Estimating quadratic variation of prices and spreads from the Swedish limit order book
(English)Manuscript (Other academic)
The realized quadratic variation is considered to be a suitable measure of volatility of financial prices since it has been shown to be a consistent non-parametric estimator of the increments of quadratic variation. In the approach presented here, measuring volatility of the prices and returns from the Swedish high-frequency limit order book data by means of a non-parametric estimator is extended to measuring volatility from the bid and ask curves. Since they are functions of both prices and quantities, these curves are likely to be more informative about volatility than the ordinary bid and ask prices. In particular, finding an optimal time interval for computations of the squared returns is a crucial step in creating more precise estimators of volatility. The main results confirm the empirical results from some other comparable studies about microstructure effects: The bias of the proposed estimator increases as the interpolation time interval approaches zero. In contrast to previous studies made on more liquid markets, the major reduction of microstructure noise is only obtained when the sampling frequency is fairly low.
Computer and Information Science
Research subject Econometrics
IdentifiersURN: urn:nbn:se:umu:diva-18753OAI: oai:DiVA.org:umu-18753DiVA: diva2:174698