Document Type


Degree Name

Doctor of Philosophy (PhD)



Program Name/Specialization



Lazaridis School of Business and Economics

First Advisor

Phelim Boyle

Advisor Role

Main Advisor

Second Advisor

Si Li

Third Advisor

Subhankar Nayak


Essay 1 deals with a topic in market microstructure. It examines a type of trading mechanism called a speed bump. The speed bump was introduced to slow down incoming orders to reduce the informational advantage of certain high-speed traders who could exploit their information to take advantage of slower traders. In 2016, the Investors Exchange (IEX) became the first securities exchange in the US to introduce a speed bump. In our first essay, we use trading data from IEX to study this issue. IEX's speed bump includes two layers of delays: a fixed delay on displayed orders and a contingent delay on some non-displayed order types. The contingent delay is triggered when the Crumbling Quote Signal forecasts an unfavorable price change. We build a simple statistical model to show why both delays are essential in reducing adverse selection and empirically assess their joint effect using the publicly available trade and quote (TAQ) data from January to March 2017. We find that compared with nine other major U.S. exchanges, IEX is among the venues with the lowest adverse selection for small- and medium-sized stocks. For large-cap stocks, IEX's performance depends on the physical time interval used for evaluation, where it outperforms most exchanges within 5 seconds after a trade occurs. The contributions of Essay 1 can be reduced to two points. First, we scrutinize the limitations of the speed bump. Second, we do a sensitivity analysis to explore how different choices of sample stocks and evaluation intervals affect the measures of adverse selection.

Essay 2 examines a special phenomenon facing US-listed, Chinese companies: short selling attacks. Compared with US-domiciled companies, Chinese firms are more susceptible to high short selling pressure from short sellers. We identify two types of short selling attacks: non-aggressive ones, identified by the appearance on the Reg SHO Threshold List; and aggressive ones, identified by the release of a negative report from professional short sellers. We find that roughly half of both types of attacks are repeated on the same firms. Using a simple model, we demonstrate that high shorting cost may explain why sellers choose to release their negative information piecemeal. Consistent with our model, first non-aggressive attacks exhibit no clear pattern, suggesting some probing of short sellers. Repeated aggressive attacks exhibit higher shorting intensity than the first attacks, consistent with the confidence of aggressive short sellers. To assess the effect of short selling attacks, we build a portfolio by longing unattacked firms while shorting attacked firms. Since the buy-and-hold returns based on this strategy are below the market index, we argue that attacks on Chinese firms are generally value-destroying. The contributions of Essay 2 are twofold. First, we create a public data set on short selling pressure. Second, we explore the decision-making mechanism of short sellers.

Essay 3 deals with Ponzi schemes and uses simple models to describe their evolution. The evolution of a Ponzi scheme depends on the number of participants (demographics) and the size of the funds available (wealth). The number of participants is increased by new entrants and reduced by those redeemed and left. The funds are increased by the investments of new entrants and reduced by redemptions and the cut taken by the promoter. In this paper, we model both processes and attempt to incorporate realistic assumptions. We adopt a modified susceptible-infected-recovered (SIR) framework to capture the demographic process. For the wealth process, we recognize the real-life practice of a lock-up period, in which investors are not allowed to redeem their investments. Our model can be calibrated to three popular stylized Ponzi schemes. The main contributions of Essay 3 is that we utilize the lock-up period to simplify investors' redemption behavior and propose a well-defined wealth process.

Convocation Year


Convocation Season