Publication

In this paper we implement threshold methods to investigate technical efficiency (how good farms are at getting the most out of their inputs relative to the very best farms in the sample) in a context in which dairy farms of different sizes may rely on different technologies.  The method developed in this paper allows us to categorize farms in different groups, the number of groups being determined by the data.  We use a data set of 302 dairy farms covering a period of 11 years.  We used two functional forms, Cobb-Douglas and Translog, and four inputs to model milk production per cow.  Results were quite similar across functional forms, but the insertion of a time trend to capture slow structural change made a difference on the number of different technological regimes identified by our statistical procedure.  We found that farms with less than 34 cows, farms with 35-44 cows, farms with 45-66 cows and farms with 66 cows made homogenous groups.  Of particular interest is the lack of significance of either labour or capital for the 3 smallest groups of farms and the high level of technical efficiency.  This suggests that most of the dairy farms in these groups have too much capital and/or labour for what they produce. Many agricultural inputs can be bought or supplied in lumps and this affects more small farms than larger ones.  Quebec (and more generally Canadian) dairy farms are small by international standards.  The mean number of cows in our sample was only 52.  The small size of the farms can be attributed to the financial constraints that make expansion difficult under Canada's supply management policy.   Our different specifications provided consistent results in the evaluation of mean technical efficiency and the identification of the most and least efficient farms.  We did not find significant differences in the level of technical efficiency across groups of farms for most years. In the last year of our sample, the group made up of the largest farms were more technically efficient.  The fact that farms' individual efficiency scores were highly correlated from one model to another suggests that we can be confident in the validity of our results.   The main implication is that costs of production of Quebec dairy farms are high because of their small size, not because of heterogeneous managerial abilities in getting the most from inputs.  As the sector is confronted to increased competitive pressure from foreign competitors, the industry will have to consolidate so that farms can drastically increase in size.