Week of January 7, 2019

1. The History of Banking: For a project I'm working on I've been thinking a lot about financial system development and have gotten a bit obsessed with the history of banking. You might think that with a topic so core to economic thinking there would be some consensus on things like what banks do and how they came to do them. But you would be wrong. I've had great fun reading conflicting accounts of the history of banking in the US and Germany over the last few weeks. At the AEA exhibit floor I stumbled on a new book about the history of banking in France, Dark Matter Credit. The short version is that informal banking was a massive part of the French economy, and worked better in many ways than French banks until World War I, and it took regulation to finally allow formal banks to displace the informal system. I also picked up Lending to the Borrower from Hell and just in the first few pages discovered that Italian "friars, widows and orphans" were buying syndicated loans to Charles the II of Spain in 1595. The bottom line is that informal finance was much more efficient and "thick" than I believed, and formal banking extended much further much earlier than I had known. There's also a new book on banking crises in the US before the Federal Reserve, Fighting Financial Crises, which is equally relevant to thinking about the much-more-grey-than-you-would-think borderland between formal and informal banking.
To tie this all more specifically to the AEA meetings than just what was on display at the book vendors' booths, one of my favorite sessions was Economics with Ancient Data. Though I'll confess I'm not sure whether to be heartened that things we are doing now can have persistent effects for thousands of years, or depressed that our present was determined by choices thousands of years ago.
   
2. MicroDigitalHouseholdFinance: There was of course a number of new(ish) papers on our favorite topics, further condensed here. Here's the session on financial innovation in developing countries and one specifically focused on South Asia. Some of these papers have appeared in recent editions of the faiV already, but I want to call out a couple specifically. Microcredit, I've argued, is in dire need of innovation. So I'm always pleased when I see papers on innovation in the core product terms, like this paper from India on allowing flexible repayment, and while it wasn't at AEA,this one in Bangladesh. In both cases, allowing borrowers to skip payments results in higher repayment rates and better business outcomes. I see these as part of an evolving understanding that microcredit is a liquidity-management product, not an investment product. Credit can also be a risk-management product, as long as you know it's going to be there when you need it. That's the story of this paper on guaranteed loans for borrowers in the event of a flood (in Bangladesh). Another cool innovation in microcredit. Of course, the next question is who is going to insure the MFI so that it has the liquidity to make good on emergency loan promises?
There was a session titled "Shaping Norms" that I almost missed out on because of the somewhat oblique title. There were some very interesting papers here on how household preferences get formed, and how they can be changed, including longer-term data on the experiment in Ethiopia that I think of as launching the "changing aspirations" theme that we see more and more of.
I was amused that there were simultaneous sessions on "Finance and Development" and "Financial Development" but the poor Chinese student beside me was very confused as apparently the translations in the official app did a poor job of differentiating between the two. Both had interesting papers, but I found this on the sale of a credit card portfolio from a department store to a bank (which has access to more credit bureau data) in Chile, and this on bank specialization in export markets particularly interesting.
But moving outside of the AEA realm, my confirmation bias prevents me from not including two other related items on Household Finance. First, Matthew Soursourian of CGAP has some pointed questions about the usefulness of "financial health" as a concept, questions I thoroughly endorse. Second, there is documentary evidence (for instance, here) that I've long been skeptical of the story about mothers in developing countries caring about their children while fathers don't. I find it more than vaguely racist as these stories typically only involve countries where the majority of fathers are black or brown. Anyway, at long last someone, specifically Kathryn Moeller, tried to track down one of the more common statistics on women spending more money on children and found that there is no source, and it was apparently made up as part of a marketing campaign. But that's just the start. Seth Gitter links to three studies that find no difference in investment in children (and I'll add the Spandana impact evaluation to his list) and Martin Ravallion points out that the "70% of world's poor are women" stat seems equally unsourced.
 
3. Entrepreneurship, Reluctant and Otherwise: Overall, the paper that left me thinking the most is a long-term update to the Blattman and Dercon experiment randomizing employment at factories in Ethiopia. If you need a catch-up, the original experiment had three arms: control, a $300 cash grant plus business training and a job in a "sweatshop"-type factory. While there were positive effects for the entrepreneurship group, the jobs didn't improve income and had negative effects on physical health. After five years, all the differences dissipate (hours worked, income, health, occupational choice). Pause to think about that for a moment--after several years of higher incomes from entrepreneurship, the average person in that arm shut down their business. And the control group started microenterprises and got factory jobs (filling the gaps left by the treatment arm participants who dropped out?). It's another piece of a growing puzzle about why microenterprises don't grow, or more specifically why people don't seem to invest in their microenterprises, even when the income is higher than the alternatives. Stuart Rutherford has been thinking about that too, and because it's Stuart, he went out and interviewed participants in the Hrishipara Diaries to try to get some answers.
If you're a regular reader of the faiV, you know that one of my standard soapboxes is the need to pay more attention to the commonalities between the US and developing countries. And this is anther example. At AEA, Fiona Grieg of the JP Morgan Chase Institute presented updated data on participation in the gig economy in the US (not publicly available yet, here's the older version). Of the various forms of gig work, driving is arguably the most similar to the low-skill self-employment options, which I generally term "subsistence retail", available in developing countries (indeed, that's one of the jobs Stuart discusses in his piece). In that sector, specifically, the striking finding is that participation is sporadic, irregular and incomes are falling, in part apparently because of competition but also because participants are spending fewer hours doing it. It's a pattern that looks to me much like the Ethiopia experiment, and Blattman's similar experiment in Uganda which also saw all effects dissipate after nine years. 
Here's a nascent explanatory theory, based on a new NBER paper about demographic change in the US. The authors show that all of the troubling changes in the US economy related to job creation, start-up rates and the labor share of income can be explained by the US's aging population. The basic idea is this: older people start fewer firms, particularly firms that grow and add employees, than young people. With fewer start-ups you get less creative destruction and more mature firms which tend hire fewer new workers and, at least partially as a consequence, have more unequal wages and less wage growth. Now apply those ideas to developing economies which tend to be quite young demographically. There young people are trying a lot of things to figure out what the best option for them is. Because of other market failures, the need for extraordinary entrepreneurial ability to succeed is much higher and therefore much fewer small firms grow to any size. And even survival takes a huge amount of effort, especially since there are so many other low-skill young people trying out the same things at the same time. So people drop out of microenterprises before learning enough to improve them, and then bounce through other options because none of them are particularly good. And that's what we are also now seeing in the US economy, with the gig economy as one example. The jobs just aren't good enough to justify investment. Any thoughts on this very welcome. 

4. Blind Spots and Privilege: The two things that generated the most attention at AEA this year had to do with blind spots. You've likely heard about the investigations into harassment and bullying of women by (former) superstar Roland Fryer. That gave real energy to the sessions on gender discrimination in the profession that were already on the agenda by the time the story broke. Here's video of a session featuring Susan Athey, Marianne Bertrand, Sebnem Kalemli-Ozcan and Janet Yellendiscussing their experiences being economists while female. The sessions and conversations certainly caught the attention of the news media with follow-up stories, from the NYT and NPR. The conversations have brought to light plenty of blind-spots and privilege. For instance, the AEA has not had any way to remove someone from the executive committee. There is now a code of conduct, but no mechanism for enforcing it. And the post-conference conversation on Twitter has been turning to more of the blind-spots, like the persistence of one-on-one job interviews in hotel rooms. It remains to be seen how much of a reckoning there will be. Case in point is the death this week of Harold Demsetz, an economist who, the consensus seems to be, should have won a Nobel (with Armen Alchian). The third comment on Marginal Revolution is a very credible story of years of harassment by Demsetz. But here's a Twitter thread lamenting his passing in which I can't help but notice an imbalance among the commenters who knew him personally.
OK, here's a huge pivot. The other session that inspired the most passionate response, at least as far as I could tell, was about coming changes to how the US Census Bureau anonymizes data. Here's some quick background: the ability to de-anonymize anonymous data is increasingly a concern in many areas of life; and the Census Bureau is moving toward something called "differential privacy" to make it harder to do, with unclear but probably negative effects on the ability of researchers to use Census Bureau data. Whether there are real threats to privacy and how the Census plan is being implemented are apparently deeply controversial. Here's a "live" thread from Gary Kimbrough, with follow-up responses from some of the participants, that reveals some of the tensions and problems. Something that emerges from the thread of particular note is an issue I was not aware of: Raj Chetty has more access to Census data than anyone else, apparently, and that is a source of a lot of tension among researchers. There is real concern that the Census' plan will create a hierarchy of who has access to useful data and put even more power in the hands of privileged researchers--and the extreme hierarchy that already exists in Economics is certainly a part of the culture problem. 

5. Replication and Causal Inference: OK, let's expand our horizons beyond things drawn directly from AEA. David Roodman has a new piece on the lessons from his work attempting to replicate two important public health economics papers over the last few years. Roodman doesn't see a replication crisis in economics similar to that in psychology, because "most...original results can be matched when applying the reported methods to the reported data." He thinks, though, that re-analysis is more important than replication and there economics has a "robustness" crisis.
There is a new study of "push button replication"--the ability to get the same results from the reported data and methods with the resources made available--of impact evaluations from low- and middle-income countries. Brown, Muller and Wood find that only 27 of the 109 studies they find are "push button replicable." Of those that were not, 59 did not provide the necessary data and code (similar to another paper from a few years ago that David cites); 30 of those were published in journals that nominally required the data and code to be posted. Not great, Bob
Finally, a clash of titans in the world of causal inference erupted this week, with Andrew Gelman posting a review of Judea Pearl's newish book, The Book of Why. AsSue Marquez notes, the comments are where it gets really interesting (which also got me to wondering, why are the comments on Gelman's blog must reads and the comments on Tyler Cowen's blog must-not-reads?). Pearl himself responds, (eventually in multiple places in the thread) and if you thought the culture in economics was unique, maybe not so much. Most of what is in the comment thread at Gelman's blog is statisticians. The economists got to discussing it on Twitter. I wish I could provide a useful guide to that, but the conversation got so fractured that even I was stymied trying to follow it. The best I can offer is to start with Marquez's tweet, and then click on various replies to see the conversation branching. It's frustrating but worth it, and if any faiV readers end up making sense of it (nudge, nudge Marc) and summarizing the conversation in a useful way, let me know.

There were a lot of things I considered including this week, but in the end I decided that everyone could use a smile to start the year with. So without further comment, click play.

There were a lot of things I considered including this week, but in the end I decided that everyone could use a smile to start the year with. So without further comment, click play.

Return to the Weekly faiV