In a study of the  effect  of civil war  exposure on local  collective action outcomes in Sierra Leone,  Bellows and Miguel (2009) employ a coef´Čücient stability approach to assess the importance of omitted variable bias building on Altonji et al. (2005a). Here  we  clarify  the econometric assumptions underlying Bellows and  Miguel (2009), and  extend their analysis using data on  dependent variable reliability ratios and  the method developed in Oster (2015).

Supplementary Materials and Data

If you would like to obtain more information about the data used in this paper, please contact Edward Miguel (emiguel@berkeley.edu).