Tracking Canadian Trend Productivity: A Dynamic Factor Model with Markov Switching

The author attempts to track Canadian labour productivity over the past four decades using a multivariate dynamic factor model that, in addition to the labour productivity series, includes aggregate compensation and consumption information. Productivity is assumed to switch between two regimes (the high-growth state and the low-growth state) with different trend growth rates according to a first-order Markov process. The author finds that labour productivity in Canada fell from the high-growth to the low-growth state towards the end of the 1970s, and that it has not yet reverted to the high-growth state. In particular, the model primarily attributes the resurgence of labour productivity growth in the late nineties to transitory effects.