Essays on Price Discovery and Volatility Dynamics in Emerging Market Currencies

This thesis investigates the price discovery and volatility dynamics in spot foreign exchange (FX) markets of emerging countries by using nearly 20 years of high frequency data. It includes three independent essays. The first essay depicts comprehensive information distribution of emerging market currencies (EMCs) by employing microstructure methods. We separate the 24-hour FX market into two sequential trading periods: daytime and overnight, and then we further divide the active daytime session into onshore and offshore markets. We find that overnight investors have contributed more to pricing EMCs in recent years. The key reason is that price discovery capacity to macro news of overnight investors have improved. Moreover, the onshore dealer information advantage to local news is decreasing, as offshore dealers become more informed in pricing EMCs. Thissuggests that EMCs are more market-determined, and international FX trading hubs now have more information and are providing more liquidity for them. However, it also suggests that EMCs are harder for central banks to manage and they are more fragile to the impact of one or two crucial dealers withdrawing supply of liquidity. In the second essay, we focus on two crucial statistical features of volatility: volatility persistence and return asymmetric effect. By using the heterogeneous autoregressive realized volatility model (HAR-RV model) and its variants, we provide new empirical evidence that EMCs have lower volatility persistence and larger asymmetric return effect than the major currencies (MCs). Furthermore, we find that the daily volatility persistence and asymmetric effect change over time: the former decreases with information flow inconsistency and the latter increases with market illiquidity. The stronger negative impact of news inconsistency and the lower market liquidity level cause EMCs to have lower daily volatility persistence and higher asymmetric volatility than MCs. The essay compares two stylized features of volatility between MCs and EMCs for the first time and suggests that information arrival pattern and market state are crucial determinants. The third chapter investigates the source of long memory in FX volatility. Inspired by Berger et al. (2009) and Patton and Sheppard (2015), we propose a new empirical specification that links volatility to good and bad news, measured as the order imbalance in the market, and to traders’ sensitivity to that news. We estimate the time-varying daily market sensitivity to good or bad news from high-frequency data. We find the explanatory power of bad market sensitivity to volatility is similar to that of good market sensitivity. This finding is different from Patton and Sheppard (2015), who find that bad volatility drives volatility persistence. Furthermore, we use Koenker and Bassett’s (1978) quantile regression model to estimate traders’ time-varying sensitivity to information across the quantiles of the conditional distribution. The empirical results also do not show a different influence on the long memory of volatility between the market sensitivity to extreme good or bad news. However, we find that sensitivity to extreme events has stronger explanatory power than that to other news, which emphasizes the importance of volatility tail persistence. Overall, this chapter expands on Berger et al. (2009) and Patton and Sheppard (2015) and finds the interesting result that price sensitivity to (extreme) good and bad news has similar importance in explaining the long memory of volatility.