## Preface — IMF Technical Report: Tonga TA on Developing FPAS at the NRBT

## Source details

**Canonical URL:** [Preface — IMF Technical Report: Tonga TA on Developing FPAS at the NRBT](https://www.imf.org/-/media/files/publications/tar/2025/english/tarea2025099-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/tar/2025/english/tarea2025099-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/tar/2025/english/tarea2025099-source-pdf.pdf.json)

---

### Project scope and timeline
- At the request of the National Reserve Bank of Tonga (NRBT), a Macroeconomic Frameworks Technical Assistance (TA) Project on developing a Forecasting and Policy Analysis System (FPAS) for NRBT was conducted by ICD during 2021-25.
- Preface statement: "A total of eleven missions (five in -person and six virtual) were conducted during 2021-25."
- Executive Summary statement: "Following a scoping mission in November 2021, a total of ten missions (five in -person and five virtual) were conducted."
- Mission team:
  - Celine Rochon (ICDFP, Project manager, who participated in all TA missions).
  - Sam Ouliaris (short-term expert, who participated in all TA missions and reviewed this TA report).
- During in-person missions the TA team met regularly with the Governor Mr. Tatafu Moeaki and the Assistant Governor for Policy, Mr. Daniel Taumoepeau.
- The mission team acknowledges the generous support of the Government of Japan under the Japanese Sub-Account (JSA) funding.

### Core objectives and workstreams
- Two broad and complementary workstreams:
  - i. Developing analytical tools for macroeconomic assessment and forecasting among NRBT staff.
  - ii. Integration of the resulting analytical reports into NRBT’s policy processes and external communications.
- Project aims:
  - Build new institutional capacity for model-based forecasting and policy analysis.
  - Streamline decision-making processes.
  - Strengthen forward-looking monetary policy communication.

### Key deliverables and tools
- Main analytical tool: the Quarterly Projection Model (QPM) as the core operational model for medium-term projections.
- Nowcasting and Near-Term Forecast (NTF) tools developed, including:
  - Simple extrapolations of past trends.
  - Univariate methods (Autoregressive Integrated Moving Average (ARIMA)).
  - Bridge equations linking GDP to monthly/daily indicators.
- Data-management work: cataloguing and saving available Tongan time series data in EViews.

### Project achievements (Executive Summary)
- Overall: The FPAS TA project has largely achieved its objectives.
- NRBT forecasting team progress:
  - Staff progress described as "remarkable progress" and "regularly delivering high-level analytical support to the policymakers."
  - Staff can use a "rich suite of analytical tools" and follow the internal forecast calendar to prepare model-based results and forecasts for key macroeconomic variables.
  - "Operational use of the core Quarterly Projections Model (QPM) in producing baseline forecasts, alternative scenarios, using expert judgements, assembling the forecast narrative and delivery to peers is performed efficiently and following best central bank practice."
- Analytical work role:
  - "Model-based results are communicated directly to the Governor."
  - "Forecasts feature, to some extent in external communications, including Monetary Policy Reports."

### Key macroeconomic facts about Tonga (as stated)
- Inflation reference rate: "currently of 5 percent."
- Foreign reserves objectives: minimum of "3 months" of imports and optimally around "7.5 months" of imports.
- Exchange rate flexibility: NRBT has flexibility "to adjust the exchange rates by 5 percent each month if necessary."
- Trade and external balances:
  - Trade deficit averaged "about 50 percent of GDP" in the past decade.
  - Remittances are "around 40 percent of GDP."

### Main recommendations and areas for further work
- Ensure tools are used regularly and steady progress continues after TA closure.
- Address macroeconomic, organizational, and capacity challenges:
  - Expand the NRBT forecasting team: current team described as "still too small to sustainably operate a fully integrated, modern FPAS."
  - Avoid capacity losses due to staff turnover; emphasize documentation and backups.
  - Allocate more staff time to practice model building and fine-tuning.
- Further improve development and communication of alternative scenarios to reflect a "shock-prone environment"; this requires additional time and resources.
- Streamline external communications to anchor market expectations, improve transparency, and build institutional credibility for forward-looking policymaking.

