Household Participation in Stock Market Varies Widely by State

From 1928 to 2016, the average annual stock return was about 8 percentage points higher than the return on three-month Treasury bills. This leads to sizable return gaps over time: $100 investments in stocks and in Treasury bills in 1928 would have yielded nearly $329,000 and $2,000, respectively, 88 years later.¹

Given the high return of stocks, it is puzzling that many households do not participate in the stock market and, hence, forgo the high return. In addition, the nonparticipation behavior is at odds with modern portfolio theory. The theory implies that all households should invest at least a fraction of their wealth in stocks in order to take advantage of the equity premium. However, the data show that many households do not participate in financial markets.

The inability of modern portfolio theory to explain what is observed in the data leads to a “participation puzzle.” A common explanation of this puzzle is the individual participation cost, which includes both monetary and nonmonetary costs. The monetary costs are relatively straightforward, including transaction or brokerage fees. The nonmonetary costs are broadly defined to be the cognitive and time costs of understanding the investment object or processing previous experiences with stock markets. The participation cost, especially the nonmonetary costs, could vary widely across the population.

While this participation puzzle has been well-documented at the national level, this article takes a first step in exploring whether the rates vary geographically. The aim is to see if the participation rate at the national level occurred homogeneously across states or if people in some states participated at a higher rate than others.

If the data indicate that there are regional differences in these rates, academic researchers could explore these disparities to better understand and resolve the puzzle. In addition, a better understanding of households’ investment decisions across regions could improve the quality and effectiveness of conducting government policy.

To answer this question, we employed the IRS’ individual income and tax data for 2014, the most recent year available. The data contain information on income from taxable dividends. We approximated the stock market participation rate as the ratio of the number of tax returns with dividend income to the number of total tax returns filed.

We used tax return data because of its availability at the state level. However, there is a shortcoming in doing so. Given that not all companies pay dividends out of their profits each year and that the dividend incomes of retirement accounts are not taxable, our figure underestimates the true participation rate and should therefore be considered a lower bound. Fortunately, this shortcoming might be less of a concern when comparing participation rates across states, as the downward bias is likely to affect states uniformly.

Wide Disparities among States

The data show a large variation in stock market participation rates across the United States. The disparities are sizable, with rates ranging from 10.5 percent in Mississippi to 26.6 percent in Connecticut. This seems reasonable, as the average household income is higher in Connecticut than in Mississippi and the existing literature shows that the participation rate increases with income.²

However, this is not the whole story. Even when controlling for household income level, the large variation in participation across states prevails. Figure 1 plots the participation rates for Connecticut, Mississippi and the United States across different income groups. The participation gap remains large for each group, indicating that household income level does not entirely lead to differences in participation rates.