Bayesian Mixed‑Frequency Quantile Vector Autoregression: Eliciting Tail Risks of Monthly Us GDP
This paper proposes a novel mixed‑frequency quantile vector autoregression (MF‑QVAR) model that uses a #bayesian framework and multivariate asymmetric Laplace distribution to estimate missing low‑frequency variables at higher frequencies. The proposed method allows for timely policy interventions by analyzing conditional quantiles for multiple variables of interest and deriving quantile‑related #riskmeasures at high frequency. The model is applied to the US economy to #nowcast conditional quantiles of #gdp, providing insight into #var, Expected Shortfall, and distance among percentiles of real GDP nowcasts.