Commentary
Reserves exposure to non-SDR sovereign debt increases
US Treasuries and gold remain the most prominent asset classes
Staff strength and salary associated with size of reserves
Managers report greater satisfaction with non-staff resources than staffing levels
Economics departments use multiple coding languages
More than half of central banks use five or more coding languages and tools
Geopolitics is fastest-rising research topic for economists
Research economists are most widely judged on their output of papers
Economists devote up to half of working hours to research
Higher-paid research economists tend to spend more time on own research
Dedicated data units widely used in economics departments
Teams make use of varied microdata in their work
Central banks predict growth more accurately than inflation
AI models may offer more precise forecasts
AI impact and usage growing in central bank economics departments
Regional variation exists, but adoption is increasing across the board
Most key projection models capture policy rates
But respondents rarely disclose policy rate forecasts alongside scenarios
Third of key models now employ endogenous financial sector
Multiple channels of monetary transmission rooted in majority models
Central banks complement traditional models with AI
Time series and semi-structural models remain economists’ choices for forecasting
Working paper production linked to number of PhD economists
Central banks averaged output of roughly 10 working papers last year
Third of economics departments say staffing ‘barely sufficient’
Forecasting, policy analysis and research are economists’ top duties
Economists’ average annual salary rises
Incomes have some relation with GDP per capita and central bank size
Economics staff strength increases over two-year period
Departmental size as share of central banks’ workforce widens to 7.7%, on average
Cross-border co-operation stronger for non-central bank supervisors
Roughly one-fifth of central banks lack supervisory college membership
Supervisors report capital and liquidity resilience across banks
Basel II remains most adopted global standard among respondents
Data quality is main focus for improved supervisory reporting
Supervisors with high level of automation rate submissions more highly, but quality still an issue
Data lakes and cloud are top tools for supervisory automation
Around half of authorities report extensive or full automation of supervisory data collection
Size and business model are key elements of risk profile
Supervisors tend to inspect systemic banks once a year
Supervisors report sharp increase in AI adoption
Summarisation ranks as topmost area of application, with biggest use in larger teams