Amazon’s Bargemen

In early 20th century China, goods were frequently transported by barges pulled by teams of six men. The men were paid only if they delivered their goods on time. Therefore they all agreed to pull as hard as possible.

This is a classic example of what economists call a Prisoner’s Dilemma — a situation where everyone wants to cheat, regardless of whether he believes the others are cheating. Any bargeman might reason that “If the others are pulling hard, we’re going to make it anyway, so I might as well relax. And if the others are not pulling hard, we’re not going to make it anyway — so I still might as well relax .” Therefore they all relax and nobody gets paid.

According to my late and much lamented colleague Walter Oi, the bargemen frequently solved this problem by hiring a seventh man to whip them whenever they appeared to be giving less than 100%. You might suppose, at least if you’re a person of ordinary tastes, that hiring a man to whip you is never a good idea. There’s a sense in which you’d be right. But hiring a man to whip your colleagues can be a very good idea indeed, and if that requires getting whipped yourself, it might prove to be an excellent bargain.

If I’d lived in China a hundred years ago, I believe I’d have gone out of my way to buy goods from the teams with whipmasters — partly because that’s where I’d expect the best service, but also partly because I’d feel a certain combination of admiration and loyalty for the teams who were working so hard to earn my business.

That’s how I feel about the folks at Amazon. Based on the fabulous service I’ve been getting, I’m confident these people are knocking themselves out to do a good job for me. In fact, it’s been widely (and perhaps accurately) reported that during a heat spell a couple of summers ago, workers in an un-airconditioned Pennsylvania warehouse continued to fill orders even as several were being treated for heat sickness.

There’s a narrative going around that tries to paint these workers as victims, though I’ve heard no version of that narrative that makes clear who, exactly, is supposed to have victimized them — the stockholders? the management? the customers? the do-nothing Congress? But there’s little point in trying to make sense of this narrative, since it’s so obviously wrong to begin with.

Imagine a team of ambitious but relatively low-skilled workers. They know that if they all push themselves to the limit, they’ll all be more productive and therefore earn higher wages. They also know that if they all promise to push themselves to the limit, they’ll all break their promises, figuring that success or failure depends almost entirely on what the others do.

What these workers need, and presumably want, is an enforcer to prevent the breakdown of the agreement. Fortunately for those Amazon warehouse workers, they found that enforcer, presumably in the person of a good warehouse foreman. And even more fortunately, the foreman did his job. A warehouse foreman who sends you home in a heat wave might be no more useful than a whipman who puts down his whip when the going gets tough.

Surely there are some circumstances on the barge trail when the going gets so tough that you’d want the whipman to ease up on you, and surely if, for example, the warehouse is burning down around you, you’d want the warehouse manager to send you home. But equally surely, the whole point of a whipman — or a warehouse foreman — is that for the most part, you want him to be a stern taskmaster.

Now it’s easy to say that the world would be a better place if nobody had to work to the point of heat prostration in order to earn a living. That’s true, and we should work toward creating such a world. But it’s also true that we do not currently live in such a world, that people will therefore try to do the best they can in this world, and that we do those people no favor by trying to hamper their efforts. Poor but ambitious people will seek and find ways to be part of productive teams. Sometimes those productive teams will require a whipmaster or a harsh foreman as an additional partner. The teams will be glad to find those partners, and I for one will be sufficiently impressed with their ambition that I’ll seek to reward them by buying their products.

In 1992, pressure from the U.S. government led to the closure of Bangladeshi factories that employed 50,000 children in conditions that you and I would surely consider harsh. An interviewer seeking a few words of appreciation from those children instead heard this from a ten-year-old girl named Moyna:

They loathe us, don’t they? We are poor and not well educated, so they simply despise us. That is why they shut the factories down.

Think about that the next time someone suggests pressuring Amazon into changing its labor standards. Once we’ve succeeded in denying those workers the opportunity to earn the rewards that can only come from high productivity, they’ll have every right to suspect that we did it because we despised them.

I don’t despise them. I admire the hell out of them. And that’s one of the reasons why I buy from Amazon.

Print Friendly, PDF & Email
Share

47 Responses to “Amazon’s Bargemen”


  1. 1 1 Harold

    There is a difference between the Amazon workers and the bargemen. The bargemen hired the whipmaster themselves, but the same third party hired the warehouse staff and the whipmaster.

