12.4 The rationale and characteristics of sanction systems

Back to contents - Previous file - Next file

The rationale of local sanction systems

In the first section of this chapter, the existence of an effective sanction system has been listed as a major determinant of success in CPR regulation schemes. In the literature, there is unanimous agreement on this crucial point. Yet the question is to be asked as to why sanctions are required to enforce CPR regulation. Indeed, one important result of the theory of non-cooperative repeated PD games is that co operation may evolve without the support of external sanctions if actors follow strategies of brave reciprocity understood as strategies of conditional co-operation. This requires, for example, that the actors' horizon is infinite or finite but indefinite, an assumption that is a good approximation to reality in many cases of local-level resource use. The basic reason why co-operation is then possible lies in the simple fact that strategies of conditional co-operation are obviously strategies of punishment and reparation (see above, Chapter 4).

Such a mechanism of purely decentralized punishment embodied in the actors' strategies is not without important shortcomings, however. In particular, even assuming that actors are all willing to follow strategies of brave reciprocity, the fact that the set of possible punishments is highly restricted creates serious problems. This is evident in N-player repeated PD games. Assume thus that there are N users of a given resource. Rule-breaking is common knowledge and, in a given round or period, one of the N actors violates management rules in front of the (N - 1) other actors who have abided by them. Why should the latter choose to retaliate by also breaking the rules during the next period, when such retaliation clearly jeopardizes the whole management scheme with disastrous long-term consequences for everybody? Rather than punishing oneself in order to effectively punish the rule-breaker, is it not more rational to inflict selective punishment on him?

It can be argued that, since no punishment occurs at equilibrium, there is no need to worry about the destructive nature of punishments: on the contrary, the severity of punishments is precisely aimed at making the threat sufficiently effective that nobody is incited to defect. Yet, this is to forget that players can make 'mistakes', that is, they may happen to deviate from their rational strategy of co-operation. And if these 'mistakes' are too frequent, there is a serious risk of a 'Nash equilibrium reversion' (Abreu, 1986), implying that players revert to single-period Nash-equilibrium behaviour. This is a serious limitation inasmuch as rural people arc typically confronted with natural and other hazards which may easily drive them, from time to time, inadvertently to violate access or conservation rules. To allow for 'irrational' moves or stressful conditions, punishment mechanisms must be flexible and tolerant enough. Such flexibility is hard to achieve with simple decentralized punishment strategies that rest on a restricted set of actions: thus, in the repeated Prisoner's Dilemma, punishment can only consist of a sequence of two possible actions co-operate or defect. Subtle, gradual, and relatively costless responses, such as enquiries, warnings, mild threats, etc., are thereby ruled out.

It bears emphasis that inflicting selective punishment on deviant players is a sensible strategy not only on rational but also on emotional grounds. As a matter of fact, people who have been cheated may strongly resent it and wish to relieve their emotional stress by punishing the cheater 'just to teach him a lesson', or by denouncing the fraud so as to bring sanction from the competent agency. Of course, the emotional drive to punish deviant behaviour will be all the more powerful if norms of fairness exist to back up the prevailing code of conduct. In gametheoretic terms, the positive role of emotions can be conceptualized as follows. Because of the vengeful reaction of co-operative players, the payoff of a deviant player is much less than the material gain from free riding on others' efforts. This may be so much so that the total payoff from defecting while others co-operate is actually less than the payoff obtained by co-operating with the other players. Under these conditions, emotions have the effect of transforming a PD game into a virtuous game in which universal co-operation is the only Nash equilibrium. Notice carefully that, when emotions perform the above function, the difficult 'public good' problem of sanctioning is automatically solved: the cheated agents do not really ponder over the private cost of sanctioning and it is actually refraining from punishing or denouncing that appears to them as a painful experience.

