Financial regulators are rarely regarded as playful. So when large numbers of them start building sandboxes, we should take notice.
Throughout the world, regulatory agencies are launching initiatives they’re calling “sandboxes,” or less fancily, “pilots,” “labs,” and even “greenhouses.”
Regardless of label, these initiatives share two key characteristics.
First, they all are broadly inspired by the flagship Regulatory Sandbox program launched in 2015 by the U.K.’s Financial Conduct Authority.
Second, they all aim to create a small, safe space where regulators can learn hands-on about financial innovation through experimentation and trial and error, in order to evaluate the regulatory implications of new technology.
This movement has implications that could remake the entire approach to financial regulation.
Growing meaning of sandboxes
A report this month by the Aspen Institute finds that at least 23 countries have launched or are exploring such testing units—only two short years after FCA’s original. While they’re all in an early stage, it is becoming clear that they are an indispensable element of any effort to shift financial regulation from the analog to the digital era.
My work in this field has convinced me of two things.
First, regulators need to build these testbeds, which are not only useful, but actually essential. They are the only way for policymakers to achieve sufficiently rapid and deep learning about how technology should change financial services and regulation—and how regulators should interact with technology.
Second, over time, regulators should build on these learning labs to launch another critical innovation: an alternative regulatory channel in which financial companies—if they opt to do so—could be efficiently supervised through high-tech, lower-cost data reporting, instead of by traditional means.
Reglab for U.S. fintech and regtech
The earliest round of current initiatives—I will call them “reglabs”—has focused mainly on fintech product innovation. Typically, regulators invite financial companies to apply to participate in controlled testing to demonstrate the likely benefits and risks of proposed innovations that raise potential regulatory questions.
More recently, a parallel focus has arisen on so-called “regtech” learning laboratories in which the agencies can test possible new ways to conduct their own work.
Both are needed because of the widening mismatch between the exponentially rising pace of technological change and the traditional, linear nature of our regulatory functions.
Regulatory processes and cultures evolved, for good reason, to be deliberative—careful, risk averse, and slow to change. Historically, this care has usually paid off. Today, though, technology shifts are rapidly emerging and converging, driven by innovation like mobile devices, big data, artificial intelligence, blockchains, and natural voice interface.
• These massive tech trends are opening the potential for previously unimaginable benefits to financial customers, especially consumers and small businesses.
• At the same time, they are seeding thorny new challenges in areas like privacy, cybersecurity, consumer data rights, fairness of artificial intelligence, and regulating business models can’t be contained in existing regulatory frameworks.
Recognizing this, several U.S. agencies have undertaken fintech initiatives. These include the CFPB’s Project Catalyst, the OCC’s Responsible Innovation program, the Commodity Futures Trading Commission’s new LabCFTC, and task forces at the Federal Reserve Board and FDIC. The Conference of State Bank Supervisors has also launched Vision 2020, and initiatives are underway in several states. While some of these contain reglab-type elements, they focus mainly on other strategies.
All these efforts are confronting the reality that innovation is inherently difficult. Harvard’s Clayton Christensen laid out the core challenge in his seminal 1997 book, The Innovator’s Dilemma, identifying the phenomenon in which organizations dismiss or resist disruptive innovation because they have too much to lose.
Established players fear that embracing change might undermine or cannibalize business models that are still succeeding. It has been said that large organizations have immune systems that instinctively attack invading innovation. These “corporate white corpusles” try to kill innovation before it can take hold and spread.
Realizing this, the private sector has learned to innovate by segregating experimental projects from the larger organization. They keep these efforts small and concrete. This enables them to learn quickly—even to “fail fast”—and thereby generate learning that feeds into a next round of trials. This idea is at the heart of government reglabs, as well.
What really happens in a “sandbox”
Some regulators have raised valid concerns about using experimental mechanisms, but all can be addressed with careful design and execution.
One worry is that tests conducted with real consumers could harm them, especially if regulators are waiving or suspending consumer protection requirements. This danger is being addressed in several ways.
First, most regulators are not waiving rules (most actually can’t), but rather are focusing on areas where current regulations are unclear in how they would affect an innovative approach.
Also, most regulators closely monitor the experiments—and they also insist on the ability o shut down a test early, if necessary, and that participants make whole any customers harmed.
Some lab testing can also be done through “virtual sandboxes” where early-stage innovation is run through data models, without touching live customers at all.
How to help reglabs work
Another concern, unique to the U.S., is that reglabs here are unlikely to produce scalable results due to our exceptionally complex and fragmented regulatory structure.
Five U.S. federal agencies directly supervise financial institutions, often in overlapping ways. About two dozen more federal agencies impact financial regulation, while another 50 states charter and supervise banks and license and oversee nonbank financial companies. This design makes it nearly impossible to achieve holistic, consistent regulation and to keep pace with technology change in the many areas requiring interagency negotiation and coordination.
Furthermore, reglabs run by prudential bank supervisors would be structurally cut off from cutting-edge innovation, most of which arises in the nonbank realm (perhaps partly due to bank regulation factors). The U.S. federal regulators (with the partial exception of the CFPB) have no supervisory role regarding nonbanks.
