What’s a reasonable withdrawal rate when living off savings? A look at how various withdrawal rates would have fared based on past market returns provides a useful guide.

Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable

By Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz

Most investors who plan for retirement eventually confront the question of how much money they should plan to withdraw annually from their investment portfolio. The dilemma is that if they withdraw too much, they prematurely exhaust the portfolio, but if they withdraw too little, they unnecessarily lower their standard of living.

Financial planners, counselors, analysts, and writers stand ready to advise investors on their dilemma, but their advice varies greatly, ranging from investing in common stocks and spending the dividend yield (roughly 3%), up to 7%, which allows for the invasion of principal. Highly risk-averse investors would likely gravitate toward the low end of the range because of their concerns about outliving their portfolio. Moreover, the larger the percentage of a retiree’s total income provided by the portfolio, the more risk-averse the retiree is likely to be. In addition, some retirees wish to bequeath a large estate to their heirs, which again argues for a low withdrawal rate. In contrast, an aggressive investor without heirs might wish to plan a financial future based on a high withdrawal rate. Because of these highly personal behavioral traits, circumstances, and goals, no single withdrawal rate appears appropriate for every investor.

What, then, can be done to help an investor in planning for a withdrawal rate? The word planning is emphasized because of the great uncertainties in the stock and bond markets. Mid-course corrections likely will be required, with the actual dollar amounts withdrawn adjusted downward or upward relative to the plan. The investor needs to keep in mind that selection of a withdrawal rate is not a matter of contract but rather a matter of planning. Thus, the question addressed here is: What is a reasonable withdrawal rate from a portfolio for purposes of planning retirement income? Or stated differently, what withdrawal rate is likely to be sustainable during a specified number of years?

To help in the selection of a withdrawal rate, the following sections provide information on the historical success of various withdrawal rates from portfolios of stocks and bonds. If a withdrawal rate proves too high based on historical year-to-year returns, then it seems likely that the rate will not be sustainable during future periods. Conversely, historically sustainable withdrawal rates are more likely to have a high probability of success in the future.

Using Historical Experience as a Guide

One approach to examining withdrawal rates is based on present value analysis and historical average rates of return. For example, if a portfolio earns 3.7% per year, the historical average return on U.S. Treasury bills, withdrawals of 6% per year can be maintained for about 26 years before exhausting the portfolio. For a $1 million portfolio, that works out to an annual income of $60,000 for 26 years. Similar exercises can be conducted for portfolios of large-company common stocks and long-term corporate bonds, which have produced annual compound rates of roughly 10.5% and 5.7%, respectively, during the period 1926 to 1995.

This analytical approach provides useful insights, but it ignores the critical short-term variations in rates of return. For an investor withdrawing assets from a portfolio, these short-term variations can have an impact on the ultimate outcome that is not reflected using long-term averages. This impact is especially significant for portfolios of common stocks, since their returns are highly variable.

An alternative approach to understanding withdrawal rates is to examine historical year-to-year experience. A sustainable withdrawal rate (as a percentage of initial portfolio value) is one that does not exhaust a portfolio of stocks and bonds despite the annual dollar withdrawals during a specified number of years (the payout period). The portfolio success rate, a useful concept for identifying sustainable withdrawal rates, is measured by the percentage of all past payout periods supported by the portfolio despite annual withdrawals. Presumably, a withdrawal rate that has worked well over the long-term past is likely to work well in the future.

Our study measured the impact of withdrawal rates on portfolio values using the following approach:

