If the proportion of wage earners with a high likelihood for sick leave increases, the percentage of wage earners with sick leave absence will increase even though the underlying trend in sick leave is unchanged. This phenomenon is called a “compositional effect”.
In this report, we study how compositional effects affect the evolution of both sick leave absence and labor market attrition after long spells of sick leave absence. To this end, we use a decomposition method to distinguish between compositional effects due to changes in the composition of wage earners, and the underlying trend in sick leave absence or attrition after long spells of sick leave absence.
By using matched employer-employee data from Statistics Norway, as well as registers from the Norwegian Labour and Welfare Administration, for the period 2015–2020 we find that compositional effects make up a small fraction of the changes in sick leave absence over time. This holds for both decompositions along the dimensions separately for the population at large, as well as decompositions separately for each industry. The reason for this is that the dimensions along which we decompose the change in sick leave absence evolve slowly over time. This implies that the compositional effects on sick leave absence are also fairly small in magnitude for the period we look at in our analysis.
For the changes in the attrition after long spells of sick leave absence (measured by an attrition indicator) the contribution from the compositional effects are larger, as a result of the indicator being calculated from a sample based on wage earners with long spells of sick leave absence. This is a group whose composition varies more over time than the composition of the population of wage earners at large, which entails that the compositional effects on the evolution of the attrition indicator are larger.