Now, it is important to recognize that equilibrium strategies may possibly exist that use selective self-enforcing punishments in the form of payoff transfers among agents. Note incidentally that these transfers need not correspond to material exchanges but may possibly consist of symbolic sanctions such as apologies. It is evident that this type of equilibrium strategy is inconceivable without some degree of collective regulation (about the amount of the transfers, the circumstances in which it must be applied, etc.), yet an essential feature is that it does not require a central enforcing agency. The self-enforcing character of punishment strategies results from the fact that the stream of future incomes the deviant player can earn by leaving the group and earning the best alternative income is lower than the stream he can obtain by remaining in the group and cooperating with the other members even after he has paid the fine. Clearly, the maximum punishment that can be imposed on individuals who deviate from the cooperative path yields the expected autarkic payoff (Fafchamps, 1992: 162-3).

Things may not be so easy, however. More precisely, the punishment path may not be 'renegotiation-proof' if the threat to impose sanctions on the deviant player is not credible because the group would be less well-off after the departure of the deviant player (see above, Chapter 8). In the words of Fudenberg and Tirole, 'equilibria that enforce "good" outcomes by the threat that deviations will trigger a "punishment equilibrium" may be suspect, as a player might deviate and then propose abandoning the punishment equilibrium for another equilibrium in which all players are better off' (Fudenberg and Tirole, 1993: 175). To put it in another way, the group would be ready to renegotiate the original agreement because it would lose from the departure of the deviant player and, since the latter knows it, he may use this bargaining strength to force the group to forgive him for his defection.

It must nevertheless be reckoned that, regarding the types of situation considered in this book, the above problem of punishments that do not satisfy the 'renegotiation-proofness 'test is probably limited. Indeed, to the extent that co-operation among the players is needed because there is dangerous pressure on the CPR, one may expect them to feel relieved rather than worried by the departure of deviant users. Even in this case, however, there are two circumstances in which this problem may remain significant. First, as pinpointed in the conclusion to Chapter 4, the use of the CPR may not be an isolated game, that is, there may be interlinkages between what is being played here and what is being played in other sectors of the CPR users' social life. Second, collective action in the group may require the co-ordinated efforts of all the members, say, for maintenance or surveillance activities (see Chapter 5).

Leaving aside the above difficulty arising from 'renegotiation-proofness', decentralized mechanisms resting on self-enforcing punishments are far from being satisfactory. There are two main reasons for this. First, as we have stressed in Chapters 4 and 5, in most non-cooperative games there exist a plethora of possible equilibria, some of which are Paretodominated. A society may therefore be justified in attempting to design a system of collective sanctions which transforms the game in such a way that the Pareto-optimal outcome is the unique equilibrium. Second, under conditions of imperfect information, decentralized punishment mechanisms provide the agents with considerable opportunities for all sorts of information manipulation that can easily lead to partial and arbitrary outcomes. For instance, an individual may be tempted to take revenge on another participant by wrongly accusing him of trespassing the rules and claiming the corresponding transfer.

This shortcoming is to be taken very seriously in the light of an aforementioned disadvantage of small groups, namely that highly personalized relationships create a fertile ground for strong negative feelings, such as envy and rivalry, which can lead to group implosion if not properly checked. In such a context, the temptation to fraudulently manipulate decentralized monitoring and punishment mechanisms is likely to be strong. The implication of this diagnosis is that, even in relatively small groups such as village communities, collective regulation through a central authority may be desirable. Obviously, such a central authority must be considered fair and legitimate by the people concerned lest it should be very costly to enforce.

Reckoning with the need for such a centralized system of sanctioning does not mean, however, that all decentralized mechanisms and procedures have to be done away with. As a matter of fact, to be effective, a collective sanction system must consist not only of formal procedures but also of informal conventions and rules that are commonly understood and accepted by the members. This is because situations on the ground are typically fraught with so many uncertainties and contingencies that no formal rule or contract could ever specify all the relevant conditions. Only general principles such as those embodied in moral norms and codes of honour allow the unpredictable to be approached with appropriate flexibility. Formal sanction systems cannot alone be expected to cover satisfactorily all the possible contingencies and room must therefore be left for complementary mechanisms grounded in the ethics of the society. Clearly, the effectiveness of a collective sanction system depends on the balance that is struck between these two inevitable components, and the more uncertain and complex the circumstances surrounding the members' decisions the larger should be the role of informal rules.

