Archives For behavioral economics

In a recent long-form article in the New York Times, reporter Noam Scheiber set out to detail some of the ways Uber (and similar companies, but mainly Uber) are engaged in “an extraordinary experiment in behavioral science to subtly entice an independent work force to maximize its growth.”

That characterization seems innocuous enough, but it is apparent early on that Scheiber’s aim is not only to inform but also, if not primarily, to deride these efforts. The title of the piece, in fact, sets the tone:

How Uber Uses Psychological Tricks to Push Its Drivers’ Buttons

Uber and its relationship with its drivers are variously described by Scheiber in the piece as secretive, coercive, manipulative, dominating, and exploitative, among other things. As Schreiber describes his article, it sets out to reveal how

even as Uber talks up its determination to treat drivers more humanely, it is engaged in an extraordinary behind-the-scenes experiment in behavioral science to manipulate them in the service of its corporate growth — an effort whose dimensions became evident in interviews with several dozen current and former Uber officials, drivers and social scientists, as well as a review of behavioral research.

What’s so galling about the piece is that, if you strip away the biased and frequently misguided framing, it presents a truly engaging picture of some of the ways that Uber sets about solving a massively complex optimization problem, abetted by significant agency costs.

So I did. Strip away the detritus, add essential (but omitted) context, and edit the article to fix the anti-Uber bias, the one-sided presentation, the mischaracterizations, and the fundamentally non-economic presentation of what is, at its core, a fascinating illustration of some basic problems (and solutions) from industrial organization economics. (For what it’s worth, Scheiber should know better. After all, “He holds a master’s degree in economics from the University of Oxford, where he was a Rhodes Scholar, and undergraduate degrees in math and economics from Tulane University.”)

In my retelling, the title becomes:

How Uber Uses Innovative Management Tactics to Incentivize Its Drivers

My transformed version of the piece, with critical commentary in the form of tracked changes to the original, is here (pdf).

It’s a long (and, as I said, fundamentally interesting) piece, with cool interactive graphics, well worth the read (well, at least in my retelling, IMHO). Below is just a taste of the edits and commentary I added.

For example, where Scheiber writes:

Uber exists in a kind of legal and ethical purgatory, however. Because its drivers are independent contractors, they lack most of the protections associated with employment. By mastering their workers’ mental circuitry, Uber and the like may be taking the economy back toward a pre-New Deal era when businesses had enormous power over workers and few checks on their ability to exploit it.

With my commentary (here integrated into final form rather than tracked), that paragraph becomes:

Uber operates under a different set of legal constraints, however, also duly enacted and under which millions of workers have profitably worked for decades. Because its drivers are independent contractors, they receive their compensation largely in dollars rather than government-mandated “benefits” that remove some of the voluntariness from employer/worker relationships. And, in the case of overtime pay, for example, the Uber business model that is built in part on offering flexible incentives to match supply and demand using prices and compensation, would be next to impossible. It is precisely through appealing to drivers’ self-interest that Uber and the like may be moving the economy forward to a new era when businesses and workers have more flexibility, much to the benefit of all.

Elsewhere, Scheiber’s bias is a bit more subtle, but no less real. Thus, he writes:

As he tried to log off at 7:13 a.m. on New Year’s Day last year, Josh Streeter, then an Uber driver in the Tampa, Fla., area, received a message on the company’s driver app with the headline “Make it to $330.” The text then explained: “You’re $10 away from making $330 in net earnings. Are you sure you want to go offline?” Below were two prompts: “Go offline” and “Keep driving.” The latter was already highlighted.

With my edits and commentary, that paragraph becomes:

As he started the process of logging off at 7:13 a.m. on New Year’s Day last year, Josh Streeter, then an Uber driver in the Tampa, Fla., area, received a message on the company’s driver app with the headline “Make it to $330.” The text then explained: “You’re $10 away from making $330 in net earnings. Are you sure you want to go offline?” Below were two prompts: “Go offline” and “Keep driving.” The latter was already highlighted, but the former was listed first. It’s anyone’s guess whether either characteristic — placement or coloring — had any effect on drivers’ likelihood of clicking one button or the other.