Table 1. Portfolio Success Rates: 1926 to 1995 (Percentage of all past payout periods supported by the portfolio)
Withdrawal Rate as a % of Initial Portfolio Value:
Payout Period 3% 4% 5% 6% 7% 8% 9% 10% 11% 12%
100% Stocks
15 Years 100 100 98 98 93 91 88 77 63 55
20 Years 100 98 96 94 92 84 73 61 47 43
25 Years 100 98 96 91 87 78 70 50 43 35
30 Years 100 98 95 90 85 78 68 54 49 34
75% Stocks/25% Bonds
15 Years 100 100 100 100 96 95 91 79 63 46
20 Years 100 100 100 96 94 88 71 51 41 33
25 Years 100 100 98 96 91 78 57 46 33 26
30 Years 100 100 98 95 88 73 54 46 37 24
50% Stocks/50% Bonds
15 Years 100 100 100 100 100 98 91 71 50 36
20 Years 100 100 100 100 96 88 61 41 25 10
25 Years 100 100 100 98 96 70 43 22 7 0
30 Years 100 100 100 98 90 51 37 15 0 0
25% Stocks/75% Bonds
15 Years 100 100 100 100 100 100 91 50 21 14
20 Years 100 100 100 100 100 71 24 12 4 2
25 Years 100 100 100 100 78 22 9 0 0 0
30 Years 100 100 100 100 32 5 0 0 0 0
100% Bonds
15 Years 100 100 100 100 100 79 43 38 14 7
20 Years 100 100 100 96 47 35 16 6 0 0
25 Years 100 100 98 52 26 7 2 0 0 0
30 Years 100 100 51 27 0 0 0 0 0 0
Note: Numbers rounded to the nearest whole percentage. The number of overlapping 15-year payout periods from 1946 to 1995, inclusively, is 36; 20-year periods, 31; 25-year periods, 26; 30-year periods, 21. Stocks are represented by Standard and Poor's 500 index, and bonds are represented by long-term, high-grade corporates.

Data source: Authors' caluculations based on data from Ibbotson Associates.

Portfolio Success Rate

The portfolio success rate responds to the variously expressed problem of an investor running out of money during the retirement years. If an investor’s portfolio outlives the investor’s planned payout period, then it is counted a success.

Table 1 presents 200 portfolio success rates resulting from different combinations of 10 withdrawal rates, five portfolio allocations, and four payout periods, all based on annual stock and bond returns from 1926 to 1995. The first entry in the table indicates that a 100% stock portfolio supported 100% of all 15-year periods in which annual withdrawals were made based on an initial withdrawal of 3% of portfolio value. The portfolio success rate drops to 98% for a 5% initial withdrawal rate, reflecting the failure of the all-stock portfolio during one of 56 15-year periods (1929 to 1943). Not surprisingly, as the withdrawal rate rises, the portfolio success rate declines.

Continuing with the all-stock portfolio and holding the withdrawal rate constant shows that portfolio success rate usually declines with increases in the length of the payout period—also not too surprising. Because the portfolio success rate declines with increases in withdrawal rates and usually with increases in payout period, the numbers in Table 1 for the all-stock portfolio generally decline proceeding from the upper-left corner to the lower-right corner. The numbers imply that young retirees who anticipate long payout periods should plan on lower withdrawal rates than their older counterparts.

Table 1 also shows the impact of asset allocation on portfolio success rates: there is a general decline in portfolio success rates caused by increases in the percentage of bonds. In contrast to stocks, bonds provide little upside potential, which causes the portfolio success rate to be small or even zero for bond-dominated portfolios at high withdrawal rates. Because of the benefits of diversification, however, the presence of some bonds in the portfolio increases the portfolio success rate for low to mid-level withdrawal rates. For example, for withdrawal rates of 7% and lower, the 50% stock/50% bond portfolio has higher success rates than the portfolios with greater stock allocations for all payout periods.

If history is any guide for the future, then withdrawal rates of 3% and 4% are extremely unlikely to exhaust any portfolio of stocks and bonds during any of the payout periods shown in Table 1. In those cases, portfolio success seems close to being assured.

For planning purposes, where should an investor draw the line between acceptable and unacceptable portfolio success rates? The answer will vary from investor to investor, but it seems clear that some investors will choose withdrawal rates exceeding the highly conservative 3% and 4% rates.

