Do Defined Benefit Concepts|
Make Sense for Defined Contribution Plans?
by Judy Markland
President, Landmark Strategies
|Stable Value Investment Association Monograph #27, June 1997|
Institutional asset allocation theories have been systematically applied to the defined contribution world where individuals, not institutions, make investment decisions. This paper examines the investment need differences between defined contribution and defined benefit plans and discusses individuals' attitudes towards risk and reward. The result is to question the utility of traditional asset allocation theory for defined contribution plans.I. Differences in Defined Benefit and Defined Contribution Plan Objectives and RisksThe conventional wisdom is that defined contribution participants invest too conservatively, thus jeopardizing their retirement security. This contention seems to have historically been based upon comparisons of defined benefit and defined contribution asset mixes showing that institutional DB investors are more heavily invested in equities. If the professionals - people who really ought to know what they are doing - invest predominantly in equities for retirement funding, then the theory is that individuals should do the same.
II. Extending Risk Analysis to the Participant's Total Portfolio
III. Inappropriateness of Standard Asset Allocation Risk Measures for DC Investors
IV. Stable Value's Positive Risk/Return Features Benefit DC Investors Only
But how relevant are defined benefit funding patterns to defined contribution investing? A close look at the purposes of the two types of plans and their relative risks illustrates major differences.
Moreover, both the theory and the traditional risk/return measures upon which asset allocation concepts are based were developed for institutional investors. The assumption has been that they also work well for participants in defined contribution plans. But are the traditional risk/return measures really appropriate for individual investors? A look at individuals' behavioral psychology and decision-making under risk indicates major problems in applying the standard asset allocation techniques to individuals.
I. Differences in Defined Benefit and Defined Contribution Plan Objectives and Risks
To understand whether defined benefit funding logic makes sense for defined contribution plans, it is first important to have an understanding of the fundamental investment goals and risk constraints for each of the two plan types to clarify the basic investment needs for each type of plan.
Defined benefit plan sponsors have two basic goals for their plan investment portfolios: to provide competitive, secure retirement benefits for employees and to minimize funding expenses (and thus to maximize asset returns). These are yield maximization objectives that by themselves indicate portfolios invested almost exclusively in common stocks. However, there are also several real-world constraints on defined benefit investment policy. Benefits must be paid when due; significant increases in unfunded liabilities and the consequent strain on the balance sheet must be avoided; and the risk of needing substantial pension inflows when profits and/or cash flow are low should be minimized.
These constraints all argue for reduced price volatility in investments and more overall asset conservatism. The result is generally the "standard" defined benefit asset mix of 60% equities and 40% debt, although differences in individual plan needs obviously mean significant variation from this aggregate total on a plan-by-plan basis.
DC investors have secondary objectives for their funds and large contingency risks.
Defined contribution investors’ primary investment funding goals are virtually identical to those for defined benefit plans: provision of retirement security with maximum investment returns. However, unlike defined benefit investors, defined contribution participants often have secondary objectives for their plan balances like home purchase or financing college education in addition to the basic goal of retirement security. Figure 1 illustrates the results of a 1992 Gallup survey, which showed that, in addition to accumulating funds for retirement, fully 38% of participants in the 25-34 age group also intend to use their plan assets to for educational spending and 28% for a home purchase. Older participants give a greater weight to contingencies, such as medical emergencies. These secondary uses all imply a shorter time horizon for investment results, indicating a need for a more conservative investment policy.
Above and beyond any such time-horizon constraints, just as with defined benefit plans, there are operational risk constraints on defined contribution funding. With DC there is only one key “real world” constraint on investment policy - the need of the investor to access plan balances when needed without fear of an eroded market value. This need may occur in retirement or before, due to the sizable contingency and secondary intention roles participants hold for plan assets. This ability to access investments with minimal risk of principal loss is clearly more important overall to the DC investor than to his DB counterpart.
Defined contribution plans cannot diversify benefit risk.
These differences in investment needs and risk tolerances are further exacerbated by the two plan types' relative ability to diversify their benefit risks. The DB investor incurs the risk that employees may opt for early retirement, unexpectedly increasing both benefit payment and funding needs. However, because this risk is spread over a large participant base, it is relatively predictable in the aggregate. And of course, the employer retains complete control over the timing and size of any formal early retirement program offered. The ability to diversify benefit risks effectively means that defined benefit plans, once having assured a cash flow stream to fund benefits, need only consider asset risks.