Organized versus decentralized monitoring

In conditions of imperfect observability of individual actions, monitoring activities are an essential component of a village control system aimed at preventing freeriding. Indeed, when monitoring is lax or observation errors are frequent, sanctions may not be applied in the way required to deter rule violation. This raises the question as to whether monitoring can be left completely decentralized or whether it is better organized at the group level. The most obvious problem with a decentralized system of monitoring is that it may not be sufficiently reliable. Monitoring is a public good and there is therefore no guarantee that someone who detects rule-violation will incur the cost of reporting it to the relevant authority. This is all the more so if he may fear incurring the wrath of the presumed violator while he cannot be sure that his denunciation will lead to actual punishment. On the other hand, a fraud detector may show partiality in his reporting, protecting his friends or allies and harassing his rivals or enemies. These considerations go a long way towards explaining why in field settings monitoring activities are often organized in a collective way.

Ease of monitoring depends on the characteristics of the natural resource considered and on the size of the user group. Thus, regarding the first factor, monitoring is particularly easy with respect to certain resources which are highly visible and/or well localized, and when users reside in close proximity to these resources. In the case of surface irrigation systems, for example, poaching of water is relatively easily observable as each irrigator keeps a constant eye on his neighbour's actions. When water rotation turns are organized, monitoring is thus costless: 'the presence of the first irrigator deters the second from an early start, the presence of the second irrigator deters the first from a late ending' (Ostrom, 1990: 95). And likewise in fishing-site rotation systems. On the other hand, the use of some resources such as open sea fisheries, forestries, and groundwater, is considerably more difficult to monitor even when user groups are rather small.

Regarding the second factor, since monitoring is a public good, it is subject to the "incentive dilution' problem and, consequently, the need to organize monitoring activities collectively should increase with the size of the user group. Thus, for a given resource such as irrigation water, one expects a significant relationship between the existence of formal monitoring mechanisms as measured by the presence of guards accountable to the water users, on the one hand, and the size of the resource system, on the other hand. This prediction seems to be confirmed by a systematic study conducted in the Philippines. It was indeed found that 80 per cent of small irrigation systems (less than 50 h.) operate without formal guards, while the proportion declines to 55 per cent for medium-scale systems (from 50 to 100 h.) and to 30 per cent for large systems (more than 100 h.) Furthermore, it is interesting to note that farmers tend to patrol their own canals even when some of them are also patrolled by appointed guards (de Los Reyes, 1980, cited from Weissing and Ostrom, 1991: 2389, 246-7).

That the presence of monitors is an important factor affecting the level of rule conformance comes out of the survey by Tang (1992, 1994) of a large sample of irrigation systems (mostly located in Asia). This is especially true of farmer-owned systems since routine rule conformance is reported in thirteen out of the fourteen farmer-owned systems with guards in the sample, but in only five out of eleven farmer-owned systems without guards. The same relationship holds true in government-owned systems although it is less significant than for farmer-owned systems: four out of the eleven government-owned systems with guards are characterized by a lack of routine rule conformance. Interestingly, this follows from the fact that local guards in farmer owned systems 'are more likely to impose sanctions on rule violations than guards hired by the government-owned irrigation systems' (Tang, 1994: 240-2), a difference which is probably to be ascribed to the relatively poor incentives provided to guards in the latter as compared to the former systems (see above, Chapter 11, sect. 1). Unfortunately, while attempting to assess the effectiveness of formal monitoring systems, Tang does not control for the size of the irrigation systems.

Do we have to infer from the above considerations about the importance of collective organization of monitoring and sanctioning activities that results from non-cooperative game theory are of little relevance for assessing the potential of village communities to manage their natural resources? The answer is negative. As we have just seen above, there is a positive relationship between the size of a user group and the probability that it regulates CPRs through a formal monitoring and sanctioning system. In so far as in many such formal systems are nested (very) small and typically homogeneous groups comprising only a few families that operate in a fully decentral ized way, the theory remains valid: by demonstrating that co-operation is a possible outcome within the framework of repeated interactions in small groups, it contributes to establish the feasibility of management systems resting on multiple layers of nested user groups.

