Kenneth French

Interview with Kenneth French

Kenneth French

Kenneth R. French is the NTU of Professor of Finance at the MIT Sloan School of Management. He is an expert on the behavior of security prices, investment strategies, and the management of financial risk. His recent research focuses on tests of asset pricing models, the trade up between risk and return in domestic and international financial markets, the cost of capital and the relation between capital structure and firm value. Professor French is past director of the American Finance Association, a research associate at the National Bureau of Economic Research and an associate editor of the Journal of Finance, the Review of Financial Studies and others.  Despite this impressive history, he defers, modestly, to Eugene Fama:  "Our partnership is supposed, total misunderstanding because Gene turned in this good stuff."

Kenneth French and Eugene Fama are credited with identifying multiple risk factors in the stock market and developing the three-factor model to measure different types of risk.  This three-factor model changed the world of finance.  "I guess we were trying to answer the question: If you were trying to form a portfolio with high expected returns or low expected returns, how would you go about doing that? At the time, the capital asset pricing model was the basic theory that said high beta stocks--high expected returns, low beta stocks--low expected returns. And so we looked at that and we looked at a bunch of other things that people had already identified and what we discovered was, gee, beta didn`t seem to work very well, knowing the stocks beta didn`t seem to tell me anything about what its average return was going to be." 

French remembers that others had already developed results indicating that small stocks tend to buy average returns more than big stocks.  "And the result was that variables, like the ratio of the book value of equity to the market value of equity, mattered a lot in terms of identifying stocks with high expected returns and stocks with low expected returns. What we`ve discovered since then is there`s no magic about book-to-market. You can measure it with dividend yield, earnings price, cash flow to price, basically anything where you have some fundamental value in the numerator and price in the denominator. So, it`s a way to scale price, basically, and the way I like to think of it is, we`re looking a discount rate. You get a discount, for example, for future cash flows at the expected return on the market. If you have a high-expected return, you get a high cash fair price. So a high cash fair price maps in higher expected return. Basically, it`s using the idea that the expected return that we as investors are looking at on the stock is the same thing as the discount rate or the cost of capital that the firm has to be thinking about.  That`s an easy way to identify differences in expected returns."

Since the three-factor model seems to be so effective, investors may be wondering if the capital asset pricing model is no longer relevant.  "That`s a tough question.  The evidence is pretty strong that as far back as we can see, there seems to be little relation between beta, the fundamental variable of the capital asset pricing model, and average returns on stocks.  Maybe it`s my upbringing, but if the argument is so compelling that stocks that vary a lot with the market bring a lot of risk to people`s portfolio, they`re bringing a lot of risk, people are going to demand a higher premium. So, I`m not willing to say no, there`s nothing that the cap end tells us about differences in expected returns, but what I think we can say is, you have to add other variables.  In addition to beta, I think what matters is sensitivity to what we call size risk and then, sensitivity to something we call distress risk . And the size risk, it`s basically the size factor we see. Small stocks, again, have more of this size risk and more of the expected return. The distress risks, that`s the book-to-market, or the cash flow to price, earnings price, that`s that variable that we`re talking about. Companies that are really sick, bad opportunities, poor investments, they have a higher expected return."  Professor French opines that investors are seeking a premium when investing in a company with poor prospects.  "Companies that have great opportunities, very robust, things are going well in their industry, it appears that the market is willing to invest at a lower expected return for those companies."

The subject French spoke of briefly, the subject of size, is more complex than one might expect.  Most people can intuitively accept size as a risk factor, but seem to have more difficulty understanding the relationship between book-to-market ratios and risk.  "Well, small stocks tend to be more volatile than big stocks, so it`s natural for people to say, oh, this higher volatility, I`m going to require a higher expected return for that higher volatility.  We don`t see that when we`re looking at stocks sorted on book-to-market. Basically, high book-to-market portfolios seem to have roughly the same volatility as low book-to-market portfolios, and what that says is you need a multi-factor model to really capture these differences and expected returns. What you need is a model that says okay, there`s risk associated with movements in the market. That`s beta risks for the capital asset pricing model. There`s risk associated with the movements of small stocks relative to big stocks. That`s the size risk that we`re talking about and then this third dimension, what we`re calling distress risk, that`s how do I move with stocks that seem to be more distressed compared to stocks that are more robust. 

