How Uber's team of economists tests new features and incentives, writes highly-cited academic papers, and provides talking points for public policy initiatives (Alison Griswold/Quartz)

0
2
- Advertisement -

It’s hard to explain just how much economists love Uber.

Economists love Uber like a mother loves her child. They love it like the internet loves cats. They love it like tech bros love Elon Musk.

Economists love Uber because it’s the closest you can get to taking the pure economic theory of textbooks and summoning it to life. Uber created a massive open market, governed first and foremost by the forces of supply and demand. Along the way it broke up the taxi monopoly, taught people to accept “surge” pricing, and ushered concepts long confined to econ 101 into the popular discourse.

“Uber is, in many ways, the embodiment of what the economists would like the economy to look like,” economist and Freakonomics co-author Stephen Levitt said on a podcast in 2016.

Economists don’t just love Uber—the company loves them back. Uber is so fond of economists that it employs more than a dozen PhDs from top programs at its San Francisco headquarters. The group acts as an in-house think tank for Uber, gathering facts from quants and data scientists and synthesizing them to arm the lobbyists and policy folks who fight some of Uber’s biggest battles. Officially, this team is known as “Research and Economics.” Internally, it’s also been called Ubernomics.

Ubernomics keeps a low profile, despite the fact that Uber has collaborated on research papers with economic superstars like Levitt and former Obama adviser Alan Krueger. Its wide-ranging mandate includes studying the consumer experience, testing new features and incentives, supporting Uber’s public policy needs, and producing peer-reviewed academic research.

Plenty of companies hire experts to bolster their business, but Uber wants prestige, too. It wants its papers authored by the best and brightest, and accepted by leading academic journals. Prestige, after all, confers legitimacy, and this sort of soft power is the perfect complement to Uber’s hard-charging politics. The mission of Ubernomics is simple: to prove, with cool and unemotional logic, that the rest of the world should love Uber as much as economists already do.

It was all but preordained that Jonathan Hall would become an economist. His father, Robert, is a professor of economics at Stanford and heads the National Bureau of Economic Research’s business cycle dating committee. His stepmother, Brownwyn, was an economist at Berkeley. His half-sister, Anne, is an economist at the US Treasury.

When Hall was nine or 10 years old, he told a friend the world would be a lot simpler if the government set a clear price for everything (his dad had to explain how a planned economy could go badly wrong). In fifth grade, he said he wanted to be an economist when he grew up (his mom still has the scrapbook entry). Hall attended Harvard, first for undergrad and then a PhD in economics. When he married in 2011, a quarter of the wedding guests were economists.

Most people who bother to get a doctorate in economics choose a career in academia and turn their noses up at corporate research. Hall considered company research “fluffy.” But rather than snub the corporate world, Hall thought he could raise the bar. He spent a few years working at Google in public policy and then, after a stint at economic consulting firm Analysis Group, joined music-streaming company Pandora as a senior scientist in 2013.

In February 2014, a job listing at Uber caught Hall’s eye. The description of the position, Hall said, was identical to one he’d written for himself when he worked at Google and carved out a niche as a policy economist. Hall contacted Uber about the job and noted the company had borrowed language from the job description he wrote at Google. “That got me in the door,” he said. He joined as Uber’s first in-house economist later that month.

Hall, 34, saw in Uber a chance to do what he’d always wanted: Get in early at a fast-growing, audacious company and build a high-quality research organization from the ground up. Uber specifically appealed because it was disrupting taxis, an industry with a rich economic literature, and which economists often cited as an example of government regulation leading to bad outcomes. Uber offered access to unprecedented data on rider and driver behavior, delivered in real time. The cherry on top was that Uber was using surge pricing—prices that changed with demand—“which economists think is the coolest thing in the world,” Hall said.

The Uber that Hall joined in February 2014 was a luxury town car service. There was no UberPool, no UberEats, no trucking or driverless cars. Uber co-founder and then-chief executive Travis Kalanick was at his bro-iest, peppering interviews with slang like “hashtag winning” and “boob-er.” The business was valued at $3.5 billion by investors, a sliver of the roughly $70 billion valuation it commands today.

In the fall of 2014, Hall reached out to Alan Krueger, a famed labor economist at Princeton who had recently chaired the White House Council of Economic Advisers for president Barack Obama, about working for Uber. Krueger had written a paper with Hall’s dad, Robert, but Hall had never met him. Krueger was interested.

The Krueger paper was the launch pad Ubernomics needed.

