## PREFACE

## Source details

**Canonical URL:** [PREFACE](https://www.imf.org/-/media/files/publications/tar/2025/english/tarea2025020-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/tar/2025/english/tarea2025020-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/tar/2025/english/tarea2025020-print-pdf.pdf.json)

---

### Mission purpose and meetings
- A Monetary and Capital Markets (MCM) Department mission visited Luanda, Angola from May 20 to May 24, 2024, at the request of the Banco Nacional de Angola (BNA).
- Mission objectives:
  - review BNA’s Forecasting and Policy Analysis System (FPAS);
  - propose enhancements to modeling and forecasting procedures;
  - introduce fiscal channels into the core forecasting model to better characterize Angolan economy dynamics.
- High-level meetings included Vice Governor Ms. Maria Juliana de Carvalho Van-Dúnem de Fontes Pereira, Mr. Sebastião Kivampolo Tuma (DEE Director), Ms. Martine Emma Dias dos Santos (Deputy Director of DEE), other senior BNA officials, and IMF AFRITAC South representatives.
- The mission worked daily with DEE experts across divisions.

### Executive summary — Key assessments
- Updated assessment of FPAS operations at BNA and TA mission objectives to strengthen analytical capacities for transition to inflation targeting.
- Progress noted: BNA has made impressive progress in recent years improving its FPAS.
- Mission activities: reviewed FPAS components; assisted BNA staff in introducing fiscal channels into the QPM; evaluated aspects of policymaking.
- Final assessment: current stage of FPAS at BNA partially meets environmental conditions to operate under inflation targeting; further progress required in use of FPAS models internally and in monetary policy communication.
- Risks and constraints: high turnover, small teams, scarcity of senior profiles pose implementation risks for recommendations.

### Mission scope and prior TA follow-up
- Mission purposes:
  - i) Review operation of QPM and follow up on previous TA recommendations;
  - ii) Assist staff to build a fiscal block in QPM to account for fiscal channels for story-telling and short-term forecasting;
  - iii) Assess adequacy of BNA resources to operate core model to produce medium-term forecasts;
  - iv) Discuss potential topics for further TA.
- 2022 TA achievements:
  - i) tracking and documenting forecast performance continuously;
  - ii) using core projection model to actively generate interest rate paths;
  - iii) fully supporting forward-looking monetary policy.
- Prior recommended changes (2022): i) make the Taylor rule active; ii) impose a UIP condition instead of an adjustment equation for the nominal exchange rate.

### Summary of FPAS operational findings
- Organizational and process:
  - DEE houses all staff involved in forecasting; workflow and communication improved.
  - Formal calendar for forecasting cycle in place; each forecasting cycle takes six weeks in the run-up to the MPC meeting.
  - Lack of an all-together meeting to present a tentative baseline scenario; MPC members do not participate in conjunctural development meetings.
- Data and modeling toolkit:
  - Relevant data inputs organized by department; databases and data workflows are still poorly integrated and lack flexibility for ad-hoc investigation.
  - Nowcasting and near-term forecasting models for selected variables are operational.
  - Core model tailored to Angola: captures non-official exchange rate dynamics, oil price effects on activity and inflation, and exposure to global food price cycles.
- Current QPM architecture and practice:
  - QPM used to support MPC decisions is based on monetary aggregates with a money block (demand and supply) operated by the central bank.
  - Real interest rate is endogenously calculated by filtering, given observable Luibor rate (Luanda Interbank Offered Rate).
  - Taylor rule is muted; official exchange rate is set to gradually reduce spread to the non-official exchange rate, affected by gaps of real exchange rate and real interest rate.
  - QPM produces internally consistent baseline and alternative scenario material with a 6-quarter horizon for main variables: output and sectoral decomposition; consumer inflation and components; monetary base and M2; official and non-official exchange rates; interbank interest rate.
  - Projections rely significantly on filtered gaps (non-oil output gap, real interest rate gap, real exchange rate gap) and on judgements that may be disconnected from reality without explicit protocol.
  - BNA undisclosed and changeable inflation objectives are used as inputs; model used to calculate money supply paths consistent with exogenous inflation paths, making model act as a calculator.

### Model performance and transmission
- Current QPM calibrated to generate a small response of inflation to a money supply shock; IRF of money supply shock shows model-filtered interest rate equivalence and weak transmission to inflation.
- BNA staff assess difficulty of stabilizing economy in short term via demand channel; non-oil output may be little affected by rising interest rates since private/public investments are driven by expectations for oil revenues, exchange rate, and fiscal results.
- Banking channels currently have limited role.

### Follow-up on 2022 TA recommendations
- Changes proposed in 2022 (activate Taylor rule, impose UIP) were produced but not deployed operationally.
- Staff implemented a version with those equations; some minor issues detected and solved during mission; forecasting exercise tried with those assumptions.
- Staff hesitation to switch to new version due to:
  - i) need to forecast and report money aggregates to MPC during transition;
  - ii) view that BNA interest rate cannot yet be taken as market benchmark;
  - iii) reduced power of interest rates to control aggregate demand;
  - iv) short-term projections for exchange rate movements and consumer inflation appear excessive/unrealistic.
- Current practice of using shocks to generate medium-term profile has clear drawbacks:
  - Underlying shocks in medium run are always unanticipated and not determined from drivers of price formation.
  - Implied shocks lack explanation or counterpart; represent abstract disinflation events with no adjustment costs.
  - Money supply policy derived from this exercise is not forward-looking nor efficient; central bank is surprised by disinflation each period.

