A good reputation can be an effective bond for honest behavior in a community of traders if members of the community know how others have behaved in the past - even if any particular pair of traders meets only infrequently. In a large community, it would be impossibly costly for traders to be perfectly informed about each other's behavior, but there exist institutions that can restore the effectiveness of a reputation system using much less extensive information. The system of judges used to enforce commercial law before the rise of the state was such an institution, and it successfully encouraged merchants (1) to behave honestly, (2) to impose sanctions on violators, (3) to become adequately informed about how others had behaved, (4) to provide evidence against violators of the code, and (5) to pay any judgments assessed against them, even though each of these behaviors might be personally costly.

DEFINE: This describes how pre-state legal institutions (judge systems) served to maintain reputation-based commercial enforcement with limited information.

How can people promote the trust necessary for efficient exchange when individuals have short run temptations to cheat? The same question arises whether the traders are legislators swapping votes, medieval merchants exchanging goods, or modern businesspeople trading promises about future deliveries. In each of these situations, one of the important ways in which individuals ensure one another's honest behavior is by establishing a continuing relationship. In the language of economics, if the relationship itself is a valuable asset that a party could lose by dishonest behavior, then the relationship serves as a bond: a trader would be unwilling to surrender this bond unless the gain from dishonest behavior was large.

DEFINE: Defines the economic concept of a relationship as a valuable bond that deters cheating, since dishonest behavior could forfeit that asset. XREF: Connects to repeated-game theory and reputation mechanisms in economics, plus game-theoretic treatments of cooperation and trust.

Even in a community in which any particular pair of people meet rarely, it is still possible (as we show) for an individual's reputation in the group as a whole to serve as a bond for his good and honest behavior toward each individual member. This illustrates the important fact that a reputation system may sometimes work only when it encompasses

We embed our study of these questions in the time of the revival of trade in Europe during the early middle ages. At that time, without the benefit of state enforcement of contracts or an established body of commercial law, merchants evolved their own private code of laws (the Law Merchanr) with disputes adjudicated by a judge who might be a local official or a private merchant. While hearings were held to resolve disputes under the code, the judges had only limited powers to enforce judgments against merchants from distant places.

XREF: Relates to institutional economics and the emergence of self-governing trade networks, comparable to modern freelance reputation economies.

The evolution and survival for a considerable period of a system of private adjudication raises both particular versions of our general questions and new questions about the details of the mechanism. What was the purpose of the private adjudication system? Was it a substitute for the reputation mechanism that had worked effectively in earlier periods (Greif, 1989)? Also, if there was no state to enforce judgments, how did they have any effect? How could a system of adjudication function without substantial police powers?

XREF: Connects to the reader's apparent interest in how order emerges without state power, resonating with reputation-mechanism literature (Greif) and spontaneous-order theory. QUESTION: Worth chasing how the adjudication mechanism gained compliance absent police powers—an intriguing puzzle in institutional economics.

The practice and evolution of the Law Merchant in medieval Europe was so rich and varied that no single model can hope to capture all the relevant variations and details. Our simple model is intended to represent certain universal incentive problems that any successful system would have to solve.

XREF: Connects to institutional economics and game-theoretic analysis of how customary law systems like the Law Merchant function as self-enforcing governance.

We argue in section 2 that even if no pair of traders come together frequently, if each individual trades frequently enough within the community of traders, then transferable reputations for honesty can serve as an adequate bond for honest behavior if members of the trading community can be kept informed about each other’s past behavior. Well informed traders could boycott those who have violated community norms of honesty, if only they knew who the violators were. It is the costliness of generating and communicating information - rather than the infrequency of trade in any particular bilateral relationship - that, we argue, is the problem that the system of private enforcement was designed to overcome.

DEFINE: Defines transferable reputations and the private enforcement mechanism where information costs, not trade frequency, are the core obstacle to honest exchange. Builds on: "Group-wide reputation can enforce honest behavior among strangers."

We argue in section 2 that even if no pair of traders come together frequently, if each individual trades frequently enough within the community of traders, then transferable reputations for honesty can serve as an adequate bond for honest behavior if members of the trading community can be kept informed about each other's past behavior. Well informed traders could boycott those who have violated community norms of honesty, if only they knew who the violators were. It is the costliness of generating and communicating information - rather than the infrequency of trade in any particular bilateral relationship - that, we argue, is the problem that the system of private enforcement was designed to overcome.

