Assessing the welfare change from a pension reform

We study the welfare implications of a hypothetical reform of the Swedish public pension system where eligibility to pension benefits is delayed by 3 years. Using an option value model, we consider the labor supply responses to the reform and develop a compensating variation (CV) measure to analytically assess the individual welfare changes in a random utility framework. We find that a purely budgetary calculation (neglecting individual labor supply responses) overestimates the welfare loss by more than 65%. We also develop a method for testing between a binary and a multinomial option value model, where the binary one is nested in the multinomial model in a Generalized Extreme Value (GEV) model framework. The binary model cannot be rejected.