This story caught my eye yesterday:
The US Federal Reserve will slash its growth forecasts and predict higher unemployment when it releases updated economic projections this week.
The Fed will release the latest forecasts made by members of its rate-setting open market committee on Tuesday, alongside the minutes of their November meeting, giving a complete picture of why they launched a new $600bn round of asset purchases....
FOMC members have made particularly aggressive upward revisions to their unemployment forecasts, with a large number now predicting that it will still be 8 per cent or above at the end of 2012, compared to the 7.1 to 7.5 per cent that they forecast in June.
“Because I expect hiring to strengthen only gradually, the unemployment rate is likely to remain elevated for quite some time. In fact, I do not expect it to fall below 8 per cent before 2013,” Sandra Pianalto, president of the Cleveland Fed, said in a speech last week.
Let's suppose that unemployment remains elevated into 2012, around the level that the Fed predicts. What will this mean?
It's hard to say. We know for sure that the economy influences political outcomes, but it is hard -- very hard -- to isolate the effect of any one variable. Statistically, this makes sense. Economic variables tend to be closely related to one another, and there are really just a handful of presidential elections where records are available. "Fuzzy factors" plus "small n" equals...uncertainty!
Unemployment is a great example. We know that unemployment matters in presidential election outcomes. The Great Depression saw unemployment exceed 25 percent, which is what prompted the public to swing into office a progressive Democratic majority under FDR, which in turn proceeded to rewrite much of the social contract between government and citizen. If unemployment had been, say 5 percent, at the time of the 1932 midterm, would Hoover have lost? No way.
And yet, look at the statistical relationship between unemployment in November of election years and the incumbent president party's margin of victory/defeat.
There is no statistically significant relationship there whatsoever. Why is that? Again, it's hard to say. It'd be foolish to write off the idea that unemployment has no effect, considering the example of the Depression. And a lack of statistical significance doesn't mean there is no relationship, only that the data set does not point conclusively toward one. And if you look at some of the elections on the left-hand side of the graph, you'll notice they tend to have funky qualities to them. 1952 saw the immensely popular Eisenhower running for office. In American history, just three candidates have enjoyed that level of credibility prior to taking office -- Washington, Jackson, and Eisenhower. It's therefore quite problematic to compare 1952 to, say, 1988. What's more, the elections of 1948, 1968, and 2000 all saw spoiler candidates, each of which robbed the incumbent party of votes. If we dropped these four elections, we'd have a statistically significant fit between unemployment and election results.
But are we accounting for outliers, or just engaging in special pleading to redeem the importance of unemployment? Who knows! When you get right down to it, we only have 16 observations here (1948 to 2008), and that's not really enough to say much of anything of interest, at least via statistics. Imagine that a buddy gives you a rigged coin that will hit on tails 60 percent of the time. In the long term, you're going to see tails more than heads, but after just 16 observations, you might actually see more heads than tails. In fact, the chances of getting 8 or more heads, even with a coin tilted to the tails side, is not too shabby -- about 28 percent. In other words, we could very easily wind up with a "false negative," calling a coin fair when it is in fact rigged.
False positives are also a possibility. They are why I am suspicious of graphs that show a tight relationship between a single economic factor and electoral outcomes, like this one which tracks election results against Q2 GDP. There are so many economic variables that we could compare to electoral outcomes, which points to a critically important, yet often forgotten, caveat about statistical analysis: if we run enough tests, at some point we will wind up with a false positive, thinking that we've hit upon a causal relationship when in fact we haven't. For instance, suppose your friend gives you a hundred coins and asks you to test to see which, if any, are rigged. You have about a 70 percent chance of getting heads 13 out of 16 times on at least one coin, even if all of the coins are fair. That would be a false positive, i.e. you conclude that the coin is rigged when in fact it isn't. By the same logic, if we run enough tests between economic variables and election outcomes -- sooner or later we are going to "find" something that isn't actually there.
All of this is why that old phrase "lies, damned lies, and statistics" is so often repeated. We always need to check statistical results against common sense, being sure to strongly favor the latter. For instance, I'd ask: how many people thought to themselves in November 1948, "Hmm...well I like the cut of Tom Dewey's jib, but boy-oh-boy GDP last May was really strong, so I'm going with Harry Truman!" Probably not that many! That's not to say that there is no relationship between presidential outcomes and Q2 GDP, in fact I think there is one. Instead, my point is that common sense suggests that the graph is perhaps overstating it, or at least inducing us to oversimplify what is in fact a very complicated relationship. (See the footnote at the bottom for more on 1948.)
