November 01, 2011

US: The Economy and the 2012 Presidential Election

Published: 10/24/2011

Based on the expected state of the economy, President Obama faces an uphill battle to win reelection in 2012. He will need a combination of an economic rebound over the next 12 months and an ineffective Republican candidate if he is to retain the White House.

Over the years, statistical attempts to explain and predict U.S. presidential election results have yielded two overarching themes. First, Americans tend to vote their pocketbooks. If the economy is growing strongly and unemployment is low, the incumbent party has a very good chance of retaining office. When the economy is faltering, U.S. voters will more likely vote for change. Second, Americans tend to favor an incumbent president running for reelection. If the economy is weak enough, however, an incumbent president can lose, as Jimmy Carter learned in 1980 and George H. W. Bush did in 1992.

Based on the likely state of the economy in 2012, President Obama faces a steep uphill task to secure reelection. Based upon our forecast for the economy, our election equation projects just a 43.5% share of the two-party vote for the president, i.e., a heavy defeat. This does not mean that the president's re-election campaign is doomed, though. The economy might perform better than we expect, and if voters perceive the economy to be on the mend, the president still could win even if unemployment is still very high (as seems inevitable). Just as important, if not more so, the election equation takes no account of the identity of the candidates—a Republican opponent lacking broad appeal could tilt the balance back in favor of the president. But it does appear that this is an election that is the Republicans' to lose.

The Election Equation: Historical Tracking Performance Is Good

IHS Global Insight has been estimating equations for presidential election results for many years, following the pioneering research of Professor Ray Fair of Yale University. The present equation is an updated version of previous equations used by IHS Global Insight and its predecessor companies.

The equation predicts the percentage share of the two-party vote won by the incumbent party, and is fitted over the 16 elections since the Second World War (Table 1). It is driven by a combination of economic and predetermined political factors.

Table 1

Election Model Results

(Percentage of two-party vote for incumbent party)













Truman (D)

Dewey (R)





Stevenson (D)

Eisenhower (R)





Eisenhower (R)

Stevenson (D)





Nixon (R)

Kennedy (D)





Johnson (D)

Goldwater (R)





Humphrey (D)

Nixon (R)





Nixon (R)

McGovern (D)





Ford (R)

Carter (D)





Carter (D)

Reagan (R)





Reagan (R)

Mondale (D)





Bush (R)

Dukakis (D)





Bush (R)

Clinton (D)





Clinton (D)

Dole (R)





Gore (D)

Bush (R)





Bush (R)

Kerry (D)





McCain (R)

Obama (D)



Obama (D)

? (R)

Note: italicized errors represent elections where the equation incorrectly predicted the winner of the popular vote.

Pocketbook concerns, traditionally viewed as key to election outcomes, enter via the growth rate in real per capita disposable income over the year preceding the election (third quarter to third quarter). Each percentage point of growth adds about 1.6 percentage points to the incumbent party’s vote (Table 2).

The equation also includes a dummy variable for an incumbent president seeking reelection. Most studies have found that an incumbent president has an advantage compared with another candidate from the same party. We find that the magnitude of that advantage depends on the health of the economy as measured by the unemployment rate. An incumbent president appears to receive credit—or blame—for the state of the labor market, while an alternative candidate from the same party does not.

We find that an incumbent president running for reelection with the unemployment rate at 5.5% gets a vote boost of 2.4 percentage points.(1) The higher the unemployment rate, the less of an advantage incumbency provides; at unemployment rates above 6.7%, incumbency becomes a disadvantage rather than an advantage.

The equation also contains a “fatigue factor.” The more consecutive terms served by one particular party, the more likely voters will decide that a change in leadership is necessary. In our equation, each additional term in office costs the incumbent party 1.9 percentage points in vote share.

The final factor in the equation is party affiliation. The equation finds that, for a given state of the economy, Republican candidates have fared better than Democrats. This advantage is worth an extra 1.2 percentage points in vote share to a Republican candidate.

