¹ Clarke, Harold D., Jenson, Jane, LeDuc, Lawrence and Pammett, Jon H., Absent Mandate: Interpreting Change in Canadian Elections (Toronto: Gage, 1991), p. 72Google Scholar. However, the role of retrospective assessments and incumbency voting is disputed. See Archer, Keith and Johnson, Marquis, ‘Inflation, Unemployment and Canadian Federal Voting Behaviour’, Canadian Journal of Political Science, 21 (1988), 569–84CrossRefGoogle Scholar; Happy, J. R., ‘Economic Performance and Retrospective Voting in Canadian Federal Elections’, Canadian Journal of Political Science, 22 (1989), 377–87CrossRefGoogle Scholar; Carmichael, Calum M., ‘Economic Conditions and the Popularity of the Incumbent Party in Canada’, Canadian Journal of Political Science, 23 (1990), 713–26.CrossRefGoogle Scholar

² Archer, and Johnson, , ‘Inflation, Unemployment’, p. 577.Google Scholar

³ Based on open-ended responses to the ‘most important campaign issue’ question in national election studies. See Clarke, et al. , Absent Mandate, p. 70.Google Scholar

⁴ A recent excellent review of the field of economics and voting that illustrates this point nicely, since taxation is not considered in it, is Lewis-Beck, Michael S., Economics and Elections: The Major Western Democracies (Ann Arbor: University of Michigan Press, 1988).Google Scholar

⁵ Attention has focused less on adding new variables to the model than on accounting for the effect of the standard variables, income, inflation and unemployment, in particular whether sociotropic or egocentric orientations dominate in the voter's calculus. See Kiewiet, D. Roderick and Rivers, Douglas, ‘A Retrospective on Retrospective Voting’, Political Behavior, 6 (1984), 369–93.CrossRefGoogle Scholar

⁶ Niskanen, William A., ‘Economic and Fiscal Effects on the Popular Vote for the President’, in Rae, Douglas W. and Eismeier, Theodore J., eds., Public Policy and Public Choice (London: Sage Publications, 1979), pp. 93–120.Google Scholar

⁷ Kiewiet, D. Roderick, Macroeconomics and Micropolitics: The Electoral Effects of Economic Issues (London: University of Chicago Press, 1983).Google Scholar

⁸ Pissarides, Christopher A., ‘British Government Popularity and Economic Performance’, Economic Journal, 90 (1980), 569–81.CrossRefGoogle Scholar

⁹ Niskanen, , ‘Economic and Fiscal Effects’, p. 111.Google Scholar

¹⁰ Kiewiet, , Macroeconomics and Micropolitics, pp. 64–5.Google Scholar

¹¹ Kiewiet, , Macroeconomics and Micropolitics, p. 97Google Scholar

¹² Kiewiet, , Macroeconomics and Micropolitics, pp. 104–6.Google Scholar

¹³ Kiewiet, , Macroeconomics and Micropolitics, p. 55.Google Scholar

¹⁴ As well as fluctuations in the exchange rate, a variable not usually included in such analyses. This variable plays no part in the current analysis and therefore will not be considered here.

¹⁵ Pissarides, , ‘British Government Popularity’, p. 572.Google Scholar

¹⁶ Clarke, et al. , Absent Mandate, p. 70.Google Scholar

¹⁷ Johnston, Richard, Public Opinion and Public Policy in Canada (Toronto: University of Toronto Press, 1986), pp. 61–5.Google Scholar

¹⁸ Clarke, et al. , Absent Mandate, p. 152.Google Scholar

¹⁹ ‘The Gallup Report’, Monday, 27 April 1987. See also Clarke, et al. , Absent Mandate, p. 35.Google Scholar

²⁰ Blake, Donald, ‘Division and Cohesion: The Major Parties’, in Perlin, George, ed., Party Democracy in Canada: The Politics of National Party Conventions (Scarborough, Ontario: Prentice-Hall, 1988), p. 47Google Scholar. The public's concern is much less for corporate tax levels, which, continuing a trend begun in the 1950s, are significantly diminishing relative to personal income taxes as a source of government income.

²¹ The data for party choice were obtained from Beck, J. M., Pendulum of Power (Englewood Cliffs, NJ: Prentice-Hall, 1968)Google Scholar for elections to 1968. For the elections from 1972 to 1988 the data were obtained from the Report of the Chief Electoral Officer. The economic data were obtained from the following Statistics Canada sources, with catalogue numbers in parentheses. Data for personal and disposable income were calculated from the National Accounts: Income and Expenditure, selected years, Cansim Matrix 000581 (13–201). The inflation series was calculated from the major cities price index. To 1975, it was computed from Historical Statistics of Canada (2nd ed., 1983, CS11–516E); for the period from 1975 to 1979 from the Consumer Price Index (62–001); and for the period from 1980 to 1988 from Cansim matrixes 1923, 1925, 1928, 1930, 1932, 1933, 1935 and 1937. The provincial-level unemployment data for the period 1953 to 1979 were obtained from tables generated internally by Statistics Canada and are available from the author on request. For the period from 1980 to 1988 the unemployment data were drawn from Cansim matrixes 2065 to 2073 and matrix 2076. With the exception of inflation, which varies predominantly over time, all of the other economic variables vary considerably over time and by province (see Appendix II).

