## 5. Impulse Responses of FCI to Monetary Policy Tightening

## Source details

**Canonical URL:** [5. Impulse Responses of FCI to Monetary Policy Tightening](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025160-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025160-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025160-source-pdf.pdf.json)

---

### Overview and context
- Objective: assess monetary policy stance in selected sub-Saharan African (SSA) emerging and frontier market economies by (i) estimating the neutral real interest rate (r-star), (ii) measuring the real interest rate gap (ex-ante real policy rate minus r-star), (iii) constructing financial conditions indices (FCIs), and (iv) estimating impulse responses of FCIs to identified monetary policy shocks.
- Sample coverage: quarterly data spanning 2002Q1 to 2023Q4 (where available). Where quarterly data were unavailable, annual data were interpolated using a quadratic interpolation method.
- Countries/economies analyzed: South Africa; frontier markets—Ghana, Kenya, Mauritius, Mozambique, Nigeria, Tanzania, Uganda, Zambia; two monetary unions—CEMAC and WAEMU.
- Monetary policy frameworks in sample (status as of end-2023): five economies classified as having an inflation-targeting framework (Ghana, Kenya, Mauritius, Uganda, and South Africa); Nigeria and Tanzania target monetary aggregates; Mozambique and Zambia follow eclectic frameworks; the two monetary unions peg their currencies to the euro.

### Neutral real interest rate (r-star): estimation approaches and key empirical details
- Definition: neutral real interest rate (r-star) = level of the real policy rate consistent with output at potential and inflation at target once cyclical shocks have dissipated.
- Methods used (six complementary approaches):
  - sample averages (historical mean of ex-ante real rate),
  - univariate filters (HP with tail correction; Christiano-Fitzgerald with band-pass 6 to 32 quarters),
  - recursive Taylor rule (rolling windows of 32 quarters; initial estimation 2002Q1–2009Q4),
  - reduced-form model (r = α + β1 g + β2 r* with potential growth g from HP-filtered real GDP trend and U.S. real federal funds rate as r*),
  - time-varying parameter VAR (TVP-VAR) with extension to include real effective exchange rate; neutral rate from five-year-ahead TVP-VAR forecast of the real interest rate,
  - semi-structural HLW-style model augmented for small-open economies (IS and Phillips curves include real effective exchange rate).
- Data and measurement choices:
  - ex-ante short-term real interest rate = nominal policy rate − one-year-ahead CPI inflation forecasts from IMF WEO (alternative checks using three-year-ahead forecasts and four-quarter moving average of past inflation).
  - HP filter smoothing parameter lambda = 1,600 for quarterly data; HP endpoint bias addressed by extending series with ARIMA projections to 2027Q4 in some applications.
- Combining methods: central r-star measure = simple average (mean) across the six methods (the paper excludes the Taylor-rule estimate from the reported mean r-star used in some figures).

### Empirical findings on r-star and neutral-rate trends
- Cross-method consistency: r-star trends are broadly consistent across methods within each country despite diverging point estimates; the study uses the mean across methods as benchmark.
- Historical patterns:
  - In three-quarters of countries, average r-star estimates peaked around 2007 and declined during the 2008–09 Global Financial Crisis (GFC) and declined further during the pandemic.
  - Some exceptions with more resilient r-star: Ghana, Kenya, and Mozambique.
  - Post-pandemic (as of end-2023) r-star estimates across countries ranged between -0.8 percent and 4.6 percent.
  - In about a quarter of SSA economies (including CEMAC, WAEMU, and South Africa), estimated r-stars have fallen to levels broadly comparable to advanced economies (AEs), although in most SSA economies estimated r-stars remain higher than in AEs.
- Cross-country heterogeneity in levels: examples cited—Mauritius had a negative average neutral rate as of 2023Q4; Ghana and Mozambique had neutral rates nearing 6-8 percent at certain points.

