More Q&As

Knowledge@Wharton: We're all familiar with the warnings we get from our brokers when we invest in something, that past performance is not a guarantee of future results. So most people recognize that there are uncertainties about the future, but when we talk about volatility and uncertainty, a lot of people automatically translate that into risk. They look at the downside. But we also have surprises on the upside, don't we? Such as technological breakthroughs that nobody could imagine earlier, things that improve productivity.... So, professor Stambaugh, according to your work, it's possible not just that results could be worse than they have been in the past, but that they could be better?

Stambaugh: Exactly. We're not making any sort of bullish or bearish statement. We're just saying that whatever projection one makes about the trend or what's likely to happen, you can have big surprises on the upside or downside. And part of what contributes to the overall uncertainty in the long run is just uncertainty about things like the trend. Indeed, we were somewhat unsure when we began this whether 200 years of data that Jeremy has very carefully assembled -- and we're very grateful to have available -- would indeed leave one with a lot of uncertainty about this trend, or whether it would resolve a lot of that uncertainty. This is where we bring statistics to bear on the problem, which allows us to quantify how much uncertainty will be left over....  And, indeed, we found out that even... two centuries of data leave one with enough uncertainty that as you look at the implied variance of stock returns over the longer horizons, the risk actually does rise significantly with [the time] horizon.

Knowledge@Wharton: Professor Siegel, as I recall from your book, you don't just say this is what the numbers average out to over time -- when you talk, for example, about the greater returns of stocks vs. bonds and cash. But you have some reasons why you think that has been the case, right? What are they?

Siegel: Our general models take individuals as what we call risk averse. People have to be paid to take on risks. Since stocks are the residual after bonds and other claimants have their first say, it's natural that stocks would have a higher return. What has tended to surprise economists -- there are a ton of papers that have been written -- is that extra return seems to be very generous over long periods of time. There has been a lot of literature about what's called the equity premium puzzle, written in the mid-1980s by a few economists. Given the macro fluctuations, it seems that equity holders actually get a very rich premium. There have been some modifications and there's been a lot of reworking on that. The premium I find over the long run on stocks over bonds is about 3% a year. And that's a compound annual premium. That premium -- although in the short run it can certainly be up and down over periods of decades and generations and even centuries -- has been relatively constant and, therefore, for someone planning 30, 40 years into the future for their retirement, that difference could obviously accumulate to a very large sum in favor of stocks.

Knowledge@Wharton: Professor Stambaugh, when you look at your findings, regardless of whether volatility may be higher in the future than people expect, what about the relative returns of stocks vs. bonds vs. cash. Do you draw any conclusions that an investor could put to some practical use?

Stambaugh: We've not yet looked at this same issue with regard to nominal bonds where you have things like inflation risk present. So I'm afraid I can't yet tell you what our conclusions would be there. But certainly the issue with regard to asset choice -- if you [want to] make a simple stock versus cash kind of choice or stocks versus a TIPS [Treasury Inflation Protected Securities] or index bond -- our ... best estimate of what the spread would be is what it's been historically.

Knowledge@Wharton: Between stocks and bonds?

Stambaugh: Between stocks and bonds or stocks and cash. All we're saying is that there is uncertainty about whether that average spread will in fact be realized by investors over time. And the uncertainty -- when you take into account all its components -- is indeed higher for longer-term investors.

Siegel: What Robert is saying is very important. These trend uncertainties apply to other assets, too.

Stambaugh: Absolutely.

Siegel: One could say that ... nominal bonds were an inflation trend. We've had inflation --we went up to 13% -- but we know countries that have gone into hyperinflation. The risk of that may be considered a trend uncertainty -- suddenly that breaks down. Rob mentioned TIPS, which are generally more protective because the government promises to pay according to the consumer price index if it isn't manipulated, if there are not price controls -- and if the government can get enough revenue to pay it. There are certainly uncertainties there in the long run. So in some sense, the trend uncertainties that Rob's work shows apply to all other assets. I don't know whether that would lead you -- and I would want to ask Rob about that -- to recommend to people that they reduce their stock allocations as a result of this uncertainty, given that it can, in fact, affect other asset classes, too.

Stambaugh: As I said, we haven't looked at a wide variety of asset classes here. But if one were making an allocation between stocks and cash, and you had a desired allocation ... and then you became aware of this additional uncertainty, you would reduce your stock allocation. Again, we're not making a prescription about what that stock allocation should be. Our point is simply that for a long-run investor this consideration has a bigger effect.

Siegel: But, Rob, have you found the same trend uncertainties affect bonds? Let's say the nominal bonds. Could you make a claim that you would reduce stocks versus bonds under those conditions?

Stambaugh: No. It's quite possible that if we were to look at nominal bonds -- and that would be an extension because there you do have substantial trend uncertainty, particularly with regard to inflation. It's quite possible that this same kind of uncertainty in nominal bonds could well make them less attractive. So that's an extension to our work that we hope to pursue but I just can't give you an answer right now.