Earmarking means dedicating specific public revenue streams to a specific purpose. These revenues then accumulate in earmarked funds. Politicians utilize earmarks to serve their clientele while voters are pleased to trace the taxes they pay. Despite the enthusiasm in politics, academics still discuss about the effects of earmarking. According to exponents of public administration earmarking prevents an optimal budget allocation. On the contrary, public choice scholars see in earmarking a mechanism that enforces the government to spend a certain share of the total budget for the intended purpose. The two opinions have provoked a rich discussion in the literature, which has not yet come to a conclusion. While these arguments are rather abstract, the thesis proposes an underlying mechanism linking earmarking with technical efficiency of Swiss cantons. It argues that the ministers have less incentive to monitor their colleagues if the latter benefit from earmarked revenues. Less monitoring allows ministers to appropriate rents which eventually leads to inefficiency. This means, then, that higher shares of earmarked revenues are predicted to diminish technical efficiency.
To test the prediction, two quantitative approaches proved successful in the literature. Albeit both use a production frontier as a reference to estimate the efficiency of each canton, their underlying assumptions are fundamentally different. The Stochastic Frontier Analysis (SFA) specifies a functional form which makes it possible to separate efficiency from stochastic noise. Also, the method already developed in the spheres of panel data and exploits them accordingly. In contrast, the Data Envelopment Analysis (DEA) forms the production frontier using the observed data points and thereby follows a deterministic logic. Even if today’s DEA estimators acknowledge a certain noise in the data, their application on panel data leaves room for improvement. The study remedies this shortcoming by proposing an algorithm to estimate the effect of environmental variables on efficiency with fixed effects. The respective Monte Carlo simulations reveal the dominance of the proposed algorithm in terms of accuracy in comparison to conventional estimators.
The empirical test of the debate prevention theory, based on data of the 26 Swiss cantons from 2000 to 2014, indeed yields significant results – both in a statistical and a practical sense. Apparently, the effect that earmarking has on efficiency depends not only on the domain under investigation but also on how well the special financings are funded, on the restrictiveness of the debt brake, on the share of tax revenues, on the fractionization within the government council, and on the concordance between the executive and the legislative branch of government. In general, earmarking negatively affects efficiency in the cultural domain whereas the opposite is true for the transportation domain. A possible explanation can be found by returning to the public choice literature. With reference to the three criteria for a successful implementation of earmarking, the transportation domain meets all of them considerably well. When looking at the cultural domain however, the correlation between the lottery (the taxed good) and the benefitting service (culture) appears rather weak. Furthermore, many different goods and services can be gathered into the category of culture, which makes it difficult to clearly differentiate from other goods. Finally, the non-paying beneficiaries of culture are difficult to clearly exclude.