### Monetary policy stance: interest rate gaps and stylized patterns
- Definition: real interest rate gap = ex-ante real short-term policy rate − r-star (positive gap typically indicates restrictive stance; negative gap indicates accommodative stance).
- Neutral stance definition used in the paper: interest rate gap within the range of -0.5 to 0.5 percentage points is classified as neutral.
- Stylized patterns across shock episodes:
  - GFC (2008–09): most countries pursued a loose (accommodative) policy stance; exceptions with tighter stances included CEMAC, Ghana, South Africa, and Zambia.
  - Commodity price slump / China slowdown (2014–15): about half of economies adopted tighter stances (e.g., Ghana, Mauritius, Tanzania, South Africa, Zambia).
  - COVID-19 pandemic (2020): policy remained broadly accommodative across the region.
  - By end-2023: half the sample had restrictive stances—Kenya, Mozambique, South Africa, WAEMU, and Zambia; broadly accommodative stances—CEMAC, Ghana, Mauritius, Tanzania, and Uganda; Nigeria’s stance was neutral.

### Financial conditions indices (FCIs): construction and role
- Purpose: incorporate the first-stage transmission of policy (policy rate → market and banking rates, credit, asset prices, exchange rates) into stance assessment, accounting for impaired transmission channels in SSA.
- FCI construction:
  - Variables grouped into four categories: price indicators (bank interest rates, government security yields), quantity indicators (monetary and credit aggregates expressed in growth rates), foreign exchange market indicators, and global financial factors (VIX, EMBI).
  - All variables standardized to zero mean and unit standard deviation; transformed so higher values indicate tighter financial conditions.
  - Baseline method: equal-weight simple average (robustness checks include PCA-based FCIs).
- Observed FCI dynamics: FCIs capture major global shocks while reflecting country-specific volatility; dynamics vary with data availability and composition of indicators (upstream vs downstream components).

### Consistency between interest rate gaps and financial conditions
- Empirical relationship:
  - Contemporaneous correlations between interest rate gaps and FCIs over full sample range between 0.2 and 0.6.
  - Stronger links (correlation ≈ 0.6, contemporaneous or slightly leading) observed in about half of economies that have adopted or are transitioning to inflation targeting (examples: Ghana, Kenya, Uganda, South Africa, and Mozambique).
  - In other economies correlations are positive but weaker (≈ 0.2), indicating a weaker link between intended policy stance and the broader financial conditions.
- Observed pattern across episodes:
  - Large interest rate gaps tended to coincide with significant shifts in FCIs (e.g., GFC, certain 2021–23 tightening episodes), but systematic cointegration across countries is limited.

### Identification of monetary policy shocks and local projection estimation
- Identification strategy:
  - Follow Romer and Romer (2004) approach adapted to quarterly data: regress changes in the monetary policy rate on current and one-year-ahead forecasts of GDP growth and inflation (from IMF WEO), lags of GDP growth and inflation, lagged interest rate, and lags of change in nominal effective exchange rate; residuals are interpreted as exogenous monetary policy shocks.
  - Rationale: financial sector fully anticipates systematic policy actions; residuals capture the non-systematic/unexpected component.
- Local projection implementation:
  - Estimate impulse responses of FCI (excluding the policy rate) to identified monetary policy shocks using Jordà (2005) local projections.
  - Horizons h = {0,1,2,...,8} (quarters); specification includes lags of FCI, four lags of changes in oil prices and commodity prices excluding fuel, and four lags of monetary shocks as controls.
  - FCI normalized by adding 100 (mean 100, variance 1) for interpretability; confidence bands computed using Newey-West standard errors.

### Impulse-response results: effects of contractionary shocks on FCIs
- General result: contractionary (tightening) monetary policy shocks lead to tighter financial conditions (positive and statistically significant FCI responses), with notable cross-country heterogeneity in magnitude, speed, and persistence.
- Magnitude and timing:
  - In economies with inflation-targeting frameworks (e.g., South Africa, Ghana, Kenya, Mauritius, Uganda), responses are larger, faster, and more persistent: a 100-basis-point monetary policy shock increases FCIs by 0.5 to 1 index point at peak.
  - Peak effects typically occur between three and six quarters after the shock and can remain evident up to two years (i.e., persistence up to eight quarters).
- Other regime patterns:
  - Monetary-aggregate-targeting regimes (e.g., Tanzania) show relatively strong effects similar in pattern to inflation-targeters but with smaller magnitude.
  - Economies with eclectic or other frameworks (e.g., Mozambique, Zambia) tend to exhibit smaller, short-lived, or muted responses.
  - Two monetary unions (CEMAC and WAEMU) display relatively strong but gradual responses (delayed peak).
- Robustness and comparison with using actual policy rate changes:
  - Re-estimating responses using actual policy rate changes yields qualitatively similar results for most countries, suggesting limited anticipatory bias in those cases.
  - In about one-third of the sample (notably Ghana, Mauritius, and Zambia), responses to actual policy rate changes are weaker and more gradual than responses to identified shocks—implying anticipatory or endogenous components can attenuate measured effects when using raw policy-rate changes.
  - In Nigeria, responses to actual policy rate changes appear somewhat stronger than to identified shocks, suggesting potential underestimation by the identification procedure or more exogenous actual rate movements.

