A Study on the Indian Market Microstructure

Student Name: Rakhi S

Liquidity has increasingly become an important tool for the investors to select their trading venue and securities. Bid-Ask spread is widely used as a measure of liquidity in the dealer market, where it is directly observable. A smaller Bid-Ask spread is an indication of a security being more liquid. In an oder-driven market like NSE, India, the Bid-Ask spread is not directly observable. Though it can be computed using Roll's formula from high frequency intra-day transaction data in an order driven market, the utility of this metric as a proxy for liquidity measure in an order driven market has not been extensively studied in the literature. In this study an attempt has been made in this direction. High frequency transaction data are collected for July-August 2004 for 50 Nifty stocks and 27 randomly selected NSE listed but outside S& P 500 stocks (and thus expected to be much less liquid). Next daily Bid-Ask spread, measured using Roll's formula, rupee volume, volatlity, median share price and number of trades are observed for these stocks. The Bid-Ask spread is then regressed on the remaining four variables for each stock using the daily data and the regression coefficients are recorded. Next these regression coefficients are studied individually for the two groups and compared between themselves using one and two sample t-tests and Wilcoxon test. It is found that for the highly liquid 50 Nifty stocks, the Bid-Ask spread decreases as the rupee volume increases, or volatility decreases, or median share price decreases, or number of trades decreases. While the first three results intuitively makes sense the last one is counter-intuitive giving rise to the question of suitability of either Roll's formula or the notion of Bid-Ask spread being a "good" measure of liquidity in an order-driven market. For the illiquid stocks however though the directions of influence of the four variables on the Bid-Ask spread are found to be similar to that of the highly liquid stocks, the results of the rupee volume and the number of trades are found to be statistically insignificant. That is for the illiquid stocks it is found that the Bid-Ask spread decreases as the volatility increases, or median share price increases, but it has no significant association with rupee volume or number of trades. No significant difference is found to exist between the liquid and illiquid stocks for all the four regression coefficients. Thus it is concluded on the whole that irrespective of a stock being liquid or illiquid, caution should be taken in using Roll's formula for the Bid-Ask spread as a proxy for liquidity for an order driven market, necessiating further research for an appropriate measure of liquidity of securities being traded in an order-driven market.