## tsicdmpfpasea

## Source details

**Canonical URL:** [tsicdmpfpasea](https://www.imf.org/-/media/files/publications/dp/2021/english/tsicdmpfpasea.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/dp/2021/english/tsicdmpfpasea.pdf.md)
- [Structured JSON version](/-/media/files/publications/dp/2021/english/tsicdmpfpasea.pdf.json)

---

### Background and Purpose
- Definition and role of an FPAS:
  - An FPAS (forecasting and policy analysis system) is a system of tools and related processes designed to support forward-looking monetary policy formulation based on economic data and analysis.
  - Any central bank or economic policymaking organization that to some degree relies on forecasts de facto has an FPAS; it typically includes a formal economic model as part of a broader system.
  - Originated in the work of early inflation targeting (IT) central banks in the second half of the 1990s.
- Main elements of a well-structured FPAS:
  - Databases and procedures for collecting, processing, managing, and sharing data.
  - A core medium-term forecasting and policy analysis model with an explicit role for monetary policy.
  - Nowcasting and near-term forecasting (NTF) tools, satellite models, ex post evaluation tools, an FT (forecasting and policy analysis team), structured forecasting and policy process, and procedures for internal/external MPRs and presentations.
- IMF CD on FPAS:
  - In the last "three to five years" IMF CD activities supported building/enhancing FPASs in 25 countries (list supplied in source).
  - Most CD delivered via multiyear, externally funded projects with frequent (quarterly) missions and HQ- or RTAC-based staff plus short-term experts.
- FPAS complementarities:
  - FPAS CD often connects with CD on: reforming monetary policy framework; monetary operations and liquidity management; foreign exchange management; financial stability; financial markets development; central bank communication.
- Paper purpose, scope, limitations:
  - Take stock of FPAS CD, share insights, outline typical tools, offer qualitative assessment of CD delivery and FPAS use.
  - Does not evaluate actual country cases nor provide cross-country comparisons; assessment of tool appropriateness at individual institutions is out of scope.

### Designing the FPAS to Support Policy Decision-Making
- Purpose and principles:
  - FPAS supports systematic, forward-looking monetary policy; projections anchored on a reaction function for the policy rate to drive inflation toward medium-term objective reflecting policymakers’ preferences.
  - Structural macroeconomic models with endogenous monetary policy and forward-looking expectations are indispensable.
- Key policymaker questions FPAS must answer:
  - (1) Is a change in the policy stance needed and if so, why?
  - (2) How strong should the policy response be initially?
  - (3) What is the optimal policy path over the policy horizon, and how fast to bring inflation back to target?
  - (4) Implications for output and exchange rate?
  - (5) What is the "story"—drivers of near- and medium-term developments and communication strategy?
- Forecast horizon and transmission:
  - Inflation measured as an annual change—"a twelfth of that outcome is already determined with a one-year horizon."
  - Eight or more quarters generally needed to keep medium-term policy horizon in view.
- Role of tools and judgment:
  - NTF/nowcasting can outperform structural models one or two quarters ahead; expert judgment is essential and must be disciplined and documented.
  - Final forecasts combine expert judgment, NTF tools, and structural model-based projections.
- Forecast usefulness and evaluation:
  - Forecast accuracy alone is a poor metric; usefulness depends on signaling policy adjustment needs and aiding communication and trust.
  - Toolkit for analyzing forecast revisions and ex post forecast errors essential; evaluate sources of errors separately and document expert "tunes".
- Organizational/process design:
  - Specify core model, NTF framework and role, organization of forecasting and decision-making, and inputs required to achieve staff–policymaker consensus (current situation, exogenous assumptions, drivers, pace/magnitude of policy adjustment, policy stance).

### Developing FPAS Capacity: Scope and Modalities for CD
- Three basic components of FPAS capacity:
  - (1) Reliable analytical tools.
  - (2) Staff trained to use tools and apply expert judgment.
  - (3) Policy framework/process that systematically uses outputs to provide policymakers alternative policy choices.
- CD activities:
  - Staffing and organization; data management; external sector data; forecasting calendars and process; coordination with economic agencies; internal/external communication materials; staff training on models/software and policy tradeoffs.
  - CD methods: mock forecasting rounds, ex post evaluation tools, outreach to management, templates for MPRs and communications.
- Sustainability, timelines, modalities:
  - Sustainability requires staff understanding and management reliance on tools.
  - Establishing a sustainable FPAS typically requires sustained intensive CD over multiple years.
  - Typical first phase for low- or lower-middle-income country: planned for two years, with three to four missions per year. Missions must be frequent to keep momentum; follow-up missions may be contingent on milestones.
  - Projects often include inter-mission tasks and online interactions; IMF classroom-style and online courses complemented CD (including an IMF online course on Monetary Policy Analysis and Forecasting).

### Forecasting and Policy Analysis Toolkit — Sectoral, NTF, and Core QPM
- Sectoral analysis, nowcasting, and NTF:
  - Purpose: extract information from high-frequency data; combine quantitative tools with sectoral expert judgment.
  - Best practice: seasonal adjustment, univariate filtering, suite of models (not single model), lower-/mixed-frequency tools.
  - Typical models: ARIMA, principal components, VAR, VEC, Bayesian VAR/VEC, factor VAR, MIDAS, bridge equations.
  - Diagnostics: analyze every new observation against most recent forecast; in data-poor environments build indicators and process qualitative data.
  - Integration to medium-term models: impose nowcasts/NTF one-to-one, weighted, or via two-stage conditioning.
- Core medium-term forecasting and policy analysis model (QPM):
  - Rationale: New Keynesian semistructural quarterly gap model (QPM) commonly used—simple, forward-looking, adaptable to country specifics; can later be supplemented or replaced by DSGE.
  - Tradeoffs: DSGE may be time-consuming and risk becoming a "black-box".
  - Typical augmentations: fiscal demand impact, TOT effects, multiple Phillips curves, money block for money-targeting central banks, exchange rate persistence, market rate deviations, credibility effects.
  - Core behavioral equations (textbook forms preserved exactly as in source):
    - Aggregate demand / open-economy IS curve (output gap ỹt): ỹt = a1 ỹt − 1 + a2 Et ỹt + 1 − a3 mci t + a4 ỹt* + εt ỹ  (equation (1))
    - Monetary conditions index (mci t): mci t = aac1 r̂t + (1−aac1)(−ẑt + 1)  (equation (2))
    - Forward-looking Phillips curve (inflation πt): πt = b1 πt − 1 + (1−b1) Et πt + 1 + b3 rmc t + εt π  (equation (3))
    - Real marginal cost approximation: rmc t = bbb1 ỹt + (1−bbb1) ẑt  (equation (4))
    - UIP exchange rate condition (st): st = Et st + 1 + (it* − it + pr_prem t)/4 + εt s  (equation (5))
    - Monetary policy rule (it): it = m1 it − 1 + (1−m1)(in t + m2(Et πt +4/4 − πT t) + m3 ỹt) + εt i  (equation (6))
    - Trend dynamics for equilibrium variables: Δx̅t = hx Δx̅t − 1 + (1−hx) Δx̅ss + εt Δx̅  (equation (7))
  - Measurement notes: All variables in 100*natural logs, except interest rates and rates of change (annualized percentage points). Output gap ỹt = yt − y̅t.

### Incorporating Fiscal Policy, TOT, Supply Shocks, Money, FX, and Credibility
- Fiscal policy:
  - Add fiscal impulse fimp t to IS: ỹt = a1 ỹt − 1 + a2 Et ỹt + 1 − a3 rmcit + a4 ỹt* + a5 fimp t + εt ỹ  (equation (1b))
  - Fiscal impulse specified via decomposition of deficit ratio or deviation of real fiscal expenditures from trend; full flow-stock fiscal blocks not implemented in country projects to date.
- Terms of Trade (TOT):
  - Decompose tot t = tot̅t + tot̃t and augment IS with tot gap a5 tot̃t (equation (1c)); caution on Lucas critique and parameter stability.
- Domestic supply shocks and multiple Phillips curves:
  - Use GDP excluding sectors (agriculture, oil, mining) or model separate agricultural block with equations (1d) and (1e); split CPI into components and model component-specific Phillips curves (equations (18a)/(18b), (19a)/(19b), (3b)-(3d), etc.).
- Money and money targeting:
  - Money growth and real money gap defined (equations (23)–(25)); use as forecasting tool or intermediate target; practical challenges due to unstable velocity and opaque operations.
- Exchange rate persistence, limited arbitrage, interventions, and pegs:
  - Model expected spot rate as weighted average of model-consistent and naïve forecasts (equation (26)); add endogenous risk premium (equation (28)); model BOP-determined exchange rate when limited arbitrage (equations (29)–(32)); intervention and peg frameworks specified (equations (33a)/(33b), (5d)-(5f), (37)–(40)).
- Market interest rate spreads and term structure:
  - Model persistent spreads explicitly: it = ig t + spreadt (equation (41)); spreadt autoregressive (equation (42)); represent market rates as weighted averages where markets segmented (equations (43)–(45)). Term premiums modeled via AR processes or credibility proxies (equations (46)–(48)).
- Central bank credibility and expectations:
  - Credibility stock and altered expectations modeled (equations (49)–(50)); credibility anchors expectations and should be incorporated where relevant.

