The effects of health care reform in 2010 and beyond

Why did Democrats do so poorly in the 2010 elections? The median academic forecast was 44 to 45 seats (PDF). However, Republicans significantly outperformed expectations in picking up 66 seats in the House and six seats in the Senate.

After the election, John Sides, Eric McGhee, and I found that Democratic incumbents who voted the most controversial legislation of the 2008-2010 period -- TARP, the stimulus, cap-and-trade, and health care reform -- performed significantly worse than those who voted no on those bills. Seth Masket and Steven Greene reached a similar conclusion about the effects of supporting health care reform on the vote share received by the most conservative House Democrats who ran for re-election.

Afterward, we joined forces Voltron-style and produced a new article (gated; ungated) that is forthcoming in a special issue of American Politics Research on the 2010 election. In it, we show that the roll call effect on vote share was driven by health care reform. Democratic incumbents who voted yes performed significantly worse than those who did not. Even among a more comparable set of members and districts that we isolate using statistical matching procedures, the estimated effect of support is -5.8 percentage points. We then provide simulation evidence suggesting that Democrats would win approximately 25 more seats if those in competitive districts had voted no, which accounts for the gap between the academic forecasts and the observed outcomes.*

Why did health care reform have such dramatic effects? Individual-level survey data shows that health care reform supporters were seen as more liberal and thus more ideologically distant from voters.

Using mediation analysis, we then show that perceived ideological distance appears to be the key mechanism linking incumbent support for health care reform with individual-level opposition among their constituents. In short, support for health care reform is associated with greater perceived ideological distance, which in turn is associated with a reduced likelihood of supporting the incumbent.

Going forward, the implications for 2012 are less clear. As Sides notes, the economy is the dominant issue in the presidential race and most of the vulnerable Democratic incumbents in Congress lost in 2010. Nonetheless, it will be interesting to see whether challengers can successfully target any remaining Democrats in competitive districts or states who supported health care reform. Will it hurt the party again?

Of course, none of this is to say that Democrats should have declined to pursue health care reform, which was arguably their party's top policy priority after the 2008 election. Parties are frequently willing to pay an electoral penalty to enact their preferred policy agenda. What our analysis shows, however, is that the costs of passing the legislation were significant.

* As Kevin Collins and Jonathan Chait point out, it's possible that news coverage and GOP ads would have focused on the other roll call votes and that they would have had correspondingly greater effects on vulnerable Democrats. In that case, the net effect of supporting health care reform on Democratic losses in 2010 might have been reduced or eliminated. For more on these counterfactuals, see the responses from John Sides and Eric McGhee. See also Seth Masket on how some media coverage has distorted our findings.

[Cross-posted to HuffPost Pollster]

Comments

[Rob]

Congratulations on your article. If it's correct that voting for the Affordable Care Act* cost a number of incumbents their jobs, the phrase that comes to mind is the one so often used after the Nixon resignation: the system works. All praise, honor and glory to the Founders.

One small cavil: Brendan and his co-authors chose to accept the rubric "reform" for the changes wrought by the Affordable Care Act. (The word is used more than a hundred times in the article.) Since "reform" has the connotation of improvement, that's a value-laden decision. When the Bush White House tried to call its privatization plan for Social Security a reform, Brendan didn't take the bait; he called it privatization. For the Democrats' health care legislation, however, Brendan not only took the bait, he swallowed the entire hook, line and sinker.
______________
* I'm tempted to modify the name of the Act with "so-called," but that might be too reminiscent of George Wallace, who once referred to "the New York Times, the Washington Post and the so-called Baltimore Sun."

[David in Cal]

Brendan focuses on politics, rather than policy. I guess that's appropriate for a Political Scientist. However, from a policy POV Health Care Reform has a number of ugly features that are responsible for its growing unpopularity:

1. This major reform was passed in a rush. It was clear that neither Obama nor any member of Congress had read it or understood it.

2. The media didn't understand it, nor did they explain it to the public.

3. The law granted special favors, in order to buy the votes of certain Senators, resulting in an unequal burden.

4. After the law was passed, we discovered that the President has the amazing authority to relieve selected organizations of the law's burden. To understand just how radical this is, just imagine that the President had the power to unilaterally relieve any organization from paying Social Security and Medicare taxes, and imagine that President Bush had used this power to benefit Halliburton and Koch Brothers.

