The forecasting and modelling process at the

1
The forecasting and modelling process at the
National Bank of Romania
Anca Galatescu Head of Macroeconomic Assessement
Models Division Monetary Policy and Macroeconomic
Modelling Department
BANCA NATIONALA A ROMÂNIEI
2
Outline

• The forecasting process
• The model
• Further developments

3
The forecasting process

• Stages
• Near term forecasts of key variables
• Assessment of current position of the economy
  over the business cycle
• Medium-term projections using the MAMTF (Model
  for Analysis and Medium-Term Forecasting)

4
Flow of information in the forecasting process at
the National Bank of Romania
Near-term forecast
Near-term models and expert forecast
Final medium-term forecast and risk scenarios
NTF Inflation, GDP, ex. rate etc.
Medium-term (core) model
Assessment of initial conditions and medium-term
trends
Trends Gaps
Anticipated shocks, fiscal impulse, etc.
Tunes
Exogenous variables forecasts
Uncertainty
Expert judgment
5
Quarterly Forecasting Decisions Schedule
Time Event Details
T-45 Task Force meeting Model meeting (recalibration and other developments) - biannually
T-35 Task Force meeting Near-Term Forecasts NTF (inflation, GDP, interest rates, exchange rate) and exogenous scenarios
T-28 Task Force meeting Initial conditions (GDP, interest rates, exchange rate gaps), medium run equilibrium conditions and exogenous variables
T-22 MPC meeting Initial conditions, equilibrium conditions, exogenous variables
T-16 Task Force meeting Final projection meeting
T-10 MPC meeting Forecasts and risk scenarios
T Inflation Report Forecasts and Inflation Report sent to MPC
T1 MPC meeting Discussion of Inflation Report
T7 Bank Board meeting Discussion of Inflation Report and monetary policy decision
Task Force set up to implement IT framework
consists of experts from Monetary Policy and
Macroeconomic Modelling Department and Research
and Publications Department
6
The forecasting process

• Characteristics
• Based on formalized models and expert judgment
• Two types of modeling approaches
• Estimation approach at the short-run horizon
• Calibration approach at the medium-term
  horizon
• Final forecast integrates information from
  short-term models, medium-term model and expert
  judgment

7
The forecasting process

• Role of near-term forecasting
• Cover short end of forecast horizon
• Input for the initial conditions of the forecast
• Role of expert judgment
• Flexibility of the NBR medium-run forecasting
  model allows direct incorporation of expert input
• Forecasts of effects of anticipated exogenous
  events (e.g. change in excise duties)
• Forecasts of variables not explicitly modeled
  (e.g. fiscal impulse)
• Model forecasts can be tuned if unrealistic,
  using idiosyncratic judgments for each projection
  round

8
The forecasting process

• Role of medium-term model
• Shapes the initial conditions of the forecast
  rounds
• Integrates all information in a consistent way
• Generates an interest rate path which can serve
  as policy guideline, together with projections
  for all relevant macroeconomic variables
• Can be used to implement risk scenarios

9
Near-Term Forecasting

• Two-quarter horizon forecasts for key variables
• ARMAX model for core inflation and ECM for GDP
  components expert judgment incorporated
• Economic theory as a basis of analysis, but
  emphasis on forecasting accuracy
• Used for analysis and for establishing the
  initial conditions for the QPM

10
Medium-Term Forecasting Framework

• 1. History of the Model
• Work on the NBRs model (MAMTF) started in
  mid-2004
• Significant progress achieved, with technical
  assistance support from several IMF missions and
  bilateral exchanges/expert visits with the Czech
  National Bank (MAMTF conceived in similar fashion
  to the CNBs QPM)

11
Medium-Term Forecasting Framework

• 2. General characteristics of MAMTF
• Small semi-structural calibrated model with a
  New-Keynesian core (ST and MT non-neutrality)
• Consistent with achieving multi-period inflation
  targets
• Economy assumed to converge to well-defined and
  stable long-run equilibrium
• Deviations from trends reflect cyclical behavior
  of the economy, paramount for this type of model
• Model open to continuous improvement, while
  maintaining the core structure in the near
  future, expected to be gradually replaced by a
  dynamic general equilibrium model

12

Medium-Term Forecasting Framework

• 2. General characteristics of MAMTF
• Use of satellite models for
• - GDP components forecasting the forecasts for
  other relevant variables (inflation, exchange
  rate, economic growth and so on) are exogenously
  imposed from the output of the MAMTF
• - fiscal impulse decompositions into cyclical
  and structural components

13
3. Transmission mechanism
NBRs monetary policy rate
Deposit interest rates
Lending interest rates
Consumption/ saving decisions
Consumption and investment borrowing
Foreign interest rate
Exchange rate (UIP)
Net exports channel
Wealth and balance sheet effect
Exchange rate pass-through
Import prices
Fiscal and income policies
Administered and volatile prices
Excess demand
CPI inflation
Balassa-Samuelson effect
CORE2 inflation
Expectations
14
3. Transmission mechanism
Interest rate channel - relatively slow impact
and limited efficiency - monetary policy
decisions transmitted through commercial banks
deposit and lending interest ratesExchange rate
channel - relatively quick through direct impact
on import prices (including fuel prices and
excise tax) indirect impact on aggregate demand
through net export channel Expectations
channel - quite significant reflects second
round effects of inflationary shocks Wealth and
balance sheet channel - important due to high
share of foreign currency loans
15
4. Model structure

• Inflation components
• Core inflation determined by its structural
  persistence, inflation expectations, output gap,
  import price inflation and Balassa-Samuelson
  effect
• Administered price inflation given by an
  exogenous scenario (discussions with the
  regulatory institutions on energy and natural gas
  prices)
• Fuel price inflation determined by its structural
  persistence, international oil price, exchange
  rate and inflation expectations
• Volatile prices inflation given by an exogenous
  scenario (seasonally pattern, exchange rate)

16
4. Model structure

• Output gap determined by its own persistence,
  real deposit and lending interest rates gaps,
  real exchange rate gap and a proxy for the wealth
  and balance sheet effect induced by the dynamics
  of the exchange rate
• Exchange rate determined according to uncovered
  interest parity relationship including a risk
  premium mixed backward and forward looking
  exchange rate expectations
• Monetary policy behavior described by a
  forward-looking policy interest rate rule that
  penalizes future deviations of inflation from the
  target, the output gap and excessive interest
  rate volatility
• Inflation expectations modeled as hybrids of
  backward-looking (inertial) and forward-looking
  (model-consistent) expectations

17
Further developments

• Implementing a DSGE model
• Advantages over the current model
• Fully structural
• Non linear
• Non-stationary steady state
• Theoretical structure derived specific features
  included
• Exchange rate appreciation in steady state
• Trends in relative prices across different
  sectors
• Administered prices are included as a component
  of the CPI index

18
Further developments

• Draft evaluation of the model including
• Calibration
• Filtering
• Forecasting
• Short term objective work with the current
  structure and provide shadow forecasts
• Medium term objective further development of the
  model, including
• Liquidity constrained agents
• Greater role for fiscal policy
• Adding a financial sector block

19

• Thank you for your attention!