But, while he or she has more control over the actual early retirement decision, there is no way for an individual DC participant to diversify early retirement risk. It's all or nothing at any point in time after early retirement becomes a possibility. This fact alone indicates more conservative DC investment patterns than the DB investor requires, because the undiversifiable benefit risk poses greater DC liquidity needs. The fact that the risk can’t be diversified also indicates that it explicitly needs to be recognized in constructing the optimal investment policy and asset mix.
Company stock funds have increased equity risk.
In 1996 investments in company stock rose to 32.3% of private DC single-employer assets overall, up from 29.4% the previous year and 23% as recently as 1993, according to Institutional Investor’s annual survey. Since less than one-third of 401(k) plans offer a company stock fund according to a RogersCasey survey, the proportion of the portfolio represented by company stock is obviously much higher where it is offered. In a recent IOMA survey of plans with company stock, almost one-third had more than 50% of plan assets invested in the company stock fund.
Because of their lack of diversification, company stock funds have significantly higher risk than other common stock vehicles. A 1995 Merrill Lynch study found that the standard deviation of returns for a single stock in the S&P 500 was approximately double that of the index as a whole. This additional risk is receiving increased attention, but it is still rarely considered when analyzing the overall debt/equity mix for a DC portfolio. A strong case can be made that a more conservative debt/equity mix is justified where a substantial company stock exposure exists.
Return to Top
II. Extending Risk Analysis to the Participant's Total Portfolio
Defined benefit plans are self-contained; defined contribution plans are not independent of and cannot be analyzed separately from the investor's overall financial and physical well-being.
Standard asset allocation theory looks at defined contribution and defined benefit asset mixes in isolation, evidently assuming that all the relevant investment considerations are found within the plans. For defined benefit plans, this approach makes sense; defined benefit plans were created to stand on their own. Risks outside the plan are theoretically not a problem. The intent of DB funding requirements is to insulate the pension plan from the financial health of the corporation, so that retiree benefits are secure. The plan has risks of unexpected asset shortfalls or unexpected benefit payments, but these are internal to the plan.
Conversely, there are primary risks to defined contribution funding that are external to the plan - risks posed by financial or emergencies that might require the participant to tap his DC assets. The biggest single risk is job loss, although health emergencies and other family disasters such as funeral expenses or house fires are also possibilities. Job loss may also pose a threat to continuation of mortgage payments on the family home, likely the family's main financial asset outside the DC plan and a key financial resource at retirement.
Like the benefit risk, these external contingency risks are not diversifiable by the employee. Nor is there any mechanism to predict when they might occur. The risks are substantial for any participant without accessible financial assets in addition to his or her DC assets. Their presence indicates a much more conservative funding pattern than would be necessary if one were to look at the retirement risks alone.
Lack of savings outside the plan.
Far too many participants lack any savings except for their DC assets. The table below illustrates the typical financial profile for families with incomes of $25,000-50,000 in 1995, the income category for the average 401(k) participant according to Department of Labor data. The typical family in this income group had about $12,000 in financial assets, only 13% of their total asset portfolio including home equity, car, etc. These financial holding are small relative to the total amount of debt of this group, and especially to the amount of mortgage debt for those with mortgages.
Source: Survey of Consumer Finances as reported in Jan. 1997 Federal Reserve Bulletin
Selected Financial Assets and Liabilities of Families with Incomes of $25-50,000/Year 1995 data; $ in thousands; median amounts for those with the asset or liability
median amounts for
% of families with
the asset or liability
Transaction accounts $2,000 94.7% CD's 10,000 13.7 Retirement accounts a 10,000 52.6 Total financial assets 12,100 90.8 Total nonfinancial assets b,c 81,500 96.6
Mortgage and home equity debt 46,000 47.3 Installment debt 6,600 54.3 Credit card debt 1,400 56.7 Total Debt c $23,400 75.2 . a including IRAs
b home equity, car, business, etc.
c Totals include items not reported separately.
Roughly half of the families in this income group had retirement accounts (including IRA’s) and the median holding was $10,000 in 1995. The Federal Reserve Bulletin did not report detail on the remaining financial assets of those with retirement accounts, but it is clear from the size of this group and its impact on the totals that retirement accounts must constitute a very large proportion of their financial or “money” assets.
