This paper tests whether the behaviour of households in different countries is homogeneous with respect to the influence of religion on income. The violation of the homogeneity assumption would have two consequences. First, results based on country studies might not be applicable to other countries. Second, one should be careful when pooling cross-country data in this type of research. Data at household level of the European and World Values Survey are pooled for 25 Western countries. We estimate simultaneously an income and a religion equation to correct for the endogeneity of religiosity. We find that estimation outcomes are different between low and high-income countries. Whereas church membership is found to have a positive effect on income for high-income countries, this effect is negative for low-income countries. This result is robust to denominational distribution, participation effects and alternative measures of religiosity. Classification-JEL: Z12; D31 Keywords: Income; Religious Membership; Religious Participation File-Url: https://papers.tinbergen.nl/08014.pdf File-Format: application/pdf File-Size: 307420 bytes Handle: RePEc:tin:wpaper:20080014