Read an Excerpt
Cambridge University Press
0521833639 - Election Timing - by Alastair Smith
Under the British system almost all elections lost by the prime ministers are ex hypothesi thought to have been held on the wrong date.
Roy Jenkins (1991, p. 367)
In many parliamentary systems the timing of the next election is at the discretion of the current government. This gives leaders in these systems the power to call elections at the most advantageous time for them - when they expect to win. It is claimed that "[t]he choice of election date may well be the most important single decision taken by a British prime minister" (Newton 1993, p. 136). Despite the apparent importance of this decision, political scientists have done little to explain when elections are called and how and why this timing affects electoral outcomes and subsequent economic performance. This book addresses these questions.
Most parliamentary systems specify a maximum time between elections, for example, five years in Britain. Yet leaders are not bound to wait the maximum number of years and may call an election whenever "the time is right." Most extant attempts to explain election timing focus on this idea of "political surfing" (Inoguchi 1979). Governments wait until their popularity and economic conditions suggest that winning would be a sure thing, at which time they call elections. This explanation assumes that the electoral outcome is simply an expression of relative support for the government at the time the election is called. As such, a party's vote share simply reflects the government's performance during its time in office. There is no conception that the timing of an election influences the election's outcome beyond the date's having been chosen when the government looked its best.
Anecdotal evidence suggests the contrary. In May 1970, for the first time in three years, the governing British Labour party overtook the opposition Conservative party in the opinion polls. Harold Wilson, then the Labour prime minister, called a snap election to take advantage of Labour's sudden recovery. Yet Labour's support slumped in the election and the Conservatives won 330 of 630 seats. This reversal of fortune is not an isolated example. In 1997, the decision of France's right-wing president Jacques Chirac to call an early election for the French lower house led to an immediate decline in the right's support and to large electoral gains for the left (Lewis-Beck 2000).
I propose and test an informational theory of endogenous election timing in parliamentary systems. I assume that political leaders, most importantly prime ministers, can make more accurate assessments of future government performance than can the public at large. This informational advantage helps leaders choose election dates when the government looks at its best. In particular, leaders want elections "today" if they expect to perform poorly in the immediate future. Unfortunately, leaders are unable to completely hide the impending decline since their very act of preempting the decline with an election indicates that the government has something to hide. The timing of elections influences electoral outcomes because a leader's decision to call an election says something about the leader's expectations for the future.
Following Britain's success in 1982 in the Falklands War against Argentina, Prime Minister Margaret Thatcher's Conservative government was extremely popular. Given that Thatcher had been elected in 1979, she was not required to call an election until 1984. Yet her enormous popularity following the war might have given her an excellent opportunity to secure another five-year term. Speculation about the possibility of an early election was sufficiently intense that polls were taken on the desirability and likelihood of an early election (see, e.g., Index to International Public Opinion, 1982-1983, p. 353). To illustrate the central arguments of this book, it is worth exploring the decision confronting Prime Minister Thatcher.
Suppose, consistent with the "surfing" hypothesis, that Thatcher's popularity would ensure her victory if she called an election in 1982. Opinion polls certainly supported such an expectation. In June 1982, a Gallup poll of voting planned showed 45% of voters planned to support the Conservatives; only 25% expressed support for the opposition Labour party. Further, 51% of voters approved of the prime minister, but approval for the opposition leader (Michael Foot) was only 14% (Butler and Butler 1994, p. 256). By waiting, Thatcher risked having her popularity undermined by policy failures. However, the extent to which she should have feared this depended upon how well she expected to perform over the coming year. If she believed she had effective solutions to the nation's problems and if she believed that her party had the appropriate policies and was competent to implement these policies, then waiting posed little threat as she could expect to get reelected in the future anyway. Yet if she were less confident about her polices or her ability to effectively implement them, then waiting jeopardized a second term in office, since policy failure would likely undermine her support. In short, the more confident she was about the future, the smaller her incentive to call an early election; the less confident she was, the greater the incentive to cash in on past successes with a snap election.
The timing of elections reveals information about how well incumbents expect to perform in the future. I wait until Chapter 2 to formalize the concepts of performance and competence. However, for the time being, performance should be thought of as the amount of public goods, such as effective economic management, that the government produces. Competence is the government's ability to produce these nonpartisan public goods. The less confident Margaret Thatcher was in her ability to rule well, the greater her incentive to call an early election when she was ahead in the polls. Competent governments wait longer before calling elections. Unfortunately, the above analytic narrative presents only half the picture. The initial supposition was that Thatcher would be reelected if she called an election immediately following the war. However, it is the incompetent, not the competent, leader who wants to take advantage of such an opportunity. What inference should voters draw upon seeing an early election? They should infer that the incumbent doubts his or her ability to continue producing good outcomes while in office. Leaders who call early elections should expect to see their support decline. This waning of support is exactly what happened to Prime Minister Wilson in 1970 and to President Chirac in 1997. The early election is a signal that the leader does not expect conditions to be as rosy in the future. In anticipation of this upcoming decline, the electorate reevaluate their assessment of the government's record.
