- Multivariate Probit (2) (remove)
- Investing at home and abroad : different costs, different people? (2009)
- We investigate US households’ direct investment in stocks, bonds and liquid accounts and their foreign counterparts, in order to identify the different participation hurdles affecting asset investment domestically and overseas. To this end, we estimate a trivariate probit model with three further selection equations that allows correlations among unobservables of all possible asset choices. Our results point to the existence of a second hurdle that stock owners need to overcome in order to invest in foreign stocks. Among stockholders, we show that economic resources, willingness to assume greater financial risks, shopping around for the best investment opportunities all increase the probability to invest in foreign stocks. Furthermore, we find that households who seek financial advice from relatives, friends and work contacts are less likely to invest in foreign stocks. This result corroborates the conjecture by Hong et al. (2004) that social interactions should discourage investment in foreign stocks, given their limited popularity. On the other hand, we find little evidence for additional pecuniary or informational costs associated with investment in foreign bonds and liquid accounts. Finally, we show that ignoring correlations of unobservables across different asset choices can lead to very misleading results. JEL-Classifications: C35, D14, G11, G35 Keywords: Foreign Assets, Household Finance, Stockholding, Multivariate Probit, Simulated Maximum Likelihood, Selection.
- Stockholding: from participation to location and to participation spillovers (2009)
- This paper provides a joint analysis of household stockholding participation, stock location among stockholding modes, and participation spillovers, using data from the US Survey of Consumer Finances. Our multivariate choice model matches observed participation rates, conditional and unconditional, and asset location patterns. Financial education and sophistication strongly affect direct stockholding and mutual fund participation, while social interactions affect stockholding through retirement accounts only. Household characteristics influence stockholding through retirement accounts conditional on owning retirement accounts, unlike what happens with stockholding through mutual funds. Although stockholding is more common among retirement account owners, this fact is mainly due to their characteristics that led them to buy retirement accounts in the first place rather than to any informational advantages gained through retirement account ownership itself. Finally, our results suggest that, taking stockholding as given, stock location is not arbitrary but crucially depends on investor characteristics. JEL Classification: G11, E21, D14, C35