Transition plan redesign of the Dutch Labour Force Survey

This report proposes a methodology to estimate discontinuities due to the transition to a new design for the Labour Force Survey via a parallel run in combination with a time series model.

Monthly figures about the Dutch Labour Force are based on the Labour Force Survey. Official figures are produced with a multivariate structural time series model, which is used as a form of small area estimation to improve the precision of monthly estimates and to account for rotation group bias. Based on a Eurostat regulation, Statistics Netherlands is preparing a redesign that is planned for 2021. In a well-designed transition process discontinuities are quantified to avoid confounding real period-to-period change from the systematic effect of the redesign on the outcomes of the survey estimate.

One way to quantify discontinuities is conduct a parallel run, where data are collected under the old and new survey designs alongside each other for some period. Because of limited budget and field capacity a large parallel run is not possible for the upcoming transition. As an alternative, the information obtained with a relative small parallel run is combined with a time series modeling approach, where discontinuities are estimated with a level intervention. In this paper simulations under different transition scenarios are conducted to obtain insight with which precision discontinuities as well as the trend under the old and new level during the transition can be estimated.