“Examining the Fintech Landscape.”
I. Introduction
The financial technology ("fintech") landscape is complex and diverse.
For purposes of this testimony, I will divide the fintech landscape into two spheres. One, incrementalist fintech, uses new data, algorithms, and software to perform classic work of existing financial institutions. This new technology does not change the underlying nature of underwriting, payment processing, lending, or other functions of the financial sector. Regulators should, accordingly, assure that long-standing principles of financial regulation persist here. I address these issues in Part II below.
Another sector, which I deem "futurist fintech," claims to disrupt financial markets in ways that supersede regulation, or render it obsolete. For example, if you truly believe a blockchain memorializing transactions is "immutable," you may not see the need for regulatory interventions to promote security to stop malicious hacking or modification of records. In my view, futurist fintech faces fundamental barriers to widespread realization and dissemination. I address these issues in Part III below.
II. Incrementalist Fintech
A. Big Data or Artificial Intelligence-based Underwriting
Many marketplace lenders are now using forms of data not traditionally used for credit underwriting, in order to offer consumer or small business loans. They may help correct some long-standing problems in US credit markets, including the problematic nature of contemporary credit scoring. However, as
Credit-scoring tools that integrate thousands of data points, most of which are collected without consumer knowledge, create serious problems of transparency.
Consumers have limited ability to identify and contest unfair credit decisions, and little chance to understand what steps they should take to improve their credit. Recent studies have also questioned the accuracy of the data used by these tools, in some cases identifying serious flaws that have a substantial bearing on lending decisions.
Big-data tools may also risk creating a system of "creditworthiness by association" in which consumers' familial, religious, social, and other affiliations determine their eligibility for an affordable loan. These tools may furthermore obscure discriminatory and subjective lending policies behind a single "objective" score. Such discriminatory scoring may not be intentional; instead, sophisticated algorithms may combine facially neutral data points and treat them as proxies for immutable characteristics such as race or gender, thereby circumventing existing non-discrimination laws and systematically denying credit access to certain groups. Finally, big-data tools may allow online payday lenders to target the most vulnerable consumers and lure them into debt traps. n2
The problem of "big data proxies" is a serious one recognized by leading privacy scholars. n3 Regulators should do much more to assure that next-generation technology does not simply reproduce old biases. n4 The alternative is a "scored society" where individuals lack basic information about how they have been treated in the credit granting context. n5
These problems are troubling in the abstract. Their concrete implications are chilling, as a recent Privacy International Report revealed. Outside
* "If lenders see political activity on someone's Twitter account in
* "The contents of a person's smartphone, including who and when you call and receive messages, what apps are on the device, location data, and more."
* "How you use a website and your location. [One firm] analyses the way you fill in a form (in addition to what you say in the form), and how you use a website, on what kind of device, and in what location." n6
Moreover, machine learning systems are constantly developing even more invasive forms of assessing creditworthiness, or factors influencing it. A recently published paper claims to infer propensity to criminality merely from the features of persons' faces. n7 Sexuality and health are also now being predicted by machine learning researchers entirely on the basis of a picture of a person's face--something relatively easy to gather via a Google image search, or
1. Neither Machine Learning Nor Predictive Analytics are too Complex to Regulate
Some fintech firms which rely on artificial intelligence may counter that the computation involved in their decisionmaking now amounts to a form of cognition as hard to explain as that of a human decision-maker. Genetic algorithms may, for instance, themselves spawn, each second, dozens of ways of processing information, which are then evaluated on some metric, and Darwinianly given a chance to persist based on their performance. Iterative machine learning processes may be similarly complex and opaque. Their view is that, just as we can't map all the brain's neurons to connect a person's decision to eat a slice of cake to some set of synapses, we can't map or unravel the sequence of events that leads to a given algorithmic score or sorting.
