The stability of the capital asset pricing model's parameters in analyzing forest investments

Abstract

Parameter instability can seriously affect the reliability and credibility of the capital asset pricing model's (CAPM's) inference results. This paper combines the cusum of square test (CST) and the log-likelihood ratio (LLR) test with event analysis to determine if the parameters of a nominal and real capital asset pricing model were unstable and, if so, to examine the resulting structural changes in the CAPM's parameters. The forestry asset was defined as the stumpage price of Pacific Northwest ponderosa pine (Pinusponderosa Laws.). The CST and LLR test results indicated that the nominal and real CAPM's parameters were unstable and that a parameter shift occurred in the fourth quarter of 1976. The β indices estimated using the event analysis model and nominal returns were insignificantly different from zero prior to the parameter shift and greater than one after it. The β indices estimated using the event analysis model and real returns were not significantly different from zero prior to as well as after the point of instability. While the CST and LLR tests indicated that the real CAPM's parameters were unstable, the estimated parameters were insignificant, indicating that the β indices calculated using real returns were stable.