### Calibration and Parameterization
- Parameters typically calibrated (not estimated) in practice for emerging markets due to data limitations and structural breaks.
- Methods to check parametrization: empirical fit, in-sample forecasting, impulse responses, Kalman filter historical decomposition, Bayesian estimation as double-check.
- Policy rule parameters represent policymakers’ preferences; implications of alternative rules should be quantified and presented. Parameterization should be reconsidered rarely.

### Forecasting Process and Incorporation into Decision-Making
- Structure and timing:
  - Core objective: ensure FPAS output has traction in policymaking via structured forecasting rounds, briefings, and real-time support.
  - Typical full forecasting round duration: five to seven weeks from start of technical work to final MPC meeting.
  - Standard recommendation: undertake full forecasting rounds quarterly (limit to four per year); align start with quarterly national accounts release.
  - Risks of more frequent rounds: overreaction to high-frequency noise, frequent reversals, noisy policy signals, high interest-rate volatility.
- Interim meetings and sequencing:
  - Interim policy meetings recommended; suggested maximum of eight policy meetings a year.
  - Preannounced published meeting calendar and MPR calendar recommended.
  - Role of forecast coordinator to schedule tasks and enable management planning.
- Pre-MPC technical meetings:
  - Recommended three pre-MPC technical meetings: initial conditions; first version of baseline and alternative scenarios; final forecasts, policy analysis and recommendations.
  - Purpose: inform policymakers, reconcile differing views, incorporate feedback, and build consensus.
  - Risks: fully staff-dictated or fully policymaker-dictated processes can both be harmful.
- Additional interactions and capacity building:
  - Seminars, workshops, startup meetings, and TA-supported seminars for management to solidify political support.
- Staffing and organization of the FT:
  - Common shortcomings: under-resourcing, FPAS treated as research only, insufficient participation from broad staff.
  - Recommended team sizes:
    - Core modeling group: recommended size at least six or seven people (with rotation: three or four in quarterly rounds).
    - Real sector team: six or seven people.
    - External sector team: two or three people (up to six or seven if producing own forecasts).
    - Fiscal sector team: three people (six or seven if producing own fiscal analysis).
    - Financial sector team: one or two people.
    - Drafting/external communication team: assigned for MPC documents and public communication.
  - Forecast coordinator: relatively senior staff with technical and communication skills and clear mandate.

### Assessment, Determinants of Impact, and Lessons
- Determinants of FPAS CD impact:
  - Maturity of monetary policy reform plans and top commitment; staff technical skills at project start; staff/management turnover; duration of CD and reform efforts.
- Observed outcomes:
  - Most projects progressed in CD and technical work (QPM, NTF tools).
  - Incorporation of model-based analysis into policymaking and impact on actual decisions were mixed.
- Countries with strong transformational impact:
  - Most adopted or moved toward IT, IT-Lite, or IT-like frameworks.
  - Projects strengthened modeling/forecasting, internal organization, policy processes, communication, and led to regular MPC meetings, forward-looking decisions, and published quarterly MPRs.
  - Enabling conditions: strong staff skills, management commitment, political backing; transitions ranged from a few years to multi-year processes.
- Countries with more muted impact:
  - Typical issues: weaker analytical capacity, erratic implementation, persistent market–policy rate deviations, limited forward-looking publication, partial management support, weak political backing.
  - Progress: more transparent and analytical internal discussion, improved policy dialogue with IMF, occasional alignment of interbank rates with policy rate.
  - Remaining weaknesses: heavy exchange rate management, incoherent internal analysis, frequent staff rotation, weak HR practices, political considerations dominating policymaking.
  - Uncertainty whether capacity will be sustained.
- Lost capacity instances:
  - Causes: reorganizations, staff relocations, resistance from middle management, gaps in CD engagement, insufficient incorporation into decision-making.
- Lessons for design, implementation, sustainability:
  - Detailed pre-project assessment, realistic objectives, scoping missions, and logframes with measurable indicators and milestones are essential.
  - FPAS projects require multiyear perspective; three to four missions a year over several years often needed initially.
  - Continuity of IMF experts and CD engagement important for hands-on follow-up and trust.
  - Integration with broader monetary policy modernization (operations, FX policy, financial markets) increases success likelihood.
  - FPAS CD is not mainly about modeling; staff skills, processes, data infrastructure, and communication are critical.
  - Keep models simple and combine with expert judgment, sectoral analysis, satellite models, and filters.
  - Organizational design: integrate FPAS into decision-making early, rethink HR, document procedures, formalize forecast coordinator role, provide backups and rotation in teams.
  - Incentives and documentation: reflect FPAS duties in terms of reference/KPIs and prepare documentation/manuals by CD recipient to ensure ownership.

### Forecast Calendar (Annex 1) — Purpose and Operational Guidance
- Purpose and components:
  - FC frames the forecasting process, ensures planning, and allows monitoring; should be available to staff and approved by management.
  - FC should include release dates for key data, national holidays, departmental and policymaker meeting timings, and precise deadlines (day and hour) with assigned responsibilities.
- Example timing and operational guidance (illustrative):
  - Example assumptions: QNA for Q1 released afternoon Friday June 28; CPI released 12 working days after month end; final MPC meeting and forecast release five working days after month-end.
  - Week-by-week timetable for a typical six-week forecasting round (Weeks 0–6) with detailed tasks for NTF, initial conditions, first and final forecasts, drafting MPR, MPC meeting, and postmortem.
  - Operational recommendations:
    - Prepare and communicate FC well ahead (preferably 12 months).
    - Determine MPC meeting timing for next 12 months and make public.
    - Schedule technique-change meeting before new QNA release.
    - Provide approximately one week for technical discussions on initial conditions and 1.5 to 2 days for MPR submission prior to MPC meeting.
    - Finalize MPR and communication materials in time for MPC review and to minimize leaks.
- Prototype Forecast Calendar example (24-VI. – 9-VIII. 2019):
  - Weekly schedule highlights and specific dated milestones, including multiple rounds of MPR inputs (1st–5th rounds) and key meetings: MPC initial conditions (15-VII.), first-version MPC (22-VII.), MPR to MPC (2-VIII.), MPC policy decision (7-VIII.), Post Mortem (9-VIII.).
  - High-frequency data update dates and abbreviations preserved as in source (CPI, MPC, MPR, NTF).

*IMF Departmental Paper — Taking Stock of IMF CD on Monetary Policy Forecasting and Policy Analysis Systems (tsicdmpfpasea)*

### 1. Background and Purpose ..............................................................................................

### 1. Background and Purpose

### Definition and role of an FPAS
- An FPAS (forecasting and policy analysis system) is a system of tools and related processes designed to support forward-looking monetary policy formulation based on economic data and analysis.
- Any central bank or economic policymaking organization that to some degree relies on forecasts for its decisions de facto has an FPAS, though components and structure differ.
- The FPAS originated in the work of early inflation targeting (IT) central banks in the second half of the 1990s to provide a more rigorous analytical foundation for interest rate settings.
- The FPAS is a consistent, well-organized framework for collecting, processing, and analyzing economic information with emphasis on providing analysis and policy recommendations to policymakers to support monetary policy decisions.
- The FPAS is not a single “model” but typically involves a formal economic model as part of a broader system.

### Main elements of a well-structured FPAS
- Databases and procedures for collecting, processing, managing, and sharing data.
- A core medium-term (forecasting and) monetary policy analysis model that has an explicit role for monetary policy.
- Nowcasting and near-term forecasting (NTF) tools and procedures.
- Satellite models and analytical tools for sectoral analysis, forecasting specific aspects/sectors, or analyzing specific phenomena.
- Tools and procedures for ex post evaluation of the forecasts and policy recommendations.
- A forecasting and policy analysis team (FT) with well-defined managerial and operational responsibilities and a sound reporting structure; the full team usually consists of the head of the team, a core modeling group, and sectoral experts.
- A structured forecasting and policy analysis process with well-defined deadlines and responsibilities, including staff-level technical meetings, technical meetings between the FT and policymakers, and monetary policy decision-making meetings.
- Processes for preparing internal and external monetary policy reports (MPRs), and structured monetary policy advice and presentations to policymakers.

### IMF capacity development (CD) on FPAS
- Developing FPASs has been a central focus of the IMF’s monetary policy technical assistance and customized training.
- In the last "three to five years" IMF CD activities have supported building or enhancing FPASs in 25 countries.
- FPAS CD recipients include: Albania, Angola, Armenia, Belarus, Botswana, Chile, Ghana, Honduras, Jamaica, Kazakhstan, Kenya, Kyrgyz Republic, Malawi, Mauritius, Moldova, Morocco, Mozambique, Russia, Rwanda, Seychelles, Sri Lanka, Tanzania, Uganda, Ukraine, Vietnam, and Zambia, as well as the East African Community.
- Most CD has been delivered through multiyear, externally funded projects with frequent (quarterly) missions using headquarters (HQ)-based or regional technical assistance center (RTAC)-based staff, supplemented by a pool of short-term experts.
- Many projects aimed at supporting a transition toward an IT-like policy framework; others aimed at enriching policy decision-making with forward-looking economic analysis as part of broader policy modernization.

### FPAS as a bridge and complementarities
- The FPAS bridges the broad objectives of the monetary policy regime and the monetary policy operations, mapping medium-term goals into high-frequency actions such as open market operations and setting short-term policy interest rates.
- FPAS CD often connects with and supports complementary IMF CD on:
  - Reforming the monetary policy framework,
  - Enhancing monetary operations and liquidity management,
  - Improving foreign exchange management practices,
  - Enhancing financial stability,
  - Financial markets development,
  - Designing effective central bank communication.
- Simultaneous capacity building in these areas generates positive externalities and more robust policy frameworks.

