نتایج انتخابات و بازده بازار مالی در کانادا

This paper investigates the relationship between federal election outcomes and expected returns and volatilities in the Canadian money, bond, equity and currency markets from 1951 to 2006. There is little evidence that investment opportunities are different in minority versus majority parliaments and only money market returns differ in Conservative versus Liberal governments. The equity market performs better in the late part of the electoral cycle than in the first two years. Furthermore, the Canadian equity investment opportunities are better in Democratic versus Republican administrations and in the late versus early parts of the presidential electoral cycle. The Canadian dollar is also affected by American election outcomes. No apparent variation in risk or expected state of the economy accounts for the differences in returns.

Canadian political parties are eager to differentiate themselves by their policies. In particular, their economic policies are often sold as being the best for the country's economy and financial markets. They furthermore argue that they are better equipped than the other parties to maintain strong relationships with Canada's most important trading partner, the United States. To be able to operate efficiently and reduce their country's economic uncertainty, they ask for ‘strong mandate’ during election campaigns. Political analysts and economists regularly comment on these important claims. They also analyze the influence of American politics on the Canadian economy. They finally discuss whether the ‘timing’ is right for the government to call an election or the opposition parties to force an election in a minority situation. In a context where these issues are as relevant today as they were throughout history, this article explores the historical relationship between federal election outcomes and Canadian financial markets returns from 1951 until 2006. We are interested in four questions. First, are financial market returns different between majority and minority governments? Second, are financial market returns different between right-leaning Conservative and left-leaning Liberal governments? Third, are financial market returns different between the early part and the late part of the variable-term federal mandates? Fourth, are financial market returns influenced by American election outcomes? To answer these questions, we estimate the expected returns and the standard deviation of returns conditional on election outcomes for ten monthly return series covering the money, bond, equity and currency markets. While the effect of election outcomes on the economy can potentially manifest itself in numerous ways, we focus on financial market returns as they are the easily measured outcome of the collective decisions of a large number of investors with strong economic incentives to act rationally. The competitive and generally efficient process of price determination in financial markets results in prices that adjust rapidly to new information so that realized returns over the long run should reflect the anticipations for risks and rewards conditional on election outcomes. Apart from being motivated by the common rhetoric of political actors and pundits, our study contributes to a growing academic literature examining the relationship between election outcomes and returns.1Hensel and Ziemba (1995), Chittenden, Johnson, and Jensen (1999), Santa-Clara and Valkanov (2003), Booth and Booth (2003) and Powell, Shi, Smith, and Whaley (2007) examine the relationship between the political ideology of the president and returns in the United States. Their results show that large and small-capitalisation equities yield higher returns under Democratic presidencies while U.S. Treasury bonds and bills produce higher returns under Republican presidencies, resulting in significantly higher equity and size premiums under left-leaning administrations.2 For example, Santa-Clara and Valkanov (2003) find annualized excess stock market return differences between 9 percent and 16 percent from 1926 to 1998. As no corresponding differences in volatility or macroeconomic conditions are found, they call the unexplained Democratic equity return premium a puzzle. Tufte (1978), Huang (1985), Hensel and Ziemba (1995), Chittenden et al. (1999), and Booth and Booth (2003) examine the returns in the early part and the late part of the United States presidential term. They find that large and small-capitalisation stocks give significantly higher returns and excess returns in the last two years of a term. For example, with data from 1926 to 1996, Booth and Booth (2003, Table 2) report annualized differences of 9.1% for large and 13.6% from small stock portfolios. This result can be interpreted as a sign of strategic election timing due to “policy timing” (e.g., Kayser, 2005, Nordhaus, 1975, Smith, 2004 and Tufte, 1978) and is known as the presidential cycle effect. Bond returns are not statistically different between the first and second halves of the presidential mandate. Tufte (1978) and Foerster and Schmitz (1997) examine whether the effects of the United States presidential election spill over into other economies. Tufte (1978, p.66) finds that “A presidential election in the United States is nearly as effective in producing accelerated economic growth in Canada, France, Germany, Japan and the United Kingdom as an election in the country itself.” Foerster and Schmitz (1997) find that the occurrence of a presidential election in the United States impacts international stock returns in 18 countries. Specifically, they document that the local currency stock returns are lower in the second year and higher in the first, third and fourth year of the United States presidential election cycle for most countries examined, including Canada. These