And one last example. Scheiber writes:

Consider an algorithm called forward dispatch — Lyft has a similar one — that dispatches a new ride to a driver before the current one ends. Forward dispatch shortens waiting times for passengers, who may no longer have to wait for a driver 10 minutes away when a second driver is dropping off a passenger two minutes away.

Perhaps no less important, forward dispatch causes drivers to stay on the road substantially longer during busy periods — a key goal for both companies.

Uber and Lyft explain this in essentially the same way. “Drivers keep telling us the worst thing is when they’re idle for a long time,” said Kevin Fan, the director of product at Lyft. “If it’s slow, they’re going to go sign off. We want to make sure they’re constantly busy.”

While this is unquestionably true, there is another way to think of the logic of forward dispatch: It overrides self-control.

* * *

Uber officials say the feature initially produced so many rides at times that drivers began to experience a chronic Netflix ailment — the inability to stop for a bathroom break. Amid the uproar, Uber introduced a pause button.

“Drivers were saying: ‘I can never go offline. I’m on just continuous trips. This is a problem.’ So we redesigned it,” said Maya Choksi, a senior Uber official in charge of building products that help drivers. “In the middle of the trip, you can say, ‘Stop giving me requests.’ So you can have more control over when you want to stop driving.”

It is true that drivers can pause the services’ automatic queuing feature if they need to refill their tanks, or empty them, as the case may be. Yet once they log back in and accept their next ride, the feature kicks in again. To disable it, they would have to pause it every time they picked up a new passenger. By contrast, even Netflix allows users to permanently turn off its automatic queuing feature, known as Post-Play.

This pre-emptive hard-wiring can have a huge influence on behavior, said David Laibson, the chairman of the economics department at Harvard and a leading behavioral economist. Perhaps most notably, as Ms. Rosenblat and Luke Stark observed in an influential paper on these practices, Uber’s app does not let drivers see where a passenger is going before accepting the ride, making it hard to judge how profitable a trip will be.

Here’s how I would recast that, and add some much-needed economics:

Consider an algorithm called forward dispatch — Lyft has a similar one — that dispatches a new ride to a driver before the current one ends. Forward dispatch shortens waiting times for passengers, who may no longer have to wait for a driver 10 minutes away when a second driver is dropping off a passenger two minutes away.

Perhaps no less important, forward dispatch causes drivers to stay on the road substantially longer during busy periods — a key goal for both companies — by giving them more income-earning opportunities.

Uber and Lyft explain this in essentially the same way. “Drivers keep telling us the worst thing is when they’re idle for a long time,” said Kevin Fan, the director of product at Lyft. “If it’s slow, they’re going to go sign off. We want to make sure they’re constantly busy.”

While this is unquestionably true, and seems like another win-win, some critics have tried to paint even this means of satisfying both driver and consumer preferences in a negative light by claiming that the forward dispatch algorithm overrides self-control.

* * *

Uber officials say the feature initially produced so many rides at times that drivers began to experience a chronic Netflix ailment — the inability to stop for a bathroom break. Amid the uproar, Uber introduced a pause button.

“Drivers were saying: ‘I can never go offline. I’m on just continuous trips. This is a problem.’ So we redesigned it,” said Maya Choksi, a senior Uber official in charge of building products that help drivers. “In the middle of the trip, you can say, ‘Stop giving me requests.’ So you can have more control over when you want to stop driving.”

Tweaks like these put paid to the arguments that Uber is simply trying to abuse its drivers. And yet, critics continue to make such claims:

It is true that drivers can pause the services’ automatic queuing feature if they need to refill their tanks, or empty them, as the case may be. Yet once they log back in and accept their next ride, the feature kicks in again. To disable it, they would have to pause it every time they picked up a new passenger. By contrast, even Netflix allows users to permanently turn off its automatic queuing feature, known as Post-Play.

It’s difficult to take seriously claims that Uber “abuses” drivers by setting a default that drivers almost certainly prefer; surely drivers seek out another fare following the last fare more often than they seek out another bathroom break. In any case, the difference between one default and the other is a small change in the number of times drivers might have to push a single button; hardly a huge impediment.

