I don't quite understand what exactly they did. How do you actually add a random error component to predicted household income?
The Impact of Remittances on Poverty and Human Capital: Evidence from Latin American Household Surveys. Acosta et al. (2007)