Imputing household expenditure for indirect tax simulation: Extending SWISSMOD using statistical matching

Indirect taxes such as value-added tax (VAT) and environmental levies play a growing role in fiscal and distributional policy in Switzerland, yet existing microsimulation tools lack the consumption information required to analyse their incidence. This paper closes this gap by extending SWISSMOD - the Swiss tax-benefit microsimulation model - through the imputation of detailed household expenditure from the Swiss Household Budget Survey (HBS) into Swiss SILC data. Building on the two-step econometric approach of Akoæguz et al. (2020), we estimate participation and conditional spending using probit and OLS models, followed by Mahalanobis distance matching to transfer observed expenditure structures at the most disaggregated level. We demonstrate how this methodology can be adapted and validated for a non-EU context characterised by distinctive survey design and consumption classifications. Validation using distributional tests, macro-level benchmarks, and income-ventile comparisons shows that the imputed data reliably reproduces key expenditure patterns, particularly for regularly consumed categories. The resulting enriched dataset enables comprehensive simulation of indirect tax reforms within SWISSMOD, providing a robust empirical foundation for evaluating VAT and environmental tax policies in Switzerland.