Assessing the correlation between labour income risk and household portfolio investment in risky assets: evidence from Italian longitudinal data

The aim of the present paper is to assess the correlation between labour income volatility and household portfolio choices in Italy. The analyses proposed are performed using SHIW data (Bank of Italy) on a representative sample of Italian households where “husband and wife” are both employees. To assess the correlation between investment decisions and labour income risk, the paper firstly analyzes the covariance structure of earnings in Italy, both at a household and at an individual level. Based on the evidence found, it proposes the construction of a labour income risk score, in turn estimated both at a household and at an individual level. This variable is then used as a regressor in a reduced form estimation modelling the determinants of household portfolio allocation in risky assets. The regression analysis, performed using novel econometric techniques for this branch of research, shows that labour income risk is negatively correlated with household portfolio allocation in risky assets. This finding is consistent with theoretical predictions, and is robust to different specifications and different econometric techniques. However, the regression results show that disentangling labour income risk for “husband and wife” does not add explicative power to the model. On the contrary, a joint measure for labour income risk is negatively correlated with household choice of investing in risky assets. This result is in line with the unitary household model predictions. What matters for household portfolio choice, at least in the light of the results displayed, is the joint volatility of labour income. That is, when incomes of the individuals in the couple are positively correlated, and labour income risk is consequently perceived as overall higher, household portfolio allocation is switched towards safer investment decisions.