Tax evasion and Productivity

Abstract

Tax evasion is a significant concern for developing and developed countries. Despite the vast efforts of the literature, measuring tax evasion remains a challenging task. Direct empirical measures are mostly unreliable because firms and individuals have incentives to conceal their behavior. Indirect structural measures must account for the productivity of firms in the case of corporate tax evasion. However, tax evasion has not been satisfactorily addressed in the estimation of productivity. I show that ignoring tax evasion leads to biased estimates of productivity. To address this gap in the literature, I provide a new estimation strategy to jointly recover tax evasion and productivity using commonly available data.

Introduction

Tax evasion is a significant concern for developing and developed countries¹. Despite the vast efforts of the literature, measuring tax evasion remains a non-trivial task². Direct empirical measures are mostly unreliable because firms and individuals have incentives to conceal their behavior ( Citation: Slemrod, 2019 Slemrod, J. (2019). Tax compliance and enforcement†. Journal of Economic Literature, 57(4). 904–954. https://doi.org/10.1257/JEL.20181437 ) . Hence, it is unlikely that evasion would be truthfully reported in surveys, for instance. Indirect structural measures have had some degree of success in the case of individual income tax evasion ( Citation: Pissarides & Weber, 1989 Pissarides, C. & Weber, G. (1989). An expenditure-based estimate of Britain’s black economy. Journal of Public Economics, 39(1). 17–32. https://doi.org/10.1016/0047-2727(89)90052-2 ) , but in the case of corporate tax evasion researchers must account for an additional unobserved variable, the productivity of firms. Having productivity estimates could potentially help, however, tax evasion has not been satisfactorily addressed in the estimation of productivity. I show that ignoring tax evasion leads to biased estimates of productivity. To address this gap in the literature, I provide a new estimation strategy to recover tax evasion and productivity using commonly available data.

Indirect structural attempts to estimate corporate tax evasion have to account for the productivity of the firms. The reason is that low productivity might be naively quantified as tax evasion. Consider the case in which firms overreport input expenses to reduce their profits to evade taxes. A practitioner with output and input data might conclude that output is a function of inputs, thus it can be considered a second measure. She might try to recover true inputs and get an estimate of tax evasion by the difference between reported and true inputs. However, for a given level of output, high input utilization by a firm could be explained by either the amount of input the firm overreports to evade taxes or by its low productivity.

Firm-level estimates of productivity could be helpful to measure tax evasion, however, tax evasion has not been satisfactorily addressed in the estimation of productivity by the literature. The literature has coped with tax evasion by treating it as a classical measurement error. In other words, it has been assumed that tax-evasion misreporting has zero mean —some firms under-report and others over-report so that the misreporting does not bias the estimates of interest— and this misreporting is independent of everything else, including the attributes of the firms. However, the classical measurement error argument is inconsistent with economic intuition³.

I show that ignoring tax evasion leads to biased estimates of productivity. Productivity is measured as the residual of a production function, where the output is a function of the inputs. A key assumption is that input demand is strictly monotonic on the productivity ( Citation: Gandhi, Navarro & al., 2020 Gandhi, A., Navarro, S. & Rivers, D. (2020). On the Identification of Gross Output Production Functions. ; Citation: Ackerberg, Caves & al., 2015 Ackerberg, D., Caves, K. & Frazer, G. (2015). IDENTIFICATION PROPERTIES OF RECENT PRODUCTION FUNCTION ESTIMATORS. 2411–2451. https://doi.org/10.3982/ECTA ; Citation: Levinsohn & Petrin, 2003 Levinsohn, J. & Petrin, A. (2003). Estimating production functions using inputs to control for unobservables. Review of Economic Studies, 70(2). 317–341. https://doi.org/10.1111/1467-937X.00246 ) . That is, more productive firms will use fewer inputs and produce more output. To reduce their declared profits and evade taxes, firms might underreport sales or overreport expenses. Therefore, if firms underreport their output to reduce sales, or if firms overreport inputs to increase expenses, then the productivity estimates would be biased downward.

To address this gap in the literature, I provide a new estimation strategy to jointly recover tax evasion and productivity using commonly available data. In particular, the method can identify the density of tax evasion through input overreporting. The key insight is that the first-order conditions of the firms’ cost minimization problem are informative about the optimal utilization of inputs and hence about tax evasion through input misreporting up to the current-period output shock. The identifying assumption is that there is a subset of firms that do not evade by overreporting inputs and the practitioner can identify them with some degree of confidence. From this subset of good firms, the production function parameters and the density of the random noise can be identified. We can then apply deconvolution techniques to learn about the distribution of tax evasion and productivity, and how they have changed over time. Furthermore, I can learn how evasion changes with productivity, and how this relationship has changed over time.

The obvious strands of the literature to which I contribute are tax evasion and productivity estimation. The contribution to the tax evasion literature is two-fold: (1) I provide an estimation strategy using commonly available data, and (2) the method identifies tax evasion through cost overreporting, an overlooked by the literature but relevant phenomenon. The contribution to the productivity literature is to show that ignoring tax evasion leads to biased estimates and to provide a method to recover the productivity density in the presence of input overreporting.

References