View article

The major premise of this study is that in federal countries voters can balance and moderate national policy by dividing electoral support between different parties in federal and sub-national elections. We compare the non-concurrent federal and provincial elections in Canada to assess the balancing properties of sub-national elections. The balancing hypothesis implies that the federal incumbent party may suffer additional electoral losses in provincial elections. We use several statistical tests - ordinary OLS, “fixed effect” and “unbalanced random effect” cross-section time series - to analyze Canadian electoral data for the period of 1949-1997. All tests sustain that the incumbent party at the federal level loses votes in provincial elections.