Week of October 25, 2019

The No Resolution Edition

1. US Household Finance (and Great Convergence/Corrupted Economy): If you've been paying attention to global news, you have no doubt deduced a pattern that many are remarking on: mass protests in many countries that are linked in more than trivial ways to the cost of living, corrupted economies and frustration with a subverted political process: Chile, Ecuador, Lebanon, Hong Kong are just a few. Here's the New York Times on that pattern with discussion of similar situations in nearly 10 more countries.
In the sense that these episodes of mass unrest stem back to "pocketbook" issues the United States is an outlier--not that the cost of living and unequal access to opportunity aren't issues--but that they haven't yet lead to mass uprisings. There are lots of reasons for that of course, including relatively low unemployment and at least some consistent economic growth. But the underlying issues just aren't that different. Here's a new report from the JP Morgan Chase Institute on how much savings US households need to weather the typical ups and downs in their income and spending needs. There's a lot to dig into in the report, and I'm still not convinced we understand volatility enough to offer prescriptions, but this is a big step in the right direction. The report finds that US families need 6 weeks of take-home income to weather a simultaneously income dip and expense spike, and that 65 percent of households don't have that.
For a more personal take on how budgets are being squeezed, here's the NYT with a in-depth look at four households' budgets.

2. SMEs: The way I see things there are two research questions at the top of the agenda: 1) What distinguishes SMEs/entrepreneurs that grow, and create net new jobs and long-lived profitable businesses (I honestly care less about high growth because I care more about short- and medium-term income effects), and 2) What are the barriers specifically for women in becoming one of those types of entrepreneurs.
There are two new-ish papers I came across this week, one on each of those questions. First, here's a paper that tries to establish some objective criteria for distinguishing between "necessity" and "opportunity" entrepreneurs, using their prior work history as the main data source. Using data from Germany and the US they find that opportunity entrepreneurs start more growing businesses (surprise!) and that 80 to 90% of entrepreneurs are opportunity entrepreneurs. The relevance to places outside of a handful of developed countries with well-functioning and tightly-integrated labor markets notwithstanding, I don't find the approach particularly convincing. I can think of lots of different ways to conceptualize what job history means in terms of "opportunity" vs. "necessity" and it doesn't take into account that a lot of "opportunity" entrepreneurs are likely just wrong about the opportunity (or their necessity). But it's a useful paper for thinking about these issues.
The second paper is a new working paper from Seema Jayachandran that I just came across this morning, so I haven't had a chance to really look at it yet. But based on the abstract, it's definitely worth taking a look at. She "reviews the recent literature in economics on small-scale entrepreneurship (microentrepreneurship) in low-income countries" with "special attention to unique issues that arise with female entrepreneurship."
3. Digital Finance: Here are some important stories about digital finance that you may not have noticed. A major German manufacturer is still down more than a week after being hit by a ransomware attackSeventeen iPhone apps have been removed from the app store after researchers discovered they were using a clever way to hide and deliver malware. Two of the most popular VPN providers in the world were hacked recently. A new information-gathering trojan is rapidly gaining popularity with hackers, in part because it's "malware as a service" where you can rent server space and get technical support all for just $200 a month. 
You may be wondering what all of this has to do with digital finance, and I'd argue that it has everything to do with digital finance. The promise of digital finance is the potential to lower costs. But as we so often find in economics, the direct costs are not the only costs. It should be increasingly clear that we can't keep talking about lower costs of digital finance and fintech without talking about the huge and on-going costs of digital security that are part and parcel of anything digital. If the VPN companies, Apple, and major manufacturers can't keep themselves safe, why should we think that telecoms, mobile money providers, firms or individuals are going to be able to do so? At least not without a lot of very expensive help. This is a topic that really needs to zoom to the top of the list of discussions of the future of digital finance. But I still don't see it really registering (a caveat: there is apparently a day-long workshop on "addressing cyber security risks in developing countries' financial sectors" at the European Microfinance Week conference. You should go--and tell me about it since I can't get there until Wednesday night).
On a more specifically digital finance note, apparently there is discussion in Kenya of "breaking up" the Safaricom monopoly. Here's an interesting take from Ali Hussein Kassim on where the Kenyan mobile money ecosystem should go from here (decidedly not in favor of limiting Safaricom). One specific claim that is interesting but I don't know whether it's true: 80% of Kenya's transactions are still in cash.  