In section 3, we introduce our basic model of a system of private enforcement and develop our core thesis that the role of the judges in the system, far from being substitutes for the reputation mechanism, is to make the reputation system more effective as a means of promoting honest trade. The formal system is more complex than the simple informal system of reputations that preceded it, but that was a natural outcome of the growing extent of trade.

XREF: Connects to literature on private enforcement and informal reputation markets in institutional economics, e.g., Milgrom, North, and Weingast.

So the system of private judges is designed to promote private resolution of disputes and otherwise to transmitjusr enough information to the right people in the right circumstances to enable the reputation mechanism to function effectively for enforcement. In order to succeed, such a system must solve a number of interconnected incentive problems: Individual members of the community must be induced to behave honestly, to boycott those who have behaved dishonestly, to keep informed about who has been dishonest, to provide evidence against those who have cheated, and to honor the decisions of the judges. All of these problems can be resolved by the system if certain institutional constraints are satisfied, as we show in section 3. Briefly, the costs of making queries, providing evidence, adjudicating disputes, and making transfer payments must not be too high relative to the frequency and profitability of trade if the system is to function successfully.

XREF: Relates to economic and game-theoretic work on reputation mechanisms and self-enforcing institutions like Ostrom's work on commons governance. Builds on: "Shared honesty reputations can enforce honesty in dispersed trading communities."

Intuitively, the system of private judges accomplishes its objectives by bundling the services which are valuable to the individual trader with services that are valuable to the community, so that a trader pursuing his individual interest serves the community's interest as well. Unless a trader makes appropriate queries, he cannot use the system to resolve disputes. The requirement that the traders make queries provides an opportunity for the judge to collect payments for his services even if no actual disputes arise.

Prior to the revival of trade in the early middle ages, few institutions underpinned commercial activity; there was no state to enforce contracts, let alone to protect merchants from pirates and brigands. In contrast, modern Western economies possess highly specialized systems of enforcing contracts and protecting merchants, resulting in widespread geographic specialization and impersonal exchange. The story of this evolution has been told elsewhere (e.g., Lopez, 1976; North and Thomas, 1973). Our purpose in this section is to suggest the outlines of an important step in this evolution, namely the early development of commercial law prior to the rise of large-scale third-party enforcement of legal codes by the nation-state.

XREF: Connects to institutional economics and the question of how informal/private ordering sustains markets before the state (cf. Ellickson's work on order without law). Builds on: "Merchants created private law systems without state enforcement"

A large number of problems had to be resolved in order to support the expansion of trade. First, as trading communities grew larger, it became harder within each community for merchants to monitor one another's behavior. New institutions were required to mitigate the types of cheating afforded by the new situation. Second, as trade grew among different regions, institutions were needed to prevent reneging by merchants who might cheat in one location, never to be seen again.

XREF: Connects to institutional economics and the classic problem of trust in impersonal exchange, echoing Greif's work on merchant guilds and reputational mechanisms. Builds on: "Commercial law preceded large-scale state enforcement of contracts"

Key to understanding the ability of merchants from widely varying regions to enforce contracts was the evolution of the L a Mercatoria or Law Merchant - the legal codes governing commercial transactions and administered by private judges drawn from the commercial ranks. While practice varied across time and space, by the end of the 1 Ith century, the Law Merchant came to govern most commercial transactions in Europe, providing a uniform set of standards across large numbers of locations (Benson, 1989). It thereby provided a means for reducing the uncertainty associated with variations in local practices and limited the ability of localities to discriminate against alien merchants (Berman, 1983; Trakman, 1983).

XREF: Connects to institutional economics and the debate over whether legal order requires state enforcement, echoing work on private governance and reputation. QUESTION: What were the actual enforcement mechanisms when a private judge ruled against a merchant? Asset seizure, ostracism, or trade sanctions? SEED: A possible essay contrasting medieval private commercial law with modern state-centric regulation as evidence that markets can self-order.

While the governments of towns supported the development of markets and were intimately involved in developing merchant law (Pirenne, 1925; Rorig, 1967), they often could not provide merchants protection outside their immediate area.* Nor could they enforce judgments against foreign merchants who had left town prior to a case being heard. Thus, merchant law developed prior to the rise of a geographically extensive nation-state. But this raises a key problem in the theory of enforcement, for what made these judgments credible if they were not backed up by the state? Ostracism played an important role here, for merchants that failed to abide by the decisions of the judges would not be merchants for long (Benson, 1989; DeRoover, 1963; Trakman, 1983).

