Tax Reform Modeled as Optimization Problem

Researchers at an unnamed institution present a new framework for designing tax reforms, casting the process as a constrained optimization problem. They demonstrate that any tax code meeting specific criteria can be represented as a piecewise-linear function, allowing reforms to be generated using linear or mixed-integer linear programming.

This approach focuses on legislatable parameters like tax rates and bracket cutoffs, guaranteeing specific income effects and budgetary costs. The team successfully reconstructed the Dutch income tax code using this method, creating reforms that reduced spikes in marginal rates, limited household income losses, and decreased the number of active rules.

Developed with the Dutch Ministry of Finance, the open-source software implementation, TaxSolver, is currently being used for ongoing tax policy work. The researchers note the framework can incorporate behavioral effects, adding complexity to the model.