1. Social Investment: You've of course seen many stories about the US college admissions bribery scandal. And if you pay any attention to the world of impact investment you likely have seen that Bill McGlashan, the very public face of one of the world's largest impact investment funds, was one of the people arrested for participating in the scheme. Anand Giridharadas, who has become the very public face of criticism of modern philanthropy and social investment, discusses why McGlashan is "the most important fish" in the story. Here's the Twitter thread versionif you prefer that over a 4 minute video.
Trevor Neilson, co-founder of the Global Philanthropy Group, says that McGlashan's behavior should not be seen as a reflection on impact investing as a whole, because...well apparently because he wrote a Medium post saying that it shouldn't. There's really no argument there other than "Our goals are too important to be worried about means!" if you consider that an argument. Here's Jed Emerson, who may have an argument, but I just don't understand what is happening in this piece. Lauren Cochran, managing director of an impact investing firm, actually has a few arguments attempting to make the same point, including that McGlashan himself was a figurehead chosen to attract investors, but who wasn't involved in actual investment decisions.
She has a nice line about Giridharadas: "using one man’s ethical failings to grab the mic is characteristically self-serving, but as usual, he forgot that there might be a baby in the bath water." It's catchy but wrong. Giridharadas whole point is that there may be a baby in the bath water, but the bathwater is toxic and everyone will be better off, even the baby, if you toss the whole thing. Moreover, the fund that Cochran administers uses this language: "dual expectation of best-in-class financial returns and maximum positive social and environmental impact." And that, to me, is a big part of the toxic nature of the current impact investment environment. On reflection, that statement illuminates what is really happening in Neilson's piece--the fear that if the myth of "no tradeoffs" is exposed then the money will dry up.
To be clear, I'm not in Giridhradas' camp but I certainly appreciate how his perspective keeps putting the "no tradeoffs" crowd on the defensive, and illustrates the inconsistency if not hypocrisy hidden there.
Kristin Gillis Moyer of Mulago points to a terrific example of the inherent tension: the new Catalytic Capital Consortium funded by MacArthur, Rockefeller and Omidyar. It aims to invest in businesses with low profit potential and/or high risk. I find it an incredibly refreshing approach--it explicitly acknowledges that the no tradeoff myth is leaving many social enterprises in the lurch. But as Gillis Moyer points out, it's not clear how catalytic it can be since there are unlikely to be that many other investors chomping at the bit to invest in low-profit, risky businesses. I'd like to think the catalytic part will be creating space for more funds and investors to say that they prioritize impact over financial returns, and that's OK.
2. Our Algorithmic Overlords: Because the faiV was so full I'd been holding on to a few things on this topic, and events have made them all the more relevant. Platforms for open sharing seemed like such a good idea for a long time. But the cost of open sharing is so so much higher than most anticipated. Not only does it enable evil, but attempting to stop evil exacts a huge toll on human beings. This is a story about the Facebook contractors whose job it is to stop the New Zealand murderer's live stream. And a Twitter thread from someone in a similar position at Google. I'm guessing many of those folks are inching toward Calvinism.
Evgeny Morozov has a different take on the costs that open platforms and big tech exact, and why the global white nationalist movement has very different views on that front. It is a helpful reminder of the costs of the old system and the structures that the liberal order created to try to limit those costs, structures that seem to not work so well in this age, and are under attack from many directions. That's in part the theme of a new book reviewed by Noah Smith, The Revolt of the Public by Martin Gurri. I haven't read the book but the review is certainly influencing my thinking on the above.
Oh, and Chinese firms are working on facial recognition of pigs, while US police forces are using bad data to train their facial recognition and other AI systems. Andwhat about "behavioral recognition"? Note that this has quite obvious connections to the use of psychometrics and other "alternative data" for creditworthiness evaluations.
3. Household Finance: There's a huge amount of new stuff here, so I'm going to be particularly eccentric this week. There's a lot more coming in the following weeks that will be more serious.
One of the questions that fascinates me these days is what is good financial advice for households that face a lot of income volatility. The foundation of virtually everything in the financial advice world is the lifecycle model--and we know that doesn't apply to a very large proportion of households. That doesn't stop the financial advice industry from thriving--but like so many other things, the internet has disrupted that world a great deal. And that disruption creates perverse incentives. Here's the story of the "Fall of America's Money Answers Man", a once-respectable financial advice columnist who turned into a con artist.
Advice on how to retire early by spending virtually nothing (while having a high-paying job, natch) has been growth industry. Here's a personal narrative from a Vice columnist who tried to follow the advice and decided the misery wasn't worth it.
Here's a new paper on the possible connection between credit availability and depression (the mental health kind). It finds that increased availability of credit to firms leads to less depression among low-income households. I'll note that this kind of paper is what made the RCT movement so attractive (see below).