Table 2. Portfolio Success Rates: 1946 to 1995 (Percentage of all past payout periods supported by the portfolio)
Withdrawal Rate as a % of Initial Portfolio Value:
Payout Period 3% 4% 5% 6% 7% 8% 9% 10% 11% 12%
100% Stocks
15 Years 100 100 100 100 100 100 97 86 69 64
20 Years 100 100 100 100 100 97 81 61 45 42
25 Years 100 100 100 100 100 88 77 46 42 38
30 Years 100 100 100 100 100 90 76 52 52 38
75% Stocks/25% Bonds
15 Years 100 100 100 100 100 100 100 86 69 53
20 Years 100 100 100 100 100 97 77 48 42 32
25 Years 100 100 100 100 100 85 54 42 31 27
30 Years 100 100 100 100 100 81 52 48 38 29
50% Stocks/50% Bonds
15 Years 100 100 100 100 100 100 94 78 56 42
20 Years 100 100 100 100 100 94 61 39 26 13
25 Years 100 100 100 100 100 69 38 19 4 0
30 Years 100 100 100 100 100 48 33 10 0 0
25% Stocks/75% Bonds
15 Years 100 100 100 100 100 100 89 53 25 17
20 Years 100 100 100 100 100 68 23 13 6 3
25 Years 100 100 100 100 73 15 8 0 0 0
30 Years 100 100 100 100 19 0 0 0 0 0
100% Bonds
15 Years 100 100 100 100 100 72 39 33 19 11
20 Years 100 100 100 94 42 29 23 10 0 0
25 Years 100 100 96 54 15 12 4 0 0 0
30 Years 100 100 48 10 0 0 0 0 0 0
Note: Numbers rounded to the nearest whole percentage. The number of overlapping 15-year payout periods from 1946 to 1995, inclusively, is 36; 20-year periods, 31; 25-year periods, 26; 30-year periods, 21. Stocks are represented by Standard and Poor’s 500 index, and bonds are represented by long-term, high-grade corporates.

Data source: Authors’ calculations based on data from Ibbotson Associates.

The Most Recent 50 Years

The portfolio success rates in the preceding section are derived from 70 years of capital market returns generated from 1926 to 1995. The most recent 50 years, frequently described as the post-war period, includes the years 1946 to 1995. Excluding the 20 years from 1926 to 1945 reveals the impact on portfolio success of excluding capital market returns generated during the Great Depression and World War II.

Table 2 presents portfolio success rates based on the methodology used in Table 1 but with the period of analysis limited to 1946 to 1995. In contrast to the 70-year period, the post-war period generally produces higher success rates for portfolios comprising at least 50% stocks. Bond-dominated portfolios, however, show little or no improvement during the post-war period.

If the most recent 50 years of capital market returns are indicative of the future, then investors with stock-dominated portfolios may be quite aggressive in planning withdrawal rates. For a 15-year payout period, withdrawal rates of 8% or 9% appear reasonably sustainable. Many investors, however, require payout periods of 20 years or longer. In those cases, sustainable withdrawal rates fall to the 7% to 8% level.

Whether portfolio success rates during the most recent 50 years are more relevant than those during the 70-year period is debatable. Restricting the analysis to the most recent 50 years excludes not only the bear market of the 1930s, but also the bull markets of the late 1920s and the early 1940s. The longer period provides a larger distribution of returns, which beneficially represents more possible states of the market. On the other hand, some of the economic conditions prevalent in the 1920s and 1930s bear little resemblance to today or the future. Whether Table 1 or Table 2 is more representative of the future is unknown, but both tables provide a richer view of past experience and perhaps future experience as well.

What About Inflation?

One big risk faced by individuals living off their portfolios is inflation. For example, an investor who plans to withdraw $70,000 per year from a $1 million portfolio of stocks and bonds (a 7% withdrawal rate) is likely to experience a decline in purchasing power; if inflation averages 3% per year, then the purchasing power of the $70,000 will be cut in half by the end of 25 years.

One way to plan for the impact of inflation is to adopt a withdrawal rate smaller than the rate of return on the portfolio; that allows the portfolio value to grow annually. If the withdrawal rate is then applied to the growing portfolio value, the annual amount withdrawn will increase.

The formula to determine this assumes a constant rate of return, which produces a constant growth rate for a given retention rate. But the rate of return on a portfolio of stocks and bonds varies substantially each year. Thus, while the formula may be useful on average, it may produce grossly misleading results in many instances.

A richer understanding of sustainable withdrawal rates in the face of inflation can be obtained by analyzing past rates of return and inflation rates. To counteract the effect of inflation, the dollar withdrawal in a given year must be increased by the inflation rate for that year. Similarly, to counteract the effect of deflation (as occurred in 10 of the past 70 years, especially frequent from 1926 to 1932), the dollar withdrawal in a given year must be decreased by the deflation rate for that year. Thus, portfolio value changes from year to year according to market return; withdrawals change from year to year according to the inflation/deflation rate, which maintains purchasing power of the withdrawals.

