Academic Research Takes Center Stage at IEX

Two Thursdays ago, IEX hosted our first ever Academic Research Conference.

By bringing together academics and industry practitioners, we hoped to foster greater discourse and collaboration around market structure topics — in effect “narrowing the spread” between these two groups of stakeholders. Judging by the conversations we heard throughout the day and the feedback we received, the conference achieved this goal beyond expectations. For that, we are very grateful to all attendees and, above all, the presenters who made this day possible.

After opening the call for papers in early August, we received nearly 60 submissions — addressing a wide range of modern market structure topics. We ultimately selected eight papers, some of which focus on the most important issues being debated today, others on those which will likely be discussed in years to come. For those who were not able to attend, we are providing brief summaries of the eight papers in this blog post. Of course, as we cannot do full justice to the papers here, we encourage you to read the papers, which are hosted on the conference website.

Algos at work

Electronification and automation are some of the most important stories in modern markets so that’s where we started the day.

In “Attention Effects in a High-Frequency World,” Pamela Moulton, together with co-authors Bidisha Chakrabarty and Frank Wang, ask if computerized trading strategies make markets better at incorporating information. The literature has already shown that markets can be slow at pricing in a new piece of information on certain days or when several news items compete for attention. If these “limited attention effects” arise from human trading, might electronic trading make the price formation process faster and more reliable? In the words of Prof. Moulton: yes, because “computers don’t golf.” The paper answers this question in a careful empirical exercise which includes building intelligent proxies for the speed of price formation and the amount of electronic trading.

The next paper, “The Cost of Exposing Large Institutional Orders to Electronic Liquidity Providers,” looks at a different corner of electronic trading: namely single-dealer platforms (SDPs). In that paper Mehmet Saǧlam, Robert Battalio and Brian Hatch study the impact of accessing SDP liquidity on the cost of trading large institutional orders. Their novel dataset from 2012 allows them to map child order routes to institutional parent orders, showing that SDP liquidity results in lower effective spreads but larger price impact. The net effect is an increase on overall trading costs. Furthermore, the trading costs are larger when SPD liquidity is accessed early in the order’s life.

All in all, this session reinforced the intuition that algorithmic intermediation makes markets more responsive to relevant information. This appears to be true whether the information is corporate news or a shift in the supply/demand balance created by a large order.

Not all liquidity is created equal

A critical component of any market is its liquidity — but not all liquidity is created equal. Current fee structures incentivize liquidity provision, but in different ways. Liquidity has also improved, but primarily in large cap stocks and not in small stocks.

Davidson Heath and his coauthors Jonathan Brogaard and Da Huang studied the connection between ETFs and this widening liquidity gap in their paper “The Rise of ETF Trading and the Bifurcation of Liquidity.” ETFs have been on the rise over the last two decades, with total assets under management increasing by multiple orders of magnitude. Looking at Russell 3000 stocks from 2006 to 2018, they find that ETF trading activity makes liquid stocks more liquid, and illiquid stocks less liquid (see chart below). Many ETFs underweight or omit illiquid, expensive stocks when replicating an underlying index. This replication strategy plays a key role in the bifurcation of liquidity.

Katya Malinova presented the paper “Maker-Taker Fees and Liquidity: The Role of Commission Structures,” coauthored with Michael Brolley. Their paper describes a theoretical model where only one exchange exists, offering two types of fee structures to access the exchange. In their model, direct-access traders directly pay (receive) per-share fees (rebates) to remove (add) liquidity. Broker-access traders access the exchange via a common broker. The broker charges all traders the same flat per-trade fee, regardless of the traders’ making vs. taking activity.

The paper studies the implications of the model on market quality, including liquidity provision, fill rates, profit distribution among trader types, and overall trading volume. The results depend in part on the number of broker-access customers and the rebate-fee disparity. For example, if the rebate-fee disparity is sufficiently wide, the direct-access traders provide more liquidity to the market, while fill rates decline for all traders. In other scenarios, the model can lead to decreased trading volume and reduced overall investor welfare.

Today’s markets do not look like yesterday’s. This session’s papers highlighted two of the most important market developments over the past few decades: ETFs and maker-taker pricing. Research plays a key role in understanding today’s and tomorrow’s market structure changes.

Putting prices under the microscope

The Price Microstructure session was an opportunity to zoom in and understand price formation and price dynamics at the sub-millisecond level.