²² Nine provinces are included for the thirteen elections. Prince Edward Island cannot be included in the analysis because inflation and unemployment data are not available for the period covered in the study.

²³ The assumption of independence among the combinations of the explanatory variables specified in the model is well justified. See Appendex III for the correlation matrix of the economic variables.

²⁴ More formally all taxes that reduce income at its source. See, Statistics Canada, National Income and Expenditure Accounts, Vol. 3, ‘A Guide to the National Income and Expenditure Accounts (Definitions–Concepts–Sources–Methods)’ (Ottawa: Department of Supply and Services, 1975), p. 173.

²⁵ In which the weights are derived from an estimate of the autoregression parameter for each of the cross-sectional units. See Kmenta, Jan, Elements of Econometrics, 2nd edn (London: Collier Macmillan, 1986), pp. 622–5.Google Scholar

²⁶ In contrast, in policy voting models the future enters the voter's calculus in a more complex fashion. Policy voters extrapolate retrospective information to assess which party will be most capable of dealing with economic problems. As a consequence, the incumbent party will not necessarily be punished for poor performance: it may escape responsibility if it is chosen as the best party to deal with a problem even if the problem arose under its administration. See Lewis-Beck, , Economics and ElectionsGoogle Scholar, for a treatment of these issues.

²⁷ Clarke, et al. , Absent Mandate, pp. 150–6.Google Scholar

²⁸ Niskanen, , ‘Economic and Fiscal Effects’, pp. 106–7.Google Scholar

²⁹ Eismeier, Theodore J., ‘Budget and Ballots: The Political Consequences of Fiscal Choice’, in Rae, and Eismeier, , eds, Public Policy and Public Choice, pp. 121–45.Google Scholar

³⁰ Department of Finance, ‘The Budget Tabled in the House of Commons by the Honourable Michael H. Wilson, Minister of Finance’, 20 02 1990, Table V, p. 142.Google Scholar

³¹ Alt, James and Chrystal, K. Alec, Political Economics (Berkeley: University of California Press, 1983), pp. 161–2.Google Scholar

³² In previous analysis these effects did not achieve statistical significance. See Happy, J. R., ‘Voter Sensitivity to Economic Conditions: A Canadian–American Comparison’, Comparative Politics, 19 (1986), 45–56CrossRefGoogle Scholar, Table 1, p. 49.

³³ Archer, and Johnson, , ‘Inflation, Unemployment’, p. 583.Google Scholar

³⁴ Kiewiet, and Rivers, , ‘A Retrospective’, p. 375.Google Scholar

³⁵ Goodman, Saul and Kramer, Gerald H., ‘Comment on Arcellus and Meltzer, “The Effect of Aggregate Economic Conditions on Congressional Elections”’, American Political Science Review, 69 (1975), 1255–65CrossRefGoogle Scholar, Table 1, p. 1263. Note that they measure change over the last year of the inter-election period only, rather than the entire period as used here. On this difference, on the superiority of the full inter-election specification for the Canadian case and for a comparison with estimates in Kramer's original model, see Happy, J. R., ‘Voter Sensitivity’.Google Scholar

³⁶ Niskanen, , ‘Economic and Fiscal Effects’, p. 111Google Scholar. Note that he uses net national product and total tax revenue as variables. Disposable income and personal direct taxes are the comparable variables used in this study.

³⁷ Pissarides uses vote intention as his dependent variable, and rate of consumption growth rather than disposable income, and the ratio of taxes to gross domestic product rather than change in tax payments, as explanatory variables. He also lags unemployment by two quarters and enters it as a non-linear relationship. See Pissarides, , ‘British Government Popularity’, pp. 571–2.Google Scholar

³⁸ Reported for Canada in Equation 5, Table 1, and for Britain in Pissarides, , ‘British Government Popularity’, p. 575.Google Scholar

³⁹ A result that Pissarides concludes is ‘surprising, given the attention focused on inflation and unemployment in previous studies’ (see ‘British Government Popularity’, p. 575). The conclusion of the current study reflects this same spirit. Economic variables beyond macroeconomic performance, in particular those such as taxation and expenditure, which are a direct government responsibility, should play a much larger role in the analysis of economic voting than is currently the case.

⁴⁰ Archer, and Johnson, , ‘Inflation and Unemployment’, pp. 572–5.Google Scholar

⁴¹ Clarke, et al. , Absent Mandate, p. 114.Google Scholar

⁴² Johnston, Richard, Public Opinion and Public Policy in Canada, pp. 208–9.Google Scholar