Tanja Kirn - SwissMod – a new tax-benefit model for Switzerland

  • Presenting author: Tanja Kirn (University of Liechtenstein, University of Freiburg)

  • Authors: Patrick Oschwald

  • Session: A02B - Tax-Benefit - Monday 16:30-18:00 - Marietta-Blau Hall

This study presents the newly developed static microsimulation model SWISSMOD, a new tax-benefit model for Switzerland. SwissMod is a spin-off of EUROMOD, the tax-benefit microsimulation model for the EU-27 plus the UK (Sutherland and Figari 2013). SwissMod simulates the main aspects of the Swiss tax and transfer system, such as income tax, as well as the main social insurance schemes, such as Old Age, Survivors’ and Disability Insurance (AHV/IV), Supplementary Benefits (EL), Occupational Pensions (BVG), Accident Insurance (UV), Income Compensation Allowance (EO), Unemployment Insurance (ALV), Family Allowances (FZ). It can be used to estimate and vary tax liability and benefit entitlement for a nationally representative sample of individuals, families and households. The microdata used in SwissMod are extracted from the Swiss SILC 2020 dataset, which is available from the Swiss Federal Statistical Office (BFS). The dataset is collected according to the EU-SILC survey guidelines of Eurostat, while households are randomly sampled from the sample register of the Federal Office of Statistics (BFS, 2023). The dataset contains about 8,000 household observations covering about 18,000 individuals. Variables are coded according to EU-SILC standards to ensure comparability (Eurostat, 2021). The variables include information on demographics, income, housing, employment, education and many other areas that reflect the general living conditions of Swiss households (BFS, 2023). The raw data provided by the Federal Statistical Office have been modified to be compatible with our model, without losing any information. To place the model in context, we provide an overview of existing Swiss microsimulation models and outline the main features of SwissMod. We describe the policies modelled and the input data used. We also provide an overview of the overview of the validation process. The article concludes with an outlook on further applications of the model.