Applications of state space models in finance
Applications of state space models in finance
Applications of state space models in finance
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
1.2 Research objectives 3<br />
1.2 Research objectives<br />
Although these advances spurred the <strong>in</strong>terest <strong>in</strong> apply<strong>in</strong>g advanced time series techniques<br />
such as the Kalman filter and Markov regime switch<strong>in</strong>g <strong>models</strong> <strong>in</strong> economic and<br />
f<strong>in</strong>ancial analysis <strong>in</strong> recent years, a transportation <strong>of</strong> these concepts from theoretical<br />
work <strong>in</strong>to applied research on a broader scale is still underdeveloped. Inspired by the<br />
quote at the beg<strong>in</strong>n<strong>in</strong>g <strong>of</strong> this chapter, this thesis analyzes the relative merits <strong>of</strong> selected<br />
elaborate econometric methods to model change <strong>in</strong> the context <strong>of</strong> widely used<br />
concepts <strong>in</strong> f<strong>in</strong>ance. The exploration <strong>of</strong> the dynamics <strong>of</strong> f<strong>in</strong>ancial markets is aimed at<br />
an improved understand<strong>in</strong>g and model<strong>in</strong>g <strong>of</strong> real-world phenomena. The follow<strong>in</strong>g three<br />
research objectives are addressed:<br />
1. Provide a notationally conformable <strong>in</strong>troduction <strong>of</strong> Gaussian <strong>state</strong> <strong>space</strong> <strong>models</strong>,<br />
the Markov regime switch<strong>in</strong>g framework and conditional heteroskedasticity <strong>models</strong>.<br />
2. Analyze which model<strong>in</strong>g technique is best able to model and forecast time-vary<strong>in</strong>g<br />
systematic beta risk as a stochastic process.<br />
3. Evaluate the practical relevance <strong>of</strong> tak<strong>in</strong>g time-variation <strong>in</strong> factor sensitivities explicitly<br />
<strong>in</strong>to account.<br />
As the different contemporaneous time series <strong>models</strong> orig<strong>in</strong>ate from different discipl<strong>in</strong>es,<br />
very different notation and term<strong>in</strong>ology is commonly employed to outl<strong>in</strong>e the respective<br />
theory beh<strong>in</strong>d these concepts. The first objective <strong>of</strong> this thesis is to <strong>in</strong>troduce the theory<br />
<strong>of</strong> the different <strong>models</strong> at hand <strong>in</strong> a unified notational framework: l<strong>in</strong>ear Gaussian <strong>state</strong><br />
<strong>space</strong> <strong>models</strong> and the Kalman filter, the Markov regime switch<strong>in</strong>g framework, as well<br />
as two <strong>of</strong> the most prom<strong>in</strong>ent <strong>models</strong> for time-vary<strong>in</strong>g volatility, namely autoregressive<br />
conditional heteroskedasticity (ARCH) and stochastic volatility <strong>models</strong>. This will allow<br />
the applied researcher to adopt the various concepts without hav<strong>in</strong>g to deal with different<br />
notation that is typical for the discipl<strong>in</strong>es <strong>in</strong> which the <strong>models</strong> were orig<strong>in</strong>ally employed.<br />
It is <strong>in</strong>tended to provide the methodology for the model<strong>in</strong>g <strong>of</strong> time-vary<strong>in</strong>g relationships<br />
<strong>in</strong> a way that is as compact and <strong>in</strong>tuitive as possible and as comprehensive as necessary.<br />
The outl<strong>in</strong>e <strong>of</strong> the respective basic ideas and estimation procedures <strong>in</strong> Chapters 3–<br />
5 illustrates that both Markov regime switch<strong>in</strong>g and stochastic volatility <strong>models</strong> are<br />
closely related to the l<strong>in</strong>ear Gaussian <strong>state</strong> <strong>space</strong> framework and the Kalman filter.<br />
The second contribution <strong>of</strong> this thesis is a systematic and comprehensive analysis <strong>of</strong><br />
the ability <strong>of</strong> the different techniques under consideration to model and forecast the timevary<strong>in</strong>g<br />
behavior <strong>of</strong> systematic market risk. The rationale beh<strong>in</strong>d start<strong>in</strong>g the empirical<br />
analysis with an application <strong>of</strong> the selected time series techniques to the s<strong>in</strong>gle-factor<br />
CAPM, is motivated by the fact that the CAPM beta is widely established <strong>in</strong> practice.<br />
It is used, for example, to calculate the cost <strong>of</strong> capital, to identify mispric<strong>in</strong>gs and to<br />
estimate an asset’s sensitivity to the broad market. As discussed by Yao and Gao (2004)<br />
betas, while traditionally employed <strong>in</strong> the context <strong>of</strong> s<strong>in</strong>gle stocks, are particularly useful<br />
at the sector level. However, <strong>in</strong> spite <strong>of</strong> various studies deal<strong>in</strong>g with the model<strong>in</strong>g <strong>of</strong><br />
conditional sector betas <strong>in</strong> other regions <strong>of</strong> the world, similar work <strong>in</strong> a pan-European<br />
context, where the advancement <strong>of</strong> European <strong>in</strong>tegration and the <strong>in</strong>troduction <strong>of</strong> a s<strong>in</strong>gle<br />
currency <strong>in</strong>creased the importance <strong>of</strong> the sector perspective over recent years, is still