This signaling argument forms the core of my explanation of election timing. Leaders call elections to censor the public's ability to observe future decline and to cash in on past successes. As with most acts of government censorship, trying to prevent people from being informed tells them that the information is worth having. Therefore, the signal of early, unanticipated elections cause voters to reevaluate their assessment of the government. Elections are more than a transfer of voting-intention opinion polls into vote shares. Voters question why their opinion is being sought at that time.
|The three major questions of this book are|
|1.||When and under what conditions are elections called?|
|2.||What are the electoral consequences of the timing decision?|
|3.||How does the timing decision influence subsequent economic performance?|
As already indicated, I contend that a major determinant of the answer to the first question is a leader's beliefs about future performance. Many other factors are also important, such as the time left before the statutory end of Parliament, the government's popularity, the size of the government's majority, the need for a political mandate to initiate new legislation, and the government's performance to date. Because these factors influencing elections are readily observable and they have predictable effects on the timing of elections, I will analyze them to determine how likely they make parliamentary dissolution and new elections. Although the leader's expectations about future performance are unknown, the timing of elections can signal this information. If, given all the observable factors, the announcement of an election is expected, then calling the election provides little indication of future decline. In such a case, the government's support should remain robust. However, if elections are announced out of the blue - when other factors predict an election to be unlikely - then the timing decision indicates that the leader anticipates a drastic decline in future performance. Voters can use this new information to reassess their evaluation of the government. Support for the government softens, and the government is likely to receive a lower vote share than pre-announcement voting intentions would indicate. These unexpectedly early elections are also likely to precede a decline in economic performance.
Had Margaret Thatcher called an election in the summer of 1982, I anticipate that her popular support would have declined drastically and that the Conservative's share of the popular vote would have been considerably less than the 45% the opinion polls suggested. The theory suggests the announcement of elections in June 1982 would have signaled that the Conservative government had little to offer in the future and did not trust its own ability to sustain economic growth and combat Britain's growing unemployment problem. The calling of elections would have been seen as a blatant attempt to cash in on the government's successful Falklands Islands policy. Thatcher's own words seem to support this conclusion. In the spring of 1983, speculation about elections was rife. In a raucous clash in the House of Commons, Denis Healey, deputy leader of the Labour party, accused Thatcher of cutting and running. As the Guardian reports, "The effect on Mrs Thatcher was awesome to witness. Allowing scarcely half a seconds for contemplation, she began to bellow back at Mr. Healey. 'Oh - the Rt. Hon. Gentleman is afraid of an election is he? Afraid afraid afraid. Frightened, couldn't take it, couldn't stand it. . . . Oh no, if I was going to cut and run I would have gone after the Falklands.'"1
Prior to the Falklands War, Thatcher's performance in office was less than stellar. Unemployment had climbed from 5% in June 1979 to over 11% in April 1982, and economic growth was on average negative. The only bright spot was the reduction of inflation, which had peaked at over 20% in the middle of 1980. By 1982 inflation appeared contained and falling. For example, in May 1982 the Retail Price Index was at 9.5%. Within a year it would fall to 3.7%. Since Thatcher's electoral platform had been to control inflation and to free up the economy through microlevel reforms, an initial economic decline was perhaps to be expected. Yet had an early election been called in 1982, signaling that the pain of economic dislocations was for naught and an economic revival was not imminent, the Conservatives might well have lost, despite opinion polls to the contrary.
Since going to the people early indicates that leaders lack confidence in their future performance, it is reasonable to ask why any leader ever calls an early election. Indeed, Margaret Thatcher did not wait until May 9, 1984 - the end of her statutory five-year term - but instead announced on May 9, 1983, elections for June 9.
Consistent with expectations, the elections preceded a perceived economic downturn. In their autobiographies, both Thatcher and her future chancellor, Nigel Lawson, mention their fear of increasing inflation. "It was pointed out that the main economic indicators would look slightly better then than in the autumn because inflation was due to rise slightly in the second half of the year" (Thatcher 1993, p. 288; see also Lawson 1992, p. 246). Their concerns were correct. Inflation did rise from its election level of 3.7% to over 5% in the fall of 1983 and the first half of 1984.
By calling the election in June 1983, the Conservative party prevented the electorate from observing this worsening of inflation, which presumably would have resulted in a decline in the party's popularity. However, if as I propose, the signal of an early election reveals that the future will not be so rosy, then the very act of calling the election conveys the information that the government is trying to conceal. This relationship is borne out in public-opinion data. In May 1983, prior to the election announcement, Gallup reported a voting intention of 49% for the Conservatives. Yet in June's general election they received only 42.2% of the vote.2 While the margin of error in the opinion data probably accounts for some of this difference, it is clear that elections are more than a direct translation of popularity into vote share.
