Distributed ledger technology and retirement savings infrastructure – part 2
In a previous article, Distributed ledger technology and retirement savings infrastructure – part 1 – the basics, we set out to define distributed ledger technologies (DLT). In this article we discuss some of the practical effects on retirement plan administration that may result from the adoption of distributed ledger technologies (DLT) by the financial services industry as a whole and by retirement plan sponsors and service providers. We then consider the policy implications of some of these changes. We conclude with a brief discussion of some legal/ERISA issues that DLT may raise.
DLT and the financial services industry
As described by the Fed (Federal Reserve Board, Washington, D.C., Finance and Economics Discussion Series, Distributed ledger technology in payments, clearing, and settlement, Mills, Wang, Malone, Ravi, Marquardt, Chen, Badev, Brezinski, Fahy, Liao, Kargenian, Ellithorpe, Ng, and Baird. 2016-095. Hereafter, “Fed 2016.”):
[Financial industry] participants and technology firms are increasingly exploring ways to develop and deploy DLT arrangements for use in payments, clearing, and settlement. Although the technology has the potential to provide a new way of storing, recording, and transferring digital assets, at present most industry participants are looking at ways to integrate the technology into existing systems and institutions. Many models may alter or eliminate some roles of current intermediaries in payments, clearing, and settlement but may not necessarily eliminate the need for coordination or centralization of certain functions by trusted intermediaries. These trusted intermediaries could still be needed to play important roles in addressing frictions beyond what DLT may be able to accomplish or may be able to use DLT arrangements to improve or evolve how they accomplish their respective missions.
Bottom line: DLT is being implemented (at least incrementally) by the financial services industry.
Effect on retirement plans
Retirement plans are not going to drive this process, but they will be affected by it, in at least two obvious ways.
First, these changes will (once, e.g., development costs are amortized) increase the efficiency and reduce the cost of managing and keeping track of assets. The cost of doing an individual transaction will go down. Perhaps more significantly, the cost of reporting and disclosure will go down.
With regard to retirement plan management, the implementation of DLT should mean, at some point, that, e.g., 5500 financial reporting could be fully automated: plan transactions, net position, etc. could be reduced to a set of inputs from the distributed ledger and software algorithms. In a perfect world, annual 5500 reporting could be eliminated entirely, and the relevant agencies could simply be given limited access to the relevant parts of the distributed ledger.
Second, plan sponsors (and even, e.g., 401(k) plan participants) will have (or may be given) access to, virtually, all relevant data. DLT implementations are characterized by “granular transparency.” This transparency will provide the opportunity for a richer set of analytic tools, most obviously useful for defined benefit plans but with possible uses in 401(k) plans as well.
Application of DLT to retirement plan administration
An obvious application of DLT is the transformation of 401(k) plan recordkeeping. A robust implementation of DLT should (it would seem) eliminate the need for trust/plan data reconciliation – the whole point of a distributed ledger is that all the relevant data is in the ledger, and everyone has a copy of that ledger. Indeed, layers of ownership, e.g., custodial and beneficial, may simply be coded into the ledger.
Elimination of these “frictions” in 401(k) plan administration may make, e.g., 401(k) discrimination testing easier. And it may allow the sponsor (as an end-user) to directly exploit the database (ledger) for a variety of purposes, e.g., targeted communications.
As data-intense and system dependent arrangements, retirement plans would seem to be a rich target for DLT implementation. Consider, for instance, the use of smart contracts – “contracts whose terms are recorded in a computer language instead of legal language” – for plan documentation and compliance with IRS qualification requirements.
The reductions in cost that result from these changes will, at the margin, increase the adoption of ERISA retirement plans (most obviously, 401(k) plans) by small employers who cannot currently afford them given the current administrative overhead burden. Thus, they may make possible a (perhaps significant) increase in retirement savings, e.g., by individuals not currently covered by a plan.
Are there situations in which friction is a feature, not a bug?
All of that seems unambiguously positive. But – as typically happens – we have adapted to our current “friction-ed” environment in a number of ways that some (at least) regard as positive. Disrupting that environment – eliminating those frictions – may raise significant policy issues.
Consider, 401(k) credit card loans. These are, essentially, a way of reducing the friction in 401(k) plan loan processing, making it easier for a participant to take a loan. There are currently bills in Congress to outlaw these arrangements – mandating friction (or at least, more friction than is necessary given current technology) with respect to loans. Why? Because some policymakers believe that these transactions, which reduce retirement savings, should generally be discouraged.
More broadly, all behavioral economics implementations in retirements savings plans – auto-enrollment, auto-escalation, default asset allocation to a target date funds – depend on participant inertia to get the default choice to stick. Inertia here is literally just a version of friction. You default the participant into saving 3% of her pay and require her to file a form to stop those contributions. Inertia (friction) may prevent her from filing that form and doing so. If all she has to do is click a button on her smart phone, maybe it doesn’t.
Thus, if plan transactions are made easier – more “frictionless” – for participants, it will be harder to nudge them in a particular direction. The benefits of implementation of DLT at the participant level – e.g., the ability to move assets between funds or take loans, hardship withdrawals or other distributions – should be evaluated against this cost.
Are there situations in which transparency is a bug, not a feature?
In a world in which data is stored as a physical object (e.g., on a piece of paper), even where some of that data is (selectively) digitized, secrecy is relatively intuitive. We have seen (e.g., in 401(k) plan fee litigation) that electronic data may be harder to keep secret. Digital is, as they say, forever.