### Key recommendations (selected, with priority and timeframe)
- Modeling and Forecasting — High priority, Near-term (< 12 months)
  - Switch immediately to the model with the Taylor rule and uncovered interest rate parity (UIP) condition; if helpful, consider running the old version in parallel for a short period when new tunes are placed to align projections; do not make tunes on medium-run inflation. — Priority: High; Timeframe: Near-term
  - Extend the forecasting window to at least 24 months ahead; check key model gaps closing in the long term. — Priority: High; Timeframe: Near-term
  - Track and document forecast performance continuously to ensure effective narrative construction and support model maintenance. — Priority: 3; High; Timeframe: Near-term
  - Build empirical evidence on the power and dynamics of key macro shocks, such as interest rate, exchange rate, and fiscal impulse. — Priority: 3; High; Timeframe: Near-term
  - Produce sensitivity analysis for key parameters whenever necessary. — Priority: Medium; Timeframe: Near-term
  - Invest time to deepen comprehension of model codes; consider regular training in MATLABMATLAB and IRIS toolbox. — Priority: Medium; Timeframe: Near-term
  - Create additional analysis and decomposition tools of scenarios to improve understanding of what is driving forecasts. — Priority: Medium; Timeframe: Near-term
  - Improve data integration between departments; enhance information sharing through secure network folders. — Priority: Medium; Timeframe: Medium-term
  - Address human resource management challenges like high turnover and scarcity of senior profiles; consider increasing the number of staff with quantitative skills. — Priority: Medium; Timeframe: Medium-term
- Policy formulation process and communication — Mixed priority, Near- to Medium-term
  - Schedule regular, pre-planned meetings between different teams along the forecast cycle; ensure the discussions are prospective. — Priority: Medium; Timeframe: Near-term
  - Enhance communication with the public immediately with a more forward-looking approach; draft monetary policy documents with a prospective analysis. — Priority: 3; High; Timeframe: Near-term
  - Start budgeting interest rate paths and discuss policy trade-offs to support MPC decision2; adjust communication to include MPC's views on monetary policy stance and timing to convergence. — Priority: High; Timeframe: Near-term
  - Evaluate the quality of survey-based market expectations, particularly for inflation in medium-term horizons. — Priority: High; Timeframe: Medium-term
- Timing definitions:
  - Near term: < 12 months
  - Medium term: 12 to 24 months

*IMF Technical Assistance Report — PREFACE (content unit from the BNA TA mission materials)*

---

### Assessment of money supply shocks and monetary policy instruments
- FT tried to calibrate money supply shock so medium-term inflation would meet objectives, but the response of inflation is small.
- In the mission baseline, forecasts showed a strong immediate devaluation of Kwanza produced by the UIP condition, even with higher interest rates, causing projected inflation above that reported by the current model.
- Illustrative point: A decrease of 1 percent qoq (4 percent qoqar) in money supply generates an implied interest rate hike of around 55 bps.
- Recommendations:
  - Move towards more efficient monetary policy instruments; the current QPM version and procedures are no longer appropriate for advising policy.
  - Switch immediately to the new version of QPM with a Taylor rule and a UIP condition; if useful, run the two versions in parallel for a few months.
  - Staff must refrain from making judgments on inflation in the medium term; observe how inflation responds endogenously to policy alternatives.
  - FT should do more work to confirm by other means that the power of monetary policy over non-oil activity is indeed so low; identifying and estimating transmission elasticities is a continuous task.

### Protocols for judgments, scenario conditioning, and horizons
- FT should organize a protocol to make judgments in selected variables: define variables and the number of quarters for short-term conditioning to organize hypothesis discussions and track changes in judgments.
- Typical practice: create baseline scenarios with trajectories for volatile variables (exchange rates, oil prices, food prices) and then explore alternative paths.
- Recommendation to extend forecasting window to at least 24 months ahead and closely monitor medium-term inflation and its drivers.
- Check if key model gaps are closing in the long term; inflation targeting requires a more forward-looking approach and consideration of trade-offs between gradual or timely responses and the buildup of central bank credibility.

### Interest-rate budgeting and scenario analysis
- TA developed alternative code for budgeting interest rates to discuss alternative interest rate paths in terms of costs to economic activity and timing of convergence of inflation to objectives.
- New script might eliminate the need for hard tunes on medium-term inflation to meet inflation objectives.
- FT should extend codes to discuss budgeting interest rates under different conditioning paths for key variables like exchange rates and oil prices.
- Illustrative exercise: presents four alternative scenarios for adjusting the policy rate, evaluating tradeoffs in convergence and costs to economic activity of immediate action versus gradualism; the blue line is the BNA’s baseline.

### Money quantity satellite and forecast evaluation
- TA added a satellite specification based on the money quantity equation for reporting money growth consistent with endogenous variables to accelerate implementation of the version with Taylor rule and UIP.
- Having an endogenous projection for the quantity of money can be useful during the transition to inflation targeting and increase staff confidence in moving to the new QPM immediately.
- Mission reinforces tracking and documenting forecast performance; evaluate sources of forecast errors and their role in changing nowcasts and near-term forecasts.
- Tracking changes in model gaps can improve storytelling; additional TA can help design a transition map to account for changes between current and previous scenarios.

### Code, diagnostics, and model development enhancements
- TA updated QPM codes with warnings and errors; medium-term model largely based on MATLAB scripts coded in old software versions.
- FT works on a 2018a version of MATLAB; mission updated scripts to be compatible with any released version and with the newest 2023 IRIS Toolbox, removing use of the 2015 IRIS Toolbox.
- Recovered code for IRFs and recommended regular use to understand overall effects beyond immediate impact; used IRFs for alternative calibrations, notably fiscal block parameters.
- Recovered code for Bayesian estimations; ran script to locally estimate some fiscal block parameters.
- Introduced a graph in simulation code to highlight all judgments made in the scenario to help screen for errors or missing items.
- Recommendation: continuous and intensive training in MATLAB and IRIS toolbox; FT should invest more time to analyze scenario outputs and increase familiarity with the code.