Examples of such 'federated' systems include irrigation, fishery, and pastoral management. Thus, most of the large, self-organized irrigation systems that have successfully survived for long periods of time are organized in three or four different tiers. This allows for clear and unambiguous rules for allocating water among relatively separable units, an absolute prerequisite for the effectiveness of any such subdivision of the system into management micro-units (Weissing and Ostrom, 1991: 259 n. 31). Likewise, in pastoral management, it has been noted that, if water points can be more effectively managed by small cohesive groups, management of viable units of rangeland requires larger groups formed by combining several such small groups (Shanmugaratnam et al., 1992: 6). As for fisheries, it will be seen in Chapter 13 that the Japanese system of coastal fishery management one of the most impressive achievements in CPR management to date—devolves substantial responsibilities to fishermen communities constituted as fishing co-operatives (known as fishermen associations) which are themselves often organized into smaller units. These small units, whether they arc fishing squads or residential groups, appear to be the main agencies for the allocation of fishing rights in coastal waters.

In the following, we consider a number of case-studies which provides us with rather detailed information about monitoring and sanctioning activities in village communities. As is evident from these examples, there can be significant variations in the degree of collective organization regarding these two activities. In the first case-study, observability of individual actions is perfect and the entire sanctioning system rests upon decentralized, informal mechanisms of social opprobrium and gossip: there is thus no organizational structure to monitor people's behaviour and to impose punishment upon deviant actors (monitoring and punishment are selfenforcing). In the other four case-studies, observability is imperfect and monitoring and sanctioning activities are collectively organized.

Sanctioning in Teelin Bay, Ireland

In Teelin Bay (Ireland), the estuarine salmon fishermen use a rotation system of access to the most valuable fishing-spots. Only exceptional and slight violations of the underlying code of conduct are observed. Abidance by the rules of the salmon rotation system within the restricted space of the local community 'is perceived as merely the "natural" expression of local behavior' and rests 'only on the egalitarian ethos of communal reciprocity, which is, in turn, understood as natural' (Taylor, 1987: 299, 302). Few would or do try to violate these rules and, when a violation takes place, communal punishment does not have to be organized because the opprobrium of informal social control that follows, under the form of 'an enormous amount of pointed whispering gossip in pub and household', is apparently sufficient to deter deviants from cheating again. The egalitarian ethos of the community is thus pre served since no one 'is perceived as having the authority to directly reprimand or punish the offender' (ibid. 299, 303).

What needs to be emphasized is that non-compliance with the rotation rule is effectively sanctioned through informal social pressure, which has the effect of transforming the original repeated PD game into an assurance game, or into a PD game with self-enforcing punishments in the form of payoff transfers. There remains, however, the second-order public good problem of who will have the incentive to punish by exerting such social pressure. Such punishment decisions must be analysed within the framework of another repeated prisoner's dilemma, since no collective sanctioning system exists to drive the agents to apply first-order punishment. The cooperative outcome corresponds here to a situation where everybody exerts social pressure on any member found to break the rotation rules.

Monitoring and sanctioning in rural Japan

In McKean's already-quoted survey of village CPR-related practices in rural Japan, we are offered exceptionally precise information about monitoring and sanctioning mechanisms and it is therefore useful to describe them in detail by closely following McKean's illuminating account.

