What Exactly is Overrated Here?

Ryan Noonan

Ryan Noonan is an economist with a small federal agency. Fields in which he considers himself reasonably well-informed: literature, college athletics, video games, food and beverage, the Supreme Court. Fields in which he considers himself an expert: none. He can be found on the Twitter or reached by email.

Related Post Roulette

29 Responses

  1. Dan Miller says:

    I actually like Yglesias, but that was a terrible post.  My theory is that because he puts out so much volume, there’s a certain number of posts that are bound to be crap.Report

    • Erik Kain in reply to Dan Miller says:

      Yglesias drives me crazy because he’ll write one thing that I really like and agree with and the next minute he’ll go off the rails.Report

    • Plinko in reply to Dan Miller says:

      Maybe I will remove myself from the cool kids club but I Yglesias is the blog I read most apart from this one and I generally get something out of every post.

      Sometimes it’s a sort of Charles Murray-esque situation where even where he’s methodically wrong, what he was trying to get at still gets me thinking of something interesting, to be sure.Report

  2. Kimmi says:

    “policy is a spectacularly inefficient machine for altering human behavior.”

    Tell that to the birds. specifically sparrows.

    Policy has MANY unintended consequences. Incentives guide behavior, and policy is all about incentivizing things. THAT said, I think many policies actually cause people to alter their behavior. Maybe it’s not always in ways we WANT, but… that’s another matter.Report

    • Ryan Bonneville in reply to Kimmi says:

      I’m not saying policy has no effect on behavior, but I think a lot of people in the Yglesias/Klein/Drum camp overrate what you can use it to do. People don’t choose the size of their families because of tax policy; they just don’t.Report

      • Kimmi in reply to Ryan Bonneville says:

        *snort* yeah. $100K (or more) committment, versus how much back in taxes??Report

        • Brandon Berg in reply to Kimmi says:

          According to the chart, $2,000.

          Per year. For eighteen years.

          You’re not thinking on the margin. If you have absolutely no interest in having children, then obviously tax credits that refund a third of the cost isn’t going to change your mind. But if you’d like to have another child but just don’t think you can afford it, then a subsidy really can make a difference.Report

      • Morat20 in reply to Ryan Bonneville says:

        But, but, but I’ve been told raising the top marginal rate back to the Clinton era levels will cause all the producers to go Galt!

        That taxing gains like income will destroy the economy!

        The problem with economic arguments in favor of tax policy is it’s pretty much a matter of “choose whatever supports what you want”.

        It’s the rare cases where someone prefers outcome X and looks at the economics and sighs and says “not worth it” or “not possible” or “will lead to Outcome Y” that are worth noting.Report

      • Jeff in reply to Ryan Bonneville says:

        But they do chose the size based on how much welfare they get?  Or do the economics put that theory to rest?Report

  3. Jaybird says:

    We don’t even know if we’re awake right now, man!Report

  4. North says:

    Yes, an astonishingly muddled littly post by MY and the parts that aren’t muddled are horrible.Report

  5. Brandon Berg says:

    He’s making a valid point–it’s just that “empirical evidence is overrated” is a terrible way to summarize it. What he’s really saying, if you read the whole post, is that shoddy, uncontrolled, N = 1 empirical evidence is overrated. Fertility rates have dropped off the cliff all over the developed world. The fact that the US fertility rate has held steady for 40 years may be significant. What you’d really have to do is look at all developed coutries and see whether fertility rates have fallen more slowly in those countries where the subsidies have increased more quickly You’d probably want to control for some other factors, too.

    Out of curiosity, how many semesters of economics have you taken?Report

    • For starters, N=1 is not what he’s critiquing. He’s critiquing this: http://www.stanford.edu/~gopi/w15984.pdf. They actually did, you know, control for other factors.

      As for the rest, I’m not sure what you want. Intro Micro and Macro, Intermediate Micro and Macro, Game Theory, two courses in Econometrics, and a seminar on philosophy and economics = 8? Plus a math major, two courses in public finance, a handful of statistics courses.Report

      • I was critiquing Kevin Drum’s post and his use of that chart and the inference he drew from the paper, not the entirety of the Crump/Goda/Mumford research which presumably is more extensive and contains additional material.Report

        • Chris in reply to Matthew Yglesias says:

          I find this to be an odd explanation of Yglesias’ post, since it’s quite clear that Drum was using the chart as an illustration of the research findings, not using the chart as the actual evidence. This is actually a good way of describing someone’s findings without having to get into the nitty gritty of the actual research: here’s what they found, and here’s a pretty little chart that illustrates it. If you want to know more, you read the research.

