The dynamics of bid-ask spread in an order driven market: the case of Indian stock market

Abstract

Bid-ask spread is the most important part of transaction costs for any asset class in the financial market. It affects the way information gets impounded into prices and thus price discovery of the asset. Bid ask spread, which is the difference between lowest sell and highest buy, is a cost an investor/trader incurs in round trip trading. Market microstructure studies show that the type of market has a significant effect on the bid-ask spread. There are mainly two types of markets: order-driven market and quote- driven market. Liquidity is provided by the dealers in a quote- driven market and by the limit orders in an order-driven market. Due to different nature of the liquidity providers, the behaviour of bid-ask spread differs in order- driven and quote-driven markets .Unlike the quote driven market such as that of the US, where bid-ask spread has been studied extensively, there is a paucity of empirical work in order-driven markets. Indian stock market provides an opportunity to investigate the behaviour of the bid-ask spread in an order-driven emerging market. This study attempts to analyze the nature of bid-ask spread in the Indian Capital Market with 3 main objectives: (1) estimation of the components of the bid - ask spread; (2) investigation of the time series variations in the bid- ask spread; (3) examination of the cross-sectional determinants of the bid-ask spread. The sample consists of tick by tick data for the time period January, 2002 through October, 2008 of 160 stocks traded on National Stock Exchange of India. In the first part of the thesis, we estimate the bid-ask spread and its components {information asymmetric costs; combined inventory & order order processing costs) and implied bid-ask spread using theoretical models. We find that all the models used in the study produce consistent estimates of the bid-ask spread as well as its components. In the Indian Stock Market, we find that the asymmetric information cost and the combined order processing and inventory holding are around 50 percent. We also find that the estimated bid-ask spread is approximately 80 percent of the quoted bid-ask spread. In our sample period, we find that the relative bid-ask spread has decreased over the years in our sample period. In the second part of the thesis, we examine time series behaviour of the bid-ask spread. We study the commonality in bid-ask spread and find that the Indian Stock Market has pervasive commonality in bid-ask spread with almost all the stocks showing this effect. Market wide bid-ask spread affects the bid-ask spread of almost all the individual stocks (>98.75 %). We also examine whether the introduction of the derivatives affects the bid-ask spread of the underlying stock. We find that there is a decrease in relative bid-ask spread of the underlying stock but this decrease is insignificant. We also examine the asymmetric nature of the bid-ask spread. Here, we investigate whether the negative returns lead to a higher bid-ask spread vis-a-vis positive returns. The relationship of negative and positive market returns as well as idiosyncratic positive or negative returns with bid-ask spread is not significantly different. Even large returns do not seem to affect changes in bid-ask spread significantly. In order to understand intraday dynamics of the bid-ask spread, we study the intraday variations in the bid-ask spread. The intraday relative bid-ask spread as well as its components pattern are found to be reverse J shaped. There is a significant decrease in the relative bid-ask spread in the first 35 minutes of trading and a significant increase in the relative bid-ask spread in the last 30 minutes of the trading. We also study co-variation of intraday relative bid-ask spread with intraday volume, intraday volatility and intraday trade size. The relative bid-ask spread is high when the volatility is high during the day. Intraday relative bid-ask spread is also high when the opening volume is high but it is low when the closing volume is high. This may be due to information based volume during the opening of the trading and less information based volume (day trading) towards the closure. Finally, it is found that intraday relative spread bid-ask spread is low when the intraday trade size is large. We examine the intra- week pattern of bid-ask spread and find that there is not much variation in relative bid-ask spread across the weekdays. In the third and last part of the thesis, we investigate the cross-section- al determinants of the bid-ask spread. Cross-sectional determinants investigated are price, market capitalization, number of shares, trade size, volatility, percentage of institutional holding, promoter holding, book-to-mark- et, and debt-to-equity. We find that relative bid-ask spread is positively related to volatility, debt to equity, percentage of promoter holding, and trade size. Relative bid-ask spread is negatively related to number of trades , market capitalization, percentage of institutional holding, and book to market. However, the results are statistically significant only for number of trades, market capitalization and volatility as the explanatory variables. We conclude that in a cross- section of stocks, large market capitalization and highly traded stocks exhibit smaller relative bid ask spread, ceteris paribus. We also conclude that more volatile stocks exhibit larger relative bid-ask spread in the Indian Stock Market. As growing number of markets are adopting electronic limit order markets as the preferred market structure, this study provides an opportunity to understand the behaviour of bid- ask spread in an emerging order- driven market. Most of the empirical findings of our study are consistent with theoretical market microstructure models & empirical research elsewhere. Further understanding and more accurate estimates of the bid - ask spread require information on buy/sell indicator which is currently not being complied by the Indian exchanges. The results have implications for both policy makers as well as practitioners. The findings related to the impact of the introduction of derivatives on the bid- ask spread of the underlying throw some light on debate of the effect of derivatives on the efficiency of price discovery. The pattern of the intraday and in traweek variation may also help uninformed investors as well as larger traders to make better trade decisions.