8. The Regulated Common Property
Back to contents - Previous file - Next file
8.1
The efficiency of regulated common property
8.2
The distributive effects of regulating common property
8.3
Common property resources in the context of pervasive
factor-market imperfections
8.4 Conclusion
In its discussion of the solution to the 'problems of the commons', the property rights school does not only overemphasize the efficiency gains which can be expected from privatization, but it also overlooks the potentialities displayed by common property arrangements. Indeed, in the arguments proposed, it is as though the group, or the community, were able to control access to the resource but not to control its use by the members of the community. In other words, it is assumed that common ownership systematically fails to satisfy the 'authority axiom' following which the well-defined group acts with a unified purpose. 'The inability of groups to act in a socially responsible manner . . . is usually blamed on the impossibility of groups to co-ordinate and co-operate on a pattern of resource use .... Thus ... the authority axiom is violated because members of the group are always presumed to have the incentive to cheat on any co-operative agreement' (Larson and Bromley, 1990: 239).
The case for or against the property rights school's view can be decided only on empirical grounds. Part II—more particularly Chapters 10 and 12—will help us better assess the claims of this school. As we shall then see, the question raised is much too complex to receive a yes/no type of answer. A convincing and credible answer must allow for subtle distinctions and carefully worded nuances so as to avoid the two opposed pitfalls of cynicism and romanticism, or of undue pessimism and excessive optimism. At this stage, what needs to be borne in mind—and this is sufficient to vindicate the kind of approach followed in the present chapter—is the following: there exist empirically significant circumstances in which rural groups or communities with an exclusive right of access to a particular resource may succeed in designing and enforcing rules or arrangements that allow them to control the use of the resource in a systematic and effective manner. In other words, human groups (of restricted size) can impose enough discipline upon their members to save common property resources from destruction or degradation: in such instances, the authority axiom is thus satisfied. We will henceforth call this property regime a regulated common property, in opposition to the unregulated common property which does not satisfy the authority axiom.
8.1 The efficiency of regulated common property
Let us first analyse the means at the disposal of the community regarding the use of the resource by its members. In economic theory, an important way of correcting externalities, such as those associated with common property, has been suggested by Pigou, and consists of taxes, subsidies, and quotas. We shall examine this approach in some detail before pointing to a number of other aspects of the efficiency issue under common property.
Taxes and quotas
Let us re-examine the example of the fishery given in Chapter 2. Total catches, Y. were given by the following expression:
where m indicates the number of fishermen in the community and n, the number of boats fisherman i operates. By letting PN stand for the rental price of a boat, as shown in Chapter 2, the profit-maximizing number of boats per fisherman, in the Nash equilibrium of the unregulated common property, is:
(1)
This is the generalization of the formula obtained for two fishermen: (a - pN)/3b. The total number of boats in this equilibrium, NN, is therefore equal to:
(2)
The optimal (efficient) number of boats is given by setting m = I in these two equations. One obtains:
(3)
Let us now introduce a tax scheme such that the new Nash equilibrium supports the Paretooptimal solution described in (3). To have such a property, the tax scheme must imply the imposition of a tax rate on the variable factor, so that the agents will have an incentive to reduce its amount in use. Let us impose a tax rate, t, per boat. Equations (1) and (2) become:
(4), (5)
The tax per boat must be such that NT = N* for the new equilibrium to be Pareto-optimal. Let us call such a tax rate, t*. It is standard algebra to show that:
(6)
The tax rate so calculated has an interesting interpretation. Indeed, assuming that (m - 1) fishermen operate the optimal number of boats, n*, what is the impact (the 'externality') fisherman i causes on them by operating one more boat? Total profit of the (m - 1) fishermen is given by:
(7)
The externality imposed by fisherman i on all other fishermen is therefore equal to:
(8)
where use is made of equation (3). Therefore, the optimal tax is that which obliges agent i to fully internalize the externality he imposes on others by not behaving in an optimal manner.
