The frequency with which I’m hearing words like “unit root” and “heterogeneous priors” while walking around my Chicago neighborhood has decreased markedly, which means that this year’s AEA has gone. Below are notes on papers that I found particularly interesting among the presentations. This post should be longer, but for a reason I can never fathom, the ASSA sets rules which force papers to be submitted way before the conference, and select many papers of marginal quality. What this means is that the vast majority of work presented is either inherently dull or else is something that many of us have already read in working paper form or seen at a previous seminar. How hard would it be to give the head of each seminar a month or two before the conference to select the newest and most interesting works she can find in a given area? It would certainly make for a more interesting AEA! In any case, onward.
Moser, Voena and Waldinger: “German-Jewish Emigres and U.S. Invention.”
A ton of German-Jewish scientists moved overseas, including to the U.S. in the 30s and 40s: the Manhattan Project gang is a famous example. We often think of research worker productivity as involving important spillovers. Are high-knowledge immigrants good for domestic scientists? From a sample of 500 or so German-Jewish chemists, the authors collect data on future patents given to those who stayed in Germany (losing their academic position, of course), who moved to the U.S., and who moved to another country like the U.K. They note the subfield of each of those patents – 166 subfields in total based on patent classification – and examine the 2 million US patents in those fields. The scientists who moved to the US were not that productive ex-ante, since super high-profile Jewish scientists tended to receive university placements in the UK after leaving Germany, so selection is not a huge concern. Over the next few decades, subfields with emigres to the US see at least a 30 percent increase in patents by domestic American inventors, while there is no change in subfields where emigres did not come to the US. This holds even controlling for the attractiveness of a given area by, for example, including patents in a given class by foreigners as an regressor. Interesting indeed, and carefully done, but I can’t help but feel the numbers are just too big to be plausible. I just can’t believe that a single, average-quality, young, German-Jewish chemist can lead to an annual increase of 50 patents by domestic workers in the subfield he works in. That’s much bigger than the biggest agglomeration spillovers I have ever seen. I definitely look forward to the promised followup where the exact method by which this huge spillover takes place is investigated further. It’s also less clear to me how much we learn about modern day immigration: the German-Jewish immigrants brought over chemistry techniques that simply weren’t known by any US scientists, whereas a modern scientist from China, say, may be an expert in some area but such knowledge is probably more substitutable with current domestic worker skills.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1910247 (December 2011 working paper)
Myerson, AAEA Schultz Memorial Lecture
Roger Myerson, ur-theorist, somehow was invited to give the Agricultural and Applied Economics Association lecture this year. Recently, he has been particularly interested in questions of political theory. Myerson is incredibly bright and has, I think, a much better grasp of history than many other theorists; surely, this lecture must have been the only one at AEA to cite Xenophon’s Education of Cyrus! The basic point of the lecture, which built on a number of his recent papers, was the following: are elections enough for democratic accountability? The answer is no, and for a simple reason. Voters will only punish the corrupt to the extent that they believe the next potential president will be less corrupt. But how are they to know who will steal and who will not? Myerson argues that strong local governments – federalism, basically – are an important method for potential national leaders to show off their quality and to build reputation. There was a lot of further talk about constitutional structure, local elite buy-in and other related topics, but the above equilibrium argument was federalism was new to me and certainly relevant to Pakistan (where strong local governments have over and over been shut down by the central leaders once they were growing strong enough to be a proving ground for future leaders), Afghanistan (with a strong central leader but little provincial control), Burma (embarking on glasnost and perestroika in a state that ought be much more multipolar) and many others.
http://home.uchicago.edu/rmyerson/research/lahore.pdf (Lecture notes not available online, but here is a recent article of Myerson’s on a similar topic)
Aghion, Farhi and Kharoubbi: “Monetary Policy, Liquidity and Growth”
We often separate questions of monetary policy which affects the business cycle from questions of long-term growth in our models. Aghion and his coauthors give a simple story where it matters: liquidity constrained companies facing an accommodative central bank need to purchase fewer liquidity-granting assets because when future credit crunches will be moderated. This gives them more money to invest. This allows their sector to grow quicker. To test whether countries with strongly countercyclical policies allow quicker growth in liquidity constrained industries, they try to get around reverse causality using the same technique as in Zingales-Rajan: country-wide monetary policy versus sectoral growth, using an international dataset. The theoretical story above can absolutely be seen in the data, and in a manner robust to a number of different specifications. I’d like to see some more comment on the politics here: if sectoral growth is heterogeneously affected by monetary policy, different sectors surely will lobby differently. Also, I highly recommend seeing Phillipe Aghion present if you get the chance: he is an exhausting – though entertaining! – whirlwind of energy.
http://isites.harvard.edu/fs/docs/icb.topic1003175.files/aghion_112111.pdf (November 2011 working paper)
Chetty, Friedman & Saez: “Using Differences in Knowledge Across Neighborhoods to Estimate the Impact of EITC on Earnings.”
This was definitely the most careful and clever work I saw presented, and it’s no surprise with Raj Chetty’s name on the front page: I take it everyone agrees that if weren’t so young, he would already have his Clark Medal? Chetty and coauthors want to know how tax changes affect behavior. Many older papers look at behavior right after a tax change is enacted, but this is problematic because many people don’t react right away to the tax change, potentially because they don’t know about it right away. So the estimates from looking at the immediate effects will be different from the steady-state impact of a tax change. For instance, does the EITC cause labor supply to increase or decrease among the poor? It would be great if we knew exactly who knew about the tax change and when, since then we can avoid the above problem. This paper is super clever here: they realize that some areas learn about changes before others: perhaps urban areas have denser information networks, or tax preparers and government bureaucrats are more likely to tell you you are eligible, or the local newspaper heavily promotes the policy. Also, lots of people cheat on their taxes, particularly the self-employed since they have no workplace telling the government how much they made. For a given family size, there is an income which maximizes the EITC rebate. It turns out that you can see a huge and obvious number of self-employed people reporting precisely that amount of income – the distribution of income is much smoother for wage earners who are not self-employed. The size of this spike proxies for knowledge about EITC in a community. Chetty showed a bunch of checks here as well: people who move from low information areas to high info areas became much more likely to cheat the next year, tax preparers are probably helping spread tax code information, there is heterogeneity even at the state level, etc. With this measure of knowledge about the tax code in hand, we can counterfactually investigate labor responses of the policy once the whole country learns about the tax code change, a process that took over a decade for the EITC. Interesting technique – and it also doesn’t hurt that Chetty and coauthors were given access to over a decade of every American tax return!