¹ The seminal work in this literature is Nordhaus, William D., ‘The Political Business Cycle’, Review of Economic Studies, 62 (1975), 169–90.CrossRefGoogle Scholar Most scholars, though, trace the political business cycle theory to Schumpeter, Joseph, Business Cycles: A Theoretical, Historical, and Statistical Analysis (New York: McGraw Hill, 1939)Google Scholar and Kalecki, M., ‘Political Aspects of Full Employment’, Political Quarterly, 14 (1943), 322–31.CrossRefGoogle Scholar

² Nordhaus, , ‘The Political Business Cycle’Google Scholar; McRae, Duncan C., ‘A Political Model of the Business Cycle’, Journal of Political Economy, 85 (1977), 239–63CrossRefGoogle Scholar; Tufte, Edward R., Political Control of the Economy (Princeton, NJ: Princeton University Press, 1978).Google Scholar

³ For a review of the empirical literature and a comprehensive test, see Alesina, Alberto, Cohen, Gerald D. and Roubini, Nouriel, Macroeconomic Policy and Elections in OECD Democracies, NBER Working Paper, no. 3830 (Cambridge, Mass.: National Bureau of Economic Research, 1991).CrossRefGoogle Scholar

⁴ Key works in this tradition include Cukierman, Alex and Meltzer, Allan H., ‘A Positive Theory of Discretionary Policy, the Cost of Government, and the Benefits of a Constitution’, Economic Inquiry, 24 (1986), 367–88CrossRefGoogle Scholar, and Rogoff, Kenneth and Sibert, Anne, ‘Equilibrium Political Budget Cycles’, Review of Economic Studies, 55 (1988), 1–16.CrossRefGoogle Scholar

⁵ Alesina, , Cohen, , and Roubini, , Macroeconomic Policy and Elections in OECD Democracies.Google Scholar

⁶ Golden, David and Poterba, James, ‘The Price of Popularity: The Political Business Cycle Re-examined’, American Journal of Political Science, 24 (1980), 696–714.CrossRefGoogle ScholarAlt, James E. and Chrystal, Alec, Political Economics (Berkeley: University of California Press, 1983).Google Scholar

⁷ Alt, and Chrystal, , Political Economics, p. 104.Google Scholar

⁸ Hibbs, Douglas A. Jr., ‘Political Parties and Macroeconomic Policy’, American Political Science Review, 71 (1977), 1467–87CrossRefGoogle Scholar; Alesina, Alberto, ‘Macroeconomic Policy in a Two-Party System as a Repeated Game’, Quarterly Journal of Economics, 102 (1987), 651–78.CrossRefGoogle Scholar

⁹ See Kerry Schott's comments appended to Chrystal, Alec and Alt, James E., ‘Public Sector Behaviour: The Status of the Political Business Cycle’, in Currie, D., Nobay, R. and Peal, D., eds, Macroeconomic Analysis (London: Croom Helm, 1981), 353–82, p. 379Google Scholar; Alesina, Alberto, ‘Macroeconomics and Politics’, in Fischer, Stanley, ed., NBER Macroeconomic Annual 1988 (Cambridge, Mass.: MIT Press, 1988), pp. 38–45.Google Scholar

¹⁰ For a discussion of the impact of rational expectations on the political business cycle theory, see Nordhaus, William D., Alternative Approaches to the Political Business Cycle, Brookings Papers on Economic Activity, no. 2 (Washington, DC: The Brookings Institution, 1989), pp. 1–68Google Scholar; and Alesina, , ‘Macroeconomics and Politics’.Google Scholar

¹¹ See Alesina, , Cohen, and Roubini, , Macroeconomic Policy and Elections in OECD Democracies, and the citations contained therein.Google Scholar

¹² See Alesina's comments appended to Nordhaus, , Alternative Approaches to the Political Business Cycle, p. 53.Google Scholar

