rise and fall of performance investing

8/11/2019 Rise and Fall of Performance Investing

1/5

November/December 2012 www.cfapubs.org 15

Financial Analysts JournalVolume 68 Number 6

2012 CFA Institute

P E R S P E C T I V E S

An Experienced View on Markets and Investing

Eugene F. Fama and Robert Litterman

At the 65th CFA Institute Annual Conference in Chicago (held 69 May 2012), Robert Litterman inter-viewed Eugene F. Fama to elicit his views on financial markets and investing.

Litterman:You were at the center of the devel-opment of two very powerful ideas: marketefficiency and equilibrium riskreturn rela-

tions. Why are they such important concepts?

Fama:Market efficiency says that prices reflectall available information and thus provide accuratesignals for allocating resources to their most pro-ductive uses. This is the fundamental principle ofcapitalism. To test market efficiency, however, weneed a model that describes what the market istrying to do in setting prices. More specifically, weneed to specify the equilibrium relation betweenrisk and expected return that drives prices. Thereverse is also true: Almost all asset pricing modelsassume that markets are efficient. So, while someresearchers talk about testing asset pricing mod-els and others talk about testing market efficiency,both involve jointly testing a proposition aboutequilibrium risk pricing and market efficiency. Thetwo concepts can never be separated.

Litterman:In developing these ideas, was therea particular point at which the light went on, soto speak?

Fama: Yes. When I was an undergraduate,I worked for a stock market forecasting service.My job was to devise mechanical trading rulesthat worked. But the rules I devised worked onlyon past returns; they never worked when appliedgoing forward or out of sample. After two years of

graduate school, I started talking to Merton Miller,Lester Telser, and Benot Mandelbrota frequentvisitor at the University of Chicago. They werethrashing around the idea of what prices wouldlook like if markets worked properly. That was my

path into research on market efficiency and equilib-rium riskreturn models.

As the market efficiency ideas took shape, itdawned on me that the reason the trading rules Id

developed earlier didnt work out of sample wasbecause price changes were random, which at thatpoint was what people thought an efficient marketmeant. We know now it doesnt. Market efficiencymeans that deviations from equilibrium expectedreturns are unpredictable based on currently avail-able information. But equilibrium expected returnscan vary through time in a predictable way, whichmeans price changes need not be entirely random.

Litterman:You and I share one rather uniqueexperience. Both of us have had an office next toFischer Blacks office. Do you have any interestingmemories that you might share?

Fama:We both would arrive at our offices veryearly in the morning. This was during the time thefirst tests of the capital asset pricing model (CAPM)were being done. Fischer, Myron Scholes, and MikeJensen were suspicious of the latest test resultsbecause it looked like the standard errors of themeasured premium for market beta were too lowrelative to what they knew to be the volatility of thestock returns. So Fischer devised this elegant tech-nique for forming portfolios that would capture thepremium and would allow for the cross-correlationof returns, which had been the major problem with

the earlier tests of the model.1

When he showedthe technique to me, I said, Thats great, Fischer.You just discovered regression. He said, No, no,I didnt. I said, Yes, yes, you did. No, I didnt.Yes, you did. And on it went.

Finally, in response to this debate, I wrote whatbecame Chapter 8 of my textbook2 to essentiallyexplain to him the portfolio interpretation of aregression that attempts to explain the cross-sectionof stock returns. The regression approach, whichwas first used in my 1973Journal of Political Economy

Eugene F. Fama is the Robert R. McCormick Distin-guished Service Professor of Finance at the Universityof Chicago Booth School of Business. Robert Littermanis executive editor of the Financial Analysts Journal.


2/5

Financial Analysts Journal

16 www.cfapubs.org2012 CFA Institute

paper with Jim MacBeth,3has since become a stan-dard method for testing for factor return premi-umsfor instance, the premium associated withsmall-cap stocks.

