CYCLES AND FORECASTING

“Have cycle, will forecast,” would be a fitting advertisement in the business of election forecasting (Jones Reference Jones2002, Stegmaier and Norpoth Reference Stegmaier, Norpoth and Valelly2013). Knowing the parameters of the cyclical dynamic lets you make a forecast without having to acquire any information on other variables such as the economy, foreign policy, or presidential approval. Hence with a cycle in hand you can also make a forecast far in advance, before data on those variables are available (Norpoth Reference Norpoth1995). Cyclical forecasts are unconditional: they are not qualified, depending on the values of the predictors, which may not be known for certain until after the event that is being forecast.

For a cyclical forecast, all you have to ascertain is how long the cycle runs (periodicity), and how it moves in between (amplitude), both derived from the past behavior of the time series. A trigonometric function would be the perfect tool to model a cycle. But that is too deterministic to be of much help for a phenomenon like the vote that is shaped by human behavior. In fact, skeptics may wonder if it is not a foolish idea to search for an electoral cycle that holds any promise for forecasting. Do past elections indicate the existence of a cycle? Does it occur with any regularity? Does the movement from beginning to end of a cycle follow a predictable path? What reasons do we have to expect that elections follow a cyclical pattern?

A FIRST GLIMPSE AT CYCLES

To keep things simple, we begin with a look in table 1 at how the White House party has fared in successive elections during the last half century. The pattern is quite clear. When the White House party is in its first term, it has won reelection six out of seven times since 1960, with an average share of 54.7% of the two-party vote. Compare that to instances in which the White House party went for reelection after two or more terms. Since 1960, it has lost six of seven such elections, averaging a vote share below 50%. It appears that the prospect of losing looms large for the presidential party when it has held the White House for two or more terms. But the White House party has little to fear when it has been there just for one term.

Table 1

How the White House Party Has Fared in Presidential Elections (Since 1960)

It is true that George H.W. Bush in 1992 and Gerald Ford in 1976 went down to defeat after having served one (or even less than one) term as president. But our concern right now is with party control of the White House. By 1992, the Republicans, with Reagan and Bush, had held the White House for three terms; by 1976, they had done so with Nixon and Ford for two terms. Neither Ford nor Bush (the elder) got into office by sweeping out the other party. They did so more or less as designated successors of presidents who had compiled a well-known record. For better or worse, the successors were measured by the legacy of their predecessors, whom they served as vice presidents, rather than as genuine first-term presidents.

The pattern of results in table 1 hints at the operation of an electoral cycle. A party that captures the White House from the other party in a presidential election can expect to hold on for two, perhaps three terms, after which the other major party is poised to return to the White House for a similar tenure. How well does the pattern of the last half century apply to presidential elections generally?

EVIDENCE FOR ELECTORAL CYCLES

Popular voting for president has been widespread since 1828. Also, since that time, just two parties have claimed the lion’s share of the vote. Throughout that history, Democrats have been the more constant party, battling a succession of National Republicans, Whigs, and just plain Republicans. Figure 1 charts the Democratic percentage of the two-party vote for president from 1828 to 2012, a period covering 47 elections altogether. It takes some, but not too much, imagination to spot numerous cycles—maybe nine—in figure 1. To be sure, these periodic fluctuations are not very well behaved. They differ a great deal in the range of the vote (amplitude) as well in the length of the cycle (period).

Figure 1

The Two-Party Vote in Presidential Elections, 1828–2012

(Color online)

To cope with “disturbed” cycles of this kind, Yule (Reference Yule, Stuart and Kendall1971) proposed a second-order autoregressive model. Yule applied it to annual sunspot numbers, which exhibit irregular periodic behavior. Applying it to the presidential vote (1828–2012), as depicted in figure 1, produces the following coefficients (with standard errors in parentheses underneath):

$${\rm{VOTE}}_{\rm{t}} = 49.2\,(0.85) + 0.525\,(0.127)\,{\rm{VOTE}}_{\rm{t} - 1} - 0.474\,(0.108)\,{\rm{VOTE}}_{\rm{t} - 2} $$

The combination of the two autoregressive coefficients, using the Yule (1971, 403) algorithm, yields this estimate for the length (period) of the electoral cycle: 5.2 terms, or roughly 20 years. The estimate confirms the impression made by figure 1, which shows about nine cycles during a time span of roughly 200 years. It is also close to an estimate derived from spectral analysis (Merrill, Grofman, and Brunell 2010, 6). Note that an electoral cycle comprises a period from the moment a party takes control until it returns to power, after being out of power for half of that period. Hence a cycle lasting five terms implies that a party can expect to control the White House for half that time, or 2.6 terms, to cite the exact estimate. For each party, getting reelected after one term in office is a safe bet. After two terms in the White House, however, the odds get longer, with the prospect of defeat looming larger than that of another victory. What theory might explain the working of such an electoral cycle?