XREF: Relates to informal enforcement mechanisms and reputation-based governance, echoing ideas from game theory and Ostrom-style commons management. Builds on: "Medieval European trade ran on private, self-governed commercial law."

The Law Merchant and related legal codes evolved considerably over time. In addition to providing a court of law especially suited for merchants, it fostered significant legal developments that reduced the transaction costs of exchange (North, 1989, ch. 13). As agency relationships became common - whether between partners in different locations or between a sedentary merchant who financed a traveling one - a new set of rules governing these agreements was required. The same also held for the new practices of credit agreements and insurance.

XREF: This connects to institutional economics literature (North's work) on how legal frameworks lower transaction costs and enable economic exchange.

In these circunistances the cheated merchant might be able to get a judgment against his supplier, but what good would it do if the supplier never returned to the Fairs? Perhaps ostracism by the other merchants might be an effective way to enforce the payment of judgments. However, if that is so, why was a legal system needed at all?

QUESTION: This raises a genuine puzzle about why formal legal systems evolved if informal mechanisms like ostracism could work. Worth chasing the historical answer. Builds on: "Merchant law thrived on ostracism before strong nation-states emerged"

Another part of the inspiration for our formal model is the system of notaries that was widely used to register the existence of certain types of contracts and obligations. Typically, notaries were used for long-term contracts such as those for apprenticeships, sales of land, and partnerships (Lopez and Raymond, 1955). The extensive use of notaries in certain areas to register agreements suggests that reputation via word of mouth alone was insufficient to support honest behavior and that a third party without any binding authority to enforce obligations was nonetheless quite valuable for promoting honest exchange.

Builds on: "Group-wide reputation can enforce honest behavior among strangers."

To represent the idea that cheating may be profitable in a simple exchange, we use the Prisoners' Dilemma (PD) game as our model of a single exchange transaction. Although this PD model is too simple to portray the richness of even simple contracts, it has the advantage that it is very well known and its characteristics in the absence of institutions have been thoroughly studied, so that the incremental contribution made by the Law Merchant system will be quite clear. Moreover, the PD game represents in an uncluttered way the basic facts that traders have opportunities and temptations to cheat and that there are gains possible if the traders can suppress these temptations and find a way to cooperate.

It is clear that if this game is played only once, it is in each player’s separate interest to play Cheat, since that play maximizes the player’s individual utility regardless of the play chosen by the competitor. Consequently, the only Nush equilibrium of the game is for both to play Cheat. Then both are worse off than if they could somehow agree to play Honest.

XREF: Classic prisoner's dilemma; connects to game theory concepts like Nash equilibrium and collective action problems.

Now suppose that the players trade repeatedly. Let ail represent the action taken by player i in period t; let ~ ; ( u , , , u2,) represent the resulting payoff earned by player i in period t; and let 6 be the discount factor applied to compute the present value of a stream of payoffs. If trade is frequent, then 6 is close to one; if trade occurs only once (or is quite infrequent), then 6 is (close to) zero.

DEFINE: Clarifies that the discount factor δ reflects trade frequency, not just time preference — a useful interpretive nuance.

Consider the Adjusted Tit-for-Tat (ATFT) strategy according to which player i plays Honest at date 0 and then plays Cheat at date t + 1 if two conditions hold: (1) i made the play at date t that was specified by his equilibrium strategy and (2) M(h,, i ) did not make the play at date t that was specified by his equilibrium strategy. If either condition fails, then the ATFT strategy calls for i to play Honest. The ATFT strategy formalizes the idea that a trader who cheats will be punished by the next merchant he meets if that merchant is honest, e v h if that merchant is not the one who was cheated.

For 6 near enough to one - specifically if 6 2 Max[p/(l + p), (a - 1)/(1 + @)I (2) - the Adjusted Tit-for-Tat strategies are a subgame perfect equilibrium in the community trading game for any matching rule M.

Our formal analysis verifies that it is not necessary for any pair of traders to interact frequently - that is, for traders to establish client relationships - in order for the boycott mechanism to be effective.

Builds on: "Adjusted Tit-for-Tat sustains cooperation across any matching rule."