Fischer was one of the rare people (MertonMiller was another) who could cause you to thinkentirely differently about something youve been

working on. Its hard to think of anything morevaluable when youre doing research. Your col-leagues play a very important role in your research.You cant work in a vacuum or you waste a lot ofvaluable time. For example, I was in Belgium fortwo years working solo. When I returned, I showedMerton Miller my research produced over that two-year period, and he put aside most of it with thecomment, Garbage. He was right on every count.

Litterman: I had exactly the same experiencewith Fischer. Now, lets turn to more recent events.What do you think caused the global financial crisis?

Fama:I think the global crisis was first a prob-

lem of political pressure to encourage the financ-ing of subprime mortgages. Then, a huge recessioncame along and the house of cards came tumblingdown. Its hard to believe that without a pretty sig-nificant recession, the financial system would havecome crashing down like it did. Subprime wasbasically a U.S. phenomenon, yet the crisis spreadaround the world. Financial institutions in othercountries were certainly holding subprime debt,but in the past, financial institutions have gonebust because they made dumb decisions withouttriggering widespread crisis. I dont think the cri-sis was a problem with markets. The big recession

was the trigger. The worst thing to come out of thatexperience, in my view, is the concept of too bigto fail.

Litterman: I totally agree. Can you elaborateon that?

Fama: Basically, the institutions that are con-sidered to be too big to fail have their debt pricedas if its riskless, which gives them a low cost ofcapital and makes it very easy for them to expandand become an even bigger problem. Plus, every-body now accepts the assertion that they are toobig to fail, which creates a terrible moral hazardfor the management of these financial institutions.Business leaders wont consciously tank their com-panies, but too big to fail will push them towardtaking more risk, whether they realize it or not. Idont think DoddFrank (the DoddFrank WallStreet Reform and Consumer Protection Act) curesthat moral hazard problem. Even if lawmakerscould devise the perfect regulation for such a cure,the chance that it will be implemented by the regu-lators in the way designed is pretty close to zero.

The simplest solution would be to raise the capitalrequirements of banks. A nice place to start wouldbe a 25% equity capital ratio, and if that doesntwork, raise it more. The equity capital ratio needsto be high enough that a too-big-to-fail financialinstitutions debt is riskless, not because of whatis essentially a government guarantee but because

the equity ratio is very high.Litterman:Do you think that the Volcker Rule(eliminating proprietary trading by commercialbanks) will be successful?

Fama:I dont believe that the Volcker Rule willaccomplish much. It wasnt just the commercialbanks that were bailed out in 2008. The investmentbanks, where the majority of risky trading was tak-ing place, were bailed out too. If the commercialbanks are prohibited from proprietary trading, theinvestment banks will still be doing it, and as prec-edent has shown, they are considered too big to fail.So they would be bailed out if something were to

go wrong. Nothing really changes with the VolckerRule.

Litterman:Let me ask you about another issue.A large fraction of state and local pension plans aresignificantly underfunded. How did that happen?

Fama:Many plans are underfunded, of course,because the sponsors simply didnt put enoughmoney into the plans. More important, the situa-tion is much worse than they admit. The reality isthat the liabilities they claim to have are about athird of the actual liabilities. They understate thecurrent value of liabilities because the sponsor dis-counts the liability stream at the assumed expected

return on the risky assets held by the fund. Thesponsor should be discounting the liabilities at theexpected return implied by the risk of the liabilities,not the expected return on the assets. The liabili-ties are basically indexed claimslike a TIPS (Trea-sury Inflation-Protected Security). Therefore, theappropriate discount rate on the high side shouldbe about 2.5%, not the 7% or 8% that the plans areusing now. Even using a 3% discount rate wouldat least double the value of liability compared withthe declared liability.

So, what does this all this mean? The plansponsors are betting that good returns on the riskyassets they hold will bail them out. The problem isthat over the lives of these plans, there will be mul-tiple instances of bad markets, and the plans haveto fulfill their pension obligations in bad marketsas well as in good markets. Thats where the riskbites. The state and local pension funds are basi-cally giant hedge funds. If that makes you uncom-fortable, it should.

Litterman:So, whats the solution?


