Key-words: contractual arrangements. Agrosystem coordination. Contractual power.

This paper aims to study inter-segment power relations in transactions in swine agribusiness system in the state of Paraná, considering Transaction Cost Economics (TCE), Measurement Costs Theory (MCT) and Industrial Organization (IO) theories. Considering the pork agrosystem (SAG), asset specificity, measurement attributes and power asymmetry turn the study relevant, given that production structures are organized mostly by integration contracts involving a large number of small producers and a small group of processors, with specific organizational and productive designs. That condition indicates an environment subject to contractual imbalances and monopoly power exercise. Such analysis within the scope of efficiency comprises a differential in understanding the relationship of law and guarantees. Moreover, the framework established for agribusiness and its scenario indicate higher pressures in world market, especially with the narrowing of margins for exporters, which may be reflected in renegotiation or, if necessary, establishing situations of contractual hold up. For the execution of this work, contractual scheme, based on Williamson (1996) proposal was considered, complemented by approaches of IO and MCT. So, measure aspects of attributes and power exercise were inserted. In the presence of asset specificity (k), the presence of safeguards (s) is complemented with the possibility of measuring attributes (m) and the exercise of power (f). Thus, considering specific assets (k>0), and absence of contractual safeguards (s = 0), the transaction is contractually unstable, and the agents are inevitably at risk of opportunistic behavior. The inclusion of safeguards in the transaction (s> 0) discourages opportunistic behavior, providing reduction of uncertainty and the introduction of some level of protection. The possibility of exercising power (f> 0) leads to the construction of hybrid forms of governance, characterized by power asymmetry and imbalance in the transaction. In the absence of power in the relationship (f = 0), two possibilities of governance are formed. If the assets are not related to specific measurable dimensions (m = 0), or if the measurement is highly expensive, the most appropriate form of governance is the hierarchy, as there would be ways to ensure property rights and avoid quasi-rent appropriation. It would prevail higher levels of control. On the other hand, if the relevant dimensions in the transaction are measurable and measured (m> 0), the transaction can occur via contract, and  the predominance of balance would be observed. The measurement of attributes, along with contractual safeguards, makes contract governance effective. In this case, contractual power is ensured in terms of balance and guarantees of property rights, demanding  higher level of coordination. Empirical work  involving 50 semi-structured interviews with producers and the use of content analysis method, allowed to ratify the new contract scheme proposed. We found that, in the presence of specific (k> 0), contractual safeguards are defined for both producers and processors (s> 0). However, the possibility of market power exercise (f> 0) leads to power asymmetry in contractual relations, specifically in favor of processors. In this dynamic, the measurement does not fulfill the function of generating contractual balance, but only presents itself as an way to govern the transaction, ensuring processorss’ property rights. Thus, there has been observed the possibility of exercising power by the buyer. Power asymmetry is clear in two distinct stages of the transaction. First, ex ante, for rigidities and the definition of contractual terms by processors and their responsibility in the measurement, with no flexibility to producers. Finally,  ex post, as terms not established in contracts are directed in favor of powerful agent, the buyer. In addition, those agents have the possibility of unanticipated changes in the contract, and producers are limited to validation of measurement, which involves variables and not attributes.