THE ECONOMETRICS OF ULTRA-HIGH-FREQUENCY DATA - Engle

Abstract:

Ultra-high-frequency data is defined to be a full record of transactions and their associated characteristics. The transaction arrival times and accompanying measures can be analyzed as marked point processes. The ACD point process developed by Engle and Russell (1998) is applied to IBM transactions arrival times to develop semiparametric hazard estimates and conditional intensities. Combining these intensities with a GARCH model of prices produces ultra-high-frequency measures of volatility. Both returns and variances are found to be negatively influenced by long durations as suggested by asymmetric information models of market micro-structure.

KEYWORDS: transactions data, point processes, hazard functions, survival models, ACD, volatility, ARCH, GARCH, market micro-structure.

Engle draws heavily on his earlier work Engle and Russell (1998). I will try to find, review, and upload this paper. He also draws upon the theories presented in O’Hara’s text, Market Microstructure Theory.

The findings are what would be expected, adding validity to the model.

The durations show very substantial differences over the day and have the typical pattern of high activity at the beginning and end of the day. At the open the average time between trades is about 10 seconds while at lunch it rises to almost 40 seconds. The volatility effect is much smaller but shows the same pattern. At lunch volatility is about 40% of the morning peak. The volatility per trade, which is not shown, is lowest at the open and highest at the end of the day.