"At this point, my thinking on this is evolving. I`m not so convinced anymore it`s really the distress risks, but rather an agglomeration of all sorts of risks. Remember, stocks with high expected returns, they`re going to have high ratios of their book value to market value or high ratios with cash flow to price. So, whatever the sources of risk, as long as there are differences in risk leading in the differences in expected returns, they ought to show up in these sorts of ratios. 

"Taking it one step further, I don`t really even need differences in risks. Whatever the reasons for differences in expected returns, they`re going to show up in these sorts of ratios. I like to think the world is an equilibrium. I like to think the market works pretty well so prices are pretty close to right. In that case, when I see differences in expected returns, it`s coming because of differences in risks, but it doesn`t have to. If in fact, the market just screws up, set some prices too high, some prices too low, those mistakes will show up in these ratios as well."

As would be expected, French credits past researchers with providing a base for his and Fama`s work. French and Fama`s work, and the risk dimensions identified, are universal enough to be recognized in markets other than the United States.  "For example, the first book-to-market research was actually done by Chan, O`Malley, DeConoshaw, in Japan, I think it was back in 1991, that they published their paper.  That was before our results on book-to-market in the US, so, the international evidence actually preceded the evidence in the US.  Since then, other people have done work showing that stocks with high book-to-market ratios have average returns. Fama and I have done it, through major markets, we`ve done it for emerging markets. It seems to show up everywhere, and in fact, in working with Jim Davis, Fama and I have gone back to 1926 and found the same results from `26 to `63.  It`s remarkable how close the premiums are from `26 to `63 versus the `63 on evidence that Gene and I did originally. Similarly, when we`re looking internationally, the US is right in the middle of the 12 international countries that have data over the whole time period that we look at. So, it looks like the US is typical of what`s going on around the world, not atypical."

Publishing the results of their research exposed French and Fama to the criticism of both the academic community as well as the investment industry.  What kind of opposition did they face with their ideas?  "The academic response was, our results, the research is screwed up!  (The academics said), clearly this is wrong, perhaps there were just flaws in the approach Gene and I used. Maybe it`s just the result of data mining.  If you have enough people searching over the same data over, over and over again, somebody`s sure to find patterns and so one claim was, this is just random, happened by chance.

"The fact that we have all of this international evidence, the fact that we have that evidence from `26 to `63, basically that (puts) the data mining complaint to rest.  The concerns about the quality of our research, that we made mistakes, a bunch of people have pursued those arguments, (and) consistently found that if they dot the I`s and cross their T`s, they get the same results we do." 

The book-to-market value effect has become widely acceptable, French says.  "I think the bottom line is that there is a book-to-market or a value effect. It`s widely acceptable. The academics have seemed to agree, the practitioners that aren`t running growth portfolios seem to agree.  And I suspect that we`re never going to convince them what they`ve been doing!  Buying low expected return stocks for the last 25 years, despite the performance of their portfolios?!

French believes the consensus is almost unanimous now in the academic market that there is a real book-to-market effect.  The new debate is over why.  "Some of us think it`s probably mostly risk; other people are thinking it`s probably mostly mistakes in the market. That`s where the academic debate is (centered).  I`m not quite sure of all the ramifications for institutional investors, but one of the things that`s come out of it, I`ve alluded to, (is) this three-factor model."  Institutions are reporting it a great way to frame their portfolio allocation decision. "Rather than worry about lots and lots of different dimensions, people have discovered we can summarize it, we can collapse it down into:  How sensitive am I to movements in the stock market?  What`s my size tilt, do I look more like small stocks or big stocks, what`s my value versus growth tilt?  Do I look more like valued stocks or more like growth stocks? With those three dimensions, you can capture an enormous amount of what`s going on in a portfolio." 

