What Can Roman Generals Teach us About the Anti-Sweatshop Movement?

We have all heard the expression, “don’t burn your bridges,” a caveat against rash behavior that can make reconciliation impossible. But to the Roman army, burning bridges or burning one’s boats was actually an effective military strategy. Though detailed historical accounts are elusive, Julius Caesar and Roman generals of the time were said to set fire to their boats during an invasion or to a bridge that could be used for retreat. By making retreat impossible, the generals intended to stiffen the resolve of their troops to conquer or be killed. This effect was not likely lost on their adversary, who fearing the must-win attitude of the mighty Roman army, would then be more likely to retreat.

To be sure, burning bridges or boats was a costly maneuver for the Roman army. Besides eliminating a critical tactical option, it made efforts of re-supplying troops a logistical nightmare. But if it led to the opposing army laying down their arms in surrender or retreating sooner than they otherwise would, such concerns would be negated.

The strategy of burning one’s bridges is one example of a commitment in the parlance of game theory. Game theory is the study of strategic interaction, using mathematical models to approximate real-world environments and predict real-world outcomes. Thomas Schelling, currently of the University of Maryland, was the first to apply the logic of game theory to the understanding of international conflict. In his 1960 book “The Strategy of Conflict,” Schelling discusses ways in which a strategic commitment—a binding promise to follow a specific course of action—can benefit the party making the commitment. With respect to the Romans, eliminating any means of retreat forced a difficult decision on the adversary: retreat or face a bloody battle. If the battle appeared unwinable for the adversary, their best response would be retreat and avoid a pointless battle. In that case, the commitment to battle by the Romans obviated the need for such a battle, much to their benefit.

Schelling was honored with the Nobel Prize for his work on “The Strategy of Conflict” due in large part to the applicability of his ideas to other economic settings, like boycotts. A movement is currently underway at Cornell University, my academic home, to convince the administration to join the Designated Suppliers Program (DSP), a consortium of universities advocating fair labor standards. The program, devised by Worker Rights Consortium (WRC), will require producers of all officially licensed apparel of member schools to demonstrate compliance with internationally recognized labor standards, allow workers to be represented by a union, and pay a living wage.

The global anti-sweatshop movement, of which the WRC is one component, is predicated on the notion that working conditions in sweatshops are inhumane. Having spent a summer in college painting houses outdoors, I can honestly say there is nothing inhumane about sweating on the job. The trouble with sweatshops is low pay and working conditions that are unhealthy and often unsafe. My earlier piece on coal mine safety argued that working conditions and pay are really two sides of the same coin, since workers in unpleasant jobs are paid more than comparable workers in more pleasant jobs. For this reason, we can boil down the demands of the anti-sweatshop movement to a call for higher wages, whether in the form of higher pay, better working conditions, or some combination of the two. In an effort to have its demand met, groups of anti-sweatshop consumers typically threaten to boycott the heartless manufacturer. However, the analysis presented herein will show that unless the consumer group can commit itself in advance to undertaking the boycott, the threat alone will fail to deter the manufacturer from its calculating behavior.

To analyze the interaction between a consortium of anti-sweatshop consumers and a manufacturer, we must utilize the tools of game theory. This involves first defining possible actions for each player and the payoffs to each potential outcome. The game involves two players, the consortium and the manufacturer. The consortium can boycott or continue to do nothing at all while the manufacturer can maintain its current level of low wages or pay its workers a living wage. The consortium gains some amount L if successful in convincing the manufacturer to pay a living wage. However, by resorting to a boycott, they incur a cost B, a result of either buying from a less-preferred manufacturer or not buying the product at all. The manufacturer loses some amount R in lost revenue from the boycott. To appease the consortium, it may decide to pay a living wage, which increases its production cost by some amount C. The payoffs are represented in the game tree found by following this link: see the game tree.

Predicting the outcome of the game requires an assumption that each player in turn chooses the best strategy for itself given the strategy of its opponent. Since the manufacturer acts first, we work backward by considering the best strategy for the consortium following each of the possible strategies for the manufacturer.

Suppose the manufacturer chooses to pay a living wage. The consortium can then boycott and receive a payoff of L-B or do nothing and receive L. Since L is greater than L-B, the consortium will choose to do nothing in the event the manufacturer raises wages.

Now suppose the manufacturer chooses to keep wages low. The consortium can either boycott yielding a payoff of –R or do nothing yielding a payoff of 0—the status quo payoff. Since 0 is greater than –R, the consortium will choose to do nothing in the event the manufacturer keeps wages low.

The upshot is, regardless of what the manufacturer chooses, the consortium is better off doing nothing than boycotting. Since the manufacturer’s decision has already been made, it does the consortium no good to boycott. The threat to boycott in this setting is nothing more than cheap talk.

This result is taken into account by the manufacturer when making its decision. Since it knows the consortium will do nothing, the manufacturer can either pay a living wage which yields a payoff of –C or keep wages low which yields a payoff of 0. Since 0 is greater than –C, the manufacturer chooses to keep wages low.

From what we have seen thus far, it would seem as though the threat to boycott cannot be effective at inducing change as it will necessarily be viewed as empty. But if the consortium of concerned consumers can take a page from the Roman’s playbook and commit to the boycott in advance of the manufacturer’s decision, their threat can no longer be ignored. Such a commitment can take the form of a written agreement like the one required of DSP members. More commonly though, commitments take the form of allegiance to specific standards of human rights that an ideologically driven organization can stake its name to.

