Orwellian Debt Collection in China

Trying to get a handle on the potential for a workable personal bankruptcy procedure in China, I've repeatedly encountered evidence that the most important element might be lacking: attitude. Successful personal insolvency systems around the world differ in design and operation, but the system architects and operators generally share a sense that default is an inevitable aspect of consumer/entrepreneurial risk, and mitigating the long-term effects of such defaults is good for debtors, creditors, and society. I don't get the sense, based on my admittedly superficial outsider perspective, that this foundation is ready in China. Indeed, quite the opposite. 

For example, for the past few years, the Supreme People's Court has run a "judgment defaulter's list" of individuals who have failed (been unable?) to satisfy judgments against them. More than 3 million names were on this list already by the end of 2015, and getting on this list means more than just public shaming; it's also a "no-fly" list, preventing defaulters from buying airplane tickets, in addition to a "no-high-speed-train" and "no-hotel-stay" list, and also a "no-sending-your-kids-to-paid-schools" list. By mid-2016, about 5 million people had been preventing from buying these services in China as a result of being on the list. This initiative is just the start of a planned "Social Credit System," which will aggregate electronic data (including not only payment history, but also buying habits, treatment of one's parents, and who one's associates are) to produce a "social credit score" for all individuals. This score will affect all manner of life events, such as access not only to loans, but also to housing access, work promotions, honors, and other social benefits. The potential problems with data integrity (including inaccurate data), among many other challenges, are discussed in this fascinating paper by Yongxi Chen and Anne Sy Cheung of the Univ. of Hong Kong. 

There may be more nuance here than I can access now, but treating all judgment defaulters as morally culpable fiends who should be punished is ... a bit outdated. Though it's dangerous to draw parallels to China's imperial past, I can hardly resist observing that bankruptcy for nearly 300 years of Qing rule up to the 20th century was a crime, punishable by an escalating number of blows of a bamboo cane depending upon the size of the defaulted debt (discussed at pp. 18-19 of this excellent thesis). Discharge does not seem to have been considered, as in pre-modern England, as a method of enticing debtors to reveal the location of their assets or mitigate the effects of economic volatility. Again, I'm not sure it's fair to equate pre-modern ideas about financial distress with 21st century attitudes, but I'm not enthusiastic about what seems to be the absence of a rehabilitative perspective on defaulting debtors in both old and new China. Comments from those who know more than I are enthusiastically welcomed! 

Comments

[Matthew Bruckner]

I think this is an interesting post. Thank you for sharing. I'd like to know more about the social stigmatization aspect that you raise.

Nothing about the social credit system strikes me as strange to the extent that it's being using to assess creditworthiness. All of the factors you note (even the treatment of your parents) seem more obviously correlated with likelihood of repayment than some data points being used by algorithmic lenders in the US.

[Jason Kilborn]

Matthew, the system is far more than a measure of (financial) creditworthiness. The plan is apparently to aggregate data including things like search engine query histories (!), social media postings (including those about someone, not only by that person), online purchasing habits (i.e., the things one purchases, not simply whether one pays for them), whether one visits one's parents regularly enough, and who one's friends are. I would hope that it would be uncontroversial that most of this has little to do with (financial) creditworthiness, and who my friends are and what I'm searching for online have absolutely nothing fairly to do with my creditworthiness. More troubling, the score generated from this data is to be used in non-financial contexts, like speeding processing of housing applications and work promotions, and perhaps job applications--a context where US (state) regulators have wisely forbidden consideration of irrelevant "creditworthiness" data. This seems to be a Big Brother monitoring system expressly designed to reward virtuousness and punish what system architects regard as signs of vice, all with precious little if any transparency about the data or its use. While credit scoring in the US is far from perfect, the Social Credit System takes it to an entirely new level in every way.

[Matthew Bruckner]

Jason, I do believe this is happening already and is not limited to China. In short, it's not a future plan but a current reality (at least in some countries). There are at least three companies doing this work that I can think of off the top of my head. Lenddo, a Hong Kong-based company, claims to be operating in more than 15 countries and using a lot of the data points you highlight (https://www.lenddo.com/pdfs/Lenddo_FS_CreditScoring_201705.pdf). Zest uses more than 3,000 data points in its scoring model (https://www.zestfinance.com/hubfs/Zestfinance_Feb_2017_files/docs/ZAML%20data%20sheet.pdf?t=1522358514090). Although they don't share which ones, I'd bet there is overlap with your list. Kabbage advertises its use of social media data in underwriting (https://techcrunch.com/2017/11/16/kabbage-gets-200m-from-credit-suisse-to-expand-its-ai-based-business-loans/).