For example, merchants engaged in long-distance trade could not be expected to know, of their own knowledge, whether another pair of merchants had honored their mutual obligations. Unless social and economic institutions developed to fill in the knowledge gap or unless other means of enforcement were established, honest behavior in a community of self-interested traders could not be maintained.

DEFINE: Defines the knowledge gap problem: distant traders cannot directly verify others' conduct, necessitating institutional substitutes. Builds on: "Trade expansion demanded new institutions to deter merchant cheating."

We now consider in more detail a model of trade in which outsiders cannot readily observe what has transpired in a given bilateral trade. While "disputes" may arise in which one party accuses the other of cheating, none of the other players have a method of freely verifying the parties' claims. Even if the dispute itself can be observed by others, they cannot costlessly determine whether cheating by one has actually occurred or whether the other is opportunistically claiming that it did.

DEFINE: Defines the model's core premise: bilateral disputes are observable but not freely verifiable, so third parties cannot reliably enforce contracts. Builds on: "Long-distance trade requires institutions to enforce honest conduct among strangers."

In Townsend's matching model, there is an infinity of traders indexed by i j where i = I or 2 and j is an integer which may be positive or negative. At period t , trader l j is matched with trader 2 , j + t. In particular, no two traders ever meet twice and no trader's behavior can directly or indirectly influence the behavior of his future trading partners. In the absence of institutions, players possess no information about their current partner's past behavior.

With limited information about the past behavior of trading partners and no institution to compensate, there are no incentives for honest behavior. It is evident that incentives could be restored by introducing an institution that provides full information to each trader about how each other has behaved. Such an institution, however, would be costly to operate. Moreover, efficient trade does not require that every trader know the full history of the behavior of each other trader.

However, there is a second problem that the institutions must overcome: Traders may not find it in their individual interests to participate in punishing those who cheat. As one simple example, if trade is expected to be profitable, a trader will be reluctant to engage in a trade boycott. The institutions must be designed both to keep the traders adequately informed of their responsibilities and to motivate them to do their duties.

Builds on: "Shared honesty reputations can enforce honesty in dispersed trading communities."

In the model we develop below, this second problem has multiple aspects. First, traders must be motivated to execute sanctions against Cheaters when that is a personally costly activity. Second, traders must be motivated to keep well enough informed to know when sanctions are required, even though information gathering activities may be personally costly and difficult to monitor. In effect, one who keeps informed about who should be punished for past transgressions is supplying a public good; he deters the traders from cheating against others. Moreover, in our model, no other trader except his current partner will ever know if a trader does not check his partner’s past history, so the trader could avoid supplying the public good without facing any sanction from future traders. Third, traders who are cheated must be motivated to document the episode, even though providing documentation may be personally costly. After all, from the cheated trader’s perspective, what’s lost is lost, and there may be little point in “throwing good money after bad.” But if players who are cheated are unwilling to invest in informing their neighbors, then, just as surely as if the neighbors are unwilling to invest in being informed, the Cheater will profit from his action and Honest trade will suffer. These are the problems that the trading institution in our model must solve.

XREF: Connects to collective action / free-rider problems in public goods economics, where monitoring and sanctioning are underprovided. QUESTION: What institution design actually solves these three monitoring/sanctioning problems in the model? Builds on: "Institutions must solve the second-order problem of motivating punishment participation."

In the model we develop below, this second problem has multiple aspects. First, traders must be motivated to execute sanctions against Cheaters when that is a personally costly activity. Second, traders must be motivated to keep well enough informed to know when sanctions are required, even though information gathering activities may be personally costly and difficult to monitor. In effect, one who keeps informed about who should be punished for past transgressions is supplying a public good; he deters the traders from cheating against others. Moreover, in our model, no other trader except his current partner will ever know if a trader does not check his partner’s past history, so the trader could avoid supplying the public good without facing any sanction from future traders. Third, traders who are cheated must be motivated to document the episode, even though providing documentation may be personally costly.

XREF: Connects to collective action/public goods problems and the free-rider problem from Olson's logic of collective action.

The institution that we model as the resolution of these problems is based on the presence of a specialized actor - a “judge” or “law merchant” (LM) who serves both as a repository of information and as an adjudicator of disputes. The core version of our model is based on the following assumptions. After any exchange, each party can accuse the other of cheating and appeal to the LM. Any dispute appealed to the LM is perfectly and honestly adjudicated at cost C to the plaintiff. (We consider the case of a dishonest LM later.)