    Your story of the warehouse staff sharing equally in the bounty produced by successfully filled orders is one narrative, but there are others. I believe exploitation is possible because markets are not perfect, but I cannot say whether or not this is such a case.

    The Bangladesh example illustrates that we should not assume everyone values things the same as we do. However, it is possible that conditions now are better than they would have been because of the intervention. For example, if factory owners were using sub-standard materials that the workers could not possibly have known about, and could not therefore include this information in their decision to work there. The USA presure could prompt the owners to use better quality materials.

  2. 2 2 Nick Janusch

    Early 20th century? Surprisingly, workers are still pulling barges today! But now they are pulling tourists and there is no whipping. See for yourself:

    http://youtu.be/4oVlqmCPrn0 (Warning: potentially NSFW–some of the clips show nude barge pullers)

    I have yet to find any valid evidence that whipping occurred back in 19th and 20th century China. There are many versions of this potentially apocryphal story, but I believe Cheung (1983) was first to give the story an economic twist. (Please correct me if I am wrong!)

    The ironic part of the story is that it is suggested that it was the communists that stopped the practice of whipping.

    Cheung, S. N. (1983). The contractual nature of the firm. Journal of Law and Economics, 1-21.

    Also Russ Roberts and Mike Munger discuss the story in this Econtalk podcast:
    http://www.econtalk.org/archives/2008/01/munger_on_the_n.html

  3. 3 3 Steve Landsburg

    Harold:

    There is a difference between the Amazon workers and the bargemen. The bargemen hired the whipmaster themselves, but the same third party hired the warehouse staff and the whipmaster.

    I don’t see the difference. When labor and management work together to produce an outcome, you can say that the labor hired the management or you can say that the management hired the labor. The fundamental arrangement remains the same regardless of what words we use to describe it.

  4. 4 4 Steve Landsburg

    Nick Janusch: I first heard this story from Walter Oi, who presumably got it from Cheung. Cheung supplies no footnote, and I know of no independent source.

  5. 5 5 Keshav Srinivasan

    “Once we’ve succeeded in denying those workers the opportunity to earn the rewards that can only come from high productivity” Steve, aren’t you exaggerating here? Are you saying that there is absolutely no policy that can possibly ensure that these people have a high income without having high productivity? What if we forced Amazon to pay them a higher wage without firing them? What if the government just supplemented their income by writing them a check?

  6. 6 6 Ken

    Keshav,

    Higher pay and income means nothing if productivity hasn’t increased. All it means is that more money is chasing the same number of products, which only pushes up price.

  7. 7 7 nobody.really

    [W]e should work toward creating [an ideal, or at least better] world. But it’s also true that we do not currently live in such a world, that people will therefore try to do the best they can in this world….

    Fair enough; let’s not make the perfect the enemy of the good.

    By the same rationale, we should all work toward making markets more efficient, because efficient markets tend to produce outcomes we favor. But that should not deter us from designing policies in acknowledgement that markets are not perfectly efficient, and that even efficient markets may not always produce the outcomes we favor.

  8. 8 8 Brian

    “This is a classic example of what economists call a Prisoner’s Dilemma — a situation where everyone wants to cheat, regardless of whether he believes the others are cheating. Any bargeman might reason that “If the others are pulling hard, we’re going to make it anyway, so I might as well relax….”

    I don’t think that this really qualifies as a Prisoner’s Dilemma, nor do I think whipping was necessary all that often. Here’s why.

    Presumably, the bargemen got paid a set fee per delivery. Since the fee would be split N ways for N barge pullers, it’s unlikely that they would form teams larger than necessary to deliver the shipment on time. If six pullers were used, it’s likely that the job couldn’t be done with five.

    Consequently, defection by one would cost everyone, including the defector. This makes the setup a coordination game, not a social dilemma.

    Then why would a whip master be needed? Well, he wouldn’t be under normal circumstances. Perhaps the claim is apocryphal. On the other hand, there might be cases where the whip master adds marginal value. If some of the pullers got injured and felt like they couldn’t pull any more, the whip master might convince them otherwise. Or perhaps teams easily lost track of time and the effort needed to get the job done. In that case, the whip master might serve as a kind of pacer, with a bite.

