Subscribe                                InvestorsFriend.com                             Home

ASSET ALLOCATION IN RETIREMENT, (PERCENTAGE TO HOLD IN STOCKS, VERSUS BONDS AND CASH)

The "asset allocation" choice is one of the toughest and most important decisions that investors face.

There is general agreement that your asset allocation has a huge impact on your long-term and short-term returns and on the volatility of returns.

Most articles that explore how your portfolio might grow if invested in different proportions of stocks, bonds and cash, rely on assumptions and averages. However, assumptions and averages can be very mis-leading. Much of the advice on this topic seems designed to make life easier for investment advisors rather than to maximize or optimize results for investors.

Rather than speculate on how investments might grow with different asset allocations, it seems to me that it is best to start by looking at results from actual past data. For illustration, I looked at scenarios where people retired in various years. I have data for stock, bond and cash returns that starts in 1926. This data is from a publication called Stocks, Bonds, Bills and Inflation which is published annually by Ibbotson Associates, now a division of Morningstar.

The following article shows what actually would have happened to people who started with a "pot" of money at the end of each year from 1926 to 1977 and who then withdrew money for 30 years (to simulate retirement). This is based on actual U.S. total return indexes for stocks, bonds and cash. Our data is also adjusted to account for inflation. By adjusting the data for inflation we can simulate a retirement income that adjusts for inflation and we insure that $100,000 at year one still has the same $100,000 purchasing power 30 years later.

How Retirement Savings Really Declined in Retirement with Different Asset Allocations based on Actual Market Returns

This article presents some rather startling conclusions about how retirement portfolios with different asset allocations would have actually declined in the past and what the annual volatility has been. Based on this, I found that "average" results are not in fact "typical" because there has been a huge variation around the average, even over 30-year periods.

The results suggest that a higher asset allocation to equities historically worked out well, over almost all historical 30 year periods (although with some truly ugly volatility along the way).

Retirement savings draw-down scenarios are often worked out using assumed average returns that are expected to be achieved. The return assumptions tend to range from 6% to 8% per year (on a nominal basis before adjusting for inflation). Usually, it is assumed that funds are invested in a balanced manner, between stocks, corporate bonds and short-term cash deposits. Usually it is assumed that a certain amount will be withdrawn from the portfolio each year.

Under most of these scenarios, your savings are projected to decline at a steady rate over the years. Often, much is made of the need to preserve capital or avoid negative returns in any given year. Usually the fact that inflation eats away at purchasing power is ignored or glossed over.

But the actual data tells a far different and rather surprising storey...

Imagine that $1,000,000 was available at the start of a retirement period that was to last 30 years. And imagine that 4% of the original amount or $40,000 per year (adjusted for inflation each year ) is withdrawn each year. (Recent studies suggest that withdrawing more than about 4% causes a risk of running out of money). Based on past results, how would the portfolio decline (or possibly grow)? What do the results look like based on investing all the money in stocks and based on dividing the money equally between stocks, bonds, and short-term cash?

When we applied this scenario to someone who began a 30-year retirement period in each of 1926, 1936, 1946, 1956, 1966 and 1976, using actual historical market index total return data, the results were quite startling.

We compared two approaches. One asset allocation approach was to invest the portfolio in 100% equities as represented by the S&P large stock index in the U.S. The other approach invested the funds in a fully balanced manner with 1/3rd in each of large stocks, corporate bonds and cash, and rebalanced each year. Our data was taken from annual yearbook, Stocks, Bonds, Bills, and Inflation, by Ibbotson Associates. Their figures show real total returns (adjusted for inflation and including re-investment of dividends). These figures ignore any transaction and money management fees. They also ignore income taxes and therefore imply a tax-free account.

We use an initial portfolio amount of $1,000,000 as a round figure. We assumed 4% or $40,000 (adjusted for inflation each year) was withdrawn each year for 30 years. (This was equivalent to starting with about $100,000 in 1926, and withdrawing $4,000 per year but the same math applies.)

First, let's look at how much money was left over after withdrawing the inflation-indexed $40,000 for 30 years, from the original $1 million portfolio, using both a 100% equity strategy and a fully balanced approach.

If the portfolio was 100% invested in large stocks (The S&P 500 index and its predecessor), then the total $1,000,000 tends to grow rather than shrink over the next 30 years. In the case of the person who retired at the start of 1966, the $1 million was exhausted prior to 30 years. In the other five cases the portfolio grew, despite the withdrawals. The ending values for the six scenarios ranged from $0 to $5.6 million in constant purchasing power inflation-adjusted dollars. One thing that this illustrates is that the variations around the "average" are huge.

