|
Abstract:
|
Household finance surveys are increasingly used for policy-making. It is therefore essential that they provide an accurate picture of the economic situation of all households. Unfortunately, research has shown that the upper parts of the wealth distribution are often missing in household finance surveys. Since rich households generally concentrate a large share of total income and wealth, survey-based estimators may be biased or affected by low precision. The ideal situation to cope with this issues would be to have access to auxiliary information on household finances at the design stage. In practice, however, this is rarely the case. In this paper we present an empirical application that uses personal tax records in the design of a major survey on household finances. We first discuss the methodological challenges to be dealt with when using administrative information for designing the sample. We then propose a method for an optimal stratification and sample allocation. Finally, we estimate the benefits of the methodology in terms of precision and bias-reduction of the estimators.
|
