Empirical Studies of the Market Microstructure on the Swedish Stock Exchange

Sammanfattning: This thesis consists of five studies on empirical aspects of the market microstructure on the Stockholm Stock Exchange (StSE). The first study presents a stock pricing model which talkes trading and non-trading time effects into account. The model provides a framework for an empirical analysis of daily returns as well as trading/non-trading time returns of the OMX-index. The results show that the process generating the OMX-prices seems to be non-normal, discontinuous and serially correlated. Most noteworthy is the fact that the behaviour of the Monday returns appear to deviate from other weekday returns. The analysis shows that the OMX-index returns appear to be generated by a more complicated process than is usually assumed in e.g. conventional option pricing theory. In the second study, it is argued that the interplay between stock market participants and trading mechanisms gives rise to friction in the market prices. This makes the observed market prices differ from the intrinsic equilibrium values of the stocks. A model for explaining the process by which prices adjust towards their equilibrium values is presented. Accordingly the presence of friction causes the returns to be more volatile and serially correlated. The theoretical model is used to emphasise the persistence of the friction on an intradaily basis on the StSE. A set of data, comprised of OMX-prices obtained evay 15th minute during the daily trading session, i used. Daily 24-hour returns are calculated in different time intervals during the day: open-to-open, intraday-to-intraday and close-to-close. Evidence of different OMX-return generating distributions is found. In particular, the variances of the returns terminating at the beginning of the day are higher compared to the variances of the returns terminating during the middle and towards the end of the day. This implies that there is comparatively more noise in the pricing process just after the opening of the StSE. The third study investigates the profitability and riskiness of stock index arbitrage between the StSE and the OM. Transactions prices of standarised forward contracts, with the OMX-index as the underlying security, and minute-by-minute transactions prices as well as bid-ask quotes of the OMX-index are observed. Several deviations from the cost-of-carry relationship are reported. However, these arbitrage signals do not always give rise to realised profits. Instead, the specific structures of the stock and forward market impose restrictions on index arbitrage The results show that the size of an observed arbitrage signal appears to be a poor predictor of the actual outcome. The restrictions on index arbitrage, especially the settlement risk, are shown to make the strategies risky rather than riskless. The fourth study examines the effects of the extended trading session at the StSE. The variance of OMX-index stock returns is estimated over trading and non-trading time periods, as well as on a 24-hour basis, before and after this event. The variances of the corresponding OMX-forward returns are also estimated and used as a benchmark. The results show that the ratio of trading/non-trading time return variances has increased. It is argued that this increase is due to the extended trading time period pes se, since the same phenomenon can not be observed at the forward market. Since the development of the overall 24-hour variance appears to be unaffected by the extension, these results are consistent with the hypothesis that information is the main source behind the variance of stock returns and inconsistent with the idea that noise trading genrates some of the variance. The fifth study investigates the dynamic behaviour of trading and non-trading time conditional variance of the OMX-index stock and forward returns, using a dynamic GARCH specification. The results show that the conditional variance during both trading and non-trading time depend mainly on the return shocks eminating from trading time. Evidence of a leverage effect in the conditional variance is found during both trading and non-trading time. The study also investigate the effects on the GARCH specification of the trading time extension at the StSE. The extension of the trading hours has affected the relative levels of the unconditional index return variances for trading and non-trading time, but not their dynamic GARCH processes.