A 100% equity portfolio, in retirement is almost universally considered extremely risky. The usual advice is to use a balanced portfolio. This is expected to offer a lower return, but less chance of ever running out of money, and less volatility along the way.

Looking at the Fully Balanced Portfolio, in the table above, it certainly delivered far less return (as promised). In fact the ending balance for the Fully Balanced approach trails the 100% equity approach by staggeringly huge amounts except for the 1966-1995 scenario where both the all-equity and the balanced approaches ran out of money.

Looking at the table above, it appears that a 100% equity approach to asset allocation may be preferable, even in retirement. But we are not able to see the volatility along the way and this is based on only six cases. (However, the six cases are 30 year retirement periods that start in six different decades so they may be reasonably representative of a range of possibilities). Below we will provide graphs for all 52 possible 30-year retirement scenarios from 1926 through 1955, all the way to 1977-2006. (We only look at calendar year scenarios and do not look at scenarios that start other than January 1 of each year).

But first, let's take a graphical look at the six retirement scenarios presented in the table.

Graphs allow us to clearly see the annual volatility through the years.

Graph below is 100% Equity Portfolio of $1,000,000 with $40,000 (growing with inflation) withdrawn annually

The above graph is for a very risk-taking retiree who decided to remain 100% in large stock equities for 30-year retirement periods starting 1926, 1936 etc. and who started with $1,000,000 and withdrew $40,000 per year adjusted up for inflation and down for deflation each year to maintain a constant $40,000 worth of purchasing power withdrawn each year.

As was apparent from the table above, the ending results for the different starting years varied across an extremely wide range. The 1966 scenario (the green line) was the lowest and actually ran out of money around year 27.  This retiree ran into the poor stock markets of the late 60's that were volatile but with no upward trend over a period years. There was also a major market crash in 1973, 1974. In addition there was very high inflation through the 70's with the result that the market needed to return around 10% just to break even after inflation.  Around 1982 a huge bull market started, but by then this portfolio had declined to about $250,000 and the bull market could only slow the decline at that point.  It's interesting to note that by about 1968, Warren Buffett was warning that stocks were over-valued and he wound down his investment partnership. The late 60's was (as Warren Buffett predicted) a very bad time to decide to take a 100% equities approach to a retirement portfolio. In fact the late 60's were a bad time to retire no matter what investment strategy was used because of the crushing inflation of the 70's.

Two of the scenarios shown feature huge drops in the portfolio value. These were truly gut-wrenching and horrific declines in the range of 50% of purchasing power. The 1946 retiree (the yellow line) saw exceptional returns for about 20 years, it then ran into weak markets around 1969 and very poor markets in 1973 and 1974, all of this compounded by brutal inflation. Still, even after the brutal decline this retiree still had about $2.3 million left after the decline because the portfolio had earlier built up to well over $4 million. Note that these figures are adjusted down for inflation.

More recently, the retiree who began with $1 million in 1976 (the light blue line) experienced increasingly good returns. By the late 90's the portfolio appeared to be headed for the moon. But the stock crash of the early 2000's was quite devastating. Still, this retiree, rather than experiencing a decline in the portfolio over the retirement period, ended up with almost $5 million in the end.

Notably, the retiree who began in 1926 (shown in the red line), did very well despite the huge stock crash of 1929 to 1932. This retiree doubled the portfolio to $2 million  by the end of 1928. There was then a devastating loss which (cushioned by deflation) reduced the portfolio back to under $1 million in purchasing power by the end of 1932. But reasonable returns were achieved thereafter and near the end of the period the portfolio sky-rockets to over $5 million with the strong markets of the early to mid 1950's. 

Overall this 100% equity approach to asset allocation did result in excellent portfolio growth (rather than the expected decline) over the full 30-year period, but had huge annual volatility and huge differences in the final amount of wealth.

Now, let's examine the graph for the Fully Balanced approach to asset allocation in retirement. Note that the balanced approach is fully re-balanced to 1/3rd in each of stocks, bonds and short-term cash investments at the start of each year.

Graph for Fully Balanced Approach with $40,000 withdrawn annually, growing with inflation, starting with $1,000,000 Portfolio

This graph uses the same scale as the 100% equity graph, to aid comparability.

As indicated in the table, there is far much less variability in the ending portfolio values, although ending differences are not eliminated. And it is immediately apparent that the ending values are on average substantially lower. In four of the six cases, the portfolio declines, in a reasonably smooth fashion to zero or near-zero by the end of 30 years.

Excepting the 1926 and the 1976 scenarios the other four scenarios follow relatively close to a text-book pattern of a smooth decline to zero at about year 30. The 1926 and the 1976 retirees end up 30 years later with portfolios worth more in real purchasing power than they started with.

In five out of the six cases, the lower volatility of the balanced approach comes at the expense of a huge amount of foregone wealth. The sixth case is very similar to the all-equity approach (both run out of money around year 27)..

Based on these 6 Scenarios, which is better 100% Equity or the Balanced Approach to Asset Allocation?

With the benefit of hindsight, we can see that the Balanced Approach usually foregoes, huge amounts of wealth in return for substantially less annual volatility. The graphs confirm the rule of thumb, that an all equity approach usually outperforms balanced portfolios in the long run, but at the cost of extreme volatility.

It is tempting then to suggest that this evidence argues strongly in favor of a very high allocation to equities even in retirement.

But the graphs show extreme volatility, in some of the scenarios, with a 100% equity approach. It's entirely possible that seeing our $1 million portfolio climb to say $4 million and then plummet to $2 million would drive many of us crazy. Would you want to live the last 10 years of your life knowing that you blew half your savings (estate), amounting to $2 million by your failure to  get out of the market at a peak?

In the end, there simply are no easy answers to the asset allocation problem. Each individual will have to choose their own allocation based on their unique circumstances.

The evidence seems to suggest that if you can both AFFORD high volatility AND YOU CAN STOMACH IT, then a very high allocation to equities might be appropriate even in retirement. This could apply for example where your portfolio is so large that losses in the market will not affect your lifestyle. And, for example, where other pension or other annuity income is enough to maintain an acceptable lifestyle.

Evidence Based on More Data

The above analysis was based on just 6 retirements dates, one per decade starting from from 1926 through 1976.

Let's now graph all the possible 30-year retirement scenarios using each retirement year from 1926 through 1977. The first graph is for the all equity approach.

Graph below is 100% Equity Portfolio $40,000 initial withdrawal, growing with inflation

The graph illustrates that with a 4% or $40,000 initial withdrawal, that grows with inflation, from a 100% equity portfolio, in most cases the portfolio grows substantially in retirement. In only 4 cases did the money run out too early (retired in 1929, 1966, 1968 and 1969), and then never before year 25. (Again, remember that the late 60's was the period that Warren Buffett closed down his equity partnership and indicated that extremely few if any bargains were to be found.)

The above graph vividly illustrates that what will happen to a 100% equity portfolio over a 30 year period, is highly uncertain. Analysis that focuses on so-called average outcomes such as a steady real 6% gain (which in fact never happens) are highly mis-leading because they do not consider the range of likely outcomes. You can calculate an average line to plot on this graph, but clearly most individual retirement experiences would be far from average.

The majority of these portfolios grow through retirement, with quite a number growing more than five fold. And this is in real dollars, adjusted for inflation.

If you look closely though you can see many huge near-vertical drops. Most of these 100% equity scenarios see a number of large drops over the 30 years (although most years see gains) and most have one absolutely devastating period (losses in the range of 35 to 50% or more) corresponding to living through the market crashes of 1929 through 1932, 1973 through '74 (this one exacerbated by huge inflation) of the early 2000's.

Given the extreme volatility illustrated with the 100% equities approach, let's now examine the fully balanced approach.

Graph for Fully Balanced Approach to Asset Allocation with $40,000 initial withdrawal, growing with inflation

Note that the same scale has been maintained in all four graphs for easier comparability.

The Balanced approach to 30-year retirements that started in 1926 through 1977 delivered a much tighter range of ending portfolio values and also less annual volatility. In most cases the portfolio would have declined through retirement. In many cases there was still significant volatility but nowhere near as large as for the 100% equity approach. Technically then, the balanced approach has less risk, less uncertainty.

However, the balanced approach actually ran out of money, twice as often, 8 times (retirement in '37, '39, '40, '64, '65, '66, '68 and '69).

Looking at the two graphs it seems that, based on this data, the Balanced Approach almost never led to more total wealth over the full 30 years. (The only case where the Balanced approach did better, regarding ending wealth, was for retiring at the start of 1929). The Balanced approach provided substantially less annual volatility but at a huge cost in terms of foregone wealth, for the vast majority of the scenarios.

