摘要：This paper examines the relations between real GDP, real labour compensation per hour, and market hours worked in
Canada over the period 1981Q1 to 2017Q4. We employ causality testing using the vector autoregression lag
augmented approach based on possibly integrated or cointegrated processes. This approach avoids the pre-test bias
associated with most other causality testing approaches. In addition, we control for the asymmetric impact of real
GDP growth on the evolution of these variables. The first finding is that of unidirectional causality from hours worked
to real GDP and bidirectional causality between real compensation per hour and market hours worked. Second, we
find that the variables have a unique long-run cointegrating relationship. Third, controlling for the non-linear dynamics
is important in analysing the causal relations among these variables.
