Does foreign aid reduce tax revenue? Further evidence

Abstract

A common criticism of foreign aid is that it reduces domestic tax effort. Empirical research on the issue has been hampered by the failure to tackle endogeneity issues effectively. We use measures of geographical and cultural distance to donor countries as instrumental variables to uncover the causal effect of aid on tax revenue in a panel of 93 countries. The tax to GDP ratio is found to decrease following aid inflows. This reduction in tax effort is statistically and economically significant; a one SD increase in aid causes a 0.52 percentage point drop in the tax-to-GDP ratio. The results indicate that the effect is driven by unconditional grants, whereas aid given as loans induces recipient governments to improve their tax effort. Our results are robust to changes in the sample and the use of a nearest neighbour matching technique to account for nonrandom assignment of aid. Our identification strategy is sharpened by the use of a difference-in-difference estimation strategy that leverages a natural experiment in which aid flows exogenously increased for some countries following the Iranian Revolution in 1979.

Keywords:

• foreign aid
• tax revenue
• instrumental variables
• difference-in-difference estimation

JEL Classification:

• F35
• H20

Notes

¹  We define tax effort as the tax revenue-to-GDP ratio and use these terms interchangeably throughout the article.

²  The aid data refer to official international development assistance in 2010 prices and is from the OECD’s Geographical Distribution of Financial Flows to Aid Recipients database. Note that not all developing countries experienced stagnant tax revenue-to-GDP ratios, however. The IMF (2011) reports that for nonresource-rich low- and middle-income developing economies, average tax revenue ratios were generally about equal to or lower at the end of the period compared to the beginning, but were somewhat higher at the end for upper-middle-income developing economies, and virtually unchanged for high-income developing economies.

³  Bulír and Hamann (2007) show that aid flows are more volatile than domestic tax revenue and that this volatility has increased over time.

⁴  See Todd et al. (2006) for a good discussion of the relevant literature.

⁵  Some studies, such as that by Gupta (2007), solve this problem by using lagged values of aid as instruments. A possible limitation of this approach is that where aid flows are serially correlated through time, the endogeneity problem will remain.

⁶  The bilateral distance data are from the CEPPI Ultimate Gravity database available at: http://www.cepii.com/anglaisgraph/bdd/gravity.asp variable, and the remaining series are the authors’ calculations based on information in the Central Intelligence Agency World Factbook available at https://www.cia.gov/library/publications/the-world-factbook/

⁷  For example, each country in the sample will have three exogenous variables that will serve as instruments for the level of foreign aid receipts, defined as: Aid-Distance_i = {(Inverse of bilateral distance_i,j)} * Aid outflows_j Aid-Religion_i = Religion_i,j * Aid outflows_j Aid-Colonial history_i = Colonial history_i,j * Aid outflows_j. We regress actual foreign aid on the exogenous instruments presented and report the respective t-statistics and for these first stage regressions in the main text.