Removing the CWB as a single desk grain marketer: Assessing the initial effects on the Canada-US feed barley market

Abstract

This thesis examines some of the economic effects associated with the elimination of single desk marketing on Canadian feed barley exports. It focuses on the interactions between Canadian and US spot feed barley markets in transition after this policy change in Canada. A brief introduction about world and regional barley markets is provided. The role of the CWB single desk and its role in Canadian barley marketing are discussed to motivate analysis about the effects of its absence. This study postulates there should be no significant change in Canada–US regional feed barley markets, based on conclusions from previous studies. This postulate is broken down into three testable hypotheses under the framework of spatial price analysis. With respect to the Canada–US regional feed barley market as single desk marketing was eliminated in Canada; (1) There should not be a significant structural break in the feed barley prices; (2) There should not be a significant change in market integration; (3) There should not be a significant change in the direction of price transmission. To test these hypotheses, the study employs econometric tests on Canadian and US prices spot prices for substitutable feed barley. The hypotheses are tested using a structural break test, a cointegration test, a Granger causality test, and associated impulse response functions. Since structural break tests do not find significant breakpoints in the data, the first hypothesis cannot be rejected. Next, the sample is split into two subsamples at the date when single desk was eliminated. An Engle-Granger procedure and the Johansen procedure are used to test cointegrating relationships between the variables. The results do not allow us to reject the second hypothesis of no significant change in market integration. In contrast, the third hypothesis is rejected, as a significant change is uncovered using the Granger causality test. Simulated impulse responses are also consistent with this finding.