The subject of this paper is to estimate consumers' welfare loss when the BSE testing age criterion for cattle is revised from all months to older than 21 months, by analyzing the change in meat demand due to this revision. However this analysis is different from ordinary revealed preference methods in that the demand data after this policy change are stated, while the data before this change are revealed as usual. This is because this paper focuses on a policy change which is not substantially executed yet. Using this demand data I estimated the generalized corner solution model of Phaneuf et al. (2000) for an incomplete demand system model. When I simulated each monitor's welfare loss based on the model estimation result, the expected compensating variation per household per month was 214 yen at the sample mean and 55 yen at the sample median.