How Portfolios Evolve For the Old: A Cross-country Comparison in Europe using SHARE Data

As aging is a serious social problem and wealth composition of the older households in Europe varies largely, studying factors influencing old people’s asset ownership rates and asset shares is urgent and important. Using five waves of panel data from SHARE, this thesis adopts random effects and fixed effects Logit models to study several variables’ effects on asset ownership rates and concludes that fixed effects fit better. I found that the effect of ageing generally follows a quadratic form in asset ownershiprates. Behavioral risks factors are associated with owning different assets but causal effects cannot be concluded. Cognitive abilities are positively related to asset ownership rates but causal effects cannot be concluded either. Secondly, this thesis uses random effects and fixed effects Tobit models to investigatefactors influencing asset shares. It is found that ageing is still very significant in many cases but not in risky financial assets in the fixed effects model. Behavior risks factors are more significant in influencing asset shares: the older households who drink more often or do more vigorous activities seem to havehigher asset shares; starting smoking has a causal negative effect on holding risky assets while stopping smoking has a causal positive effect on holding risky real asset (other real estate). Cognitive abilities are strongly causal associated with holding more risky financial assets.