Yesterday I began a two-part series trying to establish expectations for the 2014 midterms. My argument was that we won’t likely see a Republican wave this time around, and that it is probably time to scratch the “sixth-year itch” argument.
Today I want to focus more specifically on what we might expect to see in the upcoming elections. Let’s first look generally at the obstacles Democrats face in their quest to take back the House. They need a net gain of 17 seats to do so, although they might well get a start on this by picking up an open seat in South Carolina, where Mark Sanford’s troubled special election campaign has been abandoned by the NRCC.
So let’s assume they need 16. Let’s also understand that there is no “rule” against the president’s party gaining seats in a midterm election. Losing seats, however, is a very strong tendency. The president’s party has gained seats in only three midterms since the Civil War: 1934, 1998 and 2002 (it did so in 1902, but not relative to the size of the House, which was increased that year).
These elections occurred in very unique circumstances. In 1934, we were in the midst of a sharp recovery that began almost literally the day after Franklin Roosevelt was inaugurated. The economy grew by 11 percent, according to some estimates. In 1998, we were in the midst of another sharp economic boom, the Republicans had overreached with their impeachment of the president, and the Democrats had few vulnerable seats after their 1994 debacle. In 2002, President Bush was still riding high on his post-9/11 popularity, and the drums of war were being beaten again for Iraq, creating a “rally ’round the flag” effect.
In other words, an incumbent president’s party only gains seats in midterm elections in pretty outstanding circumstances. As we’ll see below, the circumstances for Democrats right now are pretty mediocre.
But even if you think that circumstances for the Democrats this year are like they were in 1934 or 1998, or like the Republicans’ circumstances in 2002, bear in mind that that probably isn’t enough. In ’34, Democrats gained nine seats. In ’98, Democrats gained five. In ’02, Republicans gained eight.
In other words, Democrats would have to almost double up on the best midterm election in U.S. history to take back the House.
Now this doesn’t mean it can’t happen. In response to yesterday’s piece, David Nir of Daily Kos Elections referred me to this article from Steve Hirdt at ESPN, which shows 10 “rules” proving that none of the teams in the NFL could possibly win the Super Bowl. Nir jokingly referred to this as the tendency of commentators to succumb to the “anything that hasn’t happened before can’t happen now” rule.
This tendency has wrecked many an otherwise fine analysis. In 1998, most observers were quite certain the Democrats wouldn’t gain seats, because everyone knew that the president’s party always loses seats in midterms (unless that president is FDR).
In 2010, we had something of a “showcase showdown” effect going on, where prognosticators seemed bent on making sure they were as close to the actual number as they could be without going over it. Almost everyone’s prediction was on the low end of what really happened. I’m convinced this was, in part, because we didn’t have any recent examples of a party losing more than 55 seats in a midterm (it’s not accidental that predictions steadily climbed upwards in 2010 until they struck this level, then stayed there until the very end).
So what’s important to keep in mind is that it is absolutely possible for the Democrats to gain 16 seats. It would just take a highly unusual combination of events for it to occur. I don’t think we’re seeing that yet.
So let’s now look more closely at the factors that influence midterms. From the Democrats’ perspective, we have the good, the mediocre, and the ugly. (Note: As I pointed out in Part 1, when put into a regression equation, all of these factors come back either significant, or awfully close to significant.)
First, the good news. Congressional elections have a tendency to revert to the mean. That is to say, over time, there’s something of an “ideal weight” for parties that they will tend to gravitate toward. The Republicans or Democrats might win majorities of 280 or 290 seats in a given election. But given the partisan composition of the nation, that level of strength in Congress just isn’t sustainable over the medium to long term.
The following chart shows how far above or below the mean the president’s party has been in each midterm election since 1962, as well as the result. It’s not perfect, but you can see the rough relationship here.
The good news for Democrats is that they’re still pretty underexposed. In other words, we should probably expect them to do better than they otherwise would, given the condition of the economy and the president’s job approval.
One caveat: The post-2010 redistricting may well have shifted the Democrats’ expected outcome downward (as did the post-1990 redistricting), which we won’t know for another few elections. For now let’s just observe that both parties’ opportunities are limited.
There are only nine Democrats in districts Mitt Romney carried, and only 17 Republicans in districts Barack Obama carried. If we look at things in terms of Partisan Voting Index (how a party performs relative to national forces), there are only eight Republicans in districts with Democratic PVIs, and 15 Democrats in districts with Republican PVIs. This is in sharp contrast to 2010, when dozens of Democrats were running in districts with Republican PVIs. Again, this limits opportunities for both parties to post big gains.
The mediocre news is the president’s job approval, which as of Wednesday was 50 percent in Gallup, and 48.2 percent in the RCP Average. Here’s how that stacks up historically:
Josh Kraushaar joked in 2011 that the president’s job approval was right at the Mendoza Line, separating acceptable mediocrity from unacceptable mediocrity. That’s about where it is now. If he stays at this level, it will probably be a quiet election night. But if he slides much more, the Democrats start to get into really problematic territory.
Note also that, historically speaking, the president’s party has only gained seats when his job approval is above 60 percent. Reaching that level is not impossible by any stretch, but it’s a long haul from where we are today.
Finally, the ugly. The condition of the economy impacts the outcome of elections. There’s a bunch of ways to measure this, but an old standby, dating back to Edward Tufte’s watershed 1975 article on midterm election prediction, is real disposable income.
Right now, year-over-year RDI growth is pretty miserable, and this correlates with some rather unfortunate outcomes for presidential parties, all other things being equal.
The good news for Democrats is that this seems to be secondary to presidential job approval: You do see years like 1990, when a relatively popular president with an underexposed party had a fairly good midterm. You also see years like 2010, where RDI growth is strong, but the party fares poorly. We should also note that RDI-based presidential models fared poorly in 2012, so the relationship might be breaking down.
What happens when we put this all together? Democrats have quite a lot of work to do to take back the House. President Obama would probably have to get his job approval into the high 60s, and the economy probably has to take off. It would probably require a pro-Obama wave akin to 2010-in-reverse. Again, not impossible, but not very likely.
At the same time, things probably have to deteriorate quite a bit for the Democrats before Republicans can expect even low double-digit gains. There just aren’t enough vulnerable Democratic seats to expect big gains, absent some significant wave. Somewhere between a five-seat Democratic pickup and a 15 seat Republican gain seems a safe prediction for now.