### Institutional roles, processes, and risks
- FPAS roles identified within NRBT:
  - Coordinator (and backup): leads forecasting team and communicates with management.
  - Model operators: maintain, use and develop forecasting tools and QPM.
  - Database managers: ensure timely and error-proof data environment.
  - Sectoral experts: real sector, inflation, fiscal, monetary, exchange rate, external, etc.
- Forecasting and policy workflow:
  - Formal internal forecast calendar tied to Monetary Policy statement releases.
  - Board of Directors normally meets "once a month."
  - Forecasts presented at least "twice a year" for Biannual Monetary Policy Statements; partial forecasts for inflation and foreign reserves produced monthly.
  - Monetary Policy decision (MPD) produced monthly followed by a press release; MPD publicly available on NRBT website.
- Identified sustainability risks and mitigation measures:
  - Team size is small; recommendations include internal mobility and new hires.
  - Staff involved in non-FPAS work; consider formal (close to) full-time FPAS assignments.
  - Risk of excessive turnover: need for backup staff, documentation (internal manuals and user guides), and staff rotation to broaden skills.
  - Key person risks: ensure analytical knowledge and skills are spread broadly and backups exist.

*IMF | Technical Report – Tonga TA on Developing FPAS at the NRBT | Preface (content unit)*

### Section II. Overview of TA missions

### A. Mission objectives and scoping (Mission 1: NOVEMBER 2021, VIRTUAL)
- Objectives:
  - Understand characteristics and components of the monetary policy framework (MPF) at NRBT.
  - Understand current NRBT policy decision-making processes.
  - Review NRBT’s existing tools and methodologies and identify complementarities with prospective FPAS TA tools.
  - Understand interactions between NRBT, MOF and the Tonga Statistics Department.
  - Provide NRBT Management and staff with an overview of the FPAS, and related ICD capacity development (CD).
  - Prepare a work program for a multi-year Macroeconomic Frameworks FPAS TA with a near-term focus on training and capacity development.
- Scoping findings:
  - NRBT is a small institution with limited resources in staff and available software.
  - Capacity constraints at NRBT are a challenge for TA implementation.
  - Authorities have weekly, monthly and quarterly economic indicators but face challenges in the compilation of data.
- Proposed implementation design:
  - Stream 1: tailored training and development of tools/techniques to improve analytical capacity at NRBT.
  - Stream 2: development of a Quarterly Projection Model (QPM) for medium-term analysis and forecasting, integration of macro analysis and forecasting into decision-making, and streamlining external communications.

### B. Capacity building in macroeconomic forecasting and nowcasting (Missions 2–6: MAY 2022–FEBRUARY 2023)
- Mission 2: MAY 2022 (VIRTUAL)
  - Focus: time series analysis and macro-econometric forecasting techniques; customized technical support and training.
  - Note: originally planned for February 2022; postponed due to the volcanic eruptions and resulting tsunami on January 15 that caused significant damage to public infrastructure.
- Mission 3: JULY 2022 (VIRTUAL)
  - Training: second part of the Macro-econometric Forecasting and Analysis (MFA) course (vector auto-regression, forecasting and forecast evaluation).
  - Hands-on EViews exercises; staff encouraged to catalogue Tongan data in EViews.
- Mission 4: OCTOBER 2022 (VIRTUAL)
  - NRBT staff formulated and estimated inflation and real GDP growth models using basic regressions, ARIMA models, and vector auto-regressions.
  - Forecasts compared with NRBT’s Excel-based moving average projections and evaluated with forecast evaluation techniques.
- Mission 5: NOVEMBER 2022 (VIRTUAL)
  - First of two missions on nowcasting; introduced nowcasting models and mixed-frequency estimators (BRIDGE, MIDAS, U-MIDAS).
  - Provided EViews command-line scripts and hands-on workshops; illustrative example used Vietnam nowcasting regression.
- Mission 6: FEBRUARY 2023 (IN-PERSON)
  - Applied nowcasting tools to Tongan data using BRIDGE approach for mixed frequencies.
  - Constructed GDPNow and InflationNow reports for Tonga (modeled on Federal Reserve Bank of Atlanta GDPNow) to confirm sources of growth and nowcast inflation.
  - These reports began feeding into NRBT internal policy discussions.

### C. Introducing Macroeconomic Diagnostics and the canonical QPM (Missions 7–9: SEPTEMBER 2023–JANUARY 2024)
- Mission 7: SEPTEMBER 2023 (VIRTUAL)
  - Introduced Macroeconomic Diagnostics tools and the canonical Quarterly Projection Model (QPM).
  - Covered potential output and output gap estimation methods (linear trend, HP filter, Bry-Boschan dating algorithm, production function), business cycle analysis, inflation targeting issues, Taylor rules.
  - Hands-on Excel workshops; canonical QPM introduced in general terms.
- Mission 8: NOVEMBER 2023 (IN-PERSON)
  - Continued Macroeconomic Diagnostics and QPM study.
  - Covered macro-financial linkages, financial soundness indicators, fiscal sector and sustainability, external sector and external sustainability.
  - NRBT staff tasked to apply MDS modules using Tongan data to produce an “IMF-style policy note.”
  - Identified need for a structured data management system outside of Excel; practical EViews demonstrations planned for January 2024.
- Mission 9: JANUARY 2024 (IN-PERSON)
  - Theoretical presentation of canonical QPM’s four building blocks: aggregate demand, inflation, uncovered interest parity, and a Taylor rule.
  - Illustrated QPM in EViews and discussed adaptation to Tonga (not an inflation targeting country).
  - Updated GDPNow and InflationNow reports; discussions on prospective public release.
  - Began implementation of a structured data management system in EViews tied to Excel.
  - Governor’s briefing: Governor emphasized enthusiastic support for macroeconomic assessment work and core macro model development.