Financial planners generally advocate that individuals accumulate funds to cover contingencies before it is prudent for them to assume investment or market value risk. A standard rule of thumb is a contingency or savings pool equal to at least six months income. For the large number of 401(k) participants without financial assets outside the plan, their plan balances must also serve as the contingency fund. Clearly these individuals need a fund option that combines both principal safety and a good return to serve the dual purpose of protecting against contingencies while accumulating retirement wealth.
Defined benefit coverage.
The explosion in 401(k) plan formation has created a misleading impression that 401(k) plans are the primary source of retirement coverage. In fact, as the figure 2 illustrates, 61% of 401(k) participants in 1993 had other primary retirement plan coverage, typically a DB plan. Although the vast majority of 401(k) plans are the sole retirement plan for the firm, most such plans are very small. Larger firms on balance have several tax qualified plans. Thus while 85% of 401(k) plans were the sole retirement plan offered by the firm in 1993, these plans covered only 39% of the active participants in 401(k) plans. Eighty-eight percent of the primary 401(k) plans had fewer than 100 participants.
Defined benefit pensions increase with the employee's length of service and wages, so that until retirement or termination there is increased coverage and reasonable inflation protection. There are valid concerns today about the lack of DB portability and resulting inadequacies of DB benefits in an economy with significant employee job mobility. However, it is a mistake to ignore the role that DB benefits play in retirement protection for employees who remain with a firm.
The largest asset holding for most US households is home equity, and historically home equity has been the largest source of retirement wealth. According to the Department of Labor’s 1995 Survey of Consumer Finances , 65% of US. households own their primary residence, with a median value of $90,000. The median amount of mortgage and home equity debt in the survey is $51,000, implying a median home equity value of $48,900. This is more than double the value of homeowners’ retirement accounts, which had a median value of $20,000 in the survey.
Although it has less market value risk than the typical common stock investment, investment real estate is classified as an equity investment because it reflects a direct ownership position which is often leveraged with debt. Defined benefit plans’ “standard” 60% equity/ 40% debt asset mix includes real estate in the equity position. However, despite the huge proportion of the balance sheet that it represents for most defined contribution investors, home equity is rarely considered as a part of the asset mix or risk profile for the participant. This is unfortunate because the home has the type of inflation protection typically associated with equity portfolios. Perhaps more critically, home equity is typically heavily leveraged (64 percent of homeowners had mortgage and/or home equity debt according to the 1995 Survey of Consumer Finances), which given the size of the investment relative to the rest of the portfolio has significant implications for the degree of overall portfolio risk and the investors’ need for steady cash flow and/or contingency reserves to support mortgage payments.
Return to Top
III. Inappropriateness of Standard Asset Allocation Risk Measures for DC Investors
Risk/return investment theory was developed for institutional investors. It is only applicable for defined contribution investors if it accurately captures their behavior and attitudes under risk. The analysis above indicated differences in DB and DC risks and in the roles of the respective portfolios. There is also evidence that plan sponsors and participants have substantially different behavioral attitudes about risk and reward that affect their respective investment behavior.
Standard institutional portfolio theory defines risk as the standard deviation of investment returns and generally expects that a riskier asset must provide a greater return over time to induce the investor to incur the additional asset risk. This is illustrated by the capital market line, in which expected returns rise with the expected volatility of return.
There are several important assumptions underlying this construct. The theory assumes perfect investor knowledge and an instantaneous time horizon for trading. Neither is especially relevant for the individual investor. More important, however, are two critical characteristics of institutional risk/return theory: 1) the measure of risk utilized places equal weight upon volatility due to gains and losses; and 2) the expected return is achieved over several market cycles, and the theory assumes that the investor's time horizon is long enough for the returns to be realized.
Behavioral research has shown that neither of these characteristics is applicable to the individual investor. The attached exhibit illustrates several examples of the type of academic research that has been undertaken on individuals' choices involving risk. These and a wide variety of similar examples have led behavioral psychologists to reach several conclusions about individuals' risk tolerance and behavior with direct impact on their investment decision making.
For instance, it is common for a household to maintain a special-purpose saving fund, while simultaneously financing one or more cars at a higher interest rate than the fund is earning.
- Individuals are risk-averse with respect to gains. A small gain with greater certainty is preferred to a hope of more money with greater risk, even when the probabilistic results are greater.
- Individuals are loss averse and will assume risks to avoid potential losses.
- Individuals evaluate choices one at a time and make choices relative to the current reference point rather than the long-term outcome.
- Individuals maintain separate mental accounts and do not combine their outcomes.