4. Research, Methods, Evidence: I was at a conference in Paris for a new book on RCTs and development economics this week, part of my travels. Drafting my chapter for that book turned out to be much more difficult than I had anticipated--the useful ways of saying something on this topic are much more limited than you might imagine. One thing that became clear to me, probably far later than it should have, is how often argumentation in research methods follows a pattern of: "Individual A made Proposition P at t1. Proposition P is wrong in context X. Therefore Group G is wrong at t2." That's a hard construct to argue with constructively. The other thing that became clear to me was that it would be very useful to have a more structured (in the economic sense) story about the use of RCTs in development economics. I plan on doing that in the next draft of my chapter, but while I was in the midst of pulling a near all-nighter in France to finish my draft before the conference began, Susan Athey produced an inadvertant history of the rise of RCTs in a single tweet: "Just think the most effective way to evangelize a new method is to demonstrate its effectiveness in a first-rate empirical application where the method clearly leads to a better quality and more credible result. Researchers will mimic a fully worked out, successful example."
That tweet was part of a "conversation" with Judea Pearl about Directed Acyclic Graphs, Pearl's preferred method for approaching causality. If you know anything about Pearl, you now why conversation is in quotes--if you don't, the whole thing begins with Pearl wondering why economists don't care about causality, as evidenced by the fact that they don't use his DAGs. If you, like me, don't really understand DAGs, here are a couple of useful tweet threads: one for those who don't mind the use of animated GIFs to provide pointless meta-commentary, andone for those who do. Just to be clear I recommend the second one which is from Scott Cunningham. Scott makes a reasonable case for the utility of DAGs--but Susan's point still stands: when someone/s start publishing papers using DAGs that are higher quality and more convincing than current practice is when their utility will be proven. And then they will quickly become ubiquitous.
Scott also pointed me to a very useful tutorial on another tool making headway in research practice: GitHub. I'm trying to wrap my head around the possibility of using GitHub for the kind of writing I do, which is often very iterative and splinters off into different directions. If anyone has used GitHub, or any other tool, that way let me know.
Here's a thread from Beatrice Cherrier on the historical debates within economics of the role of theory and data. It's worth reading for the reminder of how often the basic issues in these debates repeat. I'll be drawing on it as part of my discussion on the rise of RCTs.
Finally, here's a fun little exercise showing how bad humans are at randomizationeven when we are trying our damnedest to be random. I fully suspect someone is going to respond to this by referencing Fisher vs. Student on the value of randomization.
5. Management and SMEs: I freely admit that management, particularly in the case of SMEs and development is something of an obsession of mine. Did you remember to click on the review of evidence on management from a few weeks ago? Here's a newish paper that looks at the determinants and consequences of management practices among SMEs in Ethiopia from Abebe and Tekleselassie--of note, Ethiopians working at an Ethiopian research center. They find, consistent with the other literature that good management shows up in productivity, is distinct from human capital, and is a learned skill.
Here is an overview of two recent reports on SME financing in developing countries, that unfortunately uses the "missing middle" concept. I'm quite sympathetic to these efforts, particularly one of the reports segmentation of business types, but I generally think these things are premature. We know very very little about how these small enterprises run on a daily basis, and designing "solutions" for them before we have a better handle on that doesn't seem optimal to me. That being said, it is striking that the conclusions of both reports are essentially exclusively about improving financial systems not about interventions targeted at the firms. That's a welcome change.
In terms of better understanding small firms, there are people working on that. I didn't get to go Oxford CSAE's conference or even pay much attention to it as a consequence of my trip to Paris, but there were a number of papers on the topic that I'll be trying to catch up on in the coming weeks. For instance, an experiment on equity-style investment in microenterprises in Pakistan. Here's one on spillover effects on micro and small enterprises of infrastructure investment. Here's more evidence on heterogeneity of impact of business training and credit on micro- and small enterprises, this time in Ethiopia. The operative differences here being gender, but I think we can safely say at this point that gender, opportunity and aspirations collide (to borrow a Pearl term). And here's more reanalysis of the de Mel et al capital grants research, modeling TFP and learning effects to explain differences in outcomes and capital accumulation. But my favorite example of our collective ignorance is this paper about whether electricity shortages and outages induce firms to innovate more or less. The results aren't particularly convincing to me, but the question is important: on so many dimensions we should be very humble about what the constraints to firms are and what decisions and choices those constraints lead to.
If anyone is interested in funding a very (very) small scale, and possibly idiosyncratic experiment on small firms, technology, productivity and management in order to generate some better hypothesis on these topics, let me know.
Someone recently created a new tool for creating these time-lapse bar charts--that's the only explanation I can find for why they have suddenly been showing up everywhere. Here's one on global cities' population that is pretty interesting. And here's an explanation and one tool for creating them. But in keeping with my effort to keep these a bit lighter, here is one on leading goalscorers by age.