3/5



Fama:One solution might be to finance definedbenefit plansas Fischer Black always contendedsolely with debt in order to best match the liabilitiesand the assets.4But that wont happen because theplans would have to lower their discount rates toreflect the current level of interest rates and then itwould be obvious that the plans are so far under-

funded that they are basically bust.The crisis the country faces is that eventuallya big state is going to go bust because of its pen-sions. State constitutions typically provide that thestate first has to service its debt, then make its pen-sion payments, and then pay for services. What wedont know is whether that order will be enforced.And ultimately, the busted state is going to be look-ing to the federal government for a bailout. ThinkGreece, but on a much bigger scale.

Litterman:Youve just published research thatshows that before costs, mutual fund managerssystematically take alpha from other managers.

Fama:Yes, thats a concept I hope everybodyunderstands. Active management in aggregate isa zero-sum gamebefore costs. Good (or morelikely just lucky) active managers can win only atthe expense of bad (or unlucky) active managers.This principle holds even at the level of individualstocks. Any time an active manager makes moneyby overweighting a stock, he wins because otheractive investors react by underweighting the stock.The two sides always net outbefore the costs ofactive management. After costs, active manage-ment is a negative-sum game by the amount of thecosts (fees and expenses) borne by investors.

My research with Ken French shows that,examined before costs, the return distribution forthe universe of all mutual funds has a right tail anda left tail, both tails are about the same size, andthe distribution is centered at zero. In other words,the before-cost return distribution looks just aboutas you would expect if on average active manag-ers have no skill but there are both good and badmanagers. After costs, only the top 3% of manag-ers produce a return that indicates they have suf-ficient skill to just cover their costs, which meansthat going forward, and despite extraordinary pastreturns, even the top performers are expected tobe only about as good as a low-cost passive indexfund. The other 97% can be expected to do worse.

Litterman: So, what conclusions should wedraw from this?

Fama:That an investor doesnt have a prayer ofpicking a manager that can deliver true alpha. Evenover a 20-year period, the past performance of anactively managed fund has a ton of random noisethat makes it difficult, if not impossible, to distin-guish luck from skill.

Litterman: Let me ask you about the equityrisk premium. Why has it been so high in the his-torical data?

Fama:In a paper Ken French and I wrote,5weobserved that P/Es (price-to-earnings ratios) haverisen over the period covered by the CRSP files.P/Es can rise because expected future profitabil-

ity has risen or the discount rate (i.e., the expectedreturn on stocks) has fallen. Because we couldntfind any evidence that earnings growth is predict-able except over very short horizons, we concludedthat high P/Es must be due to a decline in discountrates. The decline in discount rates means that pastaverage stock returns were higher than expected,but going forward, a lower discount rate meansthat expected returns on stocks are lower.

I think the right equity risk premium goingforward is about 4%, but a number that high stillbothers many economists. Macroeconomists havea problem with the observed equity premium

because their consumption-based models predictthat the premium should be between 0.5% andperhaps 1.5%. My question to them is, Who wouldhold stocks if the premium were only 1%? I dontknow anybody who would.

Litterman: Does that mean that people areirrational?

Fama:No, I think it means that the economistshave bad models. Their models apparently dontcapture a risk in holding stocks that leads to highexpected returns.

Litterman: What are the implications of thevery low current real interest rates?

Fama: Investors around the world currentlyseem to have a strong preference for low-risk assets,and this drives down the real rate on such assets.

Litterman: Until 1992, when you and KenFrench published your seminal paper on the cross-section of expected returns that challenged theveracity of the CAPM,6 you seemed to be a verystaunch supporter of the CAPM. What changedyour mind?

Fama: The CAPM had been around since theearly 1960s, and tests of it started in the early 1970s.In the 1980s, research began to suggest that theCAPM had a problem explaining the returns onsmall stocks, low-P/E stocks, and high-dividend-to-price stocks. Each variable was examined in aseparate paper, and with each new paper, the pre-vailing wisdom was that the newest failing wasone problem for the CAPM but that in general itstill performed pretty well. In our 1992 paper, welooked at all the variables together and concludedthat the CAPM was fundamentally broken. I didntexpect the 1992 paper to be published because therewas nothing new in it. Every result existed in the


4/5

Financial Analysts Journal

18 www.cfapubs.org2012 CFA Institute

literature. But just by putting all the pieces together,we were able to break new ground.

