We propose a new and simple method to study intergenerational wealth correlation between two generations, which is easy to implement in wealth (and housing) surveys and aims at overcoming the strong data limitation faced in most of the countries. We show that the ownership of housing assets can be used to proxy for three wealth groups for all cohorts. Misclassification induces a low and downward bias in the estimate of the intergenerational correlation. Using France as an example, we estimate intergenerational wealth correlation for cohorts covering the 20th century and focus on the wealth positions measured at the mid-life cycle of both children and parents. First, probabilities to belong to top wealth groups are increasing with the wealth of the parents. This intergenerational correlation has increased over time for most of the top wealth groups. Second, the higher we move up along the children’s wealth distribution, the larger the role of parental wealth: the persistence in the top 50% is 38% higher than under perfect mobility and the deviations from perfect mobility are larger in higher top wealth groups. Third, 50 to 60% of the correlation is accounted for by a mix of intergenerational wealth transfers, fathers’ occupation and children’s education. Fourth, gifts and bequests explain a larger share of the link between parental wealth and the probability to belong to the top 10% compared to larger top wealth groups. We also find evidence of persistence of the effect of parental wealth over the life-cycle.
The relative importance of wealth has sharply increased in advanced economies (Piketty and Zucman, 2014). While such an increase should not necessarily be viewed as negative in itself, it raises questions about the determinants of wealth concentration and the persistence of inequality across generations. Compared to intergenerational correlations in income or education that have been widely studied, the empirical work on the intergenerational correlation in wealth is more recent. Our contribution to this literature is twofold.
Our first contribution is related to the measurement of intergenerational wealth correlation in the absence of extensive administrative data or long panel dataset. We propose a new method to overcome this lack of data and estimate the correlation of wealth across two generations (parents and children). Our analysis is based on the French Wealth Survey (Insee) which measures household’s wealth. Interestingly, this survey collects information on whether the parents of the respondents were owner of real estate assets during the respondents’ childhood. We document that the ownership of the main residence as well as other real estate properties can be used to measure the position of the parents in the wealth distribution, and that such survey questions thus provide a convincing measure of the intergenerational correlation. In France, having no real estate property is almost certainly associated with belonging to the bottom 30% of the wealth distribution, whatever the birth cohort; being owner of the main residence (with or without other real estate property) is associated with belonging to the top 70%; while being owners of other real estate, in addition to the main residence, is associated with a position in the top 50%. The bias due to misclassification is low and is a downward bias. Our method thus slightly underestimates the correlation. Using data from other European countries and the US, we show that our method can be extended to other countries to convincingly proxy for wealth positions, and thus to extend the study of intergenerational wealth mobility to numerous other countries.
Our second contribution is to use our method to study intergenerational wealth correlation and unveil new results for France for children cohorts born all over the 20th. First, the probabilities to belong to top wealth groups increase with the wealth of the parents. This intergenerational correlation has increased over time for the probability to belong to the top 75%, 50% and 25% wealth groups and remains high and stable for the the top 10%. Second, we find non-linearities in the intergenerational correlation across the wealth distribution (cf. Figure). The higher we move up along the children’s wealth distribution, the larger the role of parental wealth. Third, the effect of parental wealth appears persistent over the life-cycle. Fourth, 50 to 60% of the correlation is accounted for by a mix of intergenerational wealth transfers, fathers’ occupation and children’s education. Finally, gifts and bequests explain a larger share of the link between parental wealth and the probability to belong to the top 10% compared to larger top wealth groups.
Updated on: 07/21/2020 11:13