Income imputation in the Swiss Household Panel 1999-2007

This paper describes the methods used and the steps taken to impute missing income values in the Swiss Household Panel Survey (SHP). Missing values that result from both item- and unit-nonresponse are imputed. We impute income on the individual level, distinguishing between several income components. The imputed item- and unit nonresponse income distributions are compared with the distributions of the validly reported cases. The level of imputed income from employment resulting from item-nonresponse is similar to that of the validly reported. Other imputed income components from item-nonresponse are generally slightly, imputed income from unit-nonresponse considerably higher than that from the validly reported cases. This result shows that imputing missing income may avoid biased level estimates. Income variations of the valid cases and the imputed cases are not different.