Towards an Explanation of Household Portfolio Choice Heterogeneity: Nonfinancial Income and Participation Cost Structures

The paper uses micro data on income and asset holdings from the Panel Study of Income Dynamics to analyze reasons for nonparticipation and for heterogeneity in portfolio choice within the set of stock market participants. The focus of the paper is on non-financial income and costs of participating in the stock market. I find evidence of a strong positive effect of mean non-financial income on the probability of stock market participation and on the proportion of wealth invested in stocks conditional on being a participant. The volatility of non-financial income is found to have a strong negative impact on these two quantities. Both these results are consistent with the theoretical literature on portfolio choice in the presence of non-financial income. However, only a small or insignificant effect of the covariance of non-financial income with the stock market return on portfolio choice is present. Four different costs of stock market participation are considered, an entry cost, a fixed transactions cost, a proportional transactions cost, and a per period participation cost. The first three of these costs lead to structural state dependence in the stock market participation decision and in the proportion of financial wealth invested in stocks. A dynamic sample selection model shows evidence of strong state dependence and thus economically important entry and transactions costs. The per period participation cost does not lead to structural state dependence, but a censored regression model with unobservable stochastic threshold is estimated which allows for heterogeneity in the per period partipation cost and provides estimates of its distribution in the cross-section. I estimate the median per period participation cost to be around $100-$200.