News effects in a high-frequency model of the sterling--dollar exchange rate

This paper uses an extremely high frequency data set on the dollar-sterllng exchange rate to investigate the impact of news events on the very short-term movements in exchange rates. The data set is a continuous record of the quoted price for the exchange rate on the Reuters screen. As such it records some 130,000 observations over an 8-week period. The paper investigates the time-series properties of the data using orthodox regression models, and then by making allowance for a time-varying conditional variance. The conclusions vary significantly in moving to this more sophisticated model. The exercises are repeated now incorporating news announcement effects, letting these affect the level of the exchange rate and then the conditional variance process. Again it is found that the conclusions are radically altered in moving to the increasingly sophisticated model.