Economics
Economics Benchmarks 2024 report – making models work
The benchmarks this year include more granular details on key model design features
The past year has seen major reviews of central bank modelling frameworks, which have often found governance structures and model designs wanting. The benchmarks reveal that the most skilled staff spend around a quarter of their time on forecasting duties, though this varies widely across central banks. This year’s data also delves more deeply into the design of models, finding many of the more advanced features such as heterogeneous agents, non-linearities and wage-price spirals are relatively rare.
Charts
Key model elements
Former Fed chair Ben Bernanke urged the Bank of England to make several enhancements to its main forecast model in his 2024 review. Though the elements would add much realism to the model, benchmarking data shows that in most areas, his proposals are only in use at a handful of central banks.
To explore the data in more detail, use the benchmarking service’s interactive charts.
Model banks analysis – Economics 2024
Additional data breakdowns reveal patterns based on staffing and development
Economics data
View the full breakdown of responses to the Economics Benchmarks.
Previous years
Economics 2024
Economics 2023
Economics 2022
Economics 2021
Economics 2020
Economics Benchmarks 2024 – model banks analysis
Additional data breakdowns reveal patterns based on staffing and development
Economics Benchmarks 2024 – executive summary
The benchmark drills into model design features as well as staffing, research and AI
Economics Benchmarks 2024 report – making models work
The benchmarks this year include more granular details on key model design features
Share of economics departments with data unit rises
Web scraping and administrative data are top alternative sources for economists
Heterogeneous agents and non-linearities rare in main models
Semi-structural models continue to be dominant choice of main forecast model at central banks
Economists commonly use AI for data analysis
Lower-middle income countries, Africa and Asia-Pacific report lower use of AI
Economists least likely to research impact of AI
Inflation dynamics retains position at top of the rankings
Peer-reviewed journals still top gauge for research impact
Smaller teams more likely to consider conference presentations as assessment tool
Inflation and GDP growth forecast errors decrease further
Forecast errors used widely for key model adjustments
One in five main models has endogenous financial sector
Benchmark highlights array of methods central banks use to model financial sector
Central banks published fewer working papers and articles in 2023
Middle income and low income central banks published less
Time series and semi-structural models widely applied
High income countries more likely to use AI than peers from other income levels