Why do you think that online purchasing habits would be irrelevant to evaluating financial capacity? Target reportedly could figure out if its customers were pregnant using their purchasing habits... https://www.forbes.com/forbes/welcome/?toURL=https://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/

I agree that what's interesting ("more troubling) here is the use of this data in non-financial contexts, like speeding processing of housing applications and work promotions, and perhaps job applications. It's not clear why this data would be relevant. However, what we're learning from Big Data and machine learning is that the algorithms often see connections that humans do not. In other words, just because I don't see the connection does not mean that none exist.

Finally, I wouldn't say that US regulators have effectively forbidden the use of irrelevant creditworthiness data. I understand that FICO9 has largely abandoned the use of medical debt as being predictive of a borrower's creditworthiness. But many lenders have not advanced past the use of FICO4 and are, therefore, still using this non-predictive data to make credit determinations.

A lot of folks are already writing about these and related issues. Some articles I like are:
Andrew Tutt, An FDA for Algorithms, 69 ADMIN. L.REV. 83(2017), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2747994
Dennis D. Hirsch, That’s Unfair! Or Is It? Big Data, Discrimination and the FTC’s Unfairness Authority, 103 U. KY. L. REV. 345 (2014), http://www.kentuckylawjournal.org/wp-content/uploads/2015/02/103KyLJ345.pdf
Solon Barocas & Andrew Selbst, Big Data’s Disparate Impact, 104 CAL. L. REV. 671 (2016) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2477899
Mikella Hurley & Julius Adebayo, Credit Scoring in the Era of Big Data, 18 YALE J.L. & TECH. 148 (2016), http://digitalcommons.law.yale.edu/yjolt/vol18/iss1/5/
EVA WOLKOWITZ & SARAH PARKER, CTR. FOR FIN. SERVS. INNOVATION, BIG DATA, BIG POTENTIAL: HARNESSING DATA
TECHNOLOGY FOR THE UNDERSERVED MARKET 18 (2015),
http://www.morganstanley.com/sustainableinvesting/pdf/Big_Data_Big_Potential.pdf

Of, course, I also wrote on this: Matthew A. Bruckner, The Promise and Perils of Algorithmic Lenders' Use of Big Data, 93 Chi-Kent L. Rev. 1 (2018) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3137259

[Jason Kilborn]

It's not federal legislators I had in mind, but states, who have forbidden the use of credit reports in such contexts as hiring, and the notion that US regulators are still allowing data like medical records to influence credit scores makes me not more comfortable with the Chinese plan, but even less comfortable with the US approach.

I find it terrifying that more and more instances of correlation-causation confusion are being foisted on the lending and consuming public in the guise of supposedly reliable big data crunching. That my credit score should be impacted by what I search for in Google is ... deeply disturbing. That this phenomenon is not limited to China again makes me feel worse, not better.

But again, it's the attitude signal I'm looking for here. That national regulators, rather than private lenders, are planning such aggressive use of this sort of data, and that the resulting score will be used not in lending decisions, but in the provision of basic human services, is again deeply disturbing. That a society (governing authority) is so willing to capitulate to a computer-generated message that so-and-so is a deadbeat, rather than so-and-so having been the possible victim of global economic volatility is the problem I'm concerned with here. If that contagion is spreading, even more cause for concern.

[Matthew Bruckner]

I agree. This is scary stuff. I didn't mean to suggest that you should be comforted because this is happening more widely.

IMO, there is an intuitive appeal to using Big Data and algorithms while attempting to make sense of the world. But it's important to bear in mind the points that you've made.

[George]

It does read Orwellian, and scary; yet I doubt that restricting indebted people's freedom will allow them to pay their debt in the future.

For example, if you have a good job or business opportunity in another city, and you need a train to get there, but your debt prevents you from buying a ticket, then how is the individual going to make the money he or she needs to pay back what he owes?

I hope those measures will never be implemented here.