### Purpose, scope, and limitations of the paper
- The paper takes stock of FPAS CD, drawing on experience and lessons from developing FPAS capacity in central banks.
- Objectives:
  - Share insights gained during FPAS CD delivery.
  - Outline typical tools developed in the process.
  - Facilitate understanding of FPAS CD within the IMF and inform future CD on building macroeconomic frameworks.
  - Offer a qualitative assessment of FPAS CD delivery and FPAS use in decision-making in central banks.
- Limitations:
  - The paper does not discuss or evaluate actual country cases nor provide cross-country comparisons of FPAS CD projects.
  - Assessment of appropriateness of resulting FPAS tools and procedures at individual institutions is beyond the scope of the paper.
  - Relevant assessments are often available in joint working papers produced by CD teams and the authorities.

*tsicdmpfpasea - 1. Background and Purpose*

### 2. The FPAS: Designing It and Developing It

### 2. The FPAS: Designing It and Developing It

### A. Designing the FPAS to Support Policy Decision-Making
- Purpose and overarching design principles
  - The main reason for building an FPAS is to support making monetary policy decisions in a systematic, forward-looking fashion, informed by economic data and analysis.
  - FPAS emphasizes projections anchored on a reaction function for the policy rate designed to drive inflation toward its medium-term objective at a pace that reflects policymakers’ preferences.
  - Structural macroeconomic models with an explicit role for endogenous monetary policy and forward-looking (model-consistent) expectations are indispensable for forward-looking policy decision-making and for analyzing risks to the policy scenario.

- Key policymaker questions the FPAS must help answer
  - (1) Is a change in the policy stance needed and if so, why?
  - (2) How strong should the policy response be initially?
  - (3) What is the optimal policy path over the policy horizon, and how fast one should aim at bringing inflation back to target?
  - (4) What are the implications of the policy response for other variables such as output and the exchange rate?
  - (5) What is the “story”? That is, what are the drivers of the projected near- and medium-term development, and how to explain to the public the reason for a change in the policy stance, or the lack thereof?

- Forecast horizon and transmission considerations
  - Inflation is measured (and targeted) as an annual change—a twelfth of that outcome is already determined with a one-year horizon. Eight or more quarters are generally needed in order to not lose sight of the medium-term policy horizon.
  - Monetary policy actions affect the economy, including inflation, with long and variable lags; setting the policy stance must be based on sound forward-looking economic analysis.
  - The FPAS should align decision-making with: (1) overall strategic objectives of monetary policy; (2) analytic understanding of the economy and monetary policy transmission; and (3) real-time analysis of economic data.

- Role of different tools and judgment
  - Near-term forecasting (NTF) and nowcasting: econometric time series models and sectoral specialists may outperform structural models one or two quarters ahead; NTF’s role is twofold: inform assessment of current conditions and improve near-term forecast quality.
  - Core medium-term forecasting and policy analysis model: provides forward-looking structural analysis and allows consistent scenario construction and policy-response simulation.
  - Expert judgment is essential and must be disciplined—based on deep sectoral knowledge, documented “tunes,” and iterations between formal tools and intuition.
  - The final forecast and associated policy recommendation combine: (1) expert judgment supported by analytical tools (including near-term models); and (2) forward-looking structural model-based forecasting and policy analysis complemented with sectoral analysis and judgment.

- Forecast usefulness and evaluation
  - Forecast accuracy in isolation is a poor metric for usefulness; forecasts can be wrong for the right reason and right for the wrong reason.
  - Regular forecasting rounds aid policymaking if they:
    - (1) correctly signal the need for a policy adjustment and its direction most of the time and do that more reliably than any available alternative;
    - (2) help communicate the reasons for changing the policy stance both to decision makers and to the public; and
    - (3) ensure policymakers trust the insight provided by the FT and base their policy actions on it.
  - A toolkit for analyzing forecast revisions between forecasting rounds and ex post forecast errors is essential to facilitate learning, accountability, and building trust. Forecast evaluation should separately assess sources of errors: initial conditions, other macro policies, external outlook, model structure and parameters, and expert-judgment tunes. Thorough documentation of expert tunes is required to run ex post model simulations with and without those tunes.

- Organizational and process design elements
  - The FPAS design should specify:
    - The design and role of the core medium-term forecasting and policy analysis model.
    - The design of the NTF framework and its role (inform current conditions; impose NTF-based forecasts for first one or two quarters).
    - The organization and structuring of forecasting, policy analysis, and decision-making processes to ensure quality, consistency, and that policymakers understand, trust, and rely on the analysis.
  - Forecasting process inputs that should achieve internal consensus between staff and policymakers:
    - The assessment of the current economic situation;
    - The external and other exogenous assumptions underpinning the forecast;
    - The fundamental drivers of likely medium-term developments;
    - The pace and magnitude of required policy adjustment reflecting policymakers’ preferences and concerns not captured by the model; and
    - The policy stance underpinning the published forecasts (accurately reflected in implemented policy).

### B. Developing FPAS Capacity: Scope of CD Activities
- FPAS capacity is more than model building: three basic components required
  - (1) A set of reliable and well-designed analytical tools for economic and policy analysis;
  - (2) Staff trained in how to use such tools (including the application of expert judgment); and
  - (3) A policy framework and decision-making process in which the output is used systematically to provide policymakers with alternative policy choices.
  - FPAS CD (capacity development) revolves around these three functions.

- CD activities and institutional advice
  - CD projects focus on: staffing and organization; data storage and data management; collecting data on the external sector; forecasting calendars and structuring the forecast process (including staff forecasting meetings and policy meetings such as the monetary policy committee (MPC) or executive board); coordination with other economic agencies; internal communication and briefings; and external communication (MPRs, press releases, press conferences).
  - CD has assisted staff development by deepening understanding of macroeconomics and monetary policymaking; time-series and structural models; software (Excel, EViews, Matlab, analytic toolboxes); and policy tradeoffs.
  - CD support has included: running mock forecasting and policy rounds; implementing tools and procedures for systematic evaluation of model-based forecast performance; outreach and presentations to central bank management and policymaking bodies; and designing templates for MPRs, press releases, and press conference presentations.

- Sustainability, timelines, and modalities for CD
  - To achieve sustainability, staff must understand how to use and maintain tools, and management must find the tools useful and regularly rely on analytic inputs.
  - Establishing a sustainable and functional FPAS typically requires sustained and intensive CD over multiple years.
  - A typical first phase for a low- or lower-middle-income country with limited analytic background is typically planned for two years, with three to four missions per year. Missions must be frequent to keep momentum, while allowing time between missions for staff to implement recommendations. Follow-up missions may be contingent on authorities’ progress in achieving organizational milestones.
  - Projects often include specific tasks and training exercises to be undertaken by central bank staff between missions and provide for online interactions with CD experts. IMF classroom-style and online courses complemented CD activities, including the recently released IMF online course on Monetary Policy Analysis and Forecasting.

*Source: 2. The FPAS: Designing It and Developing It*

### 3. The Forecasting and Policy Analysis

### 3. The Forecasting and Policy Analysis Toolkit

### A. Sectoral Analysis, Nowcasting, and Near-Term Forecasting Tools
- Purpose and role:
  - Serve as a judgment support system extracting all relevant information from available data and assessing the current stance of the economy.
  - Emphasize filtering signal from high-frequency data and combining quantitative tools with sectoral expert judgment.
- Standard tools and best practice:
  - Use seasonal adjustment (decomposing into seasonal, irregular, and trend-cycle components) and simple univariate filtering to separate trend from cycle.
  - Use a suite of models and tools rather than a single model for NTF; complement nowcasting with analytic tools based on lower-frequency or mixed-frequency data.
- Typical models and techniques supported by CD:
  - Single-equation autoregressive integrated moving average, principal component, vector autoregressive, vector error correction, Bayesian vector autoregressive, Bayesian vector error correction, factor vector autoregressive, mixed-data sampling, bridge equation.
- Data and diagnostics:
  - Every new observation is analyzed against the most recent forecast; forecast errors are calculated and analyzed to improve methods and separate signal from noise.
  - In data-poor environments, build coincident and leading indicators and process qualitative data (consumer and business sentiment).
- Cautions and lessons:
  - Risk of overemphasis on technically sophisticated tools at the expense of sectoral expertise; models used mechanically without judgment underperform.
  - Basic craftsmanship (reading time series, standard decompositions, X-13 ARIMA-SEATS) is often underutilized.
- Integration with medium-term models:
  - Nowcasts and one- to two-quarter-ahead forecasts can be imposed on the medium-term core projection model via:
    - One-to-one conditioning (single shock to reproduce NTF value).
    - Weighted conditioning (multiple shocks with fixed proportion).
    - Two-stage conditioning (treat NTF as extra observed data in filter, then one-to-one conditioning).