### Data sharing, external review, and operational recommendations
- Improve data sharing and integration between areas; use secure network folders rather than email attachments to improve security, integrity, coordination, and control.
- An external review of FPAS operating during one forecasting cycle could help identify practical issues and areas for quick implementation improvements.
- FT should prioritize expanding knowledge about codes and models; in the short term focus on explaining contributions of conditioning variables to forecast changes between meetings rather than expanding model blocks.

### Human resources and infrastructure assessment
- FPAS vulnerabilities:
  - Core model forecasting activities performed by only three people.
  - No manual of usage and procedures for entire BNA forecasting process.
  - High turnover, loss of specialized knowledge, and scarcity of senior profiles threaten continuity.
- Recommendations:
  - Increase opportunities for internal training and knowledge multiplication; staff returning from external training should engage in practical forecasting activities to consolidate knowledge.
  - Resize and improve technological infrastructure: network stability, more MATLAB licenses, more computers ready to run scenarios so forecasting exercises can run in parallel.
  - Rotation of responsibilities within the group running QPM to enhance understanding of FT’s key messages.

### Preparations for inflation-targeting transition
- FPAS must be adapted to produce prospective analyses and narratives for MPC needs.
- FT needs to:
  - i) address quantitatively the contribution of scenario assumptions and the uncertainty;
  - ii) discuss prospective inflation and its drivers in different scenarios;
  - iii) prepare the MPC to deliver effective communication with the public.
- Development of new forward-looking FPAS products and decompositions is a priority over medium-term model expansion.

### Introducing fiscal channels into the QPM — rationale and focus
- BNA requested TA to introduce fiscal channels to improve storytelling and capture Treasury’s role in price determination, particularly in exchange rate markets.
- Fiscal channels expected to improve historical decomposition, support fiscal policy simulations, and infer required monetary policy reactions.
- Focus on fiscal result flows because debt is not a concern in the near term.

### Fiscal background and immediate focus (reported figures and targets)
- Angola debt exposures and composition (2022 figures as reported):
  - 79 percent of Angola's debt in 2022 is exposed to exchange rates.
  - 70 percent is foreign debt.
  - 21 percent corresponds to the overall internal debt denominated in kwanzas, with several indexation measures.
- Angola has a high and positive general government primary balance, and the current account balance is consistently positive at around 5 percent.
- Non-oil sector primary balance is highly negative and persistent.
- Government fiscal target: the primary balance of the non-oil sectors must be greater than -5 percent of GDP.

### Treasury FX operations, exchange-rate volatility, and risk channel inclusion
- Treasury collects revenues in foreign currencies (mainly, tax collections from the oil and diamond sectors) and deposits them in a BNA account denominated in US dollars, or accounts abroad.
- Treasury conversions of foreign currency to Kwanzas to pay the primary deficit in Kwanzas are not programmed.
- BNA lacks a framework and instruments to smooth large FX entries and avoid distortions in exchange-rate formation.
- Treasury FX actions create moral hazard and speculative timing incentives harming non-oil sector agents.
- Proposed extensions to the risk premium equation (three fiscal-related channels):
  1. Primary balance in Kwanzas (non-oil revenues minus current expenditure and public investment): higher deficits increase pressure to use FX balances, raise the risk premium, and lead to Kwanza depreciation (inflationary).
  2. Government revenues in hard currency: increase availability of reserves and can reduce the risk premium.
  3. Government expenditures in foreign currency: reduce international reserves and can pressure the risk premium.
- Country risk premium is a non-observable variable in the current QPM version.

### Data availability and chosen variables for fiscal channel
- Only fiscal results on a cash basis are available at a quarterly frequency; these quarterly cash-flow data do not coincide with official annual Treasury accrual-basis figures.
- Two relevant variables for explaining exchange-rate formation:
  - The primary balance in Kwanzas (government revenues from non-oil sectors minus current expenditure and public investment; excludes oil revenues collected in US dollars).
  - The government's foreign currency reserves (government USD balances).
- The overall primary balance (including oil revenues) is consistently positive and is not appropriate to address the fiscal channel.

### Modeling, estimation, and quantitative results (as reported)
- Mission incorporated a demand channel and the exchange-rate risk channel into the core model; the primary balance in Kwanzas was added to the IS curve.
- Calibration and sensitivity:
  - Several calibration options for the strength of the demand channel were considered and subject to sensitivity analysis.
  - TA fixed code generating IRFs and expanded it to compare models with alternative calibrations.
  - Empirical evidence must validate model responses; additional research is required to support parameter choices.
- IS curve estimation:
  - The coefficient of the impact of the real interest rate on the non-oil sector output gap doubled in value after estimation.
  - The primary balance coefficient was estimated to be low but relevant for short-run adjustment.
- Quantitative shocks and parameters:
  - Government spending shock: expansion of 1 percent of GDP can boost non-oil sectors, pressuring inflation; the contemporaneous impact coefficient a7 in the IS curve was estimated as a7=0.06.
  - Tax shock: expansion of 1 percent of GDP uses the same a7=0.06 coefficient for contemporaneous impact on non-oil activity.
  - Government USD balances shocks:
    - A 10-percent reduction in government foreign accounts in USD can increase the risk premium, cause Kwanza depreciation, and be inflationary. The coefficient w_cty_spendgov in the country risk premium equation controls this strength; the local estimation during the TA found w_cty_spendgov was estimated to be around zero.
    - A 10-percent increase in government foreign accounts in USD can reduce the risk premium and appreciate the Kwanza. The coefficient found in estimation for increases was more significant, indicating possible asymmetry.

