Sunday, September 19, 2010

Thoughts on Empirical Austrian Economics

Mario Rizzo has up what Peter Boettke has called a "rallying call" to Austrians. There's a lot of intro stuff about how Keynesianism is a "religion", how we "rationalize, complicate, and immunize against criticism", and how Keynesianism set us on the road to serfdom - etc., etc.. I still find it really disappointing that these smears and suspicions still go on. I take cracks at Austrians, of course. And certainly I don't prance around the issue when I find an Austrian point unconvincing. That sort of thing is fine, but this assuming the worst about the other side gets really old. Anyway - I want to get that out of the way because it's laid on pretty heavy in Rizzo's post. But then, if you just skip down to the beginning of the seventh paragraph of the post you get to the really important stuff that Rizzo brings up:

"Of course, the consequences both for policy and the future of economics depend on the interpretation of the financial crisis and the Great Recession. What caused them? What policies are conducive – or at least do not inhibit – recovery. (My late colleague Ludwig Lachmann used to say, “People no doubt learn from experience, but what do they learn?”)

Austrians failed to carry the debate within the economics profession and among the public intellectuals and economic historians after the Great Depression. Will they once again?...

[more "Keynesianism is a religion" stuff]

... Yet there is a critical deficiency. We continue to lack empirical work, on a large enough scale, to convince other economists that we have something relevant to say. The macro-economic framework has created a demand and supply for certain kinds of aggregated data at the expense of data that might be more useful to Austrians. (But I am reminded that George Stigler used to say, “It is no excuse to say the data are not available – you just must be clever.”)

This is where, perhaps, those non-Austrians with a similar mindset may be very important. We need good empirical researchers. I am, quite frankly, not interested in reviewing all of the qualms about certain kinds of econometric work. No single econometric result is definitive but little by little a case for taking a theory seriously can be built.

As I have said many times, I am not a macro or monetary economist. I entered into to all of this discussion as a political economist. I saw (and still see) the fate of free institutions and decent economic policy in the balance. With a little bit of luck, lots of hard work, and a smart sense of making intellectual alliances, we can do better than Ludwig von Mises and Friedrich Hayek did during the Great Depression and its aftermath. We have their legacy as well as the new legions to make the case."

I agree whole-heartedly with the point that the area where the Austrians could make the greatest inroads is by taking up empirical work. Notice what guys like Krugman (sorry, he's the best example despite his lack of currency in your community) say when they dismiss what he calls "liquidationism". He gets the theory wrong, of course - but he doesn't say readjustment doesn't make sense as a concept (what economist would?) he says he doesn't see it in the data. When I discount regime uncertainty arguments which have been virtually adopted as a part of the Austrian school, I don't say that it doesn't make sense. In fact I claim it makes a lot of sense and it is certainly one factor. I just say it doesn't seem to show up as a primary element in the data. Some Austrian arguments are a little unfamiliar to people, but the critique usually isn't that they're crazy or wrong-headed. It's that they're interesting and plausible but don't seem to be a major feature of what we see.

Which begs the question - is that true or not? It's not like people have been doing rigorous empirical tests on behalf of the Austrians or anything. Usually the "I don't see it in the data" claims are just eye-balling the data sort of statements. So there's a lot of potential work to be done here.

Several times before (including in the comment section of Rizzo's blog) I've advocated a specific empirical project that I think would be especially fruitful for Austrians or non-Austrians interested in the Austrian school (like me) to pursue. I think there needs to be a reunion of sorts between the Kiel school and the Austrian school. The Kiel Institute of World Economics was founded in 1914 in Germany, but the in the crucial pre-Nazi years (1926-1931) it was headed by Adolph Lowe. Lowe's research on the business cycle focused on multi-sector readjustment and endogenous business cycle theory (as opposed to the previously disparate theories of economic equilibrium and the business cycle). His work would go on to inspire two towering Nobelists that took the insights of the Kiel school into two very different directions: Wassily Leontief and Friedrich Hayek. Hayek's work on multi-sector understanding of the business cycle doesn't need elaboration. Leontief, for those of you who weren't aware, developed the now famous input-output tables produced by the Bureau of Economic Analysis and I would assume many other statistical agencies across the world. IO tables are matrices that detail the inputs for different industries from other industries and the subsequent products produced by those industries (or the subsequent industries where their outputs are used). From a policy perspective, the IO tables were instrumental in economic planning. But from a theoretical perspective the revived Walrasian general equilibrium theories most famously in the work of the Cowles Commission.

