Controlling Risk

Part 1. Keep Your Financial Affairs Simple, Diversified and Tax Advantaged

© Peter James Lingane. Revised 7/13/99.

The Appropriate Investment Vehicle depends on your time horizon. One of an adviser's responsibilities, and one of the reasons planners recommend a periodic financial review, is to advise when shifts among investment vehicles are appropriate.

• An Emergency Fund smoothes the month to month variations in your income and expenses and is a place to accumulate money for major purchases like automobiles, appliances and vacations. It should be easy to get at, free of withdrawal penalties and low risk. Income is secondary. These criteria mean bank or money market accounts and shorter term US Treasury securities.

• Long Term Savings for your children's education or your retirement should be invested to maximize return. Liquidity and downside risk are secondary. These criteria describe common stocks.

  When you are starting a long term savings plan, take advantage of tax deferred accounts like IRAs and §410k/403b plans. As your tax deferred assets grow, it may become tax wise to convert these assets to a Roth IRA.

• Stocks, then Bonds As the time for drawing upon long term savings approaches, the investment objective changes from maximizing return to insuring that the money you have already accumulated grows at a predictable rate until it is needed. This means switching to bonds (not bond funds) with maturity dates matched to when the money is needed.

• Living Off of Your Assets. Retirement savings should be invested to maximize return while you are working and accumulating assets. This means investing in a diversified portfolio of common stocks.

Diversification of stocks and bonds among issuers, industries and geographical regions reduces risk and volatility. Diversification through individual stocks and bonds requires a large portfolio and substantial research (or the cost of professional management.) Consequently, mutual funds are an attractive alternative for many investors.

The allocation of your portfolio to different market sectors and to different geographic regions can be both strategic and tactical. Strategic allocation is based on the relative capitalization of different market sectors and geographic regions or on forecasts of relative long term growth rates among the sectors or regions.

I favor a 30:30:30:10 allocation among large capitalization US stocks, smaller capitalization US stocks, foreign stocks and real estate. Reasonable variations from these averages have not made a great deal of practical difference in the past and homeowners should consider whether they may already have an adequate commitment to real estate. The portfolio is rebalanced periodically to maintain the target allocation; it is important that the rebalancing be done without large immediate tax consequences.

Tactical allocation relies on forecasts of near term relative performance.   The portfolio composition is revised quarterly or annually to respond to perceived changes in relative performance among market sectors.

It is unclear whether tactical allocation provides the retail investor a better return after costs than simply maintaining a constant allocation. Whether tactical allocation increases performance is part of the larger debate over active vs. passive portfolio management.

Active vs. Passive Portfolio Management. If all information were available to everyone (and if we all knew what to do with the information), we would all put our money into the best investments and it would be impossible for one investor to outperform another. The best way to invest in an "efficient market place" would be to invest broadly at the lowest cost. This is referred to as passive investing.

The pricing of stock in companies followed by several analysts would be an example of an efficient market place.

When information is less readily available or if you discover an advisor who knows how to obtain special insights from the information generally available, the best investment strategy is to invest selectively, or actively.

Small capitalization stocks and foreign stocks are examples of market sectors where active management would be expected to excel.

Passive investing using index funds can be the better practical approach for the retail client even in an inefficient environment because of the extra costs of active management and because it is nearly impossible task to identify superstar managers in advance.

However you decide the active vs. passive management issue, it is essential that you track investment performance against a benchmark.

A Useful Benchmark is a diversified portfolio of stock index funds:

• 30% representing US large capitalization stocks
• 30% representing smaller US stocks
• 30% representing the European, Australian and Asian stock markets.
• 10% reflecting the performance of the US real estate market.

For the five years ending June 30, 1997, this benchmark achieved an annualized 19% return with a standard deviation (volatility) of 9% a year. This is a marginally poorer return but a marginally better volatility than achieved by the S&P500 index over the same period. (My intention, in comparing the benchmark and a well known index, is to supply historical information. I do not pretend to know what will happen in the future.)

If you or your investment advisor consistently exceed the benchmark return (after costs, of course), or match the return but at a lower volatility, congratulations! If your portfolio underperforms the benchmark, remember that is possible to invest in the benchmark itself (end notes ref. 1.)

Taxable vs. Tax Deferred Investing is a further consideration when designing an investment portfolio. Those investments that produce the most potential tax should generally go into the tax deferred account.

