I gave Rodrik's book a glowing blurb in Bloomberg View, and it was well-deserved. But actually I do have one big problem: the first two chapters. These chapters consist entirely of Rodrik's very general thoughts on economic models, and what they should and shouldn't be used for.
The problem with these early chapters is the audience. Economists will already have heard most or all of these philosophical ideas. Non-economists, in contrast, will probably not understand what they're reading, because the chapters are written in sweeping, general terms, and move very quickly between a number of difficult topics that each require a good deal of background knowledge. So these early chapters suffer the same issue as Karthik Athreya's Big Ideas in Macroeconomics - they fall into an uncanny valley, too old-hat for economists but too inside-baseball for non-economists.
So I fear that many readers may get turned off early and not finish the book. Which is a shame, because the latter two thirds of it are really excellent, and should be read carefully by economists and non-economists alike.
Rodrik really shines when he talks about his own field, development econ. He gives a vivid recounting of the Washington Consensus - why it was adopted, why it went wrong, and how the mistakes could have been avoided. The story of the Washington Consensus provides the perfect backdrop for Rodrik's ideas about what economists and models should do. The episode demonstrates why it's important for policy advisors to look at a bunch of alternative models, and use personal judgment to choose which ones to use as analogies for reality. It is the perfect example of the "models as fables, economists as doctors" worldview that Rodrik is trying to lay out.
In fact, I wish more of the book had been about trade and development economics. Rodrik's blog posts and articles on these topics are always top-notch, and when you look at how Rodrik has struggled with these topics, you easily understand why he thinks about modeling and policy recommendation in the way that he does.
Anyway. Enough nitpicking. It's Rodrik's book, not my book.
Chapters 3 through 6 of Economics Rules alternate between Rodrik's criticisms of his colleagues and his responses to outsiders' criticisms of the econ profession. On all of these points, I find myself pretty much in agreement with him. It's very difficult to sum them all up (so go read the book), but here's a few that really stood out:
* Rodrik notes that economists tend to present a much more simplistic, pro-market stance to the public than they show in their research and behind closed doors. He attributes this to economists' widespread belief that the public is biased against markets. Rodrik suggests that economists give the public a little more credit, and change their public stance to reflect the true diversity of their views. That sounds right to me.
* Rodrik strongly criticizes the New Classical and RBC macro theorists of the 1980s. He essentially accuses them of trying to create a grand unified Theory of Everything, which in econ is just never going to work. That sounds about right.
* Rodrik tries to counter the criticism that economists ignore things like norms. In doing so, he basically says "The evidence shows that norms often matter, and economists pay attention to the evidence." This demonstrates Rodrik's deep respect for data and evidence. He doesn't even mess with the question of "theory vs. data" - to Rodrik, the two always go together. I admire that a lot.
* Rodrik does say one thing that kind of bothers me. He says that economics, unlike science, doesn't replace bad models with better ones - it just makes new models, expanding the menu of models that policy advisors have to choose from. That seems very true in practice. You rarely hear economists talk about models being "disproven", "falsified", or "rejected". But to think that any model is appropriate in some situation seems wrong to me. There are always many more models than real-life situations. Most of those models are just never applicable or useful to any real-world phenomenon. I think economists could stand to recognize this more.
Anyway, this is a great book, and a quick read. Get it and read it if you haven't.
There seems to be a tension between:
ReplyDeleteIn doing so, [Rodrik] basically says "The evidence shows that norms often matter, and economists pay attention to the evidence." This demonstrates Rodrik's deep respect for data and evidence.
And this:
[Rodrik] says that economics, unlike science, doesn't replace bad models with better ones - it just makes new models, expanding the menu of models that policy advisors have to choose from. That seems very true in practice. You rarely hear economists talk about models being "disproven", "falsified", or "rejected".
Paying attention to evidence to add to models (or make new ones) is important, but so is using evidence to reject models. It's my opinion, but you need both to truly respect the data.
Basically, Rodrik thinks data is there to tell you which model to use in specific cases. In my opinion, data is also there to tell you which models *never* to use, ever.
DeleteNoah, here's a related post about choosing models to fit the context. The author there (John Handley) defends neoclassical models in some contexts (which sounds like it's at odds with Rodrik). Your thoughts?
ReplyDeleteAs a non-economist who's been dipping his toe into economics blogs, I enjoyed the book - it gave me some insight into a number of questions I've been asking. I'd appreciate a couple more chapters with some specifics about how to apply his thinking to various models, though - some details.
ReplyDeleteHe convinced me that economics is indeed a science, in that (in my view) science isn't so much about a search for "answers" as it is the search for better questions. I see that in his approach to models - in contrast to "public policy" economists who are in the business of making predictions, he provided a view into his world where it's more about understanding what's really going on, to the best of our ability, then asserting "inflation is coming" or "inflation is not coming."
