Atheoretical and structural approaches to modelling Canadian-American macroeconomic interaction

Much recent work on Canadian-American macroeconomic interaction utilizes the vector autoregression methodology advocated by Sims (1980). Examples include Burbidge and Harrison (1985), Kusczak and Murray (1986) and Ambler (1988). The Sims' methodology is severely criticized by Cooley and Leroy (1985) and Leamer (1985) because, as it is generally implemented and interpreted, including in the studies cited above, the models contain strong implicit identifying assumptions. Bernanke (1986), Blanchard (1986) and Sims (1986) propose alternative estimation techniques which, while retaining some of the advantages of the vector autoregression approach, make the identifying assumptions explicit. This paper uses both approaches to estimate models of Canadian-American macroeconomic interaction. The results show the method adopted does affect the interpretation of the results. The vector autoregression methodology is not free of the imposition of untestable identifying assumptions.