Investments that yield tax free income and investments that generate foreign tax credits or depreciation losses should probably be held in taxable accounts.

These considerations mean that you should generally hold different investments in your tax deferred and taxable portfolios. It is the combination of your tax deferred and taxable portfolios that should be broadly diversified, not the individual portfolios.

Be careful, when distributing assets among taxable and tax advantaged accounts, to calculation asset allocations using after tax estimates of value.  You will not achieve the allocations you are targeting if you use the nominal valuations from your brokerage statements.

An indexed annuity unitrust distribution formula is one where the initial distribution is specified and subsequent distributions are adjusted for the effect of inflation.   The advantage of such a formula is that it assures that the income beneficiary receives inflation adjusted distributions. The disadvantage is that an indexed unitrust formula can exhaust trust corpus during the lifetime of the income beneficiary unless the initial distribution is limited to an amount consistent with future inflation, future investment return and the life expectancy of the income beneficiary.

Someone considering retirement is faced with a similar problem.  Is their retirement nest egg large enough to support the distributions needed to make up the shortfall between expenses and other sources of retirement income?

It is straightforward to calculate how long a portfolio can support a given distribution if you know what inflation and investment return are going to be.   Matrices illustrating the relationship between a "safe" initial distribution and investment return are commonplace.  The following illustrates the largest withdrawal rate which can be sustained over a given time period as a function of investment return.

Matrix 1. Allowed Initial Withdrawal, After-Tax, as Percent of Initial Assets

(assumes 3.5% inflation, 20% tax on annual appreciation and January withdrawals)

I did calculations of the type illustrated when I was considering leaving the security of an engineering career and I soon discovered that Matrix 1 resolves nothing.  I had simply transformed my search for guidance as to a safe distribution rate into a search for guidance as to future inflation, investment return and life expectancy.

Almost always, we approach future inflation, investment return and life expectancy by looking at the recent US historical record.  Over the past seventy years, the US stock market has returned about eleven percent, inflation has been about 3.5% and the life expectancy of someone approaching retirement age is currently about twenty five years.

The usual next step is to adjust these numbers to reflect individual prejudices.   Perhaps market returns are adjusted upward because of the stock market boom over the last ten years.  Inflation is adjusted up - or down -   depending on whether memories of inflation in the 1970s outweigh recent experience.   And the planning horizon is extended by five or ten years to compensate for the fact that a twenty five year life expectancy means there is a fifty:fifty chance of living longer than twenty five years.  Out comes the spreadsheet or planning matrix and the problem is solved:  You can afford to withdraw $XX,XXX annually for the rest of your life and the only uncertainly is whether or not the check to the undertaker is going to bounce.

The reality is that no one knows what your investment return is going to be nor how long you are going to live.  My primary concern with the deterministic approach illustrated by Matrix 1 is that it provides no information about the risk you are running that you will run out of money before you run out of breath.

Another concern is that any approach which assumes constant rates of inflation and of investment return is biased high. Market volatility does not affect the cumulative growth of a portfolio so long as there are no additions or withdrawals. However, there can be difficulties when you begin drawing down a portfolio in retirement if the actual return achieved in the early years is below the annualized return over the beneficiary’s life time.

Part 3 illustrates this point by considering the two 25 year intervals beginning in 1951 and in 1972. The annualized inflation adjusted market returns were about 7% over both intervals. A trust which began distributions in 1951 at a rate predicated on a constant investment return would not have experienced difficulties whereas a trust which began distributing the same initial amount from 1972 was soon impoverished. The difference is that market returns were generally falling during the first period and generally rising from 1972. 

Sampling the historical record is one way to gain some perspective about the risk of running out of money and some correction for the bias caused constant rates of return. Consider a trust invested in an all stock portfolio which began to distribute 5% after taxes in 1926 and an inflation adjusted amount each year thereafter. This trust would have been able to sustain these distributions indefinitely.

By systematically repeating this simulation for all starting dates, one discovers that trust corpus is exhausted within twenty five years in about one third of the simulations. More precisely, there has been a 31% risk of exhausting corpus in less than 25 years assuming a 5% initial distribution amount and 20% tax on the annual appreciation. The actual risk that an individual with a 25 year life expectancy will run out of money before they run out of breath is less than 31% because there is only a one in two chance that the individual will still be alive after 25 years.

