Real-Time Economic Indicators Help Better Track Activity in Africa
IMF Blog, May 27, 2022
Source details
- Canonical URL
- Real-Time Economic Indicators Help Better Track Activity in Africa
Other formats
Bibliographic details
- Authors: Seung Mo Choi, Tara Iyer
- Published: May 27, 2022
Overview
- Publication date: May 27, 2022
- Topic: Use of nowcasting and high-frequency data paired with machine learning to provide real-time economic-growth indicators for developing economies, especially in sub-Saharan Africa.
- Motivation: Many developing economies release key indicators such as gross domestic product with lengthy delays, complicating policymaking during rapid changes (for example, the onset of the pandemic or outbreak of war in Ukraine).
Nowcasting approach and data inputs
- Method: Pair high-frequency data with machine learning algorithms to generate economic-growth nowcasts that extract signals from indicators available before official GDP data are released.
- Types of high-frequency and nontraditional inputs noted:
- Tourist arrivals
- Crude oil prices (noted relevance for oil-exporting countries such as Nigeria)
- Satellite imagery of nighttime lights
- Satellite imagery of shipping vessels (to track trade volumes and disruptions)
- Web searches (example: Google search volume for a country’s name as an indicator of future tourist visits)
- Financial and macroeconomic series (examples below)
Case study — Botswana
- Nowcasting performance during pandemic:
- Nowcasting framework projection for Q2 2020: projected output would contract by about 20 percent from a year earlier.
- Official published GDP (September 2020): showed a 24 percent contraction from a year earlier.
- Interpretation: Nowcasts moved in step with actual data and accurately predicted the turning point.
- Predictors used for Botswana in the IMF work:
- Currency values
- Stock prices
- Inflation
- Imports
- Consumer lending
- Electricity generation
- Tax revenue
- Google search volume for the country’s name
- Diamond prices (because Botswana is one of the world’s largest diamond exporters)
- Economic growth for two of the largest destination countries for their shipments
Machine learning versus traditional methods
- Finding: Machine-learning algorithms are often more accurate than traditional econometric methods, especially for predicting turning points such as when an economy begins to bounce back after a recession or crisis.
- Supporting evidence: Prior IMF working paper (2020) showed web searches can help more accurately forecast tourist arrivals.
Capacity building and policy implications
- IMF contributions:
- Developed a nowcasting framework to narrow the gap between data availability and policymaking in sub-Saharan Africa.
- Held a workshop with the Bank of Botswana; the Bank of Botswana is now producing its own nowcasts after participation.
- Plans to offer related courses on developing high-frequency indicators to help authorities worldwide better track economic activity.
- Policy implication: Nowcasting can offer projections that approximate economic activity a lot more quickly than official GDP data, providing invaluable input for policy decisions during fast-moving events.
Source: Real-Time Economic Indicators Help Better Track Activity in Africa, May 27, 2022