### Policy implications and reforms on FX management
- Institutional and legal FX market reforms, better Treasury planning, and BNA instruments can regularize internal dollar supply and reduce exchange-rate formation distortions.
- Under inflation targeting with a flexible exchange rate and institutional reforms, the exchange-rate risk channel tied to fiscal flows may lose importance and could be switched off.
- BNA needs frameworks to smooth large FX inflows/outflows and to prevent Treasury operations from creating speculative incentives against the Kwanza.

### Improving monetary policy decision-making and communication
- FPAS should become more forward-looking; monetary policy conduct must aim for medium-term disinflation while anchoring expectations.
- Operational improvements:
  - Strengthen the monetary policy framework and open market operations so the official BNA rate becomes the benchmark for market rates.
  - Budget interest rate paths and discuss policy trade-offs to support MPC decisions; the code for budgeting interest rate trajectories was provided during TA.
  - Schedule regular pre-planned all-team meetings linked to the forecasting calendar and MPC meeting dates; involve senior management and MPC members.
- Survey-based expectations:
  - Expand the BNA market expectations survey beyond a 1-year horizon to monitor longer horizons; investigate the quality of long-run inflation responses.
  - Monitor projection errors, rank forecasters, and consider publicity/reward mechanisms to improve incentives and reporting quality.
- Communication:
  - Reform monetary policy statements to be more forward-looking; publish documents prominently for public consultation.
  - Consider publishing MPC minutes describing prospective scenario discussions and factors driving decisions to build credibility during the transition to inflation targeting.
- Transition to inflation targeting requires simultaneous progress on mindset change, forward-looking analysis, expectations management, and communication.

### Next steps indicated by authorities
- Authorities expressed interest in follow-up TA pending adoption of recommendations from this mission and the 2022 TA mission.
- Potential follow-up TA could include enhancing monetary policy communication under inflation targeting and support to calibrate and enrich the QPM tailored to the Angolan economy.

*IMF Technical Assistance Report — excerpts from the mission’s assessment and QPM modelling of fiscal channels and FX-related risks*

### PREFACE _______________________________________________________________________ 5

### PREFACE

### Preface — Mission purpose and meetings
- A Monetary and Capital Markets (MCM) Department mission visited Luanda, Angola from May 20 to May 24, 2024, at the request of the Banco Nacional de Angola (BNA).
- Mission objectives: review BNA’s Forecasting and Policy Analysis System (FPAS); propose enhancements to modeling and forecasting procedures; introduce fiscal channels into the core forecasting model to better characterize Angolan economy dynamics.
- High-level meetings included Vice Governor Ms. Maria Juliana de Carvalho Van-Dúnem de Fontes Pereira, Mr. Sebastião Kivampolo Tuma (DEE Director), Ms. Martine Emma Dias dos Santos (Deputy Director of DEE), other senior BNA officials, and IMF AFRITAC South representatives.
- The mission worked daily with DEE experts across divisions; the mission thanks BNA authorities and staff for cooperation and hospitality.

### Executive summary — Key assessments
- The report presents an updated assessment of FPAS operations at BNA and TA mission objectives to strengthen analytical capacities to support policymaking and communication during transition to inflation targeting.
- Progress noted: BNA has made impressive progress in recent years improving its FPAS.
- Mission activities: reviewed FPAS components; assisted BNA staff in introducing fiscal channels into the QPM; evaluated aspects of policymaking.
- Recommendations emphasize pending actions to be adopted in the near term to strengthen the transition to Inflation Targeting and increase BNA ownership of FPAS and models.
- Risks and constraints identified: mounting vulnerabilities for FPAS due to high turnover, small teams, and scarcity of senior profiles, posing implementation risks for recommendations.
- Final assessment: current stage of FPAS at BNA partially meets environmental conditions to operate under inflation targeting; further progress required in use of FPAS models internally and in monetary policy communication.

### Introduction — Mission scope
- Mission purposes:
  - i) Review operation of QPM and follow up on previous TA recommendations;
  - ii) Assist staff to build a fiscal block in QPM to account for fiscal channels for story-telling and short-term forecasting;
  - iii) Assess adequacy of BNA resources to operate core model to produce medium-term forecasts;
  - iv) Discuss potential topics for further TA.
- 2022 TA achievements and prior recommendations summarized:
  - Key achievements: i) tracking and documenting forecast performance continuously; ii) using core projection model to actively generate interest rate paths; iii) fully supporting forward-looking monetary policy.
  - Prior recommended short-term changes to QPM: i) make the Taylor rule active; ii) impose a UIP condition instead of an adjustment equation for the nominal exchange rate.
- Fiscal block purpose: incorporate traditional IS curve fiscal channels; reflect treasury operations in foreign exchange markets that affect price formation and relate to fiscal results.
- Mission explored organizational, human, and technological resource improvements to support forward-looking process and interest rate-based monetary policy advice.

