Over the last two decades, most private-sector employers stopped offering traditional defined benefit (DB) plans to newly hired salaried workers, favoring defined contribution (DC) and account-based DB plans instead.1 The shift has been fueled by several factors: a desire to reduce retirement benefit costs (perhaps due to higher compensation and benefit costs elsewhere, especially health care), a more mobile workforce, the simplicity and transparency of account-based designs, government and accounting regulations, market trends, global competition and a desire to reduce financial risk.

The prospective prevalence of DC plans as a primary retirement vehicle transfers responsibility and risk from employers to employees, who now must manage their own contributions, withdrawals, investments and retirement distributions. For employers, the shift carries other risks, such as counter-cyclical workforce trends that may necessitate larger severance payouts, raise benefit costs and reduce mobility within an organization. 

Even if DC plan participants do everything right — start saving early, save sufficient amounts and invest prudently — they remain vulnerable to financial market volatility. Our analysis shows that retirement outcomes can vary widely along with financial market ups and downs, even among workers who practice the same saving and investing behaviors.

This Towers Watson analysis looks at historical returns on stocks, bonds and interest rates on a real basis (adjusted for inflation) to estimate ending account balances and income replacement rates for retiring DC plan participants over the last 100 years. The income replacement rate is the percentage of preretirement income replaced by retirement income. For example, suppose a worker earned an average of $50,000 during the last five years of employment and, after retiring, used the DC account balance to purchase an annuity that paid out $15,000 a year for the life of the worker with no future increases in the annuity amount. The income replacement rate would be 30%.

60/40 stock/bond and life-cycle investment approaches

The study looks at retirement income results for workers who retired from 1916 to 2015. These hypothetical workers started contributing 6% of pay to their DC plans at age 25 and stopped saving when they retired at 65. For example, the first simulated worker started working and contributing in 1876 and retired at the beginning of 1916. We assume their pay increased by 2% over inflation each year.

We project results for two investment approaches. Under the first approach, workers allocated 60% to stocks and 40% to bonds, rebalancing annually to hold this mix steady over their careers. Under the second approach, workers opted for a life-cycle fund (also called a target date fund) throughout their 40-year career. In a life-cycle fund, allocations are automatically adjusted over time to reduce the investment risk as the worker nears retirement. Our analysis assumes the fund allocated 88% of contributions to stock at age 25, with the stock share decreasing to 34% in the final year of employment.2 We also assume a 100-basis-point charge annually for investment and administrative expenses.

Figure 1 shows final account balances at retirement as a multiple of the average of the final five years of earnings for workers retiring between 1916 and 2015. The results illustrate the wide range of investment outcomes over the last century. Under the 60/40 investment approach, workers’ account balances ranged from almost twice (191%) to more than six times (619%) their final average earnings. Under the life-cycle approach, while the overall outcomes were slightly more consistent, the variance was still large, with results ranging from twice (207%) to more than five times (554%) final earnings.

Figure 1. 60/40 DC and life-cycle DC account balances as multiple of pay at retirement, 1916 – 2015
Figure 1. 60/40 DC and life-cycle DC account balances as multiple of pay at retirement, 1916 – 2015

Source: Towers Watson calculations

While both investment strategies produced a wide range of outcomes, life-cycle funds offered a pronounced advantage over the last 10 years. Between 2006 and 2015, a worker with a 60/40 allocation amassed from 363% to 490% of average final earnings, while a worker with a life-cycle fund retired with between 458% and 517% of final earnings. During this tumultuous economic period, the life-cycle approach mitigated workers’ losses from the 2008 stock crash to a much greater extent than the 60/40 approach.

Many retirees worry about the possibility of outliving their retirement savings, and Figure 2 shows how long our hypothetical 65-year-old retirees’ savings would last. We assume retirees started with their ending account balance derived from Figure 1, withdrew 25% of their final average earnings annually3 and kept the remaining balance in a 60/40 stock/bond allocation.

Figure 2. Years until account balance realized at retirement is exhausted, workers retiring in 1916 – 1990
Figure 2. Years until account balance realized at retirement is exhausted, workers retiring in 1916 – 1990

Source: Towers Watson calculations

Figure 2 shows considerable variance, reflecting ups and downs in investment performance over time. Some retirees’ savings ran out in 12 years (at age 77), while others’ savings lasted for 28 years (to age 93). A 65-year-old woman today can expect to live an average 21.6 years into retirement — to age 86.6 — according to the Social Security Administration (SSA), as shown by the orange line in Figure 2. In 64% of outcomes, our hypothetical workers exhausted their account balances before age 86.6. As of 2015, roughly one of four 65-year olds will live past age 90, according to the SSA. In our model, workers’ retirement income from their DC plans lasted beyond age 90 only 16% of the time.