findings, coupled with results from Mittoo (1992), Koutoulas and Kryzanowski (1994) and Normandin (2004) that the Canadian and American financial markets are partially integrated, suggest that Canada could experience the effect of the American election outcomes beyond the effect of its own federal election outcomes. Outside the United States, there is unfortunately little evidence on the effect of election outcomes on returns. A notable exception is Bohl and Gottschalk (2006). They study whether the left-wing premium and the political cycle effect extend to other countries. Using data on 15 countries, they find that stock returns are generally not significantly different between left-leaning and right-leaning governments and between the early and late parts of the political cycle. However, their dataset for many countries is limited. For example, while they do not find significant left-wing premium and political cycle effect in Canada, their sample covers only five elections in the period 1984–2003. In fact, to our knowledge, the only article focusing specifically on election outcomes and returns in Canada is Foerster (1994). Using data from 1919 to 1992 on the Toronto Stock Exchange returns, Foerster (1994) reports that minority governments are associated with higher average returns (7.4% versus 6.2%) and lower standard deviations (14.5% versus 17.3%) than majority governments. He also finds that returns are higher under Liberal governments than under Conservative governments (8.3% versus 2.8%), a result strongly influenced by the depression period from 1930-1935, and higher under Democratic administrations than under Republican administrations (10.4% versus 4.9%). He finally finds that Canadian stock returns during the third and fourth year of the American presidential mandate are higher than those during the second year. However, Foerster (1994) presents no statistics on the significance of his estimates and offers no formal statistical tests of the differences he documents. While his results are intriguing, it is unclear whether they are significant at conventional levels and hence reflect reliable information on the relationship between election outcomes and equity returns in Canada. Foerster (1994) also does not control his results for expected economic conditions and does not examine returns in the money, bond and currency markets, which are related to election outcomes in the United States. Given the similarities between Canada and the United States, a more thorough analysis of the Canadian evidence is thus needed to provide more conclusive evidence as well as a valuable out-of-sample check on the American results. Our monthly dataset is constructed from multiple sources and contains 672 observations that cover 18 federal elections since 1951, including 8 minority and 7 Conservative governments. It includes three money market variables (the nominal and real risk-free rates and the inflation), two bond market variables (the return on long-term government bond and the term premium), four equity market variables (the return on value-weighted and equally-weighted portfolios of all stocks, and the equity and size premiums) and one currency market variable (the return on the Canada/U.S. dollar exchange rate). While our dataset does not allow us to examine the effect of election outcomes for the 1867–1950 period, the limited data available prior to 1951 are less reliable and do not provide coverage of all the four financial market categories we investigate. Our empirical results on the effect of Canadian election outcomes can be summarized as follow. First, money, bond, equity and currency market returns are not different in minority versus majority parliaments.3 Thus we do not find any financial market evidence that minority governments increase uncertainty. Second, consistent with the United States findings, the real risk-free rate is higher by 2.60% and inflation is lower by 1.13% under Conservative versus Liberal governments in Canada. However, in contrast to the results on the Democratic equity premium, equity returns are similar in Canada for both governing parties. Hence assertions that left-wing or right-wing policies are better for the stock market are not supported by our investigation. Third, as in the United States, the Canadian equity market yields significantly higher average returns in the late part of the electoral cycle than in the first two years of majority mandates. The differences for the value-weighted and equally-weighted portfolios of all stocks and for the equity premium are respectively 8.31%, 14.54% and 9.82%. They are furthermore not accounted for by a corresponding increase in risks or by the exclusion of minority months. Concerning the influence of American election outcomes, the Canadian equity investment opportunities are better in Democratic versus Republican administrations and in the late versus early parts of the presidential electoral cycle. For example, the estimated equity premium is higher by 11.56% with Democrats in the White House and by 10.77% in the last two years the presidential term. The currency market is also affected by the American electoral results. There is however no evidence that having aligned left-leaning (Liberal-Democratic) or right-leaning (Conservative-Republican) leaderships is beneficial to the financial markets. But we find that the stock markets perform by far the worst when the Canadian and American governments are simultaneously in the early part of their mandates. All documented differences are statistically significant at conventional levels using Newey and West (1987) standard errors, and are robust to the use of simulations to assess the small sample and spurious regression biases discussed in Granger, Hyung, and Jeon (2001) and Powell et al. (2007) and Powell, Shi, Smith, and Whaley (2009). Furthermore, the