This simple example shows how, through an adequate tax scheme, the community or the State is able to achieve efficiency in a common property resource. One should note that the community could have alternatively imposed a quota, q*, on the number of boats each fisherman is allowed to operate and also achieve efficiency. In that case,
(9)
The above example can be generalized so that, in general, Pareto-efficiency is achieved through quotas or taxes which allow agents to internalize fully the externalities which their own decisions are bound to cause. Therefore, the State (or the community) need not necessarily resort to privatization to solve the commons problem. It can achieve the same efficient equilibrium through an appropriate tax policy. Moreover, the following can be shown (for more details, see Dasgupta and Heal, 1979: ch. 3):
Since, through taxes, virtually any externality can be correctly internalized by the agents, one can surmise that an open-access equilibrium can also be made Pareto-optimal through an appropriate tax scheme. Let us illustrate this by way of the above example. We know that, in the open-access equilibrium, the total number of boats, N.,, is such that profits are nil:
(10)
For the open access equilibrium to be Pareto-optimal, the tax per boat, t, must be such that: N0 = N*. One gets:
(11 )
This has interesting implications. Indeed, by taxing appropriately, the State is able to correct the externality caused by open access. The problem is nevertheless that, in many cases, open access itself results from the impossibility to control the access to the resource. In these conditions, one cannot see how the State will be able to impose a tax on the use of the resource.
As illustrated above, quotas can be used in place of taxes to increase the efficiency of the common property equilibrium. However, this does not imply that taxes and quotas are always equivalent. For example, when there are some uncertainties associated with threshold phenomena (as is the case for many environmental problems), quantity controls may be more appropriate. To illustrate this, let us assume that the size of the externality, E, grows with the amount of the variable factor used, x, as described in Figure 8.1.
FIG. 8.1. Taxes versus quotas and threshold phenomena
Through quotas, the regulatory agency can ensure that the amount of variable factor, x, will always remain below x°, while, if the impact of taxes on the chosen amount of x is stochastic, or is not known with certainty by the regulating agency, the only way to ensure that x < x° is to impose a high tax rate on x, which can be undesirable because it is likely to be inefficient. In this case, quotas are preferable to taxes. However, in many instances (e.g. pollution or energysaving), taxes may be preferred to quotas because they provide incentives to technological progress aimed at reducing externalities still further. From the point of view of dynamic efficiency, taxes are often superior to quotas (see Dasgupta and Heal, 1979: ch. 13; Schokkaert, 1991; Fisher, 1981: 199-203; Baumol and Oates, 1971).
It is worth emphasizing that the amount of the tax that needs to be imposed on the users to achieve the Pareto-optimal outcome is not necessarily feasible. As a matter of fact, if these users have valuable exit options, they will threaten to leave the user group in case the tax reduces the expected benefits from their participation in the collective resource's exploitation below what they can get in their best alternative occupation. Such a threat of withdrawal may be a cause of worry for the group in so far as losing one of its members means that it is going to be less effective, for example, because the presence of everybody is desirable for certain actions (CPR maintenance or protection, risk-sharing) requiring co-ordinated efforts. In these circumstances, exit possibilities have the effect of reducing the amount of the taxes which a group can impose on deviant members. In technical terms, the threat to impose the optimal tax (in the above-defined sense) may not be credible since the group is willing to renegotiate the initial punishment scheme and the defector is well aware of it (the punishment scheme is not 'renegotiation-proof'). It is possible, however, to impose lower tax amounts or punishments that are self-enforcing, yet, precisely because they are less harsh, the amount of co-operation they can support is reduced. In that case, as pointed out by Fafchamps, 'deviants will not be able to renegotiate themselves out of their own punishment' (Fafchamps, 1992: 163). We will return to this point in Chapter 12 when we will attempt to understand the rationale of non-decentralized sanction systems.
An oft-used mechanism for regulating common property consists of temporarily excluding a rule violator. Hirshleifer and Rasmusen have shown that this mechanism can actually correspond to a subgame-perfect equilibrium strategy in a particular kind of game. The game they describe is based on the following assumptions: (i) a defector gets a higher payoff than a co-operator in the round in which he defects; (ii) payoffs per member are an increasing function of the number of members who co-operate; (iii) the payoffs per member when everyone defects do not depend on the size of the group; (iv) a member always prefers to be in the group than being ostracized; (v) the voting procedure used by the group to exclude a violator does not entail any direct or indirect benefit; (vi) for a violator to be excluded (ostracized), it suffices that one member votes for his exclusion; and (vii) ostracism lasts only one round.