Note that there is a definite asymmetry here. Unlike the standard belief that individuals are risk averse, studies show that individuals avoid risk when dealing with gains but are willing to assume it when there are potential losses. The fear of loss is greater than the comparable desire for gain. When graphed, this utility or satisfaction function shows steep declines in the loss sector but relatively shallow increases in the positive return portion.
This asymmetry about risk-taking for gains and losses is not the way risk is measured or evaluated in standard asset allocation theory. Volatility, the standard deviation of returns, assigns equal weights to gains and losses. Volatility is not necessarily a good measure to describe a situation in which investors are loss averse, rather than risk averse.
Moreover, asset allocation’s risk/return theory is very highly dependent upon the assumption of a long time horizon. The fact that individuals make decisions based on their current situation and current point of reference also casts doubt on the validity of this assumption. If the time horizon is not relevant, then both the long-term risk/return assumptions underlying asset allocation theory and even more critically, the assumption of normally distributed bond and stock returns are inappropriate.
It is evident that bond and stock returns are not normally distributed over short periods. Above and beyond the question of individual investment psychology, there is substantial evidence that DC investors have secondary uses for their funds with near-term time horizons. Only if investments are held over the long-term are the expected returns likely to be realized. There is also substantial contingency risk for the DC investor which may be exacerbated by cyclical return patterns.
The concepts of risk asymmetry, decision compartmentalization, and decisions based on the current frame of reference are all different than standard portfolio theory investment choice assumptions. Substantial work needs to be done to incorporate behavioral research about individual choices and risk tolerance into investment theory.
Return to Top
IV. Stable Value’s Positive Risk/Return Features Benefit DC Investors Only
From time to time observers note that institutional investors purchase very few GICs for defined benefit plans. Since the professionals avoid them, many infer that stable value funds must not be a good asset holding for individuals. Those who make this analysis have obviously not thought a great deal about the relative benefits that the asset offers to the two types of plans.
It is true that defined benefit plans hold very few GICs. They constituted only 0.1 percent of defined benefit assets in 1996 according to Institutional Investor, compared to stable value’s 18.7% share of defined contribution assets. (In plans where stable value funds are offered, they averaged 39% of assets according to a 1996 Foster Higgins survey.) There is obviously a major disparity between defined benefit and defined contribution plan investors' attitudes about the vehicle.
This difference is easy to understand when the risk/return characteristics of GICs and other typical stable value assets are compared for the two types of plans.
A stable value asset only offers benefit responsiveness, the ability to fund investment transfers and benefit payments at contract value without any market value risk, to participants in DC plans. It is this ability to withdraw their funds or transfer them at book that gives the DC stable value fund such risk/return superiority over other fixed income assets. This option is not available to defined benefit investors; it is only available to DC participants and uniquely available to them in their employee savings plan.
In fact, on a risk/return basis, defined benefit GICs (other stable value assets are not sold to DB investors) may be somewhat inferior to other fixed income assets because of their relative illiquidity. Contract holders are typically allowed to surrender prior to maturity only at the lesser of book or market value. Consequently, GICs are purchased in a large volume by DB plans only in high interest rate environments, when their call protection and zero-coupon features make them attractive relative to other available fixed income options.
Clearly both the professionals and the DC participants know what they're doing regarding stable value investing. Stable value offers favorable risk/return tradeoffs in defined contribution plans but doesn’t provide the same advantages to institutional investors in the defined benefit world.
Return to Top
In summary, defined contribution plan participants seem to understand their investment needs better than many investment professionals. Participants often require more asset mix conservatism as well as a fund offering good returns and principal safety because
Utilizing traditional risk/return theory for defined contribution asset allocations is suspect. Many participants are investing for a shorter time horizon than assumed by the theory. Behavioral research has shown that individuals are risk-averse for gain but risk-assuming to avoid loss, a very different tolerance than assumed by standard risk measures. They evaluate investment choices one at a time rather than for a portfolio as a whole and make decisions relative to the current reference point rather than the long term. These behavioral findings cast additional doubt on the utility of the assumption that individuals remain invested in a fund long enough to realize its longer-run expected return, which will be achieved only over several market cycles.
- they may have secondary investment objectives with shorter time horizons;
- early retirement risk is not diversifiable in DC plans and poses greater liquidity risk than in DB plans;
- external financial contingencies create large DC liquidity risk both before and after retirement;
- many lack savings outside the plan; and
- those with company stock investments have greater risk than with a diversified equity portfolio.
Return to Landmark Articles and Speeches Return Home