### B. The Core Medium-Term Forecasting and Policy Analysis Model (QPM)
- Rationale for semistructural gap models:
  - Most central banks start with a simple New Keynesian semistructural quarterly gap model (QPM); some later supplement or replace with DSGE.
  - Benefits of QPM: relatively simple to build and operate, forward-looking, malleable to country specifics, facilitates expert judgment and storytelling, embodies monetary neutrality and New Keynesian nominal/real rigidities.
- Tradeoffs versus DSGE:
  - Fully micro-founded DSGE models can take substantial time and may not yield better forecast performance; risk of becoming a “black-box”.
- Typical augmentations to the textbook QPM:
  - Incorporate demand impact of fiscal policy, terms of trade (TOT) effects, multiple Phillips curves for supply shocks/administrated prices, money block for money-targeting central banks, exchange rate persistence and managed/pegged regimes, persistent deviations of market rates from policy rate, central bank credibility effects.
- Core behavioral equations (textbook outlines):
  - Aggregate demand / open-economy IS curve (output gap ỹt): ỹt = a1 ỹt − 1 + a2 Et ỹt + 1 − a3 mci t + a4 ỹt* + εt ỹ  (equation (1))
  - Monetary conditions index (mci t): mci t = aac1 r̂t + (1−aac1)(−ẑt + 1)  (equation (2))
  - Forward-looking Phillips curve (inflation πt): πt = b1 πt − 1 + (1−b1) Et πt + 1 + b3 rmc t + εt π  (equation (3))
  - Real marginal cost approximation: rmc t = bbb1 ỹt + (1−bbb1) ẑt  (equation (4))
  - UIP exchange rate condition (st): st = Et st + 1 + (it* − it + pr_prem t)/4 + εt s  (equation (5))
  - Monetary policy rule (it): it = m1 it − 1 + (1−m1)(in t + m2(Et πt +4/4 − πT t) + m3 ỹt) + εt i  (equation (6))
  - Trend dynamics for equilibrium variables: Δx̅t = hx Δx̅t − 1 + (1−hx) Δx̅ss + εt Δx̅  (equation (7))
- Measurement notes:
  - All variables in 100*natural logs, except interest rates and rates of change (annualized percentage points).
  - Output gap defined as deviation of log real output from potential (ỹt = yt − y̅t).

### C. Incorporating Fiscal Policy
- Main approach:
  - Add a fiscal impulse variable to the IS curve: ỹt = a1 ỹt − 1 + a2 Et ỹt + 1 − a3 rmcit + a4 ỹt* + a5 fimp t + εt ỹ  (equation (1b))
- Fiscal impulse specifications:
  - Decompose fiscal deficit ratio into cyclical, long-term structural, and idiosyncratic (discrete) components; fiscal impulse defined as discrete component or change in structural+discrete (equations (8)–(10)).
  - Alternatively specify fiscal impulse as deviation of real fiscal expenditures from long-term trend (equations (11)–(13)).
- Limitations:
  - Full flow-stock fiscal blocks and debt effects on country risk premium not implemented in country projects to date; QPM being a flow model makes full stock-flow fiscal modeling difficult.

### D. Incorporating Terms of Trade (TOT) Effects
- TOT modeling:
  - Decompose TOT into trend and gap: tot t = tot̅t + tot̃t  (equation (14))
  - Trend and gap follow autoregressive processes (equations (15)–(16))
  - Real exchange rate trend Δz̅t linked to TOT trend via reduced-form relation (equation (17))
  - Augment IS curve with TOT gap: ỹt = a1 ỹt − 1 + a2 Et ỹt + 1 + a3 mci t + a4 ỹt* + a5 tot̃t + εr̃,t  (equation (1c))
- Cautions:
  - Equation (1c) is reduced form and subject to Lucas critique; parameter a5 may change if transmission channels (e.g., fiscal channel) change.

### E. Addressing Domestic Supply Shocks and Multiple Phillips Curves
- Problem and solution:
  - Supply shocks (harvests, oil, mining) can bias output gap estimates and invert expected inflation-output relationships.
  - Use GDP excluding agriculture, oil, or mining to define output gap; alternatively include separate agricultural output gap and model its direct effects.
- Agricultural block example:
  - Agriculture output gap equation: ỹt ag = α1 ỹt − 1 ag + α2 ỹt* + α3 tot̃t − α4 mci t + α5 fimp t + εt ag  (equation (1d))
  - Nonagriculture output gap can include lagged agriculture gap: ỹt nag = β1 ỹt − 1 nag + β2 Et ỹt + 1 nag − β3 rmcit + β4 ỹt − 1 ag + β5 fimp t − 1 + εt nag  (equation (1e))
- Multiple Phillips curves:
  - Split headline CPI into components (food, nonfood, transportation, administrated prices) and model component-specific Phillips curves (equations (18a)/(18b), (19a)/(19b), (3b), (3c), (3d)).
  - Model relative prices and decompose into trend and gap (equations (20a)/(20b), (21), (22)).
  - Food inflation Phillips curve may include agriculture gap and TOT gap terms (equation (3c)).

### F. Incorporating Money and Money Targeting
- Money block fundamentals:
  - Real money growth Δrm t related to real GDP growth Δy t, changes in deviation of short rates from equilibrium, real money gap rm̃ t, change in trend velocity Δv̅t, and shocks (equation (23)).
  - Real money gap: rm̃ t = rm t − (yt − γ1(it − in t) − v̅t)  (equation (24))
  - Identity in equilibrium: Δm̃t − π4 t Tggs = Δỹt − Δv̅t (identity when in equilibrium).
- Uses and policy specifications:
  - Pure forecasting tool with exogenous money target path (Δrm t T).
  - Tool to set consistent intermediate targets on monetary aggregate and short-term interest rates; allow partial adherence to money targets via parameter mpr (equations (25), (6b), (6c)).
- Practical challenges:
  - Money-interest relationship and velocity often unstable; policy operations and opaque decision processes make money-block implementation difficult.
  - Experience with money block has been mixed in CD countries.

### G. Exchange Rate Persistence, Limited Arbitrage, Interventions, and Pegs
- Exchange rate persistence:
  - Expected spot rate as weighted average of model-consistent forecast and naïve forecast: Et st + 1 = ρs s t + 1 + (1−ρs) s t + 1 NNNN  (equation (26)); naïve forecast specified in equation (27).
  - Combined exchange rate equation yields augmented UIP with persistence (equation (5b)).
- Endogenous risk premium:
  - Add endogenous premium pr_premE t = α{β(πt 4 − πT t) − γ ỹt*} (equation (28)) and include in exchange rate (equation (5c)).
- Limited arbitrage and BOP-determined exchange rate:
  - When no arbitrage, exchange rate tied to BOP gap and its components (equations (29)–(32)); partial arbitrage via weighted average of UIP and BOP-based rates.
- Pegged/managed regimes and interventions:
  - Central bank may set explicit depreciation target Δs t T consistent with Δz̅t + πT t − π̅* (equation (33a)/(33b)).
  - Market rate may be partially determined by target and UIP depending on adherence parameter h1 (equations (5d)/(5e)).
  - Market interest rates set as weighted average between UIP-implied and Taylor-type rule implied rates with weight h2 (equation (6d)).
  - Foreign exchange intervention setup includes reserve targets, intervention reaction function, reserve gap dynamics, structural interventions, and UIP with intervention effects (equations (37)–(40), (5f), (38)–(39)).

### H. Persistent Deviations of Market Interest Rates from the Policy Rate
- Causes of persistent spreads:
  - Partial adherence to money/exchange rate targets, insufficient liquidity operations to align market rates with policy rate; political pressures keeping rates lower; misconceptions about policy implementation.
- Modeling approaches:
  - Omit policy rate and use market rate (overnight interbank or 91-day bill) directly.
  - Alternatively, include both policy rate and market rates and model spreads explicitly: it = ig t + spreadt  (equation (41)), spreadt autoregressive (equation (42)).
  - Represent market rate as weighted average of several short-term rates when markets are segmented (equations (43)–(45)).
- Risks and guidance:
  - Modeling spreads as semipersistent risks masking policy implementation failures and reinforcing misconception that announcing policy rate suffices without operations.
  - For policy simulations, modeler must understand and discuss sources of spreads candidly.
- Term structure and term premiums:
  - Longer-term rates can be modeled as functions of short-term expectations plus term premiums (equations (46)–(48)); term premiums may follow AR process or be tied to credibility proxies.

### I. Central Bank Credibility and Expectations
- Credibility modeling:
  - Expectations partially backward-looking and partially forward-looking; high weight on past inflation may indicate low credibility.
  - Example explicit credibility stock dynamics:
    - lack-of-credibility stock: incrpd t = φ1 incrpd t − 1 + φ2(πt − 1 4 − πt − 1 T) + επ t  (equation (49))
    - altered expectations: Et πt + 1 = πt + 1 4 + φ3 incrpd t  (equation (50))
- Role:
  - Credibility critically anchors expectations; incorporate into Phillips curve and policy analysis where relevant.

### J. Calibration of the QPM
- Calibration rationale and practice:
  - Parameters almost always calibrated rather than estimated to reflect expected future transmission mechanism and because emerging markets often lack long data series or have structural breaks.
  - Calibration supports deeper model understanding and partly addresses the Lucas critique.
- Methods to bring model to data and check parametrization:
  - Empirical fit, in-sample forecasting performance, economic consistency via impulse responses, Kalman filter to explain historical developments, and Bayesian estimation as a double-check.
- Policy rule parameters:
  - Policy rule parameters represent policymakers’ preferences and should be well understood; quantify and present implications of alternative rules (inflation-output volatility tradeoff, inertia of policy rate changes).
  - Parameterization should be reconsidered rarely (only after major regime or committee composition changes).

