Quarterly turnover is available in two business statistics time series: it can be derived from monthly survey data and there is a quarterly census series based on Value Added Tax (VAT) and survey data. In this paper we analyse seasonal effects in the reporting of businesses of survey versus VAT turnover. Furthermore, we investigate how many businesses contribute to these differences.
Statistics Netherlands aims to benchmark monthly turnover growth rates based on sample survey data onto quarterly growth rates based on a combination of VAT and survey data. Those growth rates are published for a range of economic sectors and underlying industries. This benchmarking has been postponed since a previous study on 2014 and 2015 data suggested that the two time series have different seasonal effects: the yearly distribution of quarterly turnover tends to be shifted more towards the fourth quarter of the year for the VAT data than for the sample survey data.
In the present study, we started by fine-tuning a number of model settings of a linear regression model to explain the differences between individual VAT and survey turnover values, accounting for outliers. For the selected model, we found that the two times series consistently show seasonal effects for different economic sectors and years. Those effects could not be explained by units within a limited set of reporting patterns. Instead, a considerable proportion of the units were found to contribute to the seasonal effects. The latter implies that we cannot solve the differences by manually editing a limited number of units.