5. From a professional health care delivery POV, it's a badly designed law. It's inefficiency will increase costs, rather than reduce them.

Regarding #2, the media and public still don't understand the law. Consider Michelle Bachman's claim

we have now one person who’s a health dictator in our country and that health care dictator will decide what we get and what we don’t get...
we will see the president of the United States deciding that people won’t be getting hip replacements or knee replacements or, you name it,”
...the contraception issue was a very small example of what Americans can expect as a result of Obama health reform law... future presidents will use the authority the federal government has claimed to aid them in getting reelected.
...We’re going to see in future political elections, presidents are going to come up with a politically popular disease or a politically popular pharmaceutical drug and he or she will be pandering that product before elections.

Brendan tweeted an article from Politio describing Bachman's assertion.
Note how dreadful Politico's coverage was. First of all, their headline focused on what they called "birth panels". However, Bachman's comment dealt broadly with the President's entire range of authority.

Second, Politico missed their chance to tell the readers the facts. Does Obamacare give the President the broad authority to require or prohibit any insurance coverage, as Bachman asserts? Politico doesn't tell. Apparently, they don't know and they don't care.

Politico's coverage reminds me of what was wrong with the Death Panel debate. Critics like Brendan were happy to repeat ad nauseam that Sarah Palin and Michelle Bachman were liars. But, these critics never told us who would have the power to define coverage, what the procedures would be, what coverages were likely to be covered or exclulded, etc. These critics contented themselves with throwing brickbats at conservatives.

[Metrichead]

Brendan, do you think in the future you can post more articles with references to '80s cartoons? You just don't do it enough, I think.

I'm half-way through your paper, by the way. It's a great read, even if I give some paragraphs a double-take to understand what's being said.

[Rob]

Brendan and his co-authors attempt to explain the deviation of 2010 election results from those that were forecast by political science models. The underlying assumption seems to be that the forecast models are basically correct except for their failure to account for incumbents' votes on a prominent piece of legislation. That's fine as far as it goes.

However, in responding to criticism that if attention hadn't focused on the health care vote, it would have focused on a different bill, Brendan pithily tweets, "My take: The 'another bill would have worked' story would be more convincing if the election had not deviated so sharply from forecasts." Now the deviation from the forecasts isn't simply the starting point of the analysis, it's become the rebuttal to criticism of the analysis.

To this casual observer who's admittedly not well-versed in the science, that seems like a bootstrap argument. Moreover, the assumption that standard political science forecast models are fundamentally correct and need only be tweaked by further efforts such as Nyhan et al. is a little surprising. Are these models now regarded as "definitive" and "incontrovertible"? Isn't it possible that the models either were invalid in the first instance (over-fitted, over-simplified, whatever) or have lost validity as a result of changing conditions (e.g., the Internet, increasing polarization, etc.)?

[David in Cal]

Interesting and thoughtful comment, Rob. Coincidentally, there's an article in my actuarial magazine, A Healthy Skepticism Toward Models Although this article refers specifically to models used to regulate insurance, I think the points it makes are more generally applicable:

As models become more ubiquitous, the understanding of their proper role—and the level of healthy skepticism towards their results—is decreasing...

Models are general tools, not precision tools like surgical instruments. Models do not provide answers, they provide information....

One very important modeling issue with which I suspect most of us are already familiar is the distinction between probabilities in the classical “frequentist” sense (e.g., flipping coins, performing repeatable controlled experiments) and the Bayesian sense (e.g., claims propensities, hurricane landfalls, and future interest rates). Unfortunately, many users (and many builders!) of models do not understand this distinction; they interpret the results of models that rely on subjective Bayesian probability assumptions as facts instead of as the indicators that they are. There are many other assumptions to which a model can be extremely sensitive, and it is important for the users of the model output to understand the degree of sensitivity. Furthermore, one must be highly skeptical of model results that claim high degrees of precision (what I like to call “delusional exactitude”).