Academic journal article Journal of Economics and Finance

Specification Errors of Asset-Pricing Models for a Market Characterized by Few Large Capitalization Firms

Academic journal article Journal of Economics and Finance

Specification Errors of Asset-Pricing Models for a Market Characterized by Few Large Capitalization Firms

Article excerpt

(ProQuest: ... denotes formulae omitted.)

1 Introduction

The success of numerous models in explaining returns on the size-BM sorted portfolios has led to additional checks on the performance of the proposed models and the employed performance metrics.1 Lewellen et al. (2010) showed that fascinating results can be reported in terms of high cross-sectional OLS R2. Their study proposes to break the strong factor structure of the size-BM portfolios while adding different characteristic portfolios to the cross-section of test assets among other methodological assessments. How the standardized international evidence is attributed to other capital markets is an interesting topic in its own right. In this study, we report the specification errors of APM for a market with atypical market settings, such as the Finnish stock market.2

The peculiar attributes of the market include the small number of listed stocks and the domination of a few firms in the total market capitalization.3 The 10 largest firms comprise more than 80 % of the total market capitalization; of them, Nokia alone contributed approximately 50 % in the sample. We argue that the specification errors testing for markets like Finland may hamper the inference to have market wide significance with the usual VW portfolio testing.4 Concurring with only one type of test portfolios could produce misleading evidence regarding which model/risk consistently explains the return variability to proxy aggregate risk premium across all the stocks/portfolios in the market, given the non-normal market structure.

Generally, the empirical evidence for the U.S. and other developed markets has no particular departure for using alternating weighting schemes in the construction of test portfolios. The non-effect occurs because no single stock or small number of firms dominates market capitalization to influence model beta risks and subsequent risk premiums over a large period of time. For similar capital markets, we expect that small number of firms may dominate (or limit) the overall yields on the constructed test portfolios as well as the overall orientation of the specification testing. Henceforth, this study carries out the analysis to control for the peculiar market settings and compare the specification errors of the assetpricing models (APM) using both equally weighted (EW) and value weighted (VW) test portfolios. The study not only contributes to the relevant literature from a market with idiosyncratic characteristics but also supplements the Finnish asset-pricing literature.5

The conditional specifications are similar to Schrimpf et al. (2007). The estimations are carried for the period from 1994:07 to 2009:05 using monthly stock returns. We employ the stochastic discount factor-generalized method of moments (SDF-GMM) procedure to allow for time variation in the SDF factors and Hansen and Jagannathan (1996, HJ-distance) based performance metric. We rely on the conventional risk mimicking factors in the literature such as size and value risks proxies of Fama and French (1993) three factor model (FF3) and momentum factor of Carhart (1997). Given the evidence for the time variation in expected return over time, we allow the CAPM specification to have time varying prices of risks following Cochrane (1996). If prices of risks fluctuate over the business cycle, we can capture this effect by using variables that are associated with business cycles.

Overall, the specification errors of APM models show that unconditional CAPM is unable to explain the variations in EW and VW portfolio returns. The results display discernible patterns in the evaluation of risks for EW and VW portfolios to influence the model SDFs. The SMB and HML are important risks capturing variations in the EW expected returns. However, market factor among other model risks dominate VW return variations. The inference is further strengthened from the conditional CAPM estimations. None of the employed (conditional) factors could influence the pricing kernel for the EW test portfolios. …

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


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.