Working Papers

  • ‘‘Simple robust two-stage estimation and inference for generalized impulse responses and multiple-horizon causality’’ (with Jean-Marie Dufour). Feb 2024, 96 pages.
    We propose a two-stage estimation method for an h-horizon linear projection model and demonstrate that the two-stage estimates outperform the Least Squares estimates in two key aspects: (1) they are generally more efficient, and (2) they obviate the need for HAC standard errors. We apply this two-stage methodology to causality testing and find that the economic uncertainty index has both short-term (1-3 months) and long-term (30 months) causal effects on the economic activity index.

  • “Counterfactual analysis in macroeconomics: identification, estimation, and inference of counterfactual impulse response using instruments” sole author, March 2024, 32 pages.
    This paper examines two prevalent counterfactual analyses in macroeconomics: hypothetical trajectories and policy interventions. It introduces an innovative concept of a counterfactual impulse response. We present a local projection - instrument variables (LP-IV) based approach for identification, estimation, and inference. We apply these two counterfactual analyses to monetary policy, shedding light on the costs and benefits associated with central banks’ efforts in the post-pandemic era to reduce inflation rates by adjusting interest rates as the policy variable.

  • “Causal mechanism and mediation analysis for macroeconomics dynamics” (with Jean-Marie Dufour). Nov 2023, 30 pages.
    previous title “Intervention analysis, causality and generalized impulse responses in VAR models: theory and inference”
    We investigate the causal mechanism in a dynamic model and propose an innovative index designed to offer a quantitative measure for the mediator during causal transmission. The study focuses on the counterfactual scenario where the causal coefficients from the mediator to the output variable are suppressed to zero. The empirical application examines the causal transmission from an inflation shock to GDP through FFR, labor, consumption, and investment.

Work in Progress

  • ‘‘Honest and Uniform Inference in High-Dimensional Linear Projections: An Application of Multi-Horizon Granger Causality and Network Connectedness’’ (with Eugène Dettaa)
    We explore two types of de-sparsified estimates in a high-dimensional linear projection model and derive honest asymptotic Gaussian inference. Applying this methodology to Multi-horizon Granger causality tests, we establish network connectedness between large number of stock volatilities.
  • “Generic identification and practical specification for multivariate time series”
    This paper investigates the generic identification conditions for the vector autoregressive moving average (VARMA) model and presents a practical specification algorithm. Our goal is to offer a flexible VARMA representation as a complement to the rigorous Echelon form. We apply this methodology to impulse response estimation, achieving narrower confidence bands compared to VAR estimation.

  • “The (mis)-identification and estimation of structural impulse responses in sub-space VAR model”
    This paper investigates the effects of impulse response misidentification and estimation under a sub-space VAR model employing Structural VAR (SVAR) and Local Projection (LP). We investigate the bias-variance trade-off between SVAR and LP impulse response estimates under a subspace VAR process.