The rea­sons ex­perts are bad at fore­cast­ing

Austin American-Statesman Sunday - - PERSONAL FINANCE EXTRA - Barry Ritholtz Per­sonal Fi­nance

I re­cently pointed out an er­rant re­ces­sion fore­cast made a year ago by a per­son who has been more or less been pre­dict­ing an eco­nomic slump since 2011. I was hop­ing to spark a dis­cus­sion of why fore­casts are so coun­ter­pro­duc­tive, and why no one should make in­vest­ments based on them. In­stead, a dis­cus­sion of pas­sive bulls mock­ing ac­tive bears broke out. This was not what I in­tended. I made a few as­sump­tions, per­haps er­ro­neously: I thought that the idea I have been harp­ing on for more than a decade had been thor­oughly beaten to death, and per­haps it was time to give it a rest. By now, I fig­ured, ev­ery­one surely un­der­stands that mar­ket fore­casts — in­deed al­most all fore­casts — are folly.

Alas, my as­sump­tion was proven wrong. Thus, we go once more unto the breach, to re­mind read­ers what we know about fore­casts and pre­dic­tions, and why they are so rarely right:

Not ev­ery­thing is a fore­cast: This is es­pe­cially true in terms of mar­kets and the econ­omy, and so a rea­son­able def­i­ni­tion of a fore­cast is as fol­lows: It per­tains to a spe­cific as­set or as­set class and/or eco­nomic data se­ries, at a given price or level and a spe­cific time. It also must be dis­prov­able. Mak­ing a state­ment that can’t be proven or dis­proven is not a fore­cast; it’s a the­o­ret­i­cal aca­demic de­bate.

Con­sider the fol­low­ing state­ments: “Stocks tend to go higher” or “re­ces­sions are cycli­cal.” These are not fore­casts be­cause they lack specifics. The state­ment, “The Dow will hit 25,000 by the sec­ond quar­ter of 2018,” on the other hand, will either be proven right or wrong.

We are very bad at fore­cast­ing: Ex­am­ples are ev­ery­where: Eco­nomic fore­casts, earn­ings es­ti­mates, mar­ket fore­casts, ex­pec­ta­tions of fu­ture tech­nolo­gies, not to men­tion elec­tion pre­dic­tions. The data over­whelm­ingly show that as a species, we are sim­ply aw­ful at this.

We are even worse at pre­dict­ing our own be­hav­ior: When­ever you see some­one fore­cast­ing their own be­hav­ior, what you are get­ting is a read of their emo­tional state. Whether it’s hol­i­day shop­ping, com­pany hir­ing plans, vot­ing in­ten­tions — peo­ple say what they are feel­ing at the time the ques­tion is posed, and it is not pre­dic­tive of what they are ac­tu­ally go­ing to do.

Tech­nol­ogy is tricky: We are par­tic­u­larly bad at mak­ing pre­dic­tions about tech­nol­ogy. This goes back a long ways, to claims in the late 19th cen­tury that no one would ever need or want a tele­phone, or that cars would never re­place horses and bug­gies. And folks are still at it, de­spite their in­abil­ity to get it right. Check Mi­crosoft Inc. chief Steve Ballmer’s 2007 pre­dic­tion that Ap­ple Inc.’s iPhone would never catch on.

Ran­dom luck: Peo­ple tend to over­fo­cus on the out­come rather than pay­ing closer at­ten­tion to the process. This leads to an overem­pha­sis on guesses that were merely lucky and there­fore can­not be repli­cated.

Asym­met­ric risk: Bad fore­casts are quickly for­got­ten, while those who make ac­cu­rate pre­dic­tions that are noth­ing more than the re­sult of luck or ran­dom chance get el­e­vated to star­dom. This is the en­tire un­der­pin­ning of why peo­ple make fore­casts — and es­pe­cially rad­i­cal scary ones — in the first place. The po­ten­tial re­wards for be­ing right can be sig­nif­i­cant, while be­ing wrong has lit­tle down­side.

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.