Published in: Labour Economics 18 (4). (2011). 434-440. (with Juan Ponce)
This paper examines the earnings premium to computer use in a developing country: Ecuador. We use different approaches to examine whether the premium is causal. Controlling for an extensive set of observables, we find an earnings difference between users and non-users of around 20%. Using first differences, the premium drops and is no longer significant in a specification that includes proxies for workers' computer experience and knowledge. Estimates of the impact of the intensity of computer use are also small and in most cases insignificant. Estimates of the pencil premium are substantial in level specifications, but become insignificant in fixed effect specifications. Taken together, also in the setting of a developing country we do not find evidence in favour of the computer premium reflecting a causal impact.