4. Our Algorithmic Overlords: It's unusual for a faiV item these days to really only be about one paper/story, but this one is big enough to justify it. In a new paper published in Science, Sendhil Mullainathan, Ziad Obermeyer, Brian Powers and Christine Vogeli look at a popularly used commercial algorithm for predicting the level of care needed by patients, and find that it systematically recommends less care for Black patients, reducing the number of patients who receive an enhanced level of care by 50 percent. Here's a Washington Post summary.
How? The algorithm uses past data on the cost of future care of similar patients to determine who would benefit from more care now--essentially it tries to use spending data as a proxy for how sick someone is (or will become). But, since Black patients typically receive far less care and don't have access to the most expensive and intensive treatments, less money is spent on them. The algorithm interprets wrongly interprets that lower level of spending as an indication of lower levels of illness. This is just one algorithm in the health care system, but the use of cost data as a proxy is apparently widespread. In the WaPo story, Mullainathan says, "It's truly inconceivable to me that anyone else's algorithm doesn't suffer from this."
Perhaps the most important thing to understand is that the algorithm purposely excluded race in an attempt to minimize bias. But the bias was there in the data, and excluding race, the attempt to make the algorithm colorblind, simply made the bias harder to see. Here's an editorial from Ruha Benajamin that goes along with the paper and discusses the "new Jim Code"--an entrenching of historical racism and discrimination into the algorithms that drive more and more of the options and choices presented to service providers and individuals.
My final editorial note on this is to remember that while this is terrible, the fact that this discrimination was in the code made it possible to expose (as Mullainathan has argued elsewhere). As we learn to fight this kind of bias we also have to remember that just as algorithms that are color-blind don't remove discrimination, removing the algorithms and leaving the decisions to human beings won't either.

5. More on Nobels and Tax Incidence: Taking on the easier one first: here's another useful summary of what is happening in the Saez and Zucman tax analysis and why it is generating such controversy over whether their numbers are an accurate representation of policy reality. What I like about this one is that it explicitly addresses what Saez and Zucman do as "innovation"--they are taking a very different approach to these questions and are generating different answers that are worthy of taking into account, without necessarily accepting them as the best measure of the question. In related items, here's an Open Policy Analysis document on the Warren wealth tax plan that is based on Saez and Zucman (hattip to Leah Bridle of CEGA).
Now to a bunch of additional items on the Nobel prize and RCTs: 
We'll start with a piece by Oriana Bandiera for VoxEU of the state of play. Which is a lead-in to a critique from Ingrid Harvold Kvangraven which comes from a heterodox perspective, the first of those I've linked I think. Which in turn leads to this remarkable thread from Dina Pomeranz replying to Kvangraven (keep scrolling through the initial exchanges to get to Dina's long response and reflection). Here's one of the more bizarre critiques I've seen from an economist, and a thread from Marc Bellemare responding. And while we're on Twitter threads here's one from Johannes Haushofer on the ethics of RCTs, and the problem with the assumption that phase-in designs are more ethical.
Here's a blog post from Alex Counts, which in addition to being worth a read, is a reminder of the point I made last week about the Duvendack et al critique that bizarrely suggested that Banerjee and Duflo were proponents of microcredit. As you can see from Alex's post, the industry certainly didn't see them as allies. 
Finally, to end on an upbeat note, here's a post from David Evans which provides a summary of Michael Kremer's work, not just his work around RCTs. If you had any doubts whatsoever about his merit for the prize, you won't after reading it.
I'm sure I've still missed a lot and there will be much more to come. But for now, that's it.  

I'm not widely known for sharing good news, but I try to do it when I come across it. This map based on extensive work (the author list extends to about 4 scrolls) to document Under 5 Mortality Rates sub-nationally is some of the best news there can be. Source:  Nature

I'm not widely known for sharing good news, but I try to do it when I come across it. This map based on extensive work (the author list extends to about 4 scrolls) to document Under 5 Mortality Rates sub-nationally is some of the best news there can be. Source: Nature

Return to the Weekly faiV