### D. Customizing and calibrating the QPM; updating nowcasts and data systems (Missions 10–11: JUNE 2024–JANUARY 2025)
- Mission 10: JUNE 2024 (IN-PERSON)
  - Developed a customized QPM for Tonga on an EViews-based platform; reviewed canonical QPM theory and adaptation given Tonga’s non-inflation-targeting policy and reliance on exchange rate interventions.
  - Highlighted constraints: limited monetary policy transmission due to small economy size, openness to trade, low financial market development, heavy reliance on remittances.
  - Implemented exogenous shocks in the model and assessed impacts on key macro variables.
  - Continued updates to GDPNow and InflationNow reports and macroeconomic assessment; ongoing support for structured time-series data management in EViews.
- Mission 11: JANUARY 2025 (IN PERSON) — Final mission
  - Revised QPM and used it to examine impacts of various exogenous shocks.
  - Updated macroeconomic assessment and nowcasting reports; reviewed progress on EViews-based structured time-series data management.
  - Scenario analysis after final calibration:
    - Scenario 1: potential trade war between the United States and China.
    - Scenario 2: domestic riot (political instability) resulting in significant destruction of the capital stock, mimicking a domestic riot in Tonga in 2006 that destroyed 60 percent of the business district.
    - Scenario 3: unexpected decline in remittances and foreign aid to Tonga.
  - Nowcasting inputs and composition:
    - GDPNow high frequency indicators: travel receipts, construction permits, total agriculture, and currency in circulation.
    - InflationNow high frequency indicators: New Zealand dairy, New Zealand construction, oil, NEER, domestic and imported food prices.
  - Developed additional nowcasting models:
    - Imports model: import payments (Overseas Exchange Transaction (OET) data), container registrations, exchange rates.
    - Non-seasonally adjusted remittances nowcast: travel receipts (OET data), private transfers, social benefits, compensation of employees, crisis dummy, seasonal dummies.
  - NRBT staff outcomes:
    - Staff displayed confidence producing more elaborate reports for senior management.
    - Regular collection and updating of high frequency monthly Tongan data to nowcast annual/semi-annual real GDP growth and annual inflation.
    - Extensive scripting practice in EViews command language to replicate and document nowcast production.
  - Senior management endorsement:
    - Concluding meeting chaired by Governor Mr. Tatafu Moeaki; presentations by Ms Mefilina Tohi (Chief Manager of Economics), Mr. Daniel Taumoepeau (Assistant Governor (Policy)), and Ms. Ungatea Latu (Deputy Governor).
    - Governor and Deputy Governor emphasized ongoing enthusiastic support for continued production and use of Tonga’s macroeconomic assessment, nowcasting reports, and core macro model for medium-term policy analysis and forecasting.

### E. Key analytical and institutional findings across missions
- Data and systems:
  - Persistent need to move from Excel-based processes to a structured time-series data management system (implemented in EViews).
- Monetary policy and transmission in Tonga:
  - Tonga is not an inflation targeting country; monetary policy mainly conducted through exchange rate interventions.
  - Monetary policy transmission is limited by small economy size, openness to international trade, low financial market development, and heavy reliance on remittances.
  - Excess liquidity in the banking system has increased to over 30 percent of GDP, rendering monetary policy transmission ineffective.
  - Interest rate on excess reserves has been constant at zero percent (since 2012).
  - Reserve Bank has not issued any notes (or conducted open market operations) since 2009.
  - Model reference: Taylor rule uses a reference rate of 5 percent (assumed as a target in the model).
- Disaster and shock vulnerability:
  - Tonga experienced a volcanic eruption and tsunami on January 15, 2022, causing significant infrastructure damage.
  - Tonga has experienced a natural disaster every 2 years since 2014.
  - Historical domestic riot example: 2006 event that destroyed 60 percent of the business district.