Under these assumptions, they then prove that, even in a finite repetition of this game, the following strategy, supporting co operation till the penultimate period, is a subgame-perfect equilibrium: (a) co-operate until the last round, (b) defect in the last round, (e) vote for excluding anyone who deviates from the above strategy, including players who fail to vote for excluding defectors, those who have wrongly voted for excluding a co-operator, those who fail to ostracize those who fail to ostracize violators, and so forth. The crux of the argument lies in the fact that players can credibly threaten to exclude a defector in the last round since such a move does not entail any cost for them. The possibility to inflict a real punishment at no cost to the punisher has the effect of breaking the logic of the backward-induction argument underlying the non-co-operative outcome of the finitely repeated PD game.
If it is moreover assumed, in violation of assumption (v), that players gain a little bit of pleasure from excluding a violator, that is if they have 'a little morality', the above strategy is the unique subgame-perfect equilibrium of the game. As pointed out by the authors, 'the strategy "Always Defect, Never Ostracize", for example, is no longer an equilibrium because players would raise their payoffs by ostracizing defectors in the next-to-last round'. In this case, 'Morality achieves co-operation because some players want to reduce the welfare of others, if those others behave wrongfully' (Hirshleifer and Rasmusen, 1989: 100-1).
The logic highlighted in the above game can be extended to a wide class of problems including social dilemmas such as co ordination games or appropriation problems with respect to a common property. This means that assumptions (i) and (ii) are not essential to reach the aforementioned result. As a matter of fact, the important assumption is assumption (iii) according to which, if everyone defects, there is no benefit from a larger group size. Note that, in the commons problem, every remaining participant necessarily gains from the temporary exclusion of a violator, even when everyone 'defects', so that this assumption holds a fortiori. On the other hand, the voting procedure for exclusion is not important. Thus, the result that ostracism can enforce co operation can also be derived if exclusion is decided on the basis of majority. Moreover, ostracism works in much the same way if it is irrevocable.
However, for obvious reasons, this last extension is permitted only if the gain derived from a larger group is not too high.
There is another point which is worth making: punishment is generally more effective when it can be made escalating through a suitable design mechanism. This is particularly evident if information is imperfect since, in these circumstances, there is always a risk of imposing an unjustified sanction on a resource user who did not actually break the rule. If, however, the same individual is presumed to be a rule breaker on several occasions, then the probability that he actually behaved co-operatively becomes very low. As a result, harsher and harsher punishments can be meted out to him without incurring too much risk of misapplying sanctions. Graduated sanctions can also be justified when information is perfect. As a matter of fact, it is probably not very effective to impose harsh punishments on someone who is caught for the first time violating the rule. By imposing a small fine on him, the society wants to send him a warning (or, to reactivate the ruling norms) rather than actually punish him. As pointed out by Ostrom, a small penalty may be quite sufficient to remind the infractor of the importance of compliance. Such an approach is especially advisable when resource users may make 'mistakes', since a large fine or harsh punishment imposed on a person facing an unusual problem may produce resentment and unwillingness to conform to the rules in the future (Ostrom, 1990: 97-8). Consequently,
Graduated punishments ranging from insignificant fines all the way to banishment, applied in settings in which the sanctioners know a great deal about the personal circumstances of the other appropriators and the potential harm that could be created by excessive sanctions, may be far more effective than a major fine imposed on a first offender. (Ostrom, 1990: 98)
External authority and imperfect monitorability,
As hinted at above, in many field settings detection of free-riding is difficult and costs have to be incurred to make it possible. It can, however, be argued that imperfect monitorability as such is not an unsurmountable impediment to the intervention of an external authority which would be entrusted with the task of imposing sanctions to punish non-compliance with resourcepreserving rules. This theoretical result is important in the context of large groups in which the intervention of such an authority may be needed to provide the appropriate 'selective incentives' to the resource users. In the words of Olson, 'unless the number of individuals is quite small, or unless there is coercion or some other special device to make individuals act in their common interest, rational, self-interested individuals will not act to achieve their common or group interests' (Olson, 1965: 2; see also Taylor, 1987: 9). To demonstrate our point, we proceed in several steps. First, consider the PD game described in Figure 8.2.
Suppose an external authority enters the picture and decides to sanction defection in such a way as to transform the above PD game into another game where mutual cooperation would become a unique equilibrium outcome (see Figure 8.3).
Thus, by inflicting a punishment of 2 payoff units on each defector (whether he is found to defect singly or to defect simultaneously with the other player), the external authority has been able radically to modify the situation: the dominant strategy of each player is now to co-operate instead of free-riding as in the initial game.