    All in all,though, the claim of a whip master being needed seems improbable to me. Over time, why wouldn’t teams of six form who knew each other well enough to be trusted? Shirkers wouldn’t be able to find a team to take them on. In other words, even under Prisoner’s Dilemma conditions, defection should be avoidable through reputation effects and repeated play.

  9. 9 9 Frosty

    Now it’s easy to say that the world would be a better place if nobody had to work to the point of heat prostration in order to earn a living. That’s true, and we should work toward creating such a world.

    We agree the world would be a better place if nobody had to work to the point of heat prostration in order to earn a living.

    Will rewarding (sending money to) corporations whose employees do so cause more or fewer incidences of it?

  10. 10 10 Roger

    There is a difference between the Amazon workers and the bargemen. The bargemen hired the whipmaster themselves, but the same third party hired the warehouse staff and the whipmaster.
    I don’t see the difference.

    Does Amazon sell blinders?

  11. 11 11 Steve Landsburg

    Frosty:

    Will rewarding (sending money to) corporations whose employees do so cause more or fewer incidences of it?

    Wrong question. Will rewarding corporations whose employees do so make it any less necessary?

  12. 12 12 Frosty

    We agreed fewer incidents was better than more. In fact, zero incidents would be ideal. Therefore, how is it the “wrong question” to ask how an action will influence the number of incidents?

  13. 13 13 Steve Landsburg

    Frosty:

    I believe the world would be a better place if we were all so healthy that we never needed to take medicine. Am I forced to conclude that fewer incidents of medicine-taking are better than more and that therefore the world we live in would be a better place if medicine were harder to get?

  14. 14 14 Neil

    I have looked for a primary source and for evidence of this alleged fact for years. I have never found any. I concluded that it was a contrived anecdote, not unlike the phony antiques (http://community.seattletimes.nwsource.com/archive/?date=20030127&slug=cheung27) Cheung tried to pass off a while back. Nonetheless, it makes an interesting story and good logical point.

  15. 15 15 Frosty

    Your medicine metaphor is misguided. I asked a simple, relevant question, which has four plausible answers: more incidents, fewer incidents, no effect, or don’t know.

    Many believe if you reward a behavior, you’ll get more of it. If so, rewarding (sending money to Amazon and admiration to Amazon’s employees) working to the point of heat prostration will result in more of it.

  16. 16 16 Steve Landsburg

    Frosty:

    Many believe if you reward a behavior, you’ll get more of it. If so, rewarding (sending money to Amazon and admiration to Amazon’s employees) working to the point of heat prostration will result in more of it.

    My point exactly.

  17. 17 17 Martin-2

    “it’s unlikely that they would form teams larger than necessary to deliver the shipment on time. If six pullers were used, it’s likely that the job couldn’t be done with five.”

    Interesting point, but I think it would be risky to use the absolute minimum number of pullers. Any unexpected setback and you’re instantly late. You’d always want some power in reserve.

  18. 18 18 Brian

    “Interesting point, but I think it would be risky to use the absolute minimum number of pullers. Any unexpected setback and you’re instantly late.”

    Martin-2,

    Although my “couldn’t be done” language makes it sound all or nothing, I really view it as being more probabilistic. A team of five perhaps might have a low probability of making it on time (say less than 50%) while a team of six might have a high probability (well above 50%). In any case, assuming there’s no physical limitation on how many can pull the barge, one would expect that a team of six was the number that maximized the payoff for each puller, given whatever unexpected events typically happened.

    The reasoning behind why a whipper is unnecessary still holds. And besides, why share the money with a whipper who does no pulling? Adding a seventh puller would make more sense. The whipping story strikes me as a myth. I just don’t see any good economic reason for such a setup.

  19. 19 19 nobody.really

    1. Would the bargemen be better off with an extra puller rather than a whipper? Presumably the whipper would look like a jockey: physically small, but athletic. He might not be much good as a puller.

    2. But who cares? It’s a metaphor, much like the lighthouse metaphor for public goods. Would someone voluntarily subject himself to burdens simply to have the privilege of inducing others to accept burdens? Yes: it’s called a contract. It happens all the time — although it does not provide a very memorable image, and its enforcement mechanisms are less visceral than a whip.

    3. But this illustrates the weakness of Landsburg’s use of this metaphor as applied to Amazon workers: As others have remarked, each barge worker presumably has a lot of say in how the proceeds of his barge pulling will be divided among his team; is the same true of Amazon workers?

