Abstract
The Canadian oil industry has been the subject of several debates with respect to charges of transfer pricing. On the basis of a large data set, which includes all oil shipments into the US and Canada from 1974–84, the first direct test of manipulative transfer pricing, based on actual company behaviour, is performed. Particular attention is paid to the multicollinearity problem which arises in this context. Through regression analysis it emerges that the six largest Canadian affiliates of multinational corporations paid crude-oil import prices which were generally equal to or lower than prices of third-party transactions for a country in a given year. This is a fairly robust result which highlights the existence of manipulative transfer pricing. In the present case, the practice was beneficial to Canada.