Unlike that uninteresting yes/no question, an observation that while a certain model suggests immigration ought to lead to lower wages, say, but the empirics do not show this happening, enables an open-ended question about which assumptions are not correct, etc., and a much more useful, interesting line of inquiry.
But if you're forever looking backwards for the best explanation of what just happened, and pulling from a vast library of never-to-be-falsified models in the process (perhaps creating a new addition if none of the existing ones fit), how do you know that's not just an exercise in sophisticated just-so story telling?
DeleteMaking predictions by selecting models ahead of time (based on the context, and according to some hypothesized algorithm for model selection that anybody can use), allows you to both get feedback about your algorithm and the model, and maybe even give you confidence to shelve one or both permanently.
Failing to expose ideas to the risk of falsification seems like cutting yourself off from the only possible feedback the real world can give you.
... just to be clear, I have yet to read Rodrik's book: I'm just responding to your comment.
DeleteYes, I agree - "forever looking backwards" is not what I meant to suggest, and not what Rodrik implies. By "asking better questions" I mean essentially what you're describing - not "one model to rule them all," but a robust set of tools to describe critical assumptions and map them to appropriate models.
DeleteSo the model itself might not be falsifiable (which is why, as Rodrik explains, it's a "model" rather than a "theory") but the "better question" has to do with the rules for applying the model, which ought to be falsifiable, i.e., defining the assumptions under which the model works with increasing level of specificity.
One more point - I don't think the purpose of a model is necessarily to "predict the future" per se. If the model adds to our understanding of what's really going on in a system, that can be valuable (and falsifiable) as well.
DeleteWhat I mean is - oversimplifying a bit - even if there's no model able to say "if we implement this trade policy, it will have precisely this impact on a certain factory in Ohio," there is value in turning that around, saying OK, this trade policy went into effect, and based on the empirical data and our models we're able to form such-and-such conclusions about this specific factory.
I can't believe you've only got 9 comments on this. Is that an all time low?
ReplyDeleteI'm quite confused about the 'models can never be rejected' point. Doesn't this just invite, to use a technical term, a complete clusterfuck of models?
ReplyDeleteI don't think Rodrik says models "cannot" or "should not" be rejected, it's more that they typically "are not." What I think he's suggesting is that since a model proposes certain correlations based on a set of assumptions, it's not generally the case that the correlation doesn't exist at all, but rather that the assumptions require a set of conditions that rarely happen. So it's not "wrong" it just may not be very useful.
DeleteOne might think that the solution is to have fewer models, but extend them so each one accounts for a range of assumptions - he argues against this, though, saying the more robust, more complex model isn't necessarily "better."
Even if Rodrik does not directly endorse the clusterfuck view, is it not where his philosophy leads? And is it not essentially where we are in practice?
DeleteI'm not saying I have an answer to how and when theories in economics should be rejected. But I can't see much sense in never allowing them to be.
This comment has been removed by the author.
Delete@Unlearningecon, one possible paradigm here. I'm not saying that framework is best, just that it's an example terms of the framework and its relation to individual falsifiable models generated from it. Do you know of another example?
DeleteNoah wanted to get your view on this proposal of mine...I think I might have discovered a way to prevent regulatory capture of a regulator.The idea is simple. Start with the assumption that regulatory capture will happen and nothing you do will prevent it from happening.Then you divide the regulator into three independent wings, like the Legislative, Executive and Judiciary. Have three wings which can act as check and balances to each other. The Holy Trinity. Example in Auto Industry there are three chief regulatory captures.Industry,Workers and Consumers.Let the wings have representatives from each sector. Then we let the Coase theorem come into play, i.e. let the parties bargain with each other and you will get a more efficient result.At the end of the day, the issue is of bargaining,and you cannot capture that that information in a model,because it is in a constant state of flux.Example: when the oil prices were 120$ the bargaining would be very different when the price of oil is 30$.The consumer would be willing to pay higher taxes if the price of oil is going to go down,ask for better,safer cars at cheaper prices etc. The worker will be asking for higher wages as the cost of owning and making a vehicle around the currant market price is lower, hence greater profits,have better bargaining on their protections etc. The Industry would like to maximize the profits of shareholders, want better labor mobility, hire and fire laws,competition, trade, protectionism etc.The Government should not exist in the three mechanisms, not because of any libertarian reasons, but because even if a single team gets it, bargaining becomes asymmetric and the "game" is over.Imagine it as a game with three teams. The Government will at most have the power of a referee in a soccer match or an umpire in a cricket match.
ReplyDeleteFew seconds leaving the alternative leg in a crunched function. Alternately switch Exoslim to the alternative leg and keep it completely prolonged above the ground for a few seconds even as crunching the primary leg.For more ==== >>>>>> http://greentheorygarcinia.com/exoslim-garcinia-cambogia/
ReplyDelete