The potential endogeneity between goal satisfaction and the wealth indicator (asset per capita) was also explored as the causal relationship between the two variables could run in both directions or their relationship could be due to unobserved characteristics (Knight and Gunatilaka 2010). The fact that the wealth variable was obtained a couple of years before the measure for goal satisfaction reduced the risk of endogeneity. However, we ran a Hausman test to check whether this was the case. We found a suitable set of instrumental variables correlated with the wealth index but not necessarily affecting goal satisfaction. These were type of roof in Bangladesh and self-perceived economic status in Thailand.