Focusing again on the game at hand, suppose the consortium commits itself to the following strategy: do nothing if the manufacturer raises wages, boycott if it doesn’t. Since the consortium’s strategy has already been determined in advance, the manufacturer chooses between keeping wages low and facing a boycott or raising wages and avoiding it. What it chooses depends on which is greater, the lost revenue from a boycott or the cost of paying its workers a living wage. Thus if the consortium makes up a sufficiently large component of the manufacturer’s customer base, it can force the manufacturer to raise wages since doing so would be less costly than a boycott. Of course if the consortium is not large enough to do so, it would be foolish to commit to such a strategy.

We have thus far used a simple mathematical model to analyze two scenarios that depend upon whether the consortium can commit itself to a boycott. But in neither case does a boycott actually take place. Since we do observe boycotts in reality, it must be that the mathematical model does not adequately describe reality, right?

Actually, the mathematical model can account for boycotts only when one player knows something that the other doesn’t. To see this, consider a consortium whose membership is so zealous that they actually enjoy boycotting if it means standing up for a righteous cause. Unlike the type of consortium considered previously, this group will choose to boycott if the manufacturer fails to raise wages, and will do so without the need to commit to the course of action. Realizing this, the manufacturer will choose to raise wages as long as the potential loss of revenue exceeds the increase in wage cost and once again the boycott is averted. But if the manufacturer doesn’t know whether the consortium it is dealing with is the type considered previously or the type that enjoys boycotting, it will keep wages low until it observes people actually boycotting. In this way, the boycott signals to the manufacturer that they are indeed the type of consumers who enjoy boycotting and will do so until their demands are met.

Development economists tend to view sweatshops as a symptom and not the cause of abject poverty in the developing world. If Nike’s workers are there by choice, what does that say about alternative employment options? They must be pretty awful. For this reason, development economists tend to prefer measures aimed at improving overall economic welfare not just the welfare of those employed by large American corporations.

Lacking the funds, know-how, or foresight necessary to solve the larger development problem, organizations like the Workers Rights Consortium target American employers in the developing world through threats of boycott. A boycott or threat of one can be effective only if the cost to the employer through lower revenue exceeds the cost of increasing wages. But that is not sufficient. Just as Roman generals committed themselves so thoroughly to conquest by eliminating the option of retreat, so too must anti-sweatshop organizations commit themselves to boycott if they are to be successful.

Landsburg's Internet-Popcorn Paradox

One of my life’s greatest joys is surfing over to Slate.com and finding a new piece by Steven Landsburg. Landsburg is the Armchair Economist. His regular contributions to Slate explore the hidden economics in day-to-day activities. His latest explains why some hotels charge extra for internet access and others don’t, then leaves the reader with an unsolved puzzle. (See the link to view the article: http://www.slate.com/id/2135226/ . I assume in what follows that the reader has read the article.)

The puzzle identified by Landsburg is as follows: Some hotels charge extra for internet access while others include internet access in the price of the room. This makes sense as hotels with different types of clientele will want to charge differently for internet access. But movie theatres differ in their clientele just as much as hotels do, but seemingly all movie theatres charge separately (and exorbitantly) for popcorn. The question is, what makes hotels and movie theatres so different?

The issue Landsburg analyzes belongs to the economic literature on product bundling, where bundling refers to the practice of pricing two goods as if they were one product. Like the literature, Landsburg shows that there exists conditions under which bundling is profitable and under which it is not. One condition he doesn’t address is that of cost. According to the literature, the strategy of bundling is less likely to be profitable when the bundled good is sold to buyers whose willingness to pay for the add-on is less than the marginal cost. To see why, let’s stay with the hotel example but say the add-on is towel service rather than internet access. Suppose the hotel is charging $120 per night for the bundled good, but many of its guests stay only one night and are not willing to pay anything for towel service though they will pay $120 for the room. By charging separately for towel service, the hotel can continue to charge these people $120 for the room alone, while saving money by not hiring as many chamber maids. This is not an issue as far as internet access is concerned as the marginal cost of internet access is zero—not so for movie theatre popcorn. The lesson for the movie theatre is, if the marginal cost of popcorn is substantial enough, selling the popcorn separately is more profitable than bundling it with the movie ticket regardless of demand conditions.

This explains why bundling may be profitable for a hotel but never for a movie theatre. But why then does my local pub offer popcorn for free, implicitly bundling the popcorn with the beer? For one thing, popcorn and beer are complementary goods so the money lost on popcorn is more than made up for on additional beer sales. But there is also a cost difference between offering free popcorn in a pub and doing so in a movie theatre. Can you guess what it is? I’ll give you a hint: it involves a teenage boy and a broom. That’s right, the difference is in the cleanup cost. Movie theatre popcorn must be swept up after each showing, while pub popcorn is swept up at most once a night. Plus, the popcorn at the pub is self-serve, so you don’t bother getting it unless you plan on eating it. At the theatre, the popcorn is served to you on the way in. And any popcorn served to someone who values it little is likely to end up on the floor. The upshot is, the cost of serving free popcorn in a movie theatre may be substantial enough to preclude bundling when you take into account the clean-up cost.

I have to say I’m not completely happy with this answer. I would have preferred an explanation relying more on game theory, but that’s what I came up with and it does seem to fit. Please post a comment if you think I’ve missed something.