggests that even reforms that aim at increasing efficiency are mainly driven by political economy considerations more than by economic factors.