Uber’s first economics paper (pdf), “An Analysis of the Labor Market for Uber’s Driver-Partners in the United States,” appeared as a working paper in Princeton’s Industrial Relations Section on Jan. 22, 2015, co-authored by Hall and Krueger. The paper found that most Uber drivers had another part- or full-time job; that drivers liked setting their own schedules; and that most used Uber to supplement their income. The paper also offered the first glimpse at what drivers were earning, based on Uber’s data. Across six US major cities, Hall and Krueger reported the median driver earned between $16.20 an hour (Chicago) and $30.35 an hour (New York), after Uber’s take, but before on-the-job expenses like gas and vehicle maintenance.

The 28-page paper earned widespread media coverage. That the study was dry and academic only seemed to make it more appealing, a sort of antidote to the inflated and unsubstantiated marketing claims Uber had made about driver earnings several months earlier. But the project was tarnished by the fact that Uber had paid Krueger, an arrangement that made its credibility easier to doubt among his high-minded peers in academia.

“We shouldn’t have done that,” Hall said in an interview earlier this year, declining to say what Uber paid Krueger. “He would have done it for free had he understood what our team’s capabilities were.” Krueger didn’t respond to requests for comment. In November 2016, Hall and Krueger revised the paper—Krueger wasn’t paid for the revision—and it was accepted by the Industrial Labor Relations Review.

The Krueger paper was the launch pad Ubernomics needed. Over the next few years, the company landed unpaid collaborations with academics at top US universities, including MIT, NYU, and Yale. Hall started to receive dozens of requests a week from academics about working with Uber’s data. Earlier this year, Richard Thaler, winner of the 2017 Nobel Prize in economic sciences, approached Uber about a possible collaboration. Uber turned him down.

Before Uber’s love affair with economists, there was AT&T and Bell Laboratories. In the 1960s, AT&T faced a number of regulatory challenges but felt the research on these policy questions was limited. It caught the eye of economists when it began sponsoring conferences to increase discussion of these regulatory matters. The company soon hired three academic economists to form an internal advisory group.

In 1968, AT&T established the Bell Labs economics research team. The company enlisted a few dozen newly minted PhDs to study economies of scale, incremental costs, game theory, and whatever else they were interested in. The gig paid better than most entry-level academic work, and everyone wore jeans. “It was like a university without students,” said John Panzar, who joined Bell Labs from his Stanford graduate program in 1974 and later ran the group’s economic analysis unit. The Rand Journal of Economics, a respected academic publication, began in 1970 as the Bell Journal.

The Bell Labs economics team benefited from the company’s sterling work in the hard sciences, whose researchers had by then won two Nobel Prizes in physics. Economists could earn bigger raises and greater recognition by doing work related to AT&T’s interests, Panzar said, but the first goal was to publish in respected academic journals. “The new thing here was to hire economists who were going to do basically academic-style research in Bell Laboratories,” he said.

Modern corporate economic research began with Hal Varian. After more than a decade at the University of California, Berkeley, Varian began consulting for Google in 2002. He went on to help refine the online AdWords bidding market that made Google one of the richest companies in the world, and to become its chief economist. Varian is credited with the so-called Varian Rule, which says the rich have today what the middle class will have in 10 years, and the poor a decade after that.

It didn’t take long for the rest of the tech world to catch wind of what Google was doing, and to follow suit. Yahoo went on an economist hiring spree to help keep up with Google, poaching talent from top universities and industrial labs at Microsoft and IBM. Microsoft brought in Susan Athey, a highly regarded academic who trained at Stanford, as its consulting chief economist.

Academics who made the leap to Silicon Valley were rewarded with data unimaginable to most university researchers. Yahoo in the mid-2000s had roughly 500 million monthly visitors. Google presided over billions of AdWords auctions. Microsoft had millions of customers and hundreds of thousands of advertisers. “Economists have rarely had so much value to add in designing products and platforms that touch so many lives,” Athey said by email. “That was exhilarating—instead of hundreds of academics citing a research paper, my work could directly influence the economy as a whole.”

Like Yahoo and Google, Uber offers access to the kind of data that, as recently as 10 years ago, an economist could only dream of. Uber’s more than 3 million drivers provide roughly 15 million trips, globally, every day. Uber’s researchers can test vital questions about driver pay, customer satisfaction, and urban transit with tiny tweaks to the company’s algorithms.

Uber doesn’t police what its collaborators write or publish, but it is choosy about who it works with. It’s no coincidence that Ubernomics tends to study the thorniest matters facing the company. The goal is to “build a body of evidence and then build a global policy framework around that,” said Amit Singh, Uber’s head of global public policy. Insofar as there’s bias in the research, it comes not in the results, but from the questions that are asked.