### Interpretation, caveats, and transmission-channel considerations
- Sources of cross-country heterogeneity in transmission strength:
  - Evolution and credibility of monetary policy frameworks (transition to inflation targeting and strengthening of operational frameworks and interest-rate based procedures).
  - Financial market development and composition of FCI (indices including upstream indicators—market yields, asset prices—display stronger, faster responses; indices dominated by downstream quantity indicators—credit, monetary aggregates—show slower responses).
  - Supply-side vulnerabilities and inflation volatility (large food shares in consumption, agriculture exposure, imported fuel and food) complicate predictability and central bank communication.
  - Data limitations: lack of high-frequency market data and central bank forecasts constrain identification and can induce measurement error.
- Caveats highlighted by the authors:
  - Identified shocks rely on IMF WEO forecasts as proxies for central bank projections; this assumes similar information sets between central banks and IMF.
  - Estimates are sensitive to sample, specification choices, and the composition of FCIs.
  - Neutral rate estimates carry considerable uncertainty; combining methods reduces but does not eliminate model uncertainty.

### Conclusions and policy implications (as stated in the chapter)
- A comprehensive assessment of monetary policy stance in SSA economies should combine neutral rate estimates with measures of broader financial conditions, given impaired and heterogeneous transmission.
- Key empirical takeaways:
  - Neutral rates in most SSA economies peaked around 2007, then declined during the GFC and again during the pandemic; post-pandemic r-star estimates range between -0.8 percent and 4.6 percent as of end-2023.
  - Monetary policy stances varied considerably across SSA economies during common shocks; by end-2023 half the sample exhibited restrictive stances while the remainder were accommodative or neutral.
  - The relationship between interest rate gaps and FCIs is strongest in economies that have adopted or are transitioning to inflation-targeting frameworks (contemporaneous or leading correlations around 0.6).
  - Contractionary monetary policy shocks generally tighten financial conditions; a 100-basis-point shock typically raises FCIs by 0.5 to 1 index point at peak in economies with stronger transmission, with peak effects between three and six quarters and persistence up to two years.
- Policy-relevant implications emphasized:
  - Relying solely on the neutral rate is insufficient where transmission to financial conditions is weak; central banks should monitor broader financial conditions alongside policy-rate-based metrics.
  - Strengthening operational frameworks, improving central bank communication, and developing financial market infrastructure can enhance the speed and magnitude of monetary transmission.
  - Further research is needed to link transmission strength to structural impediments (financial system weaknesses and monetary policy framework gaps) and to analyze effects on specific transmission channels and the real economy.

*Source: IMF Working Paper — chapter "5. Impulse Responses of FCI to Monetary Policy Tightening" (authors’ calculations and figures based on quarterly data 2002Q1–2023Q4).*

### Annex A. Additional Figures and Tables .................................................................................

### Annex A. Additional Figures and Tables

### Contents listing in this unit
- Annex A. Additional Figures and Tables ........................................................................................................ 34
- Annex B. Financial Conditions Indices for SSA ............................................................................................. 41
- B.1.   The selected financial variables ............................................................................................................ 41
- B.2.   An Index of financial conditions ............................................................................................................. 43
- B.3.   Interpreting financial conditions indices for SSA ................................................................................... 45
- Annex C. Data Sources ..................................................................................................................................... 52
- References ......................................................................................................................................................... 55

### Embedded box and figures referenced
- BOX
  - 1. Trade-Offs Faced by SSA EFM Central Banks Amid Global Shocks ............................................................. 32
- FIGURES
  - 1. Neutral Real Interest Rates ............................................................................................................................. 18
  - 2. Interest Rate Gaps .......................................................................................................................................... 19
  - 3. Financial Conditions and Interest Rate Gaps .................................................................................................. 27
  - 4. Impulse Responses of FCI to Monetary Policy Shocks .................................................................................. 28

*Source: wpiea2025160-source-pdf - Annex A. Additional Figures and Tables (IMF).*

### 5. Impulse Responses of FCI to Monetary Policy Tightening ..............................................................

### 5. Impulse Responses of FCI to Monetary Policy Tightening

### Overview and context
- Objective: assess monetary policy stance in selected sub-Saharan African (SSA) emerging and frontier market economies by (i) estimating the neutral real interest rate (r-star), (ii) measuring the real interest rate gap (ex-ante real policy rate minus r-star), (iii) constructing financial conditions indices (FCIs), and (iv) estimating impulse responses of FCIs to identified monetary policy shocks.
- Sample coverage: quarterly data spanning 2002Q1 to 2023Q4 (where available). Where quarterly data were unavailable, annual data were interpolated using a quadratic interpolation method.
- Countries/economies analyzed: South Africa; frontier markets—Ghana, Kenya, Mauritius, Mozambique, Nigeria, Tanzania, Uganda, Zambia; two monetary unions—CEMAC and WAEMU.
- Monetary policy frameworks in sample (status as of end-2023): five economies classified as having an inflation-targeting framework (Ghana, Kenya, Mauritius, Uganda, and South Africa); Nigeria and Tanzania target monetary aggregates; Mozambique and Zambia follow eclectic frameworks; the two monetary unions peg their currencies to the euro.

### Neutral real interest rate (r-star): estimation approaches and key empirical details
- Definition: neutral real interest rate (r-star) = level of the real policy rate consistent with output at potential and inflation at target once cyclical shocks have dissipated.
- Methods used (six complementary approaches):
  - sample averages (historical mean of ex-ante real rate),
  - univariate filters (HP with tail correction; Christiano-Fitzgerald with band-pass 6 to 32 quarters),
  - recursive Taylor rule (rolling windows of 32 quarters; initial estimation 2002Q1–2009Q4),
  - reduced-form model (r = α + β1 g + β2 r* with potential growth g from HP-filtered real GDP trend and U.S. real federal funds rate as r*),
  - time-varying parameter VAR (TVP-VAR) with extension to include real effective exchange rate; neutral rate from five-year-ahead TVP-VAR forecast of the real interest rate,
  - semi-structural HLW-style model augmented for small-open economies (IS and Phillips curves include real effective exchange rate).
- Data and measurement choices:
  - ex-ante short-term real interest rate = nominal policy rate − one-year-ahead CPI inflation forecasts from IMF WEO (alternative checks using three-year-ahead forecasts and four-quarter moving average of past inflation).
  - HP filter smoothing parameter lambda = 1,600 for quarterly data; HP endpoint bias addressed by extending series with ARIMA projections to 2027Q4 in some applications.
- Combining methods: central r-star measure = simple average (mean) across the six methods (the paper excludes the Taylor-rule estimate from the reported mean r-star used in some figures).

### Empirical findings on r-star and neutral-rate trends
- Cross-method consistency: r-star trends are broadly consistent across methods within each country despite diverging point estimates; the study uses the mean across methods as benchmark.
- Historical patterns:
  - In three-quarters of countries, average r-star estimates peaked around 2007 and declined during the 2008–09 Global Financial Crisis (GFC) and declined further during the pandemic.
  - Some exceptions with more resilient r-star: Ghana, Kenya, and Mozambique.
  - Post-pandemic (as of end-2023) r-star estimates across countries ranged between -0.8 percent and 4.6 percent.
  - In about a quarter of SSA economies (including CEMAC, WAEMU, and South Africa), estimated r-stars have fallen to levels broadly comparable to advanced economies (AEs), although in most SSA economies estimated r-stars remain higher than in AEs.
- Cross-country heterogeneity in levels: examples cited—Mauritius had a negative average neutral rate as of 2023Q4; Ghana and Mozambique had neutral rates nearing 6-8 percent at certain points.

