The bicausal relation between religion and income

In this article the relation between religion and income is investigated using a micro-dataset for the Netherlands. Religion is measured by religious membership and by participation. Instead of estimating a religion equation and an income equation separately, joint regression is preferred since this generally yields more efficient estimates. Following the single-equation approach, both religious measures are found to decrease income significantly and income is found to affect religion negatively. However, these cross-effects become insignificant once the equations are estimated simultaneously. In contrast, the effects of socio-economic characteristics on religion and income hardly differ between the approaches.