Distortions, Efficiency and the Size Distribution of Firms∗

It has often been argued that tax administrations in less developed countries (LDCs) target larger firms more intensively. We examine the effect of this institutional distortion on the size distribution of firms and efficiency. We develop a model with entry and exit of heterogeneous firms where the tax environment acts as a selection mechanism which restricts the growth of all but the most productive firms. We calibrate the model by Indirect Inference to fit firm-level data from Uganda. The calibration strategy consists in choosing a set of parameters such that the model generates, on top of explicit targets, an endogenous size distribution of firms as well as patterns in firm’s growth and capital-labour ratios similar to the data. We test the fit of the model with a set of over-identifying restrictions and show that the model does a reasonable job at explaining the data along several dimensions. This allows conducting three policy experiments. First, we compare in terms of efficiency four types of taxes: on consumption, revenues, labour or capital. Second, we show that there are efficiency gains from redistributing audit probabilities optimally. Finally, we argue that bribes acts as a second-best mechanism because it buys entrepreneurs a tax rebate. Nonetheless there is room for efficiency improvements with the use of a reservation wage for tax officials