Validating SHARE in France with other French surveys : health and income data

The Survey of Health, Ageing and Retirement in Europe (SHARE) is cross national: the questionnaire is identical across all participating countries and, because of the modest size of the sample in each country, it is usually not feasible to proceed to solely national use of the data. Moreover, as SHARE is unique in Europe in terms of scope and target sample, its results cannot easily be validated by comparison with other similar cross national surveys. This paper attempts to relate some key SHARE variables to their counterparts in other French surveys. We concentrate on health and income data that we relate to various INSEE surveys on Health, Consumption, Housing and Income. Concentrating on France, where the SHARE survey agency is the National Statistical Institute, allows the comparison to abstract from sample design and interviewers’ quality effects. We surmise that an ex ante harmonized questionnaire such as SHARE is easier to apply in qualitative domains such as health, or in non-ambiguous quantitative measures such as weight and height, but is harder in domains where each country has its own institutions and concepts. We assess the quality of the income questions both at the extensive margin (who gets what type of income, and non-response conditional on receiving) and at the intensive margin (what are the main quantiles of income distribution for recipients). We find that the French SHARE data are of good quality when the questions are simple. For instance, the body mass index of males is the same in the Health survey and in SHARE. However, discrepancies can be larger on quantitative data. They seem less important on the extensive than on the intensive margins, and generally less in wave 2 than in wave 1. We suggest some ways to improve the quality of future waves of SHARE.