The mar­kets clearly aren’t that ef­fi­cient

The Pak Banker - - OPINION - Noah Smith

ONE of the big unan­swered ques­tions in the fi­nance world is: Do re­turns re­flect risk or mis­pric­ing? De­fend­ers of the ef­fi­cient mar­kets hy­poth­e­sis say that you can't get higher re­turns with­out tak­ing more risk, while be­havioural fi­nance says that there are of­ten un­ex­ploited anom­alies that will let wise, pa­tient, or deep-pock­eted in­vestors beat the mar­ket with­out tak­ing on more risk.

This de­bate is very rel­e­vant to the use of fac­tor mod­els. Th­ese mod­els, which are used to de­sign in­vest­ment port­fo­lios with spe­cific char­ac­ter­is­tics, have been one of the most suc­cess­ful meth­ods to come out of aca­demic fi­nance in the past 40 years. Most fi­nan­cial in­sti­tu­tions, and all of the so­phis­ti­cated ones, now use fac­tor mod­els to mea­sure their risk.

Many also use them to op­ti­mise their re­turns. For ex­am­ple, so-called smart beta in­vest­ing strate­gies, one of the most pop­u­lar in­vest­ing fads of the past decade, are mostly just the ap­pli­ca­tion of fac­tor mod­els. You can also now buy ex­change-traded funds that take ad­van­tage of a wide ar­ray of fac­tors.

Fac­tor mod­els ba­si­cally just say that any re­turn that a di­ver­si­fied port­fo­lio earns in ex­cess of the risk-free rate (the rate on Trea­sury bills) should be based on its cor­re­la­tion with a risk fac­tor. For ex­am­ple, small stocks out­per­form the mar­ket on av­er­age, but oc­ca­sion­ally they do much worse.

If you in­vest in a bunch of small stocks, ac­cord­ing to the model, you will beat the mar­ket on av­er­age, but you will also be tak­ing more risk, since you'll lose money dur­ing those times that small stocks un­der­per­form. Ad­vo­cates of fac­tor mod­els claim that there are deep eco­nomic forces at work that make small stocks, as a group, in­her­ently more risky, and thus jus­tify their higher av­er­age re­turns. Some­day, they say, we will un­der­stand what those deep forces are.But what if there is no deep un­der­ly­ing force mak­ing small stocks - or value stocks, or mo­men­tum stocks - more risky? What if th­ese stocks earned bet­ter-thanaver­age re­turns in the past not be­cause they are riskier, but be­cause in­vestors just over­looked them, or were bi­ased against them?

Ef­fects of 'go­ing pub­lic' If this is the case, we should ex­pect that pub­li­cis­ing th­ese fac­tors will lead them to shrink in im­por­tance. As more peo­ple be­come aware of a mis­pric­ing, they trade on it, and the mis­pric­ing goes away. So one way to solve the puz­zle of fac­tor mod­els is to look at how fac­tors per­form af­ter you make them pub­lic. A new re­search pa­per by R. David McLean and Jef­frey Pon­tiff fo­cuses on ex­actly this. McLean and Pon­tiff look at 97 dif­fer­ent fac­tors that fi­nance re­searchers have writ­ten pa­pers about. They find that af­ter the pa­pers are pub­lished, the ex­cess re­turns as­so­ci­ated with th­ese fac­tors go down by about 58 per cent!

One rea­son fac­tors might lose their lus­tre af­ter be­ing trum­peted to the pub­lic ac­tu­ally in­volves data min­ing, rather than any mar­ket in­ef­fi­ciency. Fi­nance re­searchers sift through data to find fac­tors, and they of­ten turn up spu­ri­ous ones. A fake fac­tor will tend to dis­ap­pear shortly af­ter it is "dis­cov­ered", be­cause it was only iden­ti­fied by ac­ci­dent in the first place. McLean and Pon­tiff take this pos­si­bil­ity into ac­count. They com­pare how much the fac­tors de­cline af­ter the dis­cov­ery to how much they de­cline af­ter pub­li­ca­tion in an aca­demic study. What they find is that al­though fac­tors lose about 26 per cent of their strength af­ter dis­cov­ery - an ef­fect that in­cludes the re­sults of data min­ing - they take an­other 32 per cent dive af­ter they ap­pear in aca­demic pa­pers.

This means that a large part of fac­tor dis­ap­pear­ance isn't due to data min­ing alone, but to the pub­lic­ity that aca­demic stud­ies pro­vide. This is a dra­matic and very im­por­tant find­ing. It means, in a nut­shell, that mar­kets aren't nearly as ef­fi­cient as many would like to be­lieve. If fac­tor mod­els re­flect a risk-re­turn trade-off, they shouldn't be af­fected by the ac­tiv­i­ties of academics. Deep-rooted eco­nomic risks shouldn't van­ish just be­cause econ­o­mists dis­cover them. Mis­pric­ings, how­ever, do van­ish when dis­cov­ered.

In other words, this pa­per is a dra­matic con­fir­ma­tion of the pre­dic­tions of be­havioural fi­nance. Can we sal­vage the idea that fac­tor mod­els re­flect risk? Per­haps. McLean and Pon­tiff find that only 58 per cent of the re­turns as­so­ci­ated with fac­tors dis­ap­pear af­ter pub­li­ca­tion - that still leaves 42 per cent. This could re­flect un­der­ly­ing risk as­so­ci­ated with the fac­tors. Al­ter­na­tively, it could be the re­sult of in­vestors' in­abil­ity to fully ex­ploit mis­pric­ings - an ef­fect be­havioural fi­nance re­searchers call "lim­its to ar­bi­trage", Mis­priced stocks could be dif­fi­cult to bor­row and sell short, for ex­am­ple. So al­though McLean and Pon­tiff's pa­per shifts the de­bate in favour of the be­havioural side, it doesn't re­solve it.

Newspapers in English

Newspapers from Pakistan

© PressReader. All rights reserved.