Observations and Conclusions

The above is the U.S. data for 30 year retirement scenarios starting 1926 trough 1977. I believe my calculations are correct but there is always a chance that I have made an error in calculations and or data entry.

Many books have been written about the proper asset allocation in retirement and about the only universal agreement is that the proper asset allocation is very much dependent on each individual's unique circumstances including particularly the financial ability to accept risk and the emotional willingness to suffer large losses along the way.

Therefore it is perhaps best if readers draw their own conclusions from the graphs and seek other opinions as well.

However, I will provide my thoughts.

The graphs confirm that the the initial percentage that can be safely withdrawn from a portfolio (where the initial amount is to be fully increased for subsequent inflation), is distressingly low. Even when only 4% was withdrawn (and adjusted for inflation each year), in 15% of the past scenarios the money ran out in the last few years with a Balanced Portfolio. This may suggest that if absolute certainty is required then an inflation-adjusted life annuity (purchased from an insurance company) may be the best option.

I believe that the graphs show that it is completely unrealistic to think that a retirement portfolio, with any exposure to inflation, stocks, and/or bonds, can be left on auto-pilot with a withdrawal level that is set at the start and then never adjusted except for inflation. That scenario just does not match what really happens. In real-life, only an inflation-adjusted annuity can be deliver that result. And the annuity approach foregoes a huge potential up-side, and also precludes any of that money being left for an estate.

In real life, the withdrawal rate, and perhaps the asset allocation, will likely be reviewed and changed every few years, depending on many things including, health, life expectancy, financial needs and wants and particularly the portfolio's performance. All of our working lives we are forced to adjust our spending in accordance with our income. In retirement if our income depends partly on a financial portfolio, then we can expect to adjust our spending in accordance with our investment results. If in retirement your original $1 million grows to $2 million, you are almost certainly going to withdraw more. Conversely, it it sinks to $700,000 after two years, you are going to have to cut back.

A high asset allocation to equities, even in retirement, appears to offer a good chance for huge rewards but at the cost of huge annual volatility. It may be that by understanding that occasional large losses are simply part of a jagged road to higher wealth, many of us could increase our risk tolerance. Looking at the graphs, the Balanced Approach does not look very appealing, though it would be far less stressful.

Other analysts have claimed that the Balanced approach can often be superior, so please check other sources.

Caveats
Again, the above is the U.S. data for 30 year retirement scenarios starting 1926 trough 1977. I believe my calculations are correct but there is always a chance that I have made an error in calculations and or data entry.

The graphs are based on annual year-end data only . Actual daily volatility and distances from peak to trough based on daily data are not shown. For example, the unfortunate scenario of getting 100% into stocks on the highest peak day in 1929, for example, is not shown.

The future is not the past. Even if equities almost always beat balanced approaches over 30 year periods in the past, there can be no assurance that the same will apply in the future. Certainly if one began a retirement with 100% invested in equities at a market bubble peak, the results could devastating.

The above analysis is based on U.S. data from 1926 onward. This was a golden age of progress and it is possible the future will be far less rosy for equities.

Tax rules might mandate withdrawals significantly larger than assumed in this analysis.

Data for Canada and other Countries may not show the same ultimately favorable (although volatile) picture for equities.

The equity allocation here is to the S&P 500 index. Allocations to less diversified or smaller cap equities would have different results.

We have ignored taxes. However in general, taxable accounts would sway the results even more in favor of equities because of the favorable treatment of capital gains historically and of dividends more recently.

We have ignored portfolio management and transaction costs. However, it is not clear that these are materially higher for equities versus bonds and cash. With the 100% equities approach, little trading might be needed if one were tracking an index. In the Balanced approach the annual rebalancing would add to trading costs. Management and trading costs would likely lower the curves in the graphs but not change their shapes much.

The bottom line is that there are no easy answers in investing. Asset allocation in retirement is a highly personal choice. However, the above analysis will hopefully allow the choice to be made from a  more educated level.

February 20, 2007 (with minor edits to April 25, 2008)

See also our related article about Asset Allocation During the Savings Phase of Life

Shawn Allen

President

InvestorsFriend Inc.

www.investorsfriend.com

We can show you how to grow rich by investing more like Warren Buffett would.  Click to see our  free monthly investment newsletter   

Check our full Track Record here.