Aggregating Animal Welfare

Karsten Klint Jensen Institute of Food and Resource Economics University of Rolighedsvej 25 DK-1958 C. Email: [email protected] Phone: +45 28738581

Technology request: Power Point

There has been increasing interest in methods to assess animal welfare on farm level in order to improve public surveillance and support labeling of welfare friendly production. The most comprehensive method to date has been developed by the EU project Welfare Quality ®. One question raised by the aggregation of animal welfare is whether fairness or concern for the worst off should play a role in the overall assessment of a farm. Welfare Quality ® concluded that it should.

In this paper, I reconsider the issue of aggregation. The framework is this. Animal welfare is assumed to be determined by an ordering of vectors the values of a fixed number of measurable indicators, each of which is assumed to be independent of the others. The overall goodness of the farm is assumed determined by an ordering of the values of such indicators for all the affected individuals.

In practice, there is limited information. It is not possible to assess the welfare of each individual. Instead, there is information on how the farm fares overall on the different indicators. More precisely, the data says how many individuals there is for each indicator value in the range. But it is not possible to discern, for different indicators, whether the same or different individuals are affected by welfare problems.

Welfare Quality ® assumes (expressed in my framework) that each indicator on farm level is separable in overall goodness of the farm. If we add the plausible assumption that the indicators of each individual is separable in overall goodness of the farm (apparently overlooked by Welfare Quality ® , it is possible to demonstrate - I do that in the paper - that overall goodness of the farm can be represented by a function which is the sum of the value contribution of each indicator for each individual.

This result contradicts the aggregation model of Welfare Quality ®, which gives more (negative) weight the worse off conditions of the animals. My result does not show that there is no room fairness or a concern for the worse off in aggregating animal welfare. Rather, the result is dependent of the nature of the information. With the underlying assumptions, I argue, a reasonable interpretation of the result would be that overall goodness of a farm is a matter of the extent of problems. Finally, I analyze the issue of comparisons across farms. Welfare Quality ® assumes this is possible; but thereby a number of implicit assumptions are made. I suggest how different number choices for animal welfare can be explicitly addressed.