Noah Smith really has it in for macroeconomists. He has recently written an article in The Week in which he claims that macro is one of the weaker fields in economics and even though there is little good work being done in macro, there is plenty of good work being done in other fields.
I think the opposite is true. Macro is one of the stronger fields, if not the strongest. At first glance it may appear to be a problematic area in economics but it is not – it is actually much healthier than most fields. [1]
A recurring theme in Noah’s article is that macro has some sort of a problem which other areas don’t have. I think this is wrong. Macro is quite productive and overall quite healthy. There are several distinguishing features of macroeconomics which set it apart from many other areas in economics. In my assessment, along most of these dimensions, macro comes out looking quite good.
First, macroeconomists are constantly comparing models to data. Now, it’s true that many of the models used by macroeconomists (that is, the way we try to understand the world) have a really tough time when they are compared to the data. Of course it is a problem that the theories are so soundly rejected but isn’t it worthwhile to make this comparison and to be candid about the results? Noah gives his readers the impression that other theories are doing much better but is this really true? In many other areas in economics the theories aren’t rejected because either the theories are never tested or the theories simply don’t exist. There are many purely empirical studies in which there is little theory to speak of. There are other areas in economics which are purely theoretical and the models in this research are rarely tested against actual data. Holding theories up to the data is a scary and humiliating step but it is a necessary step if economic science is to make progress. Judged on this basis, macro is to be commended because it so often takes this step and brings the theories close to the facts. There are other fields that make these comparisons – trade is a notable example. Modern versions of the Ricardian trade model and the Heckscher-Ohlin model are often compared with data. Similarly, labor matching models are routinely compared with data though again there are more failures of these theories than successes. [2]
Second, in macroeconomics, there is a constant push to quantify theories. That is, there is always an effort to attach meaningful parameter values to the models. You can have any theory you want but at the end of the day, you are interested not only in idea itself, but also in the magnitude of the effects. This is again one of the ways in which macro is quite unlike other fields.
Third, when the models fail (and they always fail eventually), the response of macroeconomists isn’t to simply abandon the model, but rather they highlight the nature of the failure. This is again a good research habit because mistakes and rejections have value – knowing the nature of the mismatch between the model and the data helps you to refine the theory. There are many “puzzles” in macroeconomics (the excess sensitivity puzzle, the risk-free-rate puzzle, the equity premium puzzle, the international comovement puzzles, and so on, …). At a superficial level one might be tempted to conclude that the prevalence of such puzzles shows that the field is in a constant state of disarray. In fact, these mismatches between theory and data serve as an important guide to how to modify the theories.
Lastly, unlike many other fields, macroeconomists need to have a wide array of skills and familiarity with many sub-fields of economics. As a group, macroeconomists have knowledge of a wide range of analytical techniques, probably better knowledge of history, and greater familiarity and appreciation of economic institutions than the average economist.
In his opening remarks, Noah concedes that macro is “the glamor division of econ”. He’s right. What he doesn’t tell you is that the glamour division is actually doing pretty well.
[Note 1] It is not clear that macro is a field the same way that Public Finance, Labor, Development, Trade, etc., are fields. I used to think that macro was a field but I’ve since changed my mind. Instead I think that basically every field has a component of it which is macro. So, in Public Finance you have people like Martin Feldstein, Alan Auerbach, Larry Kotlikoff, etc. These researchers work in Public Finance obviously but they also have very strong macro components to their work. In Labor there are researchers like the late Dale Mortensen, Chris Pissarides, Rob Shimer, Steve Davis, John Haltiwanger, and so forth. Again, these economists all have very pronounced macro components to their work. In fact, some of them (Mortensen and Shimer certainly) might be described most often as macroeconomists even though most of his work is on the theory of labor supply and demand. The macro component of Development, is growth, with researchers like Daron Acemoglu, Chad Jones, Debraj Ray, etc. … In Economic History you have people like Christy Romer who is obviously a very well-known macroeconomist in addition to being a historian. So, I think the proper way to view macroeconomics is not as its own field, but instead as half of all of the fields.
[Note 2] There are examples of theories in economics that do exceptionally well. What Larry Summers derides as “ketchup economics” – the application of no-arbitrage conditions to financial markets – does really well in the real world. My understanding is that if you take modern option pricing formulas and examine historical option pricing prior to Black-Scholes you find a surprising amount of agreement between the actual market prices and what the Black-Scholes formula implies. (There are aspects where the model doesn’t do as well as it should but overall it does remarkably well – if there are finance people who want to weigh-in an tell me that I’ve got this all wrong please go right ahead …). Another area in which there is a strong coherence between predicted and actual behavior, is the study of auctions. Auction theory and option pricing models are excellent examples of not only comparing theory to data but also examples in which the theory does a remarkably good job.
A further thought: It’s true that most of the theories that we have run into trouble. Noah has another recent post in which he points out that the consumption Euler equation does not fare very well in macroeconomic models. That’s true though there are more in-depth, detailed analyses in which the model does much better. For instance, there is a well-known paper by Gourinchas and Parker who use extremely detailed individual-level data together with a structural model and they show that most household’s exhibit behavior that suggests that they are liquidity constrained up to roughly age 40 – age 45 at which point they switch over and start exhibiting behavior that is more consistent with the PIH and what the Euler equation would predict. In contrast to what Noah would have you believe, this is an excellent example of a theory being refined and sharpened as a result of being compared to detailed data.