### I. Reviewing the Forecasting and Policy Analysis System — Summary of findings
- Overall progress:
  - BNA has consolidated FPAS processes and improved information sharing; DEE houses all staff involved in forecasting.
  - Workflow and communication improved; DEE positioned to make additional investments for inflation-targeting responsibilities.
- Current Set-Up (A. Current Set-Up) — operational findings:
  - Team roles well-organized; a formal calendar for forecasting cycle is in place. Each forecasting cycle takes six weeks in the run-up to the MPC meeting.
  - Calendar includes briefing meetings to explore data releases but lacks an all-together meeting to present a tentative baseline scenario and discuss each deliverable’s role.
  - MPC members do not participate in conjunctural development meetings; absence of review/retrospective meetings may limit groups’ views of input impacts.
  - Data and modeling toolkit:
    - Relevant data inputs for QPM are organized by department; databases and data workflows are still poorly integrated and lack flexibility for ad-hoc investigation.
    - Nowcasting and near-term forecasting models for selected variables are operational.
    - Core model contains specifications tailored to Angola to capture non-official exchange rate dynamics, oil price effects on activity and inflation, and exposure to global food price cycles.
  - Current QPM architecture:
    - QPM version used to support MPC decisions is based on monetary aggregates with a money block (demand and supply) operated by the central bank.
    - Real interest rate is endogenously calculated by filtering, given observable Luibor rate (Luanda Interbank Offered Rate).
    - The underlying real interest rate affects non-oil sector output gap and core inflation.
    - Taylor rule is muted; official exchange rate is set to gradually reduce spread to the non-official exchange rate, affected by gaps of real exchange rate and real interest rate.
  - Model performance issues:
    - Current QPM calibrated to generate a small response of inflation to a money supply shock; IRF of money supply shock shows model-filtered interest rate equivalence and weak transmission to inflation.
    - BNA staff assess difficulty of stabilizing economy in short term via demand channel; non-oil output may be little affected by rising interest rates because private/public investments are driven by expectations for oil revenues, exchange rate, and fiscal results.
    - Banking channels currently have limited role.
  - Forecasting practice:
    - QPM produces internally consistent baseline and alternative scenario material with a 6-quarter horizon for main variables: output and sectoral decomposition; consumer inflation and components; monetary base and M2; official and non-official exchange rates; interbank interest rate.
    - Drawbacks: projections rely significantly on filtered gaps (non-oil output gap, real interest rate gap, real exchange rate gap) that may be disconnected from reality without explicit judgment.
    - Judgments in baseline not aligned with best practices: BNA undisclosed and changeable inflation objectives are used as inputs; model used to calculate money supply paths consistent with exogenous inflation paths, making model act as a calculator.
- Follow-up on 2022 TA recommendations (B. Follow-up on the Recommendations from the 2022 TA Report):
  - Changes proposed in 2022 (activate Taylor rule, impose UIP) were produced but not deployed operationally.
  - BNA staff implemented a version with those equations; some minor issues detected and solved during mission; forecasting exercise tried with those assumptions.
  - Staff hesitation to switch to new version due to:
    - i) need to forecast and report money aggregates to MPC during transition;
    - ii) view that BNA interest rate cannot yet be taken as market benchmark;
    - iii) reduced power of interest rates to control aggregate demand;
    - iv) short-term projections for exchange rate movements and consumer inflation appear excessive/unrealistic.
  - Current practice of using shocks to generate medium-term profile has clear drawbacks:
    - Underlying shocks in medium run are always unanticipated and not determined from drivers of price formation.
    - Implied shocks lack explanation or counterpart; represent abstract disinflation events with no adjustment costs.
    - Shocks in one period may not anchor expectations in subsequent periods.
    - Money supply policy derived from this exercise is not forward-looking nor efficient; central bank is surprised by disinflation each period.
  - Policy formulation remark: model useful to find paths, but strategy must rely on tuning policy instruments and assessing effects on endogenous variables.

### Key recommendations (extracted from Table 1: Key Recommendations)
- Modeling and Forecasting — High priority, Near-term (< 12 months)
  - Switch immediately to the model with the Taylor rule and uncovered interest rate parity (UIP) condition; if helpful, consider running the old version in parallel for a short period when new tunes are placed to align projections; do not make tunes on medium-run inflation. — Priority: High; Timeframe: Near-term
  - Extend the forecasting window to at least 24 months ahead; check key model gaps closing in the long term. — Priority: High; Timeframe: Near-term
  - Track and document forecast performance continuously to ensure effective narrative construction and support model maintenance. — Priority: 3; High; Timeframe: Near-term
  - Build empirical evidence on the power and dynamics of key macro shocks, such as interest rate, exchange rate, and fiscal impulse. — Priority: 3; High; Timeframe: Near-term
  - Produce sensitivity analysis for key parameters whenever necessary, to understand general equilibrium responses beyond the immediate effect. — Priority: Medium; Timeframe: Near-term
  - Invest time to deepen comprehension of model codes; consider regular training in MATLABMATLAB and IRIS toolbox; try small changes and create new products to gain confidence and ownership of the codes. — Priority: Medium; Timeframe: Near-term
  - Create additional analysis and decomposition tools of scenarios to improve understanding of what is driving forecasts; make the case for changes in model parameters and specifications. — Priority: Medium; Timeframe: Near-term
  - Improve data integration between departments; enhance information sharing through secure network folders (beyond email). — Priority: Medium; Timeframe: Medium-term
  - Address human resource management challenges like high turnover, weak skills transfer, and scarcity of senior profiles; consider increasing the number of staff with quantitative skills. — Priority: Medium; Timeframe: Medium-term
- Policy formulation process and communication — Mixed priority, Near- to Medium-term
  - Schedule regular, pre-planned meetings between different teams along the forecast cycle to improve coordination of assumptions and validation of the baseline scenario, and to brainstorm about alternatives; ensure the discussions are prospective. — Priority: Medium; Timeframe: Near-term
  - Enhance communication with the public immediately with a more forward-looking approach; draft monetary policy documents with a prospective analysis, emphasizing main drivers and the Monetary Policy Committee (MPC) assessments behind policy decisions. — Priority: 3; High; Timeframe: Near-term
  - Start budgeting interest rate paths and discuss policy trade-offs to support MPC decision2; adjust communication to include MPC's views on monetary policy stance and timing to convergence. — Priority: High; Timeframe: Near-term
  - Evaluate the quality of survey-based market expectations, particularly for inflation in medium-term horizons. — Priority: High; Timeframe: Medium-term