    Landsburg presumes it to be true: “They know that if they all push themselves to the limit, they’ll all be more productive and therefore earn higher wages.” Yet since the 1970s, gains in productivity have greatly exceeded gains in compensation. Thus, as a general proposition, employees have little basis to conclude that their increased productivity will result in increased compensation.

    Amazon, however, is a special case. In general, recent increases in workforce productivity has tended to benefit owners/shareholders – but Amazon famously earns no money. (Indeed, last quarter it lost even more money than analysts had anticipated.) Thus, whatever productivity gains it squeezes out of employees seem to get plowed back into the operations. I tend to think this means lower prices, but it might also include higher compensation.

    Query: Do productivity increases that get plowed back into the business get reflected in the data as productivity increases?

  20. 20 20 Keshav Srinivasan

    Steve, can’t we easily move to a world where these people don’t need to work as hard to make the same amount of money, by giving them taxpayer money?

  21. 21 21 Ken

    Keshav,

    Taking money from one group of workers to give to another group of workers doesn’t reduce how hard the average worker needs to work. Taxpayer money isn’t magicked from thin air. It is taken from other workers. You’re argument is that we can move to a world were “these people” (unclear who “these people” are) unburdened by burdening others (taxpayers).

  22. 22 22 Manfred

    Speaking of child labor (as Steven mentions the case of Bangladesh). There is a story in the Wall Street Journal today (Thursday, 10/30) about child labor in Bolivia, and how Evo Morales (the very leftist Bolivian president) goes against the politically correct Zeitgeist of how to view child labor.

    Newest Legal Laborers in Bolivia: Children
    Newly Re-Elected President Morales Bucks World Trend in Legislating the Right for 10-Year-Olds to Work

  23. 23 23 Steve Landsburg

    Keshav: To whom exactly do you propose making these transfers? To everyone whose income falls below 20% of the mean? 30%? 40%? The mean for the U.S. or the mean for the world? Does Moyna get a transfer? Do we limit the transfers to those who are hardworking? How do you determine who those are? Have you done any arithmetic to figure out the magnitude of the transfers you’re talking about? Exactly what taxes do you plan to raise? Have you estimated their disincentive effects? After accounting for those, do you have any reason to believe that your plan will make poor people richer?

  24. 24 24 Ken B

    Keshav, Roger, and Harold: Steve’s argument is that by operating a social enforcement mechanism Amazon is making its workers better off, and through their united efforts they make the rest of us better off. Keshav’s scheme makes the rest of us worse off, and might make the workers worse off. Certainly if universally and symmetrically applied it would.

    Harold and Roger just miss that Amazon plays a productive role in corralling and workers on behalf of the rest of the workers.

  25. 25 25 Steve Landsburg

    Ken B: As is so often the case, you’ve expressed exactly and perfectly what I was trying to say. Thank you.

  26. 26 26 Frosty

    The enforcement mechanism should persuade employees to operate at peak productivity. At the point of heat prostration, employees are not doing so.

  27. 27 27 Harold

    “Harold: Steve’s argument is that by operating a social enforcement mechanism Amazon is making its workers better off”
    As I said, this is one narrative.

    SL “I don’t see the difference. When labor and management work together to produce an outcome, you can say that the labor hired the management or you can say that the management hired the labor.” So you would not see a difference between a barge owner hiring people to pull his barge and also hiring someone to whip them, and the staff collectively hiring someone to whip themselves? It is something of a stretch for there to be *no* difference In a perfect market possibly. However, if the market is not perfect, then there is a likelihood for the two situations to be different.

  28. 28 28 Jimbino

    If one of the bargemen was a woman, she’d be given time off, with pay, while the others carried on without her.

  29. 29 29 nobody.really

    ???

    If we assume that the other bargemen were heterosexual men, they might prefer to carry on with her. But that might create a hostile work environment — which might detract from the productivity of an friendly, constructive work environment in which people just happen to get whipped.

    Is this blog getting kinkier, or is it just me?

  30. 30 30 Harold

    #28. Or maybe there would be no time off, no children, and in 1 generation, no more bargemen, or anyone else.

  31. 31 31 Steve Landsburg

    Frosty:

    The enforcement mechanism should persuade employees to operate at peak productivity.