DEFINE: Introduces the law merchant (LM) as a specialized institutional actor combining information storage with dispute adjudication. Builds on: "Information gathering about cheaters is a public good for traders."

The institution that we model as the resolution of these problems is based on the presence of a specialized actor - a "judge" or "law merchant" (LM) who serves both as a repository of information and as an adjudicator of disputes. The core version of our model is based on the following assumptions. After any exchange, each party can accuse the other of cheating and appeal to the LM. Any dispute appealed to the LM is perfectly and honestly adjudicated at cost C to the plaintiff.

DEFINE: Introduces the law merchant (LM) model as a repository of information and dispute adjudicator in economic exchange.

Ar subsrage (b), if either player has failed to query the Law Merchant or if the query establishes that at least one player has an outstanding judgment, then both traders play Cheat (which we may interpret as a refusal by the honest trader to trade); otherwise, both play Honest. … Ar subsrage (c), if both parties queried at substage (a) and exactly one of the two players Cheated at substage (b), then the victim appeals to the LM; otherwise, no appeal is filed. … At subsrage (d), if a valid appeal was filed, the LM awards damages of J to the aggrieved party. … At subsrage (e), the defendant pays the judgment J if and only if he has no other outstanding judgments.

To establish that the LMSS is a symmetric sequential equilibrium strategy, we again appeal to the Optimality Principle of Dynamic Programming. If we show that there is no point at which a single change in the trader’s current action only (followed by later adherence to the LMSS) can raise the trader’s expected payoff at that point, then there is no point at which some more complicated deviation can be profitable, either.

DEFINE: Explains the Optimality Principle of Dynamic Programming as applied to checking sequential equilibrium: checking single-point deviations suffices to rule out more complex profitable deviations.

To establish that the LMSS is a symmetric sequential equilibrium strategy, we again appeal to the Optimality Principle of Dynamic Programming. If we show that there is no point at which a single change in the trader's current action only (followed by later adherence to the LMSS) can raise the trader's expected payoff at that point, then there is no point at which some more complicated deviation can be profitable, either.

DEFINE: Explains the Optimality Principle of Dynamic Programming — checking only single-action changes validates the whole strategy.

First, we check when it “pays to pay judgments,” that is, under what conditions a player will find it more profitable to pay any judgment rendered against him than to refuse to pay. (We ignore the sunk portion of the payoff which is unaffected by later behavior.) Paying the judgment J yields an additional payoff of - f ( J ) in the current period. In future periods, the player will spend Q to query the LM and earn a trading payoff of 1, for a total of 1 - Q. In terms of lifetime average payoff, paying the judgment leads to -(1 - S)f(J) + 6(1 - Q). If the trader refuses to pay the judgment, then his current period payoff is zero and, given the system, his payoff is also zero in every subsequent period. Therefore, it “pays to pay judgments” if and only if - (1 - 6)f(J) + 6( 1 - Q) 2 0, or equivalently,

DEFINE: Clarifies the compound condition for when a trader rationally opts to pay a judgment rather than face permanent exclusion from the market. Builds on: "A one-step deviation test verifies sequential equilibrium optimality."

Does it pay a party with an outstanding judgment to query? No, because the party’s expected payoff is -Q if he queries and 0 if he does not.

Thus, regardless of the circumstances wrought by past play, there is no situation in which a one-time deviation from the Law Merchant System Strategy that is profitable for a trader provided that conditions (7)-(9) hold.

Builds on: "Trading agents profit more buying judgments when query costs stay low"

Conditions (7)-( 10) show the relationship among the various parameters for the LM system to support the efficient cooperation. Each corresponds to one of the problems we described in introducing the model. Condition (7) requires that Cheating and then paying a judgment not be profitable; put simply, the judgment must be large enough to deter Cheating. Condition (8) requires that judgments exceed the cost of an appeal, that is, the judgment must also be large enough to encourage the injured party to appeal. Otherwise, information about Cheating will never reach the LM and Cheating will go unpunished. The two previous conditions require that the judgment be large enough, but condition (9) requires that it not be so large that the Cheater would refuse to pay, for then the injured party would not expect to collect, and so would find it unprofitable to appeal. Notice that the feasibility of satisfying all these conditions simultaneously depends on the technology of wealth transfer summarized byf. If the traders live at great distances from one another and if their principal asset holdings are illiquid (such as land and fixed capital, or reputation and family connections), then wealth transfers may be quite costly (f(J)/J may be large) and the fines required by the LM system then will not work.