I believe that we should be suspicious of the deregulatory impulse behind characterizations of machine learning as "infinitely complex," beyond the scope of human understanding. The artificial intelligence that commercial entities celebrate can just as easily evince artificial imbecility, or worse. Moreover, there are several practical steps we can take even if machine learning processes are extraordinarily complex. For example, we may still want to know what data was fed into the computational process. Presume as complex a credit scoring system as possible. Regulators could still demand to know the data sets fed into it, and, for example, forbid health data from being included in that set. We already know that at least one credit card company has paid attention to certain mental health events, like going to marriage counseling. n10 When statistics imply that couples in counseling are more likely to divorce than couples who aren't, counseling becomes a "signal" that marital discord may be about to spill over into financial distress. n11 This is effectively a "marriage counseling penalty," and poses a dilemma for policy makers. Left unrevealed, it leaves cardholders in the dark about an important aspect of creditworthiness. Once disclosed, it could discourage a couple from seeking the counseling they need to save their relationship.
There doesn't have to be any established causal relationship between counseling and late payments; correlation is enough to drive action. That can be creepy in the case of objectively verifiable conditions, like pregnancy. And it can be devastating for those categorized as "lazy," "unreliable," "struggling," or worse. Runaway data can lead to cascading disadvantages as digital alchemy creates new analog realities. n12 Once one piece of software has inferred that a person is a bad credit risk, a shirking worker, or a marginal consumer, that attribute may appear with decision-making clout in other systems all over the economy. There is also little in current law to prevent companies from selling their profiles of consumers. n13
2. The Problems of Extant Data Collectors are a Reason for More Scrutiny of
Having eroded privacy for decades, shady, poorly regulated data miners, brokers and resellers have now taken creepy classification to a whole new level. They have created lists of victims of sexual assault, and lists of people with sexually transmitted diseases. Lists of people who have Alzheimer's, dementia and AIDS. Lists of the impotent and the depressed.
There are lists of "impulse buyers." Lists of suckers: gullible consumers who have shown that they are susceptible to "vulnerability-based marketing." And lists of those deemed commercially undesirable because they live in or near trailer parks or nursing homes. Not to mention lists of people who have been accused of wrongdoing, even if they were not charged or convicted. Typically sold at a few cents per name, the lists don't have to be particularly reliable to attract eager buyers. And there is increasing risk that your spouse, friends, boss, or acquaintances could buy such data. n14
There are three problems with these lists. First, they are often inaccurate. For example, as The
Third, people aren't told they are on these lists, so they have no opportunity to correct bad information. The
It's unrealistic to expect individuals to inquire, broker by broker, about their files. Instead, we need to require brokers to make targeted disclosures to consumers. Uncovering problems in Big Data (or decision models based on that data) should not be a burden we expect individuals to solve on their own.
Privacy protections in other areas of the law can and should be extended to cover the consumer data now fueling fintech underwriting. The Health Insurance Portability and Accountability Act, or HIPAA, obliges doctors and hospitals to give patients access to their records. The Fair Credit Reporting Act gives loan and job applicants, among others, a right to access, correct and annotate files maintained by credit reporting agencies.
It is time to modernize these laws by applying them to all companies that peddle sensitive personal information. If the laws cover only a narrow range of entities, they may as well be dead letters. For example, protections in HIPAA don't govern the "health profiles" that are compiled and traded by data brokers or fintech firms, which can learn a great deal about our health even without access to medical records.
We need regulation to help consumers recognize the perils of the new information landscape without being overwhelmed with data. The right to be notified about the use of one's data and the right to challenge and correct errors is fundamental. Without these protections, we'll continue to be judged by a big-data Star Chamber of unaccountable decision makers using questionable sources.
