Essays in Asset Pricing and Market Microstructure

François Cocquemas, PhD

Abstract :

Does market incompleteness matter for market microstructure?: Market incompleteness should matter in theory, but it is difficult to identify and measure the magnitude of its effects, especially on market microstructure. We use a natural experiment at the Tel Aviv Stock Exchange (TASE) to analyze how order submission patterns, trading and hedging strategies, and overall market impact are affected by market incompleteness. Options on the dollar are traded on Sundays on the TASE, while their underlying is not, which we use in a difference-in-differences setting with options on the equity index as control. Overall, our evidence suggests that market incompleteness has two majors effects on traders behaviors. First, likely because of expected higher adverse selection, traders change their patterns, resorting to higher strategic order splitting, nonetheless some- what reducing their overall trading volumes. Second, because of actually lower information caused by the spot market closure, prices are in fact more efficient and volatility is lower, and aggregate trading costs during periods of incompleteness are slightly reduced, suggesting that informed trading plays a lesser role.

Intraday Volatility in the Absence of Public News Arrival: This paper attempts to isolate empirically the impact of private information and noise trading on the realized volatility of a synthetic spot foreign exchange rate, using a natural experiment where public information arrival is almost non-existent. Options on the dollar/shekel rate trade on the Tel Aviv Stock Exchange from Sunday through Thursday, even though the spot market is closed on Sundays. Option traders therefore need to take into account the loss of information that the spot closure creates, as well as the generally lower public information arrival on weekends. First, we confirm the quasi-absence of relevant news on Sunday and that the arrival of public information is indeed related to higher volatility using augmented HAR-RV and HAR-RV-CJ models. Unsurprisingly, the level of volatility is reduced on Sundays. Second, informed trading appears to be more marked, with a steeper U-shaped pattern of intraday volatility on Sundays. Finally, we show that the serial autocorrelation for the first and second 5-minute lags are more negative on Sundays, indicating faster price reversal and thus more efficiency in prices.

Publication date of the thesis

Thesis committee

Supervisor:  Abraham Lioui, EDHEC Business School

External reviewer: Robert Whaley, Vanderbilt University

Other committee member: René Garcia, EDHEC Business School