The magical thinking of Fred Barnes

[Updated below with more details on Clinton's win in 1996]

I take it as a given that most journalists know very little about political science. But I still assumed that almost everyone has a basic understanding of the relationship between the state of the economy and presidential election outcomes. Apparently Fred Barnes missed the memo.

As Jamison Foser points out at Media Matters, Barnes published an op-ed in the Wall Street Journal yesterday that repeatedly suggests the key to Obama's re-election is cutting spending:

If Mr. Obama wants to avert a fiscal crisis and win re-election in 2012, he needs House Speaker Nancy Pelosi to be removed from her powerful post. A GOP takeover may be the only way...

Over the past 50 years, it should be no surprise which president has the best record for holding down discretionary spending. It was President Reagan. But who was second best? President Clinton, a Democrat. His record of frugality was better than Presidents Nixon, Ford and both Bushes. Mr. Clinton couldn't have done it if Republicans hadn't won the House and Senate in the 1994 election. They insisted on substantial cuts, he went along and then whistled his way to an easy re-election in 1996...

Mr. Obama's re-election to a second term is heavily dependent on his ability to deal effectively with the fiscal mess.

The suggestion that Clinton had "an easy re-election" due to spending restraint is implausible. Here's a plot of changes in non-defense discretionary spending (Barnes's preferred metric) for recent first-term presidents (see methodological notes below for details):

The relationship between spending restraint and vote share is entirely driven by Reagan's 1984 landslide (an outlier on both dimensions). When it is excluded, there is no relationship between the variables.

By contrast, the relationship between economic growth and presidential vote share is extremely strong and robust to the exclusion of any single president. Here, for example, is the same set of presidents with their vote share plotted against the Douglas Hibbs measure of weighted per-capita growth in disposable personal income (drawn from his Bread and Peace model):

As I've previously noted, Clinton's move toward the center (which included an emphasis on deficit reduction) may have helped somewhat to boost his margin above what we would have otherwise expected, but the driving force in 1996 (as in every election) was the state of the economy.

Similarly, it's hard to imagine Reagan being re-elected based on his record of spending restraint if the economy was in the tank. The fiscal year 1982 budget cut non-defense discretionary spending substantially, but it wasn't enough to protect Reagan during the recession of 1981-1982, which pushed Reagan's approval numbers down to 35% by January 1983. Once the economy bounced back, however, approval rebounded in time for him to crush Mondale in 1984.

Barnes may wish that presidents were re-elected based on spending restraint, but that's just not the world we live in. Obama's fortunes will rise or fall depending on the state of the economy.

Update 6/10 9:33 AM: To reinforce the point that we don't need to use the Clinton's move to the center (which Matthew Yglesias emphasizes), here are plots from two other prominent forecasting models showing how closely Clinton's popular vote totals can be predicted using economic conditions alone.

First, here's a plot from a 2008 article by Alan Abramowitz predicting vote share using GDP growth in the second quarter of a presidential election year:

Second, here's a plot from a 2008 article by Tom Holbrook predicting vote sharing using average satisfaction with personal finances in the summer of a presidential election year:

In both cases, the actual outcome is quite close to the predicted value from a simple linear fit.

Update 6/10 12:07 PM: Jonathan Bernstein objects to the claim that Clinton moved toward the center after 1994. I'll certainly agree that the difference between Clinton pre- and post-1994 is overstated, but there is at least some evidence to support the claim that he moved rightward. Most notably, see this figure from Michael Bailey which uses Supreme Court cases as a bridge to compare the preferences of the president with those of Congress and members of the Court (click the graphic for a larger version):

[Methodological notes: Spending levels are drawn from the historical tables for the FY2010 budget and are in constant dollars (Table 8.2). I compare spending in the fiscal year beginning before the president takes office (e.g. fiscal year 1969 for Nixon) to the fiscal year that ends just before the president is up for re-election (e.g. fiscal year 1972 for Nixon). I start with fiscal year 1974 for Ford (since he took office in August of that year) and end with fiscal year 1976 (excluding the transition quarter in 1976).]

Comments

[The Objective Historian]

http://hotair.com/archives/2010/06/07/video-the-stimulus-failed/

http://www.heritage.org/Research/Reports/2010/01/Why-Government-Spending-Does-Not-Stimulate-Economic-Growth-Answering-the-Critics

For all your ridicule of supply-siders, I'd hope for some posts that ridicule the Obama stimulus plans. These sites should provide a start. Note the first is the opinion of Jeffrey Sachs, long-time advocate for conventional Democratic policies.

[The Objective Historian]

OK ... has it occurred to you, diehard Democrat that you seem, that it was the spending restraints previous to 1984 and 1996 that encouraged economic growth and thus permitted the elections of Reagan and Clinton.

What amazes me is that the GOP lets the US forget that Pelosi and Reid have been running the economic side since January 2007. And the real outlier is the one-shot Iraq War mishandling (really a political error rather than a strategic one) and the real estate lending derivative collapse of 2008 caused by government intervention as much as anything that elected Obama.

You'll find Obama will be the best advocate Reaganomics ever had since Reagan ... maybe better. By illustration, creating virtual half-show barely-supervised compensation-bloated unionized civil service jobs might get you campaign funds, some votes, and people to take to the streets, but it pales in comparison to a private enterprise system provided more capital to create jobs efficiently. It should be in your textbooks or perhaps it awaits me to write them.

[David]

I agree with Brendan that Barnes has weak support for his contention that spending cuts lead to votes. Brendan's analysis is much better.

Still, even Brendan's work has limitations, compared with most other applications of statistics. The number of sample points is not at all sufficient to prove cause-and-effect, according to unsual statistical standards.

That's particularly the case, because by necessity the Political Science models are poor ones. The usual statistical assumption is that the data points come from the same distribution. However, due to changing conditions -- economic, social, cultural, political, etc. -- we know that the underlying distibution is not constant. In other words, there may be changes in the relationship between votes and spending. AFAIK there is no way to measure the change in this relationship. The best we can hope for is that it's changing slowly enough so that statistical techniques have some value.

[Rob]

An interesting question is what level of 2nd quarter election year growth is sufficient to give an incumbent president confidence he'll be re-elected. Eyeballing the Abramowitz chart, it appears that even at 5% growth, there's only something like a two in three probability of re-election. (Sure, the chart isn't intended as a predictive device, but let's face it, what matters to real politicians is what they need to do to achieve 50.1% of the two-party vote, not what explains variation in the two-party vote above and below that level.)

BTW, if Abramowitz's independent variable is indeed the most meaningful predictor, that suggests an incumbent president could improve his re-election chances by holding down growth in the period through the first quarter of election year, then letting it bob up to a fat growth in the second quarter. I wonder whether looking back several quarters would improve the fit, perhaps weighting the more recent quarters higher. (I lack both the statistical skillz and sufficient curiosity to try to do alternative calculations myself.)

[pgl]

I guess Fred Barnes believes that the increase in defense spending is something we did not have to pay for in the 1980's.

[Mercer]

If economic growth is the main factor in winning elections why didn't Gore win in a landslide?

[David in Cal]

This Gallop poll shows why the impact of past spending cuts may not be a good predictor of how current spending cuts (or the lack thereof) will affect the 2010 election. Federal Debt and Terrorism are now considered the top threats to the U.S. The Republicans are perceived as the best party to deal with both.

If the Dems were to now cut spending, voters might perceive them as better qualified to deal with the threat of Federal debt. However, during some past period when Federal debt was not as urgent an issue, a spending cut by the party in power would have been less helpful.