*Italic source: IMF Departmental Paper — Taking Stock of IMF CD on Monetary Policy Forecasting and Policy Analysis Systems, Chapter 3 (3. The Forecasting and Policy Analysis Toolkit).*

### 4. The Forecasting Process, Incorporating

### 4. The Forecasting Process, Incorporating the FPAS into the Decision-Making Process

### Structure and timing of the forecasting, policy analysis, and decision-making process
- Core objective: ensure technical FPAS work has traction in policymaking by structuring forecasting, policy analysis, and decision-making processes; briefing core forecasting staff to policymakers; assisting staffing, organization, and drafting of policy materials in real-time forecasting and policy analysis environment.
- Key elements of a structured monetary policy decision round:
  - A focused discussion of topical issues and an assessment of the latest forecasts and policy recommendations, the forecast errors, and the resulting forecast revisions early in the process.
  - Sufficient time to analyze foreign and financial variables (external assumptions).
  - Sufficient time to analyze the most recent data, conduct nowcasting to fill data gaps, prepare near-term forecasts, and assess current economic fundamentals (“the initial conditions”).
  - Organization of several rounds of model simulations.
  - A series of structured meetings at technical level and with policymakers at various stages.
- Typical duration: the whole forecasting round typically takes five to seven weeks from the start of the technical work to the final MPC meeting.
- Standard recommendation:
  - Undertake full forecasting rounds only quarterly.
  - Closely align the start of the process with the release of quarterly national accounts data.
  - Limit full-fledged forecast rounds to four per year to allow time for analysis, toolkit improvement, and to focus on medium-term drivers rather than high-frequency noise.
- Risks of more frequent full rounds: insufficient time to properly analyze data; overreaction to high-frequency noise leading to frequent reversals of policy stance, noisy policy signals, high interest-rate volatility, and weakened credibility and transmission.

### Interim meetings, meeting calendar, and sequencing
- Recommendation: have interim policy meetings between full forecasting rounds; many IT central banks began with 12 policy meetings a year but had analytically weak monthly reports.
- Recommended maximum: reduce policy meetings to a maximum of eight meetings a year, with interim meetings based on limited updates between full forecasting rounds.
- Timing considerations:
  - Start of process aligned with quarterly national accounts release; first technical meeting on initial conditions and near-term forecasts typically 7 to 10 working days after.
  - Avoid having major data releases (national accounts, CPI) shortly after the MPC meeting.
  - MPC meetings should ideally take place at the beginning of a new reserve maintenance period to avoid affecting bidding in open market operations.
  - Avoid scheduling policy meetings too close to events that will take media precedence (for example, the release of a new budget).
- Communication practice: preannounced and published monetary policy meeting calendar and MPR publication calendar recommended to guide market expectations, assist media, and discipline internal scheduling.
- Role of forecast coordinator: schedule detailed tasks and enable department management to plan projects and maintenance work.

### Pre-MPC technical meetings and interaction between staff and policymakers
- Purpose of technical (“pre-MPC”) meetings:
  - Inform policymakers of the economic situation, forecast assumptions, and proposed policy action to reach the target at end of forecast horizon.
  - Allow MPC members to discuss and present views to staff for incorporation.
  - Inform MPC members of each other’s views so the MPC enters the policy meeting well informed and focused on policy discussion.
- Importance of reconciling differing views:
  - Commonly, staff and policymakers differ on assessments of current conditions, exogenous variable paths, risks, strength of transmission channels, and how factors impact the economy.
  - Active participation and close interaction improve forecast and policy analysis quality and ensure full use of available information.
  - Sufficient time for discussion helps identify model misspecification, trigger research, and build consensus on transmission mechanism.
  - Both staff and policymakers share responsibility: staff must present clearly; policymakers must state what they want and how they want analysis presented.
- Risks of inappropriate processes:
  - Fully staff-dictated process (little room for policymakers) can lead to distrust and inconsistent policy decisions.
  - Policymaker-dictated process (imposed “hard model tunes”) can produce internally inconsistent and biased forecasts.
  - Both outcomes risk suboptimal policy, demoralized staff, and weakened public trust.
- Recommended three pre-MPC technical meetings:
  - Meeting on initial conditions (first pre-MPC):
    - Provide economic interpretation of new data, external/financial developments, forthcoming policy changes, and staff’s preliminary near-term forecasts and cyclical assessment.
    - Seek policymakers’ feedback on additional analysis or reassessment.
  - Meeting on first version of baseline forecasts and alternative scenarios (second pre-MPC):
    - Seek feedback on forecasts and implied monetary policy decision.
    - Discuss specification of alternative scenarios to complement the baseline for the final MPC meeting.
    - Provide opportunity for policymakers to state views on policy tradeoffs as reflected in the QPM policy reaction function.
    - Ensure sufficient time between this meeting and the final MPC to incorporate feedback into final baseline and the MPR.
  - Meeting to present final forecasts, policy analysis, and policy recommendations:
    - Either on the same day as the MPC decision meeting or the day before (can be a segment of the MPC meeting where staff present and then leave).

### Additional interactions and capacity-building
- Additional staff–MPC meetings, joint seminars, and workshops between forecasting rounds are beneficial for:
  - Discussing forecasting and policy tools, particularly the QPM.
  - Presenting ongoing research, ex post evaluations, contrafactual simulations, and preliminary analysis for future MPR boxes/chapters.
  - Startup meetings before each round to review previous round, assess current risks, and highlight issues needing focus.
- TA practice: CD projects have supported seminars for central bank management and decision-making bodies to solidify political support and commitment to FPAS.

### Staffing and organization of the Forecasting Team (FT)
- Common shortcomings observed in CD recipients:
  - Under-resourcing FPAS-related work for establishment and operation.
  - Treating FPAS as research or training, delegating to small research-oriented teams rather than embedding in monetary policy department.
  - Insufficient participation from a broad range of central bank staff and sectoral experts, causing weak interaction between modeling staff, sectoral experts, and communications/policy document preparers.
- Typical FT composition and recommended team sizes:
  - Core modeling/forecasting and monetary policy analysis group:
    - Main coordinator of forecasting and policy analysis.
    - Required skills: mathematical, statistical, programming, macroeconomics, policy making, economic intuition, domestic economy knowledge, and ability to present simply (“storytelling”).
    - Recommended size: at least six or seven people.
    - Suggested rotation of roles; three or four participate in regular quarterly rounds with the rest on research/model development.
  - Real sector team:
    - Recommended size: six or seven people.
    - Focus: near-term forecasts (NTFs) for prices, GDP and components, labor market.
  - External sector team:
    - Tasks: prepare external sector NTFs and inputs for QPM-based medium-term forecasts and detailed BOP forecast disaggregation.
    - Size guidance: two or three people if relying on forecasts from other institutions; up to six or seven people if producing own model-based forecasts.
  - Fiscal sector team:
    - Tasks: follow fiscal policy, prepare near-term fiscal forecasts in QPM (when fiscal block exists), and assess fiscal policy interactions.
    - Size guidance: three people if relying on ministry of finance analysis; six or seven if producing own fiscal analysis.
  - Financial sector team:
    - Tasks: monitor money, credit, and FX markets and provide inputs.
    - Size guidance: one or two people.
  - Drafting and external communication team:
    - Tasks: prepare MPC documents (reports/presentations) and public communications (press releases, speeches, inflation/MPRs).
- Coordination and leadership:
  - Forecast coordinator role must be clearly assigned, preferably to a relatively senior and seasoned staff member with vision for change, technical skills, vertical and horizontal communication ability, hands-on involvement, and clear mandate to command the FT during forecasting rounds.
  - Precise seniority depends on internal circumstances and mandate to coordinate interdepartmental senior staff.

*Source: tsicdmpfpasea - 4. The Forecasting Process, Incorporating*

### 5. Assessment and Lessons

### 5. Assessment and Lessons

### A. Assessment — observable impact and determinants
- Determinants of FPAS CD impact included:
  - maturity of authorities’ monetary policy reform plans and whether the CD project was part of a clear reform strategy with strong commitments from the top;
  - technical skills of staff at project onset;
  - staff and management turnover during the project;
  - duration to date of the CD project and the authorities’ reform efforts.
- Most projects: capacity development and technical work (development of the QPM and NTF tools) progressed broadly as planned.
- Incorporation of model-based analysis and projections into policymaking, and the projects’ impact on actual policy decisions, were more mixed.

- Countries with strong transformational impact:
  - Most adopted, or were firmly on the way toward adopting, a full-fledged IT, IT-Lite, or IT-like monetary policy framework during the project.75
  - FPAS CD was described by authorities in one country as one of the most transformative CD projects they had ever received.
  - In these countries, besides strengthening modeling and forecasting capacity, the project helped to:
    - Enhance internal organization, work, policy formulation processes, and external monetary policy communication.
    - Change how central banks think about monetary policy and monetary policy analysis, with more emphasis on initial conditions, main underlying economic forces (real exchange rate, neutral interest rate, potential output, etc.), and the nature of shocks (for example, their expected persistence).
    - Enrich and better-structure macroeconomic and policy analysis, producing clearer analysis and more focused policy discussions.
    - Put in place a well-structured policy decision-making process: regular MPC or board policy meetings on a preannounced schedule; forward-looking interest rate decisions based on projections and policy analysis prepared with project tools and processes; regular publication of forward-looking quarterly MPRs highlighting analysis and projections.
  - Common enabling conditions: staff with relatively strong technical and modeling background, strong management commitment, and clear desire and political backing for reforms. Transitions ranged from a few years to more gradual, multi-year processes with prolonged CD support.

- Countries with more muted impact:
  - Typical initial conditions: weaker analytical capacity; somewhat erratic policy implementation; large and persistent deviation of market interest rates from the policy rate; little or no forward-looking information published.
  - Less institutional support: parts of management supportive but other parts less so; political backing for reform often weak or unclear.
  - Notable progress in many such countries:
    - Internal policy discussion, decision-making process, and external communication became more transparent, analytical, forward-looking, and policy-focused.
    - Policy dialogue with IMF country teams typically improved.
    - Policy implementation and liquidity management became more consistent in some countries, with interbank rates better aligned with the policy rate for prolonged periods.
  - Substantial remaining weaknesses often included:
    - Continued heavy management of the exchange rate and slippages in policy implementation because of cost concerns.
    - Incoherent internal policy analysis and decision-making, including competing or conflicting advice from different parts of the organization with no orderly consensus process.
    - Frequent staff rotation that in some cases caused the capacity built to be lost.
    - Weak human resource management practices and incomplete integration of the FPAS into policymaking; top management sometimes did not appreciate full benefits of a well-functioning FPAS; frequent changes in top management contributed.
    - Political considerations dominating policymaking, reducing the impact of FPAS-based analysis and recommendations.
  - It remains uncertain whether the capacity built in these cases will be sustained and whether central bank management will be able to implement the reforms the CD project provided a foundation for.

- Instances of lost FPAS capacity:
  - Causes included reorganizations and staff relocations following changes in the governor’s or top management’s positions, resistance from influential middle management staff, and gaps in CD engagement.
  - Insufficient incorporation of model-based analysis and projections into decision-making and lack of progress modernizing the monetary policy framework and forward-looking policy formulation were often underlying causes.

### B. Lessons — design, implementation, and sustainability
- Project planning and assessment:
  - Detailed assessment of capacity, organization, and process in the central bank prior to formulating the FPAS CD project, clarity of project objectives, and realistic assessment of what can be achieved are essential.
  - Scoping missions are important for assessing initial capacity and internal organization and processes; these should inform project objectives, timeline, and milestones.
  - Project Logical Frameworks (logframe) should lay out objectives, specify expected outcomes with measurable indicators, and include a detailed timeline with clear milestones to plan and monitor progress.76
  - Close dialogue between CD deliverer and CD recipient at all stages is essential; periodic project assessments enable reassessment and corrective measures.

- Time horizon and continuity:
  - FPAS projects should have a multiyear perspective. Medium-term planning needs a clear view about frequency of missions and their scope.
  - Interruptions and delays, and rotation of recipient central bank staff, can jeopardize the project. Capacity being built can easily be lost before practical use.
  - Three to four missions a year over several years are often needed during the initial phase. Depending on initial capacity and progress, multiple phases may be needed.
  - Continuity of IMF experts implementing projects is important; much work involves hands-on follow-up and trust-building given privileged access to sensitive information.
  - Continuity of CD engagement, even after basic capacity is established, can be important for sustainability because capacity can be lost as staff move to other assignments and new staff need retraining.

- Integration with broader monetary policy modernization:
  - FPAS projects have greater chance of success when undertaken with synergies in other aspects of monetary policy modernization:
    - (1) reform the monetary policy framework;
    - (2) enhance monetary operations and liquidity management so short-term market interest rates align with the policy rate;
    - (3) establish a foreign exchange interventions policy clarifying the role of FX interventions and protecting natural short-term exchange rate volatility;
    - (4) enhance financial stability and develop financial markets.
  - FPAS CD can be a catalyst for such reforms.

- Modeling, tools, and practical use:
  - FPAS CD is not mainly about modeling. Introducing modeling tools requires much more than developing models.
  - Building the QPM and the NTF tools may not be the most important ingredient. Models can be built remotely and relatively quickly if data are available.
  - Success requires staff skills to maintain and develop QPM and NTF tools and, importantly, skills to use them properly for preparing forecasts, policy analysis, and policy recommendations.
  - Training staff in applying the tools and communicating model-based analysis is essential but often not sufficient.
  - Keep tools simple and build a strong base for forecasting and policy analysis: data warehouse, measures and indicators, macro diagnostic framework, and skills for writing MPC briefings and presentations.

- Judgment, expertise, and team composition:
  - Proper judgment and solid understanding of current developments and initial conditions are essential. Core forward-looking models depend on hard-to-measure unobserved variables (potential output and output gap, trend or equilibrium real exchange rate and exchange rate gap, equilibrium interest rates). Errors in these estimates can increase forecast errors and produce bad policy advice.
  - Best approach: keep core model relatively simple and combine model use with expert judgment, sectoral analysis, satellite models, and filters.
  - Build the broader Forecasting Team (FT) for sustainability and depth of analysis to avoid mechanical “running of the model” by a small modeling team lacking nowcasting and sectoral expertise. High-quality forecasting requires inputs from a wide range of central bank staff and consensus within an FT that includes core modelers and many sectoral experts.

- Organizational design, roles, and conflict management:
  - Establishing and sustaining an FPAS requires rooting the system in the organization, putting tools to practical use quickly, and integrating them into decision-making so policymakers view them as useful for analytically backed decisions.
  - Rethink human resources, establish regular forecasting rounds based on forecasting calendars, and set up well-designed internal communication among staff and between staff and decision makers, including regular meetings during forecasting rounds.
  - Work on FPAS organization and related processes should proceed in parallel with technical capacity development; the two streams are mutually reinforcing.
  - Potential interdepartmental conflicts (e.g., monetary policy department vs research department ownership of the project) should be recognized and managed early. Insufficient involvement of financial stability and markets departments can cause conflicts and deviations between formal policy stance and de facto market rates.

- Leadership and staffing of the FT:
  - The role of the forecast coordinator must be clearly assigned, preferably to a relatively senior and seasoned staff member with good technical skills and economic intuition, strong communication abilities, and direct hands-on involvement in day-to-day FT work.
  - Formalizing the forecast coordinator role (for example in terms of reference) is important for accountability and for providing formal authority to manage the team.
  - Best practice for the modeling team: have two staff members who could undertake each task (a lead expert and a backup). Preferably have two full teams that rotate between running forecasting and policy analysis rounds and working on maintaining and improving tools, procedures, and related research.

- Human resources, incentives, and documentation:
  - Human resources issues should not be overlooked. Sustainability requires a conscious HR policy to ensure key FPAS functions are properly staffed and new staff are recruited and trained before key incumbents move on.
  - Establish clear incentives for FT members to prioritize FPAS work, e.g., reflect FPAS duties in staff terms of reference and/or key performance indicators.
  - Proper documentation of analytical tools, organization, and processes is critical for sustainability. Documentation and manuals should ideally be prepared by the CD recipient to ensure ownership and mastery.

*IMF Departmental Paper — 5. Assessment and Lessons*

### Annex 1. The Forecast Calendar

### Annex 1. The Forecast Calendar

### Purpose and core components of the Forecasting Calendar (FC)
- The FC frames the forecasting process, ensures careful planning, and allows monitoring by the head of the FT and management.
- The FC should be available to all involved staff and approved by central bank management.
- The FC should include:
  - Release dates for key data (important for sequencing analytic work).
  - Dates for national holidays (potentially limiting available time for producing inputs).
  - Timing of all departmental meetings and departmental meetings with policymakers.
  - Deadlines (specified in terms of day and hour) for updates of the centralized database and submissions of all key intermediate forecast inputs; ideally including personal responsibilities for delivery with deadlines.

### Example assumptions in the illustrative FC
- The quarterly national accounts data for the first quarter of the calendar year are released in the afternoon on Friday, June 28. Work (analyzing the data and preparing the forecasts) starts immediately after that (weeks 1–6 of the FC).
- The CPI data (June in this case) are released in the morning 12 working days after the end of the month. In the example, that is just after the meeting with the MPC on the initial conditions but almost a week before the meeting with the MPC on the first version of the forecasts.
- The final MPC meeting and announcement of the policy decision and release of the forecasts take place five working days after the end of the month (August in this case).
- The release of the next CPI data would take place around seven working days after that again.
- No major new data are released between the time the final version of the forecasts and policy recommendation are finalized and the time of the MPC monetary policy decision that could require last minute changes to the forecasts.
- Example organizational structure for the central bank’s Monetary Policy Department:
  - (1) Economic Policy Analysis Division with a Monetary Policy Analysis Unit and a Fiscal Analysis Unit;
  - (2) Macroeconomic Forecasting Division with a Near-term Forecasting Unit, a Medium-term Forecasting/Core Model Unit, and a Model Development Unit;
  - (3) International Economic Analysis Division responsible for preparing the forecasts and monetary policy recommendations.
- This department, together with the communication department, prepares all documents for meetings with the MPC and externally published monetary policy documents (press releases, MPR, and presentations and speeches by the governor and senior management).

### Week-by-week timetable and responsibilities
- Week 0 — Preparatory work and introductory meetings:
  - Devoted to preparatory work; staff continuously analyze/transform new data as released.
  - Work on improving forecasting techniques (nowcast/NTF and QPM-based medium-term forecasts) may occur.
  - Meeting to discuss and decide on changes to forecasting techniques should take place ahead of new QNA release.
  - Announcement of detailed time schedule for the upcoming forecasting cycle and internal briefings should occur before QNA release so staff can focus on those data once released.

