Bayesian analysis for functional coefficient conditional autoregressive range model with applications

Abstract

Financial market time series exhibit significant nonlinearity and volatility, and investors, with limited attention, are influenced by abnormal fluctuations. We propose the Functional Coefficient Autoregressive Range (FCARR) model, which extends existing asymmetric range volatility models by incorporating varying coefficient functions to better capture dynamic market changes and asymmetries. Through simulations, we show how well the model handles complicated financial data by utilizing Bayesian P-spline techniques for parameter estimation, model selection, and out-of-sample forecasting. The effectiveness of the FCARR model is underscored through its application to the Chinese stock market, confirming its capacity to capture volatility. This versatile tool helps investors and policymakers better understand and predict market dynamics, especially when information access is restricted.