Sida Li

Publication

Who Provides Liquidity and When? with Xin Wang and Mao Ye (Theory, Journal of Financial Economics 2021)

Slow liquidity providers can quote better prices than High-Frequency Traders (HFTs) when the opportunity cost is factored in.

• We model competition for liquidity provision between high-frequency traders (HFTs) and slower execution algorithms (EAs) designed to minimize investors’ transaction costs. Under continuous pricing, EAs dominate liquidity provision by using aggressive limit orders to stimulate HFTs’ market orders. Under discrete pricing, HFTs dominate liquidity provision if the bid-ask spread is binding at one tick. If the tick size (minimum price variation) is not binding, EAs choose between stimulating HFTs and providing liquidity to non-HFTs. Transaction costs increase with the tick size but can be negatively correlated with the bid-ask spread when all traders can provide liquidity.

• Presentations: University of Rochester, UCLA, Texas A&M University, the University of Florida, and Washington University at St. Louis, Carlson Junior Conference at the University of Minnesota, the NYU Stern Market Microstructure Conference, the 2nd SAFE Market Microstructure Conference, (*) the Colorado Front Range Finance Seminar, the Bank of Canada-Laurier Market Structure conference, (*) the 2019 Financial Intermediation Research Society (FIRS) conference, the Telfer Annual Conference on Accounting and Finance, the Wabash River Conference at Indiana University, and the Smokey Mountain Conference at the University of Tennessee

Refusing the Best Price?  with Mao Ye and Miles Zheng (Empirical, Journal of Financial Economics, 2023)

We analyze a proprietary message-level NYSE data and provide the first anatomy of state-of-art order types.

• Regulation National Market System (Reg NMS) links fragmented stock exchanges by routing orders to the exchange that displays the National Best Bid and Offer (NBBO). Surprisingly, we find that 57% of orders refuse routing to NBBO. The price distortion in NBBO serves as one driver. As Reg NMS defines NBBO without considering exchange fees, 62% routings lead to worse net prices. Exogenous increases in fee differences increase non-routable order usages, particularly on stocks whose fees account for a large proportion of transaction costs. Besides fueling inventions of complex order types, Reg NMS also incentivizes speed competition among exchanges and traders. 

• Presentations: Louisiana State University, the New York Stock Exchange, (*) the University of Illinois at Urbana-Champaign, Rutgers University, Citadel, Two Sigma Investments, AFA 2021, (*) EFA 2021, 3rd Future of Financial Information Conference, CICF 2021, (*) NFA 2021, and Microstructure Online Seminars Asia Pacific

Working Papers

Should Passive Investors Actively Manage Their Trades? (Empirical) Slides

• Using a novel daily holding data of ETFs, I find most ETFs’ reconstitution trades are mechanical: the entire position is traded on the reconstitution day at the closing price. Since most ETFs track public indices that pre-announce their rebalances, the predictable large trade suffers from 67 bps of execution costs, three times higher than similar-sized institutional trades. Camouflaging on either what or when to trade can help save execution costs. 37% of ETFs use self-designed indices to avoid the pre-announcement of rebalancing stocks and save 30 bps. Another 7% of ETFs track public indices, but they camouflage on their rebalance schedules and save 34 bps. Deploying less predictable rebalance strategies can help passive investors save 9.6 bps per year, which is about two-thirds of the management fees.

• Media Coverage: Financial Times, Ignites, ETF Stream, Bloomberg

The Optimal Price of a Stock: A Tale of Two Discretenesses with Mao Ye (Theory + Empirical, R&R Journal of Finance)

Both the tick size and the lot size are constraining liquidity provision and artificially widening the bid-ask spread.

• Economists usually assume that price and quantity are continuous variables, while most market designs, in reality, impose discrete tick and lot sizes. We study a firm’s trade-off between these two discretenesses in U.S. stock exchanges, which mandate a one-cent minimum tick size and a 100-share minimum lot size. A uniform tick size favors high prices because the bid–ask spread cannot be lower than one cent. A uniform lot size favors low prices because low prices reduce adverse selection costs for market makers when they have to display at least 100 shares. We predict that a firm achieves its optimal price when its bid–ask spread is two ticks wide, when the marginal contribution from discrete prices equals that from discrete lots. Empirically, we find that stock splits improve liquidity when they move the bid–ask spread towards two ticks; otherwise, they reduce liquidity. Liquidity improvements contribute 95 bps to the average total return on a split announcement of 272 bps. Optimal pricing can increase the median U.S. stock value by 69 bps and total U.S. market capitalization by $54.9 billion.

• Presentations: CICF 2021, Econometric Society North America Summer Meeting 2021, Harvard University, (*) the University of Illinois at Urbana-Champaign, (*) Econometric Society Australasia 2021, 5th SAFE Market Microstructure Conference, (*) EFA 2021, Microstructure Exchange Seminar Series 2021, Cornell University, University of Maryland, 2021 New Zealand Finance Meeting, NYU Stern/Salomon Center Microstructure Conference of 2022 (*), CICF 2022 (*), AFA 2023 (scheduled)

List of stocks that would benefit the most from a split (2023-03-31):