¹ In an election-day exit poll conducted by Voter Research and Surveys for several major news networks, 42 per cent cited ‘the economy’ as the ‘most important issue’ prompting their vote and 21 per cent cited the federal deficit. Other issues (two issues were recorded) mentioned by at least 10 per cent included health care (20 per cent) family values (15 per cent), taxes (14 per cent), and education (13 per cent). See The Public Perspective, Vol. 4, No. 2 (Storrs: University of Connecticut, The Roper Center for Public Opinion Research, 1993), p. 93.Google Scholar

² For literature reviews, see Clarke, Harold D., Elliott, Euel W., Mishler, William, Stewart, Marianne C., Whiteley, Paul F. and Zuk, Gary, Controversies in Political Economy (Boulder, Colo.: Westview Press, 1993)Google Scholar; Lewis-Beck, Michael S., Economics and Elections: The Major Western Democracies (Ann Arbor: University of Michigan Press, 1988)Google Scholar; Norpoth, Helmut, Lewis-Beck, Michael S. and Lafay, Jean-Dominique, Economics and Politics: The Calculus of Support (Ann Arbor: University of Michigan Press, 1991).Google Scholar

³ MacKuen, Michael, Erikson, Robert and Stimson, James A., ‘Peasants or Bankers? The American Electorate and the US Economy’, American Political Science Review, 86 (1992), 597–611.CrossRefGoogle Scholar

⁴ Brody, Richard A., Assessing the President: The Media, Elite Opinion, and Public Support (Palo Alto, Calif.: Stanford University Press, 1991)Google Scholar; Iyengar, Shanto and Kinder, Donald R., News That Matters: Television and American Opinion (Chicago: University of Chicago Press, 1987).Google Scholar

⁵ Key, V. O. Jr, The Responsible Electorate: Rationality and Presidential Voting, 1936–1960 (New York: Vintage Books, 1968).Google Scholar

⁶ See Monroe, Kristen R. and Erickson, Lynda, ‘The Economy and Political Support: The Canadian Case’, Journal of Politics, 48 (1986), 616–47, p. 619.CrossRefGoogle Scholar

⁷ Bloom, Howard S. and Price, H. Douglas, ‘Voter Response to Short-Run Economic Conditions: The Asymmetric Effects of Prosperity and Recession’, American Political Science Review, 69 (1975), 1240–54.CrossRefGoogle Scholar

⁸ See Clarke, Harold D. and Elliott, Euel W., ‘New Models of Presidential Approval: Carter, Reagan and their Predecessors’, European Journal of Political Economy, 5 (1989), 551–69CrossRefGoogle Scholar; Clarke, Harold D., Elliott, Euel W. and Seldon, Barry F., ‘Key Assumptions Reconsidered: A Utility Function Analysis of Competing Models of Party Support’, Journal of Theoretical Politics, 6 (1994), forthcoming.CrossRefGoogle Scholar

⁹ Previous research indicates that such perceptions have an empirical basis since, historically, levels of unemployment and inflation have been lower under Democratic and Republican administrations, respectively. See, for example, Hibbs, Douglas A., ‘Political Parties and Macroeconomic Policy’, American Political Science Review, 71 (1977), 1467–87CrossRefGoogle Scholar; Hibbs, Douglas A., The American Political Economy: Macroeconomics and Electoral Politics (Cambridge, Mass.: Harvard University Press, 1987), chap. 7Google Scholar; Whiteley, Paul F., ‘Party Incumbency and Economic Growth in the United States, 1929–1988’, Political Behavior, 10 (1988), 1–23.CrossRefGoogle Scholar

¹⁰ For a formal proof of this possibility in the context of a utility-maximization model, see Clarke, , Elliott, and Seldon, , ‘Key Assumptions Reconsidered’.Google Scholar

¹¹ Whiteley, , ‘Party Incumbency and Economic Growth’.Google Scholar

¹² Real GDP (in 1987 $) grew during five of the first six quarters of Bush's presidency (the 1989.Q3 quarterly change was 0.0 per cent). Starting in 1990.Q3, however, the successive quarterly changes were 1.0 per cent, −3.9 per cent and −3.0 per cent. From 1991.Q2 onwards, growth resumed, albeit slowly, with the average quarterly change from that date to 1992.Q3 being 1.6 per cent. These and other economic data are obtained from the Council of Economic Advisors, Economic Indicators (Washington, DC: Government Printing Office, January 1989–November 1992)Google Scholar. Unemployment is the seasonally adjusted percentage of civilian workers unemployed; inflation is the annualized percentage increase in the consumer price index; real income is measured as the monthly average of weekly earnings in 1982 $, and as the annualized percentage change in this average. Unlike the other series, GDP figures are available only quarterly and, hence, cannot be included in our monthly presidential approval model.