QUESTION: The upper and lower bounds on judgment size (deter cheating, but not so large the cheater refuses to pay) suggest an interesting Goldilocks problem for institutional design. What happens when clear deterrence and collectability can't be satisfied simultaneously?

Notice that the feasibility of satisfying all these conditions simultaneously depends on the technology of wealth transfer summarized byf. If the traders live at great distances from one another and if their principal asset holdings are illiquid (such as land and fixed capital, or reputation and family connections), then wealth transfers may be quite costly (f(J)/J may be large) and the fines required by the LM system then will not work.

XREF: Connects to monetary theory and legal enforcement mechanisms where transaction costs constrain institutional design.

The LM institution encourages queries by making them a condition for appealing to the LM. and, as we have seen, querying deters Cheating. At equilibrium, traders who fail to query are constantly Cheated by their trading partners.

Builds on: "Querying with an outstanding judgment yields negative expected payoff."

If traders do not query the LM, then they will have insufficient information to administer punishments, so once again Cheating will go unpunished. The LM institution encourages queries by making them a condition for appealing to the LM. and, as we have seen, querying deters Cheating. At equilibrium, traders who fail to query are constantly Cheated by their trading partners.

For example, in the last line of Table 1, Cheating is seven times more profitable than playing Honest at each current round, the cost of querying the LM consumes one-third of the profits of Honest venturers, the cost of complaining is three times the profits of the venture, and half of any assets transferred in settlement of a judgment are lost. The judgment itself is six times what the Cheater could expect to earn from Honest trade with his next partner (nine times net of transaction costs). Nevertheless, if the inter-trade discount factor is at least 0.9, the LM system is in equilibrium and supports honest behavior, filing of valid complaints, and payment of judgments.

DEFINE: Explains the threshold discount factor of 0.9 that sustains equilibrium honest behavior in a repeated-trade game despite strong short-term cheating incentives. Builds on: "At equilibrium, traders who fail to query are constantly Cheated by their trading partners."

Theorem 3 shows that the LM system restores cooperation even when the players know little about their partners' histories. There are transaction costs necessary to maintain this system, however: That the average payoff per period is 1 - Q reflects the transaction cost of Q per period incurred by each trader to support the Law Merchant system.

Builds on: "LM-governed systems stay honest when future trades matter enough"

However, once these costs are low enough that the threat to file claims with the Law Merchant is credible, they act only as a deterrent: These costs are never actually incurred at equilibrium in our model of the Law Merchant system.

DEFINE: Explains the economic mechanism by which credible legal threats function as pure deterrents at equilibrium.

Notice that the cost, C, of making and investigating a claim and the costf(J) - J of making the transfer do not appear in the expression for the average payoff. These costs do appear in condition (6): The Law Merchant system is not viable if the cost of making and investigating a claim or the cost of paying a judgment is too high, for then the traders cannot reasonably expect that the others will make claims and pay judgments when they should. However, once these costs are low enough that the threat to file claims with the Law Merchant is credible, they act only as a deterrent: These costs are never actually incurred at equilibrium in our model of the Law Merchant system.

DEFINE: Clarifies how a legal system's viability condition differs from its equilibrium behavior — costs matter for credibility but vanish from equilibrium payoffs.

Theoretically, any institution that restores incentives for Honest trading by restoring the effectiveness of decentralized enforcement must inform a player when his partner has cheated in the past. If the temptation to Cheat is small and the value of continued trading is high, then this information need not be perfect, as in our model. So it may be possible to induce honest behavior using a less costly information system - one that costs only q < Q to inform a trader adequately well - and correspondingly to increase the traders' average payoffs from 1 - Q to 1 - 9.' However, using imperfect information to economize on information costs calls merely for a refinement of the Law Merchant system - not for something fundamentally different. It is not possible to provide correct incentives without incurring some information cost of this kind

DEFINE: Clarifies that imperfect information systems can substitute for costly perfect enforcement while still maintaining honest behavior, so long as temptation is small and trading value is high.