In addition to investment uncertainty, workers also face challenges in managing withdrawals from their accumulations during retirement. One way workers can manage withdrawals while ensuring they don’t outlive their money is by converting their DC account balance at retirement to an annuity, which provides regular income for life. While an annuity provides income security, its purchase introduces another level of variance and complexity due to fluctuations in interest rates, which are used to calculate annuity values. To illustrate this challenge, we assume that retiring workers had identical account balances in all 100 years of simulations4 and converted their savings into immediate life-income annuities, based on a unisex life table and interest rates at the time.5 We also assume constant life expectancy and a 15% load on the annuities for insurer expenses/profit. We then compare the annuity payout with the average of the worker’s final five years of earnings to arrive at income replacement rates. 

Even after controlling for stock and bond performance, retiring workers still faced uncertainty in retirement income due to fluctuations in the interest rates used for annuity conversions (Figure 3). For example, a worker retiring in 1982 — when interest rates were at historical highs — was able to replace 31% of working income. But a worker retiring five years later — after interest rates had declined — could replace only 19% of preretirement earnings, despite spending the same amount. Over the 100-year period, income replacement rates ranged from 12% to 31% of preretirement pay for workers who all started with the same DC account balance at retirement. 

Figure 3. Income replacement rates assuming equal DC account balances at retirement, 1916 – 2015
Figure 3. Income replacement rates assuming equal DC account balances at retirement, 1916 – 2015

Source: Towers Watson calculations

Figure 4 looks at replacement rates based on both historical investment performance and interest rates.6 The lump sums are derived from Figure 1 and then converted to an annuity using the methodology derived from Figure 3. Results are shown for both the 60/40 stock/bond and life-cycle investment approaches.

Figure 4. Income replacement rates reflecting DC investment results and interest rates, 1916 – 2015
Figure 4. Income replacement rates reflecting DC investment results and interest rates, 1916 – 2015

Source: Towers Watson calculations

Income replacement rates achieved by converting DC account balances to annuities over the last 100 years ranged from 6.7% to 39.6% of pay under the 60/40 allocation and from 7.2% to 34.5% under the life-cycle approach. Over the 100-year period, median replacement rates were 21.4% of pay under the 60/40 allocation and 22.4% of pay under the life-cycle approach.

More extreme investment approaches

The analysis so far has focused on common investment approaches that reflect some type of balanced portfolio. A relatively high number of workers, however, are much less diversified. Fifteen percent of workers avoid stocks entirely, while 22% invest everything in equities, according to a recent Towers Watson analysis.7 Avoiding equities entirely might rule out large losses, but it also limits retirement savings growth. On the other side of the spectrum, investing everything in equities has the potential for very high returns and growth, but also poses the risk of major losses, which would be particularly problematic for workers nearing retirement.

Figure 5 shows income replacement rates for workers who invested everything in bonds, those who invested everything in stock and those who invested 60% in stocks and 40% in bonds.

Figure 5. Income replacement rates for DC all-bonds, all-stock and 60/40 investment approaches, 1916 – 2015
Figure 5. Income replacement rates for DC all-bonds, all-stock and 60/40 investment approaches, 1916 – 2015

Source: Towers Watson calculations

Workers who invested everything in stocks realized higher replacement rates compared with the other strategies, but the variance in results over the last 100 years is striking. For example, over the last 20 years — after reaping the rewards of consecutive bull markets in the late 1990s — a worker invested entirely in stocks who retired in 2000 could replace 66.2% of preretirement income. A decade later, however — after two major bear markets — a younger but otherwise identical saver/investor could replace only 23.7% of preretirement income — almost two-thirds less than the older retiree.

Figure 6 illustrates the range of income replacement rates at retirement over the last 100 years under four investment scenarios: 60/40 stock/bond, life cycle, all bonds and all stock. Investing in all bonds produced the most stable results but income replacement rates remained relatively low. All-stock investing resulted in the highest peaks, along with a much broader range of outcomes. The 60/40 and life-cycle strategies offered comparable moderate results, with slightly higher upward potential for life-cycle investing.

