Perils of past per­for­mance

Finweek English Edition - - Property Compass - Al­liance Bern­stein

GIVEN THE SIG­NIF­I­CANCE of un­der­per­for­mance of ac­tive man­agers be­tween 2007 and mid-2008 it would hardly be sur­pris­ing if clients were ques­tion­ing whether man­agers are worth their fees. A few may be ready to ditch the pi­lot. But are clients in dan­ger of jump­ing to hasty con­clu­sions? Given the com­plex­i­ties of as­sess­ing man­agers’ per­for­mance sack­ing them may be sat­is­fy­ing but ul­ti­mately self-de­feat­ing. Aca­demic stud­ies sug­gest even ex­pe­ri­enced in­vestors may not be very good at man­ager as­sess­ment.

Two Amer­i­can pro­fes­sors ‒ Amit Goyal and Su­nil Wa­hal ‒ ex­am­ined the se­lec­tion and ter­mi­na­tion of man­agers for 3 700 US in­sti­tu­tional funds over a 10-year pe­riod. They found in­vestors’ ten­dency was to end the con­tracts of in­vest­ment man­agers who had un­der­per­formed and take on man­agers who out­per­formed.

If, how­ever, in­vestors had stayed with their fired man­agers, out­per­for­mance would have been larger than that de­liv­ered by the newly hired man­agers. And that’s be­fore tak­ing into ac­count the costs of chang­ing man­agers, which can be hefty.

So how much weight re­ally can be put on the short-term ‒ even medium-term ‒ per­for­mance of pro­fes­sional in­vestors? The ev­i­dence sug­gests cau­tion is needed. This pa­per looks at some of the is­sues in­volved in in­ter­pret­ing the past per­for­mance of ac­tive eq­uity man­agers and sug­gests it’s es­sen­tial to ac­com­pany quan­ti­ta­tive anal­y­sis with qual­i­ta­tive as­sess­ment. The past as a poor guide to the fu­ture We looked at per­for- mance statis­tics for a large num­ber of ac­tive man­agers recorded on the data­base of con­sult­ing firm Mercer. We tracked the per­for­mance of the best-per­form­ing 25% for ev­ery rolling three-year pe­riod be­tween De­cem­ber 1985 and De­cem­ber 2005. For those manag­ing US shares, only 42% of the pre­vi­ous top quar­tile turned in a per­for­mance above the me­dian per­for­mance in the fol­low­ing three years. Man­agers of non-US shares did slightly bet­ter, with 47% per­form­ing above the me­dian. But the key point is in nei­ther case did the man­agers’ three-year track records pro­vide a re­li­able guide to their fu­ture re­sults.

An­a­lyz­ing the prob­lem

So given those lim­i­ta­tions, how should you think about a man­ager’s re­cent per­for­mance?


First, it’s crit­i­cal to re­mem­ber a per­for­mance tar­get rep­re­sents only an av­er­age ex­pected re­turn. For ex­am­ple, take a man­ager claim­ing to of­fer a 3%/year re­turn over rolling three-year pe­ri­ods. That means half the time the per­for­mance will range above that re­turn and half the time it will range be­low. While, over time, the av­er­age re­turn should cen­tre on 3%, it can mean that over shorter pe­ri­ods per­for­mances can de­vi­ate sharply from that mean and still be in line with our 3%/year long-term re­turns.


A man­ager’s re­turns can’t be as­sessed without con­sid­er­ing how much risk he’s tak­ing. Al­though there are many ways of quan­ti­fy­ing risk, per­haps the most rel­e­vant is tracking er­ror: a mea­sure of how far a port­fo­lio is de­vi­at­ing from the cho­sen bench­mark, whether due to stock se­lec­tion, in­dus­try weights or other fac­tors.

Tracking er­ror mea­sures the volatil­ity of a port­fo­lio’s re­turns rel­a­tive to its bench­mark: in other words, it mea­sures the con­sis­tency a port­fo­lio hits its re­turn tar­gets. In sta­tis­ti­cal terms, that’s the “stan­dard de­vi­a­tion” of a port­fo­lio’s rel­a­tive re­turns. It’s nor­mally used to mea­sure how much risk a man­ager is cur­rently tak­ing. It can also be used to show how volatile the port­fo­lio’s rel­a­tive re­turns have ac­tu­ally been in the past.

The way the maths works means a tracking er­ror is ex­pected to en­com­pass a lit­tle more than 66% of out­comes. So a man­ager with a 5% ex­pected tracking er­ror who aims to gen­er­ate 2% more than the bench­mark should ex­pect to pro­duce a per­for­mance of be­tween +7% and -3% in roughly two years out of three. For most of the rest of the time sta­tis­ti­cal the­ory sug­gests the out­comes should be cov­ered by re­turns within two stan­dard de­vi­a­tions from the tar­get, from +12% to -8%. But in one in 20 years it would be per­fectly nor­mal to ex­pe­ri­ence re­turns be­yond those lim­its. (See­graph).

Put an­other way, in any one year of a rolling three-year per­for­mance pe­riod, sta­tis­ti­cal the­ory sug­gests even a skilled man­ager with a tar­get pre­mium of 2% and an ex­pected tracking er­ror of 5% has about a 7% chance of pro­duc­ing a per­for­mance worse than -8%, rel­a­tive to the bench­mark.

In re­al­ity, man­agers’ re­sults only ap­prox­i­mate that ideal pic­ture, with ex­treme out­per­for­mance and un­der­per­for­mance hap­pen­ing more of­ten than a sta­tis­ti­cal nor­mal dis­tri­bu­tion would sug­gest.

The odds of any one man­ager beat­ing the mar­ket con­sis­tently through luck may seem re­mote. None­the­less, given the num­ber of man­agers and strate­gies that have ex­isted the odds of one or more

Ni­cholas David­son

Newspapers in English

Newspapers from South Africa

© PressReader. All rights reserved.