CFDP 2334

Weak Identification of Long Memory with Implications for Inference


Publication Date: June 2022

Pages: 41


This paper explores weak identification issues arising in commonly used models of economic and financial time series. Two highly popular configurations are shown to be asymptotically observationally equivalent: one with long memory and weak autoregressive dynamics, the other with antipersistent shocks and a near-unit autoregressive root. We develop a data-driven semiparametric and identification-robust approach to inference that reveals such ambiguities and documents the prevalence of weak identification in many realized volatility and trading volume series. The identification-robust empirical evidence generally favors long memory dynamics in volatility and volume, a conclusion that is corroborated using social-media news flow data.

Supplemental material

Supplement pages: 9

Keywords: Realized volatility; Weak identification; Disjoint confidence sets, Trading volume, Long memory

JEL Classification Codes: C12, C13, C58

PDF icon d2334.pdf