A survey of sampling-based Bayesian analysis of financial data |
| |
Authors: | James M. Sfiridis Alan E. Gelfand |
| |
Affiliation: | Tokyo Branch , The First National Bank of Chicago , 7th Floor, Tokyo, 105, JapanHibiya Central Building, Minato-ku, 1-2-9 Nishi-Shimbashi |
| |
Abstract: | The capability of implementing a complete Bayesian analysis of experimental data has emerged over recent years due to computational advances developed within the statistical community. The objective of this paper is to provide a practical exposition of these methods in the illustrative context of a financial event study. The customary assumption of Gaussian errors underlying development of the model is later supplemented by considering Student-t errors, thus permitting a Bayesian sensitivity analysis. The supplied data analysis illustrates the advantages of the sampling-based Bayesian approach in allowing investigation of quantities beyond the scope of classical methods. |
| |
Keywords: | Event Studies Inference Bayesian Markov Chain Monte Carlo Gibbs Sampler |
|
|