Al­go­rithms Aren’t Just For Coders

Tech com­pa­nies are on a hir­ing spree for econ­o­mists There are “these new com­pa­nies with tons of dig­i­tized data”

Bloomberg Businessweek (Europe) - - TECHNOLOGY - Pa­trick Clark Edited by Dim­i­tra Kessenides and Cristina Lind­blad Bloomberg.com

“Some of the pri­vate data is garbage. It’s not that the peo­ple pro­duc­ing it are not as smart or that they don’t do hard work. The mo­ti­va­tions are dif­fer­ent.” ——Bill McBride, Cal­cu­lated Risk

At an April meetup or­ga­nized by the National As­so­ci­a­tion for Busi­ness Eco­nomics (NABE), a Face­book re­searcher named Michael Bailey pre­sented a work­ing pa­per sug­gest­ing that a buyer in Detroit might be will­ing to pay more for a home if he had lots of Face­book friends liv­ing in a high-priced hous­ing mar­ket like San Fran­cisco. For the project, Bailey and his co-au­thors matched public records of 525,000 home sales to anonymized data for 1.4 mil­lion Face­book users.

The day­long meet­ing, held at the Fed­eral Re­serve Bank of San Fran­cisco, was the first for­mal gath­er­ing of tech com­pany econ­o­mists, ac­cord­ing to NABE Executive Direc­tor Tom Beers, and in­cluded nu­mer­ous stars of the con­sumer in­ter­net. Hal Var­ian, the Google econ­o­mist who helped de­velop the AdWords mar­ket­place, was there; Keith Chen, of Uber, pre­sented a pa­per on the com­pany’s surge-pric­ing policy that re­futed ear­lier re­search that said taxi driv­ers won’t work in the rain. Econ­o­mists from Ama­zon.com, Net­flix, and LinkedIn elab­o­rated on their work as well. “It was like a geek dream come true,” says Nela Richard­son of the real es­tate bro­ker­age Redfin.

The meet­ing gave par­tic­i­pants a chance to trade notes about what it’s like to be at the fore­front of a trend in the pro­fes­sion. U.S. com­pa­nies went on an econ­o­mist hir­ing spree in the late 1950s and 1960s, says Beers, as com­put­ers made econo­met­ric anal­y­sis pos­si­ble and com­pa­nies sought ex­perts to fore­cast swings in the busi­ness cy­cle. To­day, busi­nesses are once again ramp­ing up their hir­ing of econ­o­mists, this time spurred by a boom in web-gen­er­ated data and tools for stor­ing and sort­ing it. Their job is to ex­tract in­sights that can help busi­nesses im­prove their prod­ucts or user ex­pe­ri­ence. Some also pro­duce re­search to shape public policy. “Now you have all these new com­pa­nies with tons of dig­i­tized data, and not only that, it’s data that de­scribes hu­man be­hav­ior,” says An­drew Cham­ber­lain, chief econ­o­mist at the jobs and re­cruit­ing web­site Glass­door.

There were 11,500 econ­o­mists work­ing in the pri­vate sec­tor as of May 2015, ac­cord­ing to the Bureau of Labor Statis­tics, up from 5,580 in May 2010. Face­book’s data science team em­ploys about 25 Ph.D.s in eco­nomics, says Bailey. That’s about the same num­ber em­ployed at a large U.S. bank, NABE’s Beers says. Ama­zon em­ploys more than 60 econ­o­mists, ac­cord­ing to at­ten­dees at the NABE net­work­ing event, and its ca­reers page lists more than 30 open po­si­tions. (The com­pany didn’t re­spond to re­quests for com­ment.)

Stan Humphries joined Zil­low in

2005 to de­velop al­go­rithms for es­ti­mat­ing home prices. When the hous­ing mar­ket started to crater, he emerged as a fa­vorite source for jour­nal­ists look­ing for data and com­men­tary. “At a time when you still had in­dus­try peo­ple say­ing ‘Yes, we’ve had some cor­rec­tion in prices, but there’s noth­ing to see here, move on,’ I’d be the guy who came out and said, ‘No, we’re go­ing to see an­other two years in hous­ing re­ces­sion; here’s why,’” says Humphries, whose cur­rent ti­tle is chief an­a­lyt­ics of­fi­cer and chief econ­o­mist. “We felt be­ing as ac­cu­rate as we could would gar­ner re­spect from con­sumers.”

Pub­lish­ing data-driven re­search has be­come a pop­u­lar strat­egy for web mar­ket­places and list­ings sites to show­case their depth of knowl­edge about a par­tic­u­lar in­dus­try. That in­cludes home ren­o­va­tion star­tups such as Houzz and BuildZoom, and jobs sites like In­deed and Glass­door. More data is gen­er­ally a good thing, says Bill McBride, who blogs about the hous­ing mar­ket at Cal­cu­lated Risk, but it pays to con­sider where it comes from and how it’s com­piled. “Some of the pri­vate data is garbage,” he says. “It’s not that the peo­ple pro­duc­ing it are not as smart or that they don’t do hard work. The mo­ti­va­tions are dif­fer­ent.”

In the early days of the data boom, tech com­pa­nies sought to en­tice big brains by al­low­ing them to keep one foot in academia, says Su­san Athey, a for­mer chief econ­o­mist at Mi­crosoft who now teaches at Stan­ford. Re­cently, Ama­zon has emerged as an ex­cep­tion to that rule, says Athey: It keeps a tight leash on re­search pro­duced by its in-house econ­o­mists. None­the­less, it’s man­aged to at­tract a team whose size and qual­ity ri­vals the eco­nomics de­part­ments of top uni­ver­si­ties, Athey says, in part be­cause the com­pany of­fers ac­cess to unique data. “I can’t run an ex­per­i­ment on a cou­ple of mil­lion peo­ple at Stan­ford. If you want to be aware of what in­ter­est­ing ques­tions are out there, you al­most have to go and work for one of these com­pa­nies.”

The bot­tom line The gi­ants of the web are as­sem­bling teams of econ­o­mists that ri­val those at banks and uni­ver­si­ties.

Newspapers in English

Newspapers from Bahrain

© PressReader. All rights reserved.