Figure 6. Range of income replacement rates for all DC investment approaches, 1916 – 2015
60/40, life cycle, all bonds, all stock
Figure 6. Range of income replacement rates for all DC investment approaches, 1916 – 2015

Source: Towers Watson calculations

Cash balance plans

As stated earlier, most employers offer an account-based retirement program to newly hired salaried workers, either a DC plan alone or a DC plan and a hybrid DB plan. Cash balance plans are the most popular hybrid DB plan today. In a cash balance plan, employers contribute a specified percentage of pay annually, which accumulates at a variable (or fixed) rate of interest. A major difference between a DC plan and a hybrid DB plan is that, in the former, the employee shoulders all the investment risk. In a hybrid plan, the employer assumes the investing and funding risk.

We next estimate ending account balances and income replacement rates from cash balance plans for 65-year-old workers retiring between 1916 and 2015. Our assumptions are similar to those for our DC plan estimates: The worker’s pay grew by 2% a year after inflation, and the employer started providing a 6% pay credit when the worker was 25. Cash balance accounts grow annually by an interest crediting rate. Our analysis uses Shiller’s long-term interest rates plus 30 basis points to derive the interest crediting rate.8 The balance at retirement is then converted to an annuity in a similar fashion to that in our DC model, except that no load is applied for the annuity conversion since the lifetime income option is offered through the plan. Figure 7 compares income replacement rates from a DC plan using a 60/40 allocation, a DC plan using a life-cycle approach and a cash balance plan.

Figure 7. Preretirement income replacement rates for 60/40 DC, life-cycle DC and cash balance plan, 1916 – 2015
Figure 7. Preretirement income replacement rates for 60/40 DC, life-cycle DC and cash balance plan, 1916 – 2015

Source: Towers Watson calculations

Income replacement rates (in real terms) from cash balance plans ranged from 7% to 22% over the last 100 years. While in most years, replacement rates were lower in the cash balance plan than in the DC plan, the outcomes were also more stable and predictable. 

Figure 8 shows the average and standard deviation of results over the last five years, the last 20 years and over the entire 100-year period for a 60/40 DC plan, a life-cycle DC plan and a cash balance plan. The cash balance plan had less variance in outcomes than the DC plans. Over the last 20 years, the standard deviation of results for the life-cycle DC plan fell between the 60/40 DC plan and the cash balance plan, yet yielded the highest average replacement rates.

Figure 8. Mean and standard deviation of preretirement income replacement rates
60/40 DC, life-cycle DC and cash balance plan
Figure 8. Mean and standard deviation of preretirement income replacement rates

Source: Towers Watson calculations

Conclusion

As sponsors continue to concentrate retirement benefits in account-based retirement structures, workers must plan for retirement with less secure retirement income guarantees. In today’s retirement world, workers can follow all the saving and investing guidelines but still come up short (or ahead) of their income goals due to the ups and downs of financial markets and timing. 

To mitigate financial market risk, many companies now offer life-cycle funds in their DC plans. While these efforts are helpful, even best-practice tools cannot shield workers and retirees from market uncertainty. Some employers choose to offer both a hybrid DB plan and a DC plan instead of a DC plan alone to enhance retirement security for their workers.


Endnotes

1. See Towers Watson, “Retirement in Transition for the Fortune 500: 1998 to 2013,” Towers Watson Insider, September 2014.

2. The glide path for our life-cycle fund is derived from the average reported in Poterba, James, Joshua Rauh, Steven Venti and David Wise, “Lifecycle Asset Allocation Strategies and the Distribution of 401(k) Retirement Wealth,” in Developments in the Economics of Aging, David A. Wise (ed.), University of Chicago Press, 2009.

3. This assumption is based on a targeted 75% replacement ratio at retirement with one-third of that met by the DC plan.

4. For this purpose, the median account balance from the 60/40 allocation in Figure 1 was used in all cases.

5. The annuity conversions were calculated using Robert Shiller’s long-term interest rates.

6. All investment returns and interest rates in this analysis were adjusted for inflation.

7. See Towers Watson, “Asset Allocations: How American Workers Are Investing Their Retirement Savings,” Towers Watson Insider, April 2015.

8. Because the 30-year Treasury bond was not issued consistently during the last few decades and not at all in earlier years of our analysis, we used Robert Shiller’s long-term interest rates based mainly on 10-year Treasury bond rates. To derive comparable historical 30-year Treasury rates for this analysis, we examined the relationship between 10- and 30-year rates since 1990 and added the difference of 34 basis points to the Shiller rate in all years.