(Security regulations require that I point out that past performance does not guarantee future results. In addition to this obvious flaw, the described procedure oversamples the middle years of the twentieth century, undersamples the results at the beginning and at the end and probably undersamples large gains and declines. I suspect that these sampling issues could be resolved by a competent statistician.)

If one accepts historical probabilities as a working hypothesis of what the future might bring, the true risk of outliving corpus is the product of the probability of running out of money times the life expectancy of the income beneficiary. This first matrix quantifies the risk in terms of initial withdrawal rate, the age of the income beneficiary and the marginal tax rate (MTR) applied to annual appreciations. The calculations assume that the tax liability is paid each year.

Matrix 2. Chance of Outliving Trust Corpus (white female life expectancies), percent

Practically speaking, a 3.5% initial distribution has eliminated the risk of exhausting trust corpus during the life time of a sixty year old beneficiary who is in a 20% MTR, whereas corpus was exhausted six percent of the time for a sixty year old beneficiary in a 45% MTR.

Constructing the probability matrix to reflect the life expectancy and tax situation of the income beneficiary, provides perspective to assist the grantor in deciding the tradeoff between more income and increased risk of exhausting corpus.

The retiree may be more interesting in viewing the results assuming that the tax is deferred and only paid when money is withdrawn, as is the case for an IRA. The results suggest that it is probably wise to limit initial after tax withdrawals to between 2.5 and 3.5% of the initial account.  (Both withdrawal rates correspond to about 4.5% on a pre-tax basis.)

Matrix 3. Chance of Outliving an IRA (white female life expectancies), percent

Adding bonds to a stock portfolio lowers the expected return. This reduces the likely wealth transfer to the remaindermen. However, the security of the income beneficiary is usually of higher priority and bonds are added to the portfolio in hopes of increasing the financial security of the income beneficiary in spite of the impact on the remaindermen. Distributions are often reduced when adding bonds to compensate for the reduced return, guided by the kind of calculations illustrated by Matrix 1.

By examining probability matrices for stock and bond portfolios, one gains a historical perspective on how bond content affects financial security. Contrary to the usual assumptions, adding a moderate amount of bonds does not increase the security of the income beneficiary.  Indeed, the risk that the income beneficiary will outlive trust corpus is unchanged.  Consequently, there is no reason to reduce distributions unless bonds exceed 25 – 50%.

(Sixty year old white female)

The historical evidence is that moderate amounts of bonds reduce volatility (good) and hurt the remaindermen (bad) without increasing the financial security of the income beneficiary. Adding more than a moderate amount of bonds reduces volatility further but adversely affects both the income and remainder beneficiaries.

Forecasts which neglect the possibility of low returns in the early retirement years underestimate the risk of outliving trust corpus. It is probably for this reason that the probability approach leads to lower initial distribution limits than does the spreadsheet. For example, the historical market return has been about 11% before-tax and the life expectancy of a sixty year old is about twenty five years. Entering Matrix 1 at 11% return and twenty five years, we see that the initial distribution limit should be 6½% of corpus. Matrix 2 suggests that this is excessive since a 6½% initial distribution has, historically, led to a larger than one in four chance of outliving corpus.

This approach is fundamentally the same as "Determining Withdrawal Rates Using Historical Data," William P. Bengen, J. Portfolio Management, October 1994, pp. 171 – 180, "Investing for Retirement: Using the Past to Model the Future," Larry Bierwith, J. Portfolio Management, January 1994, pp. 14 – 24 or "Retirement Savings: Choosing a Withdrawal Rate That is Sustainable," Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz, AAII Journal, February 1998, pp. 16 - 21.

See also

• "Determining a Realistic Spending Rate in Retirement," T. Rowe Price's Fall 1999 Newsletter at www.troweprice.com/viewpoint/pr99fall.pdf
• Bibliography "How Much Can You Safely Withdraw From Your Savings During Retirement" at www.bylo.org.
• "Making Retirement Income Last a Lifetime," John Ameriks and Mark J. Warshawsky, Journal of Financial Planning, December 2001, p.60.
• Various articles at www.retireearlyhomepage.com/
• "How Long Can Retirement Savings Last?" By Jonathan Clements, The Wall Street Journal. Available at www.newstimes.com/archive96/dec0396/bzc.htm.
• Searching www.google.com for "What much can you safely withdraw during retirement" led to over 3000 hits!