    It appears to me that you are not stopping to think for even a moment before posting. There is nothing optimal about stopping at the point of peak marginal productivity. If you’re at your most productive at 10AM, does that mean you shouldn’t work for the rest of the day?

  32. 32 32 Conor

    Could information asymmetry lead to the opposite conclusion?

    Say consumers prefer excellent customer service all else equal, but will forgo it if that service requires anyone to work in especially difficult conditions.

    If consumers can’t easily figure out which companies employ harsh foremen, companies will ignore their preference for kindness towards employees and only aim to provide excellent service.

    Someone (either you or David Friedman, I’m pretty sure) made a similar point about the minimum wage. It very likely becomes a good policy if consumers prefer to frequent firms that pay at least $X per hour, but in the absence of some way to verify that wage rate, they’ll settle for best price and product.

  33. 33 33 nobody.really

    Ken B: “Steve’s argument is that by operating a social enforcement mechanism Amazon is making its workers better off, and through their united efforts they make the rest of us better off….

    Amazon plays a productive role in corralling and workers on behalf of the rest of the workers.”

    Landsburg: “Ken B: As is so often the case, you’ve expressed exactly and perfectly what I was trying to say.”

    I find Ken B’s statement — “by operating a social enforcement mechanism Amazon is making its workers better off, and through their united efforts they make the rest of us better off” – compelling. And if that’s the message Landsburg intended to convey, great.

    But both Landburg and Ken go further. Landsburg says, “They [workers] know that if they all push themselves to the limit, they’ll all be more productive and therefore earn higher wages.” Ken echoes this (somewhat ambiguously) in saying “Amazon plays a productive role in corralling … workers on behalf of the rest of the workers.”

    Why should we conclude that the benefits of the increased productivity flow to Amazon’s workers? I can’t find where either Landsburg or Ken B. supports this assertion.

    So let’s clarify the bargemen analogy to make it more like Amazon: The whipman decides how to allocate the benefits of the bargemen’s efforts, including having the discretion pocket the benefits of increased productivity himself. This is the incentive structure that influences how the whipman wields the whip.

    I don’t mean to suggest such an arrangement would be unfair, provided all parties consent. But it changes the appeal of the bargemen analogy – precisely because it becomes a more apt analogy for conventional employment arrangements.

  34. 34 34 Steve Landsburg

    nobody.really: Why would people be working under Amazon’s (allegedly) harsh labor conditions if not because they can earn higher wages there than elsewhere?

  35. 35 35 Bob Lince

    This whip-master story explains the acceptance of coaches by athletes.

  36. 36 36 Steve Landsburg

    Bob Lince: Great point!

  37. 37 37 nobody.really

    Why would people be working under Amazon’s (allegedly) harsh labor conditions if not because they can earn higher wages there than elsewhere?

    1. I don’t know whether Amazon has harsh labor conditions or not. Landsburg implies that they do when he applies the bargeman analogy to explain Amazon’s superior performance. For all I know, Amazon achieves superior performance by exploiting a first-mover advantage, ruthless wholesaler negotiations, and having great logistics. But those advantages would not provide Landsburg with an opportunity to use his bargeman analogy.

    2. More generally, I expect that harsher labor conditions can lead to greater productivity, and that in general management would need to provide workers with higher compensation to endure harsher working conditions.

    My point is that there’s no basis for knowing how much higher compensation. The bargemen analogy suggests the laborers control how much of the benefits of the productivity increases flow to the whipmaster. Yet as I noted previously, even as labor has grown more productive in recent decades, the benefits of these productivity increases have accrued primarily to owners, not workers. It’s far from obvious that this is the allocation workers would have chosen.

  38. 38 38 Dave Fernig

    This analysis, like a lot of what one might label as “socio economics” fails miserably to take into account some of the basic biology of Homo Sapiens. Even as toddlers, humans cooperate without any possibility of gain – that is to enrich the other, but not oneself. Chimps, our closest relatives do not do this.
    Another issue that is not addressed: why do we need harsh conditions to get the workers to deliver, yet for the CEO we need softer and softer conditions to ensure greater productivity. Since, perhaps surprisingly to some, both workers and CEO are all Homo Sapiens, what is it in one group that requires whipping and in the other massive reward?
    Until this is explained, then I am afraid the arguments in this post do not work and one can only fall back on an explanation based on exploitation.