TERM LIMITS

A seemingly obvious clue to change in elections after roughly two terms is the presidential term limit. The limit, whether by tradition or law, clearly impacts the electoral prospects of the White House party. With a two-term president not on the ballot, those presidential contests become, to use a congressional expression, open-seat races. This should diminish the value of the incumbent president’s record and standing, giving the opposition candidate a better shot at winning the election (Norpoth Reference Norpoth, Dorussen and Taylor2002). Several elections come to mind in which popular presidents such as Eisenhower in 1960 and Clinton in 2000 most likely would have won a third term for their party that was lost without them on the ballot.

At the same time, not all two-term presidents end their tenure with a glowing standing that promises electoral victory. Would President George W. Bush in 2008, given his dismal popularity, public discontent over the Iraq War, and the economic collapse, have had any chance of keeping the White House in his party’s hands? Or Truman in 1952, Johnson in 1968? For any election in which the White House party loses an edge with the departure of a popular two-term president, there is another where it gains an edge as an unpopular one leaves the stage.

Moreover, the two-term limit does not explain how first-term incumbents do so well winning a second term. Even more remarkable than the difficulties of the White House party in securing a third term is its unqualified success in winning reelection after the first term. Consider these facts of electoral history: Since 1828, the White House party has lost such a bid only three times (1848, 1888, and 1980), with only the 1980 case an unambiguous exception. In the 1848 election the sitting president, James K. Polk, declined to run for a second term, while in 1888 the loser, Grover Cleveland, actually won the popular vote. What is it that incumbency in presidential election bestows after the first term that diminishes after two terms?

TIME FOR A CHANGE

Any time the incumbent party is defeated in a presidential election the winning side typically ran a campaign with a variant of the theme, “It’s Time for a Change.” Sometimes it is so stated explicitly, as in Eisenhower’s campaign in 1952, following one of the longest stretches of party control in presidential politics (“The Man from Abilene” commercial, http://www.livingroomcandidate.org/commercials/1952). Change is what the electorate wanted and what it got, at least in personnel and party. So now the newly installed president and his party have to make good on the pledge, from Roosevelt’s New Deal platform in 1932, to settling the war in Korea in 1952, to Kennedy’s “New Frontier in 1960, to Reaganomics in 1980, and Obama’s “Hope and Change“ in 2008. But change often takes time to be implemented. FDR was unable to restore prosperity by 1936, the “New Frontier” agenda largely languished in Congress, Reaganomics produced the worst recession since the Great Depression, and Obama’s woes in overcoming the Great Recession are familiar. So how did all these first-term incumbents manage to win reelection, often in landslide fashion?

At such moments in history, one may suspect, the public is willing to show some patience with the efforts of the new administration to work its magic. Voters seem willing to give the implementers of change the benefit of the doubt. They are inclined to heed a call as expressed by the 1984 Reagan reelection campaign, “Stay the Course.” Or they are persuaded by claims that bad conditions are mostly the fault of the preceding administration that lost the last election, not the current one. Few probably blamed FDR for the Depression that he inherited from the Hoover administration, and more voters blamed Bush than Obama for the poor economic conditions under Obama’s watch in the 2012 election (D’Elia and Norpoth Reference D’Elia and Norpoth2014).

But when two terms are up the public may no longer be inclined to give the administration the benefit of the doubt. By then blaming the predecessor for any problems besetting the current incumbent no longer sounds persuasive. Now public patience with progress is wearing thin. While a positivity bias in public reactions may help a new incumbent get by during the first term, a negativity bias may undermine it in subsequent terms (Bloom and Price Reference Bloom and Price1975). Or it may simply be that the novelty of the previous change has worn out and the public’s desire for another change is irresistible. This may be a universal dynamic, not rooted in term limits, realignment eras, or even presidential elections. The same type of second-order autoregressive behavior found in US presidential elections (Norpoth Reference Norpoth1995) also manifests itself in parliamentary elections in Britain (Lebo and Norpoth Reference Lebo and Norpoth2007). Incumbents worldwide appear to be subject to the “cost of ruling” (Lewis-Beck and Stegmaier Reference Lewis-Beck and Stegmaier2013; Paldam Reference Paldam, Norpoth, Lewis-Beck and Lafay1991). Another view credits ideological shifts between liberalism and conservatism for the swings of the electoral pendulum (e.g., Merrill, Grofman, and Brunell 2010; also Jones Reference Jones2002, for a review).