The Limits of Self-Enforcing Agreements

I agree with many of the stepping stones in Leeson’s argument, but his bottom line—that a regime of state-less self-organization can generate the institutional underpinnings of economic prosperity—represents a huge leap of faith. It is supported neither by theory, nor by any systematic empirical evidence that I am aware of.

Here is how I would restate Leeson’s conclusions (my additions are in bold):

I have argued that anarchy sometimes works better than you think. In the face of obstacles that stand in the way of individuals’ ability to cooperate for mutual gain, individuals have the incentive to develop solutions to overcome these obstacles. This is as true in society ruled by government as one that exists without government. Where the state does not provide law, order, or the institutions required to produce these things, there is a tendency for private institutions to emerge to perform these roles instead. The overwhelming balance of evidence suggests, however, that the large-scale division of labor on which modern economic prosperity depends can be sustained only with state-like institutions—i.e., public enforcement of laws and contracts.

I do not have any trouble with the idea that self-enforcing agreements (what Leeson calls “anarchy”) can sometimes substitute for third-party (i.e., government) enforcement. Such self-enforcing agreements are maintained through the force of repeated interaction (“if you cheat me now, I will cheat you in the future,”) through reputational mechanisms (“see, I am not the cheating kind of guy”), and collective punishment schemes (“if you cheat me, I will bring the wrath of my colleagues on you”). The literature is replete with examination of such informal institutions. See for example Avner Greif’s work on medieval merchant guilds, John McMillan and Chris Woodruff’s work on commercial dispute settlement in Vietnam, Marcel Fafchamps’ work on firms’ relations with their suppliers in Africa, and Elinor Ostrom’s work on the management of common property resources around the world. Leeson’s own account of how pirates have developed self-enforcing arrangements to elicit cooperation fits squarely in this larger literature.

The problem with self-enforcing agreements is that they do not scale up. One of the findings from Elinor Ostrom’s extensive case studies is that self-enforcing arrangements to manage the “commons” work well only when the geographic scope of the activity is clearly delimited and membership is fixed. It is easy to understand why. Cooperation under “anarchy” is based on reciprocity, which in turn requires observability. I need to be able to observe whether you are behaving according to the rules, and if not, I have to be able to sanction you. When the size of the in-group becomes large and mobility allows opportunistic behavior to go unpunished, it becomes difficult to maintain cooperation. Imagine that the pirates numbered in the millions and they could easily jump ship to join competing groups mid-voyage; would the arrangements Leeson describes have been sustainable?

Unlike in pirate societies or pre-colonial Angola, modern economies require an elaborate and ever-evolving division of labor—among owners of firms, managers, and their employees, among producers up and down the value chain, and between producers and providers of supporting services such as finance, accounting, and legal services. The complexity, fluidity, and geographic non-specificity of these activities leave too much room for opportunistic behavior for self-enforcing arrangements to work well. They require an external backstop in the form of government-enforced rules.

This point is underscored in the work of Avinash Dixit and J.S. Li, who have compared the mechanics behind informal, self-enforcing arrangements with those of formal institutions with external enforcement. Here is Dixit (2006):

A relation-based system of networks and norms has low fixed costs, but high and rising marginal costs. Trading on a small scale naturally starts among the most closely connected people who have sufficiently good communication and common understanding to sustain honesty. No fixed costs need be incurred to establish any formal rules or mechanisms of enforcement. But as trade expands, potential partners added at the margin are almost by definition less well-connected, making it harder to communicate information with them and to ensure their participation in any punishments. By contrast, formal or rule-based governance has high fixed costs of setting up the legal system and the information mechanism, but once these are incurred, marginal costs of dealing with strangers are low. Therefore relation-based governance is better for small groups and rule-based governance better for large groups.

A modern post-industrial economy requires rule-based governance because it needs to elicit cooperation within a large group—that of the nation-state.

Which is why the scatter plot below, showing the relationship between per-capita GDP and the size of the public sector, should not be a surprise. There is a strong, statistically highly significant, and positive association between countries’ income levels and the share of their economy that the government consumes. This highlights the complementarity between markets and the state. Those societies in which markets work best are the ones where the reach of the state is longer—not shorter. Prosperity is achieved when states are effective in setting and enforcing the rules of the game, not when they wither away. It is possible that the Somali people are better off under anarchy than under a predatory state. Certainly no one believes that “any government is always superior to no government.” But to think that the Somali can ever develop economically without an effective state apparatus is to commit the flip side of the Nirvana fallacy: to believe that a better state than what exists currently is not achievable just because a perfect state is not possible. There is no example of a society that has become prosperous without a state machinery.