The NBBO midpoint is often taken to be the “most efficient” or “fairest” price. On the other hand, the direction of the next price change can at times be predicted with relative confidence from order book imbalances or other variables. In “The Micro-Price: A High Frequency Estimator of Future Prices,” Sasha Stoikov proposes an alternative concept of fair price that considers these other variables. The paper describes a specific micro-price model based on the actual midpoint, the spread and a notion of order book imbalance. The methodology of the paper can be extended to include other information-rich variables.

In the next paper, “Price Discovery in High Resolution,” Joel Hasbrouck revisits the topic of price dynamics and the direction of information flows in fragmented markets — an area where some of his previous research has been widely influential. This current paper tackles the challenges that comes with high-frequency data. His new methodology facilitates studying the flow of information at the microsecond level.

He applies this methodology to price formation in symbols IBM and NVDA on October 3, 2016. Specifically, the paper studies information flows between SIP vs. exchange feeds, primary vs. non-primary venues, and quotes vs. lit trades vs. dark trades. In his talk, Prof. Hasbrouck described how in each case the picture changes as one “turns up the power of the microscope.” For example, SIP and exchange feeds appear to be equally contributing to price formation at second-level resolutions, but exchange feeds are revealed to be massively dominant at microsecond-level resolutions.

High-frequency time series are challenging to work with when studying modern financial markets. But analyzed with the right tools, they can reveal a tremendous amount of information that less granular data simply cannot.

Let’s talk about exchange competition

Our final session of the day looked at the market structure landscape through the lens of competition: What inhibits competition among exchanges? What introduces competition into the market? And how does this impact investors?

In his paper “Is Equity Market Exchange Structure Anti-Competitive?” Chester Spatt, a former SEC Chief Economist, compares the rebate tiers in modern exchanges to airline frequent flyer programs. Both are opaque, with multiple thresholds, many ways to quality for “status,” and increased benefits to the largest participants. In addition, trading rebates grow with volume, whereas data and co-location prices are fixed. This means that there’s a marginal incentive to route any additional share to an exchange for rebates — which only further increases the value of the exchange’s market data and co-location services.

Prof. Spatt suggests that more direct restrictions on rebates or enhanced disclosures could help. For example, knowing which pricing tier a broker falls into, or how many brokers are in each tier, could go a long way towards increasing transparency into the competitiveness of exchange pricing.

Sean Foley presented “Forming an orderly line — Does protecting empty queues drive excessive fragmentation?”, coauthored with Elvis Jarnecic and Anqi Liu. By analyzing Canadian broker-identified data during the introduction of the trade-through prohibition in Canada in early 2011, as well as its partial removal in late 2016, the authors look at how orders are placed at different exchanges by individual brokers.

They show that queue jumping is a key driver of fragmentation. Reusing Prof. Foley’s analogy from his talk, small venues without order protection rules are akin to cashier-less checkouts in a supermarket: you wouldn’t want to get in line there. However, the trade-through rule prevents brokers from ignoring any venue. In other words, it’s a sure guarantee that the cashier will show up if you get in line. This means that liquidity providers can get a spot at or close to the front of the line across all markets by routing orders to small venues that have little to no existing queue. Their results suggest that policy initiatives like the trade-through rule can further fragment liquidity, but at the same time may also help introduce competition into the market and reduce barriers to entry.

All told, competition between trading venues raises a number of interesting questions around the shape of liquidity in a fragmented market, the role of intermediation, and the ways venues compete with each other via pricing changes or technological innovation. The regulatory framework has a large impact through a variety of rules such as tick size regimes, order protection rules, and various constraints on pricing.

Promoting transparency & innovation through academic research

At IEX, we believe in the power of research and in transparency. By furthering knowledge and understanding, market structure research helps explain real-world phenomena, the interrelationships between participants, and the impact from regulatory and technology changes. Democratizing access to research allows more of the investing public to understand how markets work. Data-driven research and insights are also key to promoting, introducing, and evaluating new innovations in the marketplace.

We hope to hold this event on an annual basis going forward. Collaboration between academics and practitioners is essential as markets evolve. To foster continued discussion, we invite you to email with any interesting research you author or come across, potentially for consideration in a future conference or for academic seminars held at IEX throughout the year. And if you have any ideas on topics for the market structure community to research and explore, don’t hesitate to write as well!