¹³ Fiorina, Morris, ‘Short- and Long-term Effects of Economic Conditions on Individual Voting Decisions,’ in Hibbs, Douglas A. Jr., and Fassbender, Heino, eds, Contemporary Political Economy: Studies on the Interdependence of Politics and Economics (Amsterdam: North-Holland Publishing Company, 1981), p. 99.Google Scholar

¹⁴ Friedman, Milton, ‘The Role of Monetary Policy’, American Economic Review, 58 (1968), 1–17.Google Scholar

¹⁵ Nordhaus, , ‘The Political Business Cycle’, pp. 184–5Google Scholar; Tufte, , Political Control of the Economy, p. 27.Google Scholar

¹⁶ Bernanke, B. S., ‘Irreversibility, Uncertainty, and Cyclical Investment’, Quarterly Journal of Economics, 98 (1983), 85–106CrossRefGoogle Scholar; Cukierman, Alex, ‘The Effects of Uncertainty on Investment under Risk Neutrality’, Journal of Political Economy, 88 (1980), 462–75.CrossRefGoogle Scholar

¹⁷ Frey, Bruno S. and Ramser, Hans-Jürgen, ‘The Political Business Cycle: A Comment’, Review of Economic Studies, 43 (1976), 553–5.CrossRefGoogle Scholar

¹⁸ Most governments probably possess better indicators of their re-election chances than simply the aggregate poll numbers. These would include district-by-district poll numbers, and especially the party's standing in marginal districts. After all, the aggregate figures are only reliable to the extent that there is little regional variation in party support. Nevertheless, in the absence of more detailed polling figures, the national aggregate is the best indicator we have.

¹⁹ It is important to emphasize that this inverse relationship between policy manipulation and the incumbent's lead in the polls is only predicted to hold in the period immediately preceding the election. As noted above, the model presented here depicts a purely ‘opportunistic’ cycle in which politicians care only about how they fare at election time. This aspect distinguishes the model from so-called ‘popularity-maintenance’ models which predict that governments formulating economic policy always respond to the incumbent's popularity, whether an election is near or not. See, for example, Golden, and Poterba, , ‘The Price of Popularity’.Google Scholar

²⁰ Frey, Bruno S. and Schneider, Friedrich, ‘An Empirical Study of Politico-Economic Interaction in the United States’, Review of Economics and Statistics, 60 (1978), 174–83CrossRefGoogle Scholar; Frey, Bruno S. and Schneider, Friedrich, ‘A Politico-Economic Model of the United Kingdom’, Economic Journal, 88 (1978), 243–53.CrossRefGoogle Scholar

²¹ Golden and Poterba, for example, criticize Frey and Schneider's idea that governments only manipulate the economy when they are low in the polls. They suggest that risk-averse, vote-maximizing politicians would never forgo policy manipulations that improve their chances of victory, regardless of their current standing. The present analysis, however, suggests that Golden and Poterba's critique overlooks the costliness of this kind of behaviour. See Golden, and Poterba, , ‘The Price of Popularity’, p. 709.Google Scholar

²² Indeed, the vast empirical literature on political business cycles includes only two major articles that include poll figures in their regression models (not including Frey and Schneider's original articles). These are Golden, and Poterba, , ‘The Price of Popularity’Google Scholar, and Keil, Manfred W., ‘Is the Political Business Cycle Really Dead?’, Southern Economic Journal, 55 (1988), 86–99.CrossRefGoogle Scholar In these models, as in Frey and Schneider's, governments are thought to respond to their standing in the polls at all points in the electoral period. The present analysis differs from these in that it only predicts policy manipulations in the immediate run-up to an election.

