Rafael di Tella: This paper is concerned with one of the most important questions in the corruption literature, namely, would raising bureaucratic wages reduce corruption? This is a key policy question that also has important theoretical implications. If the answer is yes, it would provide an indication that the basic incentive model is applicable to the problem of corruption, as argued in a seminal paper by Becker and Stigler.1 The hypothesis, however, has received relatively little empirical attention. Panizza looks at the correlation between the public sector wage premium and corruption across countries in Latin America and the Caribbean. The author also discusses the relationship between the wage premium and measures of bureaucratic quality, but I am unsure about the interpretation of these results.
The paper starts by documenting the public-private wage differential in Latin America. Since presumably only a fraction of public sector workers have opportunities for corruption, the overall premium is only of moderate interest. Panizza then partitions the sample to obtain information concerning public sector workers who are in charge of making decisions and who thus have the opportunity to become corrupt. This approach is promising, and it is certainly an improvement over the previous literature, which just looks at the average premium for the public sector. The problem is that the results show a very small premium for male public sector workers with little education and a wage penalty for those with high levels of education. Thus as the opportunities for corruption increase, the premium decreases. Since Becker and Stigler predict no relationship between bureaucratic wages and corruption when there is a public sector wage penalty, the results do not present a direct test of the theory. Even if there were some premium to working in the public sector, we would still need some reassurance regarding the presence of auditing, given that, again, the theory does not offer many interesting predictions concerning the relationship between wages and corruption in the absence of auditing. [End Page 104] As to why the results on corruption and wage premiums are so weak, the lack of auditing is as equally good an explanation as Panizza's citing of the poor quality of the corruption data. This is particularly true considering that the ICRG data correlate so well with other corruption indexes, the exceptional cases pointed out by the author notwithstanding.
Another issue is the simultaneous determination of corruption and public sector wages. This is a serious problem in previous work using cross-country data, and it is also a concern in this paper. Clearly, the government cannot afford high wages when corruption is a drain on public resources, so the presumed correlation is again negative. The author seems to be aware of this problem, as well as of the difficulty of finding a plausible instrument. Without some reassurance regarding causality, however, the exercise is bound to generate few policy implications.
Perhaps the most interesting finding of the study concerns the pattern of wage premiums. Although informal knowledge indicates that public sector employment performs some kind of redistributive role, Panizza documents this process across gender and education. He finds that the group that is worst off actually suffers a wage penalty. This has important implications for students of public sector dynamics in Latin America. The analysis suggests that if public sector promotions are biased in favor of those who have formal educational achievements and who are male, as is generally held to be the case, then the system is not likely to place the more talented and honest professionals at the top of the public sector hierarchy. This is because these two characteristics are associated with the lowest wage premiums—and sometimes even a penalty. Thus the question is why would a highly educated male seek employment in the public sector in the first place. It must be that the public sector offers perks not found in the private sector. Maybe the public sector subjects workers to low monitoring, and these workers either like to get away with low effort or like to take advantage of any opportunities for extra income. Whatever the case, the author is suggesting that if there...