Using On-Chain Data for Policy Research, Part 2: Stablecoin Risk Rundown

1. Introduction

The first post in this series provided an overview of one of the many ways to access on-chain data for use in policy research. This piece builds on that work by introducing a set of policy issues relevant to stablecoins, including what can and can’t be probed using on-chain data. It also covers a few practical steps related to data cleaning. It is intended for policy professionals operating at the academic-practitioner frontier.

2. Framing the policy space

2.1 Risk overview

Fiat-backed stablecoins act as bridges between fiat and crypto. As with intra-crypto bridges, they play a crucial role in supporting the fluid movement of assets across the ecosystem today. But they can also concentrate financial and operational risk depending on their design.

Accordingly, policymakers are worried about stablecoins. Within the past few years, there have been a number of speeches, recommendations, and proposed laws or regulations targeting the perceived risks associated with fiat-backed stablecoins. A lot of the risks concern the speed and reliability of converting between stablecoins and commercial bank deposits, and the broader impact on the financial system if things go awry.

For example, if fiat-backed stablecoins are perceived as “safer” than commercial bank deposits, there could be runs into stablecoins during times of stress as people sell their bank deposits to buy stablecoins.¹ This could cause instability as bank funding (i.e., customer deposits that the banks use to make loans) is drawn down. On the flip side, if people are concerned about the stablecoin’s solvency, there could be mass redemptions — runs out of stablecoins — that are operationally destabilizing to the stablecoin and could affect collateral prices in the extreme.

Additionally, regulators have raised concerns about traditional payment system risks, such as liquidity and operational risk. For instance, what happens if time-critical payments must be made with stablecoins, but the payor does not have any and the mechanisms for converting fiat to stablecoins are temporarily unavailable? This could happen off-hours when the traditional payment systems that serve as on- and off-ramps are closed. The failure of time-critical payments due to liquidity shortfalls or operational issues can transform into other risks and threaten solvency and financial stability.

Designing safe fiat-backed stablecoins means grappling with these challenges. This is something the largest fiat-backed stablecoins have been doing since launch, in a less-than perfect environment, with little regulatory clarity or access to some of the traditional safeguards that benefit and buttress incumbents.

2.2 Structural mechanics

When a customer purchases a fiat-backed stablecoin, one of two things usually happen: (A) an existing stablecoin is transferred to the customer, or (B) a new stablecoin is first created and then transferred to the end-user. Both scenarios are interesting, but scenario (B) is most illustrative for thinking about risks.

What happens when $5 of stablecoins are created? The exact steps vary from stablecoin to stablecoin, but the following steps should occur in theory:

1. Customer A instructs their bank to transfer $5 to the Stablecoin Issuer’s bank.
2. Upon receipt of funds, Stablecoin Issuer purchases collateral to back issuance.
3. When collateral purchase settles, Stablecoin Issuer mints $5 of stablecoins and transfers them to Customer A.

In a perfect world, the three steps would happen sequentially, with limited latency between them, and conditionally, such that a step happens if and only if the other two happen. In the real world, that is not always possible.

Steps 1 and 2 rely largely on legacy financial infrastructures, such as ACH and card networks or the market for U.S. Treasuries. Step 3 relies on the issuer’s internal systems and the blockchain on which the stablecoins are minted. Here, there is a mismatch in control and operational resiliency and efficiency, which is a thorny risk vector for the issuer to have to manage.

The pain points in U.S. retail payments are well-documented. Until the U.S. has ubiquitous, reliable real-time payments, retail payments often settle with a lag that often lasts multiple days. This is in part due to technological factors, but also because of structural and regulatory reasons.

Today’s financial system bundles banking, money, and payments and creates dramatic inefficiencies for consumers and a windfall for large banks. Fiat-backed stablecoins present an opportunity to introduce innovation and competition to the retail payments landscape, but today they sit atop a fragmented system that has significant drawbacks. If a customer pays an issuer in exchange for stablecoins (Step 1), when will the issuer receive the funds, and what happens if the funds-transfer does not occur as planned?

