The religious transition: a long-run perspective

We use factor analysis to derive a robust measure of religiosity from items reported in five waves of the World Value Survey. Our measure of religiosity is negatively correlated with per capita income. Development apparently causes religiosity to fall to about half its pre-modern level. Most components of the demand for religion are reduced by development. The supply of religion declines once churches lose control over the institutions providing collective goods like education, health, and social security. These goods used to be supplied by churches jointly with religious services but tend to be supplied by the state with rising levels of development. Aspects of supply and demand are integrated in a CES production function framework that can explain the direction of causality in the observed negative correlation between income and religiosity.