Table 3. Inflation-Adjusted Portfolio Succes Rates: 1926 to 1995 (Percentage of all past payout periods supported by the portfolio after adjusting withdrawals for inflation)
Withdrawal Rate as a % of Initial Portfolio Value:
Payout Period 3% 4% 5% 6% 7% 8% 9% 10% 11% 12%
100% Stocks
15 Years 100 100 100 91 79 70 63 55 43 34
20 Years 100 100 88 75 63 53 43 33 29 24
25 Years 100 100 87 70 59 46 35 30 26 20
30 Years 100 95 85 68 59 41 34 34 27 15
75% Stocks/25% Bonds
15 Years 100 100 100 95 82 68 64 46 36 27
20 Years 100 100 90 75 61 51 37 27 20 12
25 Years 100 100 85 65 50 37 30 22 7 2
30 Years 100 98 83 68 49 34 22 7 2 0
50% Stocks/50% Bonds
15 Years 100 100 100 93 79 64 50 32 23 13
20 Years 100 100 90 75 55 33 22 10 0 0
25 Years 100 100 80 57 37 20 7 0 0 0
30 Years 100 95 76 51 17 5 0 0 0 0
25% Stocks/75% Bonds
15 Years 100 100 100 89 70 50 32 18 13 7
20 Years 100 100 82 47 31 16 8 4 0 0
25 Years 100 93 48 24 15 4 2 0 0 0
30 Years 100 71 27 20 5 0 0 0 0 0
100% Bonds
15 Years 100 100 100 71 39 21 18 16 14 9
20 Years 100 90 47 20 14 12 10 2 0 0
25 Years 100 46 17 15 11 2 0 0 0 0
30 Years 80 20 17 12 0 0 0 0 0 0
Note: Numbers rounded to the nearest whole percentage. The number of overlapping 15-year payout periods from 1926 to 1995, inclusively, is 56; 20-year periods, 51; 25-year periods, 46; 30-year periods, 41. Stocks are represented by Standard and Poor’s 500 index, and bonds are represented by long-term, high-grade corporates, and inflation (deflation) rates are based on the Consumer Price Index (CPI).

Data source: Authors’ calculations based on data from Ibbotson Associates.

Table 3 presents portfolio success rates based on the methodology used in Table 1 but with the addition of withdrawals adjusted for inflation and deflation. Immediately noticeable is the dramatic decline in many of the portfolio success rates, especially for mid-level and high withdrawal rates. Despite the adjustment, however, withdrawal rates of 3% to 4% continue to produce high portfolio success rates for stock-dominated portfolios. Even the 5% withdrawal rate produces reasonably high portfolio success rates for all payout periods, but the 6% and 7% rates perform reasonably well only for short payout periods. All withdrawal rates above 7% perform poorly for all payout periods.

Adjusting withdrawals for inflation substantially reduces near-term withdrawals in favor of much larger ones in later years. Whether such adjustments are justifiable depends on investor preferences. Each investor must judge individually which of the possible patterns of consumption produces the most benefit. Because of health considerations, some investors might prefer a consumption pattern tilted toward the early years of retirement. Others might derive more utility from the increased financial security that postponed consumption produces.

A second issue revolves around the inflation/deflation calculation itself. Table 3 presents portfolio success rates that reflect withdrawals adjusted for changes in the Consumer Price Index (CPI). Many economists believe, however, that inflation as measured by the CPI overstates the actual increase in cost of living by 1.0 to 1.5 percentage points per year. If so, then the portfolio success rates in Table 3 are biased downward, especially those for the longer payout periods. Planning for CPI-adjusted withdrawals places great demands on the portfolio and requires the investor to reduce the withdrawal rate, perhaps more than necessary. As a result, the investor may forgo more current consumption for future consumption than is necessary to maintain a given standard of living.