¹³ The Gallup question is: ‘Do you approve or disapprove of the way — is handling his job as President?’ For months with multiple polls, Bush's approval score is calculated as the average of these polls. Polls beginning in one month and ending in another are excluded.

¹⁴ Unemployment also became increasingly ‘sticky’ in the last half of Bush's term. Before 1991, the per centage of long-term unemployed (out of work twenty-seven weeks or more) was quite stable and averaged 10.0 per cent. Thereafter, it grew steadily and reached 22.2 per cent in October 1992. The correlation between the per centage of long-term unemployed and the unemployment rate was very strong (r = + 0.89).

¹⁵ One Gallup poll conducted in March 1991 gave Bush an 89 per cent rating – the highest score since Gallup began measuring presidential approval in the 1930s.

¹⁶ The unemployment increases were strongly correlated (r = −0.80) with declining income, as real weekly earnings (1982 $) fell from $265.90 in January 1989 to $255.20 in October 1992.

¹⁷ Lewis-Beck, , Economics and Elections, chap. 8.Google Scholar

¹⁸ See, for example, Kinder, Donald R. and Kiewiet, D. Roderick, ‘Sociotropic Politics: The American Case’, British Journal of Political Science, 11 (1981), 129–61CrossRefGoogle Scholar. See also Lewis-Beck, , Economics and Elections, chaps 2, 10.Google Scholar

¹⁹ Retrospective evaluations of national economic conditions are measured using the question: ‘Would you say that at the present time business conditions are better or worse than they were a year ago?’ Three questions measure prospective assessments of national conditions: (i) ‘Now turning to business conditions in the country as a whole – do you think that during the next 12 months we'll have good times financially, or bad times or what?’ (ii) ‘… a year from now, do you expect that in the country as a whole business conditions will be better, or worse than they are at present, or just about the same?’ (iii) ‘Looking ahead, which would you say is more likely … that in the country as a whole we'll have continuous good times during the next 5 years or so, or that we will have periods of widespread unemployment or depression, or what?’ The question tapping retrospective judgements about personal financial circumstances is: ‘We are interested in how people are getting along financially these days. Would you say that you (and your family living here) are better off or worse off than you were a year ago?’ Prospective evaluations of personal financial circumstances are measured with the question: ‘Now looking ahead, do you think that a year from now you (and your family living here) will be better off financially, or worse off, or just about the same as now?’ (emphasis in the original in all cases). The subjective economic evaluation indices based on responses to these questions range from 0 to 200, constructed as 100 plus the percentage offering a positive assessment minus the percentage offering a negative one. See Surveys of Consumers: Monthly Time Series Data Base (Ann Arbor: University of Michigan Survey Research Center, 1987).Google Scholar

²⁰ See, for example, Mueller, John E., War, Presidents, and Public Opinion (New York: Wiley, 1973)Google Scholar; MacKuen, Michael, ‘Political Drama, Economic Conditions, and the Dynamics of Presidential Popularity’, American Journal of Political Science, 27 (1983), 165–91CrossRefGoogle Scholar; Norpoth, Helmut, ‘Economics, Politics and the Cycle of Presidential Popularity’, in Eulau, Heinz and Lewis-Beck, Michael S., eds. Economic Conditions and Electoral Outcomes: The United States and Western Europe (New York: Agathon Press, 1985)Google Scholar. The inclusion of political events is important for methodological reasons as well. As Hibbs observes: ‘it is not possible to obtain accurate (unbiased and consistent) estimates of the relationship between macroeconomics and electoral politics if variables correlated with the economy that affect political support are omitted from the equations.’ See Hibbs, , The American Political Economy, p. 163Google Scholar. There is no ‘canonical list’ of political events and, thus, the analyst must choose which ones to model, taking into account limitations imposed by degrees of freedom.

²¹ The inclusion of the Tiananmen Square demonstrations and the attempted coup in the former Soviet Union as rally events is prompted by the assumption that the public would perceive that serious political turmoil in these major Communist regimes had the potential to threaten American security.

²² Although it is reasonable to hypothesize that unemployment, inflation and income affect presidential approval contemporaneously or at short lags, it is difficult to hypothesize a priori which specification is preferable. Specifying a one-month lag is consistent with the idea that media reporting of economic statistics for the previous month is an aspect of the process by which economic conditions affect support for incumbent parties and their leaders. See, e.g., Brody, , Assessing the President, chap. 6Google Scholar; Mosley, Paul, ‘“Popularity Functions” and the Role of the Media: A Pilot Study of the Popular Press’, British Journal of Political Science, 14 (1984), 117–29CrossRefGoogle Scholar; Sanders, David, Marsh, David and Ward, Hugh, ‘The Electoral Impact of Press Coverage of the British Economy, 1979–87’, British Journal of Political Science, 23 (1993), 175–210CrossRefGoogle Scholar. Specifying alternative lags (0,2,3 months) shows that the strongest effects occur at lag 1. Also, cross-correlations of the Bush approval residuals from the model using the one-month lag for unemployment, inflation and income with the latter three series (in differenced form) show that correlations at the other lags are insignificant.