But such claims persist, nevertheless. Setting a trivially different default can have a huge influence on behavior, claims David Laibson, the chairman of the economics department at Harvard and a leading behavioral economist. Perhaps most notably — and to change the subject — as Ms. Rosenblat and Luke Stark observed in an influential paper on these practices, Uber’s app does not let drivers see where a passenger is going before accepting the ride, making it hard to judge how profitable a trip will be. But there are any number of defenses of this practice, from both a driver- and consumer-welfare standpoint. Not least, such disclosure could well create isolated scarcity for a huge range of individual ride requests (as opposed to the general scarcity during a “surge”), leading to longer wait times, the need to adjust prices for consumers on the basis of individual rides, and more intense competition among drivers for the most profitable rides. Given these and other explanations, it is extremely unlikely that the practice is actually aimed at “abusing” drivers.

As they say, read the whole thing!

Smoothing Demand Kinks

Steve Salop —  4 April 2011

One criticism of the unilateral effects analysis in the 2010 Merger Guidelines is that demand curves are kinked at the current price.  A small increase in price will dramatically reduce the quantity demanded.  One rationale for the kink is that people over-react to small price changes and dramatically reduce demand.  As a result of this behavioral economics deviation from standard rational behavior, it is claimed, merging firms will not raise prices when the merger increases the opportunity cost of increasing output.  (The opportunity cost increases because some of the increased output now comes from the new merger partner.)  It has been argued that such kinks are ubiquitous, whatever the current price is.  For some recent views on this issue, see the recent anti-kink article by Werden and the pro-kink reply by Scheffman and Simons.

A story in today’s New York Times nicely illustrates one of the problems with the kinked demand story.  Instead of raising prices, consumer products firms can and commonly do raise per unit prices by reducing package sizes.  Changes in package sizes do not create a disproportionate reaction, perhaps because they are less visible to busy shoppers.   Whatever the reason, this smaller package size raises the effective price per unit while avoiding the behavioral economics kink.  Of course, this is not to say that firms never raise prices; they do.  Moreover, even a kink did exist for reasons grounded in behavioral economics or menu costs, any kink likely is just temporary.  In contrast, a merger is permanent.

It is for these reasons that this kinked economics has gotten much traction in the current debate.  But, these presumptions do not mean that kinked economics arguments can never be raised in a merger.  If there were evidence of a low pass-through rate of variable cost into higher prices over a significant period of time, that evidence would be relevant to a more refined analysis of upward pricing pressure.

 

Geoffrey A. Manne is Executive Director of the International Center for Law & Economics and Lecturer in Law at Lewis & Clark Law School

The problem with behavioral law and economics (and its behavioral economics cousin) is not that it has nothing interesting to say, but rather that the interesting things it has to say do not mean what its proponents think they mean.  It is one thing to claim that people are less rational than we thought.  It even one thing to claim that people are systematically less rational than we thought, in predictable and important ways.  But it is entirely another to presume that the implication of this is a larger scope for government regulation to protect the market and market actors from the depredations of this irrationality.

Why?  Well, the market, of course.  Just because individuals may be less-rational than we thought does not mean that the complex and nuanced activities of markets can’t account for these deviations (particularly if they are predictable).  Add to this well-canvassed problems like government actors subject to the same biases, the problem of competing and conflicting biases, and the problem of unacknowledged, contrary implications, and the case for doing anything about behavioral quirks is extremely weak.

Thus, for example, let’s grant that, as many behavioralists aver, hyperbolic discounting exists.  Um, so, if that’s right, what should we do about it?  Force everyone to save more of their paychecks for retirement?  Insist on opt-out rather than opt-in retirement investing?  Ban cigarettes? Raise tax rates? (I don’t know if anyone has argued this one yet, but it seems like a plausible implication, and it’s only a matter of time)

Here’s the problem, as I see it:  Let’s say the behavioralists are right that, in the abstract, people save less money for future consumption than they would like.  Richard Thaler’s solution to this problem is the “Save More Tomorrow plan (pdf),” which takes advantage of people’s alleged current hyperbolic discounting to commit them to future savings that they actually want but can’t otherwise adhere to when the future actually arrives.  This is a “libertarian paternalist” (pdf) solution to the problem.