Table 4. Terminal Value of a $1,000 Initial Portfolio After All Annual Withdrawals: 1926 to 1995
Payout Period 75% Stocks/25% Bonds Withdrawal Rate*: 50% Stocks/50% Bonds Withdrawal Rate*: 25% Stocks/75% Bonds Withdrawal Rate*:
4% 5% 6% 7% 4% 5% 6% 7% 4% 5% 6% 7%
15 Years
Average $2,964 $2,631 $2,297 $1,970 $2,285 $1,992 $1,698 $1,405 $1,755 $1,496 $1,236 $977
Minimum 493 249 5 0 855 615 375 135 969 756 542 327
Median 2,727 2,328 1,909 1,543 2,086 1,770 1,472 1,175 1,422 1,198 951 727
Maximum 6,417 5,919 5,421 4,923 5,554 5,103 4,652 4,202 5,321 4,898 4,474 4,051
20 Years
Average 4,239 3,628 3,026 2,435 2,954 2,449 1,944 1,443 2,026 1,606 1,185 765
Minimum 536 108 0 0 975 587 199 0 1,019 744 451 110
Median 4,481 3,752 2,914 2,076 2,755 2,291 1,798 1,309 1,505 1,164 824 502
Maximum 9,484 8,672 7,859 7,047 7,512 6,769 6,025 5,282 5,965 5,168 4,422 3,746
25 Years
Average 6,031 4,995 3,991 3,016 3,815 3,007 2,199 1,416 2,307 1,672 1,036 424
Minimum 785 0 0 0 1,340 655 0 0 1,203 736 269 0
Median 5,574 4,483 3,710 2,636 3,568 2,706 2,058 1,381 1,850 1,325 787 200
Maximum 11,534 10,418 9,301 8,185 8,109 6,624 5,138 3,652 6,795 5,492 4,188 2,997
30 Years
Average 9,031 7,367 5,779 4,262 5,171 3,936 2,712 1,553 2,645 1,724 803 122
Minimum 1,497 0 0 0 2,151 870 0 0 1,428 729 29 0
Median 8,515 6,868 5,586 3,745 5,171 4,041 2,610 1,251 2,245 1,481 806 0
Maximum 16,893 14,980 13,067 11,245 8,423 7,212 6,001 4,790 5,407 3,451 2,080 1,330
*As a percentage of initial value

Note: Numbers rounded to the nearest dollar. The number of overlapping 15-year payout periods from 1926 to 1995, inclusively, is 56; 20-year periods, 51; 25-year periods, 46; 30-year periods, 41.

Terminal Value

Portfolio success rates provide useful information for the question "Is my portfolio likely to last as long as I do?" A corollary question is: "What is the likely value of my portfolio after making all of those annual withdrawals during my retirement years?" Portfolio value at the end of a payout period, or terminal value, depends on length of the payout period, portfolio composition, and withdrawal rate.

Reflecting the methodology used in Table 1 for calculating portfolio success rates, Table 4 presents terminal values for a $1,000 portfolio (for a $1 million portfolio, multiply by 1,000) after making annual withdrawals. The terminal values are for portfolios containing both stocks and bonds, which exclude the most extreme allocations; and for payout periods ranging from 15 years to 30 years. Based on all past payout periods from 1926 to 1995, the statistical values in Table 4 for each case include the average, the minimum and maximum terminal values, and the median, which is the midpoint value (half of all values are below, and half are above).

As an example, assume a 75% stock/25% bond portfolio allocation, a 7% withdrawal rate, and a 20-year payout. Table 4 shows that the average terminal value for all 51 20-year periods from 1926 to 1995 is $2,435-in other words, after paying out 7% of the initial portfolio value each year for 20 years, the portfolio has $2,435 remaining, presumably to pass on to heirs. The worst 20-year period would have resulted in a terminal value of $0, while the best 20-year period would have resulted in a terminal value of $7,047. The median, or midpoint of all the results, is $2,076, which is smaller than the average and implies a distribution of terminal values that is skewed upward, which is also suggested by the large maximum value.

For stock-dominated portfolios, the median terminal value generally increases as the payout period grows longer, but so does the frequency of a zero minimum. Investors with longer planning horizons potentially will have larger terminal values, but without mid-course reductions in the withdrawal rate, in some cases, they will experience higher frequencies of portfolio failure. And, as the percentage of bonds increases, the median terminal value decreases, but the minimum terminal value increases, and the frequency of zeros is reduced.

Conclusion

What is the appropriate annual withdrawal rate from a portfolio during the retirement years?

It is clear from the results in Tables 1 through 4 that the answer depends on the mix of stocks and bonds in the portfolio, a planned payout period, and on a retiree’s degree of risk aversion and preferences for consumption patterns. Nonetheless, there are some general conclusions:


Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz are professors of finance in the Department of Business Administration, Trinity University, San Antonio, Texas.

© AAII Journal February 1998, Volume XX, No. 2