Development of an Empirical Model
- PDF / 1,720,603 Bytes
- 30 Pages / 441 x 666 pts Page_size
- 0 Downloads / 229 Views
-AfterPidd(1996), p.l5
58
SUCCESS FACTORS OF CORPORATE SPIN-OFFS
3.2 PRECEDING ANALYSES AND UNDERLYING MODELS The following sections describe preceding models for the analysis of the relationship between factors influencing Corporate Spin-Offs.
3.2.1 Univariate Models: Event-Date Analysis A number of authors have investigated different aspects of Corporate Spin-Offs using event date analysis. This type of analysis goes back to Fama et al. (1969) and is extensively described in Brown and Warner (1985 and 1980)'". The methodology analyses excess stock returns around the announcement date, t(o), of the Spin-Off Stock market data are usually drawn from publicly available sources*, from statistical offices** or Universities***. In event date analysis, excess returns are calculated and then tested for significance.
3.2.1.1 Calculation of Excess Returns An estimation period before the Spin-Off announcement serves for calculating the 'normal' return behaviour of the stock without any influence of the Spin-Off announcement. The 'normal' behaviour is compared with the actually observed behaviour during the event period, which includes the announcement date. The underlying assumption is that, if there are abnormal returns around the announcement date, these can be detected by testing the significance of the difference between the expected normal and the observed returns during the event period. This difference is also called prediction error, or abnormal, residual, excess, or announcement date returns. Three different ways have been used for calculating announcement date returns: (a) Mean adjusted returns For the estimation period, for example for the 121 trading days ending 60 days before the announcement date, t(.is(i .^o> the average market return of the stock is calculated:
^'^ See Krishnaswami and Subramaniam (1999); Allen et al. (1998); Daley et al. (1997); Johnson et a!. (1996); Allen et al. (1995); Cusatis et al. (1994); Johnson et al. (1994); Seifert and Rubin (1989); Copeland et al. (1987); Linn and Rozeff (1985); Zaima and Hearth (1985); Hearth and Zaima (1984); Hite and Owers (1983); Kudla and Mclnish (1983); Schipper and Smith (1983); Kudla and Mclnish (1981) * Like The Wall Street Journal, the New York Stock Exchange, or the Capital Changes Reporter. ** For example Compustat, Standard and Poor's, or Moody's. *** The CRSP (Center for Research in Security Prices, University of Chicago) is compiling such data.
DEVELOPMENT OF AN EMPIRICAL MODEL
59
-60
For the estimation period:
r„„ =(1/121)* 2] ^u /=-180
r^ : actual observed return for stock i on day t, Tj^j : average market return on the stock exchange for stock i For the event period, for example for the 80 trading days beginning 59 days before the announcement, /^.jy, ^20)^ the daily prediction error, jUu, is calculated by subtracting the average market return from the actual observed return: For the event period:
ju^ = ri, - r^^^
jUif : daily prediction error (excess stock return), rif : actual observed return for stock i on day t, Fj^j : average
Data Loading...