Apparently, Japanese villages widely resorted to selective inducements and punishments in order to ensure due respect for the written codes which most of them had to govern their CPRs. Regarding inducements, we are thus told that 'there was an intrinsic pride in the importance of doing one's duty by the commons and in preserving the village's well-being; a young man brought credit to his family and future by doing the job properly' (McKean, 1986: 564). Regarding punishments, the evidence is that 'violating rules that protected the commons was viewed as one of the most terrible offences a villager could commit against his peers, and the penalties were very serious' (ibid. 561). In order to detect rule infractions, purposeful monitoring was practiced in the form of groups of detectives destined to constantly patrol the commons: 'the detectives would patrol the commons on horseback every day looking for intruders, in effect enforcing exclusionary rules'. Their job was considered 'one of the most prestigious and responsible available to a young man' (ibid. 560-1). According to villages, these positions changed hands more or less frequently and, in some of them, all eligible males had to take a turn, so that no family was without its full labour supply for very long. In the poorest villages, specialized but rotating detectives did not exist (probably because people were too poor to spare the required labour), yet anyone could report violations (ibid. 561).

As we have mentioned in the first section of this chapter, sanctions obeyed a number of principles which can be summarized under three main headings. First, the specific punishments stipulated in the village written codes for specific violations followed a built-in scheme of escalating penalties for non-co-operation. This escalating scale of penalties began with confiscation of the contraband taken from the CPRs, as well as the equipment and horse, which a violator could retrieve only after having paid a fine and apologized to the detectives who apprehended him (interestingly, the sake given as payment of a fine was referred to by special terms indicating the humility of the rule-breaker). The contraband harvest was of course retained by the village. If the offence was relatively great or the apology unsatisfactory, the head of the culprit's household or his kumi (an intermediate grouping comprising several households and constituting a very important unit of accounting and distribution of responsibilities and benefits connected with the CPRs) or temple priest was called to make the apology on his behalf and offer a larger fine in his stead (McKean, 1986: 550, 561-2).

When necessary, Japanese villages did not hesitate to threaten to use their more powerful sanctions: 'ostracism in increasingly severe stages, followed by banishment'. Ostracism—which implies that the village community 'cuts off all contact with the offender except for assistance at funerals and fire-fighting'—was thus resorted to in gradual stages, 'starting with social contact and only escalating to economic relations if the offender did not express remorse and modify his behavior'. Moreover, 'to ensure that the villagers would remember to shun contact with someone subjected to ostracism, that person might be expected to wear distinctive clothing (a flashy red belt or pair of unmatched socks)' (McKean, 1986: 562). McKean reckons that this was 'a horrible punishment for the Japanese villager, not only because it cut him off from a highly group-oriented society and made daily life unpleasant, but because it actually deprived the villager of tangible services essential to daily living', including access to the CPRs. It is therefore not surprising that ostracism was an effective deterrent to repeated and serious rule infractions, especially for ordinary villagers who would never jeopardize the survival of their household and their family's reputation for many generations when a simple apology could extinguish the controversy (ibid. 562).

It is worth stressing the obvious fact that the effectiveness of social ostracism as a weapon against rule-breaking hinges very much on the degree of physical immobility of CPR users. In present-day circumstances where mobility is much higher than before, this effectiveness has no doubt been significantly reduced (see above, Chapter 11, sect. 2).

A second basic feature of sanction mechanisms in rural Japan is their flexibility or tolerance. According to McKean, there were two kinds of circumstance in which inspectors or village leadership were inclined to manifest a good deal of tolerance towards detected rule-breakers: (a) when violation followed an obvious error on the part of CPR managers or village leaders which could have damaging consequences for the rule-breakers (for instance, the village chief set the day of mountain-opening too late, thereby preventing some people from cutting in time the poles needed to support garden vegetables raised on private plots); (6) when the rule violators stole from the CPRs only out of desperation. In the former case, tolerance arose from awareness of the fact that rule-breaking was an act of understandable protest against a mistaken decision. In the latter case, 'inspectors or other witnesses who saw violations maintained silence out of sympathy for the violators' desperation and out of confidence that the problem was temporary and could not really hurt the commons' (McKean, 1986: 566). When a collective disaster occurred, as in the village of Yamanaka during the Depression of the 1930s, tolerance vis--vis widespread rule-breaking was generalized for the same reason. There was no fear of an irreversible collapse of the whole CPR conservation system: indeed, 'instead of regarding the general breakdown of the rules as an opportunity to become full-time free riders and cast caution to the winds, the violators themselves tried to exercise self-discipline out of deference to the preservation of the commons' (ibid.).

