Monday, June 25, 2012

SED Meetings

A short report on the Society for Economic Dynamics meetings in Cyprus:

Plenary Talks This was perhaps unusual for the SED, in that the three plenary talks were pretty light on theory. Andy Atkeson's Friday evening talk was purely a measurement exercise. The idea is to measure a firm's financial soundness by "distance to insolvency." An increase in leverage or in risk will decrease distance to insolvency. Andy and his coauthors (Andrea Eisfeldt and Pierre Oliver-Weill) isolate three "solvency crises," which occurred in 1932-33, 1937, and 2008. During these crises, you can see what happens to the whole distribution of firms (ranked according to distance to insolvency). Essentially the distribution collapses - all firms get a lot closer to insolvency. Andy makes a big deal of the fact that what we see happening to financial firms is similar to what happens to nonfinancial firms during crises. He wants to question the view that financial firms are especially vulnerable during a financial crisis. I'm not sure. The financial intermediation sector can be fragile, and transmit this fragility everywhere, so that insolvency is observed in both financial and nonfinancial firms. Bailing out the financial firms because they are viewed as "systemically" important may be wrong, but I don't think the work of Andy and his coauthors necessarily suggests that.

The SED has become a very large meeting, with 12 parallel sessions running. That's about 450 papers presented over three days. In spite of its size, though, the conference has retained its midwestern sensibility. The papers are mainly in modern macro and structural applied micro. However, one of the plenary talks each year is typically devoted to a presentation by someone outside the tribe, and this year's was by Christopher Udry, from Yale. Udry talked about work on field experiments in development that related to some of the financial frictions mechanisms that macroeconomists like to think about. Udry characterized the results as mixed - in some experiments it appears that credit frictions seem to matter, and in other cases not. Here, I think the development experimenters and the macroeconomists could benefit more from talking to each other. In some venues, I think this is happening already. For example, Rob Townsend at MIT has made attempts to get the two groups together. One benefit from cross-fertilization would be the integration of more serious theory into the design of field experiments and the interpretation of the evidence. In particular, it wasn't clear that some of the field experiments Udry discussed could actually tell us much about the role of credit and financial arrangements in the economy.

Finally, Monika Piazzesi presented some interesting preliminary work, joint with Juliane Begenau and Martin Schneider, on measuring bank risk. They focus in particular on making inferences about derivative positions, which are in principle difficult to measure. Some of the results indicated that the derivative positions of large banks were increasing risk rather than reducing it. Not sure how we think about this in a systemic context.

RBC is not dead. We would have to go back years to find anything that would resemble a baseline real business cycle (RBC) model in a paper presented at the SED meetings. This year's crop includes plenty of models with heterogeneous firms, heterogeneous consumers, banks and other financial intermediaries, search frictions, etc. A common view of the recent recession is that the standard representative agent RBC model does not fit the facts, particularly with regard to the behavior of labor productivity. However, this paper, by McGrattan and Prescott makes the case that we can solve the "productivity puzzle" by thinking about measurement error in the national income accounts. McGrattan and Prescott argue that the key mismeasurement involves intangible investment. What's that?
Intangible capital is accumulated know-how from investing in research and development, brands, and organizations, which is for the most part expensed by companies rather than capitalized. Because it is expensed, it is not included in measures of business value added and thus is not included in GDP.
The argument is that intangible investment is a significant fraction of correctly-measured GDP, and that it is volatile and procyclical, just like tangible investment. MacGrattan/Prescott claim that intangible investment helps us understand both the 1990s boom and the 2008-2009 bust as TFP-driven. You may think that factors other than TFP are important for business cycles, or that TFP is some kind of reduced form for those other factors, but people with alternative ideas need to be as serious about the data as MacGrattan and Prescott are. Popular discussions about the role of Keynesian phenomena in the recent recession are, in this respect, particularly loose.

11 comments:

  1. Here's some evidence for intangible capital:

    http://oz.stern.nyu.edu/cite05/readings/brynjolfsson3.pdf

    ReplyDelete
    Replies
    1. An important point is that including the intangible investment makes the recession look even worse. As is, you might think that the United States is at least highly productive (though measured increase in labor productivity might just be a composition effect), though we're not producing very much. With the intangible investment included, production and productivity are both pathetic.

      Delete
    2. "With the intangible investment included, production and productivity are both pathetic."

      Any chance business people aren't investing due to a known lack of aggregate demand?

      I thought not.

      The modern business man has such a keen, analytical mind, that she can look to the future and will invest and pay taxes for more defense and a war with Iran . . ., but will not invest and pay taxes to finance a Keynesian stimulus.

      We are now all lead by Russ "we have a shortage of savings" Roberts (paradox of thrift, never heard of that) and Laffer, of which SW is a leading disciple.

      SW, when is inflation going to kick in?

      If only you had to pay fees and commissions to short Keynes. You would be feeling the pain by now.

      Delete
  2. the advantage of their approach is they use techniques that were developed in other papers before the crisis. there is no after the fact redesigning to protect the core hypothisis from refutation

    ReplyDelete
  3. 'A common view of the recent recession is that the standard representative agent RBC model does not fit the facts, particularly with regard to the behavior of labor productivity. However, this paper, by McGrattan and Prescott makes the case that we can solve the "productivity puzzle" by thinking about measurement error in the national income accounts.'

    So the argument is that if the theory doesn't fit the data, something must be wrong with the data. Shouldn't be so hard for people with alternative ideas 'to be as serious about the data as MacGrattan and Prescott are.'

    ReplyDelete
    Replies
    1. You're not understanding the point (possibly because you think you're being clever by being deliberately disingenuous, but I'll give you the benefit of the doubt here). The problem isn't that the data is wrong, it's that it is the wrong data. If we systematically exclude one kind of output from GDP, we obviously miss something about the economy. Ed and Ellen show that this omission matters for our measurement of productivity, which is constructed using the flawed measure of output.

      Delete
    2. As Ed would say, economics is about the interplay between theory and measurement. Measurement helps you come up with better theories, and sometimes theory can direct the measurement in useful ways. This is a case of the latter, I think.

      Delete
    3. I think this applies to science in general.

      Delete
  4. Which is the best session? I vote for Violante

    ReplyDelete
    Replies
    1. You mean the macro/labor session, right? Yes, I liked that one.

      Delete
  5. Didn't get much from the plenary addresses. I thought they were weak by the standards of past addresses.

    ReplyDelete