Shiqing Ling

1 month ago | onlinelibrary.wiley.com | Yaosong Zhan |Shiqing Ling |Zhenya Liu |Shixuan Wang

1 Introduction The noncausal autoregressive (NAR) model has gained attention in time series modelling for its ability to capture nonlinear patterns within a stationary framework. Unlike causal models, which rely solely on lags, noncausal models incorporate both lags and leads, enabling dependence on future values. This makes them well suited to model the nonlinear and asymmetric features often observed in financial time series, such as asset price bubbles (Nyberg et al. 2012).