The third and last feature of Japanese rural sanction systems to which we wish to draw attention concerns the reliability of CPR monitors or inspectors. This issue involves the question of the monitors' incentive to perform their job adequately and the related question as to who will watch the watchers. Monitors had two kinds of incentives both of which have already been mentioned: on the one hand, they gained social prestige and, on the other hand, they derived material advantages since they were entitled to keep the small fines paid by detected rulebreakers (in the form of cash and sake). Regarding the second issue, three observations deserve to be made. First, as we have pointed out earlier, the duty of watching usually rotated through the body of male members of the community so that everyone got his turn to exercise the monitoring power and to obtain the advantages associated with it. Second, the monitors or detectives patrolled in teams and were collectively responsible for their actions. Third, whenever exceptional circumstances arose which significantly increased the temptation to break the rules on the part of both users and monitors, mutual monitoring by all users replaced specialized monitoring by appointed guards. For example, we are told the following story. 'When the village of Shiwa suffered a drought, farmers at the downstream end of the irrigation system, including the water guards on patrol, were sorely tempted to alter the dikes so as to receive more than their allocated share of water. During such times, the collective response was for all adult males to patrol the dikes all night long in mutual surveillance' (McKean, 1986: 565). The above arrangements were apparently effective. The most important abuse of power by the (young) detectives consisted of ignoring repeated offences in exchange for sexual favours from attractive girls whom their fathers purposefully sent to collect grass (ibid. 564).

Let us now turn to three other case-studies. The first of these studies confirms the effectiveness of social opprobrium as a sanctioning device in rural communities and, moreover, it explicitly points to an intergroup punishment mechanism that makes social excommunication extremely costly. The second study provides valuable details about monitors' incentives and it lays stress on the importance of accountability for the effectiveness of any sanction system, especially in times of crisis. Finally, the third study is especially interesting because it supplies information not only about monitoring and sanctioning mechanisms but also about the incidence of rule-breaking, the effectiveness of monitoring, and the state of the CPR.

Monitoring and sanctioning in the Solomon Islands

In Marovo (Solomon Islands) as in traditional Japanese villages, banishment is also the ultimate and most serious step along the punishment scale for free riders. According to a recent study:

When people not so entitled are observed to enter certain fishing grounds, use certain restricted fishing methods or harvest protected species, a response usually follows from the guardians of the area or resource in question. In addition to receiving communications of a more or less heavily rhetorical nature, habitual trespassers are often subjected to wider social pressures aimed at creating personal shame. This is a serious measure in Marovo society, where there is generally a strong wish to avoid open conflict and public criticism. In some instances, chiefs demand compensation from offenders, in the form of cash or the resource 'stolen'. In certain serious cases, most notably dynamite-fishing, offenders may even be reported to the local police, resulting in court cases and heavy fines. The most serious measure applied towards really notorious poachers and trespassers is a form of social excommunication through which the offended group effectively revokes the poacher's status as a kinsman. This punishment may be meted out on a joint basis by several groups and may have grave consequences for Marovo individuals, who typically rely on a multitude of primary, secondary and tertiary use rights in land and sea resources through wide-ranging kinship privileges. (Hviding and Baines, 1994: 24)

In the above example, excommunication is a particularly tough measure because the sanction is applied multilaterally, leaving practically no chance of survival to the unrepentant free rider. For another thing, if the management responsibility (the so-called nginira, or 'power to speak about' fishing resources) is vested in the elders who are each kin group's spokesmen and guardians of the territory, their decisions are often directly enforced by men in their twenties and thirties. As a matter of fact, 'since these younger men are more often out on the fishing grounds they are better informed about shifts in resource availability and environmental change, and most likely to detect any trespassers' (Hviding and Baines, 1994: 26). Subsequently, elders take up trespassing cases through traditional procedures of consultation.