In 2015, when public officials condemned surge pricing as price gouging, Hall wrote a short paper that explained how surge moved drivers to where riders needed them most, using trip data from an Ariana Grande concert in New York. Uber policy operatives used the paper to talk down politicians who were interested in restricting how much it could lift prices during busy periods. Hall and Krueger’s original paper on driver satisfaction developed during a bitter legal battle over whether Uber had misclassified its workers as independent contractors. The study provided Uber with the perfect talking points to defend the contractor model, for example, that drivers liked to “be their own boss” and “set their own schedule.”

Insofar as there’s bias in the research, it comes not in the results, but from the questions that are asked.

“They’re pretty clear that they’re not going to give access to the data to people who are likely to find things that are not favorable to Uber,” Lawrence Mischel, a labor economist and former president of the pro-labor Economic Policy Institute, said of Uber’s researchers. Mischel has criticized how Hall and Krueger presented driver earnings data without explicitly deducting on-the-job expenses. An analysis he released with EPI in May found Uber drivers made $11.77 an hour after both commission and expenses, and $9.21 an hour when adjusted to compare to the wages of a regular W-2 employee—significantly less than the headline numbers from Hall and Krueger.

“I think we should give [Uber] some but not complete credibility,” Mischel said.

In July 2016, Uber hired John List, a big-deal economist at the University of Chicago, as chief economist (a title he shared with Hall). Amazon had approached List with a similar offer in 2010, but he declined after the company said it wouldn’t allow academic publication. Travis Kalanick lured List to Uber by making clear that publishing was part of the deal.

List assembled a small team of researchers to work on pay and pricing, and dubbed it Ubernomics. Not everyone at Uber was fond of the moniker. “We all cringe at that name,” said Michael Amodeo, a spokesperson. List published papers exploring why a gender pay gap existed among Uber drivers and how Uber should apologize to riders who had bad trips. Behind the scenes, he also pushed Uber to add tipping to its app, something Kalanick considered a disguised price increase and strongly opposed. List made an economic case for tipping, arguing it would improve Uber’s service, make drivers happier, and put the company on even footing with Lyft, which had tips.

Uber introduced tipping on June 20, 2017. The next day, Kalanick resigned as CEO after failing to contain months of scandal and board infighting at the company. (Uber has said the two events weren’t connected.) Two months later, Uber named Dara Khosrowshahi CEO. The regime change was felt by List, who said everyone he worked with at Uber at a high level either quit or was fired. List left in May 2018. He’s now the chief economist at Lyft.

Uber doesn’t have a lock on gig economy research, but the company’s economists have deftly woven their findings into the literature. Hall and Krueger’s paper, for instance, has hundreds of academic citations and a handful of footnotes in third-party corporate research (pdf). Krueger referred to his work with Uber in a policy paper (pdf) he co-authored on modernizing labor laws. Krueger and Harvard economist Lawrence Katz then cited that policy paper in a widely read study (pdf) on structural changes in the workforce that was published by Princeton and the National Bureau of Economic Research.

This past spring, MIT published a brief of a working paper from researchers at Stanford that claimed the typical Uber or Lyft driver made just $3.37 an hour, a shockingly low sum. The study received instant attention from media outlets, which proclaimed things like “Uber drivers often make below minimum wage” and “MIT study shows driving for Uber or Lyft is worse than literally any real job.”

Less than a day after the brief gained traction, Hall rebutted it in a post on Uber’s blog. Khosrowshahi shared Hall’s analysis on Twitter, commenting, “MIT = Mathematically Incompetent Theories.” A handful of Uber’s economic collaborators, including Krueger and Yale economist Judy Chevalier, also rallied to the company’s defense on Twitter.

The researchers’ findings were indeed flawed. They based their analysis on self-reported earnings data from a 2017 survey of about 1,100 Uber and Lyft drivers. Because several questions in the survey were worded poorly, the Stanford researchers made incorrect adjustments to the monthly wages that drivers reported, artificially deflating their final earnings number.

A few days later, Stephen Zoepf, the paper’s lead author, said he would revisit the findings and update the paper. He thanked Hall for his analysis, while also issuing a subtle critique. “Transparency and reproducibility are the foundation of any academic endeavor,” Zoepf wrote in a statement he shared on Twitter. “What Hall and Khosrowshahi’s assessment laid bare was an assumption about revenue that I made in the absence of public ride-hailing data and a paucity of independent studies outside Uber’s own analyses.”

The episode provided a glimpse of how Hall’s team springs into action and the academic soft-power it can muster. The company’s research is a virtuous circle. Uber’s high-quality data attracts talented academics and researchers, who write papers that raise the company’s profile and increase its credibility, attracting more academics who write more papers and garner more legitimacy. Upending global transportation is a gritty business. The next policy fight is just around the corner. Ubernomics is ready.


This post was originally published here
- Advertisement -