Cointegration analysis with structural breaks and deterministic trends: an application to the Canadian dollar

Abstract

This article applies recent developments in cointegration analysis with structural breaks and deterministic trends to analyse the relationship between the real Canada–US exchange rate and commodity prices. Previous empirical studies disagree on whether these variables are cointegrated. The root of disagreement could be in the handling of deterministic trends and potential structural breaks. I find that even after controlling for these matters, the question of whether the real exchange rate and commodity prices are cointegrated for Canada remains unresolved.

Acknowledgements

I thank an anonymous referee for helpful suggestions. I am grateful to Robert Lafrance and seminar participants at the University of Saskatchewan for useful comments and to Jiming Gu for excellent research assistance. Financial support of the President's SSHRC Fund, University of Saskatchewan is greatly acknowledged. All remaining errors are mine.

Notes

¹ There are also panel studies on unit root and cointegration in real exchange rates. See Taylor and Sarno (1998), Taylor (2006), Hooi and Smyth (2007) for a review of this literature.

² See Bailliu et al. (2007), Cashin et al. (2004), Murray et al. (2001) and references therein.

³ The following discussion is based on Issa et al. (2006). Initially, Canada reacted to the price of oil shock in 1973 by introducing a policy of supply management that lasted until 1985. In response to the second price of oil shock in 1979–1980, Canada introduced the National Energy Program which included new energy taxes and some policy initiatives. In the second half of the 1980s, the Canadian government has started to move into the direction of deregulation of the energy market. First, the Western Accord was signed in 1985. Then, the Canada–US Free Trade Agreement in 1989 and the North American Free Trade Agreement in 1994 have resulted in the creation of the integrated North American market.

⁴ I cannot replicate the results in Chen and Rogoff (2003) exactly, see their Table 1 (Chen and Rogoff, 2003, p. 142). It could be due to the fact that they construct their own commodity price indexes, while I use official series. The difference among the results is not substantial though.

⁵ The formula for is based on simulations in Perron and Rodriguez (2003).

⁶ Oil accounts for ≈60% in the Bank of Canada energy index.