Monitoring and sanctioning in Andhra Pradesh

In the village of Kottapalle (Andhra Pradesh) studied in detail by Wade, monitors like common irrigators or field guards are appointed by and responsible to the village council. The common irrigators' job is to distribute water between the paddy-fields, apply it to each field, and help bring more water down the distributary to the village. The role of field guards is to enforce village regulations intended for minimizing losses to standing crops in unfenced fields, and to guard against crop thefts. They are hired for a period extending from a few months (common irrigators) to most of the year (field guards) and the procedure of appointment is repeated every year; this enables the council to rotate the monitoring groups (if demand for this job exceeds the supply) and to get rid of agents whose work was deemed unsatisfactory in the past. A reputation for being hard-working and conscientious is important in their selection while faction allegiance is probably not. Monitors are paid by the farmers or herders at a rate set by the village council. In addition, they work in small groups, partly because violating the procedures, say for equitable distribution of water, is more difficult with more than one monitor present. Field guards are empowered to take straying animals to the village pound, from which the owner has to pay a standard fine set in cash by the council to get them back. However, where many animals are involved, the case is brought before the council, which fixes the fine (Wade, 1988a: 63-8, 75-86). The fines are of non-trivial amounts—a day's field wages for an animal caught at night, and much more for water infringements (ibid. 193).

Incentive considerations clearly underlie the system of monitors' payment. For example, 'the field guards' salaries are set at less than the daily wage of an agricultural labourer so as to give them a strong incentive to collect the fines, for they keep all of the small fines and a fixed percentage of the larger ones' (Wade, 1988a: 193). As for common irrigators, they are paid from acreage levies. Free riding on their services, Wade observes, is held in check by knowledge that termination of these, when water is scarce, would produce an immediate crisis for everyone with land in a tail-end location within the village (ibid. 194). In other words, abidance by the rules is more forthcoming here than it tends to be in the case of CPRs whose mismanagement produces only long-term consequences. For another thing, people know that nonpayment of the common irrigators (which is done after the first harvest and in kind) can be penalized the following year; it is indeed possible for common irrigators to interrupt a nonpayer's water supply until his crops suffer yield-reducing stress (ibid.).

Another interesting feature of sanction mechanisms highlighted by Wade is that, as in the Japanese villages studied by McKean, the effects of fines are reinforced by the threat of reputation loss:

Whether because the desire for social acceptance by a group is a fundamental principle of social behaviour or because reputation loss has material consequences for an individual in terms of contracts foregone, reputation in a small agricultural community is not lightly exposed to attack. We have seen the council deliberately seeking to activate reputation sanctions, as in its strategy of bringing the maximum number of 'influentials' to council meetings at times of crises to signify by their presence and non disagreement their acceptance of the decision.... Council meetings are in a public place, so that anyone with the confidence to do so can monitor the proceedings from the sidelines and even take part in the discussions. . . council meetings on contentious issues often turn into de facto general meetings.... In these various ways ... accountability is maintained. (Wade, 1988a: 193, 195)

Monitoring and sanctioning in Uttar Pradesh

A recent aforementioned study by Agrawal (1994) provides us with interesting information about monitoring and sanctioning mechanisms in a sample of community-managed forests in Almora district which belongs to the middle Himalayan ranges north of Uttar Pradesh, known as Uttarakhand (India). The sample comprises six villages. In three of them, the village forest is in excellent, or excellent-to-good condition, while in the other three the resource condition is poor to fair. Note that all six villages have elected village forest councils so as to be able to take advantage of the Van Panchayat Act (1931) which provides for the creation of communitymanaged forests from the forests controlled by the Revenue Department (see above, Chapter 11, sect. 1).

Bear in mind that most of these villages have adopted rather precise rules prescribing how the forest is to be used (see above, Chapter 10, sect. 2). Yet there are considerable variations in the extent to which villages resort to monitoring and sanctioning to enforce these rules. In some villages, such activities are well organized. Guards are assigned different compartments of the forest and their rewards are made dependent on their performance. The panchayat may pay the guards a lower salary when many rule violations occur and it may even dismiss them and refuse to pay their salary if the levels of these violations are too high. (To detect lax behaviour on the part of the guards is relatively easy in so far as freshly cut grass or tree branches provide evidence that the guard concerned has not been guarding properly the section of the forest for which he is responsible) (Agrawal, 1994: 275).