### F. Outputs, modeling approaches and forthcoming work
- Nowcasting methodology and limitations:
  - Nowcasting provides timely forecasts using high-frequency data but standard reduced-form nowcasting estimators cannot assess the impact of policy changes on baseline nowcasts.
  - IMF Working Paper WP/23/153 (“Assessing the Impact of Policy Changes on a Nowcast”, July 2023) outlines two methodologies to address this:
    - Partial equilibrium approach using baseline nowcasting regression and single-equation ARMA(p,q) forecasting models for high-frequency data.
    - Non-parametric structural VAR estimator (Ouliaris and Pagan (2022) approach) illustrated with Tongan data.
- Forthcoming working paper:
  - “A Quarterly Projection Model for Tonga” co-authored by the ICD mission team and NRBT Assistant Governor (policy).
  - Calibrates a gap model for Tonga to generate medium-term quarterly projections of output, inflation, interest rate, and exchange rate.
  - Model key calibration features:
    - Aggregate demand: moderate persistence in the output gap; monetary condition index assigns more weight to the exchange rate gap than the interest rate gap; relatively low policy passthrough; low impact of external demand on domestic output.
    - Aggregate supply: high persistence of core inflation; low persistence of domestic food and energy prices; nominal wage stickiness and price stickiness considered.
    - External price stability defined by exchange rate stability; exchange rate arrangement serves as the nominal anchor and responds slowly to external shocks; Reserve Bank can adjust the peg to address misalignments.
    - Monetary policy: tied to exchange rate stability; Taylor rule displays strong persistence to domestic interest rates, low persistence to output and inflation gaps relative to a reference rate of 5 percent; interest rates are not sensitive to world interest rate movements.
  - Scenario analysis in the calibrated QPM:
    - Scenario set 1: shocks proxying for a bank failure (reflecting financial sector fragility and rising non-performing loans).
    - Scenario set 2: shocks capturing consequences of a natural disaster.
    - Scenario set 3: shocks representing a significant negative external shock to the Tongan economy.
  - Rationale: scenarios reflect sensitivity to adverse financial shocks, trading partners’ macroeconomic policies, and extreme weather events.

*Italic: Section II. Overview of TA missions — tarea2025099-source-pdf*

### Section IV. Aide-Mémoire (Post January 2025 Mission)

### Summary of training outcomes and main outputs
- Training and tools delivered between 2022 and 2025 (ten missions in total) led to development of 4 main outputs:
  - Nowcasting reports for GDP and Inflation.
  - Macroeconomic review.
  - Quarterly Projection Model (QPM) for Tonga.
  - Structural time-series data management systems.

### Nowcasting reports for GDP and Inflation
- Findings and recommendations:
  - These reports should be revised monthly and become key inputs in communication for upper management.
  - These reports could also be published on the NRBT website for broader dissemination.

### Macroeconomic review
- Findings and recommendations:
  - Analysis of vulnerabilities and risks, and associated recommendations extend beyond existing NRBT reports provided to management.
  - This new report, together with current monthly staff reports, could enrich discussions on intersectoral linkages, thereby supporting monetary policy formulation.
  - The Macroeconomic review should be revised at least quarterly and become a key input in the monetary policy meetings.

### Quarterly Projection Model (QPM) for Tonga
- Capabilities and uses:
  - The model allows for the assessment of various exogenous shocks on key macroeconomic variables over the medium term.
- Implementation and maintenance requirements:
  - The calibration of the model will need to be assessed and refined on a regular basis, depending on the acceptability of the baseline projections and forecasting performance.
  - Consistency across sectors will be key to this exercise.
  - Input data for the QPM needs to be maintained and updated as new Tongan data is released.
  - Staff should continue to use on a regular basis the EViews implementation of QPM to remain proficient in solving the model and interpreting the outcomes with and without shocks.

### Structural time-series data management systems
- Findings and recommendations:
  - Such systems are critical for the conduct of sound economic policy.
  - They will improve the efficiency of NRBT staff in preparing various high-quality notes as requested by management, compared to the use of an error prone and inefficient Excel based data system.

### Staff preparedness and communication tasks (in response to management questions)
- Staff should be ready to:
  - Formulate appropriate exogenous shocks to the baseline using the QPM methodology and interpret the results proficiently.
  - Prepare briefs and presentations for upper management addressing the questions asked.
  - Prepare notes to be published, conditional on upper management approval, on the NBRT website for external audiences, to promote communication outside of NRBT, to foster financial inclusion and financial literacy among the public, and the credibility of NRBT policy making in the public domain.

*IMF | Technical Report – Tonga TA on Developing FPAS at the NRBT | Section IV. Aide-Mémoire (Post January 2025 Mission)*

### Preface ................................................................................................................