### Monetary policy stance: interest rate gaps and stylized patterns
- Definition: real interest rate gap = ex-ante real short-term policy rate − r-star (positive gap typically indicates restrictive stance; negative gap indicates accommodative stance).
- Neutral stance definition used in the paper: interest rate gap within the range of -0.5 to 0.5 percentage points is classified as neutral.
- Stylized patterns across shock episodes:
  - GFC (2008–09): most countries pursued a loose (accommodative) policy stance; exceptions with tighter stances included CEMAC, Ghana, South Africa, and Zambia.
  - Commodity price slump / China slowdown (2014–15): about half of economies adopted tighter stances (e.g., Ghana, Mauritius, Tanzania, South Africa, Zambia).
  - COVID-19 pandemic (2020): policy remained broadly accommodative across the region.
  - By end-2023: half the sample had restrictive stances—Kenya, Mozambique, South Africa, WAEMU, and Zambia; broadly accommodative stances—CEMAC, Ghana, Mauritius, Tanzania, and Uganda; Nigeria’s stance was neutral.

### Financial conditions indices (FCIs): construction and role
- Purpose: incorporate the first-stage transmission of policy (policy rate → market and banking rates, credit, asset prices, exchange rates) into stance assessment, accounting for impaired transmission channels in SSA.
- FCI construction:
  - Variables grouped into four categories: price indicators (bank interest rates, government security yields), quantity indicators (monetary and credit aggregates expressed in growth rates), foreign exchange market indicators, and global financial factors (VIX, EMBI).
  - All variables standardized to zero mean and unit standard deviation; transformed so higher values indicate tighter financial conditions.
  - Baseline method: equal-weight simple average (robustness checks include PCA-based FCIs).
- Observed FCI dynamics: FCIs capture major global shocks while reflecting country-specific volatility; dynamics vary with data availability and composition of indicators (upstream vs downstream components).

### Consistency between interest rate gaps and financial conditions
- Empirical relationship:
  - Contemporaneous correlations between interest rate gaps and FCIs over full sample range between 0.2 and 0.6.
  - Stronger links (correlation ≈ 0.6, contemporaneous or slightly leading) observed in about half of economies that have adopted or are transitioning to inflation targeting (examples: Ghana, Kenya, Uganda, South Africa, and Mozambique).
  - In other economies correlations are positive but weaker (≈ 0.2), indicating a weaker link between intended policy stance and the broader financial conditions.
- Observed pattern across episodes:
  - Large interest rate gaps tended to coincide with significant shifts in FCIs (e.g., GFC, certain 2021–23 tightening episodes), but systematic cointegration across countries is limited.

### Identification of monetary policy shocks and local projection estimation
- Identification strategy:
  - Follow Romer and Romer (2004) approach adapted to quarterly data: regress changes in the monetary policy rate on current and one-year-ahead forecasts of GDP growth and inflation (from IMF WEO), lags of GDP growth and inflation, lagged interest rate, and lags of change in nominal effective exchange rate; residuals are interpreted as exogenous monetary policy shocks.
  - Rationale: financial sector fully anticipates systematic policy actions; residuals capture the non-systematic/unexpected component.
- Local projection implementation:
  - Estimate impulse responses of FCI (excluding the policy rate) to identified monetary policy shocks using Jordà (2005) local projections.
  - Horizons h = {0,1,2,...,8} (quarters); specification includes lags of FCI, four lags of changes in oil prices and commodity prices excluding fuel, and four lags of monetary shocks as controls.
  - FCI normalized by adding 100 (mean 100, variance 1) for interpretability; confidence bands computed using Newey-West standard errors.

