Appropriate asset allocation is extremely important, but unless almost no stock investments are in the portfolio, in a secular bear market risk levels are still going to be too high to eliminate the lurking danger of large drawdowns and long periods to recover them. That is why another step is needed, and where traditional Wall Street thinking is woefully lacking.

That step is using a "mechanical" model to reduce the risks to the point where the lurking danger is essentially gone. In this context, mechanical means not requiring judgment because it consists of mathematical formulas that say when investing in stocks (or bonds or other asset classes) is too risky and when the risks are at acceptable levels. There may be a few investment analysts who have judgment in this regard that is consistently accurate and who can apply it unemotionally and without letting their egos come into play. Most investors and analysts can't do that, so mechanical models are preferable. It still requires discipline to follow the models and execute the signals they generate.

Another advantage of mechanical models is that their historical performance can be evaluated and analyzed. It is best if this evaluation can be performed over a long period and a wide range of market conditions. The strengths and weaknesses of the models can be identified, which leads to an understanding of how and why they are effective and under what circumstances signals may not be reliable. That is all but impossible to do, even if there is a long and stellar track record, for signals that results from someone's judgment.

The pages that follow show the effects of two mechanical risk reduction models: one for stocks and one for bonds. MDP Associates employs these models in the management of retirement and other portfolios. However, the models were developed and first tested by others, who are respected analysts and successful money managers. We did make some minor modifications and perform our own testing and analysis of the models.

Both of these models are quite simple. Each has only three data elements as inputs and each employs simple formulas rather than sophisticated mathematical or statistical tools. You may wonder if a simple model can be effective or if it would not be better to add complexity to improve the model's performance. The answer to the first is that these simple models have proven to be effective because they capture the major trends in stock price and interest rate movements. A more complex model might work better, but experience has taught that the more elements in a model, that is the more complex it is, the greater the danger that it will not work as well in the future as it has in the past. In other words, It is better to possibly sacrifice some potential model effectiveness in return for more confidence that the model will continue to serve its primary purpose--risk reduction--in the future.

Risk Reduction Model for Stocks
The three inputs to this model are 1) stock prices as measured by the S&P 500 index, 2) short-term interest rates (90-day Treasury Bills), and 3) long-term interest rates (10-year Treasury Bonds). The formulas employed determine the major trends--of a duration that is expected to change at most two or three times a year--of these three components. The logic is simple yet compelling. The risks of stock ownership are acceptably low when a) stock prices are trending up and b) at least one of the interest rates is trending down. Otherwise, the risks are too high, so stocks should not be owned.

Risk Reduction Model for Bonds
The structure of the model for bonds is similar to the one for stocks. Instead of stock prices as an input, the value of the Dow Jones Utility Average is used because utility stocks are more sensitive to interest rate levels and anticipated future levels than are stocks in general. The two interest rate inputs are the same. Once again, the model evaluates the current trends in the three inputs although not exactly the same way as the stock model. Keep in mind that bond prices move in the opposite direction as interest rates and that longer term bonds are more sensitive to interest rate changes so they will have larger percentage moves than shorter term ones. The logic is similar to that of the stock model. The risks of bond ownership are acceptable when long-term interest rates are falling and either short-term rates are also falling or utility stock prices are rising.

It is important to realize that the trend evaluation methods in these models are not likely to identify that the trend has changed until several weeks after the high or low values occur. That is by design. The models do not try to and are highly unlikely to "buy at the low" or "sell at the high." That is not their purpose. Reducing risk to protect portfolios from significant equity drawdowns does not require that. Moreover, I do not think any model or person can consistently identify the highs and lows of stock and bond prices. Worse yet, trying to do that is quite likely to lead to going broke and going crazy, not necessarily in that order.

The next page show how these models would have worked in conjunction with the three example asset allocations over a long period of time in both secular bull and bear markets. You will see how effective they are in reducing risk.

Next page: long-term histories

Previous page: appropriate asset allocation

Return to retirement investing main page