Policymakers are also free to restrict the scope of computational reasoning too complex to be understood in a conventional narrative or equations intelligible to humans. They may decide: if a bank can't give customers a narrative account of how it made a decision on their loan application, including the data consulted and algorithms used, then the bank can't be eligible for (some of) the array of governmental perquisites or licenses so common in the financial field. They may even demand the use of public credit scoring models, or fund public options for credit. Finally, they should look to
B. Emerging Issues in Preemption and Regulatory Arbitrage Some fintech advocates advocate radical deregulation of their services, to enable their rapid entry into traditional banking markets. However, there is a risk of the fintech label merely masking "old wine in new bottles." The annals of financial innovation are long, but not entirely hallowed. n18 When deregulatory measures accelerated in the late 1990s and early 2000s, their advocates argued that new technology would expertly spread and diversify risk. However, new quantitative approaches often failed to perform as billed. Most fundamentally, a technology is only one part of a broader ecosystem of financial intermediation. n19
I do believe that some fintech may promote competition and create new options for consumers. But we should ensure that it is fair competition, and that these options don't have hidden pitfalls. In my research on the finance and internet sectors, I have explored patterns of regulatory arbitrage and opaque business practices that sparked the mortgage crisis of 2008. n20 I see similar themes emerging today.
In the run-up to the crisis, federal authorities preempted state law meant to protect consumers. n21 The stated aim was to ensure financial inclusion and innovation, but the unintended consequences were disastrous. Federal authorities were not adequately staffed to monitor, let alone deter or punish, widespread fraudulent practices. Agencies like the
For example, such fintech charters could enable regulatory arbitrage around state restrictions on payday lending. As 270 entities--community, labor, civil rights, faith-based, and military and veterans groups--observed earlier this year, 90 million Americans "live in jurisdictions where payday lending is illegal." n26 These state consumer protection laws help consumers "save billions of dollars each year in predatory payday loan fees that trap people in long-term, devastating cycles of debt." n27 OCC should not take action to preempt them. n28
These are not mere hypothetical concerns; as the
Nor should the
One more aspect of regulatory arbitrage is now in fintech news: recent applications by Square and SoFi for
However, in this case, neither SoFi nor Square appear to be the type of commercial firms which would fit Baradaran's account, since they would not inject the source of strength that was praised by Baradaran in the
III. Futurist
Though sober reports from the
Futurist fintech envisions "smart contracts," which would be executed via some degree of automatic, code-based enforcement. n41 As one article puts it, "Where a smart contract's conditions depend upon real-world data (e.g., the price of a commodity future at a given time), agreed-upon outside systems, called oracles, can be developed to monitor and verify prices, performance, or other real-world events." n42 However, until robotic assessments of physical reality are far less delayed, corroded by a lack of data, and contestable (thanks to the messy complexity of discordant human meanings), the prevalence of totally automated, smart contracts is likely to be limited.
There are many contractual relationships that are too complex and variable, and require too much human judgment, to be reliably coded into software. Code may reflect and in large part implement what the parties intended, but should not itself serve as the contract or business agreement among them.
Still, some technologists and lawyers aspire to that subsumption, echoing older movements for financial deregulation. n43 The rise of Bitcoin as an alternative currency has sparked an interest in automation of transactions and recordation. n44 Software can allow distributed computers to transfer information en masse and monitor one another. n45 Bitcoin is a particular case of using blockchain technology to ensure a durable record of ownership, which is intended to be regulated by code. n46 Blockchain enthusiasts envision it scaling en masse to serve as a distributed ledger of all manner of transactions.
Given enthusiasm expressed for blockchain at the highest levels of international finance, n47 governments may soon explore more extensive use of blockchain-based, public ledgers of ownership transactions, such as land records. n48 Such a digital transition would cut out a fair number of time-consuming steps in current financial processing. Using technology to modernize transactions would seem to be a huge opportunity for saving personnel costs and reducing inconvenience.
Yet there are also reasons for caution. As
Moreover, some early adopters of this ideal of self-executing or coded law have experienced troubling and telling failures. n54 Investors in a "decentralized autonomous organization" (DAO) run on code have already experienced the turbulent and troubling aspects of software-governed legal orders. In early 2016, a hacker managed to take millions of dollars in a fashion unanticipated by the drafters of the code governing the organization. The main organizer of the DAO,
According to Buterin and other organizers of the DAO, this intervention was a success story: it proved the recoverability of their system. But for advocates of futurist fintech, this was a Pyrrhic victory. The post hoc intervention violated the principle of autonomy supposedly at the core of the DAO. n56 Persons managed the smart contract--not mere code. n57 In other words, the only way the supposedly smart, incorruptible, automated, and immutable contract actually protected investors was by allowing human intervention to change its terms and consequences. Rather than demonstrating the dispensability of human interventions, the DAO has proved the opposite--the vital necessity of human governance over even extensively coded and computerized forms of human cooperation.
