We ask whether a PAYG-financed social security system is welfare improving in an economy with idiosyncratic and aggregate risk. We argue that interactions between the two risks are important for this question. One is a direct interaction in the form of a countercyclical variance of idiosyncratic income risk. The other indirectly emerges over a household's life-cycle because retirement savings contain the history of idiosyncratic and aggregate shocks. We show that this leads to risk interactions, even when risks are statistically independent. In our quantitative analysis, we find that introducing social security with a contribution rate of two percent leads to welfare gains of 2.2% of lifetime consumption in expectation, despite substantial crowding out of capital. This welfare gain stands in contrast to the welfare losses documented in the previous literature, which studies one risk in isolation. We show that jointly modeling both risks is crucial: 60% of the welfare benefits from insurance result from the interactions of risks.
social security idiosyncratic risk aggregate risk welfare