Working Paper: NBER ID: w28445
Authors: Ayegl Ahin; Murat Tasci; Jin Yan
Abstract: This paper presents a flow-based methodology for real-time unemployment rate projections and shows that this approach performed considerably better at the onset of the COVID-19 recession in the spring 2020 in predicting the peak unemployment rate as well as its rapid decline over the year. It presents an alternative scenario analysis for 2021 based on this methodology and argues that the unemployment rate is likely to decline to 5.4 percent by the end of 2021. The predictive power of the methodology comes from its combined use of real-time data with the flow approach.
Keywords: unemployment; COVID-19; real-time projections; flow-based methodology
JEL Codes: E24; E32; J6
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| unemployment inflow rates (st) (J60) | unemployment rate (J64) |
| unemployment outflow rates (ft) (J69) | unemployment rate (J64) |
| initial unemployment claims (J65) | unemployment inflow rates (st) (J60) |
| vacancy data (J60) | unemployment outflow rates (ft) (J69) |
| economic reopening (F41) | unemployment rate (J64) |
| flow-based methodology (C67) | accuracy of unemployment projections (E27) |
| historical data and dynamics of unemployment (J60) | projections of unemployment rate (J64) |