Chart the Future of Your Retirement

There is little doubt that the past two years have been tough on retirement savings. Indeed, the market downturn has taken some diversified stock and mutual fund portfolios down 30%.

In 1999, many people were considering retiring early. But in 2002, many of those same people are wondering whether they can retire at all.

In my May 2001 article "Portfolio Survivability: Will Your Assets Outlast You?," I discussed how retirees could draw down annually a conservative 5% to an aggressive 8% of their assets and never run out of money, if their portfolios were invested 75% in stocks and 25% in fixed income. I provided tables from an academic study showing survival rates for different asset mixes and time frames.

Recently, I started using a software program that models retirement savings outcomes more precisely by incorporating different investment allocations and drawdown rates, and by including income from Social Security or defined benefit plans.

The program was developed by Bill Swerbenski, a chartered financial analyst who uses an analytical technique called "Monte Carlo simulation," based on Microsoft Excel. (You can download a limited demonstration copy by clicking here, or pay $25 for the full version.)

The simulation uses the ranges of historic rates of return for inflation (the CPI index), stocks (the S&P 500), bonds (all government-issued) and cash (the T-bill index) to generate "trials" of overall returns.

Each "trial" represents the growth of assets over a given time frame (for example, 20 years). The simulated change year over year is guided by the historical average returns and standard deviations [a statistical measure of historical volatility typically computed using 36 monthly returns] of the asset classes.

For example, the S&P 500 has gained an average of 1% per month since January 1946, with a standard deviation of 4%, which means that 68% of the time, the monthly gain in the S&P 500 will range from -3% to +5%.

Even though past performance is no guarantee of future results, the information contained in the historical average returns used in this program, particularly the historical volatility expressed as a standard deviation, is more realistic than plugging an assumed constant rate of return into your retirement calculator. You can also add or substitute your own historical data (for example, corporate bond returns).

Because each trial produces a different result, an average of multiple trials will show ranges of likely outcomes for different drawdown rates and asset allocations.

To explain how this program works, let me take you through an example. Say you have $1 million saved, have $4,000 per month in pension benefits and $2,000 a month in Social Security benefits, and you intend to take out $12,000 each month for the next 20 years and then $7,000 per month during the following 10 years. To see how you would enter such cash flows, click here.

Next, you need to specify how you would like your assets allocated. To see an example, click here.

The "Options" page of the program lets you include additional parameters, such as tax rates. For an example, click here.

The iterations parameter on this page, which controls how many trials are run per simulation, deserves particular attention. The fewer trials you include (for example, 50), the faster the simulation will run. However, more trials reduce the variability in the results from simulation to simulation. (Not only will each trial produce a different outcome, the averages of multiple trials will rarely be the same from run to run.) I recommend including at least 250 iterations.

After you've entered all the parameters, click "Run Simulation." To see an example, click here.

The solid line in the graph shows the average value of the portfolio over your expected life span. The bands within the graph, indicating the +/- 1, 2 and 3 standard deviation bands, illustrate the likely ranges of values.

The program also calculates the chance your portfolio will survive through a given time frame. In this example, the portfolio has only a 6.75% chance of surviving 30 years. Using a 7.2% drawdown rate from a $1 million portfolio, the retiree is netting $6,000 every month ($12,000 in expenses minus $6,000 in benefits). However, because the portfolio is invested 25% in cash, 25% in bonds and 50% in stocks, the drawdown (net of taxes) exceeds the average 7.68% annual return of the portfolio. In my opinion, the initial allocation is heavy in cash and light in stocks. If you followed this strategy, you'd be taking more money out of the portfolio than it could support. As a result, the portfolio would run down to zero before your time frame ends.

When I changed the asset allocation to 75% equities and 25% bonds, the probability of survival for 30 years rose to 20%, and the average returns rose to 10.3%.

Then I reduced the draw rate from $12,000 per month to $10,000 ($48,000 each year from assets, or just under a 5% per year draw rate), and the probability of survival for 30 years rose to 49.2% with the average value of the portfolio at the end of the time frame just under $1 million. This lower drawdown rate preserves the principal value of the portfolio through the period. As I described in my previous article, a 5% draw rate is about as conservative as you need to be.

Note: You can use Swerbenski's program for more than retirement forecasting. For example, if you estimated that your newborn's college expenses were likely to be $35,000 each year for four years and you intended to set up a 529 College Plan, you could estimate how much money you need to put in over 18 years to have a good chance of being fully funded. If your plan has an allocation of 80% equity and 20% fixed income, a $3,000 contribution each year, growing tax-free, yields a 58.8% chance, while a $2,000 investment offers only a 23.2% chance.