Excessive nitrogen (N) surpluses from mineral fertilizers and animal manure, appear to be a major pollutant in many European underground and surface wa- tersheds, posing a threat to the environment and the human health. To provide a general level of protection for all waters against nitrate pollution, the European Council established in 1991 the Nitrates Directive (91/676/EEC), defining a series of codes of good agricultural practice (EC, 1991). The non-point-source characteristics, however, of agricultural pollution pose a substantial problem in the effective regulation of reported water pollution problems. The inability of regulatory authorities to directly observe individual decisions (i.e. nitrogen usage) provides the farmers incentives to deviate from statutory requirements and retain nitrogen usage at the unregulated profit maximizing levels, with the associated adverse consequences. To ensure that regulated farmers comply with statutory nitrogen performance standards and that foreseen sanctions are im- posed on those deviating so that compliance is further enforced, Member States are required to incorporate a substantial monitoring mechanism in their policy design. It is evident thus, that the effectiveness of the existing regulatory policies to induce restricted usage of nitrogen input is heavily dependent on the ability of the monitoring and enforcement mechanism to provide adequate compliance incentives, and thus implement the Nitrates Directive.
The purpose of the present paper is to examine the effectiveness of a monitoring and enforcement mechanism to induce in the long-run a large population of homogeneous farmers to comply with the statutory requirements of a regulatory regime under different assumption regarding the way that farmers choose to comply or not with regulation. The examined regulatory regime falls into the category of public voluntary environmental programs and involves a combination of "carrot" financial inducements provided through the agri-environmental programs of the second pillar of the communal agricultural policy (CAP) and "stick" legal binding features of the action programs of the European Council Nitrates Directive (91/676/EEC).9
In our approach, and in contrast to the majority of the enforcement literature, farmers do not necessarily adopt an optimizing behavioral rule in their decision to comply or not with the suggested nitrogen usage constraint, but may follow evolutionary rules modeled by imitation dynamics. Most economic models assume that agents are "infinite in faculties", they act "as if" unboundedly rational (Conlisk, 1996). If farmers are characterized by full rationality then they adopt optimizing behavioral rules and they behave as though they had all the necessary information when they decide about complying or not. In such a case farmers have full knowledge of the structure of payo§s and after comparing the payoff that each strategy entails they define their optimal response to the regulation. This response is maintained across time and space if there is no modification of the policy parameters by the regulator. On the other hand, under bounded rationality agents "are no longer assumed to be mathematical prodigies with access to encyclopedic mannuals written by omniscient game theorists" (Binmore, 1992). Farmers cannot choose their individual strategy in an optimal manner, and their decision about whether to comply or not is adapted to the information revealed via their interaction over time. We assume that such passive decision making is based on the imitation of the better-off performing strategy and is modeled by the replicator dynamics, imitation rule. Under such an evolutionary process more successful agents and activities gradually increase their share in the population at the expense of less successful agents and activities (Conlisk, 1996), leading potential agents who have no clear idea what is going on to behavior that may look very rational indeed to a Kibitzer (Binmore, 1992).
