Can Algorithmic Market Makers Safely Replace FX Dealers as Liquidity Providers?

By Jack Krupinski
(This Money View piece is by my students, Jack Krupinski. Jack is currently a fourth-year student at UCLA, majoring in Mathematics/Economics with a minor in statistics.)

Financialization and electronification are long term economic trends and are here to stay. It’s essential to study how these trends will alter the world’s largest market—the foreign exchange (FX) market. In the past, electronification expanded access to the FX markets and diversified the demand side. Technological developments have recently started to change the FX market’s supply side, away from the traditional FX dealing banks towards principal trading firms (PTFs). Once the sole providers of liquidity in FX markets, dealers are facing increased competition from PTFs. These firms use algorithmic, high-frequency trading to leverage speed as a substitute for balance sheet capacity, which is traditionally used to determine FX dealers’ comparative advantage. Prime brokerage services were critical in allowing such non-banks to infiltrate the once impenetrable inter-dealer market. Paradoxically, traditional dealers were the very institutions that have offered prime brokerage services to PTFs, allowing them to use the dealers’ names and credit lines while accessing trading platforms. The rise of algorithmic market markers at the expense of small FX dealers is a potential threat to long-term stability in the FX market, as PTFs’ resilience to shocks is mostly untested. The PTFs presence in the market, and the resulting narrow spreads, could create an illusion of free liquidity during normal times. However, during a crisis, such an illusion will evaporate, and the lack of enough dealers in the market could increase the price of liquidity dramatically. 

In normal times, PTFs’ presence could create an “illusion of free liquidity” in the FX market. The increasing presence of algorithmic market makers would increase the supply of immediacy services (a feature of market liquidity) in the FX market and compress liquidity premia. Because liquidity providers must directly compete for market share on electronic trading platforms, the liquidity price would be compressed to near zero. This phenomenon manifests in a narrower inside spread when the market is stable.  The FX market’s electronification makes it artificially easier for buyers and sellers to search for the most attractive rates. Simultaneously, PFTs’ function makes market-making more competitive and reduces dealer profitability as liquidity providers. The inside spread represents the price that buyers and sellers of liquidity face, and it also serves as the dealers’ profit incentive to make markets. As a narrower inside spread makes every transaction less profitable for market makers, traditional dealers, especially the smaller ones, should either find new revenue sources or exit the market.

During a financial crisis, such as post-COVID-19 turmoil in the financial market, such developments can lead to extremely high and volatile prices. The increased role of PTFs in the FX market could push smaller dealers to exit the market. Reduced profitability forces traditional FX dealers to adopt a new business model, but small dealers are most likely unable to make the necessary changes to remain competitive. Because a narrower inside spread reduces dealers’ compensation for providing liquidity, their willingness to carry exchange rate risk has correspondingly declined. Additionally, the post-GFC regulatory reforms reduced the balance sheet capacity of dealers by requiring more capital buffers. Scarce balance sheet space has increased the opportunity cost of dealing. 

Further, narrower inside spreads and the increased cost of dealing have encouraged FX dealers to offer prime brokerage services to leveraged institutional investors. The goal is to generate new revenue streams through fixed fees. PTFs have used prime brokerage to access the inter-dealer market and compete against small and medium dealers as liquidity providers. Order flow internalization is another strategy that large dealers have used to increase profitability. Rather than immediately hedge FX exposures in the inter-dealer market, dealers can wait for offsetting order flow from their client bases to balance their inventories—an efficient method to reduce fixed transaction costs. However, greater internalization reinforces the concentration of dealing with just a few large banks, as smaller dealers do not have the order flow volume to internalize a comparable percentage of trades.

