Thin Trading and Betas Risk Estimators in an Emerging Country: The Case of the Tunisian Stock Market (Transactions Manquantes Et Mesure Du Risque Des Titres Cotés Sur Le Marché Boursier Tunisien)
18 Pages Posted: 7 Dec 2007
Date Written: December 1, 2007
Abstract
It's frequently reported that the transactions on the Tunis Stock Exchange (BVMT) are irregular et infrequent. The non-trading problem would render the estimation of beta biased. This has implications for researchers using the Market Model (MM) et/or investigating the Capital Asset Pricing Model (CAPM). Indeed, when some of the stocks comprising the stock market index are intermittently traded, the returns of the market index would be serial correlated. Estimates of betas will be biased upward for more frequently traded stocks et biased downward for infrequently traded stocks. This problem has been discussed by several studies on currently most liquid stocks markets: the New York Stock Exchange (Scholes et Williams, 1977), the London Stock Exchange (Dimson, 1979), et the Toronto Stock Exchange (Fowler et Rorke, 1983). These researchers have devised techniques for obtaining unbiased estimates for beta in infrequently traded environment. The aim of our paper is to investigate the impact of non-trading on Tunisian data et to provide several beta estimates of 34 stocks traded on the BVMT from January 1st, 2002 to December 31, 2006. In order to identify which model better explain the behavior of stocks, MM, Dimson, et Fowler et Rorke methods of estimation have been used in the analysis. We find that the instrumental-variable estimator for the market-model proposed by Fowler and Rorke (with a lag of t = - 3 et a lead of t = 3) exhibits more significant results et seems to be most suited to estimate betas on Tunisian market.
Note: Downloadable document is in French.
Keywords: Market model, non-trading, beta, BVMT
JEL Classification: C13, C22, G11
Suggested Citation: Suggested Citation