Orderstatistic 
I suspect the true state of macro is a convex combination of your and Noah’s views. I have two complaints with your upbeat take on macro:
1) Macroeconomists may use a lot of data but it is very hard to “know” much empirically. Good identification is rare in macro. We only get one run at the macro economy and we don’t (often thankfully) get to see the counterfactuals. Often micro studies can look at variation across smaller units and sometimes even run (natural) experiments. There is an incredible amount of observational equivalence in macro studies, so sensible economists can hold wildly different views about basic dynamics. All of economics is a mix of art and science, but macro above all needs the art … and sometimes we are godawful artists.
2) Which is a good segue to my other compliant: The financial crisis and Great Recession should have shaken every macroeconomist to the core. When you taught me macro, ‘we’ had solved the problems of macro stabilization … the Great Moderation. I really felt for the job market candidates who still came to the Fed in the winter of 2009 with Great Moderation papers. It was a sign of a discipline that was not only flat footed but complacent right before the macroeconomy nearly fell apart. Yes, macroeconomists (or any any economist who thinks about economy-wide, general equilibrium effects) are important … but with great ‘power’ comes great responsibility.
All that said, I agree with your points … and see Noah’s critiques more as a reason to work harder, not to throw up our hands. There is a lot to be done now and we got a lot of macro and policy variation to sort through.
(One last thing I think the macro-micro dichotomy is unhelpful … we economists are all in this together.)
Like Claudia, I’m macroeconomist in the Fed System, and I found your this post useful. I found especially valuable your observation that: “Instead I think that basically every field has a component of it which is macro. ‘’ This seems like a key point, and in part, is why I wish macro would not so routinely get defined as “business cycles”. There are many questions of a scope, ones often dealing with issues of outcomes at lower frequencies, that one could regard as “aggregate.” To my taste, this is all that makes the inquiry in question “macro.”
Beyond this, I’ve found in (more than a decade of doing macro) the business that the referees and colleagues I deal are very concerned with the real world, and typically hard to persuade re: the appropriateness of models for addressing the things we claim they’re good for. The whole thing looks “healthy” as you say—often bad in an absolute sense at making sense of the world, but where these failures reflect more the difficulty of both theory ( and empirical work as pointed out by Claudia), rather than the median macro person being a signatory to some weird pact to be mendacious.
So I’d rather that critics just *did* some macro already, and made it better instead of talking so incessantly about how bad the whole thing is. If they think its stinks, it’s probably full of low-hanging fruit, no?
“Noah Smith really has it in for macroeconomists.”
Shouldn’t have assigned so much homework, prof! 😉
Response posted, btw.
http://noahpinionblog.blogspot.com/2014/01/chris-house-sticks-up-for-macro.html
Your telling of the Black Scholes model is true. See MacBeth and Merville (1979) and Rubenstein (1985). The model worked very well, but the constant volatility assumption is violated in real life, so even early on the studies showed some of the problems. Rubenstein showed that for the average option this wasn’t economically important and the BS model worked well. Of course, a tremendous amount of work has been done on the vol smile and relaxing the assumption–jumps, time varying volatility and so on…
First, I would refer you to Paul Krugman’s post on this today, at:
http://krugman.blogs.nytimes.com/2014/01/25/none-so-blind-macroeconomics-division/?module=BlogPost-Title&version=Blog%20Main&contentCollection=Opinion&action=Click&pgtype=Blogs®ion=Body
I really hope, you’ll read it, and a response to it would be great. If there’s something wrong with his criticisms, it would be great to have a macroeconomist explain it, explain that side, in a way that’s intuitive and clear to economists who don’t have a career in “freshwater” macro, and hopefully to any well-educated laypeople. In my opinion, we don’t currently have a macro blogger who does this well. Perhaps because it’s not really possible to explain well points that are truly very wrong, but perhaps not, and you can explain this side well. I think you’d be the first in the econoblogosphere.
Next, my comments….
An absolutely crucial thing, which I always say, is that a model is only as good as its interpretation. The model may be very different from reality, and empirical evidence may show this. But using perfect logic from mild assumptions to reality while utilizing other evidence we have about the real world can give us valuable conclusions about at least some important factors in reality, at least qualitatively.
It’s this interpretation to reality that’s the most crucial part, and that takes the most true, high level intelligence. It is also this part where people can really and intentionally bias their interpretations to make their ideology look more attractive. Usually this is libertarian and/or plutocrat economists making extremely literal interpretations of their “freshwater” models to reality, because this makes government action look less desirable (And another enormous profound thing is having Pareto optimality as the optimality, rather than maximizing total societal utils, or at least considering both. Pareto optimality is virtually saying we’ll be only extreme libertarian and only make changes if there’s unanimous consent. It’s incredibly biased to use that as your only criterion.)
Finally, along the lines of my last comment, the crucial interpretation to reality, you defended the use of rational expectations in models. And I agree that in the real world people do, by and large, do some thinking ahead (at least eventually), and learn from experience (at least eventually). What this misses is, that’s not, I think, the biggest problem here. It’s the extremely unrealistic amount of public information and high expertise with which it’s assumed people do this – and then the interpretation to reality is often literal! There’s no, ok, people are going to have massively less public information and expertise, let alone numeracy, so how will that affect what happens in the real world.
If you look at the mountain of surveys, tests, and other evidence, you see extreme lacks of important public knowledge (and misknowledge as well), necessary expertise to optimize, and just numeracy, that will cause most people to make very different decisions than the perfect optimizers in the model. For more on this, including some of the empirical evidence on expertise and knowledge, I have a brief post, at:
http://richardhserlin.blogspot.com/2013/12/surveys-showing-massive-ignorance-and.html
Basically black scholes works if you don’t assume constant vol. for eg you might use vols implied from quoted premiums (implied using BS) which exhibit a smile (ie non constant in the moneyness axis). Or you might calibrate a stochastic vol model to market prices.