Based on empirical data, a two-equation game-type corruption reaction function model was developed. A 'data to model' approach was used rather than the usual a priori approach. The general hypothesis tested was the 'monkey see, monkey do' principle. The latest data on corruption among developing countries was obtained from the Enterprise Surveys done by the World Bank Group in 2010. The key variables were the percent of domestic firms expecting to make informal payment to public officials to 'get things done,' and the percent of foreign firms doing like wise. The time span is from 2002-2010. A variety of econometric methods were used. The statistical results were quite good and supported the hypothesis. Both reaction equations were positively sloped. Time had a reducing effect on the frequency of domestic corruption, yet it had an increasing effect on foreign corruption. Variations in the frequency of corruption across regions of countries were generally not significant.