Bread, Peace, and the 2008 Election

Douglas Hibbs’ “bread and peace” model has been extremely effective at predicting the outcomes of U.S. presidential elections. This chart, covering elections from 1952 to 2004, gives you the gist:

On the vertical axis is the incumbent-party candidate’s share of the two-party vote. On the horizontal axis is the growth rate of per capita real disposable personal income (DPI) over the three and a half years leading up to the election. The growth rate is adjusted so that more recent periods (quarters) are weighted more heavily. Income growth — “bread” — is a strikingly good predictor of the vote outcome.

There are two main exceptions: 1952 and 1968. (These aren’t included in calculating the regression line in the chart.) This is where the “peace” component of the model comes in. In those two years the incumbent (Democratic) party suffered from a large number of American casualties in a war for which it was viewed as responsible — Korea in 1952 and Vietnam in 1968. Those two wars were still relevant in the ensuing elections, in 1956 and 1972, but the incumbent (Republican) party didn’t suffer much because it hadn’t started the wars.

In analyses in a 2000 article and a recent update (here; hat tip to Andrew Gelman), Hibbs finds that other factors neither add to the explanatory utility of the bread and peace model nor reduce the estimated impact of income growth and war casualties.

What does the model predict for the 2008 election? It’s early yet, but nevertheless interesting to take a look. Through the end of 2007, Hibbs’ weighted income growth rate measure is 1.4% (my calculation). If that continues and “bread” dominates, as it has in most prior elections, the model projects a victory for the Republican candidate. The next chart shows this as “2008a.”

Surprised? Many citizens and pundits expect a Democratic victory. And seemingly with good reason. The current Republican president is extremely unpopular, the Democrats did very well in the 2006 mid-term elections, Democratic voters appear to be more energized than their Republican counterparts, and the two issues voters say are most important to them, the economy and the Iraq war, seem likely to favor the Democratic candidate. Still, some recent polls (such as this and this and this) suggest a Republican nominee might well defeat either Hillary Clinton or Barack Obama in the general election.

Lots of things could result in a Democrat winning in November. In the context of the Hibbs model there are three relevant scenarios.

  1. The 2008 election is an exception. In politics there are no immutable laws. Even if average income growth and war casualties have been and continue to be key factors in presidential elections, they might not determine the outcome of this one.
  2. Like in 1952 and 1968, the incumbent party is hurt by American casualties in a war it initiated. As a result, its candidate receives less than 50% of the two-party vote despite relatively rapid income growth. This is shown in the chart as “2008b.” Hibbs reports that approximately 29,000 Americans had been killed in Korea at the time of the 1952 election, and almost exactly the same number had died in Vietnam by November of 1968. As of February 2008 nearly 4,000 Americans have been killed in the Iraq war. That’s far fewer than in 1952 or 1968, but the public’s tolerance threshold may be considerably lower now (see here, for instance).
  3. Income growth slows in 2008. That would reduce the predicted vote share of the incumbent-party (Republican) candidate, especially since growth in recent periods is weighted more heavily in the model. This is shown as “2008c” in the chart. At the moment, with the economy teetering on the brink of recession, a slowing of income growth appears extremely likely.

Another possibility is that the model no longer works well. You can see in the first chart above that 1996 and 2000 are not predicted very well by income growth, and unlike 1952 and 1968 they cannot be explained by war casualties. The model predicts 2004 accurately, but perhaps that is a fluke, due to fear of terrorism and the incumbent president’s at-that-time seemingly successful prosecution of the Afghanistan and Iraq wars.

Why might the model no longer work well? One hypothesis is that as a society gets richer, pocketbook issues recede in importance for voters.

A second consideration is that growth of per capita personal income is no longer a useful indicator of how voters have fared economically. In recent decades the bulk of income gains have gone to a small slice of the population — those at the top of the distribution. Measuring growth via an average misses this. The next chart illustrates the point. From the late 1940s through the mid-1970s per capita (i.e., average) DPI and median family income both increased steadily. Since the mid-seventies per capita income has continued to do so, but median family income has been relatively flat.

If the model’s effectiveness has indeed declined, other factors may matter more in 2008 than they have in the past. Which factors? And with what outcome? Your guess is probably as good as mine.

14 thoughts on “Bread, Peace, and the 2008 Election

  1. This chart

    shows real median household income steadily increasing, in contrast to your last image which shows it to be stagnating. What is the source of the disparity?

  2. Pingback: It’s McCain for President! at Jacob Christensen

  3. Tapen:

    It can’t just be a matter of scale. The time series for median income on this page is equal to 2 at some point in the early 1970s and also at some point in the early 1990s (as well as at other times). The graph over at wikipedia is strictly larger at any point during the 90s than it ever was during the 70s.

    Perhaps the difference occurs because “household income” is not the same thing as “family income?”

  4. Alex and Tapen:

    My chart uses families, whereas the one Alex links to uses households. Households are in principle a better unit to use, but the Census Bureau data for them only begin in 1967, so it’s impossible to see the contrast between the early post-WW2 decades and the period since the mid-1970s. In any event, median household income too has increased far less rapidly than per capita DPI since the mid-70s.


  5. Isn’t a more important reason why the Hibbs model is inapplicable to 2008 that no sitting president or vice-president is contesting the election, and therefore can neither claim credit or be attacked for his administration’s record? Having said that, both 1928 and 1920 elections seem to fit the bread and peace model, with the first returning a Republican campaigning on the basis of Coolidge prosperity (Hoover) and the second a Republican campaigning against the late-Wilson recession…

  6. Pingback: Looking Toward November, Part II: Forecasting The Election… « Professors R-Squared


    The Election Model utilizes the latest state and national polls to determine the winner of the True Vote.
    Of course, it assumes that the election is held on the date of the most recent polling update.

    In projecting the True vote, the implicit assumption is that the election will be fraud-free.
    But the recorded vote is never equal to the True Vote.
    The evidence is overwhelming: millions of mostly Democratic votes are uncounted in every election.

    Therefore, the Election Model cannot accurately project the official, recorded vote.
    If it did, then the polling data was wrong and/or the projection assumptions were invalid.

    In 2000, 110.8 million votes were cast, but only 105.4 million recorded,
    In 2000, 125.7 million votes were cast, but only 122.3 million recorded.
    Why should 2008 be any different?
    It is a certainty that millions of votes will be uncounted.
    It is also a certainty that more than one million Democratic voters will be disenfranchised.

    Can we expect that DRE touch screens, without a verifiable vote count, will not be rigged?
    Can we expect that central vote tabulating software will not be tampered with?

    Obama’s True Vote (T) will be reduced by uncounted (U), switched votes (S).
    The recorded vote formula is: R = T – U – S (not including the disenfranchised)
    The bottom line is that Obama will need a landslide to overcome the multiple levels of fraud.

    The Election Model does not use historic econometric time-series data.
    Interest rates, employment rates, commodity prices, consumer confidence are important factors.
    But don’t expect political pundits to quantify the fraud factor in their regression models.

  8. Pingback: Elections, Campaigns, and Fundamentals « Consider the Evidence

  9. Pingback: On the Identification of Parameters « Epanechnikov's Blog

  10. Pingback: Carbon Logic · The Fuzzy Math, and Logic, of Election Prediction Models

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s