Litterman:It is probably the most cited paperin finance.

Fama:It is the most cited paper in finance overthe last 20 years. After that, in 1993, we publishedanother paper that introduced the so-called three-

factor model. Here, we added to the CAPM mar-ket beta a size factor and a valuegrowth factor.7This model is now widely used both in academicresearch and by practitioners.

Litterman: Do you view the value premiumand the size premium as risk factors?

Fama: I do. They are both associated withcovariation in returns that cant be diversifiedaway. These sources of variance seem to have dif-ferent prices than market variance has; otherwise,the CAPM would work. Because these factors arescaled price variablessuch as the book-to-market,earnings-to-price, and dividend-to-price ratiosthe

implication is that the price contains informationabout expected return that isnt contained in themarket beta. These ratios are a good way to identifyvariation in expected returns across stocks, even ifwe dont know the true sources of the variation.

Litterman: What about momentum? Isnt itpretty hard to argue that its a risk factor?

Fama: There is covariance associated withmomentum. But the real challenge from momen-tum is that if it represents time variation in the risksof stocks, it is discouraging because the turnoverof momentum stocks is so high. Of all the potentialembarrassments to market efficiency, momentum

is the primary one.Litterman:In the past, youve said that behav-

ioral finance is ex post storytelling and doesntgenerate new testable hypotheses. With the devel-opment of this literature, has your view changedabout that?

Fama:I think the behavioral finance literature isvery good at the micro levelindividual behavior.But the jumps that are made from the micro levelto the macro levelfrom the individual to mar-ketsarent validated in the data. For example, thebehavioral view is that a value premium exists andits irrational. If its irrational, it should go away,but it doesnt seem to have gone away. Behavioralfinance also claims to explain momentum andreversal. Thats too flexible in my view. Its not ascience. In Daniel Kahnemans book Thinking, Fastand Slow,8he states that our brains have two sides:One is rational, and one is impulsive and irrational.What behavior cant be explained by that model?

Litterman:Whats your view of the purportedexcess return of low-volatility stocks?

Fama:The excess return is really a result of lowbeta, not low volatility, and this potential sourceof return has been well known for 50 years. Whenthe first tests of the CAPM were done, the problemalways out front was that the market line, or theslope of the premium as a function of beta, wastoo low relative to what the model predicted. This

meant that low-beta stocks had higher returns thanpredicted and high-beta stocks had lower returnsthan predicted.

Litterman:Wasnt one of Fischer Blacks origi-nal ideas to exploit this anomaly?

Fama: Fischers idea was that the problem inthe original CAPM was that it assumed that bor-rowing and lending were at the risk-free rate. Hesaid if that assumption were thrown out, then allwe would know about the premium for marketbeta is that its positive.

Litterman: If all investing were to be passive,what would make markets efficient?

Fama: In the extreme, all a market needs inorder to be efficient is an epsilon (a tiny proportion)of wealth held by perfectly informed investors whotrade actively and aggressively. If an efficient mar-ket needs more good active investors, it is to cor-rect the mistakes of bad active investors. And thisdoes not at all mean investors should be buyingactive management. Because of the large amountsof random noise in the returns of active manag-ers, investors wont be able to tell the good (butunlucky) from the bad (but lucky) active managers.To the extent that they mistakenly give money tobad active managers, investors make markets less

efficient. Perhaps most important, as Bill Sharpewarned us years ago (in the pages of the FAJ), it isa matter of arithmetic that investors who go withactive management must on average lose by theamount of fees and expenses incurred.

Litterman:What do you think of the risk parityasset allocation strategy? A risk parity asset alloca-tion strategy chooses weights in all asset classes ina portfolio so as to create the same level of volatilityor risk in each. That portfolio can then be leveragedto equal the risk of, say, a 60/40 equity/bond allo-cation.

