Working Paper: CEPR ID: DP17111
Authors: Thomas Hasenzagl; Filippo Pellegrino; Lucrezia Reichlin; Giovanni Ricco
Abstract: A mixed-frequency semi-structural model is used for estimating unobservable quantities such as the output gap, the Phillips curve and the NAIRU in real time. We consider two specifications: in one the output gap is observed as the official CBO measure, in the other is unobserved and derived via minimal theory-based restrictions. We find that the CBO model implies a smoother trend output but the second model better captures the business cycle dynamics of nominal and real variables. The methodology offers both a framework for evaluating official estimates of unobserved quantities of economic interest and tracking them in real time.
Keywords: realtime forecasting; output gap; phillips curve; semistructural models; bayesian estimation
JEL Codes: C11; C32; C53; E31; E32; E52
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| output gap estimates (E23) | inflation dynamics (E31) |
| output gap (E23) | unemployment rate (J64) |
| model structure (C52) | estimation of economic trends (E66) |
| misinterpreting size of output gap (E23) | costly policy mistakes (E65) |
| output gap (E23) | inflation (E31) |
| model structure (C52) | dynamics of inflation based on output gap estimates (E31) |