Speaking academically, this is all very interesting and valuable.  However, on a practical level, what would the relevance of such research be to financial advisors and their clients?  Professor Fama believes the model equally useful for academics and investors.  "Again, I think it`s a great way to frame the portfolio allocation decision. I can look at it and say, am I comfortable with this exposure to the overall stock market? I can look at it and say, am I making the right trade-off, between the expected return I get from buying small stocks and the risk that brings? And then, am I making the right trade-off between the expected return I get from buying distressed stocks and the risk that that brings? By answering those three questions, I frame that portfolio decision in a really easy way, at least for me, to think about." 

With such an important tool, it may be possible to create the one thing that all investors quest for all their lives.   It may be possible to construct optimal portfolios using this three-factor model!  Unfortunately, no!  You should know by now, there is no such thing!  "I can construct a large set of portfolios that are optimum, but the model won`t tell you this is the right portfolio.  In the end, it comes down to a question of taste.  How or what is your taste for risks versus expected return? How scared are you and how greedy are you? And I can`t tell you that. So& "  It is reassuring to know that Professor French agrees individual goals will still be involved.

Asked what the expected premium is for investing in high book-to-market stocks versus growth stock, Professor French thinks aloud.  "It depends on how you define high book to market or value stock, compared to a growth stock, but we typically talk about the top 30 percent, for example, sorted on book-to-market and the bottom 30 percent. If I look at that spread between a valuated portfolio at the high end and a valuated portfolio at the low end, the historical evidence is that spread somewhere on the order of, oh, five or six percent.  Gene and I tend to be a bit more conservative, and we expect something like three and a half or four percent. So we always like to shrink back toward more typical numbers, so my guess is three and a half, four percent." 

A fear for some may be that many people are aware of this premium now, and it could threaten to disappear.  Professor French does not indulge this fear.  "If it is simply mistakes in the market, one might expect some of the premium to go away, but if any mistakes were made, you`d expect this sort of ratio. Remember what we`re looking at here, is a ratio that`s going to discount those cash flows back to the present. If there are differences in expected returns, they ought to show up in that ratio.  So, if there are mistakes in the market, identifying them ought to make some of them go away, but I suspect that we`re never going get to a world where there are never any mistakes in the market.

"On the other hand, if what we would have done here is simply identified differences in risk, there`s no reason for the differences in expected return to go away. Any more than when Bill Sharp invented the capital asset pricing model, and it said high beta stocks should have high-expected returns, that didn`t make anybody feel like, gee, my portfolio ought to adjust behind beta stocks.  It was a statement that said, there`s going to be the correct trade-off between risk and expected returns and if I`m willing to take the risk, I get the premium. If I`m not willing to take the risk, I don`t get the premium. None of that should drive the premium away." 

Just how long does it take, how many years of data, to identify a risk premium?  "There`s really two questions there. One has to do with riskiness, which academics call covariances. The inclination of one stock to move with another portfolio, the tendency of stocks to move together. You can identify covariances answers with relatively short periods. For example, people often use five years of monthly data to estimate beta, so if I wanted to know, is my stock very sensitive to movements in the market, I could use five years of data to answer that question very confidently. If on the other hand what we`re trying to say is not is this a risk factor in the sense that it tends to move with something, but rather is there a reliable risk premium? That takes a long time. It depends on the magnitude of the premium and the volatility of the factor, but, you typically would need, perhaps 25, 30, 40 years to be able to confidently say yeah, this premium here is really different from zero. Again, it depends on the magnitude of the premium and the volatility, but 20 to 30 years is not an unreasonable number."

Value an investor to be confident for investing right!  Another long term puzzle French:  How many years would it take?

"But, if you want to be absolutely certain, you are going to have wait until infinity."