Foxes Rabbits Scorpions CFA 2006

Embed Size (px)

Citation preview

  • 8/4/2019 Foxes Rabbits Scorpions CFA 2006

    1/4

    ,ffi*.Wffim*ffimffiffimffiffiru-BY.HR,'T,NAGROTHEERcord ;

    Evolut ionaryinancesoftendescr ibedsa branch f behav-ioral inance, ut University f ZurichProfessor horstenHens nsistshatevolut ionaryinance tands lone. l f youthin l

  • 8/4/2019 Foxes Rabbits Scorpions CFA 2006

    2/4

    :+ , .Critics ave omplainedhatbehavioralinanceails o sin-gleoutmarketnefficienciesnadvance,rovidingnlyafter-the-factxplanations.s here nyworknprogressthatmight hangehis?

    I hope so, but one should never forget that the so,called nullhlpothesis - the thing that has to be beaten - is thar youcannot predict the market in any way. Given that, I thinkthere is some progress,because here are specific things thatactually do work.There are some behavioral hedge funds, for example,that play very specific strategies,and they generateexcessreturns. For example, at Fuller & Thaler Asset Managementthey play on the earnings,surprise effect, in which the marketreactsmore to negative surprises than to positive surprises.

    There are other very specific things. A colleagueof mine,Bill Ziemba in Canada, s playing on what is cailed probabili-ty weighting - that people care more about small probabili-ties relative to what they should. So he plays on what iscalled the favorite long-shot bias on stock options, and hedoes it verv successfullv

    What are some addit ionalpract ical pplications fbehavioral inance?

    I think one should make a distinction between assetmanage-ment and private banking. Asset management has to do withunderstanding the market as a complex system, which I do inevolutionary finance. Behavioral finance is more tailored tothe intuitive mistakes that you make when investing, and themost interesting area of application is private banking.

    Switzerland is a big place for private banking, righr?So a lot of people in the world give their money to banks inSwitzerland. Then they have to find a way to allocate themoney, not in the Swiss market but in the world market.

    Typical mistakes are that they don't want ro do a iot ofplanning. This is matched by the behavioral concept ofhyperbolic discounting, where you always think it's better tostart your diet tomorrow, so you keep postponing and post-poning. People don't want to face the initial cost of thewealth management process.That is the first major mistake.

    The second mistake is there is a lot of misframing of thesituation. Here, behavioral finance is useful because t can usebalance sheets and personal asset4iability management torevea l o c l ien ts he misf raming .

    The third thing is diversification. There are so many mis-takes that you make when you are naive and you try to diver-sify, and behavioral finance helps to point our those mistakes.Finally, people have difficulty holding to a srraregy.They change their mind every day or every week, and theybasically go south in the market because hey don't followa strategy that has been proven to suit their risk ability andtheir risk preferences.

    Whatsyour olenallof his?We have helped Swiss banks l ike Credit Suisseand DeutscheBank Asset Management do a structured wealth managementprocess,which is based on behavioral finance. They gothrough the processwith clients from one step to the next.

    They point out the common pitfalls and show clients howto avoid them.

    Its quite a successstory, especialiy or Credit Suisse. nprivate banking in Switzerland alone, Credit Suissehas some-thing like 100,000 clients, so you need a strucrured process;otherwise, it is totally arbitrary and the customers wonderwhat they are paying for.

    Where does evolutionary finance fall within the spectrumof behavioral finance?My point of view is you should distinguish between tra-

    ditional, behavioral, and evolutionary finance. Maybe I amtoo stdct on this, but behavioral finance is tailored to individ-ual mistakes and not so much to the markets.

    Evolutionary finance, on the other hand, is useful forassetmanagement because t takes an i.ntegrated iew of inter-actions of strategies,no t only individuals but also institutions- hedge funds and delegatedassetmanagement. Evolutionaryfinance looks at lnteracting strategiesand not at individuals.

    Why oes volutionaryinanceot ocus n ndividualinvestors?Because he logic of the market is that individual investorsdon't matter. Its psychologically difficult becauseyou wanr rosee yourself in the model, but for the market, it only matterswhich strategy s followed by how much capital.

    If you look at the models, what you uncover pretty soonis that assetprices are not derermined by individuals; theyare determined by cash flow between the strategies. n theDarwinian view, what is going on in the forest is not deter-mined by the individual fox and the rabbit but by the relariveproportion of foxes and rabbits.

    Can volutionaryinanceeusedocomplementradition-al assetmanagementrdoest take tsplace?It's a question of what you think traditional assetmanage-ment is. If you think traditional assetmanagement is doingmean-variance calculus and looking at risk factors, as n thecapital assetpricing model, I'm pretty sure ir will take itsplace because he CAPM and mean-variance theory are basedon cross-sectionsof returns. But evolutionary finance is basedon time ss1is5 it's based on dynamical systems. f you wantto invest, in order to get returns, you have to understand thedynamical system you are acring n.

    There asa ot of nformationbout ehavioralinancenthe iterature,utnotmuch bout volutionaryinance.Why s hat?The main reason s becausepeople think there is traditionalfi.nanceand then there is "something else." People don'tunderstand the difference between evolutionary flnance andbehavioral finance because hey already have difficulty under-standing the distinction between traditional and behavioral.So, many of the results from evolutionary finance get suckedup in behavioral inance.

    My second guess s that the mathematics are not onlynew but they are also hard. I'm not a mathematician myself.

  • 8/4/2019 Foxes Rabbits Scorpions CFA 2006

    3/4

    I work with highly trained co-authors: a Russian mathemati-cian and another mathematician based n England. So youn e e d o m e n v e s t m e n tn t o t .

    Would you expect that more people will beg in to follorrvevolutionary finance as it becomes better known?

    I'm pretty sure about this. When we invited DanielKahnneman. winner of the 2002 Nobel Prize for behavioralfinance, to give a presentation here at the University ofZurich, a student asked him about evolutionary finance, andhe said that it is the future. And I'm pretty sure that heknows. The rest of us have to do something else, becauseweall want the Nobel Prize [aughs] and this does seem to bethe future for assetmanagement.

    Howdoes evolutionary finance apply to asset management?First, you need to identify the set of strategies was talkingabout. That's pretty easy because f you are an investmentbanker or a hedge fund manager, you may have tried manysuch strategies.So you make a list of such strategies:activeand passive, distinguished by time hori.zons.

    Whatwere heconclusionsfyour esearch?It does not naively depend on return. That was the first guess- that when a strategy has a high return, everybody umpson the high return. But it depends more on th e rank. Whena strategy is ranked first, it gets more inflows. Ultimately,investors want t o divide their money acrossvarious of thosestrategies,and then they take the best-ranked strategy.

    The first-r anked strate gy will get more money next peri-od. And if you understand what the fund is doing - and thisis known because f they want to attract money, then theyhave to disclose n a sensewhat they are doing - then youcan make predictions on a medium-term horizon.

    Back to the forest analogy: if we have so many f oxes andsuddenly the rabbits die out, we can do sort of a prediction.So last year was a good year for the Swissmarket, not somuch for the US market, and then you can imagine wherethe flows go this year. When you see the interaction of thestrategies,you can guess t. Of cour se, you need a lot ofeconometrics in order to find the corre ct values, but youcan make some educated guesses.

    There are some successstories.The Duke of Liechtenstein

    Then what's more difficult is you h ave to get some num-bers for the relative importance of those strategies.Fortunate-ly, we have some good data now from State Street on whichtrades have been done by which investors, so we can classify .the strategiesby the investors and the wealth.

    Then, you try to calibrate the model so that it fits to .what you have observed. Finally, the more difficult step is togo one period ahead- to look at the next week or the nextmonth. We have good results on a monthly basis but not somuch on a weekly basis.

    So you have strategies,you have the wealth at any pointin time, and then you have to figure out what you call the jwealth-flow function: when does a certain strategy attractmore wealth than some other strategy?

    This is quite well known, and its quite robust, actually.We can observe this with hedge funds or mutual funds, forexample. We have some nice criteria to seewhen one strategy .of a hedge fund attractswealth and when a certain strategydries out. We did it in the laboratory as well to look at otherflow functions - they look quite similar, pointing at a"uni.versallaw" almost like those known in physics. :H#j+

    has a bank called LGT (Liechtenstein Global Tiust). Wehelped create a fund for him based on these evolutionaryprinciples. For five years now it has worked great.

    Hpwdoes evolutionary finance relate to geneticalgorithms?A genetic algorithm is one strategy to solve complicated opti-mization problems. There has been some strand of the litera-ture - Holland and others at the SantaFe Institute - whothought maybe the mean-variance calculus is too simple andwe should do some more sophisticated things, like geneticalgorithms, in order to frnd the best investment strategies.But the problem with this is, as with traditional or behavioralfinance, it's only a partial view; it's only focusing on one classof strategiesnow generated by genetic algorithms.

    Evolutionary finance takes nto account those strategies,and they don't always do very well. They do better thanmean-variance, which we can observe and also prove mathe-matically, but it's only one ingredient in a pool of strategiesinterac tng wi th other s l ra teg ies .

    40

  • 8/4/2019 Foxes Rabbits Scorpions CFA 2006

    4/4

    Workingrom nevolutionaryerspective,IT'sAndrewohas aid hat innovationis hekev o survival."Would ouagree?I think it depends a bit on the time horizon. lf you are a hedgefund manager - if you want to generate return over a short-term horizon - you h ave to be verf innovative because t mayjust happen that your strategv s cover ed by others so thateverybody umps on it and the -r he excess eturns go away.

    But you should also keep in mind that if you are a differ-ent animal, you can do ven- clifferent strategiesand they willalways work. For example. \\hrren Buffett is doing simplethings, and th ey alu,a vs s orli. This is again similar to evolu-tionary biology Did lou knon that scorpions have been theway they are for more tl-ran iOtlmillion years? They neverchange; thev ar e r,rstgooclat ri'hat they do.

    Why has he role of delegated nvesting argelybeenignored n f inance?

    In the standard theor1,, here is onl,v one person who decideson evervthing as a representatir,eagent. It cannot build indelegated assetmanagement, becauseyou need at least one

    How ften oanomaliesoundn he aboratoryndupbeing xploitablen real-life arkets?I onLy know the two or three successfulbehavioral hedgefunds mentioned earlier. I know others who have tried to godirectly from the laboratory to the market, but they basicallywasted their mone1..The strength of laboratory research sthat )rou can isolate the aspectsyou are interested in. But areal market is just too complex compared with what you aredoing in the laboratory.

    I would be wary of betting my money on a partialaspect found in a laboratory experiment, because here maybe a dozen other aspectsgoing exactly the other way around.I'm saying this even though I know my colleagueswill hateme for saying it, because t's so nice to say,"Look, we havefound this in the laboratory, and now we'll go to Wall Street."flaughsl /Christina Grotheer s a contnbu ting editor and an editorialconsultantto CFA Masazine.

    ' l;,t . .

    who delegatesand one who does the management. I thinkthis methodology is total nonsense.becauseyou have to studythe interactions of various decisions that are taken. A singlerepresentativeagent ust doesn't fit.

    If you look at what people in phl,srcsor biology aredoing, they would say what we are dorng are not really "mod-els." They would say it's a Mickey Mouse thing. Here at theUniversity of Zurich, they have 20 researcherswith variousbackgrounds doing one model to study a very specific type ofcancer that you get in your belly In flnance, we do a modelon our ou'n and we maybe have two students helpi.ngus, soit's not a coordinated effort as n the natural sciences.

    I think n'e har.e o improve on this. We need to do moreobservations and computer simulatlons in order to get goodmodels. The Iinancral market is even more complex than acancer, hough th e rnathematics s the same.That'swhy I metwith colleagues rom the cancer research,because hey are alsousing random dynarnical s\'sterns in this case.describingthe relative speed of grorvth of healthy and cancerous ceils.

    Why ocus none ieldwhen oucando wo?ThorstenHens, professort he nstituteorEmpirical esearchin Economicst he Universityf Zurich,urrentlyspursuingesearchn bothbehavioralndevolutionaryfinance3opapers ndcounting).

    Hens lso erves sanadjunct rofessorn heNorwegianusinesschool'separtmentf inancenBergen, orway.t he National enter f Competencein Research,e s hescientificirector f FinancialValuationndRiskManagementFlNRISK),orks saCenteror Economicolicy esearchellow, ndcollabo-rates n a Europeancienceoundationroject.

    Having arned isPhD romBonnUniversity(Germany)n r9gz,Hens asheldpositionstseveraluniversitiesn Germany,n additiono work n Parisandat Stanford niversity.M A R C H - A P R I Lo o 6F * M A G a Z r n el :