In some cases spot markets failure to govern to whole or a part of the marketing channel effectively and contractual relations are gaining more importance. It is especially true in case of agricultural markets, since these markets became more differentiated and market players are vulnerable in most of the cases. Examination of Hungarian dairy sector is an actual issue, so that one could understand how contractual systems work in the situation when crises appear thanks to governance insufficiency. Our research’s aims are to present a theoretically structured framework of contracting arrangements of milk producers based on Transaction Cost Economics’ (TCE) predictions and economics of contracting and an empirical analysis of the key determinants of governance structure between farmers and dairy processors in Hungary. The source of the research is a theoretical argument based partly on review of Hungarian and international literature on relevant market channels, economics of contracting and governance structures. These gave the theoretical determinants of testable prepositions. In the framework of a postal survey in the second quarter of 2005 we carried out a survey among milk producers. A total of 300 questionnaires containing closed and open questions were sent out for milk producers, 65 of them could have been evaluated. The results have been structured, electronically converted for applying SPSS-software. The data base has been analysed by employing multivariate techniques (cluster analysis, linear regression, multidimensional scaling, etc). First, to be able to decide the number of clusters, we applied a hierarchical clustering. The formation of starting clusters was made by giving the number of the future groups which based on hierarchical cluster method and dendrogram. Hence asset specific investment, price determination and bargaining power proved to be significant in dividing the cases into three groups focusing on governance structure. We revealed the main characteristics of clusters focusing on con-tracting attributes. Groups’ means comparison obtains the result that there is no significant difference in partner change, neither in the whole sample nor in the sub-groups. Unlike from this, the reasons for selling to a particular buyer are different in the clusters, the most important factors are reliability, valid contract and the ones based on geographic reasons. The variables mentioned above were suitable for further investigations, so with the help of linear regression we attempted to see their effect on the contract period. Taking into consideration the t-values of the variables, neither asset specific investment and bargaining power, nor price determination have role in the explanation of contract period. Since the variables applied in the whole survey measure same theoretical concepts, we had the possibility to reduce their number by multidimensional scaling. The aim of this scaling is to gain information about the differences among the respondents reducing the dimensions of the variables. The goodness of fit was good in case of three and two dimensions, so we found that the six-dimensional space can be reduced into two or three dimensions without giving up the differences among cases.