That's all history, though. What is most relevant, I think, is that the Bureau of Economic Analysis publishes tables of the structure of production and the production and use of progressively higher order goods. My empirical analysis of Austrian business cycle theory would look something like this:
1. Take the columns of a "Use of Commodities" table and scale them by the "sum of final uses" column minus the "private investment" and "inventories" column. This is the y-intercept of the Hayekian triangle.

2. Subtract out value added by these scaled sectoral contributions. This leaves the value of the intermediate goods used as inputs.

3. Use the IO table to allocate the scaled inputs left after subtracting the value added to the various sectors supplying inputs to final goods producers.

4. Repeat steps 2. and 3. until subtraction of the percent of output that is value added reduces output to zero.
This iterative process should leave you with a rough view of the capital structure. It will incorporate the complications that Garrison and Hayek mention (that different industries can appear in different stages). It will also produce the value of output at each stage (so you don't have to assume a linear Hayekian triangle), and the sectoral composition of each stage (this way you can attach things like labor market information to each stage of production to test ABCT labor market theories). I see one major problem with this method: you don't have a measurement of time. You can get several iterative stages, and you can get the incremental reduction in output for each stage, but you can't get the incremental increase in time for each stage.

I don't think this is a major obstacle, although it's certainly a caveat to include. The way I read ABCT, a lot of the lengthening of the capital structure is due to compositional shifts. Producers rely on higher and higher order goods, they don't just take a longer time working through the same production process. If that's the case, then the lack of a specific time increment for each stage of production in my method shouldn't be too problematic.

So what to do with this? Well, the BEA produces this information for 1997-2008, which is a good place to start. I'd simply map out a Hayekian triangle for those years and see how it shifts over time, to begin with. Then I'd try to see whether changes in the interest rate influence the shift. After that, I'd see how labor markets at each stage change in response to changes in the Hayekian triangle.

Ultimately, ten years of data is not nearly enough to establish an empirical relationship between changes in the interest rate and changes in the capital structure. If ABCT is right you oughta be able to see it pretty clearly in the 1997-2008 period - and that's good - the problem would be that that would remain more anecdotal without more data. The BEA produces benchmark IO tables back to the forties, but they only did them every couple of years. You could impute intermediate years, but that may be harder. The data might not be high frequency enough to look at a period beyond 1997-2008. The next place to look would be other countries that more regularly collect this information. The problem is, the countries that were on the ball collecting this information were usually countries where central planning was going on, which makes them poor test cases for ABCT dynamics.

That's my idea - it's been stewing for a long time. When I dangled the idea out on Rizzo's blog in this older post it didn't get any bites. Maybe people will be more interested now.

If any young scholar (or hell, even a more established one) wants to collaborate with me on this and write a paper on it, I would be happy to. I have a pretty busy fall, but could at least get started - and then the spring definitely looks clearer. Empirical work is the way to go. As I've said in the past, I (as a Keynesian) think Austrians overstate the extent to which a lot of their stuff is really rejected by the mainstream. All the market process stuff, all the subjectivity stuff, all the efficiency of free markets stuff maybe emphasized or phrased in slightly different ways, but there's really no monolithic conflict with the mainstream on that stuff. The biggest difference and biggest contribution is ABCT. It's worth testing. That let's you understand the veracity of your theory, perhaps improve upon it, and it also lets you convince others and most importantly assess how important it is for actual economic downturns.


  1. This is the best thing I've seen you write on the internet. Cheerio, from an "Austrian" who would like to remain anonymous.

  2. Nice idea.

    Any takers yet? Maybe they sense a Trojan Horse!

    Timeo Danaos et dona ferentes...

  3. Haha - no Trojan Horse, and no takers either.

    I'm open to criticism that this is a bad idea too. I've already identified at least one problem with it. I don't think the existence of caveats makes it a bad thing to pursue.

  4. but then again... if there WERE a Trojan Horse would I say that there was?!?!?!!?!?!

  5. Seriously, though... I'm not brilliant enough to be that dastardly


All anonymous comments will be deleted. Consistent pseudonyms are fine.