Part 3. This Thing Called Risk

© 1998, Peter James Lingane

There are a variety of things that can go wrong in an investment portfolio, and therefore a variety of steps that need to be taken to reduce risk. The most challenging risk is the possibility of a serious market decline in early retirement.

Knowledgeable investors know that they can protect their family from the financial impact of their premature death by purchasing life insurance and they know that annuities protect against the risk of living too long. They know that they can reduce the risk of unnecessary taxes at death by carefully drafted and periodically reviewed legal documents.

Short term market movements tend to be random. Not totally random, or momentum strategies which rely on short term persistence ("The trend is your friend") would never succeed. There is also a reversion to the norm; one of the scary aspects of the current market is that the recent splendid performance may be the prelude to a serious market decline.

Because market fluctuations tend to be random, longer term price changes tend to follow the normal or bell shaped distribution. Statistics play a large role in the modeling of market performance. Option pricing is only one such application.

To the extent that market changes are random, the average performance becomes more predictable the longer the time interval due to the balancing out of shorter term variations. The daily variation may be thirty times the trend while the variation in annual returns is closer to twice the trend and the variation in ten year returns is smaller than the trend.

It would be a mistake to conclude that longer term investments are safer. The longer term is only more predictable if there are no additions or withdrawals from the portfolio.

Suppose that you buy an investment which appreciates 10% the first year but falls ten percent in the second. Would you have been better off if the investment had declined the first year and appreciated the second? No, you lose exactly 1% in either scenario. Convince yourself of this result that by multiplying 1.1 (a gain of 10%) times 0.9 (a loss of 10%) and comparing the answer to 0.9 times 1.1.

This specific observation can be extended to any number of years: the order of investment gains and losses makes no difference to the performance of an investment which is bought and held.

Before retirement, we are adding to our portfolio. Additions when the market is down buy more shares than additions when the market is up. Dollar cost averaging can therefore sometimes make money in volatile markets.

Predicting portfolio growth before retirement requires, in principle, a knowledge of both the underlying trend and the ordering of the gains and losses. But we shall see that forecasts based on the underlying trend alone don't come out too badly.

The order of large market movements is important when portfolio value is declining, as during retirement, even if the net change is small. Consider a market which declines by half and subsequently doubles and another which doubles and then declines by half. The net change is zero in both instances.

If you withdraw $10 and the market falls 50%, your $100 portfolio is worth $45. If you withdraw another $10, leaving $35, and the market doubles, your portfolio is worth $70. Your portfolio declines by more than your withdrawals because the market went down before it went up.

If the market changes were to occur in the reverse order, the final portfolio would be worth $85. Your portfolio declines less that your withdrawals because the market went up before it went down.

The conclusion is that, if there is to be a substantial market decline, you would prefer that it occur late in life rather than shortly after you retire.

Figure 2

If one had invested a thousand dollars each year from 1951, the savings would have grown to about $45,000 by the end of 1976. If savings had begun in 1972, the portfolio would have been worth about $130,000 in 1997. Had the investment return been a constant 7%, the savings would have accumulated to a little less than $70,000. The effective investment returns would have ranged from 5 to 11%. The possibility that savings might undershoot the forecast by one third is hardly a celebration of forecasting prowess but it is not be a matter of huge concern either. Perhaps $45,000 is adequate for your needs, perhaps you could work a few years more, perhaps you can reduce retirement spending. The point is that you have options during the accumulation phase that you do not have after retirement.

The solid line in Figure 3 illustrates how retirement savings would decline if the investment return were 7% each year, if annual withdrawals were 8% of the initial portfolio and if these withdrawals were adjusted each year for inflation. Retirement savings are exhausted at the end of 25 years which is a plausible planning horizon for someone in their middle sixties. Unfortunately, this reassuring forecast obscures the potential for a financial disaster if there were a serious market decline shortly after retirement.

Figure 3

The solid points are the performance of a hypothetical portfolio based on the historical performance from 1951-76. Even though the annualized return over these 25 years of generally declining market returns is 7%, the same return used to calculate the solid curve, the portfolio value at the end of the period is larger than the starting valuation.

The asterisk-like symbols in Figure 3 are the results of a second hypothetical scenario based on the historical performance from 1972-97. Even though the annualized return over these 25 years of generally rising market returns is 6.8%, nearly the same as the return used to calculate the solid curve, retirement savings are quickly exhausted.

Return to beginning of Part 3; return to top.