The answer: central reglab
This unique American complexity can be addressed in part by creating an interagency reglab as a shared learning laboratory and springboard toward new models for interagency collaboration.
A joint reglab would foster alignment in how regulators are viewing innovation trends and would enable deep understanding of emerging models. That understanding would include how they work; what risks they present; and why and how fintech founders and visionary bankers are trying to change finance.
Only labs can build this depth.
Fintech to regtech to “supertech”
A joint reglab should address not only the challenges of regulating fintech, but also the opportunity to foster and adopt regtech, which is rapidly emerging throughout the world but has been slower to take shape in the U.S. The term refers both to new technologies for industry compliance and to use of “regtech for regulators,” sometimes called “suptech.”
Policymakers are realizing that the same trends driving change in finance, including new, plentiful data and machine learning, are making it possible to redesign regulatory systems as well. Determining precisely how to do so will require testing, small-scale experiments, and new ways of learning.
U.S. regulators are facing very difficult questions as technology changes them and the financial industry.
Among the questions:
1. Can regulators move toward machine-readable regulations?
2. How should regulators treat use of alternative data in loan underwriting in light of fair lending disparate impact policy?
3. How could data be better shared for anti-money laundering, including know-your-customer requirements?
4. How could the Community Reinvestment Act be modernized to reward technology-enabled innovation that widens access to affordable financial services?
5. How should federal agencies regulate digital currency?
6. Should capital requirements be reexamined based on changing technology?
7. Can blockchains simplify some regulatory activities?
8. What are safe and sound models for partnering between fintechs and community banks?
9. Can community banks compete, if regulatory and economic factors prevent them from innovating?
10. Could some regulations eventually be promulgated as computer code, and be self-implementing?
To answer questions like these and more, it would help to study them in the lab.
Alternative regulatory channel
If U.S. regulators can build a shared reglab and use it to reshape both use of regtech and regulation of fintech, they can lay a foundation for a more ambitious new strategy—creation of a new, alternative regulatory channel, as an option, for financial companies that prefer to adopt it.
Today’s technology is creating a possibility that has never before existed—the opportunity to improve the public policy results of financial regulation, and reduce the costs of achieving them, at the same time.
This is a sharp departure from tradition, which has normally framed regulatory policy as a binary choice: either more regulation with better results (one hopes) at higher costs, or less regulation with lower cost but worse results.
It is becoming possible to break this zero-sum dynamic and make huge strides toward achieving regulatory goals that, despite absorbing massive resources, have been elusive for decades.
Imagine truly curtailing financial crime; achieving virtually full financial inclusion; fostering widespread consumer financial health; and reducing systemic risk—through a new regulatory model that also slashes costs for industry and government.
If this could be done, in a verifiable and trustable way, a new alignment could emerge in which industry and advocates groups that normally clash would begin to join in support of better approaches.
The technology is the key.
Our financial regulatory system was built for an analog era in which data was scarce and computing power was scarce and expensive. Today, both are cheap and abundant. This makes it possible for both industry and regulators to make massive gains in effectiveness and efficiency, following the same digital transformation paths that are reshaping other fields, from communication and photography to health science and taxis.
Importantly, the regtech tools emerging today are qualitatively different from older ones.
Traditional regulatory technology has typically started with an existing linear process and overlayed it with a layer of automation, to speed it up or make it easier to use.
The new systems are instead starting with a blank screen. Developers ask what the objectives and requirements are. Their goal: Building a “systems” solution that deeply redesigns the process, using better data and usually machine learning.
Think of the difference in this way: You can ask a career regulator to design rules in the traditional way. Or you could hand the job over to someone who’d never written a bank rule but had other skills—say someone like Steve Jobs.
What supertech could bring
Determining how best to build such a new system will require time and a great deal of work. It’s neither desirable nor practical for policymakers to attempt to adopt it through large waves of omnibus regulatory reform. Instead, policymakers should build gradually on the small experimental work of the reglabs, and begin to identify areas in which financial companies could be evaluated based on objective metrics that they can prove through data.
For example, suppose a community bank was prepared to demonstrate that some or all of its consumer products are:
• Simple, transparent, and easily understood.
• Perform as promised.
• Do not rely for profitability on high levels of penalty income.
• Generate low levels of valid complaints.
In a regtech/supertech system, such a bank could make the choice to opt into a new high-tech regulatory channel to be evaluated against known quantitative metrics that show it meets these objective standards.
By opting in, the bank would then be relieved of most traditional oversight of its compliance management system.
The government’s position in the process would evolve. Instead of getting into minutiae and human examinations, they could conclude that if a financial company can prove its product meets the objective standards for fair customer outcomes, regulators need not care how this was accomplished.
Think of it—results-oriented regulation.
If the regulators gradually build this alternative channel, starting small and learning through experience, the alternative approach could gradually be refined, prove itself, and attract more and more financial companies into the new channel.
If the new channel produces demonstrably better public policy outcomes and/or lower costs, it would grow and might eventually become the mainstream regulatory system.
Call this the “Reg 1776 Initiative”
Consider this: it’s doubtful that the American experiment could have worked if its founders had stayed in Europe. They needed new, unploughed ground where they could seed something new, cultivate it, and grow it into a system that began to make people want to join it.
Maybe we can change regulatory systems in the same way.