Groups Excluded from "Representative" Household Surveys: An Analysis Based on Remittances Sent and Received in Vietnam (Research Notes) (Report‪)‬

I. Introduction The Vietnam Living Standard Surveys (VLSS) in 1992/93 and 1997/98, which were conducted by the General Statistics Office of Vietnam (GSO) along with other international agencies as a part of the World Bank's Living Standard Measurement Surveys, are assumed to be representative of the entire Vietnamese population (Dollar and Kraay 2004; Cox 2004). However, this paper uncovers an anomaly in the surveys, that is, the identified amount of remittances received by households from domestic sources is significantly larger than the amount these same households send to others in Vietnam. This should not be the case, because if the surveys provide a representative sample of the entire Vietnamese population, the amount of remittances sent from domestic households to other domestic households should be approximately equal to the amount received by domestic households from other domestic households. Recently, Pincus and Sender (2008) have argued that the survey design methodology leads to the exclusion of many internal migrants and cannot be considered representative, particularly for a country undergoing rapid economic transformation and urbanization. They argue that migrants are missing from the survey design both because of the passing time between drawing sample lists and conducting interviews, and because many migrants are not legally registered at their new locations. Our contribution is that, by further exploring the questions about remittances in the surveys, we are able to use the surveys themselves to confirm that particular groups are underrepresented and thus to provide direct empirical support for the conclusions of Pincus and Sender (2008). Our results demonstrate that female and urban-dwelling remittance senders are underrepresented in the surveys.