02.12.2012 Views

Applications of state space models in finance

Applications of state space models in finance

Applications of state space models in finance

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.

124 7 A Kalman filter based conditional multifactor pric<strong>in</strong>g model<br />

switch<strong>in</strong>g framework. In this thesis, time-vary<strong>in</strong>g sensitivities will be modeled directly<br />

as cont<strong>in</strong>uous stochastic random walk processes based on the <strong>state</strong> <strong>space</strong> framework<br />

<strong>in</strong>troduced <strong>in</strong> Chapter 3.<br />

The research goal <strong>of</strong> this chapter is to study the practical relevance <strong>of</strong> time-vary<strong>in</strong>g<br />

factor load<strong>in</strong>gs <strong>in</strong> a multifactor pric<strong>in</strong>g framework. Based on the conditional multiple<br />

beta series to be estimated via the Kalman filter, it will be analyzed whether an explicit<br />

consideration <strong>of</strong> the time-vary<strong>in</strong>g impact <strong>of</strong> macroeconomics and fundamentals on<br />

European <strong>in</strong>dustry portfolios can be exploited <strong>in</strong> a pr<strong>of</strong>itable way, either from a risk<br />

management or from a portfolio management perspective. The present chapter contributes<br />

to the literature a synthesis <strong>of</strong> the classical cross-sectional regression approach<br />

by Fama and MacBeth (1973) and conditional factor load<strong>in</strong>gs.<br />

The chapter is organized as follows. After briefly summariz<strong>in</strong>g the concept <strong>of</strong> factor<br />

model<strong>in</strong>g, Section 7.1 reviews the anomalies literature with a focus on macroeconomic<br />

and fundamental factors. Section 7.2 outl<strong>in</strong>es the conditional multifactor methodology.<br />

It will be differentiated between time series and cross-sectional regressions. Section 7.3<br />

<strong>in</strong>troduces the set <strong>of</strong> common risk factors. Section 7.4 discusses the empirical results<br />

and evaluates the relative out-<strong>of</strong>-sample forecast<strong>in</strong>g ability <strong>of</strong> the proposed conditional<br />

multifactor specification. It will be analyzed whether the statistical superiority <strong>of</strong> the<br />

Kalman filter based model can be exploited <strong>in</strong> practice. Section 7.5 concludes.<br />

7.1 Factor model<strong>in</strong>g<br />

As outl<strong>in</strong>ed <strong>in</strong> ¢ 6.1 the most widely used factor model is the CAPM, <strong>in</strong> which the excess<br />

return <strong>of</strong> the overall market is chosen as the s<strong>in</strong>gle common factor. The measure <strong>of</strong><br />

systematic risk is referred to as market beta. Despite the CAPM’s popularity, various<br />

studies over the last three decades have suggested that a s<strong>in</strong>gle beta model, while describ<strong>in</strong>g<br />

a large portion <strong>of</strong> the common variation <strong>in</strong> returns, may not be sufficient to expla<strong>in</strong><br />

the cross-section <strong>of</strong> returns. The biggest challenge to the CAPM <strong>in</strong>cludes the empirical<br />

evidence that macroeconomic sources <strong>of</strong> risk and company-specific characteristics are<br />

priced beyond market risk. In accordance with these well-documented f<strong>in</strong>d<strong>in</strong>gs, the empirical<br />

deficiencies <strong>of</strong> the CAPM are most commonly expla<strong>in</strong>ed by miss<strong>in</strong>g risk factors or<br />

by a misapproximation <strong>of</strong> the total wealth portfolio. 22 This leads to multifactor pric<strong>in</strong>g<br />

<strong>models</strong> as motivated by the Intertemporal Capital Asset Pric<strong>in</strong>g Model (ICAPM), <strong>in</strong>troduced<br />

by Merton (1973), or the Arbitrage Pric<strong>in</strong>g Theory (APT), developed by Ross<br />

(1976). The basic idea <strong>of</strong> a multifactor pric<strong>in</strong>g model is that the common variation <strong>in</strong><br />

asset returns can be accounted for by multiple common components, or risk factors. 23<br />

Without explicitly differentiat<strong>in</strong>g between the ICAPM and the APT <strong>in</strong> the follow<strong>in</strong>g,<br />

22 Alternatively, <strong>in</strong>stead <strong>of</strong> expla<strong>in</strong><strong>in</strong>g the violations <strong>of</strong> the CAPM us<strong>in</strong>g risk-based arguments,<br />

nonrisk-based explanations have been proposed. They <strong>in</strong>clude, among others, data-snoop<strong>in</strong>g<br />

biases, the existence <strong>of</strong> market frictions, transaction costs and liquidity effects. As nonriskbased<br />

explanations will not f<strong>in</strong>d any consideration <strong>in</strong> this chapter, the reader is referred to<br />

MacK<strong>in</strong>lay (1995) for more details.<br />

23 An <strong>in</strong>troduction to the basic conception <strong>of</strong> asset pric<strong>in</strong>g theory is provided <strong>in</strong> Appendix A.<br />

As the theory on multifactor <strong>models</strong> is well-established, it is not <strong>in</strong>tended to derive the underly<strong>in</strong>g<br />

assumptions <strong>of</strong> the various multifactor pric<strong>in</strong>g <strong>models</strong> <strong>in</strong> this thesis; for any details and

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

Saved successfully!

Ooh no, something went wrong!