Announcement: SOA congratulates the new FSAs for June 2019.

Managing Risks In Defined Contribution Plans

Managing Risks in Defined Contribution Plans

by Raymond J. Murphy

Tips on how to best deal with the risks inherent in defined contribution plans.

Americans will rely on defined contribution plans more than any other retirement program besides Social Security to provide their retirement income. The Bush administration is now proposing to gradually shift Social Security to a program with individual accounts. Defined benefit plans have been on the decline due to employer concerns about funding levels, accounting disclosure and administrative burdens.

If there is no turning back from this movement to DC plans, the question is then how to best manage the investment and longevity risks inherent in a DC design. This article will present some unusual design elements that can help control these risks and potentially lower the cost.

In most DC plan designs, employees have the ultimate control in deciding how much to save and how to invest their contributions. Accordingly, they also bear the risk of poor investment performance and the risk of outliving their funds. If employees give up some of this control, I would argue that some of the risk can be reduced.

Description of Model
A simplified stochastic model will be used to illustrate the investment and longevity risk. This model can be run with many different assumptions and scenarios, but for the purpose of this paper, only the scenario in Table 1 will be presented.

The contributions and investment returns will accumulate for 40 years, and then the initial withdrawal, equal to the replacement ratio, will be made in the retirement year and indexed for inflation thereafter.

The stochastic model will be run 1,000 times for each variation, and the key output results will be the age that the account is exhausted and the ratio of the account balance to final salary at the retirement age.

The replacement ratio of 50 percent was chosen to represent the portion funded by employer and employee contributions. Studies suggest that retirees will need between 75 percent and 90 percent of final pay to maintain their standard of living. It is assumed that the balance above 50 percent will be provided by Social Security.

The contribution rate is based on a hypothetical 401(k) plan with the employee contributing 5 percent of pay, a 50 percent match and a profit sharing contribution equal to 5 percent of pay.

Modeling Asset Allocation
Participants in 401(k) plans are often poor investment managers of their own accounts. Some are too conservative, even with many years to go before retirement, and others take too much risk with significant equity exposure even after retirement.

The results for these two extremes are very different. The tables show the 90th, 50th and 10th percentile results of the stochastic model.

It is difficult to define a "successful" plan. For purposes of this analysis, I will define a "success" if the 10th percentile age of exhaustion is at least 85 and the 50th percentile age of exhaustion is 100 or more. Life expectancy at age 65 is approximately to age 81 for males and age 84 for females (See Table 2).

These results show that this asset allocation and contribution strategy would not be sufficient to cover the risk of living to old age since there is a high probability of outliving the retirement assets.

Financial advisors recommend that new retirees limit their initial withdrawal to 4 percent or 5 percent of assets to ensure that the account will last through retirement. However, even the 50th percentile result would have an initial withdrawal equal to 6.7 percent of assets (50 percent replacement income divided by 7.5 account ratio).

If we assume 100 percent invested in the stock market at all times, the results improve, but no one would recommend full equity exposure to a retiree in their 70s or older. (See Table 3)

This scenario shows wild variations. On the plus side, the retiree is rich if the market performs well. In this case, the account at retirement would be well over 15 times final pay which should be more than enough to meet their retirement income needs. However, there is downside risk and the retiree could be out of funds by age 86.

Most experts recommend high equity allocation early in a saver?s career and gradually allocate more to fixed income as the individual approaches retirement. This scenario starts at 100 percent equity at age 25, gradually moves down to 40 percent equity at age 60 and is 100 percent invested in fixed income at age 80. (See Table 4)

This allocation produces a similar 10th percentile result to the 100 percent equity scenario, but has less variation, since the balance ratio range is 7.6 to 22.1. It has basically achieved the same income security result with substantially less risk. The median initial withdrawal rate is 4 percent (50 percent divided by 12.4).

Please note that this is not intended to be an optimal allocation. It is only included to show how an age-based allocation produces a favorable income result with lower risk.

Professional Investment Management
Some 401(k) plan sponsors are offering professionally managed accounts or life cycle accounts to employees who do not want to make investment decisions on their own. The investment strategy for these funds follows an age-based allocation similar to the scenario described above. Of course, employees can either accept this management or continue to do it themselves.

One suggestion would be to require that employer-paid contributions be invested in a professionally managed or life-cycle fund. This would take the control away from the employee, but would eliminate some of the investment and longevity risk borne by the employee.

Mandatory Annuities
Another radical suggestion would be to require that the employer-paid balance be used to purchase an annuity. Again, this would take the control away from the employee and they would forego any potential positive investment performance, but they would secure their retirement income needs for that portion of their account. The feasibility of this strategy would depend on the annuity marketplace.

Both of these suggestions would not be popular with many employees because they enjoy the freedom to make investment decisions. Employee communications would need to focus on the security aspects rather than the prospect of wealth creation.

Variable Employer Contributions
Even with the age-based asset allocation, there is a good likelihood that the account will have far exceeded the balance needed to provide a secure income. An argument could be made that a balance ratio above 12.4 exceeds our target, in which case too much has been saved. In this way, DC plans are not cost-effective.

One novel approach would be to vary the employer contribution based on investment performance. If the investment performance has been very strong, the employer could cut back on contributions. If the performance has been poor, the contributions could be raised to help the individual catch up.

This could be designed many ways. For this model, I?ll assume that the 5 percent profit sharing contribution has been changed to a fixed contribution of 2.5 percent with a variable contribution based on the three-year average performance of the stock market index. The 401(k) design (5 percent employee with 2.5 percent match) would be unchanged. The new total contribution schedule would be: (See Table 5 )

This scenario produces some interesting results. The average contribution rate is a new output variable. (See Table 6)

The results are very similar to the scenario with a fixed 12.5 percent contribution. However, the median contribution is 12 percent and the probable range is from 11.5 percent to 12.5 percent. In other words, this achieves roughly the same income objective at a lower cost.

Early Retirement
It is very difficult to accumulate enough retirement assets to fund early retirement through a DC design. Employees who want to retire early must either hope for superior investment returns or increase their savings rate significantly. In early retirement, there are fewer years in the accumulation phase and more years in retirement. In addition, Social Security is not available until age 62, so the retiree would need to cover all income needs from the account prior to age 62.

The model has been modified for early retirement to fund 80 percent of pay prior to age 65 and 50 percent thereafter. Using the model with the same age-based allocation indicates that the median total contribution rate would need to be increased from 12.5 percent to 19.0 percent to generate similar results for retirement at age 60, and the median balance ratio would need to increase from 12.4 at 65 to 15.6 at 60. Retirement at age 55 would require a median contribution rate of 27.5 percent and a median balance ratio of 17.4. Employers would not be willing to accept this additional cost, so employees would have to fund the difference on their own.

This analysis indicates that a contribution of 12 percent to 13 percent of pay for 40 years with prudent investment management should be sufficient to meet the income needs for the majority of future retirees at age 65. The cost to fund an earlier retirement age is prohibitive in a DC design.

The risk control mechanisms discussed here, namely professional investment management, mandatory annuitization and employer contributions based on investment performance are regular features of defined benefit plans. These strategies shift risk from employee back to the employer, although the employee would need to forego some of the control they enjoy today.

If Social Security eventually adopts a defined contribution design, I would hope that some of these rules would be incorporated to help promote benefit security, particularly mandatory professional investment and annuitization. If not, then many Americans will mismanage their retirement both before and after retirement.

Raymond J. Murphy, FSA, is director, Pension & Savings Plans for Bristol-Myers Squibb Company. He can be contacted at: