Research Topics

Microstructure, Securities Lending, Short Selling, Information Economics, Regulatory Impact

Working Papers 

You can Only Lend What you Own: Inferring Daily Institutional Trading from Lendable Equity Inventory

revise and resubmit Journal of Finance - With Yashar Barardehi, Zhi Da, and Junbo Wang

We develop a new measure of daily institutional trading activity from securities-lending data. Institutions own nearly 80% of publicly traded U.S. equity, but researchers generally observe their holdings only through quarterly 13F filings. Existing daily proxies rely on trade size or inferred trade direction and have very low performance, while databases of actual institutional transactions cover only a subset of institutions and end in 2015.

The paper’s central insight is that institutions generally make available for lending the shares they own. Consequently, changes in lendable equity contain information about changes in institutional ownership. The proposed measure scales daily changes in lendable inventory by the proportion of institutional holdings typically made lendable for each stock. Because that proportion is highly persistent, the resulting measure can track stock-level changes in institutional positions at a daily frequency.

We validate the measure by aggregating it quarterly and comparing it with observed changes in institutional ownership. It substantially outperforms proxies based on large trades and non-retail trading and even performs better than ANcerno data. Its performance improves further when the sensitivity of ownership changes to lendable inventory is allowed to vary with stock characteristics. 

The paper then applies the measure in three settings. First, we show that institutions reduce their holdings before earnings announcements and rebuild them afterward, potentially contributing to the earnings-announcement premium. Second, institutional buying and selling generate temporary price pressure: stocks experiencing strong net institutional purchases subsequently underperform those experiencing net sales as prices reverse. Third, momentum profits arise only when institutional trading opposes contemporaneous intraday returns. When institutional flows reinforce price movements, prices appear to incorporate information more fully and momentum disappears. These findings support theories in which momentum reflects underreaction to price signals that are not confirmed by institutional trading. 

Shrinking the Market-Inside-the-Market: The Effects of Reducing the Round Lot 

With Su Li, John Ritter, and Kevin Roshak

Standard finance theory treats stocks as having one equilibrium price. However, in reality a continous limit order book means that a stock could have dozens of prices that are constantly changing. The national best bid and offer, or NBBO, is commonly used to make sense of these prices and identifies the best prices at which traders can obtain one round lot of liquidity. Historically, a round lot consisted of 100 shares, allowing better-priced odd-lot orders to exist inside the NBBO and creating a “market inside the market” only visable to market participants with full market data resulting in some market participants having less complete information than others. 

In November 2025, the round-lot size was reduced from 100 shares to 40 shares for stocks priced between $250 and $1,000. The $250 threshold provides a relatively clean regression-discontinuity difference-in-differences setting for estimating the effects of reducing the size of the market within the market. However, the change also creates an important measurement problem. Reducing the number of shares required to establish the NBBO mechanically narrows conventional quoted spreads. We also show that it reduces noise in the midpoint which mechanically reduces computed effective spreads, even if underlying trading conditions remain unchanged.

The paper addresses these problems using MIDAS order-level and depth-of-book data. These data allow us to reconstruct the consolidated limit order book and calculate liquidity measures using constant depth thresholds before and after the reform. This produces an apples-to-apples comparison that separates mechanical changes in measured spreads from genuine changes in market quality.

The results show that liquidity moved closer to the midpoint following the round-lot reduction, although total depth did not necessarily increase. Improvements were modest for already-liquid stocks but considerably larger for less-liquid stocks, where smaller round lots reduced frictions associated with odd-lot liquidity and inventory constraints. On-exchange effective spreads declined meaningfully primarily for less-liquid stocks, while off-exchange effective spreads fell for both more- and less-liquid stocks. The reform also reduced trade-throughs of better-priced orders and produced a modest shift in trading volume toward exchanges.

The paper’s broader contribution is methodological as well as economic. It shows that changes in market definitions can mechanically alter standard liquidity measures and that seemingly technical market-structure reforms can materially affect competition, venue choice, and execution quality.

Published Research

Detecting Informed Trading Risk from Undercutting Activity

Journal of Finance (2026) - With Yashar Barardehi and Qiyu Liu

This paper develops a new measure of informed-trading risk called QIDres. The paper addresses a central empirical problem in finance: informed traders try to conceal their activity, making their behavior difficult to observe. Rather than attempting to identify informed traders directly, the paper measures how liquidity providers respond when they believe they may be trading against informed investors.

The measure builds on “undercutting runs,” in which liquidity providers repeatedly improve the best quote by the minimum price increment as they compete to execute against an expected uninformed marketable order. When informed-trading risk rises, liquidity providers become less willing to compete for incoming orders, causing undercutting activity to decline. Thus the prevalence of undercutting runs is an inverse measure of informed trading risk. The paper measures undercutting activity through the quote-improvement-to-deterioration ratio, or QID, which compares single tick improvements in the national best bid and offer with trade-driven quote deteriorations.

Because ordinary liquidity conditions also affect undercutting, we adjust QID for its relationship with bid-ask spreads and standardize the resulting residual (times -1). The final measure, QIDres, is a measure of informed trading risk that is easy to construct from top-of-book trade and quote data, can be calculated intraday, is orthogonal to liquidity, and performs at least as well, and usually better than other measures of informed trading risk such as PIN, MIA, and ITI.

We show that QIDres increases during five-minute intervals containing known illegal informed trades and around earnings announcements, unscheduled press releases, large changes in short interest, and informed mutual-fund trades. It also captures informed-trading spillovers into industry peers and predicts imminent unscheduled press releases after controlling for alternative measures and market conditions.

When Does the Tick Size Help or Harm Market Quality? Evidence from the Tick Size Pilot

Journal of Financial Markets (2025) - With Yashar Barardehi and Qiyu Liu and Ariel Lohr

This paper examines how minimum pricing increments affect liquidity. The paper argues that tick sizes create a tradeoff between pricing fidelity and undercutting. Smaller ticks allow prices to adjust more precisely to supply and demand which improves market quality. At the same time, smaller ticks make it cheaper for traders to step ahead of displayed orders by offering economically trivial price improvements, which harms market quality.

The SEC’s Tick Size Pilot, which increased the minimum tick from one cent to five cents for selected stocks, provides a natural experiment for studying this tradeoff. The paper shows that the effect of the larger tick depended critically on a stock’s prevailing spread. For stocks with spreads below approximately 10 cents, the five-cent tick generally harmed liquidity by constraining price competition. For stocks with spreads above approximately 15 cents, it improved liquidity by reducing undercutting and encouraging displayed liquidity. Effects were limited or ambiguous for stocks between those thresholds.

These heterogeneous effects help reconcile conflicting findings in the prior Tick Size Pilot literature. Many studies pool near-tick-constrained and wide-spread stocks even though those groups responded in opposite directions. The paper therefore recommends that researchers using the Pilot for causal inference account for prevailing spreads, at minimum dividing samples near the 10-cent threshold. More broadly, the findings show why a uniform tick size is unlikely to maximize market quality: stocks with few ticks inside the spread benefit from finer pricing, whereas stocks with many ticks may benefit from a larger increment that limits undercutting.

Business Cycle Variation in Short Selling Strategies: Picking During Expansions and Timing During Recessions 

Journal of Financial and Quantitative Analysis (2022)- With Eric Kelley

We present evidence that short sellers alternate between stock picking during expansions and market timing during recessions. First, firm-level short interest is a much stronger negative predictor of the cross-section of stock returns during expansions than it is during recessions. High short interest also only predicts negative future earnings announcement returns during expansions. We attribute these findings to short sellers’ emphasis on collecting firm-specific signals. Second, short sellers appear to make factor bets more so during recessions than they do in expansions. These bets tend to pay off as we observe a strong negative relation between the betas of highly shorted stocks and future stock market returns, a result which disappears during expansions. Together, these findings are consistent with theories of information acquisition under attention constraints, endogenous information production, as well as theories of time variation in aggregate overconfidence amongst traders. 