There are also difficulties on the collateral side of the equation. Risk-free assets are not always easy to come by. Most stablecoin issuers today claim to back issuance with cash or cash-like equivalents, such as short-duration Treasury Bills. While Treasury securities are close to risk-less from a market risk perspective, the Treasury market has had its own share of issues due to structural factors that affect market liquidity and operational problems caused in part by outdated technology systems.²

Not to mention the fact that the market settles in T+2, creating a lag between when a collateral purchase is executed by the stablecoin issuer and when the issuer becomes the legal owner of the collateral. If a stablecoin issuer attempts to purchase collateral to back issuance, will the market be open and functioning with enough liquidity to support the purchase?

In theory, these problems could be mitigated in part if issuers of fiat-backed stablecoins were allowed to have master accounts at the Federal Reserve, which is not possible for most nonbanks in the U.S. today.³ Another option would be to allow or encourage stablecoins to be backed by some form of on-chain collateral, which could take the shape of a tokenized high-quality liquid assets (“HQLA”) that can be traded and settled 24/7 using an automated market maker (“AMM”), or another on-chain asset with appropriate safeguards for mitigating market risk and volatility.

Given the real challenges they face, how are stablecoins currently managing risk vis-à-vis inflows and outflows? And what can we glean from the concerns raised by policymakers?

3. Data considerations

3.1 The power of open data

A great advantage of public blockchains is their openness.

This means that stablecoins that exist on public blockchains can be used by other projects or services without any special configuration or approval by the stablecoin issuer. It also means that stablecoin transactions are publicly visible, giving an unprecedented degree of transparency to payments activity on public blockchains.

When someone transfers money to their Venmo account, Venmo keeps that information in a proprietary database — and restricts access. Users implicitly delegate control of their funds to Venmo in the process, and external services have to comply with Venmo’s access policy if they want their services to interoperate with the Venmo payment rail. In addition, regulators have almost zero real-time insight into such proprietary recordkeeping systems.

By comparison, the simple query at the end of the first post is an openly-accessible way to read the records in the public stablecoin database. It extracts data on all on-chain events for the stablecoins in scope. That includes minting and burning activity, which gives a real-time snapshot of stablecoin supply and demand, something that is not currently available from any traditional provider of money and banking services. It also includes transfer activity, which shows how the stablecoins in question are being used with other open, permissionless protocols. The current web payments tech stack is a collection of black boxes and proprietary systems that does not offer anywhere near this level of granularity and transparency.

Here is a CSV that shows all of the raw data for June 16, 2022.⁴

It cannot be understated how valuable this information is, both as a barometer of economic activity and a mechanism for monitoring and containing risk in the financial system.

3.2 Data cleaning and prep for analysis

The scope of this particular tutorial is limited to minting and burning activity as it is the simplest base case for analysis.⁵ After running the query and parsing the topics column into separate fields for topic0, topic1, and topic2, a few additional steps are helpful to get the data ready for inspection.

This Jupyter Notebook shows each step in turn.

First, it is helpful to replace the information in topic0 — which, recall, is the hash of the function signature — with a human-readable description (i.e., “mint” or “burn”), depending on the function hash referenced by topic0. samczsun created a helpful tool for looking up signatures. There is some variation among the five stablecoins with respect to how function names map to the actual act of minting/burning activity, which needs to be accounted for. The tables below provide the mappings, for ease of use.⁶

When topic1 and topic2 include addresses, the fields contain extra padding in the form of leading zeros, which need to be removed before referencing against address records in other locations.

Finally, the amounts in the data field need to be converted from hexadecimal to decimal and scaled appropriately.

4. Concluding thoughts

On-chain data provides only a snapshot into the lifecycle of stablecoins. For example, the minting and burning activity that is observable in the data is a crude measurement for total stablecoin activity, considering that it is not possible to see exactly how the stablecoin issuers are managing deposits and withdrawals — or any other activity that happens off-chain. Still, when compared with the traditional financial sector, the openness and transparency of public blockchains enables a new level of insight into stablecoin supply and demand.

The next and final post in this series will use the data that was extracted and cleaned to generate graphs and run simple analyses. By combining the data with contextual information, we can begin to draw larger takeaways regarding the current functioning of the stablecoin ecosystem. With a better practical understanding of how the system works, including its fragilities, policymakers will be in a better position to create policies that encourage innovation and limit the downside risk of new technologies.