About This Event
“Know where growth is today, not where it was two quarters ago.”
A five-day virtual programme on mixed-frequency techniques for estimating GDP in real time. You will combine monthly indicators such as credit, tax receipts, trade and mobile money volumes with quarterly national accounts, using bridge equations, MIDAS regressions and dynamic factor models to nowcast growth ahead of official releases.
What You'll Explore
A working GDP nowcasting model built on your own monthly indicators
Reusable R and Python code for bridge, MIDAS and dynamic factor models
A pseudo real-time evaluation comparing model accuracy for your economy
A news decomposition template to explain each nowcast revision
A nowcasting dashboard and briefing note format for policy meetings
Who Should Attend
Open to all qualifying staff, particularly: Research Economists, Forecasting & Modelling Staff, Real Sector Analysts, National Accounts Liaison Officers, Monetary Policy Analysts.
Why This Course Matters
Close the Data Gap
GDP arrives months after the quarter ends, yet rate decisions are made today. A nowcast fills that gap with evidence, so the committee is not setting policy on growth figures that are already out of date.
Use Every Indicator
Mixed-frequency models let monthly and weekly indicators speak alongside quarterly accounts. You extract signal from data you already collect, rather than waiting for one headline number each quarter.
Track the News
Nowcast news decomposition shows which data release moved your estimate and why. That makes it far easier to explain a revised growth view to the Governor in plain, defensible terms.
Programme
Day 1
Nowcasting foundations & building an indicator dataset
You will set out the nowcasting problem, select and transform high-frequency indicators for your economy, and handle ragged edges, publication lags and missing observations in a real-time dataset.
Day 2
Bridge equations & benchmark models
You will build bridge equations that link monthly indicators to quarterly GDP, and set up simple benchmark models so you can judge whether more complex approaches actually improve accuracy.
Day 3
MIDAS regressions & mixed-frequency estimation
You will estimate MIDAS and unrestricted MIDAS regressions in R or Python, choosing lag structures and weighting schemes that make the best use of monthly and weekly data without overfitting.
Day 4
Dynamic factor models & news decomposition
You will estimate a dynamic factor model with the Kalman filter, extract a common growth factor from many indicators and decompose each update to see which data release moved the nowcast.
Day 5
Real-time evaluation & a nowcasting dashboard
You will run pseudo real-time evaluation to compare models, combine forecasts, and assemble a nowcasting dashboard and briefing note ready for your forecasting round or policy committee.
Standards & Faculty Benchmark
Eurostat Handbook on Rapid Estimates
Eurostat's 2017 guidance on flash estimates and nowcasts, including mixed-frequency methods.
IMF Data Quality Assessment Framework (DQAF)
The IMF framework for assessing accuracy, reliability and timeliness of official statistics.
IMF Quarterly National Accounts Manual
IMF guidance on quarterly GDP compilation, the benchmark your nowcasts are ultimately judged against.
Nowcasts are only useful if they are built on sound data and judged against official GDP. Anchoring your methods to Eurostat and IMF guidance means your statistics office and external reviewers recognise the approach, and your estimates carry weight in policy discussion.
Is This Right for You?
- ☑You forecast or analyse growth for your policy committee
- ☑Your institution relies on GDP data released with long lags
- ☑You want practical models you can run every month in-house
Good to Know
Suited to economists comfortable with regression; R and Python templates are provided, so basic scripting is enough. Delivered live online with hands-on labs. You leave with a working nowcasting model on your own indicators and a template dashboard for policy briefings.
The Bottom Line
Walk away with a nowcast you can update every time new data lands, and explain to your Governor in a single chart.
Recommended For
Open to all qualifying staff, particularly: Research Economists, Forecasting & Modelling Staff, Real Sector Analysts, National Accounts Liaison Officers, Monetary Policy Analysts.