What we can say is that perceptions of the economy do indeed make a difference in election results. The following chart tracks final election outcomes against the Gallup question of "who do you trust more to handle the economy." There are even fewer observations here, but the relationship is much less fuzzy.
The only observation that doesn't "fit" here is 2004, when John Kerry outpaced George W. Bush in terms of evaluations of the economy, yet he still lost. Two thirds of the Bush vote was from people whose most important issue that year was either terrorism or moral issues. Voters most concerned about the economy voted for Kerry 80-20. Other than this, we see a pretty good relationship in the above graph -- in which 6 out of 7 times the candidate who is most trusted on the economy wins the election, and 2000 being an even split in terms of economic-based evaluations and the final result.
How can we tie all this data together? Here's my take. The economy is a very important factor in presidential elections (duh!), but voters form holistic judgments about it that are not reducible to a line on a chart. Voters are also not nearly as well informed as economists; they don't head to BEA.gov the day before the election to check on Q2 GDP. They vote on a general, diffuse sense of where the economy has been, where it is headed, and how much credit/blame each side deserves for the situation. So, unemployment is an exceedingly important factor -- insofar as people care above all about having jobs -- but it is also a trailing indicator of economic strength, meaning that unemployment matters in the broader economic context. So, for instance, while it was still high by Election Day 1984, it was on its way down and the leading indicators were all very strong, thus contributing to a blowout Reagan victory.
What we appear to be headed for by November 2012 is an economy that is not as weak as it is now, but still not as strong as it could be, especially in terms of the labor market. To me, that signals a political gray area, one in which evaluations of the two major party candidates could be determinative. Perhaps a good metaphor will be 1976. The economy that year was not terrible but not great, having recovered from the 1973-75 recession but soon to suffer two more contractions (in 1980 and 1981-1982). That election more than any in recent history strikes me as one in which evaluations of the two nominees played a decisive role. Neither candidate was very inspiring. A stronger Democratic nominee than Jimmy Carter could have defeated Gerald Ford by a larger margin, and a stronger Republican nominee could have squeaked out the win against Carter. In the end, Carter won a narrow victory that depended heavily on his native South coming in overwhelmingly for him, while the rest of the country was lukewarm at best about the former Georgia governor.
For 2012, I think the Obama campaign will not be able to claim "Morning In America," at least not if the Fed's prediction turns out to be correct. Instead, they'll be singing a different verse from the same old tune they've been belting out for the last year -- the trend line is good, but there is more work to be done. The Republican message will inevitably be, "We can do better!" That suggests the GOP needs to find a nominee who can sell that message, somebody a majority of voters believes really can deliver on the economy better than Obama has. If the presidential election were held today, I'd say the message would sell itself, much as it did in 2008. But by 2012, I think Republican salesmanship is going to be a critically decisive factor in the outcome.
Footnote: Regarding the graph that tracks Q2 GDP against electoral outcomes, Jonathan Chait and Brendan Nyhan recently argued that the strength of the economy in 1948 was a big reason why Truman was bound to win, and it was a "myth" that he came from behind. Indeed, Q2 GDP was very strong that year. But Michael Barone -- who frankly knows more about modern American politics than anybody I've ever read -- makes exactly the opposite claim. In My Country, he writes:
Dewey's above-the-battle strategy, an overreaction to the criticism he had endured for campaigning too bitterly after the Fala speech in 1944, gave Truman the opening to charge that Republicans were going to shut down successful government programs and refuse to protect ordinary citizens when hard times returned. For ordinary citizens did not know that twenty years of prosperity and vibrant economic growth lay ahead. Around 1947 they may have begun to suspect that that was the case, but the slowdown in the fall of 1948raised just enough apprehension about another recession that a critical number of voters decided to keep the Democrats in for another four years. In October, Truman started attacking Herbert Hoover, even though he had brought the former president back into government to head a commission on reorganization. [Emphases Mine]
If the only thing you knew about the 1948 election was where Q2 GDP fell on a scatterplot against the election results, you'd conclude that it was the strength of the economy that favored Truman, not the weakness. Yet Barone, as has been his stock in trade for decades, supplies the needed context and turns conventional wisdom on its head. In the election of 1948, millions of people who voted still had painful memories of the Great Depression and the seeming indifference of Hooveresque Republicanism. They also had different expectations of what a healthy economy should look like, not having been through the amazing growth of the 1950s-1960s or the 1980s-1990s. Dewey's aloofness gave Truman the opportunity to exploit these sentiments as economic growth slowed from 7.6% in Q2 to 0.6% in Q4, and thus the CW has it exactly backward: it was the weakness of the economy that favored Truman, not the strength.