Contrary to our expectations, 401(k) defined contribution plans in the United States earned a higher gross average return at a lower average operating cost than their defined benefit cousins in 1997.
These surprising findings came from the Cost Effectiveness Measurement database, which now includes asset-related information on 59 401(k) plans, valued in aggregate at $145 billion, as well as on 153 U.S. defined benefit plans, worth in total $1.1 trillion.
In 1997, the estimated average return of the 59 401(k) plans in the sample was 22%, and the average operating cost was 29 basis points. The same averages for the 153 defined benefit plans were 19.2% and 37 basis points, respectively. Thus the average 401(k)'s gross return was 2.8 percentage points higher than its defined benefit counterpart, and its average cost was eight basis points lower.
At first blush, thus, it would seem the defined contribution funds outperformed the defined benefit funds on a cost effectiveness basis in 1997. However, as is often the case, things are not as they at first seem. It turns out the higher average return realization of the defined contribution funds was entirely due to their higher average levels of systematic and specific risk. The lower average operating cost experience turned out to be entirely due to the fact that company stock and GIC-type options -- which have very modest operating costs -- accounted for an average of 50% of total assets for the 401(k) plans in 1997. What about the other 50% of plan assets?
When the other average 50% of defined contribution fund components managed through diversified investment portfolios are compared with their defined benefit counterparts on an asset class by asset class basis, the defined benefit components produced higher average gross returns at lower average fees. Thus, in the fund components that are directly comparable, defined benefit funds outperformed their defined contribution counterparts in cost effectiveness. This is the relationship we had, in fact, expected to find.
These findings have implications for defined contribution plan participants and sponsors, and for the government agencies with oversight responsibilities for these plans.
POLITICAL ECONOMY
Defined contribution plans have become the darlings of the pension world, and nowhere more so than in their 401(k) manifestation in America. In a recent piece, Peter Bernstein observed that in the 1980s, defined benefit plans began to reach out for more risk in order to meet their high return targets.
Then, "in the midst of that process came the 401(k) phenomenon," he noted. ". . . Individuals joined the institutions in succumbing to the inevitability that taking on risk was the only choice. . . . In the long run, everything would come out roses."
Peter Bernstein is not the only thoughtful person who has begun to worry where "this burgeoning momentum" eventually will lead us. Speaking recently about the phenomenal growth of 401(k) plans, Securities and Exchange Commission Chairman Arthur Levitt Jr. saw a "financial literacy crisis," with many defined contribution plan members having minimal knowledge about the nature of financial markets and how they are participating in them. For its part, the Department of Labor has begun to encourage employers to play a more active role in raising that minimal level of knowledge in the context of 401(k) plans.
Indeed, large employers themselves are beginning to realize that just because an increasing number of their employees are now DC rather than (or in addition to) DB participants, they are not off the hook. In the presence of pervasive "informational asymmetry" between the ultimate buyers and sellers of 401(k)-related services, it is the knowledge of the employer that offers the best hope to reduce that asymmetry, and thus get DC plan participants into the right investment program at a reasonable cost. Any employer who does not use its knowledge to that end is on moral and legal thin ice.
AMBIVALENCE TO DATABASE
Efforts by CEM to create a database to measure the cost effectiveness of 401(k) pension plans offer a telling example of how the attitudes of large U.S. employers have changed in the past four years. When CEM surveyed its U.S. corporate clients four years ago to see if there was an interest in creating a defined contribution counterpart to the defined benefit cost effectiveness measurement service, the answer was an almost universal "no." In 1998, the answer was an almost universal "yes."
As CEM prepares individual cost effectiveness studies for the 59 survey participants, it is stressing a very important point. The 59 401(k) plans under study are not a random sample. Quite to the contrary, this sample is likely to exhibit a material "best practice" bias for two reasons. First, these 59 plan sponsors were able to meet the exacting data requirements of the CEM 401(k) survey. Second, they were prepared to have the cost effectiveness of their plans formally evaluated. This best-practice sample bias should be kept in mind as we evaluate and interpret the findings set out in this commentary. Thus, if we find room for improvement in this best-practice sample, there likely will be a great deal more room for improvement in the rest of the 401(k) universe.
DB VS. DC
Good research starts with a prior hypothesis. Research findings are either consistent with, or inconsistent with the "prior." Our "prior" in this study was that the average defined benefit fund is managed more cost effectively than the average defined contribution fund. We expected this to be the case even where an employer sponsors both types of plans. This expectation follows directly from our previous observation about "asymmetrical information."
When a market for any good or service exhibits asymmetrical information between buyers and sellers, a predictable outcome is the result. The buyers will pay too much for too little value, while the sellers will earn too much for too little effort. There is a great deal of evidence that the market for investment management services is such a market. Specifically, most buyers behave as though they believe there is a high association between brand name suppliers and "performance," and that fees are unimportant. Despite the fact this performance belief has no empirical basis, many sellers reinforce it in their marketing strategies and charge fees as though the belief were justified.
Defined contribution participants are mired at the bottom of the financial foodchain because they have the greatest asymmetrical-information problem. Sponsors of defined benefit plans should do better because many understand that it is in the corporate interest to reduce informational asymmetry by employing full-time professional pension fund management teams. Such teams, when properly empowered and compensated, should be able to level the informational playing field, and thus produce higher risk-adjusted returns at lower costs. For understandable organizational reasons, many corporations with defined benefit plans thus far have not taken full advantage of this in-house expertise to ensure their DC arrangements are structured equally cost effectively. To date, the emphasis has been on "giving participants what they want," rather than cost effectiveness. Thus we expected the defined benefit funds to outperform the defined contribution funds in cost effectiveness even in cases where employers sponsor both types of plans.
SURPRISE FINDINGS
Why did the average 401(k) plan in the database have a higher return and lower cost than defined benefit plans? The average 401(k) mix in the CEM database has six percentage points more equity exposure than its DB counterpart (i.e. 28% in company stock plus 44% in diversified stock portfolios). This alone made a material contribution to the 401(k) funds' higher average return, as stocks significantly outperformed fixed-income investments in 1997. Also, the 401(k) equity exposure was virtually all to U.S. stocks, which performed better than foreign stocks, where defined benefit funds have significant equity exposure.
In 401(k) plans, sponsor-company stock exposure has a "roll the dice" effect. Good company stock performance will give participant returns a considerable lift, while poor performance will have the opposite effect.
HIGHER VALUE ADDED
The asset mix policy returns provide important benchmarks against which to evaluate the realized actual fund returns. If the actual returns exceed the benchmark returns, active management is adding value. If not, active management is reducing value.
How good is the investment performance a typical 401(k) plan buys, compared to that of a defined benefit plan? The key finding from the CEM databases is that the average gross value added of each defined contribution mandate generally was lower than its defined benefit counterparts -- even before the higher DC fees are taken into account. The most direct way to assess this performance question is to compare actual fund component performance vs. benchmark performance where the benchmark is an investible portfolio reflecting the same investment goals and risks. The CEM databases show the defined contribution value added against benchmarks generally trailing that of defined benefit plans for the following active mandates (while both DC and DB, as the negative figures show, often underperform benchmarks):
Domestic large cap:
* DC externally managed, -5.1%.
* DB externally managed, -3.1%
* DB internally managed, -2.7%
Domestic small cap:
* DC externally managed, -2.6%.
* DB externally managed, -2.3%
* DB internally managed, +0.2%
Foreign equity:
* DC externally managed, +3.9%.
* DB externally managed, +3.9%
* DB internally managed, +5.2%
Bonds:
* DC externally managed, +0.2%.
* DB externally managed, +0.8%
* DB internally managed, +0.5%
Defined contribution fees are about 50% above defined benefit fees for comparable external active mandates. The DC: DB fee ratio exceeds 2: 1 for comparable passive mandates.
These findings support the concerns being expressed by such organizations as the SEC and the DOL about the need to educate employees about not only return and risk, but also costs. Very few are aware that lower plan costs usually translate directly into higher net returns and higher ultimate pensions. A common rule is that one percentage point of extra long-term return produces a 20% higher pension.
For employers, these findings highlight the importance of applying all of the knowledge and skill corporations employ in managing their defined benefit plans to the defined contribution side as well. There is a moral and legal obligation to offer employees a cost-effective approach to converting retirement savings into a growing pool of pension assets.
In the case of the 59 401(k) plan sponsors in the CEM database, 54 offered a defined benefit plan as an excellent starting point of that process. However, even in their case, the structure of the defined contribution arrangement may deserve more attention. Consider these questions, for example:
* Should there be some maximum limit on company stock exposure to control risk?
* Does the DC plan offer a series of low-cost passive investment options?
* Do employees really understand the importance of controlling plan costs?
* Do we know how the cost effectiveness of our DB and DC pension fund investment programs compare?
* Do we know how the cost effectiveness levels of our pension fund compare against those of our peers?
* Should we not know these cost effectiveness levels from both management and legal due-diligence perspectives?
We invite sponsors, consultants and managers to ponder these questions.
Keith P. Ambachtsheer is principal of K.P.A. Advisory Services Ltd., Toronto.
