Description: Labor-Public-Applied Economics Workshop

Location: 368A Heady Hall

Contact: John Winters

Title: Asset market participation and portfolio choice: The role of parental wealth

Abstract: A sizeable positive correlation between child and parent wealth accumulation and risky-share is well documented.2 The mechanisms driving this could be due to selection or are causal (Fagereng et al., 2021). Selection occurs when parent abilities and preferences for saving or risk taking are passed on through genetics. Causal mechanisms include parent actions that increase child wealth, such as direct child investment, wealth transfers and learning behaviors. The seminal work of Chiteji and Stafford (1999) posit a learning mechanism: adult children are more likely to invest in stock if their parents do. This is confirmed by Black et al. (2017) and Fagereng et al. (2021), who find adoptive parent financial risk-taking has a positive effect on adult adoptees’ stock market participation and portfolio risk. Fagereng et al. (2021) highlight that genetics are not the only cause of the observed positive correlation between child and parent wealth.

A child’s decision to allocate her portfolio to risky financial assets can be influenced by parental wealth through mechanisms beyond genetics, learning or transfers. In this paper, we examine two as-yet unexamined mechanisms: a collateral effect and decreasing relative risk aversion (DRRA) effect. A child may view their parent’s wealth as ‘collateral’ (Yao and Zhang, 2005) as a support in the event of adverse shocks. This can lead a child to save less or invest in more risky assets. Further, a child might view their parent’s wealth as part of their own portfolio. Because relative risk aversion falls with wealth (Morin and Suarez, 1983), a child might thus invest in more risky assets as parent wealth increases. By ignoring these other channels, previous research may have attributed more importance to the learning channel in determining the causal share of risky assets that a child holds.

In this study, we use a twenty-two-year panel (1999-2021) of household data from the Panel Study of Income Dynamics (PSID), measures of non-retirement wealth and asset ownership, and the restricted-use location data to test how changes in parent portfolio composition influence a child’s portfolio. We append parental wealth and other parental data to the child’s record so that we can estimate child risky share, controlling for child and parent characteristics. Since a child’s parents may separate into different households, we consider alternative measures of head (and spouse parent if present) parent wealth metrics (i.e., the average and maximum value). Controlling for key household wealth determinants, we estimate equations for total net wealth (TNW), stock market participation and risky share.

We first test if child portfolio composition – level of wealth, stock market participation and risky share – are sensitive to variations in parent wealth using OLS and IV approaches. We instrument for parent metric of interest – level of wealth, stock market participation and risky share – using exogenous changes to parent home equity. We examine two IV approaches. First, we use current and year of move state home price index (HPI) for the parent state of residence to instrument parent wealth. While the current year HPI is a strong predictor of house value, controlling for both current and year of purchase HPI generates variation in home equity (Chetty et al., 2017). Second, we use a local shift-share measure of predicted local employment growth as an exogenous demand shifter of local housing demand in the parent MSA location.

To identify a collateral effect, we examine if children shift out of risky assets when a parent experiences a negative wealth shock due to divorce, separation or the onset of a debilitating health event, such as pre-mature memory loss. The adverse shock diminishes the likelihood that a parent can provide collateral in the event the child needs assistance. We examine child stock market participation and risky share in a stacked difference-in-difference framework with the treatment defined as a negative shock to parent wealth that is likely to be long lasting.

To examine a DRRA mechanism, we examine the association of child stock market participation and risky share with a measure of discounted parent wealth using OLS and IV approaches. We compare children of a similar age to those that have older or younger parents, and thus are more or less likely to receive inheritance in the near future, to determine how the discounted value of parental wealth influences the allocation of risky assets. Additionally, we investigate whether a t+1 gift/bequest receipt alters child investment behavior at time t.

Our analysis sample includes roughly 3000 households per year with parent data during the 1999 to 2019, comprising roughly 24,000 household-year observations with a rich set of covariates non-missing. Dramatic differences in household wealth accumulation by parent wealth exist. The stock market participation rate of households with high-wealth parents is nearly twice the rate of those with low-wealth parents: 24% versus 13%. The risky share held by households with high-wealth parents is nearly five times the rate of those with low-wealth parents: 14.5% versus 3%. Lifecycle profiles show that child TNW, stock market participation and risky share trajectories are substantially steeper if a parent is high-wealth rather than low-wealth and invests in stocks (rather than not). Finally, we find evidence that children pursue different investment strategies based on parent wealth via collateral and DRRA effects.