|
Abstract:
|
In many large data sets of economic interest, some variables, as wages, are right-censored. In order to analyze wages with the IAB-employment register we first have to solve the problem of censored wages at the upper limit of the social security system. We treat this problem as a missing data problem and use multiple imputation approaches to impute the censored wages by draws of a random variable from a truncated distribution, based on Markov chain Monte Carlo technique. In general, the dispersion of income is smaller in lower wage categories than in higher and the assumption of homoscedasticity in an imputation model is highly questionable. Therefore, we suggest a new multiple imputation method which does not presume homoscedasticity of the residuals. Finally, in a simulation study different imputation approaches are compared to confirm the necessity and validity of the new approach.
|