When
IV. Conclusion
This testimony has presented reasons to be cautious about legislative or regulatory efforts to federally preempt state laws now applying to both incrementalist and futuristic fintech. I know that advocates for deregulation will likely argue that imposing a level playing field on fintech and non-fintech firms will harm innovation in the fintech sector. But innovation is not good in itself. The toxic assets at the core of the financial crisis were innovative in many ways, but ultimately posed unacceptable risks. n61 So, too, may the superficially attractive services of many fintech firms.
To be sure, promoters of fintech deregulation may claim that such worries are anecdotal. But many tech firms have only themselves to blame for obscuring what we know about the sector. As I explain in my book
Data gathering is important, because nearly every story of technologized "financial inclusion" can be countered with other stories of exclusion, via digital redlining. As
We should not have faith that accelerated deregulation will free the financial sector to solve important social problems. The value proposition of some fintechs merely points out larger problems in existing credit provision that could be solved by more direct action. For example, if fintechs can make a hefty profit by refinancing student debts owed to the
In conclusion:
n1 The Government Accountability Office has described fintech as follows: "The financial technology (fintech) industry is generally described in terms of subsectors that have or are likely to have the greatest impact on financial services, such as credit and payments. Commonly referenced subsectors associated with fintech include marketplace lending, mobile payments, digital wealth management, and distributed ledger technology." GAO,
n2
n3 See, e.g.,
n4 For an up-to-the-minute overview of this and related problems, see
n5
n6
n7
n8
n9 To be clear, I am not alleging any particular fintech firm in
n10
n11
n12
n13
n14
n15
n16 Note that information generated for or within a credit context may spread outside it--and vice versa.
n17 See, e.g.,
n18 FINANCIAL CRISIS INQUIRY COMMISSION, FINAL REPORT OF THE NATIONAL COMMISSION ON THE CAUSES OF THE FINANCIAL AND ECONOMIC CRISIS IN
n19
n20
n21 FCIC Report, 112 and passim ("Once OCC and OTS preemption was in place, the two federal agencies were the only regulators with the power to prohibit abusive lending practices by national banks and thrifts and their direct subsidiaries."); id., at 350 ("The
n22 Fortunately, the
n23
n24 Id., at 2.
n25
n26
n27 Id.
n28 Americans for Financial Reform,
n29
n30
n31 Madden v. Marine Midland Funding, No. 14-2131 (2d Cir. 2015).
n32
n33 Id.; see also
n34
n35
n36
n37 Mehrsa Baradaran, Reconsidering the Separation of Banking and Commerce, 80 George Washington Law Review 385 (2012).
n38
n39 Press Release, Waters Calls on
n40
n41
n42
n43
n44
n45
n46
n47
n48 It is at this point unclear whether decentralization via distributed ledger technology would address or exacerbate key problems identified in the
n49
n50 See
n51 Paul Merrion,
n52
n53
n54
n55
n56
n57 Id.
n58
n59 Id.
n60
n61
n62
n63 CATHY O'NEIL, WEAPONS OF MATH DESTRUCTION (2016).
n64
n65
n66 MEHRSA BARADARAN, HOW THE OTHER HALF BANKS (2015). Over 25% of US households are unbanked or underbanked.
Read this original document at: https://www.banking.senate.gov/public/?a=Files.Serve&File_id=0A92AD09-6834-4D7E-901A-6AE5C51572AE
This legislation would provide relief to Americans who are trapped in overpriced and unreliable Obamacare plans.
“Health Care: Issues Impacting Cost and Coverage.”
Advisor News
Annuity News
Health/Employee Benefits News
Life Insurance News