engleski

Volatility estimation using high frequency data: Challenges and issues

Availability of high-frequency data enabled the use of more information to estimate ex-post the true but unknown volatility. Old fashion approaches use only closing prices and assume that underlying distribution is time-invariant. However, time-varying realized counterparts support findings that returns are not identically distributed across trading days. Moreover, when dealing with ultra-high frequency data or tick-by-tick observations the enormous amount of data needs to be processed prior to moment’s estimation for two reasons: (1) eliminating microstructure noise and (2) finding appropriate unbiased estimators. This task is challenging because of high-frequency unique characteristics, particularly at emerging markets with non-frequent trading and lower liquidity. Employing data sampled at very high frequency to compute the ex-post measure of volatility at a lower frequency can be helpful in such circumstances, but if data are sampled too sparsely the variance of the estimator increases. This leads us to a conclusion that sparse sampling frequency shouldn’t be chosen arbitrarily. The research objective is to find appropriate data-driven ex-post volatility estimator of integrated variance by usage of realized counterparts which are robust to both microstructure noise and price jumps. Thus, two time scale estimator is utilized and appropriate slow time scale frequency is recommended for emerging markets under consideration, while the fast- time scale sampling frequency is held fixed.

Emerging markets ; ex-post volatility ; fast time scale frequency ; microstructure noise ; price jumps ; slow time scale frequency

nije evidentirano