IS TIME-VARYING INDUSTRY BETA RISK BROKEN STATIONARY? PANEL UNIT ROOT TEST WITH MULTIPLE STRUCTURAL BREAKS

Dynamics and the instability beta risk have been recognized in previous studies, but as to whether or not there exist structural breaks in time-varying risk there is little known in spite of the issue being addressed in various empirical studies. Therefore, this paper aims to empirically investigate the stability of industry beta risk over time and identify potential and unkown mean revision using the panel data unit root test with multiple structural breaks approach developed by Carrion-i-Silvestre et al. (2005). Time-varying beta is generated by utilizing the Kalman filtering approach for eight main monthly industry portfolio indexes over the period from January 1981 to October 2007, covering several sectors of banking, cement, construction, electrical machinery, food, plastics, pulp and paper, and textiles. The results differ from previous findings in that industry beta risk is found to exhibit time-varying characteristics as well as stationarity based upon the panel unit root test for beta series. Besides the pulp and paper, plastics, and textiles sectors, the empirical evidence reveals three distinct regime changes for the months July 1988, December 1996, and April 2000 in the case of the electrical machinery sector, as well as three distinct regime changes for the months of July 1988 and March 2001 in the case of the food sector.