Trading Frequency and the Efficiency of Price Discovery in a Non-Dealer Market

The increasing popularity of non-dealer security markets that offer automated, computer-based, continuous trading reflects a presumption that institutionally-set trading sessions are economically obsolete. This theoretical paper investigates the effect of trading frequency, a key feature of the trading mechanism, on the efficiency of price discovery in a non-dealer market. The effect of diverging expectations on error-based and overall return volatility is isolated by tracing the market pricing error to the correlation structures of arriving information and pricing errors of individual traders. The analysis reveals that, due to a portfolio effect, an increase in the trading time interval has contradictory effects on the portion of return volatility arising from pricing errors. A greater accumulation of information increases error-based return volatility, but a greater volume and number of traders per session have the opposite effect. The net effect on overall return volatility can go either way. The results show that return volatility of heavily-traded securities is likely to be minimized under continuous trading, but volatility of thinly traded securities may be minimized under discrete trading at moderate time intervals. The probability that the latter will occur increases with the divergence of expectations among traders. These findings challenge the presumption that automated continuous trading in a non-dealer market is more efficient than discrete trading for all securities regardless of trading volume. Findings are applicable to all economies, but have a special relevance for developing countries where often a single market is dominated by small issues and a low volume of trade. As part of the analysis, we show how to correct the biased estimate of inter-session price volatility when observations are less frequent than the trading sessions themselves