What is the Optimal Market Structure for Digital Assets?

It is a hopeful sign of the beginning maturation of the digital asset markets that various players are staking out their positions and arguing their opinions about market structure. The topic often becomes mired in technical details so for this piece, we have taken a step back to discuss a broader set of market structure points, informed by history and experience, aiming to see the forest in addition to the trees.

Market Structures

There are a variety of market structures that have emerged in capital markets. Each has trade-offs and an adopted market structure is a consequence of some combination of the history of the market and the nature of the asset itself. For example, futures trading has become mainly electronic with an order-driven, multi-lateral, central limit order book (CLOB), bonds trade quote-driven by telephone over-the-counter (OTC) bi-laterally, foreign exchange has elements of both OTC and CLOB, and physical gold has an interdealer wholesale OTC market alongside a dealer-to-consumer retail e-commerce market.

None of these market structures is perfect and one might be more “appropriate” depending on the nature of the asset in question (e.g. how often it trades, its fungibility, etc.). Each requires trade-offs in the competing interests and incentives of market participants; for whom is the market optimized? Policy and regulation play an important and influential role in shaping markets and often seek to balance these competing interests.

Examples of abuses emerging from misaligned incentives and interests embedded in some of these structures resulted in fines levied that run into the tens of billions of dollars in the aggregate:

• Dark Pools & principal trading ($13M, $1M, $154.3M, $14.4M, $20.3M)
• FX OTC “last look” ($150M)
• FX ($90.5M, $1.2B & $2.8B, $7M)
• Bond OTC ($26.5M)
• Gold fix ($44M)
• Bond OTC ($2.1M)

A careful observer will note that the above cases often resulted from the actions of the operators of these non-exchange markets and venues, all of which are less transparent than a regulated displayed exchange, and often have structural conflicts of interest that need to be managed. For example, an operator of a venue that also trades principally for its own account, especially when trading versus its own customers, faces a conflict of interest of its own trading profits versus the best interests of its customers (a model that is fairly common in the crypto market).

While regulated exchanges are not exempt from examples of manipulative trading behavior, such behavior is more typically engaged in by participants on the exchange that are caught through the exchange’s surveillance program — it is difficult to find an example where nefarious behavior resulting in fines was engaged in by the operator of a regulated exchange. One could reasonably conclude that the structure, rules, and transparency of regulated exchanges is a contributing factor to this track record. We are not suggesting absolute causality, but our view is that transparency and oversight, with minimized inherent conflicts of interest, create conditions that make it easier to prevent, or catch and root out, malicious behavior.

A lot of the market structure debate revolves around the costs of competition between various types of professional trading firms. Circling back to the question about optimization, asymmetric speed bumps are examples of market operators making a conscious decision to favor one category of participant over another versus acting as a neutral venue. Matt Levine at Bloomberg did a great job with this complicated subject so we will not rehash it here.

So, the question that the crypto market should be asking itself is, “What is the optimal market structure for trading cryptocurrencies?” A lot of crypto trading currently takes place via opaque OTC trading desks. This might serve the counterparties that have established exclusive bi-lateral trading relationships for both large trades, and increasingly, small electronically routed orders. In an environment where there is much debate and discussion about when more institutional and mainstream investors will enter the market, it is worth considering to what degree the opacity, exclusivity and lack of oversight of the OTC market structure is making it harder and/or less attractive for broader market participation.

Displayed Central Limit Order Books and Block Trading Facilities

In our view, a displayed (“lit”), fair access, multilateral CLOB is the market structure that optimizes to serve the broadest set of participants for liquidity access and price discovery. An exchange-based block trading facility that enables block trades to be reported and broadcast to the market solves for large sized trades that would otherwise impact the market on a CLOB, while integrating valuable trade information into the market. This transparent market structure is optimized to make trading, and trading data, accessible broadly and openly to market participants.