FIG. 8.2. A 2 X 2 prisoner's dilemma game
FIG. 8.3. A 2 X 2 external-authority game with complete information (adapted from Ostrom, 1990: 10)
However, things are apparently more difficult when the external authority possesses only incomplete information about the relevant parameters. Indeed, it can then be shown, with the help of an example drawn from a recent book by Ostrom (1990), that its intervention may turn out to be an outright failure for a given level of punishment. If the information about the particular actions of the resource users is incomplete, the external authority cannot avoid making errors in imposing punishments. Let yd be the probability with which the agency punishes defections (a correct response) and, yc, the probability with which it punishes co-operative actions (an erroneous response). Note incidentally that (1-yd) represents the probability with which it fails to punish defections (an erroneous response) and (1-yc) the probability with which it does not punish co-operative actions (a correct response). The fine imposed on a defector amounts to 2 payoff units. The payoff structure of the game is thus as described in Figure 8.4.
Notice that there is complete information when yd = 1 and yc = 0, in which case the game displayed in Figure 8.4 is strictly the same as that shown in Figure 8.3. Information is incomplete when both yd end yc lie (strictly) between zero and one. Assume, for example, that the external agency imposes sanctions correctly with a probability of only 0 7 ( yd = 0 7 and yd = 0.3): in other words, there is a 30 per cent chance that a defector escapes punishment and that a co operator is being imposed an unjustified sanction. We then have the specific payoff matrix shown in Figure 8.5.
It is easy to verify that, given this payoff structure, the resource users again face a PD game: their dominant strategy is to violate the resource-preserving rules.
For the Pareto-superior outcome to materialize, it suffices that the payoff obtained by a player when he free-rides while the other co-operates is smaller than the payoff he would receive by co-operating with a co-operator. We therefore require that the following condition be satisfied:
10 — 2yc > 11 — 2yd.
In more general terms, if C and D stand for the payoffs, respectively, of co-operation and defection in conditions where the other player co-operates and before any sanction has been imposed, and if F stands for the fine imposed by the external authority, efficient monitoring requires that
F > (D - C)/(yd - yc).
In other words, the higher the incentive to defect, and the more imperfect the monitoring (the less likely a correct punishment is meted out to the agents), the higher the amount of the fine must be in order to make monitoring, activities effective. In small groups, since the difficulty of monitoring is obviously much less than in large groups, the probability of incorrect punishments is correspondingly lower. Therefore, the amount of the fine to be imposed to induce compliance is also lower. To take advantage of group size, it is clearly advisable that monitoring and sanctioning responsibilities be delegated to local or communitybased authorities.
Monitoring costs can also be reduced by shifting to methods of resource conservation that are relatively easy to check out. Indeed, monitoring being less imperfect, it could become effective with a lower amount of fines. The point can be made as follows. Difficult-to-monitor conservation methods imply that lengthy investigation procedures are necessary to provide adequate evidence that a suspected violation has actually occurred. These lengthy procedures have the effect of reducing the number of cases that can be investigated and, as a result, the perceived risk of detection by violators is low (yd low so that the right-hand term of the above equation is high) (Sutinen, Rieser, and Gauvin, 1990: 360-1).
In fishing, for example, gear restrictions are typically viewed as inefficient regulatory measures because they increase the costs of production (Crutchfield, 1961). However, since they may be less costly to enforce than other measures, gear restrictions could turn out to be the most efficient method of regulation when enforcement costs are taken into account (Sutinen and Andersen, 1985: 384). In particular, gear restrictions are less costly to monitor than the control of individual vessel quotas or minimum-size or minimum-weight restrictions. In the case of the latter, indeed, the alleged violator can challenge the techniques used by enforcement agents in sampling and measuring the size of the catch (Sutinen, Rieser, and Gauvin, 1990: 360). Alternative methods of monitoring are particularly advisable when the amount of the fine required is too high to be practically feasible: if the penalty meted out by enforcement authorities is perceived as too severe (i.e. unfair) by the community of users, the latter are likely to resist and social pressure against violators will be weakened (ibid.: 341).
