Week of November 13, 2017

The Trolling Edition

Editor's Note: In this week's edition I'm using the world "trolling" broadly, and less negatively than it is often used. I'm using it to describe pieces that raise the ire or the fears of readers, not to suggest that the ideas are purposefully wrong or misleading. 
Thankfully (pun intended) I have some time to come down from the ire that writing this edition produced in me. There won't be a faiV the next two weeks because of vacation and travel, though I may produce a special edition coming out of European Microfinance Week which I'll be attending. If you're there, feel free to troll me--Tim Ogden


1. Our Algorithmic Overlords: I saw someone joke on Twitter recently that the best way to do a literature review was to complain on Twitter that "no one is studying..." and just use the incensed replies that come pouring in. It's an interesting form of trolling. This week Cathy O'Neil, author of Weapons of Math Destruction and perhaps better known as mathbabe.org, had an Op-Ed in the New York Times saying, "Academics aren't paying attention to our algorithmic overlords." Of course, I agree with the need to pay attention--hence this regularly featured topic--but it's a curious framing that academics aren't paying attention. In fact, all of the examples she gives of areas where academics need to be paying attention to algorithms and their effects are areas of intense academic work. Say the use of sentencing algorithms. Or teacher assessments. Or dynamic scheduling.  
And it's not just the specific instances. There's also work on the big picture of the use of algorithms and big data in policy making. Or simply understanding how companies will approach gathering and using data and algorithms (How could I not link to a paper so excellently titled as "Seeing Like a Market"?). Or how about a whole academic center "examining the social implications of artificial intelligence"? The conspiracy theorist in me couldn't help noting that the center, at NYU, was officially announced the day after O'Neil's op-ed which proposes an academic center, though they have a 2nd annual report on the use of AI and 10 recommendations to guide research and accountability.  
I can hardly be opposed to academic research centers, but it seems to me that what's missing is not academics paying attention or research centers devoted to the topic, but a Consumer Algorithm Protection Board. Yes, this is a pipe dream given the dire outlook for the Consumer Financial Protection Board, but it is a pipe dream I'm particularly fond of. Anyone want to help me make the case for it?

2. Household Finance: Before the algorithmic overlords item gets ridiculously long, let's move on to something that could fit either under algorithms, protection boards, or household finance. Entrepreneurial Finance Lab, which uses psychometrics to assess creditworthiness, has a piece on the FICO blog about how their testing for personality traits like impulsiveness and delayed gratification predicts default rates. It's such a good example of why I've been a fan of EFL while being queasy at the same time, it almost felt like I was being trolled. On the positive side it's an operationally relevant way of assessing borrowers who otherwise would be shut out of access to credit. On the queasy side, there's apparently huge variation in different cultures (while the metric remains predictive), and real questions about the immutability of the features they are testing--which cuts both ways. If they're mutable there's a question about what we are measuring; if they're immutable, what do we do about people who lost the "present bias" lottery? It's a good thing to protect people from themselves by not offering them credit they are likely to default on, but it still leaves me queasy nonetheless.
In terms of others being trolled, here's a piece about Refinery29's ongoing series where women share a week-long financial diary, and then readers rip apart their life choices. I'm not entirely sure whether it's the ones sharing or the ones critiquing that are the trolls, perhaps both.
And since we're on the topic of diaries and I've already gone fishing for help on one research interest of mine, here's another: I read this week that more than 100,000 puertoriquenos have migrated to Central Florida since this fall's hurricanes. Wouldn't it be great to do financial diaries of those households? It's a really unique opportunity, wouldn't be very expensive to do, and it breaks my heart to see it go to waste. If you think so too, call me.

3. US Inequality: The big news this week in the US is the Republican tax plans. Here's Saez and Zucman's take. And here's Tyler Cowen arguing that argument that corporate tax cuts will lead to more investment is empirically sound (which some will undoubtedly believe is trolling on Cowen's part).
Beyond the current tax issues, here's a new paper that attributes growth in income inequality to relatively small-scale business owners (as opposed to corporate executive pay or passive investment income). And here's a thought-provoking post about a different impediment to fighting extreme inequality: the very high cost of investing in low-income neighborhoods because of land-use restrictions. And on that theme, this piece from Felix Salmon on who gets hired to lead organizations is also an important part of the story on inequality and mobility. 

4. Evidence-Based Policy and Methods: One could easily argue that this new paper from Alwyn Young is trolling the entire community of economists who "grew up" in the age of instrumental variables. It certainly has generated a lot of concerned responses about the need to reassess a lot of work relying on IV. Nancy Cartwright, meanwhile, has been similarly trying to punch holes in people's faith in the use of RCTs for guiding policy (and I'm sure some in the movement would think of her in trolling terms). Here's a new paper (with co-authors) on the challenge of moving "evidence-based policy" from ideas to local implementation
I know a lot of MFI executives who have felt trolled by the research community over the years. But for anyone who is ready for more on the question of who does microcredit help and who doesn't it help, here's a podcast with Alice Evans and Rachael Meager discussing Meager's paper (covered way back in January) on how to better understand the results of the microcredit impact evaluations.

5. Philanthropy and Social Investment: Finally, this week there were several pieces on philanthropic practice that felt like they were directly trolling me. First, Gabe Kleinman wonders "Why aren't foundation's actually helping their grantees like VCs?" Kleinman is apparently unaware this is an idea that is more than 20 years old. By the time I came into the space a little over 10 years ago, one of the first pieces I tried to commission was tentatively titled, "picking over the corpse of venture philanthropy" because it was already "old news" by then. And like Cathy O'Neil's piece, Kleinman gives lots of examples of things foundations "should be doing" where foundations are actually doing those things, he is simply unaware of them. Overall, it's a fine example of the annoyingly common, but intellectually bankrupt "non-profits should act more like businesses" idea that fails in two ways: it holds up how businesses operate in theory rather than reality; and shows a profound ignorance of the challenges of social investing. 
Speaking of the challenges of social investing, there is a new philanthropic effort (with big time funding including Gates) targeted at "systemic change," lamenting that it is hard to for funders to collaborate and there are few "big ideas" or non-profits able to effect systemic change. Again, grrrrrr. One of the reasons for the challenges noted is the "venture philanthropy" mindset which if taken at face value discourages collaboration and emphasizes short-term metrics. But it's also true that there have been (and continue to be) lots of collaborative efforts--these ideas are not new--and they often struggle because it's virtually impossible to arrive at agreement on how systems should change, who gets a seat at the table and what priorities should be. Co-Impact will make grants of up to $50 million for up to 4 years, which is nice, but hardly seems like the scale necessary for systemic change if those words mean what I think they mean. For comparison, the grants are at most half of 100&change, which has narrowed down to surprisingly prosaic ideas that I don't think anyone would describe as systemic.
Finally, to end on a positive note, here's a piece about how the Hewlett Foundation has retooled it's grant reporting process so that's its more useful for everyone. The trolling element? I'm not a Hewlett grantee.

Duck of Minerva also known as Steve Saideman writes about the peer review process and who is doing the work in various social sciences. It's a whole new form of inequality!

Duck of Minerva also known as Steve Saideman writes about the peer review process and who is doing the work in various social sciences. It's a whole new form of inequality!


Return to the Weekly faiV