|
Portfolio selections and investment returns play an important role in accumulating wealth. Studies have established that many workers need to allocate their assets more wisely, particularly participants in defined contribution (DC) retirement plans. Measures are being proposed or are under way to encourage better informed, smarter and more effective investment. For example, on the regulatory side, the Department of Labor has encouraged a greater use of equities as part of qualified default investment alternatives in retirement accounts. The effects of these measures and policies on individuals’ investment behavior remain to be seen.
Asset allocation patterns in DC accounts as reported by the 2007 Survey of Consumer Finances (SCF) provide a valuable reference. This analysis uses regression techniques to identify factors that are closely correlated with investors’ portfolio choices. Considering these salient factors should enable plan sponsors and policymakers to choose financial education tools, policies and default programs to better guide and improve DC plan participants’ asset allocations.
Regression analysis: identifying influences on asset allocations
This analysis considers households whose total DC wealth is greater than zero. Individual retirement accounts (IRAs) are ignored. An econometric regression is most enlightening because it disentangles the influences on portfolio choices, simultaneously controlling for a host of factors. The regression runs the dependent variable — the value-weighted equity percentage combining all DC accounts — on factors that potentially explain variations in equity exposure. Since the percentage is between 0 and 100 (i.e., censored at the two ends), a Tobit regression specification is appropriate to gauge the coefficients and signs on the explanatory variables.1 Figure 1 reports the results.
Equity investing exhibits a hump shape in age: The equity percentage rises with age, but the increase slows or even reverses at later ages, as indicated by the significantly positive coefficient on age and negative coefficient on age squared. For instance, equity exposure increases by roughly 0.25 percentage points when an age-30 investor gets one year older but decreases by 0.59 percentage points between ages 60 and 61.
The wealthy tend to invest more in equity, but higher income is associated with lower equity exposure, other things being equal. When net worth, including retirement and other financial wealth but excluding Social Security, increases by 1 percent (for instance, from $600,000 to $606,000), equity exposure increases by about 0.02 percentage points, on average. When household income increases by 1 percent (for instance, from $100,000 to $101,000), equity exposure decreases by about 0.03 percentage points.
The composition of retirement wealth, i.e., defined benefit (DB) wealth as a percentage of DB and DC wealth, does not seem to significantly affect equity allocations. This finding is both somewhat surprising and plausible.2 On the one hand, the generally secure DB benefit should accommodate greater equity investment. On the other hand, countervailing factors could come into play: Employers sponsoring DB plans might attract and retain more risk-averse workers; and some workers fail to accurately assess the nature and level of their DB benefits and thus do not realize the optimal scope for equity. Also, very generous DB benefits (achieving near-100 percent replacement of preretirement income) might eliminate the need to save in DC accounts or take risk.
DC plan participants tend to hold more equities in nonretirement assets than in retirement wealth, although both types of accounts move in the same direction, as indicated by the significantly positive, and less than one, coefficient on the equity share of nonretirement assets. That is, a 1 percentage point increase in equity in nonretirement accounts is associated with an increase of about 0.1 percentage points in DC accounts. This finding suggests that investors are perhaps exploring the tax arbitrage opportunities that arise from the tax rate differentials between equity and bond returns. Theory suggests that the optimal locations are tax-deferred DC accounts for higher-tax bonds and regular taxable accounts for lower-tax equities.3
Equity holdings in retirement accounts are strongly correlated with the presence of company stock, probably because some plan sponsors provide company stock as matching or nonmatching contributions. Inertia and/or limited recognition of the value of asset diversification may lead to sustained concentrations of employer stock. Thus the default funds selected by plan sponsors play a major role in forming participants’ retirement portfolios. For instance, risk-averse investors who default into relatively aggressive life-cycle or balanced funds — or risk-tolerant investors into a conservative fund — might not fully assess their investment portfolios, inadvertently taking more risks than desirable or forgoing opportunities for wealth growth.
Offering investment choices — or enhancing workers’ awareness of their choices with better communications — is more likely associated with significant increases in equity investment. For instance, equity holdings are on average 14 percentage points higher in households with full choice in all DC plans than in those with no choices. Equity exposure is nearly 8 percentage points higher for participants with limited choices in some plans.
Higher education, likely as a proxy for financial literacy, generally means a larger allocation to equities. College graduates hold roughly 5.7 percentage points more equities in their retirement accounts than do high school graduates. This finding suggests that if participants are making unduly conservative investment choices, plan sponsors might want to strengthen financial education to raise participants’ knowledge and comfort in equity investing. To be effective, the education tools need to be easy to understand and carefully designed and targeted.
Risk tolerance is one of the most prominent factors linked to equity allocations. Households willing to take on more risks in the hopes of greater returns invest approximately 15-25 percentage points more in equities than their more cautious counterparts.
Households whose financial planning horizons extend five-plus years are more likely to invest in stocks, despite the puzzling phenomenon that those planning for the “next few months” hold more equities. Longer-term investing probably enables investors to outlast short-term volatilities of equity returns.
The regression controls for several other factors. Households of public-sector workers tend to invest more in equities than private-sector households. This conforms to the perception that public-sector wages are more secure, comparable to a bond (the human capital theory), and therefore can accommodate more equities.
Union members, however, are less inclined to equities. This might reflect more risk-averse workers’ being attracted to unions, in part by the greater sense of job security. Self-employed workers invest more aggressively in equities, which is presumably linked to their entrepreneurial spirit.
Figure 1
Influences on asset allocations by Tobit regression analysis
Dependent variable: equity share in DC accounts
Notes: The sample includes households with at least one member working. IRAs are excluded. *, ** and *** indicate significance of 10%, 5% and 1%, respectively; insignificant otherwise.
Source: Authors’ regression analysis based on Survey of Consumer Finances 2007.
Conclusions
This regression analysis empirically identifies influences on household asset allocations in retirement accounts. Among the statistically significant factors, the equity share in DC accounts is positively associated with household net worth, equity share in nonretirement savings, company stock holdings, investment choices, education level, risk tolerance and financial planning horizon. These insights should help plan sponsors and policymakers design plans and programs to help DC plan participants make more effective investment choices.
The financial crisis has likely deepened workers’ comprehension of the risk-reward tradeoffs in the stock market. This may effectively mobilize some workers to more carefully assess their economic situations and build up portfolios that reflect their preferences and financial needs. Informed decisions by workers themselves would be the best way to improve their asset allocations. But plan designs and employer-sponsored programs can significantly influence participants’ asset allocations and thus their wealth trajectories. In particular, financial education, long-term planning, employee investment choices and default equity exposures in life-cycle or balanced funds will have lasting effects, given the increasing prevalence of DC plans.
1 An ordinary least square (OLS) regression would otherwise assume that the equity percentage could fall outside of the 0-100 percent range and thus yield biased estimates of the coefficients.
2 Including a dummy variable for DB coverage yields similar results.
3 See Daniel Bergstresser and James Poterba, 2004, “Asset Allocation and Asset Location: Household Evidence From the Survey of Consumer Finances,” Journal of Public Economics, 88 (9-10): pp. 1893-1915; and Robert Dammon, Chester Spatt and Harold Zhang, 2004, “Optimal Asset Location and Allocation With Taxable and Tax-Deferred Investing,” Journal of Finance, LIX (3), pp. 999-1037.
September 2009
|
Back
Issues 
