What’s with that?

Often times I’ll be listening to a trade seminar and the presenter makes the following argument:

  1. Here’s an economy with some friction (e.g. trade restrictions)
  2. Here’s an economy without said friction, see how its best
  3. The difference between countries that are more like the first and countries more like the second can be attributed to the friction

Here’s my problem with that argument. Its silly.

Oh, you need more detail? Its silly because no country is completely without the friction and there’s no guarantee that going from lots of the friction to a little bit less will move you in the direction of the frictionless world. Maybe some other friction balances out the first friction and when you reduce the first, the second over compensates and you’re in a worse state than you began in.

Well, that was abstract. Example: let’s say I would be the most handsome man in Davis if I just took my girlfriend’s advice about how to dress myself. She has a perfect sense of style and boy would I be cool if I would dress exactly how she tells me. So, here we have two scenarios: me dressed as my usual college casual (read: boring) self or me dressed to the nines. Obviously, it would be best for me to do 100% what my girlfriend tells me.

Does it then follow that if I put the girlfriend-approved power tie over my faded tee-shirt but kept everything else in my attire the same that I’d be better dressed? No. In fact, I’m sure it would reduce my sartorial rating.

This argument doesn’t work in that context and I don’t think it works in economics, either. Why, then, do I see this argument so often? ((See chapter 8, in Prescott and Parente (pdf) for a variation on this argument))

12 thoughts on “What’s with that?”

  1. What you’re saying is that this policy advice, in another, unspecified model, would yield ambigous results. That’s true of *any* policy advice, no? (Including monetary, optimal taxation, etc.)

    You can either take this route and end up in policy nihilism (fine with my political preferences) or you can stick to explicit models, I think.

    In the P&P model, a lowering of the barriers has an unambiguous effect. To advance the discussion productively I think you need to come up with a model that embeds their plant-level barriers but has ambiguous effects for their policy prescriptions.

    In this particular case, of growth, I think we can measure barriers by protection from expropriation indexes, the Heritage index and the Transparency International index and see how this relate to TFP, as extracted with Solow’s method. I think we’ll find a strong, linear-ish, positive connection.

    There could be endogenous mischief, of course, so it’s not causality… but it makes your thesis of theoretical ambiguity less relevant in practice, I think.

  2. I’m saying they’re comparing apples and oranges. We don’t observe, in reality, frictionless economies. We observe high and low frictions in economies. Comparing frictions to no frictions doesn’t tell you much about comparing high and low fictions per my example.

    What I’m saying is that the U.S. isn’t the no-monopoly case of their model. So when their no-monopoly economy shows TFPs of 2.7 times the monopoly economy’s, its says nothing about the experience of the U.S. versus low-income countries. I think the best you can say is that the 2.7 figure is an upper bound on the explanatory power of their model. We can’t even say zero is a lower bound because they haven’t given us a reason to believe TFP increases monotonically with the level of monopoly power.

    Their conclusion is that the 3 times TFPs seen in reality can be mostly explained by monopoly power as demonstrated by their model… it just ain’t so.

    Of course, there’s a possibility the Nobel Laureate knows something I don’t…

  3. But, but… that’s what their theory is all about… to have TFP as a (continuous) function of the barriers. They compare economies along a continuum, given by the barriers parameter. They give that model exactly to argue that TFP is nonincreasing (at all values) in barriers. If that’s not a reason, it’s not because they weren’t trying.

    Also, they deal explicitly with the issue of relative vs. absolute TFP. The leader/US is normalized, so they don’t rule out, as far as I understand, the idea that the US could increase its own TFP by further reducing barriers.

    In any case, re: what you “observe in reality”… you don’t observe an auctioneer, competitive labor markets, a single good, a single price for each good, dynastic households, etc. etc. so I’m not sure where you can go from there.

  4. I think you’re going all Dani Rodrik on us. Which is cool. But what if policy is not like dressing the best but eating well. If I take my doctor’s advice about nutrition 100% I’ll definitely be healthier. But if I only cut out McDonalds french fries … well, I’ll still be healthier, just not as much. What kind of “symptoms” or “signals” can we get that would allow us to discriminate the policy cases between your and my analogies?

  5. Fashion might not be the best example. Some men can get away with tennis shoes and a tuxedo. Or a five o’clock shadow. Or leaving the fedora at home.

    Please, carry on.

  6. No, no, the more I think about it, pushmedia1 has a point, and I don’t think it’s the Rodrikian one… I’m not sure he should make this a criticism of the modeling strategy, though.

    His point seems to me to be more Joan Robinson. Comparative statics with macro models where parameters are calibrated conveniently (this is a problem for P&P and sometimes Lucas) are a awkward approach to the business of strategic, firm-level interaction.

    I’m doing a post on growth, on what I learned recently, and Barriers is going to be featured prominently, so I need to think about this more.

  7. I think Rodrik questions the policy implications of models along these lines. I’m questioning the positive implications of these models. To be clear, I’m not saying the assumptions of their modeling are wrong. Models are necessary simplifications, blah blah blah.

    My point is that P&P haven’t shown us if labor monopolies in their model are like dressing better (non-monotonic in “better” policy) or like eating right (monotonic in better policy) and this is the crux of the issue in comparing rich and poor countries.

    Maybe I’ll admit to learning this lesson from Rodrik. He’d stop here, though, and say see we need second best policies. I don’t think he is right. All we have here is uncertainty about the effect of marginal reductions in frictions but in principle, I’m all for removing government mandated monopolies. My point is that monopolies haven’t been shown to account for that much of a difference in TFPs and at this point we should go back to the drawing board.

    Why can’t we model the degree of the friction with some parameter and then do comparative statics? Or calibrate one model once for each type of economy and compare the results?

  8. I don’t understand your second paragraph. In the P&P model you get an inverse relation between barriers and TFP. Isn’t that the result you’re looking for?

    There’s something in your critique that escapes me. If it’s just me, it’s understandable. But I can’t tell.

    So, let me ask you another way… what would constitute satisfactory proof of a connection between factor monopolies/coalitions and income differences?

  9. “what would constitute satisfactory proof of a connection between factor monopolies/coalitions and” TFPs?

    A negative partial derivative on efficiency with respect to the monopoly parameter.

    There was no monopoly parameter in their model, you say? Exactly.

Comments are closed.