### Preface

### Project scope and timeline
- At the request of the National Reserve Bank of Tonga (NRBT), a Macroeconomic Frameworks Technical Assistance (TA) Project on developing a Forecasting and Policy Analysis System (FPAS) for NRBT was conducted by ICD during 2021-25.
- Preface statement: "A total of eleven missions (five in -person and six virtual) were conducted during 2021-25."
- Executive Summary statement: "Following a scoping mission in November 2021, a total of ten missions (five in -person and five virtual) were conducted."
- The mission team consisted of Celine Rochon (ICDFP, Project manager, who participated in all TA missions) and Sam Ouliaris (short-term expert, who participated in all TA missions and reviewed this TA report).
- During in-person missions the TA team met regularly with the Governor Mr. Tatafu Moeaki and the Assistant Governor for Policy, Mr. Daniel Taumoepeau.
- The mission team acknowledges the generous support of the Government of Japan under the Japanese Sub-Account (JSA) funding.

### Core objectives and workstreams
- Two broad and complementary workstreams:
  - i. Developing analytical tools for macroeconomic assessment and forecasting among NRBT staff.
  - ii. Integration of the resulting analytical reports into NRBT’s policy processes and external communications.
- The project aimed to build new institutional capacity for model-based forecasting and policy analysis, streamline decision-making processes, and strengthen forward-looking monetary policy communication.

### Key deliverables and tools
- Main analytical tool emphasized: the Quarterly Projection Model (QPM) as the core operational model for medium-term projections.
- Nowcasting and Near-Term Forecast (NTF) tools developed, including:
  - Simple extrapolations of past trends.
  - Univariate methods (Autoregressive Integrated Moving Average (ARIMA)).
  - Bridge equations linking GDP to monthly/daily indicators.
- Data-management work: cataloguing and saving available Tongan time series data in EViews.

### Project achievements (Executive Summary)
- The FPAS TA project has largely achieved its objectives.
- NRBT forecasting team progress:
  - "remarkable progress" and "regularly delivering high-level analytical support to the policymakers."
  - Staff can use a "rich suite of analytical tools" and follow the internal forecast calendar to prepare model-based results and forecasts for key macroeconomic variables.
  - "Operational use of the core Quarterly Projections Model (QPM) in producing baseline forecasts, alternative scenarios, using expert judgements, assembling the forecast narrative and delivery to peers is performed efficiently and following best central bank practice."
- Analytical work plays a key role in NRBT policy processes and forward-looking monetary policy formulation.
  - "Model-based results are communicated directly to the Governor."
  - "Forecasts feature, to some extent in external communications, including Monetary Policy Reports."

### Key macroeconomic facts about Tonga (as stated)
- Inflation reference rate: "currently of 5 percent."
- Foreign reserves objectives: minimum of "3 months" of imports and optimally around "7.5 months" of imports.
- Exchange rate flexibility: NRBT has flexibility "to adjust the exchange rates by 5 percent each month if necessary."
- Trade and external balances:
  - Trade deficit averaged "about 50 percent of GDP" in the past decade.
  - Remittances are "around 40 percent of GDP."

### Main recommendations and areas for further work
- Ensure tools are used regularly and steady progress continues after TA closure.
- Address macroeconomic, organizational, and capacity challenges:
  - Expand the NRBT forecasting team: current team described as "still too small to sustainably operate a fully integrated, modern FPAS."
  - Avoid capacity losses due to staff turnover; emphasize documentation and backups.
  - Allocate more staff time to practice model building and fine-tuning.
- Further improve development and communication of alternative scenarios to reflect a "shock-prone environment"; this requires additional time and resources.
- Streamline external communications to anchor market expectations, improve transparency, and build institutional credibility for forward-looking policymaking.

### Institutional roles, processes, and risks
- FPAS roles identified within NRBT:
  - Coordinator (and backup): leads forecasting team and communicates with management.
  - Model operators: maintain, use and develop forecasting tools and QPM.
  - Database managers: ensure timely and error-proof data environment.
  - Sectoral experts: real sector, inflation, fiscal, monetary, exchange rate, external, etc.
- Forecasting and policy workflow:
  - Formal internal forecast calendar tied to Monetary Policy statement releases.
  - Board of Directors normally meets "once a month."
  - Forecasts presented at least "twice a year" for Biannual Monetary Policy Statements; partial forecasts for inflation and foreign reserves produced monthly.
  - Monetary Policy decision (MPD) produced monthly followed by a press release; MPD publicly available on NRBT website.
- Identified sustainability risks and mitigation measures:
  - Team size is small; recommendations include internal mobility and new hires.
  - Staff involved in non-FPAS work; consider formal (close to) full-time FPAS assignments.
  - Risk of excessive turnover: need for backup staff, documentation (internal manuals and user guides), and staff rotation to broaden skills.
  - Key person risks: ensure analytical knowledge and skills are spread broadly and backups exist.

*IMF | Technical Report – Tonga TA on Developing FPAS at the NRBT | Preface (content unit)*

### Section II.   Overview of TA missions

### Section II.   Overview of TA missions

### A. Mission objectives and scoping (Mission 1: NOVEMBER 2021, VIRTUAL)
- Objectives:
  - Understand characteristics and components of the monetary policy framework (MPF) at NRBT.
  - Understand current NRBT policy decision-making processes.
  - Review NRBT’s existing tools and methodologies and identify complementarities with prospective FPAS TA tools.
  - Understand interactions between NRBT, MOF and the Tonga Statistics Department.
  - Provide NRBT Management and staff with an overview of the FPAS, and related ICD capacity development (CD).
  - Prepare a work program for a multi-year Macroeconomic Frameworks FPAS TA with a near-term focus on training and capacity development.
- Scoping findings:
  - NRBT is a small institution with limited resources in staff and available software.
  - Capacity constraints at NRBT are a challenge for TA implementation.
  - Authorities have weekly, monthly and quarterly economic indicators but face challenges in the compilation of data.
- Proposed implementation design:
  - Stream 1: tailored training and development of tools/techniques to improve analytical capacity at NRBT.
  - Stream 2: development of a Quarterly Projection Model (QPM) for medium-term analysis and forecasting, integration of macro analysis and forecasting into decision-making, and streamlining external communications.

### B. Capacity building in macroeconomic forecasting and nowcasting (Missions 2–6: MAY 2022–FEBRUARY 2023)
- Mission 2: MAY 2022 (VIRTUAL)
  - Focus: time series analysis and macro-econometric forecasting techniques; customized technical support and training.
  - Note: originally planned for February 2022; postponed due to the volcanic eruptions and resulting tsunami on January 15 that caused significant damage to public infrastructure.
- Mission 3: JULY 2022 (VIRTUAL)
  - Training: second part of the Macro-econometric Forecasting and Analysis (MFA) course (vector auto-regression, forecasting and forecast evaluation).
  - Hands-on EViews exercises; staff encouraged to catalogue Tongan data in EViews.
- Mission 4: OCTOBER 2022 (VIRTUAL)
  - NRBT staff formulated and estimated inflation and real GDP growth models using basic regressions, ARIMA models, and vector auto-regressions.
  - Forecasts compared with NRBT’s Excel-based moving average projections and evaluated with forecast evaluation techniques.
- Mission 5: NOVEMBER 2022 (VIRTUAL)
  - First of two missions on nowcasting; introduced nowcasting models and mixed-frequency estimators (BRIDGE, MIDAS, U-MIDAS).
  - Provided EViews command-line scripts and hands-on workshops; illustrative example used Vietnam nowcasting regression.
- Mission 6: FEBRUARY 2023 (IN-PERSON)
  - Applied nowcasting tools to Tongan data using BRIDGE approach for mixed frequencies.
  - Constructed GDPNow and InflationNow reports for Tonga (modeled on Federal Reserve Bank of Atlanta GDPNow) to confirm sources of growth and nowcast inflation.
  - These reports began feeding into NRBT internal policy discussions.

### C. Introducing Macroeconomic Diagnostics and the canonical QPM (Missions 7–9: SEPTEMBER 2023–JANUARY 2024)
- Mission 7: SEPTEMBER 2023 (VIRTUAL)
  - Introduced Macroeconomic Diagnostics tools and the canonical Quarterly Projection Model (QPM).
  - Covered potential output and output gap estimation methods (linear trend, HP filter, Bry-Boschan dating algorithm, production function), business cycle analysis, inflation targeting issues, Taylor rules.
  - Hands-on Excel workshops; canonical QPM introduced in general terms.
- Mission 8: NOVEMBER 2023 (IN-PERSON)
  - Continued Macroeconomic Diagnostics and QPM study.
  - Covered macro-financial linkages, financial soundness indicators, fiscal sector and sustainability, external sector and external sustainability.
  - NRBT staff tasked to apply MDS modules using Tongan data to produce an “IMF-style policy note.”
  - Identified need for a structured data management system outside of Excel; practical EViews demonstrations planned for January 2024.
- Mission 9: JANUARY 2024 (IN-PERSON)
  - Theoretical presentation of canonical QPM’s four building blocks: aggregate demand, inflation, uncovered interest parity, and a Taylor rule.
  - Illustrated QPM in EViews and discussed adaptation to Tonga (not an inflation targeting country).
  - Updated GDPNow and InflationNow reports; discussions on prospective public release.
  - Began implementation of a structured data management system in EViews tied to Excel.
  - Governor’s briefing: Governor emphasized enthusiastic support for macroeconomic assessment work and core macro model development.

