Innovation in the Stock Market: Exchanges and ATSs

Is something wrong with the structure of the stock market? Both industry participants and scholars have recently faulted the equity market for its lack of innovation. Economists at Harvard, Chicago, and elsewhere have argued that the continuous nature of modern trading bakes in a problematic arms race among high-frequency traders for speed. Stock exchanges process incoming instructions to trade in the order they arrive and as quickly as possible, which can mean in millionths of a second or less. The result, they argue, is a wasteful race for speed in order to trade first on public information. This race would be eliminated if continuous trading was replaced with discrete, periodic auctions (“frequent batched auctions”), say once per thousandth of a second. The market, these scholars also claim, will not fix itself because the nation’s stock exchanges lack robust incentives to appropriately innovate. (See, e.g., Budish, Lee & Shim, 2020).

In a forthcoming paper, I start off with the fact that there are other important markets for trading stock besides the national stock exchanges. I argue that the innovation calculus for alternative trading systems (“ATSs”) differs markedly from exchanges. ATSs, like stock exchanges, are marketplaces in which traders interact to purchase and sell stock from one another. In fact, ATSs and exchanges often function very similarly, with the same trading mechanics, technology, and participants, and both satisfy the statutory definition of a stock exchange. The defining difference between them is that exchanges choose to register as self-regulatory organizations, with closer supervision by the Securities and Exchange Commission (“SEC”), while ATSs make use of an exemption from registration as exchanges to operate in a less regulated environment.

It is important to discuss ATSs and innovation because of ATSs’ distinctive innovation calculus. My central claim is that ATSs have stronger incentives to innovate than stock exchanges. There are several reasons for this. First, ATSs face far lighter regulatory hurdles to beginning operations and changing structure than an exchange, and some important regulations that limit exchange experimentation do not apply to ATSs at all. Second, ATSs are subject to fewer disclosure obligations than exchanges, and as a result they can, in practice, retain their innovations as “trade secrets,” while this is exceptionally difficult for exchanges. Third, ATSs and exchanges have different revenue sources. This is important because ATSs benefit less from the status quo market design than exchanges and thus lack reasons to preserve that structure against desirable innovations, especially those that would reduce opportunities for latency arbitrage. I use newly available data on ATSs to inform and illustrate these arguments.

Strikingly, at least two ATSs already operate a market structure that offers frequent batched auctions. They illustrate the conceptual argument’s point—ATSs have stronger incentives to innovate, and in fact they innovate more. For instance, IntelligentCross offers traders two trading structures, both batched auctions in which orders executing at discrete, scheduled times. Timing of the discrete auctions or “matching events” is randomized by an artificial intelligence model within defined intervals that are determined on an individual security basis, and the intervals last between 200 milliseconds and 900 microseconds. OneChronos also offers a batched auction: “Rather than matching orders continuously as they arrive at the matching engine (as in a continuous limit order book) the ATS periodically holds auctions designed to seek an optimal matching between buyers and sellers across all eligible orders.” OneChronos Form ATS-N, 11(a). Beyond batched auctions, ATSs have introduced multiple other innovations, including different ranking systems for executing orders beyond exchanges’ price, display, and time priority rules; order-initiated auctions; and innovative order types, including ones that allow execution instructions across multiple securities.

So, will ATSs fix the stock market? Are Budish and his co-authors wrong to doubt the equity market’s capacity to innovate? Unfortunately, I doubt it. The reasons lie both in market structure and in law. All currently operating ATSs are dark pools that do not provide order quotations to the public data stream, i.e., they are “dark.” They only provide quotes to their subscribers, if at all. Relatedly, because ATSs’ quotes are dark, they are not protected by the stock market’s regulatory scheme. Even if the best priced quotes anywhere are at an ATS, broker-dealers can ignore them without violating the so-called “trade through” rule. As a result, it seems unlikely that the amount of volume that would flow to an exchange offering a specific innovation will flow to an ATS offering the same structure, because of the difficulty of accessing dark quotes and the lack of a strict prohibition on failing to do so. The ATS could transition to an exchange, but as Budish and his co-authors note, doing so requires the ATS to obtain SEC approval of their innovations under exacting standards. Approval is costly and uncertain, and if the ATS is approved, other exchanges may be able to rapidly copy its innovation. ATSs, it seems, will not fix the stock market’s innovation problem.

Nonetheless, appreciating ATSs’ innovation dynamics has important implications for theory and policy. In terms of policy, ATSs’ many innovations provide a reason, though perhaps not a decisive one, for a classification system in which some trading venues enjoy a lighter regulatory touch. Recent regulatory moves impose greater disclosure obligations on ATSs. They represent a healthy development, but too much disclosure could have a serious downside for ATS innovation, cautioning against an ATS disclosure system modeled on exchanges. In terms of theory, I suggest that economic models should push further than the state-of-play to endogenize trading venues’ choice of legal status. Excluding ATSs from analysis leads to an incomplete picture of stock market innovation and industrial organization. It fails to ask difficult and important questions about why, historically, some ATSs have become exchanges, while others have not, about why there are no ECNs, and why ATSs generate innovations that fail to resolve market structure pathologies.

The paper, written as a chapter for the forthcoming book, Financial Market Infrastructure: Law and Regulation (OUP, Jens Hinrich Binder and Paolo Saguato, eds) (forthcoming 2021), briefly provides an overview of ATSs before making this argument in greater detail.

The complete chapter is available for download here.