  39. 39 39 Mark Cancellieri

    It’s really very simple. Amazon employees choose to work for Amazon of their own free will. They work there because they WANT to work there.

  40. 40 40 Charles G. Phillips

    Do Amazon warehouse workers operate as teams or as individual pickers? Are Amazon warehouse workers compensated on an hourly wage basis, on a piecework basis, on team performance or on warehouse performance? My guess is that Amazon has the data and has done the engineering required to analyze the performance of each warehouse employee, and employees understand how they are being evaluated. Those who value their jobs and wish to progress work hard to meet or exceed evaluation standards, those who do not are replaced by more highly motivated individuals–no whipping required

  41. 41 41 Frosty

    Enforcement mechanisms can work and can also error. Errors, including lower productivity, could result from being either too lax or too harsh. My point is employees suffering from heat prostration could reasonably be a clue that the foreman, at least in that particular case, was too harsh. None of us has sufficient information to conclude whether Amazon should (or did) calibrate the enforcement mechanism due the alleged incident.

  42. 42 42 Steve Landsburg

    Dave Fernig:

    Even as toddlers, humans cooperate without any possibility of gain – that is to enrich the other, but not oneself.

    I’m sure we can agree on this. Can we also agree that humans sometimes behave less then 100% altruistically, and that in those cases enforcement mechanisms can be useful?

  43. 43 43 Jim Rose

    Sports related professions hire their own boss or whip.

    The Coxswain in professional rowing is an example where the crew hire someone to shout at you.

    Tennis and chess players hire coaches to stiffen their play.

    University faculty have been known to have the power to elect their head of Department who in turn has the power to fire faculty for misconduct.

  44. 44 44 Jim Rose

    Ed Lazear wrote extensively on labour managed firms such as in the professions where the workers were appointed a chief executive.

    see Peer Pressure and Partnerships, Eugene Kandel and Edward P. Lazear
    Journal of Political Economy, Vol. 100, No. 4 (Aug., 1992)

    cooperatives such as of taxi drivers also hire management teams to discipline them and fire drivers who misbehave.

  45. 45 45 Cyril Morong

    This article backs up Landsburg. From the NY Times business section, 9-27-14, “Looking at Productivity as a State of Mind”
    http://www.nytimes.com/2014/09/28/upshot/looking-at-productivity-as-a-state-of-mind.html

    “Greg Clark, a professor of economics at the University of California, Davis, has gone so far as to argue that the Industrial Revolution was in part a self-control revolution. Many economists, beginning with Adam Smith, have argued that factories — an important innovation of the Industrial Revolution — blossomed because they allowed workers to specialize and be more productive.
    Continue reading the main story Continue reading the main story
    Continue reading the main story

    Professor Clark argues that work rules truly differentiated the factory. People working at home could start and finish when they wanted, a very appealing sort of flexibility, but it had a major drawback, he said. People ended up doing less work that way.

    Factories imposed discipline. They enforced strict work hours. There were rules for when you could go home and for when you had to show up at the beginning of your shift. If you arrived late you could be locked out for the day. For workers being paid piece rates, this certainly got them up and at work on time. You can even see something similar with the assembly line. Those operations dictate a certain pace of work. Like a running partner, an assembly line enforces a certain speed.

    As Professor Clark provocatively puts it: “Workers effectively hired capitalists to make them work harder. They lacked the self-control to achieve higher earnings on their own.”

    The data entry workers in our study, centuries later, might have agreed with that statement. In fact, 73 percent of them did agree to this statement: “It would be good if there were rules against being absent because it would help me come to work more often.””

  46. 46 46 Tjd

    Not sure how you can jump from adults imposing unto themselves a whip master to child labor in harsh circumstances.. I wonder, do you advocate re-introducing child labor in this country as well? I assume by your logic it would be a step forward?

  47. 47 47 Kirill

    Correct me if i m wrong but looks like nobody has mentioned the following that this whipmaster argument can be easily extended to justify the existense of state redistribution of income as opposed to private charity… Say some rich people are keen to proceed with their charity activity but are uncomfortable that other rich people do not join enthusiastically… Solution – hire whipmaster (the state) and force everyone to be charitable…

Leave a Reply