Confidentiality/secrecy – from competitors, possible plaintiffs and regulators – is intuitive with paper: you limit possession of it and hide it. Where all data has been digitized and, indeed, distributed to all participants in a distributed ledger network, literally everyone has a copy of your data. E.g. (and with regard to 401(k) plan fee litigation), they have data on what recordkeeping and investment management fees your plan is paying. Confidentiality/secrecy is implemented by encrypting that data and limiting access to the “keys” that unlock it.
That works, up to a point. But, presumably, in a heavily regulated system like retirement savings plans, regulators will be able to demand access to this (richer) data. And courts, at, e.g., the request of participant-plaintiffs, will be able to order that access be given to plaintiffs lawyers.
Will this increased transparency increase the regulatory/litigation burden on sponsors? Is there a “secure” way of addressing that concern?
Disintermediation and decreasing returns to scale
Why should disintermediation stop at, e.g., custodians and recordkeepers?
DLT is often described as a way of democratizing information, empowering end users and disintermediating intermediaries. With respect to retirement plans, to what extent is the employer “just another intermediary?”
There are certainly those (typically, Republicans) who would like to see our current system of employer-mediated retirement benefits turned over to individuals, by, e.g., significantly increasing IRA contribution limits a la the Bush administration’s ERSA, RSA and LSA proposal.
Interestingly, the state plan auto-IRA initiative (primarily supported by Democrats) is also targeted at increasing the use of IRAs, with employer-involvement necessary only to implement auto-enrollment and payroll withholding.
In a “flat” data world, with minimal securities data management costs, what value do employers add? Because DLT is decentralized, in a fully implemented DLT financial services system there should (theoretically) be minimal economies of scale. If DLT is deployed in payroll (and one has to ask, why wouldn’t it be?), auto-enrollment/payroll withholding could be implemented at the system level.
Still, there are studies indicating that employers (especially, larger employers) add significant value by curating retirement savings and investment. And that value is not just the (marginal) reduction of cost due to scale. Employers encourage savings, limit participant asset allocation choices to an efficient set, implement defaults that improve outcomes and provide education and a culture of retirement awareness.
Moreover, if significant retirement design choices (e.g., about default savings rates or asset allocations) are made at the system level (instead of, as currently, at the employer level), isn’t there a risk of over-centralization – where one mistaken policy choice will negatively affect all participants?
We raise these issues not because there are clear solutions but simply to begin consideration of them as the financial services industry (at least) begins its process of DLT implementation.
Most expect that process to be incremental and (relatively) slow. But experience (e.g., with the Internet and smart phones) indicates that, at some point, the pace of change/adaptation may accelerate.
We conclude with a brief discussion of the ERISA issues implementation of a DLT system may present.
ERISA “in-writing” requirements: ERISA requires that “[e]very employee benefit plan shall be established and maintained pursuant to a written instrument.” Moreover, the Department of Labor has been reluctant to broadly authorize the default use of electronic communications for ERISA-mandated participant communications. And there may be a (to some degree understandable) bias among policymakers against forcing individuals (many of whom are older) to adapt to a new technology.
Thus, the implementation of DLT at the institutional level – among providers, sponsors and regulators – may be easier than at the participant level. And there may have to be a “paper utility” that outputs hardcopy of required statements, etc. for some time.
ERISA “in trust” requirement: Under ERISA, with certain exceptions, “all assets of an employee benefit plan shall be held in trust by one or more trustees.” A “trustee” of a trust, for this purpose, is not necessarily the person with physical custody of the plan’s assets but rather the person who – unless an investment manager has been appointed or the trustee is subject to the direction of a named fiduciary – has responsibility for management of plan assets. Indeed, typically multiemployer (Taft-Hartley) plans are trusteed by a joint union-management board, while custody of the assets is held by a professional custodian. DLT should be able fit within these ERISA constraints.
ERISA indicia of ownership requirement: ERISA requires that the “indicia of ownership” of plan assets be held subject to the jurisdiction of the district courts of the United States. This requirement has presented issues in the past – e.g., with respect to American Depositary Receipts (ADRs) – that have been accommodated by regulators. We would expect that, in any case, the ability to enforce contractual rights (e.g., with respect to a security) will be solved in the financial services industry implementation of DLT.
Tracking prohibited transaction risk: Dealing with the risk of a prohibited transaction should – as long as the risk itself is properly understood – be (if anything) easier under a DLT system. Parties-in-interest can be identified and the software can be programmed to disallow (or hold and flag) transactions with them. The critical element here is an understanding of the underlying legal risk and communication of it to the persons operating the DLT system. No doubt something like that happens today, in a hybridized part-human, part-paper and part-electronic environment. Adaptation of those procedures to an all-electronic environment seems possible.
In this regard, obviously, open (non-anonymous) participation is absolutely necessary. The sort of anonymity/psuedonymity provided by Bitcoin would be unacceptable in any retirement plan DLT implementation or, indeed, in any retirement plan investment.
Adapting fiduciary practice to a DLT environment: DLT will challenge plan fiduciaries, by providing opportunities to reduce costs and increase the efficiency of data use. Like those 401(k) plan fiduciaries who have been slow to adopt more lower-cost institutional fund menu policies, fiduciaries that are slow to take advantage of lower-cost DLT solutions may risk litigation. We are, emphatically, nowhere near this point today, but it may be on us faster than some anticipate.
We will continue to follow this issue.