Impact of high frequency trading on volatility in short run and long run

Computers have overtaken the most of tasks in intraday trading on modern exchanges. From stock picking to deal timing, optimized algorithms are crucial in trading process. This phenomenon is apparent on the spot as well as on derivative markets. In this paper, we consider the effects of high frequency trading on the short term volatility. The aim of the paper is to investigate the links between high frequency trading (HFT) and spot volatility. High frequency with presence of market microstructure noise and also low frequency data from German stock market are considered. We employ Markov switching models to estimate the relationship of dynamics in stock returns and changes in the activities of high frequency traders. Activity of algorithmic traders is estimated by proxy variables based on the average size of trades. The problem of optimal sampling biases is avoided by incorporating Bundi-Russell (2008) test and test of Lagrangian multipliers. Market microstructure noise can cause biasness in the estimates of the empirical volatility measures and models based on such variables. It is mostly caused by bid ask bounce and technical realization of trading on certain exchanges. Most actively traded stocks listed on the German stock exchange (Deutsche Borse) are selected for the empirical analysis. Analyses of optimal sampling suggest that highest frequency without market microstructure noise should be approximately hourly. Results from models confirm the hypotheses of positive impact of high-frequency trading on market volatility. Interesting are conclusions that aggressive trading using market orders have smaller impact on realized volatility than market making using limit orders.