by Rob Bennett

Buy-and-Hold and Valuation-Informed Indexing are both data-based models. How is it that the strategic recommendations made under the two models are so different? There is only one set of data. How can it be used to support such different conclusions, depending on whether the person interpreting it happens to favor the Buy-and-Hold Model or the Valuation-Informed Indexing model?

The difference is that Buy-and-Holders believe that stocks offer a stable value proposition. Valuation-Informed Indexers believe that stocks offer a variable value proposition. That one difference is so fundamental that it causes followers of the two models to arrive at different strategic recommendations on just about every question examined.

Say that you want to know how much you can take out of your portfolio in each year of your retirement. A Buy-and-Holder looks at the historical record, notes that a 4 percent withdrawal has never failed, and concludes that it is safe to take out 4 percent. Makes sense, no?


Not if you believe that the value proposition of stocks varies.

If you believe that the value proposition is constant, it makes perfect sense to look at the entire historical record for guidance on what withdrawal rate is safe. If the value proposition is stable, the entire historical record offers relevant data points for your to consider in your analysis.

But if you believe that the value proposition varies, you can only look to time-periods when stocks offered a similar value proposition to the value proposition that applies on the day your retirement begins to learn what withdrawal rate is safe for your particular retirement. It makes no sense to look at what happened in time-periods in which entirely different realities applied. If the value proposition of stocks changes from time to time, the safe withdrawal rate too must change from time to time.

The fact that a 4 percent withdrawal worked for someone who retired in 1980 is not relevant information for the person who retired in 2000. The valuation levels that applied at those two times were very different and thus, the 10-year return from 1980 forward was likely to be much higher than the 10-year return from 2000 forward. The Year 2000 retiree taking comfort in the fact that a 4 percent withdrawal worked for the Year 1980 retiree is fooling himself.

The Year 2000 retiree needs to look at what happened to retirees who retired in circumstances similar to those that apply for him. What he learns when he does so is that retirements calling for a 4 percent withdrawal came within a whisker of failing in both of the two cases in which valuation levels similar to those that applied in 2000 applied in earlier times in history. The data is not telling us that that withdrawal rate is a safe one but that it is a high-risk one. A very different finding!

It’s the same with just about any other strategic question you might examine.

Does a 60 percent stock allocation make sense for the typical middle-class investor? The historical data shows that on average it does.

But if you consider the effect of valuations, you find that there are some P/E10 levels for which a 60 percent stock allocation makes good sense, some for which it is too low a stock allocation, and some for which it is too high a stock allocation. If the value proposition of stocks varies, the asset allocation advice we give to investors must vary with it.

Are stocks risky?

Most people think the are. And, if you want to know whether stocks are risky on average, the data does indeed support the conventional thought on this question.

But there has never been a time when stocks performed poorly in the long term starting from a time of low valuations or starting from a time of moderate valuations. Start from the premise that the value proposition of stocks varies with changes in valuation levels, and you are led to a conclusion that the riskiness of stocks is concentrated: Stocks are insanely risky at times of high valuations but are not nearly as risky as most perceive them to be at all other times.

This reality (that the value proposition of stocks varies) explains why investing advice became so messed-up not long after stock analysis became an object of academic study. The benefit of systematic analysis is that it drills deep. We have learned more about how stock investing works in the past 40 years than we did in the prior 140 years. Unfortunately, most of what we have learned is wrong. Deep drilling is not such a good thing when you are drilling in the wrong place!

Our mistake was in coming to a hasty conclusion in the 1960s that the value proposition of stocks is a stable thing. Now that we (well — some of us!) know better, we can get to work drilling at spots where there really is a good chance of hitting a gusher. The value proposition of stocks varies according to the valuation level that applies. Once all researchers adopt a practice of considering the effect of valuations in all studies they perform, analysis of the historical return data will begin bearing good fruit at last.

Rob Bennett is always trying to figure out what is different about budgets that work. His bio is here.