Algorithmic traders could also intensify the riskiness of the market for FX derivatives. Compared to the small FX dealers they are replacing, algorithmic market makers face greater risk from hedging markets and exposure to volatile currencies. According to Mehrling’s FX dealer model, matched book dealers primarily use the forward market to hedge their positions in spot or swap markets and mitigate exchange rate risk. On the other hand, PTFs concentrate more on market-making activity in forward markets and use a diverse array of asset classes to hedge these exposures. Hedging across asset classes introduces more correlation risk—the likelihood of loss from a disparity between the estimated and actual correlation between two assets—than a traditional forward contract hedge. Since the provision of market liquidity relies on dealers’ ability to hedge their currency risk exposures, greater correlation risk in hedging markets is a systemic threat to the FX market’s smooth functioning. Additionally, PTFs supply more liquidity in EME currency markets, which have traditionally been illiquid and volatile compared to the major currencies. In combination with greater risk from hedging across asset classes, exposure to volatile currencies increases the probability of an adverse shock disrupting FX markets.

While correlation risk and exposure to volatile currencies has increased, new FX market makers lack the safety buffers that help traditional FX dealers mitigate shocks. Because the PTF market-making model utilizes high transaction speed to replace balance sheet capacity, there is a little buffer to absorb losses in an adverse exchange rate movement. Hence, algorithmic market makers are even more inclined than traditional dealers to pursue a balanced inventory. Since market liquidity, particularly during times of significant imbalances in supply and demand, hinges on market-makers’ willingness and ability to take inventory risks, a lack of risk tolerance among PTFs harms market robustness. Moreover, the algorithms that govern PTF market-making tend to withdraw from markets altogether after aggressively offloading their positions in the face of uncertainty. This destabilizing feature of algorithmic trading catalyzed the 2010 Flash Crash in the stock market. Although the Flash Crash only lasted for 30 minutes, flighty algorithms’ tendency to prematurely withdraw liquidity has the potential to spur more enduring market dislocations.

The weakening inter-dealer market will compound any dislocations that may occur as a result of liquidity withdrawal by PTFs. When changing fundamentals drive one-sided order flow, dealers will not internalize trades, and they will have to mitigate their exposure in the inter-dealer FX market. Increased dealer concentration may reduce market-making capacity during these periods of stress, as inventory risks become more challenging to redistribute in a sparser inter-dealer market. During crisis times, the absence of small and medium dealers will disrupt the price discovery process. If dealers cannot appropriately price and transfer risks amongst themselves, then impaired market liquidity will persist and affect deficit agents’ ability to meet their FX liabilities.

For many years, the FX market’s foundation has been built upon a competitive and deep inter-dealer market. The current phase of electronification and financialization is pressuring this long-standing system. The inter-dealer market is declining in volume due to dealer consolidation and competition from non-bank liquidity providers. Because the new market makers lack the balance sheet capacity and regulatory constraints of traditional FX dealers, their behavior in crisis times is less predictable. Moreover, the rise of non-bank market makers like PTFs has come at the expense of small and medium-sized FX dealers. Such a development undermines the economics of dealers’ function and reduces dealers’ ability to normalize the market should algorithmic traders withdraw liquidity. As the FX market is further financialized and trading shifts to more volatile EME currencies, risks must be appropriately priced and transferred. The new market makers must be up to the task.

Jack Krupinski is currently a fourth-year student at UCLA, majoring in Mathematics/Economics with a minor in statistics. He pursues an actuarial associateship and has passed the first two actuarial exams (Probability and Financial Mathematics). Jack is working to develop a statistical understanding of risk, which can be applied in an actuarial and research role. Jack’s economic research interests involve using “Money View” and empirical methods to analyze international finance and monetary policy.

Jack is currently working as a research assistant for Professor Roger Farmer in the economics department at UCLA and serves as a TA for the rerun of Prof. Mehrling’s Money and Banking Course on the IVY2.0 platform. In the past, he has co-authored blog posts about central bank digital currency and FX derivatives markets with Professor Saeidinezhad. Jack hopes to attend graduate school after receiving his UCLA degree in Spring 2021. Jack is a member of the club tennis team at UCLA, and he worked as a tennis instructor for four years before assuming his current role as a research assistant. His other hobbies include hiking, kayaking, basketball, reading, and baking.