What determines the length of a typical Canadian parliamentary government?

In this paper we examine the length of political tenure in Canadian federally elected parliamentary governments since 1867. Using data on tenure length, we categorize the distribution of governing tenures in terms of a hazard function--the probability that an election will arise in each year, given that an election has not yet been called. We then ask whether that distribution responds in a systematic way to characteristics of the political and/or economic environment. Our particular focus is on whether there is evidence of electoral timing and whether governing parties have used economy policy in conjunction with federal elections. Finally we investigate whether partisan effects emerge. The results suggest that, independent of party affiliation, governing parties do engage in election timing. The data also suggest that election calls coincide with periods of monetary expansion and more with tax decreases than with expenditure increases, supporting the Persson and Tabellini (2003) hypothesis that under parliamentary systems, it is tax cuts (rather than expenditure increases) that will be most closely associated with elections. Unlike the case in other parliamentary systems, however, Canadian data also support the hypothesis that tough measures (expenditure cuts) are postponed until after elections.