### D. Customizing and calibrating the QPM; updating nowcasts and data systems (Missions 10–11: JUNE 2024–JANUARY 2025)
- Mission 10: JUNE 2024 (IN-PERSON)
  - Developed a customized QPM for Tonga on an EViews-based platform; reviewed canonical QPM theory and adaptation given Tonga’s non-inflation-targeting policy and reliance on exchange rate interventions.
  - Highlighted constraints: limited monetary policy transmission due to small economy size, openness to trade, low financial market development, heavy reliance on remittances.
  - Implemented exogenous shocks in the model and assessed impacts on key macro variables.
  - Continued updates to GDPNow and InflationNow reports and macroeconomic assessment; ongoing support for structured time-series data management in EViews.
- Mission 11: JANUARY 2025 (IN PERSON) — Final mission
  - Revised QPM and used it to examine impacts of various exogenous shocks.
  - Updated macroeconomic assessment and nowcasting reports; reviewed progress on EViews-based structured time-series data management.
  - Scenario analysis after final calibration:
    - Scenario 1: potential trade war between the United States and China.
    - Scenario 2: domestic riot (political instability) resulting in significant destruction of the capital stock, mimicking a domestic riot in Tonga in 2006 that destroyed 60 percent of the business district.
    - Scenario 3: unexpected decline in remittances and foreign aid to Tonga.
  - Nowcasting inputs and composition:
    - GDPNow high frequency indicators: travel receipts, construction permits, total agriculture, and currency in circulation.
    - InflationNow high frequency indicators: New Zealand dairy, New Zealand construction, oil, NEER, domestic and imported food prices.
  - Developed additional nowcasting models:
    - Imports model: import payments (Overseas Exchange Transaction (OET) data), container registrations, exchange rates.
    - Non-seasonally adjusted remittances nowcast: travel receipts (OET data), private transfers, social benefits, compensation of employees, crisis dummy, seasonal dummies.
  - NRBT staff outcomes:
    - Staff displayed confidence producing more elaborate reports for senior management.
    - Regular collection and updating of high frequency monthly Tongan data to nowcast annual/semi-annual real GDP growth and annual inflation.
    - Extensive scripting practice in EViews command language to replicate and document nowcast production.
  - Senior management endorsement:
    - Concluding meeting chaired by Governor Mr. Tatafu Moeaki; presentations by Ms Mefilina Tohi (Chief Manager of Economics), Mr. Daniel Taumoepeau (Assistant Governor (Policy)), and Ms. Ungatea Latu (Deputy Governor).
    - Governor and Deputy Governor emphasized ongoing enthusiastic support for continued production and use of Tonga’s macroeconomic assessment, nowcasting reports, and core macro model for medium-term policy analysis and forecasting.

### E. Key analytical and institutional findings across missions
- Data and systems:
  - Persistent need to move from Excel-based processes to a structured time-series data management system (implemented in EViews).
- Monetary policy and transmission in Tonga:
  - Tonga is not an inflation targeting country; monetary policy mainly conducted through exchange rate interventions.
  - Monetary policy transmission is limited by small economy size, openness to international trade, low financial market development, and heavy reliance on remittances.
  - Excess liquidity in the banking system has increased to over 30 percent of GDP, rendering monetary policy transmission ineffective.
  - Interest rate on excess reserves has been constant at zero percent (since 2012).
  - Reserve Bank has not issued any notes (or conducted open market operations) since 2009.
  - Model reference: Taylor rule uses a reference rate of 5 percent (assumed as a target in the model).
- Disaster and shock vulnerability:
  - Tonga experienced a volcanic eruption and tsunami on January 15, 2022, causing significant infrastructure damage.
  - Tonga has experienced a natural disaster every 2 years since 2014.
  - Historical domestic riot example: 2006 event that destroyed 60 percent of the business district.