### Impulse-response results: effects of contractionary shocks on FCIs
- General result: contractionary (tightening) monetary policy shocks lead to tighter financial conditions (positive and statistically significant FCI responses), with notable cross-country heterogeneity in magnitude, speed, and persistence.
- Magnitude and timing:
  - In economies with inflation-targeting frameworks (e.g., South Africa, Ghana, Kenya, Mauritius, Uganda), responses are larger, faster, and more persistent: a 100-basis-point monetary policy shock increases FCIs by 0.5 to 1 index point at peak.
  - Peak effects typically occur between three and six quarters after the shock and can remain evident up to two years (i.e., persistence up to eight quarters).
- Other regime patterns:
  - Monetary-aggregate-targeting regimes (e.g., Tanzania) show relatively strong effects similar in pattern to inflation-targeters but with smaller magnitude.
  - Economies with eclectic or other frameworks (e.g., Mozambique, Zambia) tend to exhibit smaller, short-lived, or muted responses.
  - Two monetary unions (CEMAC and WAEMU) display relatively strong but gradual responses (delayed peak).
- Robustness and comparison with using actual policy rate changes:
  - Re-estimating responses using actual policy rate changes yields qualitatively similar results for most countries, suggesting limited anticipatory bias in those cases.
  - In about one-third of the sample (notably Ghana, Mauritius, and Zambia), responses to actual policy rate changes are weaker and more gradual than responses to identified shocks—implying anticipatory or endogenous components can attenuate measured effects when using raw policy-rate changes.
  - In Nigeria, responses to actual policy rate changes appear somewhat stronger than to identified shocks, suggesting potential underestimation by the identification procedure or more exogenous actual rate movements.

### Interpretation, caveats, and transmission-channel considerations
- Sources of cross-country heterogeneity in transmission strength:
  - Evolution and credibility of monetary policy frameworks (transition to inflation targeting and strengthening of operational frameworks and interest-rate based procedures).
  - Financial market development and composition of FCI (indices including upstream indicators—market yields, asset prices—display stronger, faster responses; indices dominated by downstream quantity indicators—credit, monetary aggregates—show slower responses).
  - Supply-side vulnerabilities and inflation volatility (large food shares in consumption, agriculture exposure, imported fuel and food) complicate predictability and central bank communication.
  - Data limitations: lack of high-frequency market data and central bank forecasts constrain identification and can induce measurement error.
- Caveats highlighted by the authors:
  - Identified shocks rely on IMF WEO forecasts as proxies for central bank projections; this assumes similar information sets between central banks and IMF.
  - Estimates are sensitive to sample, specification choices, and the composition of FCIs.
  - Neutral rate estimates carry considerable uncertainty; combining methods reduces but does not eliminate model uncertainty.

### Conclusions and policy implications (as stated in the chapter)
- A comprehensive assessment of monetary policy stance in SSA economies should combine neutral rate estimates with measures of broader financial conditions, given impaired and heterogeneous transmission.
- Key empirical takeaways:
  - Neutral rates in most SSA economies peaked around 2007, then declined during the GFC and again during the pandemic; post-pandemic r-star estimates range between -0.8 percent and 4.6 percent as of end-2023.
  - Monetary policy stances varied considerably across SSA economies during common shocks; by end-2023 half the sample exhibited restrictive stances while the remainder were accommodative or neutral.
  - The relationship between interest rate gaps and FCIs is strongest in economies that have adopted or are transitioning to inflation-targeting frameworks (contemporaneous or leading correlations around 0.6).
  - Contractionary monetary policy shocks generally tighten financial conditions; a 100-basis-point shock typically raises FCIs by 0.5 to 1 index point at peak in economies with stronger transmission, with peak effects between three and six quarters and persistence up to two years.
- Policy-relevant implications emphasized:
  - Relying solely on the neutral rate is insufficient where transmission to financial conditions is weak; central banks should monitor broader financial conditions alongside policy-rate-based metrics.
  - Strengthening operational frameworks, improving central bank communication, and developing financial market infrastructure can enhance the speed and magnitude of monetary transmission.
  - Further research is needed to link transmission strength to structural impediments (financial system weaknesses and monetary policy framework gaps) and to analyze effects on specific transmission channels and the real economy.

*Source: IMF Working Paper — chapter "5. Impulse Responses of FCI to Monetary Policy Tightening" (authors’ calculations and figures based on quarterly data 2002Q1–2023Q4).*

### Box 1. Trade-Offs Faced by SSA EFM Central Banks Amid Global Shocks

### Box 1. Trade-Offs Faced by SSA EFM Central Banks Amid Global Shocks

### Stylized-fact analysis and scope
- Focus: behavior of policy rates during four major global episodes: the Global Financial Crisis (2008–09), the China slowdown (2016), the COVID-19 crisis (2020), and the economic fallout from Russia’s invasion of Ukraine (2022–23).
- Empirical approach: scatter plots (Box 1.1) showing changes in policy rates relative to the output gap and inflation gap; output gap measured as deviation of log real GDP from HP-filter trend; inflation gap defined as actual inflation minus official target (or proxied by sample average for countries without explicit target).
- Change in policy rate defined as difference in the main policy rate between the indicated year and two years earlier.