But there is a problem, even with a libertarian brand of paternalism here. Continue Reading…

Josh’s ongoing series on “Nudging Antitrust” and FTC Commissioner Rosch’s recent thoughts on behavioral economics has been excellent and I look forward to the next installment.  Rosch’s speech, not surprisingly, also elicited a strong response from me.  What follows are my thoughts on Rosch’s speech, focusing on some of the same issues Josh addressed in his first post.

Rosch puts forward four reasons, purportedly identified by behavioral economics, that explain why “human beings sometimes act irrationally in making commercial decisions.”  Each of his four reasons and/or his understanding of its implication is problematic.  Finally, Rosch proposes a policy response to his assessment of the behavioral literature—one that, unfortunately, bears almost no relationship to the behavioral findings form which he purports to draw it.

In the interest of space (and assuming readers have seen Josh’s posts on the speech) I will jump right in with my assessment of Rosch’s claims without trying to restate them.  The speech is readily-available here for those interested in checking my characterizations of Rosch’s comments.

Rosch’s four claims about behavioral economics:

1.  Information asymmetry.  It is deeply problematic that Rosch seems to think that this isn’t something that classical economics has discussed for generations.  More importantly, the idea that it presents a special problem is valid (at least as Rosch seems to see it) only if you act like information is or should be free and that there are no (or small) adverse effects to forcing disclosure of information.  In reality, the creation, assessment and disclosure of information are costly, and it makes no sense to act like there should be no return to these activities or to view as illegitimate the protection by commercial actors of the value they create.

2.  Instant gratification may be more important than long-run maximization.  As Josh points out, we have to assume this refers to hyperbolic discounting.  But even so, the implications of this, if true, are unclear, especially if we take into account (as Rosch seems not to) that regulators are also subject to the effect.  Implicit in much of the behavioral literature (derived, as it is, from constrained and manufactured experimental settings) is that institutions little affect the found effects.  It may even be the case that regulators operating within bureaucratic constraints indeed could be less affected by hyperbolic discounting.  But certainly actors in markets are less constrained than the naked assumption implies, and, either way, the policy implications would seem to be significantly affected by the institutional setting, consideration of which is essential to any legitimate assessment of the value of behavioral economics to the antitrust enterprise.  Unfortunately this consideration seems to be completely lacking here.

3.  Status quo bias.  Again, a large conventional literature on this sort of effect (arising from different sources, perhaps) exists, and the extent to which Rosch is talking about something new is unclear to me.  Even to the extent that it is a new idea, its assessment here falls prey to the same problems I mention above.  Moreover Rosch’s comments fail to reflect the difficulty in differentiating a “normal” or “rational” preference for the status quo arising from, for example, risk aversion, avoidance of transaction costs and the obtaining of costly information from their “irrational,” status-quo biased counterparts.

4.  Sellers may not always behave rationally.  As Josh suggests, this is, at least as Rosch seems to think relevant, an absurd claim on his part.  First off, claiming the theory of the firm for the behavioralists, or ignoring the essential ways in which “conventional” economics accounts for the very things Rosch thinks it fails to account for, seriously calls into question the quality and clarity of Rosch’s thinking on this topic.  Second, everyone knows that perfection is an assumption and not a description of reality.  The constant refrain, sometimes implicit and sometimes explicit, that classical economics “presumes that people will always behave rationally to achieve the best outcome” is, I think, willfully misleading.  Especially along the lines Rosch suggests here (agency costs?  Really? This is something behavioralists invented?), economics has never assumed this kind of perfection, and entire fields revolve around its rejection.

Moreover, there is also a sly and perhaps willful disconnect in emphasis.  The critique ends up sounding like a disparagement of classical economics’ (presumed) description of actual, individual psychology and behavior rather than a critique of a methodological assumption used to achieve better analysis (Josh’s reference to Friedman here is apt, of course).  Saying the behavioralists have a good handle on certain aspects of human psychology is not at all the same thing as saying they have a good approach to making those bits of knowledge systematic and useful.

Rosch’s policy recommendation:

Finally, Rosch pulls all of this together to make a recommendation: Merger review should rely more heavily on documentary evidence of merging parties’ “actual intent.”  In Rosch’s words, neoclassical economics has a problem dealing with behavioral economics because it

suggests [that neoclassical] models are imperfect and that better evidence (i.e., the parties’ documents and testimony) may be just as accurate at predicting competitive effects.