These findings have important market efficiency implications.  If short sellers collect and trade on fewer firm specific signals, then prices may be less informative with respect to firm specific information during recessions.   This finding may help explain recent empirical evidence presented by Loh and Stulz (2018) and Schmalz and Zhuk (2018) which suggests that markets are more responsive to the release of public information during recessions.  If less information is being traded into prices by sophisticated traders like short sellers during recessions, then the revelation of new public information contains a greater amount of un processed information.  Additionally, short sellers play an important role as external monitors of the firm.  If sophisticated traders such as short sellers collect fewer firm specific signals during recessions, then their role as monitors of the firm may diminish. Contrary to the predictions of Povel, Singh, and Winton (2007) this diminished monitoring may increase the likelihood of fraud during down economic times. 

To Own or Not to Own: Stock Loans Around Dividend Payments

-Journal of Financial Economics (2021) - With Corbin Fox and Eric Kelley

A securities lending agreement transfers the legal right to receive dividends – and any tax liability thereof – from the lender to the borrower, the borrower must then reimburse the lender any dividend payments that occur during the life of the loan and return the shares upon recall. In a world where dividends and income from other sources are taxed differently, this distinction matters as it gives investors an opportunity to earn dividend-like income from either dividends paid by the company or substitute dividends paid by a third party. Some investors who ordinarily lend shares may prefer not to do so around dividends if the tax levied on substitute payments exceeds that on dividends themselves, constricting the supply of lendable shares. This is currently the case for many in the U.S. Others seeking ways to avoid receiving dividends paid out by firms may choose to lend and receive a substitute dividend instead. These may include foreign investors—even those domiciled in tax havens—who face mandatory U.S. withholdings taxes of up to 30% on dividends. Demand for equity loans may increase as an arbitrageur with favorable tax status could borrow shares, receive a dividend, and split the tax savings with the foreign investor only to quickly return the shares to their original owner. This setting gives rise to a series of interesting questions. First, do equity lending markets experience meaningful supply and demand shocks when firms pay dividends? Second, do these changes result in lending constraints that are presumably unrelated to stock fundamentals? Third, since the availability of equity loans underpins the economically important role short sellers play in liquidity provision, do these constraints result in a deterioration of liquidity? We answer each question with a resounding “yes”.

Why Do Short Selling Bans Decrease Price Efficiency and Increase Adverse Selection 

- Review of Asset Pricing Studies (2021)

This study examines a puzzle concerning short-selling bans, adverse selection, and price efficiency. Short sellers are generally considered among the most informed traders in markets. Consequently, removing them from the market via a short selling ban should reduce both adverse selection and price efficiency. However, empirical research instead finds that short-selling bans increase adverse selection while reducing price efficiency. These results appear contradictory: greater adverse selection implies that trades are more informative, whereas lower price efficiency implies that prices contain less information.

I explain this puzzle using a model with endogenous information acquisition. When short selling is permitted but costly, investors who already own an asset have stronger incentives to acquire information because they can act on negative signals without first borrowing shares. Information acquisition therefore becomes concentrated among asset owners. A short-selling ban magnifies the disadvantage faced by nonowners, further discouraging them from becoming informed and reducing the total amount of information produced.

This concentration of informed investors among owners explains why a ban increases sell-side adverse selection. The ban prevents nonowners—both informed traders and uninformed liquidity traders—from selling. Because informed investors are disproportionately likely to own the asset, however, the ban removes relatively fewer informed investors than liquidity traders from the pool of sellers. Consequently, the proportion of informed traders remaining among sellers rises. Sell orders become more informative even though the market produces and incorporates less information overall. Price efficiency declines because reduced information in buy orders and the conversion of potential sell orders into uninformative no-trade outcomes outweigh the increased informativeness of executed sell orders.

Evidence from the 2008 U.S. short-selling ban supports these predictions. The ban reduced price efficiency and increased adverse selection, but the increase occurred only on the sell side of the market. This asymmetry also had economically important liquidity effects. Increased sell-side adverse selection was the dominant contributor to higher transaction costs during the ban, causing trading costs for sellers to increase approximately 50% more than trading costs for buyers.