Working Paper: CEPR ID: DP15854
Authors: Jacopo Cimadomo; Domenico Giannone; Michele Lenza; Francesca Monti; Andrej Sokol
Abstract: Monitoring economic conditions in real time, or nowcasting, and Big Data analytics sharesome challenges, sometimes called the three "Vs". Indeed, nowcasting is characterized bythe use of a large number of time series (Volume), the complexity of the data covering varioussectors of the economy, with different frequencies and precision and asynchronous releasedates (Variety), and the need to incorporate new information continuously and in a timelymanner (Velocity). In this paper, we explore three alternative routes to nowcasting withBayesian Vector Autoregressive (BVAR) models and find that they can effectively handlethe three Vs by producing, in real time, accurate probabilistic predictions of US economicactivity and a meaningful narrative by means of scenario analysis.
Keywords: big data; scenario analysis; mixed frequency; real time; business cycles; nowcasting
JEL Codes: E32; E37; C01; C33; C53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| BVAR models (C32) | accurate real-time probabilistic predictions of US economic activity (E37) |
| BVAR models (C32) | manage the three Vs of big data (C55) |
| BVAR models (C32) | handle data irregularities and mixed frequencies (C22) |
| LBVAR, BBVAR, CBVAR (C29) | comparable predictions and impulse response functions (C22) |
| BVAR models (C32) | capture genuine data features (C55) |
| BVAR models (C32) | high correlation with benchmarks like New York Fed's dynamic factor model (E17) |
| differences in performance among BVAR methods (C51) | significant only in the initial weeks of the quarter (G14) |
| BVAR models' predictions (C29) | reflect complex dynamic interactions among macroeconomic variables (E13) |
| mixed-frequency data (C32) | enhance the timeliness of the analysis (C41) |