Fama:I dont think much of that approach. Younever start with a proposition like thatequalizingrisk of assets in a portfolioas the way to solve theportfolio problem. A meanvariance investor aimsto form asset allocations that minimize the varianceof the return of the entire portfolio given the levelof expected return, which almost surely does notimply levering risk up to equate volatility across allthe asset classes in the portfolio. A risk parity port-folio almost surely represents a suboptimal solu-tion to the portfolio optimization problem.


5/5



Litterman:What impact will the big expansionin the Federal Reserves balance sheet have on themarkets?

Fama:It has basically rendered the Fed pow-erless to control inflation. In 2008, when LehmanBrothers collapsed, the Fed wanted to get the mar-kets moving and made massive purchases of secu-

rities. The corollary to that activity, however, isthat reserves issued by the Fed and held by banksexploded. An explosion in reserves causes anexplosion in the price level unless interest is paidon the reserves. So, the Fed started to pay intereston its reserves, which means that the central bankissued bonds to buy bonds. I think its a largelyneutral activity.

Before 2008, controlling inflation was a mat-ter of controlling the monetary base (currency plusreserves). But when the central bank pays intereston its reserves, it is the currency supply that deter-mines inflation. But banks can exchange currency

for reserves on demand, which means the Fed can-not control the currency supply and inflation, orthe price level, is out of its control. The Fed had thepower to control inflation, but I dont think it doesunder the current scenario.

Litterman: How does that relate to the debtissues that the United States is facing?

Fama: The debt issues are entirely different.The debt issues are about how much we want tosacrifice the future for the present and whether we

get anything in the present for the future were sac-rificing. This has been the big debate between theKeynesians and the non-Keynesians since 2008.

Litterman: But isnt one way out of our debtproblem to inflate it away?

Fama:Yes, thats one way to handle it, but itsfar from a great solution. If the Fed were to stoppaying interest on its reserves, wed probablyhave a big inflation problem. The monetary basewas about $150 billion before the Fed stepped inin 2008. Currency plus required reserves are stillin that neighborhood, but the Fed is holding $2.5trilliontrillion!worth of debt financed almostentirely by excess reserves. The price level couldexpand by the ratio of those two numbers, and thattranslates into hyperinflation. Economies typicallydo not function well in hyperinflation. The realvalue of the government debt might disappear, butthe economy is likely to disappear with it.

Litterman:What would your suggestion be for

monetary or fiscal policy at this point?Fama: Simple. Balance the budget. I heard avery prominent person say in private that we couldbalance the budget by going back to the level ofgovernment expenditures in 2007. The economyis currently about the size it was then. If you justrolled expenditures back to that point, I think itwould come close to balancing the budget.

This article qualifies for 0.5 CE credit.

Notes

1. Fischer Black, Michael Jensen, and Myron Scholes, TheCapital Asset Pricing Model: Some Empirical Tests, in Stud-ies in the Theory of Capital Markets, edited by Michael Jensen(New York: Praeger Publishers, 1972).

2. Eugene F. Fama, Foundations of Finance: Portfolio Decisions andSecurities Prices(New York: Basic Books, 1976).

3. Eugene F. Fama and James D. MacBeth, Risk, Return, andEquilibrium: Empirical Tests, Journal of Political Economy,vol. 81, no. 3 (MayJune 1973):607636.

4. Fischer Black, The Plan Sponsors Goal, Financial AnalystsJournal, vol. 51, no. 4 (July/August 1995):67.

5. Eugene F. Fama and Kenneth R. French, The Equity Pre-mium,Journal of Finance, vol. 57, no. 2 (April 2002):637659.

6. Eugene F. Fama and Kenneth R. French, The Cross-Sectionof Expected Stock Returns,Journal of Finance, vol. 47, no. 2(June 1992):427465.

7. Eugene F. Fama and Kenneth R. French, Common Risk Fac-tors in the Returns on Stocks and Bonds,Journal of FinancialEconomics, vol. 33, no. 1 (February 1993):356.

8. Daniel Kahneman, Thinking, Fast and Slow (New York: Farrar,Straus and Giroux 2011).

rise and fall of performance investing

Documents