Regime-Switching Factor Models and Nowcasting with Big Data

IMF Working Papers, September 6, 2024

Source details

Demonstrates that the Expectation-Maximization (EM) algorithm for regime-switching dynamic factor models:
- Provides satisfactory performance relative to other estimation methods.
- Delivers a good trade-off between accuracy and speed, making it especially useful for large dimensional data.
Advantages over traditional numerical maximization approaches:
- Benefits from closed-form solutions for parameter estimation.
- Enhances practicality for real-time applications and historical data exercises with focus on frequent updates.
Suitability:
- The EM algorithm proposed is suitable for various regime-switching configurations.

Scope:
- Studies the information content and relative performance of regime-switching models after each data release in a fifteen year period.
- The time efficiency of the proposed estimation methodology made this sequence of updates feasible.
Key finding:
- Confirms existing literature that nowcasting models under regime-switching improve performance.
- Shows that superior nowcasting performance is observed particularly when key economic indicators are released.

Findings on recession dating:
- The model can closely match the recession starting and ending dates of the NBER despite having less information than actual committee meetings.
- The fit between actual dates and model estimates becomes more apparent with additional available information.
- Recession end dates are fully covered with a lag of three to six months.

Provides economists and policymakers with a valuable tool for:
- Conducting comprehensive analyses ranging from point estimates to information decomposition and persistence of recessions in larger datasets.
Practical strengths for policy and real-time monitoring:
- Computational efficiency suitable for frequent updates and large datasets.
- Closed-form parameter estimation improves applicability for real-time nowcasting and historical exercises.