Saturday, April 19, 2014

Apples and oranges

There’s a brouhaha about the Census Bureau changing the questions, asked annually, about whether a respondent has health insurance. Since the change is occurring just as ObamaCare is implemented and since preliminary tests of the new questions show that fewer people end up counted as uninsured, many on the Right are making two claims:

1) The fact that the new questions show fewer people are without health insurance will be an advantage for ObamaCare supporters.

2) The change and the timing of the change are deliberate; that is, the Administration set out to design questions that would show fewer uninsured and timed the implementation of the new questions to coincide with the implementation of ObamaCare.

I disagree with the first claim and think the second claim is over-simplified. I also think that the quality and comparability of data really don’t matter much.

First, I do agree that when the new questions show that fewer people are without health insurance, ObamaCare supporters will claim ObamaCare has succeeded in reducing the number of uninsured. However, ObamaCare opponents will be able to claim that the drop is solely a result of the new questions. I think ObamaCare supporters will get the worst of these dueling claims. The New York Times quotes officials and official documents as saying:

... it will be difficult to say how much of any change is attributable to the Affordable Care Act and how much to the use of a new survey instrument. [snip]

... “it is likely that the Census Bureau will decide that there is a break in series for the health insurance estimates,” ... This “break in trend” will complicate efforts to trace the impact of the Affordable Care Act, it said. [snip]

... the data for 2013 and 2014 would not be directly comparable with the long series of data for prior years.

Given these “official” disclaimers, it will be difficult for ObamaCare supporters to rebut opponents’ claims that any reduction in the number of uninsured is an artifact of the new questions. In fact, one could argue that the change in questions is at least as much of an advantage for ObamaCare opponents as it is for ObamaCare supporters.

Second, I don’t think the revision of the questions was a totally political move. The Times claims that the Census Bureau has been “trying to make [the insurance questions] more accurate” “for more than a decade”. Therefore, this is not something the current Administration cooked up five years ago when it took office. However, the Times also states:

The Department of Health and Human Services and the White House Council of Economic Advisers requested several of the new questions, and the White House Office of Management and Budget approved the new questionnaire.

I don’t think HHS and the CEA set out to write questions that would show more people with health insurance. But I do believe that if the preliminary tests had shown the questions showed fewer people with health insurance, the White House would not have approved the new questionnaire or at least would have insisted that the Census Bureau wait a few years to implement the new questions.

That said, there is one thing that makes me more likely to accept the argument that the changes to the questions and the timing were designed solely to serve the Administration’s purposes: the Census Bureau seems unwilling to make any effort to provide data that would “be directly comparable with the long series of data for prior years.” Even if the Census Bureau is bound and determined to start using the new questions, honestly believing the results will more accurately reflect reality, it would be pretty simple to, for example, do what Greg Mankiw suggests:

Why not, for a few years, give half the sample the old questionnaire and half the new one?  This procedure would provide a basis for eventually splicing together the old and new time series.

That the Census Bureau is apparently not even considering this - or any approach like this - does give weight to the “White House conspiracy” storyline.

Finally, while policy analysts are (rightfully) horrified at losing comparable data, I don’t think the lack of continuity matters a bit so far as political arguments about policy are concerned. Let’s imagine that the questions didn’t change and we are now comparing the uninsured picture of 2014 with the strictly comparable uninsured picture of 2013. We would find one of three scenarios:

1) There are fewer people without health insurance. ObamaCare supporters will claim this is proof the law is succeeding. ObamaCare opponents will claim that this is a result of an improving economy so more people have employer insurance or could afford to buy their own anyhow; that this is a result of people going on Medicaid who could have gone on Medicaid even without ObamaCare; that even if ObamaCare is reducing the number of uninsured the effect is so small that neither the monetary cost nor the coercion cost is worth it; and/or that any number of other approaches could have resulted in a greater decrease at less cost.

2) There are more people without health insurance. ObamaCare opponents will claim this is proof the law is, at best, ineffective and, at worst, is actually detrimental. ObamaCare supporters will claim that the problem is a weakening economy; Republican scare tactics; technical problems with the website; people not being aware of the individual mandate; the insurance companies charging too much; that the law didn’t go far enough; and/or that the full effect will only be felt as people learn more about the law and come to rely on it. They will probably talk about a slow embrace of Medicare.

3) There are the same number of people without health insurance. The arguments over this scenario will look a lot like those for Scenario Two although ObamaCare opponents will probably drag out the the arguments about costs not being worth it and about any number of other approaches being better.

And you know what? Reasonable people will be able to reasonably accept any of these arguments. The truth is that this type of economic and social engineering is so incredibly complex that there is no simple causal connection between what the government does over here and what the result is over there; we cannot hold all other variables constant. Therefore, we can always come up with apparently valid reasons why a law this complex seems not to have worked but really did, and with equally apparently valid reasons why a law this complex seems to have worked but really didn’t.

Perhaps this is part of the reason our current politics are fiercely ideological and bitterly partisan. We are so far past the point at which the effects of policies can be accurately measured and intelligently evaluated that all we have to go on is what we believe about how the world works and, perhaps more important, what we believe about how it should work.



Judd Gregg was right (Via Neoneocon) - Wall Street Journal piece about Gregg withdrawing his name from consideration as Secretary of Commerce “due in part to his concern with the Obama administration's decision to have the next Census director report to senior White House staffers as well as the commerce secretary.” I had forgotten all about this dust-up and, yes, it does give more weight to the “White House conspiracy” storyline. (I have to say, though, that I find it hard to believe that the same Administration that couldn’t get up and running even minimally had enough intelligence, foresight, and competence to grab the Census Bureau five years ago in order to try to make ObamaCare look good now.)

No comments: