Week of May 2, 2019

The Workers of the World Unite Edition

1. Microfinance/Household Finance: I mentioned the Hrishipara Financial Diaries last week--it's a project Stuart Rutherford has been running in central Bangladesh for four years now. That's a truly unique data set of high frequency data on the financial lives of households. I also mentioned that Stuart is now funding the continuation of the diaries out of his own pocket. Don't make me beg for someone to step in with more funding so this dataset gets even more valuable. It's incredibly cheap by the way---hmm, maybe the first faiV GoFundMe? See, don't make me resort to such things!
Continuing in the wave of revisiting ideas about microfinance and it's impact, Bruce Wydick has "3 reasons the impact of microcredit might be bigger than we thought." Of course, the "we" in that sentence matters a lot. Mushfiq Mubarak and Vikas Dimble have a short review of microfinance research with handy links to the research we talk about most these days: evidence for ways that microfinance could innovate to increase impact. Of course, I have to return to the binding constraint on microfinance innovation: funding appropriate for investment in innovation

2. Replication: I know what you're thinking: "Hey, I haven't heard about Worm Wars in a long time. What happened?" And so, let me bring you a new paper from Owen Ozier that reviews the history of the Worm Wars in an effort to understand the state of reproducibility in Economics and related topics. Here is Owen's Twitter thread with some "wild things" he learned working on the paper. And here's Annette Brown's replies (onetwothree) pointing out some longstanding errors in the literature on replication in economics--one lesson is that if you don't read the variable definitions you're likely to draw the wrong conclusions and others won't be able to replicate your work.
Here is an interesting argument that theory constrains degrees of researcher freedom more than experiment--that in fact one of the sources of the replication crisis is a lack of theoretical frameworks around empirical research. Oh, and that empirical work needs more formal mathematical models. In case you haven't figured it out yet, this is coming from the perspective of "behavioral sciences" which apparently does not include economics, where alot of recent argument has been about the need for experiments to constrain degrees of freedom and that "mathiness" is a problem. And here's Dorothy Bishop on "reining in the four horsemen of irreproducibility".
Inherent variability is not one of those four horsemen, but it is a plausible source of irreproducibility that has nothing to do with bad practices or researcher misbehavior. If reactions to stimuli vary a lot based on minor contextual factors (which is in fact one of the findings of behavioral sciences, albeit one that is itself subject to lots of questions about replication), then you should expect that the exact same experiment conducted at a different time and place with different subjects will yield different results. Whether that's the case is the subject of this debate between Simmons/Simonsohn, McShane/Bockenholt/Hansen (not that one), and Judd and Kenney (also not that one), all hosted by Andrew Gelman. It's worth the time to read through.
3. Research and Communications: Taking that conversation as a leaping off point, here's a new paper on demand effects in survey experiments. On the one hand, it may come as a relief to know that the paper doesn't find much evidence of experimenter demand effects. On the other hand, a lot of economics lab experiments are built on the idea that the experimenter can induce people to behave in certain ways with incentives--and when those incentives don't work, it's evidence of some other important factor operating. But, "Even financial incentives to respond in line with researcher expectations fail to consistently induce demand effects." I feel like this paper could not have been published in an economics journal, because the theory constraints (I'm particularly proud of this callback).
In other backed-up research methods links, I've carried around an open tab for this very useful post from Berk Ozler on alternatives to recruiting a control group for more than a month. As usual Berk lays out the issues and questions, and there's bonus follow-up via Susan Athey, linking to some other recent papers on related issues that I've also been carrying around in open tabs, so I'm feeling good about slaying all those giants at once.
The questions Berk is asking and the responses from Susan stirred something deep in my memory bank and led me back to this 2014(!) post from David McKenzie onwhether the impact evaluation production function is best understood via O-Ring Theory or Knowledge Hierarchy theory. And it seems to me increasingly like the answer is O-Ring.
Finally, there is another part of the production function: communicating the results of the work. That is a place where it seems the dominant model is Knowledge Hierarchy--leave the communications side to the comms experts. (OK, now I have to pause for moment and figure out if that explains the faiV or the faiV is contradictory evidence...). Here is David Evans making an argument that becoming a better communicator should be high on the list of priorities for economists. And here's a paper that finds that high quality communication is contagious, so there are positive externalities if you follow David's advice.
I'm going to confidently predict though, that this last link on communications is going to get the most clicks this week: Bullshitters: Who Are They and What Do We Know About Their Lives? 

4. US Inequality: Here's another topic I've been neglecting of late. And there is some good news: wage growth is finally picking up for the bottom half of the distribution. And minimum wages may be the highest they have ever been.
On the other hand neither of those two trends are going to make a meaningful impact in the racial wealth gap. Here's an essay on that topic which I broadly agree with but have some quibbles. Particularly the statement that "there is no buying your way out of racism." On the individual level that is obviously true, and the systematic worse outcomes for people of color no matter what their wealth is powerful evidence of it, even if you aren't persuaded by the numerous stories. On the other hand, from a historical perspective, it seems to me that the only two things that have worked in dissolving societal racism are a) integration during war, and b) economic growth/power. But I'm very interested in research on this question if you know of it. On that point, here's a discussion of the findings in the most recent GSS of significant positive shifts in racial attitudes.
One of the main policies that "works" to fight poverty in the US is the Earned Income Tax Credit. I have a separate issue with how we talk about the EITC "lifting people out of poverty," when it is delivered in a lump sum and most of the recipients live below the poverty line for 11/12ths of their year. It's almost as if people in the US have no sense of seasonal poverty. But that aside, the widely held idea is that part of the anti-poverty impact of the EITC is encouraging people to enter the labor force. Here's a thread on some emerging research that questions that widely held view.
As the Trump Administration tries to have courts cancel the ACA, here's a reminder that the health insurance plans low-income people have access to through work are much worse than the plans they have access to via the ACA marketplaces. And Texas appears poised to waste an enormous amount of time and money limiting what foods SNAP participants can buy.  

5. Philanthropy: It's late in the day so I'm going to wind things up here quickly. The Notre Dame donations have kicked up the flames of a newly revived debate about the role of philanthropy, especially large scale philanthropy, in democratic societies. Here's some historical perspective on what that debate used to look like, and it involved riots.
Here's Rob Reich, political scientist, on the corruption of college admissions philanthropy and the idea of auctioning admittance to elite schools. And a conversation about the ethics of big donations to arts and culture, and the Sacklers. Here's an interview with Melinda Gates. Here's a perspective on what is next for the Hewlett Foundation, one of the larger US foundations giving to global causes and an incubator of sorts for a lot of the "better philanthropy" and evidence-based policy efforts, as Ruth Levine steps down.

Breaking my own recent rule, here's a fun thing that actually relates to faiV themes. If you can guess which coffee maker I have in one of my offices, you win a free annual subscription to the faiV.

Breaking my own recent rule, here's a fun thing that actually relates to faiV themes. If you can guess which coffee maker I have in one of my offices, you win a free annual subscription to the faiV.

Return to the Weekly faiV