The panchayats have recourse to various sanction mechanisms and, in this context, the effectiveness of apologies is not to be underestimated:

They ask offenders to render written or public apologies, confiscate cutting implements such as scythes, strip villagers of use rights, impose fines, report villagers to government officials, and sometimes, seek redress in courts. The sanctions they impose depend on a number of factors: the severity and nature of the offense, the economic status of the offender, whether the person is known to be a troublemaker, the attitude that the rule breaker displays towards the panchayat and its authority, and so forth.... the panchayats often excuse even repeat violators from paying fines imposed on them, if the offender is willing to render a written or public apology. Such an apology reinforces the authority of the panchayat to manage the forest and to punish other individuals who commit rule infractions. (Agrawal, 1994: 278)

In the villages where such sanction mechanisms exist and are impartially applied, they tend to be effective in the sense that 'many of the users pay their fines, appear before the panchayat when summoned, render apologies, and promise not to break rules in the future' (Agrawal, 1994: 280). Agrawal explains that: 'Even if the panchayat does not have formal legal powers to extract fines from rule breakers, in courts of law its word carries greater weight than that of an ordinary villager. Since it has been created by a statute of law, its mere existence has the support of law' (ibid.). To this explanation must be added the presence of informal retaliation mechanisms that threaten any unrepentant rule infractor with a variety of more or less subtle ostracization responses.

One of the central conclusions reached by Agrawal is that, as expected, 'a significant relationship exists between enforcement and resource condition': villages that commit substantial resources to monitor and sanction rule-breakers are more likely to prevent degradation of their local CPRs (Agrawal, 1994: 281-2). However, as the same study shows, a comparatively high incidence of rule-breaking may also occur irrespective of the resources devoted to monitoring. Thus, the characteristics of the CPR may make a given amount of monitoring comparatively ineffective, such as when the panchayat forest compartments are highly dispersed. Or, caste prejudice may lead to biased reporting of rule-breaking, as a result of which dominant castes 'get a licence to break rules' end 'the resentment against the Brahmins would goad Harijans to break rules as often as possible' (ibid. 276-7).

Another important finding of Agrawal's study is that rule violations occur routinely even in villages with relatively high levels of monitoring and sanctioning:

Villagers illegally entered the panchayat forests, cut grass and leaf fodder from trees, grazed their animals, collected twigs and branches, and in some instances even felled trees. I heir activities occurred in violation of the rules, and in spite of the presence of guards who could discover and report them to the panchayat, which would then try to force them to pay fines. The records, while documenting high levels of abuse, underestimate the extent of illegal grazing and cutting. (Agrawal, 1994: 274)

It has been proved in Chapter 8 (sect. 1) that, if monitoring is imperfect, it is possible that rulebreaking always occurs at equilibrium (there is a Nash equilibrium in mixed strategy in which the resource user violates the rule with positive probability and the guard monitors also with positive probability). To recall, this situation happens when the payoffs are such that a resource user has no incentive to violate the rule if his behaviour is monitored (since, otherwise, rulebreaking would be a dominant strategy and would therefore occur systematically rather than probabilistically). That this equilibrium in mixed strategy provides a good description of the situation observed in the villages surveyed by Agrawal can be inferred from the fact that he mentions not only the imperfect monitoring ability of the guards (due to dispersion of community forests) but also the possibility for them to evade their duties: 'The guards are often absent from the forest and even when at their posts cannot monitor all compartments of the panchayat forest simultaneously' (Agrawal, 1994: 274). While opting for occasional rule violations, forest users therefore know that guards have a monitoring tendency smaller than one and that, when they actually monitor, the probability of detection of rule-breaking is also smaller than unity.

Of course, violations of the prevailing rules are less frequent or are of relatively minor importance in villages where more resources have been devoted to monitoring (for given CPR characteristics) and, as a consequence, community forests are less subject to degradation in these villages.