So says a new paper by de Haan et al. (2013). They argue that government wages have a significant negative impact on corruption when income is $8,842 or less. Very interesting findings that is intuitive as well.
Using this new database, we construct a relative government wage indicator as the ratio of government wages to the average wages in the manufacturing sector and estimate the relationship between corruption and this indicator, controlling for other variables that previous studies found to be related to corruption. When the impact of government wages on corruption is assumed to be linear, as previous studies do, we find that one unit increase in the wage indicator, which is equivalent to raising government wages by the average of wages in manufacturing reduces corruption, measured on a scale from zero to six, by only 0.35. This result suggests that using government wages to combat corruption is rather costly, similar to the conclusion by Van Rijckeghem and Weder (2001).
However, the nature of corruption might be different at different levels of economic development. In low-income countries, corruption often consists of petty corruption, which involves tiny amounts of money, appears in a rampant manner and is easy to detect. Road bribery in India, where traffic policemen openly collect bribes from passing trucks (Bardhan 2006), or the extra payment to obtain a birth certificate in Cambodia (Feinberg 2009) are typical examples. In high-income countries, petty corruption is less common because wages are above subsistence level. Corruption in these countries, if present, involves more secret deals, brings about larger payoffs, and is more difficult to detect. Government wages will arguably be less effective to combat the latter form of corruption.
By including an interaction term between government wages and economic development, we allow the impact of government wages on corruption to vary with the level of economic development. Controlling for a large number of other determinants of corruption and country specific effects (which in our view are important because corruption changes very slowly and appears to be country specific), we find that the estimated coefficient of government wages is negative but the coefficient of the interaction term is positive. This suggests that the role of government wages in reducing corruption decreases as countries become richer.
Government wages have a significant negative impact on corruption when income is $8,842 or less. Above this income level, no significant relationship can be established. The impact of government wages on corruption is moderated by the level of income per capita. The poorer a country is, the stronger the negative impact of higher government wages on corruption is.