²³ Following MacKuen, et al. ‘Peasants or Bankers’, pp. 600–1Google Scholar, we assume that subjective economic evaluations are exogenous to presidential approval.

²⁴ Box, George E. P. and Jenkins, Gwilym, Time Series Analysis: Forecasting and Control (San Francisco: Holden-Day, 1976)Google Scholar; Box, George E. P. and Tiao, G. C., ‘Intervention Analysis with Applications to Economic and Environmental Problems’, Journal of the American Statistical Association, 70 (1975), 70–9.CrossRefGoogle Scholar

²⁵ Political events are measured as dummy variables, scored 1 in the month in which an event occurs and 0 in other months.

²⁶ The analyses reported in Table 1 use annualized percentage change in real income. Analyses using monthly real income yield similar results – all variables that are significant in the former analyses are significant in the latter, and all variables that are insignificant continue to be insignificant. Note also that the coefficient for unemployment remains very large in the latter analyses, again indicating its influence on Bush's approval.

²⁷ For stationary processes, δ parameters will vary between 0 and 1, with the size of the parameter indicating the speed at which the adjustment occurs (the larger the δ, the slower the adjustment). For temporary effect interventions occurring at time t, the size of the effect in period t + i is calculated as co δⁱ.

²⁸ The δ coefficient (0.41) for the Panama invasion indicates that its effect had decreased to less than 1 per cent three months later.

²⁹ The index is constructed by averaging the monthly scores for the one-year (better – worse) and five-year measures.

³⁰ Unemployment also remains significant in models using the Surveys of Consumers overall index of consumer sentiment (ICS) – a composite of six items measuring national/personal, retrospective/prospective economic evaluations. Modelling the ICS at 0 to 3 lags shows that it has its strongest effect (ω = 0.13, t= 1.22) at a 0 lag. The unemployment coefficient in this analysis is −10.18, t=−3.91.

³¹ Including subjective economic evaluations reduces the size of the coefficients for Desert Storm somewhat, thereby suggesting that its effect was partially indirect, i.e., the success in the Gulf War bolstered public expectations about the country's economic prospects, and these expectations, in turn, enhanced presidential approval. In this regard, Figure 2 shows that long-run (five-year) expectations moved sharply upwards in February and March 1991 and, although they subsequently eroded, they remained, on average, above the figures for late 1990. An analysis of these expectations is consistent with this observation – controlling for income, inflation and unemployment, Desert Storm was associated with a sizeable permanent increase (ω = 31.7, t = 4.46) in the long-term expectations index. Note also that this increase was partially offset by the effects of income (ω = 3.4, t = 2.37) and, especially, unemployment (ω = −18.9, t = −2.41).

³² See, for example, Granger, Clive W. J. and Newbold, Paul, ‘Spurious Regressions in Econometrics’, Journal of Econometrics, 2 (1974), 111–20CrossRefGoogle Scholar; Hendry, David F., ‘Econometrics: Alchemy or Science’, Economica, 47 (1980), 387–406.CrossRefGoogle Scholar

³³ See, for example, Beck, Nathaniel, ‘Comparing Dynamic Specifications: The Case of Presidential Approval’, in Stimson, James A., ed., Political Analysis, Vol. 3 (Ann Arbor: University of Michigan Press, 1992), pp. 51–88.Google Scholar

³⁴ See Engle, Robert F. and Granger, Clive W. J., ‘Co-integration and Error Correction: Representation, Estimation and Testing’, Econometrica, 55 (1987), 251–76CrossRefGoogle Scholar; Johansen, Soren, ‘Statistical Analysis of Cointegration Vectors’, in Engle, Robert F. and Granger, Clive W. J., eds., Long-Run Economic Relationships: Readings in Cointegration (Oxford: Oxford University Press, 1991), chap. 7.Google Scholar

³⁵ See, for example, Stewart, Jon, Econometrics (Deddington, Oxon: Philip Allan, 1991), p. 204.Google Scholar

³⁶ See, for example, Kennedy, Peter, A Guide to Econometrics, 3rd edn (Oxford: Blackwell, 1993), p. 264.Google Scholar

³⁷ See, for example, Clarke, Harold D. and Suzuki, Motoshi, ‘Partisan Dealignment and the Dynamics of Independence in the American Electorate’, British Journal of Political Science, 24 (1994), 57–78.CrossRefGoogle Scholar