The market model as a Markov switching model using the kalman filter.
Item
-
Title
-
The market model as a Markov switching model using the kalman filter.
-
Identifier
-
AAI9618096
-
identifier
-
9618096
-
Creator
-
Rosenberg, Menahem.
-
Contributor
-
Adviser: Salih Neftci
-
Date
-
1996
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Economics, Finance
-
Abstract
-
This study addresses the problem of testing the Capital Asset Pricing Model (CAPM). CAPM is essentially a one period model, but when tests or estimations are conducted, multiple periods of historical times series data are being used. Previous researches have shown the CAPM parameters to instable. Various attempts have been made to tackle that issue, starting with detection of the parameters shifts through estimating of the continuously shifting parameters.;The purpose of this study is to propose a technique that will provide maximal flexibility to the CAPM estimating procedure. This is achieved by allowing those parameters to change overtime smoothly and continuously, while recognizing the possibility of a nonlinear change in the parameters.;A simple state space form for CAPM is presented, with an additional transition equation for the nonlinear indicator. Once the model is in a state space form, a Kalman Filter is utilized. The Kalman Filter then produces a time series for the CAPM parameters as well as for the nonlinear indicator, the later points to possible shifts in the parameters.;Results of the nonlinear indicator are then compared with other shift detection method (Cumulative Sum of Squared Residuals and Log Likelihood Ratio). The nonlinear indicator was able to detect all the shifts the other method detected. Moreover, it can detect occurrences at the edges of the data set, making it a candidate for real time detection procedures.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