²³ Friedrich, Robert J., ‘In Defense of Multiplicative Terms in Multiple Regression Equations’, American Journal of Political Science, 26 (1982), 797–833.CrossRefGoogle Scholar

²⁴ This is not to deny that there has been some significant third-party activity in Great Britain, especially since the 1980s. Fortunately, third-party effects are mitigated by measuring the government's security as its lead in the polls over the major opposition party. The other alternative – measuring security simply as the percentage of voters who say they intend to vote for the governing party – is easily confounded by a strong third-party showing. After all, a 50 per cent support level can either mean a very close race – if the remaining 50 per cent is committed to one party – or a potential landslide – if the remaining 50 per cent is split evenly between two parties. Only by looking at the differential between the governing party's support level and that of the major opposition party can we distinguish these two cases. In the first case, the lead will be close to zero, while in the second, the lead will be 25 percentage points.

²⁵ The effect of coalition governments is a topic worthy of further research. The general logic laid out in this article should still be valid: that is, the magnitude of politically-motivated policy cycles should still vary with the incumbents' political needs. However, how these needs are defined is more problematic. When looking at the polls, for example, should we consider the entire coalition, the main party within the coalition, or all members of the coalition individually? As a first cut, we would probably want to analyse the coalition as a whole, positing that the partners co-operate in order to ensure their collective re-election. This approach, however, will be confounded if such co-operation does not take place and each party looks out for itself. In that case, it will be important to identify which parties control which aspects of economic policy and determine whether they can employ those instruments in such a way as to improve their individual re-election chances, such as by targeting crucial constituencies.

²⁶ Mosely, Paul, The Making of Economic Policy: Theory and Evidence from Britain and the United States Since 1945 (Brighton, Sussex: Wheatsheaf, 1984), chap. 3.Google Scholar

²⁷ In one case, the 1979 election, the government was forced to call an election when it lost a vote of no confidence. Because the government lost control of election timing, it may not have had a chance to prime the economy sufficiently before the election. As will be seen in Figures 1 and 2, there was a modest pre-election boost in 1979, but not one of the magnitude that might be expected given the governing party's dismal poll ratings. Unfortunately, we cannot know when the government would have called the election had it retained control. On the one hand, if it had waited until October 1979, the last possible date for the election, the pre-election quarter would have come later. On the other hand, there is some speculation that then-Prime Minister Callaghan was planning to call the election earlier, meaning that the no-confidence vote only moved the election date up by a few weeks. In that case, the pre-election quarter would have been unchanged. I have decided not to treat this election as a special case because of this ambiguity and because it does not appear to influence the results greatly. For a discussion of election timing in 1979, see Leonard, Dick, ‘The Election Campaign’, in Penniman, Howard R., ed., Britain at the Polls, 1979 (Washington, DC: American Enterprise Institute for Public Policy Research, 1981), pp. 95–105.Google Scholar

²⁸ Tufte, , Political Control of the Economy, p. 29.Google Scholar

²⁹ The classic work here is Kramer, Gerald H., ‘Short-Term Fluctuations in US Voting Behavior, 1896–1964’, American Political Science Review, 66 (1973), 131–43.Google Scholar For findings on the relationship between real income and popularity, see Hibbs, Douglas A. Jr., Rivers, R. Douglas and Vasilatos, Nicholas, ‘On the Demand for Economic Outcomes: Macroeconomic Performance and Mass Political Support in the United States, Great Britain, and Germany’, Journal of Politics, 42 (1982), 426–62.CrossRefGoogle Scholar

³⁰ Because transfers can be targeted to specific constituencies, there may be systematic partisan differences in the kinds of benefits each party manipulates. The Labour party, for example, is more likely to manipulate unemployment benefits since they go directly to its main support groups. Conservative governments, on the other hand, might choose to increase pensions or child benefits. The actual pattern depends, of course, on where the governing party thinks the money would be best spent. While a party might often want to funnel benefits to its core supporters, there may also be times when the core constituency can be counted on and the money can be more effectively spent on swing voters. Hence, it is difficult to know a priori which specific types of transfer payments the government will decide to boost. It is unlikely, though, that an increase in one kind of transfers will be offset by a decrease in others, given the political costs of such a move. For that reason, any politically-motivated increase in one type of benefits should show up in the aggregate numbers. I am grateful to James Alt for raising this issue.

