Nowcasting Growth Using the Bayesian Structural Time Series Model: Application to Tanzania
IMF Working Papers, March 20, 2026
Source details
- Canonical URL
- Nowcasting Growth Using the Bayesian Structural Time Series Model: Application to Tanzania
Other formats
Bibliographic details
- Authors: Sunwoo Lee
- Published: March 20, 2026
- Series: IMF Working Papers
- DOI: https://doi.org/10.5089/9798229042116.001
Overview
- Title: Nowcasting Growth Using the Bayesian Structural Time Series Model: Application to Tanzania
- Author: Sunwoo Lee
- Date: March 20, 2026
- Series: Working Paper No. 2026/049
- Issue: 049
- Volume: 2026
- Pages: 27
- DOI: https://doi.org/10.5089/9798229042116.001
- Stock No: WPIEA2026049
- ISBN: 9798229042116
- ISSN: 1018-5941
- Format availability: PDF (Download PDF), More Formats on IMF eLibrary, Order a Print Copy, Create Citation
Research objective and context
- Purpose: Examine application of the Bayesian Structural Time Series (BSTS) model to nowcasting quarterly economic growth in Tanzania using high-frequency economic indicators.
- Motivation: Rising external volatility and recent global shocks increase need to effectively monitor short-term economic fluctuations, especially in countries with limited access to high-frequency growth data.
Methodology: Bayesian Structural Time Series (BSTS) framework
- Model components:
- Incorporates trends
- Incorporates seasonal variations
- Incorporates regression effects
- Variable selection:
- Uses spike-and-slab variable selection to identify relevant indicators
- Flexibility and extensions:
- Capacity to adapt to longer forecast horizons
- Supports dynamic regressors
Model selection, evaluation, and robustness
- Framework components:
- Outlines a framework for model selection and evaluation
- Includes robustness checks
- Includes sensitivity analysis
- Performance assessment:
- Demonstrates the model’s relative performance (comparative performance details are presented in the paper)
Findings and utility
- Nowcasting application:
- BSTS applied to nowcast quarterly economic growth for Tanzania leveraging a range of high-frequency indicators
- Benefits highlighted:
- Flexible framework for trend and seasonality decomposition
- Spike-and-slab aids in isolating informative predictors
- Adaptability to changing economic environments through dynamic regressors and longer horizons
- Intended use:
- Enhances capacity to monitor short-term growth trends in low-frequency-data environments
Subjects and keywords
- Subject: Agricultural sector, Econometric analysis, Economic and financial statistics, Economic forecasting, Economic sectors, Sensitivity analysis, Time series analysis
- Keywords: Africa, Agricultural sector, application of the Bayesian Structural Time Series, Bayesian models, Bayesian Structural Time Series model, economic activity, GDP, Global, IMF working papers, inclusion probability, low-income countries, Nowcasting, prediction error, Sensitivity analysis, Sub-Saharan Africa, Time series analysis
IMF Working Papers — Nowcasting Growth Using the Bayesian Structural Time Series Model: Application to Tanzania, Sunwoo Lee, March 20, 2026.
Content in this bundle
- Working Paper