- Timing definitions provided:
  - Near term: < 12 months
  - Medium term: 12 to 24 months

*IMF Technical Assistance Report — PREFACE (content unit from the BNA TA mission materials)*

### 17. The FT argued that money supply policies in the current QPM cannot be used to

### 17. The FT argued that money supply policies in the current QPM cannot be used to generate the inflation paths because the response of inflation is too low

### Assessment of money supply shocks and model performance
- The FT tried to calibrate the money supply shock such that medium-term inflation would meet objectives, but the response of inflation is small.
- The FT considers it hard to find consistent trajectories for aggregates to achieve monetary objectives, since inflation feeds back strongly into the demand for money.
- In the baseline scenario explored during the mission, forecasts showed a strong immediate devaluation of Kwanza produced by the UIP condition, even with higher interest rates, which caused projected inflation above the reported by the current model.
- Figure 1 illustrative point: A decrease of 1 percent qoq (4 percent qoqar) in money supply generates an implied interest rate hike of around 55 bps.

### Recommendations on monetary policy instruments and model structure
- Move towards more efficient monetary policy instruments; the current QPM version and procedures are no longer appropriate for advising policy.
- Switch immediately to the new version of QPM with a Taylor rule and a UIP condition.
- If useful for communication, run the two versions in parallel for a few months while refining initial conditions and short-term judgments that improve model forecasts.
- Staff must refrain from making judgments on inflation in the medium term; it is paramount to observe how inflation responds endogenously to policy alternatives and short-term conditionings.
- The FT should do more work to confirm by other means that the power of monetary policy over non-oil activity is indeed so low; identifying and estimating transmission elasticities is a continuous task.

### Protocols for judgments and scenario conditioning
- FT should organize a protocol to make judgments in selected variables: define variables and the number of quarters for short-term conditioning to organize hypothesis discussions and track changes in judgments.
- Typical practice: create baseline scenarios with trajectories for volatile variables (exchange rates, oil prices, food prices) and then explore alternative paths.
- For the mission’s baseline scenario, FT could evaluate setting a near-term trajectory for official and informal exchange rates and external variables for a few quarters.
- FT assessment of mechanical model outputs is part of the process, but governance must include the ability to subject forecasts considered non-realistic to exogenous conditioning and to identify sources of unrealistic effects.

### Model gaps, filtering, and forecasting horizon
- When building the baseline scenario, monitor key filtered gaps; model gaps are essential in determining dynamics and some gaps appear excessively open.
- The dynamics of the exchange rate and non-oil sector activity have shown important imbalances; the model may project short-run rebalancing that is unrealistic.
- FT calculates an out-of-model estimate of the effective real exchange rate gap for the Kwanza used for MPC monitoring; staff could provide an educated guess for the real exchange rate gap before filtering to obtain a better trend/cycle decomposition.
- Recommendation to extend forecasting window to at least 24 months ahead and closely monitor medium-term inflation and its drivers.
- Check if key model gaps are closing in the long term; inflation targeting requires a more forward-looking approach and consideration of trade-offs between gradual or timely responses and the buildup of central bank credibility.

### Interest-rate budgeting, trade-offs, and reporting
- TA mission developed and explained an alternative code for budgeting interest rates to discuss alternative interest rate paths in terms of costs to economic activity and timing of convergence of inflation to objectives.
- The new script might eliminate the need for hard tunes on medium-term inflation to meet inflation objectives.
- FT should extend codes to discuss budgeting interest rates under different conditioning paths for key variables like exchange rates, oil prices, and other external variables.
- Illustrative exercise (Figure 2): presents four alternative scenarios for adjusting the policy rate, evaluating tradeoffs in convergence and costs to economic activity of immediate action versus gradualism. The blue line is the BNA’s baseline; other lines show alternative adjustment speeds and specific 2024: Q1 and 2024: Q4 timing configurations (as described in the source).

### Money quantity satellite and forecast evaluation
- TA mission added a satellite specification based on the money quantity equation for reporting money growth consistent with endogenous variables to accelerate effective implementation of the version with Taylor rule and UIP.
- Having an endogenous projection for the quantity of money can be useful during the transition to inflation targeting and increase staff confidence in moving to the new QPM immediately.
- Mission reinforces recommendations for BNA to track and document forecast performance; evaluate sources of forecast errors and their role in changing nowcasts and near-term forecasts.
- Tracking changes in model gaps can improve storytelling; additional TA can help design a transition map to account for changes between current and previous scenarios.

### Need for empirical estimation of transmission elasticities
- Immediate need for out-of-model estimates of the power and dynamics of interest rate shocks, exchange rate shocks, and fiscal impulses to validate calibration and strengthen channels.
- Key parameters include coefficients determining pass-through of transmission channels of monetary policy; additional TA can help estimate and/or recalibrate these parameters.

