Academic journal article Journal of Financial Management & Analysis

Market and Financial Performance as Related to Idiosyncratic Risk and the Effect of Outlier Screening in Market Studies

Academic journal article Journal of Financial Management & Analysis

Market and Financial Performance as Related to Idiosyncratic Risk and the Effect of Outlier Screening in Market Studies

Article excerpt

Introduction

This is a follow-up research to a study by Lusk, Halperin and Bern1 that examined various extensions of the following conjecture of Roll (p.542)2 as he speaks about the low explanatory power of the CAPM model as reflected through the R^sup 2^ value:

The paucity of explanatory power represents a significant challenge to our science. We ought to discover

(a) measurable influences that will explain the remaining sixty percent, or

(b) a coherent reason why it should forever remain unexplained.

Simply put: Roll offers that the residuals of the OLS onestage two-parameter linear regression - termed Idiosyncratic Risk [IR] - could be a fruitful venue of exploration for discerning latent measurable influences that could enrich the CAPM information. Various researchers including most recently Campbell, Lettau, Malkiel and Xu3, Goyal and Santa-Clara4 and BoutinDufresne and Savaria5 have reported information on IR and to this extent have contributed information consistent with Roll's conjecture. This is the point of departure for this extended research study.

If the extension of Roll's conjecture were true, it provides a clear path to a better understanding of information that one may extract from the OLS-filter by assuming that there is firm-related performance structure in the random residuals left by the OLS filter. However, in this context where one is examining the IR one must exercise care to be attentive to the assumptions underlying the OLS regression filter in particular respecting outliers.

It is well understood that outliers are data that do not follow the OLS two-parameter linear regression model, hereafter OLS-Fi Iter, assumption that the Y. are independent N(μu = β^sub 0^ + β^sub 1^-x^sub 1^, σ^sub 2^) random variables (see Tamhane and Dunlop) such outliers may change the character of the regression fit by repositioning the regression OLS line. Outliers thus may

* affect the orientation of the fitted linear slope,

* affect the intercept of the fitted line, and

* have an impact on IR - i.e., the residuals of the OLSftlter.

Therefore, in the analysis of IR where the intention is to draw performance related inferences from the IR or the CAPM profiles of organizations, one is required to attend to the assumptions of the OLS-filter. For this reason, it is the case that outliers should be eliminated from the dataset used to generate the CAPM and the related IR profiles.

Objectives

In this paper, we will examine the effect of the recommended screening for outliers on IR and the related performance profile of the firm. To do this we will use market returns from three industry groups listed on the NYSE: The Old Economy [OE], and two benchmarks: IPOs and firms from the New Economy [NE]. For accruing the Old Economy firms, we followed guidelines one may intuit from Jungquist7 and Chen8 and Zanini, Lusk and Wolff. OE firms were in the durable soods sectors such as Metals, Heavy Manufacturing, Mining and Chemicals. The Zanini et al. study documents the expected financial performance differences between the Old Economy and the New Economy firms. We used for the New Economy firms that are in the Technology, Electronics and related Light Manufacturing, Software, and Systems Development sectors. In addition, we added a set of non-dot.com technology related firms that were 1993-IPOs traded on the NYSE during the build-up of the so called "Internet-Bubble" from 1 January 1 994 to 31 December 1999. The New Economy and the nondot.com IPO firms seemed ideal benchmarking groups for the OE firms because they are most likely to be more highly sensitive to the developing survival strategies to remain actively traded during the "hyperactive" market during the mid-to late 1990s (see Jungquist)7.

As the accrual period, we selected firms that were traded on the NYSE continuously from 1 January 1994 to and including 3 1 December 1999; this was the time where the market was exhibiting extraordinary growth in particular for the IPO and NE firms. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.