Weak Identification of Long Memory with Implications for volatility modelling
Professor Jun YU
Lee Kong Chian Professor of Economics and Finance
Singapore Management University
This paper explores the implications of weak identification in common ‘long memory’ and recent alternative ‘rough’ approaches to modeling volatility dynamics of financial assets. Our analysis unveils asymptotic near observational equivalence between a long memory model with weak autoregressive dynamics and a rough model with anti-persistent shocks and a near-unit autoregressive root. A data-driven semiparametric and identification-robust approach to inference is developed, revealing the effect of these model ambiguities and documenting the prevalence of weak identification in many realized volatility and trading volume series. The forecasting performance and economic value of these two models are examined across a wide range of tradable financial assets.