### Code, diagnostics, and model development enhancements
- TA updated QPM codes with warnings and errors; medium-term model largely based on MATLAB scripts coded in old software versions.
- FT works on a 2018a version of MATLAB; mission updated scripts to be compatible with any released version and with the newest 2023 IRIS Toolbox, removing use of the 2015 IRIS Toolbox, and explained syntax differences.
- Recovered code for IRFs and recommended regular use to understand overall effects beyond immediate impact; used IRFs for alternative calibrations, notably fiscal block parameters.
- Recovered code for Bayesian estimations and explained elements used for built-in maximum likelihood estimation in the new IRIS version; ran script to locally estimate some fiscal block parameters.
- Introduced a graph in simulation code to highlight all judgments made in the scenario to help screen for errors or missing items.
- FT should invest more time to analyze scenario outputs, check for inconsistencies, and increase familiarity with the code; lack of familiarity limits progress in model analysis.
- On the road to inflation targeting, FT needs new analysis and decomposition tools to improve understanding of scenarios and model channels; additional TA can help build tools and richer reports.

### Data sharing, external review, and operational recommendations
- Improve data sharing and integration between areas; use secure network folders rather than email attachments to improve security, integrity, coordination, and control.
- An external review of FPAS operating during one forecasting cycle could help identify practical issues and areas for quick implementation improvements.
- FT should prioritize expanding knowledge about codes and models; in the short term focus on explaining contributions of conditioning variables to forecast changes between meetings rather than expanding model blocks.
- The mission recommends continuous and intensive training in MATLAB and IRIS toolbox.

### Assessment of BNA human resources and infrastructure
- FPAS at BNA has substantial vulnerabilities related to human resources: small team sizes increase risk of burnout and turnover; core model forecasting activities performed by only three people.
- No manual of usage and procedures for the entire BNA forecasting process; deficiency of expertise prevents greater progress.
- FT members have doubts about code steps and do not know how to address warnings/errors or implement deeper changes like adding new variables.
- BNA needs to address high turnover, loss of specialized knowledge, and scarcity of senior profiles; evaluate forecasting team size to ensure a well-functioning FPAS.
- Increase opportunities for internal training and knowledge multiplication; staff returning from external training should engage in practical forecasting activities to consolidate knowledge.
- Resizing and improving technological infrastructure recommended: network stability, more MATLAB licenses, more computers ready to run scenarios so forecasting exercises can run in parallel.
- Rotation of responsibilities within the group running QPM recommended to increase awareness of FT’s key messages.

### Preparations for inflation-targeting transition
- Shortages in proficiency and ownership will be exacerbated in transition to inflation targeting; FPAS must be adapted to produce prospective analyses and narratives for MPC needs.
- FT needs to:
  - i) address quantitatively the contribution of scenario assumptions and the uncertainty;
  - ii) discuss prospective inflation and its drivers in different scenarios; and
  - iii) prepare the MPC to deliver effective communication with the public.
- Development of new forward-looking FPAS products and decompositions is a priority over medium-term model expansion.

### Introducing fiscal channels into the QPM — mission rationale and focus
- BNA requested TA to introduce fiscal channels in the core model to improve storytelling and capture Treasury’s role in price determination, particularly in exchange rate markets.
- Fiscal channels expected to improve historical decomposition, support fiscal policy simulations, and infer required monetary policy reactions.
- Focus is on fiscal result flows because debt is not a concern in the near term.

### Fiscal background and immediate focus
- Angola debt exposures and composition (2022 figures as reported):
  - 79 percent of Angola's debt in 2022 is exposed to exchange rates.
  - 70 percent is foreign debt.
  - 21 percent corresponds to the overall internal debt denominated in kwanzas, with several indexation measures.
- Angolan debt is on a downward trajectory as fiscal consolidation is underway.
- Angola has a high and positive general government primary balance, and the current account balance is consistently positive at around 5 percent.
- These good numbers are strongly influenced by oil sector results; non-oil sector primary balance is highly negative and persistent.
- Debt sustainability is not an issue now; focus is on flows and the impact of fiscal channels on short-term dynamics.

*IMF Technical Assistance Report | content unit 17–44 excerpts*

### 45. Treasury movements in the foreign exchange market cause high volatility and distort

### 45. Treasury movements in the foreign exchange market cause high volatility and distort

### Treasury FX operations and exchange-rate volatility
- The Treasury collects revenues in foreign currencies (mainly, tax collections from the oil and diamond sectors) and deposits them in a BNA account denominated in US dollars, or accounts abroad (e.g., Escrow accounts related to debts to China).
- Treasury conversions of foreign currency to Kwanzas to pay the primary deficit in Kwanzas are not programmed.
- The BNA lacks a framework and instruments to smooth large FX entries and avoid distortions in exchange-rate formation.
- Treasury FX actions create moral hazard and speculative timing incentives: selling hard currency when domestic FX liquidity is low to obtain larger gains in Kwanzas, harming non-oil sector citizens and companies that lack hard-currency revenues.
- Previous TA recommendations focused on reforms in the Angolan foreign exchange markets and adoption of new BNA instruments to tackle volatility.

### Demand for the Kwanza and institutional environment
- Foreign companies and investors are not required to convert dollars to local currency and therefore have low structural demand for Kwanzas.
- The Kwanza is not used as a store of value by private agents; demand for Kwanzas is structurally lower than in other countries.

