Économie et Statistique n° 469-470 - Statistics on Income and Living Conditions (EU-Silc/SRCV) Income and poverty - longitudinal approach and international comparisons; quality of life
Attrition in the SRCV survey: determinants and effects on the measurement of monetary variables
The aim of this study is to measure attrition and its effects on disposable income in the Statistics on Income and Living Conditions survey in France. This survey is based on a rotating sample whereby the individuals are tracked for nine years. Six individuals out of ten are no longer part of the survey by the eighth year. Residential mobility is the prime cause of this attrition. Nonresponse is also encouraged by situations of poverty in the previous year, as well as inactivity and poor health. Nonresponse also increases when the survey investigator changes. The effects of attrition are correcting by re-weighting. The efficacy of these adjustments is tested by comparing 2003 incomes, as measured across all 2004 respondents, with those measured in the increasingly small samples of respondents in the waves of 2005 to 2011, after re-weighting. As regards overall income, the deviations are systematically below 2%, with a slight peak in the fourth year. The biases after adjustment are also small for most components of income: wages, family allowances, housing benefit, unemployment benefit. A bias does however subsist in pensions, and grows continuously from one year to the next. This bias remains even after differential mortality per social category is taken into account and does not appear to be explained by the intensity of institutionalisation of the elderly.