### F. Outputs, modeling approaches and forthcoming work
- Nowcasting methodology and limitations:
  - Nowcasting provides timely forecasts using high-frequency data but standard reduced-form nowcasting estimators cannot assess the impact of policy changes on baseline nowcasts.
  - IMF Working Paper WP/23/153 (“Assessing the Impact of Policy Changes on a Nowcast”, July 2023) outlines two methodologies to address this:
    - Partial equilibrium approach using baseline nowcasting regression and single-equation ARMA(p,q) forecasting models for high-frequency data.
    - Non-parametric structural VAR estimator (Ouliaris and Pagan (2022) approach) illustrated with Tongan data.
- Forthcoming working paper:
  - “A Quarterly Projection Model for Tonga” co-authored by the ICD mission team and NRBT Assistant Governor (policy).
  - Calibrates a gap model for Tonga to generate medium-term quarterly projections of output, inflation, interest rate, and exchange rate.
  - Model key calibration features:
    - Aggregate demand: moderate persistence in the output gap; monetary condition index assigns more weight to the exchange rate gap than the interest rate gap; relatively low policy passthrough; low impact of external demand on domestic output.
    - Aggregate supply: high persistence of core inflation; low persistence of domestic food and energy prices; nominal wage stickiness and price stickiness considered.
    - External price stability defined by exchange rate stability; exchange rate arrangement serves as the nominal anchor and responds slowly to external shocks; Reserve Bank can adjust the peg to address misalignments.
    - Monetary policy: tied to exchange rate stability; Taylor rule displays strong persistence to domestic interest rates, low persistence to output and inflation gaps relative to a reference rate of 5 percent; interest rates are not sensitive to world interest rate movements.
  - Scenario analysis in the calibrated QPM:
    - Scenario set 1: shocks proxying for a bank failure (reflecting financial sector fragility and rising non-performing loans).
    - Scenario set 2: shocks capturing consequences of a natural disaster.
    - Scenario set 3: shocks representing a significant negative external shock to the Tongan economy.
  - Rationale: scenarios reflect sensitivity to adverse financial shocks, trading partners’ macroeconomic policies, and extreme weather events.

*Italic: Section II. Overview of TA missions — tarea2025099-source-pdf*

### Section IV. Aide-Mémoire (Post January 2025 Mission)

### Section IV. Aide-Mémoire (Post January 2025 Mission)

### Summary of training outcomes and main outputs
- The training and tools in macro-forecasting, nowcasting, macro-diagnostics, and quarterly projection models delivered between 2022 and 2025 (ten missions in total) led to the development of 4 main outputs:
  - Nowcasting reports for GDP and Inflation.
  - Macroeconomic review.
  - Quarterly Projection Model (QPM) for Tonga.
  - Structural time-series data management systems.

### Nowcasting reports for GDP and Inflation
- Findings and recommendations:
  - These reports should be revised monthly and become key inputs in communication for upper management.
  - These reports could also be published on the NRBT website for broader dissemination.

### Macroeconomic review
- Findings and recommendations:
  - The analysis of vulnerabilities and risks, and associated recommendations included in this report extend beyond the existing NRBT reports provided to management.
  - This new report, together with the current monthly staff reports, could enrich discussions on intersectoral linkages, thereby supporting monetary policy formulation.
  - The Macroeconomic review should be revised at least quarterly and become a key input in the monetary policy meetings.

### Quarterly Projection Model (QPM) for Tonga
- Capabilities and uses:
  - The model allows for the assessment of various exogenous shocks on key macroeconomic variables over the medium term.
- Implementation and maintenance requirements:
  - The calibration of the model will need to be assessed and refined on a regular basis, depending on the acceptability of the baseline projections and forecasting performance.
  - Consistency across sectors will be key to this exercise.
  - For this to be done successfully, the input data for the QPM needs to be maintained and updated as new Tongan data is released.
  - Staff should continue to use on a regular basis the EViews implementation of QPM to remain proficient in solving the model and interpreting the outcomes with and without shocks.

### Structural time-series data management systems
- Findings and recommendations:
  - Such systems are critical for the conduct of sound economic policy.
  - They will improve the efficiency of NRBT staff in preparing various high-quality notes as requested by management, compared to the use of an error prone and inefficient Excel based data system.

### Staff preparedness and communication tasks (in response to management questions)
- Staff should be ready to:
  - Formulate appropriate exogenous shocks to the baseline using the QPM methodology and interpret the results proficiently.
  - Prepare briefs and presentations for upper management addressing the questions asked.
  - Prepare notes to be published, conditional on upper management approval, on the NBRT website for external audiences, to promote communication outside of NRBT, to foster financial inclusion and financial literacy among the public, and the credibility of NRBT policy making in the public domain.

### References
- Berg, A., Karam, P. D. & Laxton, D. (2006). A Practical Model-Based Approach to Monetary Policy Analysis – Overview. IMF Working Papers 06/80.
- Ouliaris, S., & Rochon, C. (2023). Assessing the Impact of Policy Changes on a Nowcast. IMF Working Paper 23/153.
- Ouliaris, S., Rochon, C. & Taumoepeau, D. (2025). A Quarterly Projection Model for Tonga. Forthcoming IMF Working Paper.

*IMF | Technical Report – Tonga TA on Developing FPAS at the NRBT | Section IV. Aide-Mémoire (Post January 2025 Mission)*

---


_Source: https://www.imf.org/-/media/files/publications/tar/2025/english/tarea2025099-source-pdf.pdf_