significant return differences are not materially affected by the inclusion of various controls for the expected state of the Canadian or American economies, and are not accompanied by corresponding differences in standard deviation of returns. In fact, no test finds robust difference of volatilities across election outcomes. Although there is little well-developed theory relating election outcomes to financial market performance, we discuss some explanations of our results. In particular, the relationship between the Canadian electoral cycle and equity investment returns is consistent with the hypothesis of opportunistic election calling that could include policy timing. Furthermore, the influence of the U.S. presidential ideology and electoral cycle demonstrates the spill over effects of American election outcomes into the Canadian equity markets. Even though these explanations are non exhaustive, it is difficult to think of a rational explanation for the historically profitable equity trading strategies that could have been implemented based on election outcomes. As no apparent variation in risk or expected economic conditions account for the results, the large and persistent differences in equity returns based on election outcomes are therefore somewhat puzzling. Overall, the findings presented in this paper contribute to an increased understanding of the relationship between politics and financial markets in Canada. The rest of the paper is structured as follows. Section 2 introduces the data and the empirical method used in the paper. Section 3 presents the empirical results while Section 4 concludes the paper.

Federal elections in Canada and their related campaigns attract tremendous attention from citizens. Political parties spend large sums of money to get the message across that they should be allowed to form the next government because their policies are the most appropriate for the country. Once election results are known, there is continued interest (and media coverage) on whether the government has the strength needed to lead the country and is able to deliver on its promises. Ultimately, election outcomes can thus have wide-ranging implications for the country's future economic and social conditions. As financial markets generally reflect changing conditions rapidly, they represent a natural experiment to observe the effects of election outcomes. This article examines the historical relationship between federal election outcomes and Canadian financial markets returns from 1951 until 2006 by asking four questions. First, are financial market returns different between majority and minority governments? We find no significant difference in means and standard deviations for money, bond, equity and currency market returns. Second, are financial market returns different between right-leaning Conservative and left-leaning Liberal governments? Except for the average inflation and real risk-free rate, we again discover no significant difference. Third, are financial market returns different between the early part and the late part of the variable-term federal mandates? We find that the equity market yields significantly higher average returns in the last part of the federal mandate than in its first two years, consistent with the hypothesis of opportunistic election calling that could include policy timing. Fourth, are financial market returns influenced by American election outcomes? We show that the equity market performs better in Democratic versus Republican administrations and in the late versus early parts of the presidential electoral cycle, an apparent spill over effect of American election outcomes into the Canadian financial markets. The currency market is also affected by the American electoral results. We furthermore show that the stock market performs the worst when the Canadian and American governments are simultaneously in the early part of their mandates, and much better otherwise. However, having ideologically aligned leaderships in Canada and the United States does not lead to a difference in investment opportunities. All differences in mean returns are not accounted for by expected state of the economy control variables, are not associated with corresponding increases in standard deviations and are robust to small sample and spurious regression biases. Our findings thus document some stylized facts on the relation between election outcomes and financial market returns in Canada. Our investigation is principally motivated by claims on these issues anecdotally offered by politicians and political commentators, as well as by a growing empirical literature on the subject in the United States. While we discuss some possible hypotheses for our results, we leave open numerous issues for future research. For example, we do not know the mechanism through which political variables impact financial market returns. Tracing the effect of economic policies on returns would necessitate data to measure shocks to fiscal, monetary and regulatory policies that are unexpected by market participants, a significant undertaking. The policy timing hypothesis in the spirit of Tufte (1978) would then imply more stimulating shocks in the late part of the mandates. Another fundamental question is whether political variables cause fluctuations in returns or whether it is the other way around. This article has implicitly assumed that election outcomes are exogenous events, an assumption that is given some support by our findings that election outcomes do not proxy for the expected state of the economy. However, economic variables might be helpful in predicting election outcomes. Endogenizing the political cycle is a complicated problem that deserves further attention. Future work on these issues would be welcomed from both a theoretical and empirical perspectives.