On the robustness of balance statistics with respect to nonresponse

Abstract

A general problem for survey conductors is the fact that the response decision can be connected to the intended answer of the non-respondents. This nonresponse bias might have a substantial effect on the aggregated results. In this paper, a participation framework for the widely used business cycle balance statistics indicators is examined. An extensive simulation study is performed to analyse their effects. The analyses show that these indicators are extremely stable towards nonresponse biases.