In operation, the Law Merchant system would appear to be a low cost way to disseminate information, for two reasons. First, the LM system centralizes the information system so that, for information about any partner, a player need only go to one place.

XREF: Relates to reputation systems and institutional economics literature on how centralized trust infrastructure lowers transaction costs.

What we can say confidently is that the kind of costs incurred by the LM system are inevitable if Honest trade is to be sustained in the face of self-interested behavior and that the system seems well designed to keep those costs as low as possible.

DEFINE: Defines the LM system as a mechanism that minimizes but cannot eliminate the costs inherent to sustaining honest trade among self-interested parties.

One need not look far in history (or, for that matter, in the modern world) to see that judges are not always so perfect. Within our model, there are many small amendments that could be made to insert opportunities for bribery and extortion. Although we do not provide a systematic treatment of these, we shall give a brief development of one of them to emphasize the simple idea that the Law Merchant business is itself valuable and that LMs may wish to maintain their reputation for honesty and diligence in order to keep the business active.

QUESTION: The brief development of one amendment for bribery/extortion isn't shown here — worth chasing to see how corruption enters the model.

The most obvious problem with this reputation based account is that it seems to presume that a trader who is extorted by the Law Merchant can somehow make his injury widely known to the community of traders. It might be that the Law Merchant is a more sedentary merchant than the long-distance traders whom he serves, so that idea is perhaps not so far-fetched. Nevertheless, we shall argue that even if, in the spirit of our earlier analysis, there is no way for the trader to inform others about his injury, it may still be an equilibrium for the LM to behave honestly, due to the “client” incentives in the long-term relationship between the LM and each individual trader. More precisely, we will show that there is an equilibrium of the system in which every trader expects that if he pays a bribe he will be subjected to repeated attempts at extortion in the future; this dissuades the trader from paying any bribe. Then, a Law Merchant who commits to his threat to damage the reputation of a trader succeeds only in losing business, so he does not profit from making the threat.

Builds on: "Judges in the model are subject to corruption opportunities."

Theorem 4 pertains to a model in which only one kind of dishonest behavior by the LM is possible. The problem of discouraging other kinds of dishonest behavior may require other strategies. From our preliminary analysis, it appears that the most difficult problem is to deter the LM from soliciting or accepting bribes from traders who have an unpaid judgment but wish to conceal that fact. By concealing the judgment, cheating. and refusing to pay the new judgment, the trader could “earn” Q - Q and a portion of that might be offered as a bribe to the LM. As we add richness to the possibilities for cheating, it is natural to expect that the necessary institutions and strategies must respond in a correspondingly rich way.

QUESTION: The suggestion that bribery is the hardest form of dishonesty to deter is underexplored here — worth investigating what makes bribery structurally resistant to institutional fixes. Builds on: "Institutions must solve the second-order problem of motivating punishment participation."

We began our analysis by studying an environment in which private information about behavior in exchanges is a potential impediment to trade. Under complete information, even if meetings among particular pairs of traders are infrequent, informal norms of behavior are theoretically sufficient to police deviations. But when information is costly, the equilibrium may potentially break down and informal means may not be sufficient to police deviations.

DEFINE: Clarifies the distinction between complete information and costly information in sustaining informal norms as a trade-policing mechanism. Builds on: "Reputation enforcement requires low costs relative to trade profitability."

The Law Merchant enforcement system that we have studied restores the equilibrium status of Honest behavior. It succeeds even though there is no state with police power and authority over a wide geographical realm to enforce contracts. Instead, the system works by making the reputation system of enforcement work better. The institutions we have studied provide people with the information they need to recognize those who have cheated, and it provides incentives for those who have been cheated to provide evidence of their injuries. Then, the reputation system itself provides the incentives for honest behavior and for payment by those who are found to have violated the code, and it encourages traders to boycott those who have flouted the system. Neither the reputation mechanism nor the institutions can be effective by themselves. They are complementary parts of a total system that works together to enforce honest behavior.

Builds on: "A law merchant resolves trade disputes as an information repository and adjudicator."

Our account of the Law Merchant system is, of course, incomplete. Once disputes came to be resolved in a centralized way, the merchants in Western Europe enhanced and refined their private legal code to serve the needs of the merchant trade - all prior to the rise of the nation-state. Without this code and the system for enforcement, trade among virtual strangers would have been much more cumbersome, or even impossible.