### Episode-by-episode findings
- Global Financial Crisis (2008–09)
  - Transmission: mostly indirect and delayed via weaker export demand, commodity prices, tourism receipts, remittances, and FDI; stronger financial linkages (e.g., South Africa) transmitted the shock through portfolio flows and banking-system disruption.
  - Central bank response: most central banks loosened monetary policy and lowered policy rates, even in countries with positive output gaps and relatively high inflation (examples: Uganda, Zambia).
  - Notable exception: Ghana raised the policy rate amid high inflation despite a negative output gap.

- China’s Slowdown (2016)
  - Transmission: slump in commodity prices, reduced China demand for African exports, and weaker capital inflows.
  - Exchange rate impacts: commodity-exporting countries with flexible exchange rates (examples: Ghana, South Africa, Zambia) experienced substantial currency depreciations.
  - Central bank response heterogeneity:
    - Tightening: commodity-exporting economies with inflation and exchange-rate pressures raised policy rates, particularly where output gaps were positive (examples: Mozambique, Nigeria, South Africa, Zambia). Ghana raised rates despite a negative output gap due to persistently high inflation.
    - Loosening or unchanged: countries with limited inflationary pressures lowered rates (examples: CEMAC, Mauritius) or left rates broadly unchanged (examples: WAEMU, Tanzania).

- COVID Crisis (2020)
  - Shock nature: simultaneous demand and supply shocks causing sharp contraction in economic activity.
  - Central bank response: across SSA, central banks cut policy rates—even with elevated inflation—and many implemented complementary measures (liquidity support via open market operations and purchases of government securities in secondary markets).

- Economic Fallout of Russia’s Invasion of Ukraine (2022–23)
  - Inflation drivers: global food and energy price spikes, compounded by COVID-era supply-chain disruptions and Russia’s invasion disrupting global agricultural commodity markets—particularly cereals and fertilizers.
  - Central bank response: raised policy rates to address persistent inflation despite negative output gaps, aiming to prevent inflation expectations from becoming unanchored.
  - Country heterogeneity: Ghana recorded the steepest rate hikes reflecting surging inflation; Tanzania kept rates unchanged corresponding to relatively stable price dynamics.

### Cross-cutting patterns and trade-offs
- Policy priority depends on shock type:
  - Global supply shocks (example: Russia’s invasion of Ukraine): SSA central banks tended to prioritize price stability and tightened policy in face of persistent inflationary pressures.
  - Global demand shocks (examples: GFC, COVID-19): growth stabilization predominated and most central banks adopted countercyclical easing.
- Responses reflect country-specific exposures and institutional differences:
  - Key shaping factors: heterogeneous effects on terms of trade, capital flows, exchange-rate pressures, and degree of pass-through to inflation and output.
  - Role of economic structure: distinctions between oil exporters (examples: CEMAC, Nigeria), non-oil resource exporters (examples: Ghana, Tanzania, South Africa, Zambia), and countries with stronger financial linkages to global markets (examples: Ghana, South Africa) influenced policy choices.
  - Institutional and mandate differences across central banks affected monetary policy reactions.

### Box 1.1 and data notes
- Box 1.1 presents scatter plots of output-gap and inflation-gap deviations against changes in policy rates for 2009, 2016, 2020, and 2023 across the following economies/regions: ZAF, GHA, KEN, MUS, MOZ, NGA, TZA, UGA, ZMB, CEMAC, WAEMU.
- Data sources: Haver, Authors’ calculations.
- Methodological notes:
  - Output gap: deviation of log real GDP from HP-filter trend.
  - Inflation gap: actual inflation minus official target (or explicit price stability objective); where no quantitative objective exists, inflation target proxied by sample average inflation.
  - Change in policy rate: difference between main policy rate in the indicated year and two years earlier.