- Week 1 — Analyze new QNA data and prepare first NTF estimates:
  - Dedicated to analyzing new QNA data, identifying issues affecting forecast preparation, and preparing first NTF estimates.
  - Departmental-level issues meeting identifies important domestic and international developments, especially events not captured by NTF models and/or QPM.
  - Departmental-level meeting on NTFs, BOP, fiscal developments and proposed trajectory of exogenous model variables (foreign output gap, foreign inflation, foreign interest rate, commodity prices, fiscal, etc.) scheduled for the last day of the week.
  - Sectoral experts have approximately four or five days to analyze national accounts, produce first estimates, and prepare presentations.
  - Iterative internal team meetings and forecast iterations are expected to distill and synthesize team knowledge.

- Week 2 — Determine initial conditions for medium-term forecasts:
  - Reserved for setting initial conditions for medium-term forecasts and preparing background materials for the first preparatory meeting with the MPC on initial conditions.
  - NTFs and sectoral assessments feed into a consensual view on the current cyclical state of the economy.
  - Setting initial conditions requires close interaction between sectoral experts and the core modeling group, and group-level, FT-level, and departmental-level discussions.
  - Approximately one week is typically needed for technical-level discussions and preparation of material for the preparatory MPC meeting.
  - The meeting with the MPC is scheduled for the Monday of week 3 to allow MPC members time to review materials.

- Week 3 — Prepare first version of medium-term forecasts and policy recommendations:
  - Dedicated to preparing the first version of medium-term forecasts and associated monetary policy recommendations, starting immediately after the MPC initial conditions meeting.
  - Policymaker feedback on interpretation should be obtained to reduce risk of later fundamental disagreements.
  - Departmental-level meeting on the first version of forecasts and policy recommendations is scheduled for the middle of week 3.
  - Core modeling unit should begin work before the MPC initial conditions meeting to identify bottlenecks and assist preparation.
  - Multiple iterations and potentially several team-level forecasting meetings (sometimes not formally scheduled) are expected.
  - Draft of first version of forecasts and policy recommendations will typically be revised following departmental-level meeting; two days are provided to prepare material for the second MPC meeting (scheduled for the Monday of subsequent week).

- Week 4 — MPC meeting on first version and preparation of final version:
  - Devoted to the meeting with the MPC on the first version and preparing the final version reflecting that meeting’s outcome.
  - This meeting is the most important preparatory meeting with the MPC; feedback must be incorporated into the final forecast and policy recommendation.
  - Meeting should lead to agreement on alternative scenarios to be produced (requested by policymakers or proposed by staff).
  - Departmental meeting on the final version of the forecast is scheduled two days after the MPC meeting; it may approve the draft final version or provide final fine-tuning observations.
  - Head of the FT informs department and staff preparing MPC materials, MPR, and press release that the final forecast is completed and database updated.

- Week 5 — Prepare background material for MPC policy meeting and documents to be published:
  - Dedicated to preparing background materials for the MPC meeting and documents to be published.
  - Completing the MPR may take two or three days and typically requires several drafting sessions—one for each major part and one final session for the executive summary.
  - Authors of individual sections attend sessions covering their sections and the final session; editors and core drafting team attend all sessions.
  - FC should include clear deadlines for submission of drafts of individual sections to allow time for review.
  - Some report parts are not dependent on the final forecast and should be submitted first (first-round, second-round inputs).
  - The MPR should ideally be submitted to the MPC at least 1.5 to 2 days before the MPC policy meeting.
  - Drafts of all communication material (especially press releases) consistent with the MPR should be completed before the MPC meeting so they can be discussed, revised, and approved during that meeting.
  - Staff may prepare an internal document for the MPC elaborating on policy recommendation, risks and tradeoffs, alternative scenarios, and communication considerations. A high-level/department management meeting on the last day of week 5 may finalize this document.

- Week 6 — MPC policy decision meeting and postmortem:
  - Final week is dedicated to the MPC policy decision meeting.
  - Release of policy decisions should ideally take place in the middle of the week to give markets and public time before the weekend and to reduce market distortions.
  - Some central banks schedule the meeting over two days to allow staff presentations, closed-door MPC member-only meeting, and last-minute changes to the MPR.
  - Final documents should be submitted to the MPC at least one day before the meeting; typically include draft press release, summary presentation, internal document elaborating policy recommendation, and staff presentation.
  - On the day of the monetary policy decision, staff incorporate all requests regarding final wording of communication documents from MPC; drafting suggestions from policymakers or advisors should be collected and incorporated as soon as possible.
  - Some central banks schedule the release on the morning after the meeting so press release and MPR can be published together shortly after the MPC meeting; this requires tight control to minimize leaks.
  - Process ends with a postmortem technical-level meeting aimed at drawing lessons from systematic or ad hoc errors and adopting procedures to prevent repetition.

### Operational recommendations and timing guidance
- Prepare, approve, and communicate the FC well ahead of the forecasting cycle; preferably for the next 12 months.
- Ideally determine timing of MPC monetary policy decision meetings for the next 12 months and make timing available to the public.
- Include key activities involving forecasting staff planned between interim MPC meetings in the internal calendar.
- Schedule the meeting to discuss changes to forecasting techniques before the release of new QNA data.
- Provide approximately one week for technical-level discussions on initial conditions and 1.5 to 2 days for MPR submission prior to MPC meeting.
- Ensure sufficient time between data releases and MPC meetings (example: CPI 12 working days after month end; final MPC meeting five working days after month end; next CPI around seven working days after that).
- Maintain iterative forecast rounds and team-level discussions to synthesize collective expertise and produce analytically backed materials for policymakers.
- Finalize MPR and communication materials in time to allow MPC members to review and to incorporate last-minute edits while minimizing leaks.

*Source: tsicdmpfpasea - Annex 1. The Forecast Calendar*

### Annex Table 1.1. Prototype Forecast Calendar: 2019 24 June – 9 August

### Annex Table 1.1. Prototype Forecast Calendar: 2019 24 June – 9 August

### Calendar overview
- Period covered: 24-VI. (24 June) – 9-VIII. (9 August) 2019.
- Purpose: Detailed timetable of forecasting-cycle events, meetings, submission deadlines, data release dates, and database updates for a monetary policy forecasting cycle.

### Weekly schedule highlights
- WEEK 0 (24-VI. – 28-VI.)
  - 24-VI. Mon 09:00 — Announce time schedule for the upcoming forecasting cycle.
  - 25-VI. Tue R #1 13:00 — Meeting on forecasting techniques - Department. Data release: 9:00 High freq. data #1.
  - 28-VI. Fri 9:00 — Q1 National accounts; 15:00 — National accounts.

- WEEK 1 (1-VII. – 5-VII.)
  - 1-VII. Mon — NTF and trajectory for exogenous variables.
  - 3-VII. Wed R #3 13:00 — Issues meeting - Department. Data release: 10:00 Nowcast.
  - 4-VII. Thu — NTF and trajectory for exogenous variables; 9:00 High freq. data #2; 10:00 Near Term Forecast.
  - 5-VII. Fri R #3 9:00 — NTF + BoP + Fiscal + other exogenous variables - Department.

- WEEK 2 (8-VII. – 12-VII.)
  - 8-VII. Mon — National Holiday.
  - 9-VII. Tue — Initial conditions; 14:00 External sector.
  - 10-VII. Wed R #1 13:00 — Initial Conditions - Department.
  - 11-VII. Thu 10:00 — Fiscal policy.
  - 12-VII. Fri 14:00 — Documents for MPC meeting.

- WEEK 3 (15-VII. – 19-VII.)
  - 15-VII. Mon R #2 14:00 — MPC Meeting: Initial Conditions.
  - 16-VII. Tue — Produce the First Version of the Forecast; 9:00 June CPI; 15:00 CPI.
  - 17-VII. Wed R #2 13:00 — First Version of the Forecast - Department.
  - 18-VII. Thu 9:00 — External assumptions.
  - 19-VII. Fri 16:00 — 1st round inputs into the MPR; 12:00 External forecast; 14:00 Documents for MPC meeting.

- WEEK 4 (22-VII. – 26-VII.)
  - 22-VII. Mon 9:00 — High freq. data #3. R #3 14:00 — MPC Meeting First Version of the Forecast, alt. Scenarios. 16:00 — 2nd round inputs into the MPR.
  - 24-VII. Wed R #3 13:00 — Final Version of the Forecast - Department; 16:00 — 3rd round inputs into the Monetary Policy Report. 9:00 — High freq. data #4.
  - 25-VII. Thu 16:00 — 4th round inputs into the MPR; 16:00 — Final forecast data distributed.
  - 26-VII. Fri 16:00 — 5th round inputs into the MPR; 14:00 — Documents for Tuesday drafting session.

- WEEK 5 (29-VII. – 2-VIII.)
  - 29-VII. Mon 16:00 — Documents for Wednesday drafting session.
  - 30-VII. Tue R #4 9:00 — Drafting Session - MPR; 16:00 — Documents for Thursday drafting session.
  - 31-VII. Wed R #4 9:00 — Drafting Session - MPR.
  - 1-VIII. Thu R #4 9:00 — Drafting Session - MPR; 16:00 — Finalizing the MPR.
  - 2-VIII. Fri 14:00 — Monetary Policy Report to MPC.

