Bureaucratic corruption as a constraint on voter choice

Abstract

This paper has attempted to explain variations among countries in the tax level, in the importance of progressive taxes and transfers, and in the after-tax share of middle-income groups. Particular attention was paid to the declines in each of these variables which are observable in cross-section data in the intermediate ranges of development. It was argued that these phenomena are difficult to explain by variations in the demand for public spending or in the degree of tax evasion. The hypothesis set out in this paper was that these nonlinearities are a result of the inability of elected representatives to observe taxable activities at certain levels of income. The result is a loss of potential tax revenues through bribes to officials. Such bribes, it was suggested, are most likely in the income brackets where total income is increasing rapidly in developing countries. A theoretical model based on expected vote maximization by political candidates indicated that voter-taxpayers will be unwilling to make up the short fall by additional taxes. To the extent that the problem of observing taxable activities is more serious for direct than for indirect taxes, direct tax revenues should fail to keep pace with total income in the early stages of development. The presence of supply-side constraints on the capacity of the fiscal system to generate direct-tax revenues was tested by means of a simultaneous-equation model. Since in practice transfer payments or negative taxes form an important part of the direct tax system, it was necessary to include them in the empirical analysis. It was decided to define an unobservable variable, the overall fiscal structure, which reflects the effects of both tax structure and transfer payments. Use of the LISREL technique allowed this new variable to be treated as a latent variable in statistical estimation. The results indicated that overall fiscal structure varied in nonlinear fashion with the logarithm of per-capita income, as the theoretical model predicted. This variable was found in turn to be a significant determinant of the level of taxes and the distribution of after-tax income. Little evidence was found of feedback from these other variables to the overall fiscal structure. These findings were interpreted as being consistent with the hypothesis that voter choices may be constrained by the revenue-generating capacity of the fiscal system if a significant portion of taxable activities is not easily observable. Copyright Martinus Nijhoff Publishers 1987