Decentralized taxes and imperfect monitorability: an illustration
Monitoring activities can also be carried out in a completely decentralized way by the user groups concerned. In a recent paper, Weissing and Ostrom (1991) have analysed this issue in the context of water management systems. Towards that purpose, they have constructed an (N + 1)-agent 'irrigation game' in which, at each period, there is one turn-taker who pumps water out of the collective channel and N turn-waiters. There is a predetermined set of rules which regulate access of each agent to a turn-taking position according to a rotating pattern. Potential conflict exists between these two types of agent since the turn-taker can steal water and the turn-waiters are negatively affected by this event. Also, sanctioning activities are assumed to be performed by the users themselves without the intervention of any external authority (there are transfers of payoffs from agent to agent), even though we are not told how the participants come to agree on the amount of the (uniform) fine to be imposed and why the rule-breakers agree to pay the fine under conditions where it is not evident that punishment is self-enforcing. Something essential is obviously left out of the model (the punishment mechanism is not made explicit) and we have therefore to conclude that, implicit in the irrigation management system modelled by Weissing and Ostrom, a regulating agency is at work.
Fig. 8.6. Normal form of the two-person irrigation game
In the following, we shall present a modified version of their most basic model. We assume that there is one turn-waiter (TW) and one turn-taker (TT). TT has two possible actions, 'stealing' (S) or 'not stealing' (-S), and TW, without knowing TT's decision, has to decide between 'monitoring'(M) and 'not monitoring' (-M). Stealing can only be detected by monitoring, but monitoring is not completely efficient. Assume that the probability of detection is equal to a and µ is the 'monitoring tendency', that is, the probability with which TW monitors TT. Accordingly, the probability for TT to remain undetected is given try: h (µ, a ) = (1 - a µ), which the authors call the monitoring deficiency of the system. The payoff matrix associated with this game is given in Figure 8.6.
As can be seen above, when TT does not steal and TW does not monitor, their payoffs are equal to zero, reflecting a status quo situation. When TT does not steal and TW monitors, the latter incurs the fixed monitoring cost C. If TT steals and TW does not monitor, the former gets a payoff of +B (the value of the water stolen), and the latter gets -B (the value of the water lost). Now, when TT steals and TW monitors, TT earns a (-P) + (1 - a )B, where P represents the fine paid by TT to TW if his stealing is detected. The payoff accruing to TW is a (P) + (1 - a )(-B) - C. In such a game, the characterization of equilibria obviously depends on the values of the parameters. Here, we shall assume that, when TT steals, it cannot be TW's best response to abstain from monitoring, which implies a > C/(P + B).
Two cases must be distinguished. First, consider the situation in which a < B/(P + B), meaning that, when TW monitors, TT's best reply is to steal. In such a situation, there is a unique Nash equilibrium: TT always steals (it is a dominant strategy) and TW monitors. Second, we have the situation in which a > B/(P + B). In this case, it is no more rewarding to steal if the other monitors. As a result, there is no equilibrium in pure strategies in the above game. There is a Nash equilibrium in mixed strategies, in which TT steals with probability s * = C/(a P + a B) and TW monitors with probability m * = B/(a P + a B).
At the equilibrium, the stealing probability of TT is a positive function of the relative cost of monitoring with respect to the expected benefits. Conversely, the monitoring probability of TW is a positive function of the benefit of stealing water relative to its expected cost. The striking result is that, at equilibrium, stealing always occurs with positive probability, and, yet the turn-taker is not fully monitored. This derives from two key assumptions of the model. 'On the one hand, monitoring is only profitable if there is a positive chance to prevent a stealing event. If no stealing occurs, the costs of monitoring outweigh its benefits. On the other hand, stealing is always profitable if no monitoring occurs.... A strategy combination without stealing cannot be in equilibrium since a zero stealing rate induces the turnwaiters not to monitor, and a zero monitoring rate in turn gives the turn-taker a positive incentive to steal' (Weissing and Ostrom, 1991: 240).
Weissing and Ostrom show that the above result (some stealing occurs at equilibrium) also obtains when there are many turn-waiters. There are two different kinds of equilibria. In the asymmetric case, some agents are relatively specialized in monitoring activities. An interesting finding derived by the authors in a companion paper (1992) is the following: the introduction of a specialized guard will sometimes but not always make an irrigation system more efficient in the sense of reducing both the stealing and the monitoring rates. (Note that, in their corresponding model' the rewards of the guard consist not only of the value to him of the avoided water loss, but also of a special reward obtained in the case where he himself has detected the stealing event.) This suggests that the relative specialization of one agent in monitoring activities may reduce the turn-waiters' inceptive to monitor the turn-taker's behaviour. Between the turnwaiters, the structure of the monitoring game being played is that of a chicken game.