An exclusive-access dark pool does not contribute to pre-trade price discovery and its liquidity is available to only a select subset of participants. The more than 45 dark pools in U.S. equities let their broker-dealer operators “internalize” matching buy and sell orders to avoid executing those orders on lit exchanges where they would be subject to exchange trading fees, while simultaneously relying on those exchanges’ prices as references for trade matching. Those dark markets cannot work without the lit markets. It is worth noting that dark pools are not as prominent in Canadian equities, where rules and market structure make them less attractive; brokers can “internalize” on exchange via the broker-priority matching logic, and rules require that dark pool trades provide meaningful price improvement over the displayed quotations. So even in a market that looks a lot like the U.S., and where about 50% of trading volume is on interlisted stocks (companies listed in both the U.S. and Canada), Canada has optimized differently and proves that dark pools are not, by definition, necessary.

Clearinghouse-Based Settlement

Both CLOB and block trade execution facilities benefit from a post-trade clearinghouse model that optimizes for the elimination of settlement risk; every participant can trade with every other participant without requiring each to assess the risks and creditworthiness of their counterparts.

A bi-lateral OTC market structure requires participants to establish and monitor credit relationships with all other participants with whom they wish to trade and these participants face settlement risk, also known as Herstatt Risk, named after a German bank that went bankrupt and failed to send U.S. dollars to settle trades with counterparties after those counterparties had already sent it their Deutsche Marks. Due to this risk, the price and volume at which a participant is willing to trade is impacted by the risk of a given counterparty in an OTC bi-lateral market. The result is liquidity and prices that vary from participant to participant that can be dislocated from the “fair value” of the asset and impede price discovery.

What’s Past is Prologue

With experience building, trading in and on, and operating markets around the world in mature asset classes, we have seen similar versions of what is now emerging in crypto asset market structure play out before and are reminded of the adage, “Those who cannot remember the past are condemned to repeat it.” The current trend in crypto trading of aggregating multiple OTC/dark liquidity providers is a structure that benefits those that currently have the most information and, in fact, preserves their privileged advantage of information asymmetry.

During the past decade select groups working in OTC FX trading came to believe that FX prices and liquidity to other participants in the market were “theirs” to uniquely determine, and the previously referred to practice of “last look,” was a structural means for them to asymmetrically protect themselves from “predatory” trading by sophisticated counterparties.

Sunlight is the Best Disinfectant

In instances such as those noted above that resulted in fines, there was outright malfeasance taking place in the opacity. But a lack of transparency also creates profitable information asymmetries that don’t necessarily amount to malfeasance. For example, by building trading profiles across counterparts, OTC liquidity providers can optimize prices to be “just tight enough” to extract the maximum willingness to pay from counterparts. It is conceptually similar to dynamic pricing on the internet for consumer goods where location, browser, computer type, and previous purchasing history can be compiled by sellers to create profiles of customers and charge higher prices to those most likely to pay them. In a lit CLOB, liquidity providers must compete outright in an open market where other liquidity providers may be willing to offer superior pricing.

In Conclusion

A fair access displayed CLOB exchange design, that enables the largest most diverse set of participants with different market views and time horizons to interact transparently has been time tested and proven in U.S. equities and U.S. futures, two of the most demanding and electronic markets in the world.

An OTC or private dark pool market structure encourages exclusive, fragmented pools of trading with prices and liquidity that are not fully exposed to competition, while being fully-exposed to the settlement risk of bi-lateral credit relationships. If that is not a desirable outcome then the next logical question is whether the current cloud-based, retail-focused exchanges are capable of supporting institutional trading moving from the OTC markets to the CLOB markets for the maximum, most efficient interaction of participants broadly. If not, for some of the reasons we described in a piece for CoinDesk, then the market may remain bifurcated between retail cloud-based exchanges and institutional OTC desks, with the attendant negative impact on liquidity and price discovery. That market structure may, in turn, present ongoing obstacles to greater participation and stunt the growth of the overall market. In conclusion, we would prefer to see the benefits of price discovery and liquidity aggregation in a market structure based on a fully displayed, transparently operated, broadly accessible CLOB and clearinghouse, such as the model ErisX operates, that removes settlement risk that can serve all participants.