A market microstructure perspective on the price formation of cross-listed stocks

Over the past two decades, globalization in capital markets has led to the growth of equity listings in more than one market. Such growth has heightened the levels of competition among stock exchanges, especially in terms of attracting more foreign listings and the associated business opportunities. Hence, finding ways to achieve a competitive advantage over other markets is becoming more crucial for exchanges. This has emphasized the need to understand how prices are formed in multiple markets. In that respect, this thesis intends to add to the understanding of the price formation process for stocks with foreign listings through three empirical studies. In terms of application, this thesis focuses on Canadian stocks which are listed on the Toronto Stock Exchange (TSX) and cross-listed on the New York Stock Exchange (NYSE). The first essay contributes to our understanding of the impact of news arrival on price discovery. It employs macroeconomic news announcements as proxies for new information and examines the impact of these announcements on price discovery of cross-listed stocks. This study reveals that price discovery shifts significantly from Canada to the U.S. during days with a macroeconomic news announcement, regardless of the origin of the news. This finding shows that markets differ in terms of information processing capability, particularly with regard to the processing of market-wide information. The second essay examines the dynamics of price discovery for cross-listed stocks. We model the interactions between daily price discovery measures, trading volume, bid-ask spread, and algorithmic trading activity using a vector autoregression, taking into account lagged and contemporaneous relations among the variables. We observe that price discovery exhibits a trend and persistence over time. Improvements in liquidity increase an exchange's contribution to price discovery, while at the same time, an increase in price discovery leads to better liquidity. We also find that algorithmic trading activity is negatively related to price discovery of cross-listed stocks, which we attribute to the crowding out effect as arbitrageurs make use of computers to trade aggressively and compete for arbitrage opportunities that exist in their respective markets. As a consequence, high-frequency trading by these arbitrageurs push away informed investors, who are disadvantaged in terms of speed. The third essay assesses how information is incorporated into prices in multiple markets. We develop a general model to assess how quotes in dual markets react to information coming from quotes and trades. We further develop this model to extract an implied model for the spreads, the efficient price, and the relative premium between the two markets. We observe that quotes of cross-listed stocks are linked directly to each other. We find evidence of intermarket competition between liquidity providers as indicated by significant impacts of bid-ask spreads on quotes in both markets. We also find that while prices adjust primarily to trades in their respective market, there is some impact by trades from another market. This finding suggests that there is some degree of informational segmentation between markets. On the whole, the above findings describe the mechanisms of how information is incorporated into prices for dually-listed stocks