In the symmetric case, all turn-waiters monitor at the same equilibrium rate. It is then possible to demonstrate that the equilibrium rate of stealing increases with the number of turnwaiters. This is because the total benefit of detection of a stealing event is evenly spread among all turn-takers, whether they monitor or not. As a consequence, the individual benefit from monitoring and detection decreases with the number of turn-waiters (Weissing and Ostrom, 1991: 227). Other comparative-static results) obtained are rather straightforward: for example, the equilibrium stealing rate increases with a decrease in the detection probability of monitoring (a), in the size of the fine, (P) or in the loss of water due to undetected stealing, (B), and with an increase in the cost of monitoring, (C).
Since rule violation occurs at equilibrium when enforcement is imperfect, one must expect the optimal steady-state size of the stock of a renewable resource to be all the smaller as enforcement is more costly. This has been formally demonstrated by Sutinen and Andersen for a fishery. When enforcement is imperfect, the optimal steady-state stock size lies between the smaller open-access stock size and the larger stock size where catch rates are assumed to be perfectly controlled at zero cost (Sutinen and Andersen, 1985).
Enforcement costs of a regulated common property
As has been explained above, common property, whether regulated or not, differs from open access because the community detains an exclusive right of use on the resource, and is therefore entitled to exclude non-members. The enforcement of common property rights therefore entails costs, similar in nature to those resulting from the establishment of private property. However, it should be noted that, on the one hand, common property may be more effectively established than private property. It has indeed been argued by Bruce and Fortmann that 'the combined social and physical force of a community may be better able than single individuals to protect a resource against incursions by outsiders' (Bruce and Fortmann, 1989: 9). On the other hand, enforcement costs of common property are likely to be lower than those of private property, since, in most instances, former users of the resource simply get their rights recognized through common property. In other words, recognizing common property amounts in many cases to enacting an état de fait. Furthermore, in those situations where the privatization process has resulted in a parcelling out of the resource, the costs of enforcing privated property rights (such as the costs of fencing, or those involved in surveillance activities) may be high relative to those incurred under common property.
This being said, regulation of common property may also entail significant transaction costs since a centralized decision unit, namely a political authority, has to be established. Demsetz (1967) thus argues that: 'negotiation costs will be large because it is difficult for many persons to reach a mutually satisfactory agreement.... But even if an agreement among all can be reached, one must yet take account of the costs of policing the agreement, and these may be large also' (Demsetz, 1967: 354-5). There are many aspects to this issue and many of them will be addressed in Chapter 12 in the light of empirical evidence. At this stage, it is sufficient to say that the costs involved will be lower if there is adequate leadership to create the necessary consensus or, at least, to drive enough resource users to adopt co-operative behaviour. Traditional leadership seems to be particularly welcome in so far as, since it is already in existence, it entails no set-up costs. Yet, as will be seen in Chapter 12, this solution may also present important shortcomings. Furthermore, if there are costs associated with the enforcement of political authority to regulate common property, once established, this authority will be in a better position than private owners to cope with 'residual' externalities and with the new externalities which may later emerge (see above, Chapter 3).
In the above, we have assumed that the regulation task under the common property regime is entrusted with the user community. This need not be so since the State is also a potential candidate for such a responsibility. The question as to whether the State or the community (or any intermediate agency) is more effective as regulating agency is, of course, a crucial question that deserves careful consideration. It has actually many facets which will be discussed throughout Part II, more specifically in Chapters 11 (sect. 1) and 13.
At this stage, suffice it to say that information is a central dimension of any choice of the appropriate level and method of regulation. As demonstrated above, by imposing Pigovian taxes, a regulating agency equipped with the necessary powers can achieve efficiency in a common property resource. What is worth bearing in mind is that this solution is feasible only if the regulating agency possesses satisfactory infor mation about the state of the resource (e.g. the stock of a fishing population), the flow of its current use, and the identity of its users. Only then can the right fee be levied on the right people. Yet, the more centralized this agency the more difficult the task of collecting the relevant information and, therefore, the more serious the problem of implementation of Pigovian taxes (or quotas). As pointed out by Deacon (1992) in his discussion of the means of controlling tropical deforestation, such a difficulty is likely to preclude the State from directly managing common property resources. Being closer to the resources to be managed, user communities possess an informational advantage compared with any centralized agency. Yet, on the other hand, the information-processing ability of the latter is typically higher than that of the former. Hence the need for a co-management approach as further explained and illustrated in Chapter 13.