### Exchange-rate risk channel for fiscal flows (inclusion in QPM)
- Rationale:
  - Active Treasury participation in FX markets is highly relevant for spot exchange-rate formation and transmission to inflation and non-oil sector activity.
  - A risk channel improves storytelling and captures richer short-term dynamics by affecting official and informal exchange rates and linking to Treasury financing needs and FX cash reserves.
- Proposed extensions to the risk premium equation (three fiscal-related channels):
  1. Primary balance in Kwanzas (non-oil revenues minus current expenditure and public investment): higher deficits increase pressure to use FX balances, raise the risk premium, and lead to Kwanza depreciation (inflationary).
  2. Government revenues in hard currency: increase availability of reserves and can reduce the risk premium.
  3. Government expenditures in foreign currency: reduce international reserves and can pressure the risk premium.
- The strength of these channels is determined by coefficients of the fiscal measures in the risk premium equation.
- The country risk premium is a non-observable variable in the current QPM version.

### Data availability and variable choice
- Only fiscal results on a cash basis are available at a quarterly frequency; these quarterly cash-flow data do not coincide with official annual Treasury accrual-basis figures.
- The staff concluded that two relevant variables for explaining exchange-rate formation are:
  - The primary balance in Kwanzas (government revenues from non-oil sectors minus current expenditure and public investment; excludes oil revenues collected in US dollars).
  - The government's foreign currency reserves (government USD balances).
- The overall primary balance (including oil revenues) is consistently positive and is not appropriate to address the fiscal channel.
- The government fiscal target: the primary balance of the non-oil sectors must be greater than -5 percent of GDP.

### Modeling and estimation results
- The mission incorporated a demand channel and the exchange-rate risk channel into the core model; the primary balance in Kwanzas was added to the IS curve.
- Calibration and sensitivity:
  - Several calibration options for the strength of the demand channel were considered and subject to sensitivity analysis.
  - TA fixed code generating IRFs and expanded it to compare models with alternative calibrations.
  - Empirical evidence must validate model responses; additional research is required to support parameter choices.
- IS curve estimation:
  - The coefficient of the impact of the real interest rate on the non-oil sector output gap doubled in value after estimation.
  - The primary balance coefficient was estimated to be low but relevant for short-run adjustment.
- Risk premium estimation:
  - Augmented country risk premium equation with fiscal risk channels showed little effect of incorporating fiscal channels on endogenous risk-premium dynamics.
  - Identification of these channels in Bayesian estimation is not reliable given historical changes in exchange-rate regimes.
  - The mission recommended further empirical evidence and staff assessment to calibrate parameters for forecasting horizons.

### Quantitative shocks and estimated parameters (as reported)
- Government spending shock illustrated: expansion of 1 percent of GDP can boost non-oil sectors, pressuring inflation; the contemporaneous impact coefficient a7 in the IS curve was estimated as a7=0.06.
- Tax shock illustrated: shock to tax revenues (expansion of 1 percent of GDP) contemporaneous impact on non-oil activity uses the same a7=0.06 coefficient.
- Government USD balances shocks:
  - A 10-percent reduction in government foreign accounts in USD (e.g., a USD selloff to settle the primary deficit in Kwanzas) can increase the risk premium, cause Kwanza depreciation, and be inflationary. The coefficient w_cty_spendgov in the country risk premium equation controls this strength; the local estimation during the TA found w_cty_spendgov was estimated to be around zero.
  - A 10-percent increase in government foreign accounts in USD can reduce the risk premium and appreciate the Kwanza. The coefficient found in estimation for increases was more significant, indicating possible asymmetry and the importance of oil-related USD balances for exchange-rate dynamics.

### Policy implications and reforms
- Institutional and legal FX market reforms, better Treasury planning, and BNA instruments can regularize internal dollar supply and reduce exchange-rate formation distortions.
- Under inflation targeting with a flexible exchange rate and institutional reforms, the exchange-rate risk channel tied to fiscal flows may lose importance and could be switched off.
- The BNA needs frameworks to smooth large FX inflows/outflows and to prevent Treasury operations from creating speculative incentives against the Kwanza.

### Improving the monetary policy decision-making process (policy-making steps summarized)
- FPAS should become more forward-looking; monetary policy conduct must aim for medium-term disinflation while anchoring expectations.
- Operational improvements:
  - Strengthen the monetary policy framework and open market operations so the official BNA rate becomes the benchmark for market rates.
  - Budget interest rate paths and discuss policy trade-offs to support MPC decisions; the code for budgeting interest rate trajectories was provided during TA.
  - Schedule regular pre-planned all-team meetings linked to the forecasting calendar and MPC meeting dates to improve forward-looking analysis and scenario validation; involve senior management and MPC members.
- Survey-based expectations:
  - Expand the BNA market expectations survey beyond a 1-year horizon to monitor longer horizons; investigate the quality of long-run inflation responses.
  - Monitor projection errors, rank forecasters, and consider publicity/reward mechanisms to improve incentives and reporting quality.
  - Evaluate survey responsiveness to shocks and consider synchronizing collection dates with MPC meetings.
- Communication:
  - Reform monetary policy statements to be more forward-looking; publish documents prominently for public consultation.
  - Consider publishing MPC minutes describing prospective scenario discussions and factors driving decisions to build credibility during the transition to inflation targeting.
- The transition to inflation targeting requires simultaneous progress on mindset change, forward-looking analysis, expectations management, and communication; failing any step can compromise the process.

### Next steps indicated by authorities
- Authorities expressed interest in follow-up TA pending adoption of recommendations from this mission and the 2022 TA mission.
- Potential follow-up TA could include enhancing monetary policy communication under inflation targeting and support to calibrate and enrich the QPM tailored to the Angolan economy.

*IMF Technical Assistance Report — excerpts from the mission’s assessment and QPM modelling of fiscal channels and FX-related risks*

---


_Source: https://www.imf.org/-/media/files/publications/tar/2025/english/tarea2025020-print-pdf.pdf_