A NOTE ON ENFORCEMENT SPENDING AND VAT REVENUES.

Publication: Review of Economics and Statistics

Publication Date: 01-MAY-01

Bookmark this article Print this article Link to this article Email this article Digg It! Add to del.icio.us RSS

COPYRIGHT 2001 MIT Press Journals

I. Introduction

ONE OF THE main fiscal problems in developing countries is the high level of tax evasion. For example, in the case of Chile, where evasion levels are low by developing-country standards, studies suggest that currently about 50% of the income tax and 23% of the value-added tax (VAT) are evaded.(1) All in all, it is estimated that evasion amounts to about 6% of GDP, which is quite high when compared with tax revenues of approximately 18% of GDP.

A recurrent explanation for these high levels of evasion is that enforcement is lax because the agencies in charge of collecting taxes are understaffed, underpaid, and poorly equipped. Quite often, it is argued that their budgets should be increased. Yet whether that is a good idea depends, among other things, on the yield of the additional expenditures, and thus is an empirical question. Furthermore, in developing countries, data on evasion is often not available or suspect, so that it is difficult to evaluate the effect of changes in enforcement spending or any other policy intervention aimed at reducing evasion.

The purpose of this note is to show how revenue data can be used instead of evasion data to estimate the impact of changes in enforcement spending. This is important because revenue data are regularly collected in most countries and are far more reliable than evasion estimates. The idea is to exploit the elementary accounting identity that relates revenues and evasion to obtain a simple equation that can be estimated with observable data. An estimate of the elasticity of evasion with respect to enforcement spending follows from estimating the coefficients of this equation.

We use the method proposed in this note to quantify the effect of enforcement spending on VAT revenues in Chile, showing that the yield of increasing enforcement spending is substantial. Other things equal, had the budget of the IRS in 1997 been $1 (USD) higher, VAT revenues would have increased by $31.2. In other words, the cost of raising an additional $1 of VAT revenues was slightly more than $0.03 at the margin. Moreover, these figures are not valid only at the margin: the IRS budget (as a fraction of GDP) would have remained within the sample range even if it had been 40% larger in 1997, which suggests that VAT evasion could be significantly reduced by increasing enforcement spending.

This note is related to Agha and Haughton (1996) who studied the determinants of VAT compliance in a cross section of seventeen OECD countries and found that increasing administrative expenditures by $1 raises revenues by $12. Our study differs from theirs in that we use VAT revenues instead of an estimate of compliance as the left-side variable. By doing so, we avoid using evasion estimates.

This note is also related to Dubin, Graetz, and Wilde (1990), who studied the effect of audit rates on reported federal income tax per return in different U.S. states over time. They found that the audit rate has an economically large impact on reported taxes per...

Read the full article for free courtesy of your local library.