Abstract
The consequences of sampling errors in estimating a simple labour demand model, using panel data of firms is discussed. It is found that the sampling errors of the variables at the firm-level due to the sampling process at the employee-level, have a substantial influence on the estimates of the elasticity of substitution between white-collar and blue-collar workers. It is shown that because of the large noise-to-signal ratio of the explanatory variable no evidence of substitution can be found.