Transcript Document

Forecasting models and procedures
Bank of Lithuania
April 23, 2008
Sofia
Rūta Rodzko
Bank of Lithuania
Economics Department
Forecasting models
• Short term:
– Inflation: monthly, disagregate (92 COICOP), naive,
12m ahead.
– GDP (production approach, monthly, quarterly
nowcast)
• Medium-term:
– MCM style “traditional” macroeconometric model.
2
Exogenous information
• Technical (oil price, exchange rate, short-term interest rate).
• Labour force: natural demographics and migration.
• Fiscal: revenue – effective rates, expenditure – (mainly) SCP, G
budget.
• Foreign demand and price variables: based on GDP and CPI
projections from Consensus Economics. TCI – more precise, but
no breakdown, availability does mot match the BoL needs.
• EU funds.
3
Timing
• Strategy changed since 2007.
• Forecast produced in parallel with the quarterly
review in Jan, Apr, Jul, Oct
• Timing and cut-off date for data influenced by
data availability (last day of the quarter).
4
Production process
Day 1: Data and assumptions
Day 4: 1st meeting (latest developments, changes in judgement)
Day 10: 2nd meeting (internal presentation, feedback)
Day 12: Final forecast
Day 15: Quarterly report submitted to the Board
Day 18-19: Forecast presented to the Board and published.
5
Publication
• Since 2008 published quarterly:
– “Short-list” of indicators complemented with a note.
– Risk assessment qualitative.
– Endorsed by the board.
6