XREF: Connects to institutional economics and theories about private governance preceding state authority, such as Ellickson's work on order without law or the historical development of lex mercatoria.

However, our core contention that institutions sometimes arise to make reputation mechanisms more effective by communicating information seems almost beyond dispute. The Mishipora, described in the Hebrew Talmud, according to which those who failed to keep promises were punished by being publicly denounced; the use of the “hue and cry” to identify cheaters in medieval England; the famed “Scarlet Letter,” described in Hawthorne’s famous story; and the public stocks and pillories of 17th century New England, which were sometimes used to punish errant local merchants, are all examples of institutions and practices in which a principal aim is to convey information to the community about who has violated its norms.

Builds on: "Neither the reputation mechanism nor the institutions can be effective by themselves."

It is our contention that an enduring pattern of trade over a wide geographical area cannot be sustained if it is profitable for merchants to renege on promises or repudiate agreements.

Many of the key characteristics of our model correspond to practices found at the Champagne Fairs. While merchants at the Fairs were not required to query prior to any contract, the institutions of the Fair provided this information in another manner. As noted above, the Fairs closely controlled entry and exit. A merchant could not enter the Fair without being in good standing with those who controlled entry, and any merchant caught cheating at the Fair would be incarcerated and brought to justice under the rules of the Fair. So anyone a merchant met at the Fair could be presumed to have a “good reputation” in precisely the sense of our model. It did not indicate that all free merchants had never cheated in the past; but it did indicate that anyone who had been convicted of cheating had made good on the judgment against him. Moreover, because merchants might disappear rather than pay their judgments, judges at the Fairs had to balance the size of their judgment so that the value of being able to attend future Fairs exceeded the award.

XREF: Connects to the repeated-dealing/reputation model of informal enforcement vs. formal legal institutions.

Other, smaller fairs that tried to compete with the Champagne Fairs on an equal footing would have to contend with merchants who participated only long enough to make a profitable cheating transaction and then return to the Champagne Fairs where their participation rights were intact.

Even if the Law Merchant and related systems were effective underpinnings for local trade, how was information about a trader's dishonesty in one location transmitted to another? The model in this paper is too simple to handle this problem, but we hope to extend our approach to the institutions that developed during the middle ages to protect against the added problems raised by spatial separation. This includes the merchant gilds in northern Europe, the consulates of the Italian city states, and the organization of alien merchants into colonies (like the Steelyard in medieval London) with local privileges and duties. These institutions can also be understood from the perspective developed in this paper - they are designed to reinforce reputation mechanisms that alone are insufficient to support trade.

XREF: Connects to institutional economics and the notion that organizations arise to support trust where simple reputation fails. Builds on: "Entry control at fairs enforced merchant reputations"

The Law Merchant system of judges and reputations was eventually replaced by a system of state enforcement, typically in the late middle ages or the early modern era in Western Europe. Enforcement of the private codes by the state added a new dimension to enforcement, especially in later periods when nationstates exercised extensive geographic control. Rather than depend for punishment upon the decentralized behavior of merchants, state enforcement could seize the property of individuals who resisted paying judgments, or put them into jail.

XREF: Relates to North's work on institutions and the shift from private-order to public-order enforcement. Builds on: "Medieval merchant institutions scaffolded reputation mechanisms for distant trade."

Thus our approach suggests that the importance of the role of the state enforcement of contracts was not that it provided a means of enforcing contracts where one previously did not exist. Rather, it was to reduce the transaction costs of policing exchange.

Builds on: "State enforcement replaced merchant-run reputation systems in early modern Europe."

In complete information settings, institutions are frequently unnecessary because decentralized enforcement is sufficient to police deviations. However, this conclusion fails in environments where information is incomplete or costly. In the context of our model, the Adjusted Tit-for-Tat strategy requires that a trader know his current partner's previous history. When such information is difficult or costly to obtain, decentralized enforcement mechanisms break down. Institutions like those of the Law Merchant system resolve the fundamental problems of restoring the information that underpins an effective reputation system while both economizing on information and overcoming a whole array of incentive problems that obstruct the gathering and dissemination of that information.

DEFINE: Positions a theory of why institutions like the Law Merchant system exist — they restore reputation signaling where information is expensive. Builds on: "State contract enforcement mainly lowers transaction costs of policing."