³¹ Golden, and Poterba, , ‘The Price of Popularity’, pp. 697–704.Google Scholar

³² Price, Simon and Sanders, David, ‘Modeling Government Popularity in Postwar Britain: A Methodological Example’, American Journal of Political Science, 37 (1993), 317–34, pp. 330–2.CrossRefGoogle Scholar

³³ Two prior studies of British transfer payments are worth mentioning. As discussed above, Frey and Schneider test a model in which governments base economic policy decisions on their ‘popularity deficit’, or the difference between their lead in the polls and some desired lead in the polls, which increases as the election approaches. When governments have a deficit, they increase transfer payments to bolster their standing; when they have a surplus, they pursue partisan goals – i.e., Labour increases transfers, and Conservatives cut them. See Frey, and Schneider, , ‘A Politico-Economic Model of the United Kingdom’.Google Scholar These findings were criticized by Chrystal and Alt, who tested various models but found no evidence of a politically-motivated cycle in transfer payments. While Chrystal and Alt tried a modified version of Frey and Schneider's model, none of their models directly tested the hypothesis advanced in this article. See Chrystal, and Alt, , ‘Public Sector Behaviour’.Google Scholar

³⁴ All economic data came from Central Statistical Office, Economic Trends: Annual Supplement 1993 (London: HMSO, 1993)Google Scholar and earlier editions. The level of real transfers was obtained by deflating nominal figures by the retail price index. Poll data is based on the Gallup voting intention poll, with the government's lead equalling the difference between it and the main opposition party (Labour or Conservative). These figures came from Craig, F. W. S., British Electoral Facts 1832–1987 (Aldershot, Hants: Parliamentary Research Services, 1989)Google Scholar and were updated through 1992 by the Gallup Political Index. As Gallup poll data is reported monthly, the quarterly figures used here are three-month averages.

³⁵ The analysis focuses exclusively on the quarter before each election. This quarter was selected based on the assumption that any manipulation of transfer payments would have to take place close enough to the election so that the resultant inflation would not be felt until afterwards. The election quarter itself was rejected due to the fact that a number of elections took place in the first several weeks of the quarter, so a significant portion of that quarter's transfer payments would have been paid out after the election.

³⁶ Mosely, , The Making of Economic Policy, pp. 76–7Google Scholar; Alt, and Chrystal, , Political Economics, p. 222.Google Scholar

³⁷ Brittan, Samuel, Steering the Economy (London: Secker and Warburg, 1969), pp. 265–7Google Scholar; Mosely, , The Making of Economic Policy, pp. 95–7.Google Scholar Mosely points out, though, that the balance of payments was less important during the period of floating exchange rates.

³⁸ Alt, and Chrystal, , Political Economics, chaps. 9–10.Google Scholar

³⁹ The current account balance, real GDP and the unemployment rate all entered the regressions lagged by one quarter.

⁴⁰ Friedrich, , ‘In Defense of Multiplicative Terms in Multiple Regression Equations’.Google Scholar

⁴¹ When using interaction terms, there is always a danger of strong collinearity between the interaction term and the variables being interacted. In this case, the correlation between the interaction term and both the pre-election dummy (−0.13) and the poll lead (0.24) are small enough to be of no concern.

⁴² Wallis, K., ‘Lagged Dependent Variables and Serially Correlated Errors: A Reappraisal of Three-Path Least Squares’, Review of Economics and Statistics, 49 (1967), 555–67.CrossRefGoogle Scholar

⁴³ For purposes of comparison, the October 1974 dummy was added to the first model as well. Given that transfers were increased markedly in the quarter before that election, it should come as little surprise that this addition served to decrease the coefficient on the pre-election dummy variable. The coefficient on the pre-election dummy fell to 51.69 with a t-statistic of 0.31. The adjusted R ₂ for this model was 0.993.

⁴⁴ The values of the poll lead in this table were selected roughly to reflect the range actually observed in pre-election quarters. The observed value of the pre-election poll lead ranges from − 14 per cent to 11.5 per cent.