FiveThirtyEight.com will officially relaunch in very early 2014 in cooperation with ESPN. I've been spending most of my time recruiting and interviewing job candidates for the new site. It's a labor-intensive process, and we hope that your patience will be rewarded as we begin to tell you more about our plans.
While most of my focus has been on building the new site, the idea was never for me to stop writing completely during the transition period. Instead, Grantland has set up an interim website for me and other FiveThirtyEight contributors to write articles from time to time.
I would have liked to write something about Bill de Blasio's comeback, or the methodological miscues of the Democratic polling firm Public Policy Polling, for instance. Perhaps I would have done so if there were a few extra hours in the day. (I've used this same excuse for not going to the gym since 2004.)
But the shutdown is another matter. I've been avoiding the topic for another reason: I'm not sure I have all that much to say about it.
During last year's election campaign, some readers who followed our coverage came to the conclusion that the truth behind what the polls said was relatively simple once you stripped away all the bullshit that accompanies campaign coverage. And that's more or less correct when it comes to presidential elections. They're relatively predictable — more so than most people might expect intuitively, and more so than most of the mainstream media lets on. (Let me qualify that some: Presidential elections in the United States are quite predictable in the late stages of the race — they're not especially predictable months or years ahead of time.) We know this because there is a very rich track record of polling in the United States, and we can go back and look at how accurate the polls were in the past. The polls haven't been perfect, but they've generally done a very good job.
However, presidential elections are more the exception than the rule. As I discuss in my book, the more common tendency instead is that people (and especially the "experts" who write about the issues for a living) overestimate the degree of predictability in complex systems. There are some other exceptions besides presidential elections — sports, in many respects; and weather prediction, which has become much better in recent years. But for the most part, the experts you see on television are much too sure of themselves.
That's been my impression of the coverage of the shutdown: The folks you see on TV are much too sure of themselves. They've been making too much of thin slices of polling and thinner historical precedents that might not apply this time around.
There's been plenty of bullshit, in other words. We really don't know all that much about how the shutdown is going to be resolved, or how the long-term political consequences are going to play out.
So what can we say? What follows are a series of points that I consider to be on relatively firm ground. Some are critiques of the conventional wisdom; some are points of context; some concern relatively fine details of the situation; some are obvious things that I don't think have been emphasized quite enough. None of them constitute a prediction of how the shutdown is going to turn out, or exactly what the political fallout will be. But perhaps they can serve as useful guidance as you read coverage of the shutdown elsewhere.
1. The media is probably overstating the magnitude of the shutdown's political impact.
Remember Syria? The fiscal cliff? Benghazi? The IRS scandal? The collapse of immigration reform? All of these were hyped as game-changing political moments by the news media, just as so many stories were during the election last year. In each case, the public's interest quickly waned once the news cycle turned over to another story. Most political stories have a fairly short half-life and won't turn out to be as consequential as they seem at the time.
Or consider the other story from President Obama's tenure in office that has the most parallels to the shutdown: the tense negotiations, in 2011, over the federal debt ceiling. The resolution to that crisis, which left voters across the political spectrum dissatisfied, did have some medium-term political impact: Obama's approval ratings declined to the low 40s from the high 40s, crossing a threshold that historically marks the difference between a reelected president and a one-termer, and congressional approval ratings plunged to record lows.
But Obama's approval ratings reverted to the high 40s by early 2012, enough to facilitate his reelection. Meanwhile, reelection rates for congressional incumbents were close to their long-term averages.
None of this applies if the United States actually does default on its debt this time around, or if the U.S. shutdown persists for as long as Belgium's. But if the current round of negotiations is resolved within the next week or so, they might turn out to have a relatively minor impact by November 2014.
2. The impact of the 1995-96 shutdowns is overrated in Washington's mythology.
But what about the pair of government shutdowns in 1995 and 1996? It's common to find articles asserting, without qualification, that they were a major factor in prompting President Clinton's reelection.
However, the empirical evidence for this claim is thin. Clinton's approval ratings were somewhat higher a few months after the shutdown than a few months beforehand — but this was part of a relatively steady, long-term trend toward improved approval ratings for Clinton, probably because of solid economic growth.
Nor was Clinton's victory over Bob Dole in 1996 anything unexpected. Incumbent presidents generally win reelection even under marginal conditions (as Barack Obama did last year) — and they're overwhelming favorites during peacetime elections when the economy is robust, as it was during 1996. Furthermore, Clinton did not have much in the way of coattails: Democrats gained just two seats in the House that November, and wouldn't win back the chamber for another decade.
3. Democrats face extremely unfavorable conditions in trying to regain the House.
Even if the shutdown were to have a moderate political impact — and one that favored the Democrats in races for Congress — it might not be enough for them to regain control of the U.S. House. Instead, Democrats face two major headwinds as they seek to win back Congress.
First, there are extremely few swing districts — only one-half to one-third as many as when the last government shutdown occurred in 1996. Some of this is because of partisan gerrymandering, but more of it is because of increasingly sharp ideological divides along geographic lines: between urban and rural areas, between the North and the South, and between the coasts and the interior of the United States.
So even if Democrats make significant gains in the number of votes they receive for the House, they would flip relatively few seats because of the way those votes are distributed. Most of the additional votes would come in districts that Democrats were already assured of winning, or where they were too far behind to catch up.
Consider that, between 2010 and 2012, Democrats went from losing the average congressional district by seven percentage points to winning it by one percentage point — an eight-point swing. And yet they added only eight seats in the House, out of 435 congressional districts.
