16.11.2014 Views

McGraw-Hill

McGraw-Hill

McGraw-Hill

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

184 Seledine Securities<br />

11. Further confirming evidence is provided by various criteria that<br />

determine the order of a stochastic process. Akaiie's (1969) AIC<br />

criterion determines the order of autoregressive processes. The<br />

Hannan-Rissanen (1982,1983) criterion determines the order of<br />

autoregressive/moving-average (ARMA) processes. Both indicate<br />

that this return series cannot be successfully modeled as an ARh4A<br />

process. Significant autocorrelation patterns exist for pure returns to<br />

some other equity attributes. See Jacobs and Levy (1988b).<br />

12 Makridakis, Wheelwright, and M&% (1983) observe (p. 18): "Unlike<br />

explanatory forecasting, time-series forecasting treats the system as a<br />

black box and makes no attempt to discover the factors affecting its<br />

behavior."<br />

13. Chan and Chen (1988b) form two mimicking portfolio-ne of firms<br />

in distress, as measured by reductions in dividend payments, and<br />

another of highly leveraged smaller firms. They find smaller firms<br />

have higher sensitivities to the mimicking portfolios than do larger<br />

firms, even after controlling for firm size.<br />

14. Siar variables were used by Chan, Chen, and Hsieh (1985) to<br />

investigate links with the size effect, by Chen, Roll, and Ross (1986)<br />

to investigate links with stock returns, and by Fama and French<br />

(1987) to investigate links with stock and bond returns. These studies<br />

transform the variables; for instance, default spread measures are<br />

formed from the difference between yields, or returns, on lowquality<br />

corporate and government bonds. The default spread, and<br />

other spreads, are implicitly incorporated our in approach as<br />

differences between the independent variables.<br />

15. Jacobs and Levy have demonstrated, for example, that value<br />

considerations alone are insufficient to explain security pricing. For<br />

instance, the effectiveness of the dividend discount model is<br />

dependent on market psychology [see Jacobs Levy and (1988c)J. And<br />

Japanese investments in U.S. stocks, which are generally<br />

concentrated in larger companies, are influenced by the dollar/yen<br />

exchange rate. For expository purposes, however, liited we our<br />

investigation to the six valuation variables.<br />

16. Although standard VAR models use a uniform lag length for all<br />

variables, Hsiao (1981) has proposed a stopping-point criterion for<br />

choosing the optimal lag length for each variable in each equation.<br />

17. For a critique of the usefulness of VAR models for understanding<br />

macroeconomic relationships, see Runkle (1987). VAR modeling for<br />

small-scale applications was first proposed (1980).<br />

by S i

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!