- WEEK 6 (5-VIII. – 9-VIII.)
  - 7-VIII. Wed R #1 9:00 — MPC policy decision meeting.
  - 9-VIII. Fri R #3 13:00 — Post Mortem - Department.

### Key meetings and deliverables
- MPC-related meetings and deliverables:
  - 15-VII. Mon 14:00 — MPC Meeting: Initial Conditions.
  - 22-VII. Mon 14:00 — MPC Meeting First Version of the Forecast, alt. Scenarios.
  - 7-VIII. Wed 9:00 — MPC policy decision meeting.
  - 2-VIII. Fri 14:00 — Monetary Policy Report to MPC.
  - Multiple Document inputs into the MPR: 1st round (19-VII. Fri 16:00), 2nd round (22-VII. Mon 16:00), 3rd round (24-VII. Wed 16:00), 4th round (25-VII. Thu 16:00), 5th round (26-VII. Fri 16:00).

- Forecast production milestones:
  - 16-VII. Tue — Produce the First Version of the Forecast.
  - 17-VII. Wed 13:00 — First Version of the Forecast - Department.
  - 24-VII. Wed 13:00 — Final Version of the Forecast - Department.
  - 25-VII. Thu 16:00 — Final forecast data distributed.

- Drafting and finalization of Monetary Policy Report (MPR):
  - Drafting Sessions: 30-VII. Tue 9:00; 31-VII. Wed 9:00; 1-VIII. Thu 9:00.
  - 1-VIII. Thu 16:00 — Finalizing the MPR.
  - 2-VIII. Fri 14:00 — Monetary Policy Report to MPC.

### Data releases and database updates
- High-frequency data updates:
  - 25-VI. Tue 9:00 — High freq. data #1.
  - 4-VII. Thu 9:00 — High freq. data #2.
  - 22-VII. Mon 9:00 — High freq. data #3.
  - 24-VII. Wed 9:00 — High freq. data #4.
- Official national accounts and price data:
  - 28-VI. Fri 9:00 — Q1 National accounts; 15:00 — National accounts.
  - 16-VII. Tue 9:00 — June CPI; 15:00 — CPI.
- Other data/calendar items:
  - 3-VII. Wed 10:00 — Nowcast.
  - 9-VII. Tue 14:00 — External sector.
  - 11-VII. Thu 10:00 — Fiscal policy.
  - 19-VII. Fri 12:00 — External forecast.

### Roles, meeting types, and coding
- Color coding (legend):
  - Departmental level meeting (full forecasting team).
  - Meetings with the MPC.
  - Drafting sessions. Meeting of the drafting sub-team.
- Abbreviations used:
  - CPI = consumer price index
  - MPC = monetary policy committee
  - MPR = monetary policy report
  - NTF = near-term forecasting

*Source: Authors' calculations.*

### 1.    Developing a fully operational FPAS and incorporating it into the decision-

### 1.    Developing a fully operational FPAS and incorporating it into the decision-making process

### Monetary policy framework status and governance issues
- The MP framework is transitioning from money targeting to one focused on inflation and with interest rates as the operational target.
- An MPC and an MP technical committee is already in place, with regular quarterly two-day MPC meetings based on a preset, but not published, meeting calendar.
- MPC meeting minutes already published; the economic content of the minutes can be enhanced.
- An annual monetary policy statement is published and can be enhanced.
- While price stability is the stated policy objective, a numerical medium-term inflation objective is not publicly communicated.
- The legal framework governing monetary policy objectives and operations could benefit from an update.
- CB operational independence could benefit from a strengthening.

### Objective 1 — Strengthening the Forecasting and Policy Analysis System (Log Frame)
- Strategic Objective: Developing a fully operational FPAS and incorporating it into the decision-making process.
- Projective Objective 1 Verifiable Indicators and General Risks/Assumptions:
  - Stream 1: Developing capacity for MP analysis and forecasting, including NTF.
  - Stream 2: Developing a fully operational FPAS and incorporating it into the decision-making process.
  - NOTE: The two streams are overlapping and should be implemented in parallel.
- Verifiable Indicators:
  - Stream 1: Adequate NTF tools are used to produce regular short-term forecasts that are feeding into a QPM-based analytical and forecasting framework for producing regular medium-term projections and policy analysis.
  - Stream 2: Near and medium-term forecasts and staff analysis are used by the MPC in their policy decision-making.
  - MP communication is strengthened to enhance transparency (as evidenced, for example, by medium-term forecasts being published in the quarterly analytical reports).
- General Assumptions:
  - Long-term commitment by the CB management.
- Risks:
  - Stream 1: Lack of human resources, inadequate software. Resistance from the CB staff to endorse organizational changes, causing barriers to internal communication and information sharing.
  - Stream 2: CB management is not willing to change the established decision-making practices and/or to adopt a coherent transparent forward-looking MP framework.

### Stream 1 — Building FPAS Capacity (Outcomes, Indicators, Milestones, Risks/Assumptions)
- Expected outcomes:
  - Economic analysis and forecasting capabilities at the CB improved.
  - FT is setup with representatives from all relevant divisions within Research and Statistics Department.
  - QPM-based analytical and forecasting framework is used to produce regular medium-term projections.
  - Recommendations provided on the implementation of FPAS, based on review of existing practices.
- Milestones (selected activities and dates unspecified in source):
  - Detailed work plan for the FPAS technical assistance and training activity prepared in coordination with CB management.
  - The role of macroeconomic analysis and forecasting in decision-making process re(de)fined.
  - Create economic forecasting and modeling team within Research and Statistics Department, responsible for maintaining and operating the core model infrastructure.
  - Data release calendar and templates created.
  - Data collection and recording process is in place.
  - Forecast calendar created.
- Training and model development milestones:
  - Training provided to develop core QPM.
  - FT staff is trained in using software for macroeconomic modeling and forecasting.
  - Core FT members attend relevant ICD courses.
  - First version of core QPM and adjacent codes infrastructure is developed in close cooperation with the FT staff.
  - FT staff is trained in operating the model codes infrastructure.
  - Core QPM is recalibrated and extended.
  - Core QPM is fully operational.
  - Training provided to develop and/or improve satellite models.
  - Existing NTF tools and current practices reviewed.
  - Satellite models are fully operational and are used to produce inputs for the core QPM.
  - Training provided to do MP analysis and produce forecast using core QPM.
  - FT staff is trained in using QPM to produce baseline projection, conditioning on all available information (domestic and foreign sector data, NTF) and expert judgment.
  - FT staff is trained in conducting scenario analysis.
  - FT staff is trained in analyzing forecast errors.
  - A process for systematic evaluation of forecast performance is established.
- Assumptions:
  - Well-defined internal processes.
- Risks:
  - Data collection is not adequate.
  - Lack of hardware and software.
  - Lack of human resources.
  - Staff turnover.

### Stream 2 — Incorporating FPAS into the Decision-Making Process (Outcomes, Indicators, Milestones, Risks/Assumptions)
- Expected outcomes:
  - Decision-making process streamlined and responsibilities within the CB clarified.
  - Organizational changes necessary to integrate FPAS into decision-making process are implemented, as evidenced by a revised forecasting calendar.
  - Existing structure of the CB and responsibilities of various internal departments reviewed, and recommendations provided to improve decision-making process.
- Country-specific risks/assumptions:
  - Risks: Budgetary constraints, Lack of human resources, Staff turnover.
- Status and verifiable indicators (implementation signals):
  - MPC and MP technical committee already in place.
  - MPC understands the structure of the core QPM and the underlying FPAS mechanisms, and uses staff projections in their policy decision-making.
  - FT delivers a presentation to MPC on the structure of QPM and the details of the integrated forecast process.
  - At least one MPC member participates regularly in the cornerstone FT meetings.
  - Regular interactions between MPC/MP technical committee and FT take place according to the forecast calendar.
- Milestones and capacity-building activities:
  - Training provided to produce consistent well-focused forecast report and presentation for MPC.
  - A senior management member is appointed as FPAS coordinator.
  - FT prepares an experimental QPM-based medium-term projection and policy analysis.
  - FT is trained in producing a story-driven forecast presentation.
  - FT prepares a mock forecast presentation for MPC.
  - Assistance provided during a real-time forecasting round.
- Outcomes for decision-making:
  - Outcomes (that is, baseline forecast, scenarios, risks, forecast error) are presented to the MPC to provide analytical basis for decision-making process and help frame the final policy decision.
  - FT provides policy recommendation to MPC.
- Assumptions:
  - Well-defined internal processes.
  - Open communication channel between CB staff and decision makers.
- Risks:
  - Lack of buy-in from the top management.

### Outputs, Activities, and Inputs
- Outputs:
  1. Assessment of current practices
  2. Advice on organizational changes needed
  3. Training in modeling and forecasting
  4. Advice on FPAS implementation
  5. Advice on enhancing transparency and strengthening monetary policy communication channels
- Activities/Inputs:
  - Short-term (multiple) HQ staff missions and/or short-term (multiple) expert missions (backstopped by HQ staff).
  - HQ-based hands-on workshops.
  - Off-site practical assignments.
  - ICD courses.
  - Regional joint workshops and/or higher-level retreats.

*Source: Authors’ calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/dp/2021/english/tsicdmpfpasea.pdf_