The chart shows the relationship between (the logarithm of) per-capita GDP and the share of government consumption in GDP in 2003, for all countries for which data exist in the World Development Indicators of the World Bank. The estimated slope coefficient is 1.5, and is statistically significant at more than 99%.

It may be objected that the operation of the global economy is proof in itself that a high level of economic activity can be maintained without political institutions. After all, we do not have a global central bank, a global anti-trust authority, a global court, and so on—not to mention a global government. Leeson writes: “we do not observe perpetual world war in the absence of global government, [nor] shriveling international commerce in the absence of supranational commercial law…” But Leeson is overlooking several things. First, there is in fact a significant global institutional architecture that supports the international economy: globalization would not have reached this far in the absence of the WTO, IMF, World Bank and a host of regional supranational institutions. The global public sector is not non-existent. Second, it is often national legal and political institutions that provide the backstop where international institutions prove inadequate. International commercial arbitration often takes place and is enforced in third country legal jurisdictions. And third, precisely because the arrangements just mentioned in the preceding are inadequate, the world economy remains full of transaction costs and is subject to all kinds of syndromes that were banished long ago from national economies. Financial panics are a thing of the past at home thanks to the Fed; but they remain rampant in international finance because there is no international lender-of-last-resort.

Leeson writes: “Most of the world, for most of its history, has existed without effective governments.” Indeed. That is why most of the world for most of its history has remained poor, with lives that are nasty, brutish, and short.


Dixit, Avinash, “Economic Governance.” To appear in S. Durlauf and L. E. Blume eds., The New Palgrave Dictionary of Economics, forthcoming, Palgrave Macmillan, March 2006.

Dani Rodrik is professor of international political economy at the John F. Kennedy School of Government, Harvard University.

Also from this issue

Lead Essay

  • Everybody seems to know we need government … But pirates didn’t! How did they manage without the state? In this issue’s thought-provoking lead essay, Peter T. Leeson, the BB&T Professor for the Study of Capitalism at George Mason University, explores what pirate “constitutions,” credit institutions among 19th century African bandit traders, and the well-being of Somalians after the collapse of the Somalian state have to tell us about the possibility of practical anarchy. It works better than you think, Leeson concludes. “As long as there are unrealized gains to realize, people will find ways to realize them” — state or no state.

Response Essays

  • Bruce L. Benson, author of The Enterprise of Law: Justice Without the State, argues Peter Leeson’s defense of anarchy is too moderate. Governments in developed nations, Benson maintains, are not better than ordered anarchy. Drawing on Franz Oppenheimer’s classic account of the state as a protection racket, Benson argues that the state only seems necessary because it offers “solutions” to problems the state itself creates. Benson claims that even well-constrained states are essentially parasitic, leading him to conclude that “even when a relatively ‘good’ government exists, there still is way too much government and not nearly enough anarchy.”

  • Harvard economist Dani Rodrik is willing to accept a number of steps in Peter Leeson’s argument for anarchy, “but [Leeson’s] bottom line … represents a huge leap of faith.” Citing the work of several important thinkers, Rodrik argues that “the problem with self-enforcing agreements is that they do not scale up.” Both theory and data show that complex, well-functioning social and economic systems require the enforcement of rules by government. “Those societies in which markets work best are the ones where the reach of the state is longer, not shorter.”

  • Florida State University economist Randall Holcombe argues that even if Leeson is right about anarchy, it doesn’t much matter. “Regardless of its merits,” Holcombe writes, “anarchy has no prospect as an actual policy option.” The bottom line is that government is popular in developed nations. Furthermore, anarchy may not be a “stable equilibrium,” in which case it might “coalesce into governments … potentially more oppressive and more destructive than those we see in prosperous areas today.” According to Holcombe, if we’re going to get a government anyway, the best approach to policy is to “make it smaller, less intrusive, and more libertarian,” not to make it go away.