Life Insurance, Housing, and Household Portfolios

This thesis extends the conditional portfolio efficiency framework of Gouriéroux and Jouneau (1999), as applied by Pelizzon and Weber (2008) to a single illiquid constraint, to the empirically relevant case in which households simultaneously hold two constrained asset positions: residential real estate and life insurance and pension products. Observed Italian household portfolios deviate substantially from mean-variance efficient allocations, yet it remains unclear whether such deviations reflect genuine irrationality or rational responses to large illiquid positions that generate hedging demands in the tradable financial portfolio. The theoretical contribution is to show that the optimal financial-asset allocation decomposes additively into the Markowitz tangency portfolio and two independent hedge terms, one per constrained asset. The framework is applied to 3,462 eligible households from the 2022 Survey on Household Income and Wealth, using asset returns spanning 2005–2024. Conditioning on housing alone reduces the share of conditionally efficient households from 38.3 percent to 19.7 percent at the ten percent test size, while conditioning on insurance leaves efficiency broadly unchanged at 51.1 percent, confirming that housing, not insurance, constitutes the binding illiquid constraint. A pronounced and persistent North–South efficiency gradient is documented, with 42.2 percent of North-Eastern households classified as conditionally efficient against only 2.7 percent in the South and Islands, and a monotone age gradient shows younger households to be the most conditionally inefficient.