In 2014, likewise, it will require not just a pretty good year for Democrats, but a wave election for them to regain the House. But wave elections in favor of the party that controls the White House are essentially unprecedented in midterm years. Instead, the president's party has almost always lost seats in the House — or at best gained a handful.
One might be able to construct an argument for why the precedent could be violated. The pattern of the president's party losing seats in the midterms has been very strong in the past — but political scientists aren't quite sure why this is the case. One theory is that voters may elect members of Congress from the opposite party as a check on the president's power. But if Congress instead is seen as the more powerful entity, voters might desire to curb its power instead.
Essentially, Democrats will have to persuade swing voters that having Republicans in charge of one chamber in one branch of government is more dangerous than yielding unilateral control of the government to the Democrats — at a time when President Obama is fairly unpopular, and when the signature initiative of the last Democratic Congress has been rolled out badly. Moreover, the voters that Democrats have to persuade about this are somewhat right of center, since the median congressional district is somewhat Republican-leaning and since the voters who reliably turn out at midterm elections are older, whiter, and otherwise more conservative than those who vote in general elections. It's not an impossible task for Democrats, but the terrain is all uphill.
4. The polling data on the shutdown is not yet all that useful, and we lack data on most important measures of voter preferences.
There is an array of polls that ask voters which party they blame for the shutdown. For the most part, they show Republicans taking somewhat more blame than Democrats, although the differences aren't as stark as in 1995 and 1996.
The unanswered question is how this abstract notion of blame, on just one issue, might translate into tangible changes in voter preferences 13 months from now. Republicans are taking more blame for the shutdown — but they were extremely unpopular to begin with. How many people's votes will be changed by the shutdown?
The best measure of this might be the generic congressional ballot, which measures overall preferences for Democrats or Republicans in congressional races around the country. However, very few generic ballot polls have been released since the shutdown began, and the exceptions are from dubious polling firms like Public Policy Polling and Rasmussen Reports.
That isn't to say Republicans are without any reason for concern: The most recent Gallup poll shows a much sharper drop in Republican favorability ratings than in those for Democrats, which could presage a shift in the generic ballot.
But measures that put the parties head-to-head are much more valuable. I'll be more convinced about the electoral downside for Republicans if and when we see such a shift in the generic ballot, or, say, in a number of Senate races around the country. (One irony is that while the House has been the focal point for GOP intransigence on the shutdown, Republican candidates for the Senate may have much more at risk, since the race for that chamber is much closer and contains a much higher proportion of competitive races.)
5. President Obama's change in tactics may be less about a change of heart and more about a change in incentives.
Democrats seem pleased that Obama has held a relatively firm line against negotiating on the shutdown or debt ceiling so far. There is something to be said for his change in tactics. His position is much clearer than the story of shifting imperatives that has characterized the Republican Party's stance. For everyday Americans who are paying only a moderate amount of attention to the shutdown debate, and who apply the heuristic that the party with the more unified message has a stronger position than the one that seems to be at war with itself, that could matter.
But Obama also has one advantage that he lacked during the debt ceiling debates of 2011: He no longer has to face reelection. One risk to Obama during 2011 was that even if a debt ceiling breach endangered many congressional Republicans, it could also have injured his reelection chances because of the long-term damage to the economy. Voters could have chosen both a new Congress and a new president.
This time around, Obama has less to lose. Whether or not he "wins" the battle with Republicans over the shutdown, he is unlikely to persuade Republicans to sign on to any of his other policy priorities, such as immigration reform. Further damage to Obama's approval ratings could still have some knockoff effects on races for Congress. But the relationships aren't nearly as direct as when the president is on the ballot himself.
Furthermore, the historical relationship between the economy and races for Congress is relatively ambiguous. (For example, the GOP's huge gains in the 1994 midterms came despite a growing and healthy economy.) A quick-and-dirty chart comparing GDP growth and the performance of the president's party at the midterm elections reveals almost no correlation, for instance. (While more complicated formulas that account for factors such as which party holds control of Congress and how many seats it began with may perform slightly better, these models can be prone toward backfitting, and they've done a mediocre job of predicting congressional results in real time.) This is not to imply that there's no downside for President Obama and the Democrats: It's hard to imagine Democrats gaining many seats in the House, let alone enough to win back the chamber, if Obama's approval ratings are mired in the 30s at this point next year. But voters will not have the option in 2014 on casting a pox on both the president and the Congress: They'll have to choose whether they hate congressional Democrats or congressional Republicans less.
6. The increasing extent of GOP partisanship is without strong recent precedent, and contributes to the systemic uncertainty about political outcomes.
Congress has gone through periods of relatively high partisanship before — for example, at the turn of the 20th century. But the degree of polarization in the Congress is higher than at any point since the Great Depression by a variety of measures, and is possibly at its highest point ever. (Most of the evidence suggests the trend is asymmetric: Republicans in Congress have become much more conservative, while Democrats have become only somewhat more liberal.)
What this means is that, whether they assume the form of statistical models or more anecdotal takes on the evidence, conceptions based on recent history of how the negotiations might play out may not be all that reliable. That there were 17 government shutdowns between 1976 and 1996, for example, none of which persisted for more than three weeks, may not be all that meaningful since none of those came at a time when Congress was nearly as polarized as it is now. Similarly, the fact that an aggregate limit on federal debt has been in place since 1939 [PDF] may not tell us all that much. This is not to imply that the risk of a debt ceiling breach is all that high, especially given the reports of progress in budget talks as of Thursday morning.
But there's a lot we don't know.