Canadian Newsprint in the United States: a Multivariate Cointegration Analysis

In this paper, Johansen multivariate cointegration tests are adopted to investigate the U.S. demand for Canadian newsprint using monthly data from May 1988 to December 1996. Preliminary data analysis shows that all data are non-stationary which implies that previous results based on simple ordinary least squares are spurious. Johansen multivariate cointegration techniques allow for identifying a long-run relationship as well as a short-run relationship via an error correction model. Among the determinants are the export price of Canadian newsprint to the U.S., the exchange rate between the two countries, U.S. personal disposable income, and U.S. newsprint price.