Um, no.  Actually there is nothing in behavioral economics, and certainly nothing in Rosch’s speech, that supports the claim that the parties’ documents may be just as accurate as neoclassical models (nor do I think that most neoclassical economists, other than a few misguided post-Chicagoans, think they should base their analysis of cases solely on abstract theories) at predicting competitive effects.

Of course regular readers know that this is a hot button for me.  My paper with Marc Williamson on the serious problems with relying on documents to infer intent, and on relying on expressions of intent as meaningful to relevant antitrust analysis, must have made little impression on Rosch (in case you haven’t looked at it in a while, the paper, Hot Docs vs. Cold Economics, is here).

Not only does Rosch’s assertion overstate the claims that even behavioralists would make for their ability to understand, interpret and systematize human and group psychology and sociology, it completely glosses over the incredibly thorny problem of the disconnect between that which is intended and that which actually is.  Nothing in the behavioral literature says anything at all about this problem, and it is a serious lapse for a Commissioner of the FTC to suggest that better knowledge (assuming it is better knowledge) about intent permits stronger conclusions about competitive effects.  This is just a non sequitur.

Overall I find the Commissioner’s speech to be seriously lacking.  I understand and applaud the effort to find knowledge useful to the antitrust enterprise wherever it may lie.  But one can’t help feeling that Rosch is a bit too enthusiastic, combative, naïve and, thus, careless in his support for the usefulness of this particular bit of knowledge.

Chris Hoofnagle writing at the TAP blog about Facebook’s comprehensive privacy options (“To opt out of full disclosure of most information, it is necessary to click through more than 50 privacy buttons, which then require choosing among a total of more than 170 options.”) claims that:

This approach is brilliant. The company can appease regulators with this approach (e.g. Facebook’s Elliot Schrage is quoted as saying, “We have tried to offer the most comprehensive and detailed controls and comprehensive and detailed information about them.”), and at the same time appear to be giving consumers the maximum number of options.

But this approach is manipulative and is based upon a well-known problem in behavioral economics known as the “paradox of choice.”

Too much choice can make decisions more difficult, and once made, those choices tend to be regretted.

But most importantly, too much choice causes paralysis. This is the genius of the Facebook approach: give consumer too much choice, and they will 1) take poor choices, thereby increasing revelation of personal information and higher ROI or 2) take no choice, with the same result. In any case, the fault is the consumer’s, because they were given a choice!

Of all the policy claims made on behalf of behavioral economics, the one that says there is value in suppressing available choices is one of the most pernicious–and absurd.  First, the problem may be “well-known,” but it is not, in fact, well-established.  Citing to one (famous) study purporting to find that decisions are made more difficult when decision-makers are confronted with a wider range of choices is not compelling when the full range of studies demonstrates a “mean effect size of virtually zero.”  In other words, on average, more choice has no discernible effect on decision-making.

But there is more–and it is what proponents of this canard opportunistically (and disingenuously, I believe) leave out:  There is evidence (hardly surprising) that more choices leads to greater satisfaction with the decisions that are made.  And of course this is the case:  People have heterogeneous preferences.  The availability of a wider range of choices is not necessarily optimal for any given decision-maker, particularly one with already-well-formed preferences.  But a wider range of choices is more likely to include the optimal choice for the greatest number of heterogeneous decision-makers selecting from the same set of options.  Even if it is true (and it appears not to be true) that more choice impairs decision-making, there is a trade-off that advocates like Hoofnagle (not himself a behavioral economist, so I don’t necessarily want to tar the discipline with the irresponsible use of its output by outsiders with policy agendas and no expertise in the field) typically ignore.  Confronting each individual decision-maker with more choices is a by-product of offering a greater range of choices to accommodate variation across decision-makers.  Of course we can offer everyone cars only in black.  And some people will be quite happy with the outcome, and delighted also that they have avoided the terrible pain of being forced to decide among a wealth of options that they didn’t even want.  But many other people, still perhaps benefiting from avoiding the onerous decision-making process, will nevertheless be disappointed that there was no option they really preferred. Continue Reading…