          Anyway, Yglesias’ post is odd even if Drum were using the graph by itself (which would make it seem that Drum just doesn’t understand how statistics work). The first question any scientist will ask you if you say to him or her, “There are good theoretical reasons to believe X,” will inevitably be, “Where’s your data?” You know why, because there are good theoretical reasons to believe a whole lot of things, particularly when we’re talking about complex human behavior, and the only way to make sure that you’ve got the right assumptions and have accounted for all of the relevant variables is to actually test your theory empirically. If your theory is that tax incentives for having children lead to families with more children, but the data says that there’s no difference, then either your theory is wrong or you’ve missed something. This doesn’t mean data is overrated. In fact, it means the opposite: your theory is only as good as your data.Report

          • Michael Drew in reply to Chris says:

            You realize that was Yglesias explaining Yglesias’ post, right?  Not that that means it can’t be odd.Report

            • Chris in reply to Michael Drew says:

              I do realize that. I assume, however, that he’s gone from this thread forever, so I wasn’t really talking to Yglesias.Report

            • Plinko in reply to Michael Drew says:

              Yeah, I kinda thought the same thing, MD!

              I went back and read both MY’s and Drum’s post, and I think in the end MY’s overall point is valid, it’s just muddled in that the reference to Drum seems kind of incidental when it is, in fact, the whole point.

              Matt would have better made his point id he’d draw out Drum’s post more and respond to it specifically rather than mention it as an aside and then go off about people doing what he does in his posts every day.


              • Chris in reply to Plinko says:

                But Drum doesn’t just use the graph. He references the research, links to it, then uses the graph to illustrate their finding. Nowhere does he suggest that the graph is how they got to their conclusion. I really have no idea what Yglesias is thinking, to be honest.Report

              • James Hanley in reply to Chris says:

                I have to agree with Chris here.  Yglesias’ argument, boiled down, is that evidence that contradicts theory should be rejected.  I don’t think he actually believes that, or would dare to argue that directly, but it is essentially what he argued in this case.

                A much better critique of Drum would be to note that two different measurements of fertility are being conflated: one is fertility of the general population over time, and the other is the variation in fertility within the general population at a given moment in time.  There’s actually nothing contradictory about overall fertility dropping across generations while child-bearing subsidies create fertility differences between different sectors of the current population at any given moment in time.

                The study cited by Drum uses only the general fertility rate as a dependent variable, so it does not get into the issue of sectoral differences within a particular generation, and so it does not rebut the hypothesis suggested by this critique.  I seem to remember some reports suggesting that recipients who have run out of welfare have fewer children, so there’s some implicit support for this hypothesis.  But, it remains a hypothesis, an alternative line of study that could be verified, or not, by an empirical study.  In the end, it, too survives or falls on the actual data.Report

    • Jaybird in reply to Brandon Berg says:

      I stayed at a Holiday Inn Express Last Night.

      I suspect that when you start controlling for more than one issue, you start controlling for issues you have no idea you’re controlling for.Report

  6. Katherine says:

    Yeah, that seems like a poor argument.

    The correct argument to make regarding the chart he shows is that there’s too many confounding factors for a simple chart (or a simple regression) to determine the impact of child tax subsidies on fertility rates, so we don’t have any firm empirical evidence one way or another.  But that’s not a case against using empirical evidence, just against assuming it conveys more useful information than it truly does.


  7. I don’t get the incredulity.  Why is it so absurd to suggest that the number of variables flying around any macroeconomic data that we get is so vast that a priori induction is more reliable than empirical data?  It’s a fulfillment of, not an abandonment of, scientific principles to note the difficulty of isolating variables in the field of economics.Report

    • that we get is so vast that a priori induction is more reliable than empirical data?

      What this says is, “The potentially relevant variables are so myriad that it’s better to just ignore them altogether.” Or at least, that’s what it amounts to.

      It is of course true that “isolating variables” is difficult in any study of human behavior, but the people who do it do have a few tricks up their sleeves. What’s more, using those tricks, we don’t have to choose between induction and data.Report

  8. Damon says:

    The more I read of that guy’s work, the more I think  he’s an idiot.  I’m not even gonna respond to his question.  I mean, really.Report