Using Register Data in Income Statistics in the Austrian EU-SILC: (Why) Do People Get Poorer?
In our paper, we take advantage of the fact that for the Austrian SILC 2011 to 2008 both register-based and questionnaire-based income data is available. We aim at providing explanations for changes in different aggregate poverty indicators by investigating the underlying changes in the distribution of household income as a consequence of using register data. It is asked which component (income type, weighting) contributes most to the change of the poverty headcount if register data is used instead of questionnaire data. We also look at overall and group-specific differences in measured household income and poverty headcounts between the two data source. Furthermore, by estimating multinomial logit and linear models with covariates referring to the interview situation (f.i. CATI vs. CAPI) and household characteristics we try to explain which households tend to underreport respectively overreport their household income.
Preliminary results show that the distribution of most income components at the personal and household level become more unequal if register data is used. We find that mean reverting errors - particularly for employee income - seem to drive the change of the poverty rate most: high income households tend to underreport whereas low income households tend to overreport their income. An open question remains to which extent this is due to social desirability and/or cognitive errors.