Tuesday, January 20, 2015

The War on Data Continues

Andrew Gelman and I wrote a piece for the ASA a while back called "The War on Data." It discussed what appears to be a disturbing trend of powerful interest groups trying to discredit and/or defund major sources of important, high-quality data, ranging from the Census Bureau to the National Oceanic and Atmospheric Administration.

Now we can add the CBO to the list. Jonathan Chait spells out the ugly details. You should read the whole thing but here are a few key paragraphs.

The Congressional Budget Office is a 40-year-old institution that has acquired enormous clout within Washington by virtue of its reputation for ideological neutrality. It furnishes Congress and the public with budgetary estimates that, if necessarily imperfect (as all predictions must be), are arrived at fairly. It is also a perfect modern expression of an old Progressive Era–ideal: that policymakers should be informed by the work of impartial experts. That the conservative majority has set out to corrupt this institution as one of its first major acts is, therefore, perfectly fitting.

The old methods CBO used to measure legislation would account for changes in behavior that a new law might create. (Say, higher cigarette taxes would lead to less smoking.) They did not attempt to measure legislation’s impact on the economy as a whole. This is because the two parties disagree completely over what policies make the economy grow faster. Democrats, for instance, believe that tax rates on the rich have little effect on economic growth, but that investing in public infrastructure or education has a lot. Republicans believe the opposite. Congress voted yesterday to require the CBO’s measurement of the budgetary cost of legislation to incorporate assumptions about how it will affect economic growth. Specifically, the GOP's assumptions.
The whole reason the Republican Congress is instituting dynamic scoring comes as a response to its attempt to write a tax reform bill last year. The idea was to lower tax rates while eliminating loopholes and preferences. But Republicans discovered that, while lowering rates is easy, eliminating preferences is hard. After Representative Dave Camp produced a tax reform bill that failed to cut tax rates for high-income taxpayers enough for their liking, Republicans abandoned it en masse. Paul Ryan openly declared his plan to change the forecasting rules so that Republicans could cut tax rates without having to pay for every dollar by ending preferences. The first step was kicking out Douglas Elmendorf, the CBO director widely respected by both sides. The second step was yesterday’s vote.
The new, “dynamic” CBO will be systematically biased to make conservative proposals appear misleadingly cheap and liberal proposals misleadingly costly to the public fisc. This would be true even if the Republicans were soliciting a fair range of forecasting perspectives. By its design, the dynamic scoring rule allows the party in power to game its effects. It applies “dynamic scoring” only to legislation affecting 0.25 percent of Gross Domestic Product. As Chye-Ching Huang and Paul Van de Water point out, congressional leaders can manipulate this requirement easily: They can break up large pieces of legislation into smaller bills to avoid dynamic scoring, or combine smaller pieces into a major bill, if needed to make their agenda appear more affordable. Dynamic scoring is subject to abuse by its very design.


  1. Do Democrats really "believe that tax rates on the rich have little effect on economic growth"? My impression was that Democrats (at least of the Krugman persuasion) believe that increasing taxes on the rich would boost economic growth. Maybe the issue is that Chait is separating taxes and spending. Like the song says, you can't have one without the other.

    1. Andrew,

      With all the usual caveats (not an economist, working from memory, etc.), here's my understanding of the Krugman, Thoma, et al. position:

      Moderate tax increases on the upper classes (somewhere between Kennedy and Reagan rates) raise revenue and reduce inequality while having negligible effect on growth. Tax increases on the lower and middle classes do have an impact on spending and productivity but that drops off as you get to the top of the income scale;

      What you do with that revenue can very much affect growth, particularly if you spend the money in areas such as education, research and infrastructure, but that's a separate question.


      The debate here is over the direct effects of tax cuts and over where we are on the Laffer curve. The larger economic debate has to address both the revenue and the spending sides, but that debate needs to consist of lots of smaller specific discussions and, in this specific discussion, I think that Chait's assertions are well-grounded.


  2. I disagree that this is part of a war on data. Projections are not data in the sense of the usual term. They are the results of applying models to observed data. Those models inevitably incorporate assumptions about the underlying processes, assumptions that typically cannot be verified.

    The way in which this particular change is being proposed, which allows it to be gamed by packaging legislation, is, of course, logically indefensible. But the notion that dynamic scoring is inherently the wrong way to go is just an ideologic preference. Surely there is some effect of changes in taxation on economic growth. The direction may be unclear and the magnitude may be small (at least within the range of changes that are under consideration), but it is unlikely to be zero. So really what the CBO should be doing is providing a variety of projections based on a broad range of plausible different assumptions about the impact on economic growth. That would be the honest approach.

    I would even say that a Bayesian synthesis of those forecasts would make sense, but in the current environment, it is hard to imagine a usefully informative prior that would get any buy-in from more than a handful of people.

    Finally, I'll indulge my cynicism. Do you really believe that these politicians care about data?