Older versions of the equation have also included the inflation rate (with higher inflation subtracting from the incumbent party’s vote share), but we found that with the inclusion of all of the factors listed above, the inflation rate lost statistical significance, and it is not included in the present version of the model.

The equation explains 88% of the variation in the vote going to the incumbent party, and correctly forecasts the winner of the popular vote in most elections. It errs in the close election of 1968 by predicting a narrow win for the incumbent-party candidate Hubert Humphrey rather than a close victory for Richard Nixon. This may reflect the turmoil in the Democratic Party over the Vietnam War—a factor not picked up by the model. The model also errs in 1976, overpredicting Gerald Ford’s vote share by 1.6 percentage points. The Watergate scandal may have wiped out the “incumbent advantage” built into the equation.

Table 2

Regression Results

Dependent Variable: Percentage of presidential election two-party vote for incumbent party



Std. Error






Growth in real per capita disposable income




Incumbent President*




Times unemployment rate




Fatigue factor**




Incumbent party***








Adjusted R-squared


Durbin-Watson stat


S.E. of regression


* 1 if incumbent president running, 0 otherwise

** Number of terms served by incumbent party minus 1

*** 1 if Republican, -1 if Democrat

In the 2000 election, the equation correctly predicts a majority of the popular vote for Gore, but anticipated that he would poll about 2.4 percentage points higher than he actually did. The equation suggests that Gore should have won the popular vote by a comfortable enough margin that his fate wouldn’t have rested on a handful of votes in Florida.

The equation gives an insight into the elections of 1980 and 1984, showing how an incumbent can win even with a high unemployment rate. The unemployment rate in the third quarter of 1984, at 7.4%, was little different from the 7.6% rate in the third quarter of 1992. So how did Ronald Reagan storm to victory in 1984 while George H.W. Bush was rejected in 1992? The equation's explanation is—income growth. Real per capita income growth was 6.8% over the year preceding the 1984 election, but just 2.1% over the year preceding the 1992 election. So President Reagan had solid income growth to back up his "Morning In America" campaign theme.

In 2008, the IHS Global Insight equation in use at that time predicted that incumbent-party candidate John McCain would receive 46.9% of the two-party popular vote; in the event, McCain achieved 46.3% of the vote—a significant margin of defeat.(2) By November 2008, the economy was grinding to a halt, and McCain's election chances were hurt both by the poor performance of incomes and by the fatigue factor. Real per-capita disposable income rose by just 0.7% from the third quarter of 2007 to the third quarter of 2008, and only by 0.1% excluding the impact of temporary economic-stimulus payments.(3) Even with the stimulus payments, income growth was lower than for any other election since 1980, when it was minus 0.5%. Jimmy Carter, the incumbent president, paid the price in 1980 and lost heavily to Ronald Reagan. The fatigue factor also hurt McCain, since the Republicans had held the White House for two consecutive terms.

The Election Equation: 2012 Projection Is Bad News for the President

What does the equation say about the 2012 election? President Obama is an incumbent running for reelection without a fatigue factor (Democrats have held the presidency for just one term). Since World War II, three of ten incumbent presidents seeking reelection have failed: Ford in 1976, Carter in 1980, and George H.W. Bush in 1992. But of the seven incumbent presidents without a fatigue factor working against them, only Carter failed to win reelection.

IHS Global Insight’s October forecast for the U.S. economy anticipates real per-capita disposable income growth of just below 1.3% over the year to the third quarter of 2012. This is troubling for Obama, since it is nearly 1.6 percentage points below the average rate for the year preceding other postwar elections. In fact, no incumbent party since 1948 has won reelection with real per capita personal disposable income growth below 1.75% year-over-year.

The labor market is even more worrying for the president. IHS Global Insight projects an unemployment rate of nearly 9.4% in the third quarter of 2012, exceeding the present rate and well above the 5.5% average rate immediately preceding other postwar elections. The high unemployment rate turns incumbency into a disadvantage for Obama. Not once since World War II has the unemployment rate been above 9.0% for the three months leading up to the election, and no incumbent party has won reelection with unemployment above 7.4% in that period.

Plugging in these economic assumptions, the equation predicts a 43.5% share of the two-party vote for President Obama, a significant margin of defeat. In fact, 43.5% would be the fifth-lowest vote share garnered by a candidate on either side during that period. That the equation has overpredicted the incumbent share of the vote in each of the past three elections does not help Obama’s cause.

Even using assumptions from our optimistic alternative (10% probability; 2.5% real per capita income growth instead of 1.3%, and an unemployment rate of 8.3% instead of 9.4%) increases the projected Obama vote by just 4.1 percentage points, to 47.6%. An Obama victory even under these circumstances would still mean a bigger equation error than in all but 3 of the previous 16 elections, but a large out-of-sample forecast error is possible.

A downside scenario would be disastrous for the president. It is nearly inconceivable that Obama could secure reelection in the midst of or immediately after a second recession (even though he inherited the first one). Using assumptions from our pessimistic alternative (40% probability; -0.5% real per capita income growth instead of 1.3%, and an unemployment rate of 10.2% instead of 9.4%) decreases the projected Obama vote by 4.6 percentage points, to 38.9%. This share of the two-party vote would qualify as the third-lowest for any major party candidate since World War II (Goldwater in 1964 and McGovern in 1972 were lower).

How well would the economy have to do for the equation to predict an Obama victory? If year-on-year GDP growth were to reach 4.0% in the third quarter of 2012, per-capita income gains could strengthen to just over 3.0% year-on-year and the unemployment rate could fall to 7.5% in the third quarter. Under this scenario, the equation delivers Obama the slightest of victories, with a voting share of 50.01%. Such a turnaround in the economy seems unlikely. It has been more than seven years since year-on-year GDP growth eclipsed 3.9%.

How Could President Obama Win Reelection?

Despite the results of our election equation, the Republicans are not guaranteed the White House in 2012. It may be that voters will respond to economic factors in a different way than in the past. Having seen the unemployment rate climb as high as 10.1% per year into Obama’s presidency, voters may be more tolerant of an unemployment rate around 9.0% as the election nears. Some voters may discount Obama’s accountability for the current state of the economy and still assign much of the blame to his predecessor George W. Bush. If the president can deliver faster income growth over the next year, voters may assign more weight to income growth and less weight to the unemployment rate than in past elections, opening the door for the president to win a second term.

In addition, non-economic factors also matter. The equation assumes that the only factors influencing the election result (other than objective, predetermined political factors) are income growth and unemployment. It ignores factors such as foreign policy and social agendas, as well as any qualifying data about the candidates on either side. The topic of electability has become increasingly relevant in recent years, and the equation is built to predict election results based on the average Republican or average Democrat, making no special considerations for African-American, female, or politically extreme candidates (or for their vice-presidential nominees).

The Republicans do not yet have a candidate who commands anything close to majority support in their party. Front-runner Mitt Romney has not decisively separated himself from the rest of the field, suggesting that Republicans are seeking an alternative but have not found it. And their candidate must win broader appeal in the general election. The traditional unwritten rules say that candidates swing to the right or left in the primaries to capture their party base, and then swing to the center in the general election to capture independents and disaffected members of the other party. President Obama must hope that the Republican candidate, whoever that may be, cannot make that transition.

by Nigel Gault and Erik Johnson

1. The unemployment rate is calculated as the average rate of unemployment for the three months ending September of the election year.

2. The reestimated equation predicts a tighter race, but a McCain defeat nonetheless.

3. Note: we excluded the stimulus payments from the income calculation in our election forecast on the grounds that they were temporary rebates bunched into the second and third quarters of 2008, and did not reflect the underlying trend in incomes.

No comments: