A Markov Switching Causal-Noncausal Autoregressive Model with Application to Economic Bubbles

• Activity Number: 168 - Bayesian Models for Gaussian and Point Processes
• Type: Contributed
• Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Anton Lobach* and Gavino Puggioni
• Companies: University of Rhode Island. Dept. of Computer Science and Statistics and University of Rhode Island, Department of Computer Science and Statistics
• Keywords: Mixture models ; Heavy tails errors ; Forecasting ; Bayesian models ; Exchange rates ; Noncausal processes

Abstract

Economic bubble phenomena may result in explosive trends followed by a steep decline in financial time series. A bubble can be viewed as a positive feedback loop, when investors' future profits expectations influence the present market value of securities and vice versa. These dynamics can be better captured by mixed causal-noncausal autoregressive processes in comparison to traditional ARIMA models. In this work we propose Markov switching mixed causal-noncausal autoregressive processes (MSMAR) to account for changes in regime at different times. Parameter estimation is performed in a Bayesian framework via MCMC algorithms. The model is tested for performance with a simulation study and then applied to Bitcoin/USD exchange rate and US inflation data.