The objective of politicians is not to maximize vote share but to remain in power. Despite their decline in popularity, the Conservatives won 397 of 650 seats. The opposition was split between their traditional opponent, the Labour party, which obtained 209 seats with a vote share 27.6%, and an alliance of the Liberals and Social Democrats, which obtained 23 seats with vote share of 25.4%. The Conservative victory, the largest since 1945, was the result not of overwhelming Tory popularity but of the Tories having faced a divided and demoralized opposition. According to Lawson, "Labour was in such a mess with an unelectable leader, left wing policies which the country would never stomach, and suffering badly from the Social Democrats defection . . . " (1992, p. 246). In fact, Lawson goes on to state that at that time he thought Labour was in such a poor position that the Conservatives could have won at anytime. But with hindsight, he also admits that a "bird in the hand" is a powerful argument for an election.
Given the first-past-the-post, plurality electoral system in Britain and its large divisions, the opposition had little hope of unseating the Tories (Duverger 1963; Lijphart 1994; Rae 1967; Riker 1982). However, had the opposition overcome their differences and presented a unified opposition, the Tories' reign would have looked much more assailable. Had the 1979 Parliament continued toward its statutory termination (May 9, 1984), the impetus of the impending election might have enabled the opposition to present a unified front. But Thatcher forestalled such a development by going to the polls before the opposition could reorganize.
A pervasive feature of the British political system is the shortness of election campaigns. For example, in 1983, Thatcher announced the election on May 9, Parliament dissolved on May 13, and the general election was held on June 9. The opposition had only one month in which to adopt a policy platform, prepare a manifesto, find candidates for each seat, and organize a campaign. Given such time pressures, the opposition needs to be ready. But parties have only limited resources with which to prepare for office. If they use them immediately following one election, then by the time a new election is called their manifesto appears dated. If they save all their resources until the election is called, then they risk having insufficient time to prepare. This dilemma between husbanding resources and being prepared becomes easier to resolve as the statutory five-year limit approaches, since an election becomes inevitable. Yet early in the electoral cycle, the opposition's preparedness is lacking, suggesting a relationship between the timing of elections and the opposition's subsequent performance. All else being equal, early elections are fought between incompetent incumbents and ill-prepared challengers.
Election timing is only one aspect of the more general theoretical problem of government turnover. Indeed, in the political science literature, comparativists have spent much energy examining the fall of governments. The topic of this book is the announcement of elections, which, as I shall argue, is the important political event in majoritarian parliamentary systems. In more general settings, scholars have studied the survival of individual leaders in office (Bueno de Mesquita and Siverson 1995), the fate of political leaders (Goemans 1995, 2000), and the survival of coalition governments and the breakup and reshuffling of cabinets (Diermeier and Stevenson 1999; Bienen and van de Walle 1992; Browne et al. 1986; Diermeier and Merlo 2000; Grofman and van Roozendaal 1994; King et al. 1990; Strom 1988; Warwick 1992, 1995). In many political settings, these topics make more sense than the study of election timing. For instance, in autocratic polities, elections are typically rigged, if they occur at all. In proportional representation (PR) parliamentary systems, coalition dynamics make and break governments. (Italy is notorious for having averaged nearly a government per year in the postwar period.) Of course, elections still play an important role in the making and breaking of governments in PR systems, but they are not the exclusive mechanism or even the most common. Strom and Swindle (2002) find that as the extent of cabinet and parliamentary participation in the dissolution decision increases, early dissolution becomes less likely.
I believe that the incentives to time elections that I shall describe in this book also exist in multiparty PR systems. However, the multiple paths to dissolution and the ability of numerous actors to bring about government collapse make the study of election timing in those systems more complex. Lupia and Strom (1995), for instance, explain legislative dissolution in terms of coalition dynamics. They suppose that an exogenous event changes the electoral prospects for different parties and examine the consequences for internal coalition arrangement. Their results suggest that for shocks to lead to dissolution, a majority must prefer an election to a continuance of the existing government and the governing coalition members cannot form new coalitional arrangements they prefer to an election. Under all other circumstances, shocks or critical events result in either the maintenance of the status quo or the reorganization of the governing coalition. While their model seeks to examine how the threat of dissolution shapes coalition bargaining, the underlying electoral dynamics are independent of the political process. This is to say, in their model, each party's expected returns from an election are unaffected by how the election itself comes about. In contrast, the model I present, although not addressing the problem of coalition dynamics, examines how and why the circumstances surrounding legislative dissolution do affect the electoral outcome.
I focus primarily on majoritarian parliamentary systems and, in particular, on Britain. The political events referred to are from Britain unless stated otherwise. Majoritarian systems are typically characterized by single-membered electoral districts, with first-past-the-post (plurality) voting. Such institutional arrangements promote two-party competition, a result often referred to as Duverger's rule (Duverger 1963; Riker 1982). Given the strength of parties in such systems, elections devolve into a contest over which party will gain the majority of seats and form the next government. In this context, the party in office remains in office until either it loses its majority and is forced from office or the incumbent prime minister believes his or her party has its best chance of winning and so calls for new elections. Elections dominate the survival of governments in majoritarian systems, which makes the timing of elections a key weapon in the armory of the incumbent.
© Cambridge University Press