*Source: IMF Working Paper — Box 1. Trade-Offs Faced by SSA EFM Central Banks Amid Global Shocks*

### 0.5 percentage points.

### wpiea2025160-source-pdf - 0.5 percentage points.

### Estimated Taylor Rule (Table A.3)
- Regression estimated by OLS for: 푖
௧ = 훼 + 훽
௜ 푖
௧ିଵ + 훽
గ (휋
௧ − 휋
௧
∗) + 훽
௬ 푦෤
௧ + 휀
௧, where 푖 is the nominal short term policy rate, 휋 is the inflation rate, 휋
∗ denotes the central bank’s desired level or inflation target, and 푦
௧ − 푦
௧
∗ is the HP-filtered output gap.
- The constant neutral interest rate is obtained by mapping estimates from this equation to equation (1).
- Standard errors shown in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.
- Reported coefficients and standard errors by country (columns (1)–(11)):
  - Lag 1 of nominal policy rate:
    - CEMAC: 0.969*** (0.0169)
    - GHA: 0.884*** (0.0323)
    - KEN: 0.804*** (0.0446)
    - MOZ: 0.863*** (0.0318)
    - MUS: 0.763*** (0.0379)
    - NGA: 0.904*** (0.0336)
    - ZAF: 0.938*** (0.0381)
    - ZMB: 0.625*** (0.0700)
    - ZWE: 0.928*** (0.0225)
    - AUA: 0.914*** (0.0255)
    - WAM: 0.775*** (0.0680)
  - Inflation gap:
    - CEMAC: 0.006520 (0.0105)
    - GHA: 0.0673*** (0.0155)
    - KEN: 0.207*** (0.0293)
    - MOZ: 0.232*** (0.0309)
    - MUS: 0.0345* (0.0205)
    - NGA: 0.0620*** (0.0214)
    - ZAF: 0.002600 (0.0480)
    - ZMB: 0.243*** (0.0542)
    - ZWE: 0.0317*** (0.00723)
    - AUA: 0.0603** (0.0265)
    - WAM: 0.0496 (0.0448)
  - Output gap:
    - CEMAC: -0.0146 (0.0154)
    - GHA: -0.0008950 (0.0789)
    - KEN: 0.310*** (0.0928)
    - MOZ: 0.008070 (0.142)
    - MUS: 0.0445** (0.0216)
    - NGA: 0.06660 (0.0734)
    - ZAF: 0.183 (0.279)
    - ZMB: -0.01440 (0.173)
    - ZWE: 0.0303* (0.0173)
    - AUA: 0.125*** (0.0296)
    - WAM: 0.0419 (0.167)
  - Constant:
    - CEMAC: 0.1201 (0.0757)
    - GHA: 1.633*** (0.558)
    - KEN: 1.317*** (0.425)
    - MOZ: 1.678*** (0.400)
    - MUS: 0.963*** (0.189)
    - NGA: 1.183*** (0.430)
    - ZAF: 0.6623 (0.439)
    - ZMB: 3.697*** (0.754)
    - ZWE: 0.192*** (0.0695)
    - AUA: 0.561*** (0.192)
    - WAM: 2.308*** (0.787)
  - Observations:
    - CEMAC: 919
    - GHA: 585
    - KEN: 576
    - MOZ: 727
    - MUS: 287
    - NGA: 959
    - ZAF: 692
    - ZMB: 987
    - ZWE: 878
    - AUA: (value appears as 878 in table context)
    - WAM: (value appears as 878 in table context)
  - R-squared:
    - CEMAC: 0.975
    - GHA: 0.948
    - KEN: 0.844
    - MOZ: 0.937
    - MUS: 0.890
    - NGA: 0.915
    - ZAF: 0.878
    - ZMB: 0.672
    - ZWE: 0.953
    - AUA: 0.949
    - WAM: 0.670

### Estimated Reduced-Form Model (Table A.4)
- Reduced-form OLS regression: 푟 = 훼 + 훽
଴ g + 훽
ଵ 푟
௪, where g denotes growth rate of the HP-filtered trend component of real GDP (potential growth), r is the ex-ante real short-term interest rate, and the U.S. real federal funds rate is used as the proxy for 푟
௧
௪.
- Standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.
- Reported coefficients and standard errors by country (columns (1)–(11)):
  - Trend GDP growth coefficient:
    - CEMAC: 0.369*** (0.0310)
    - GHA: -1.022*** (0.319)
    - KEN: 1.903* (1.018)
    - MOZ: -0.773*** (0.245)
    - MUS: 0.778*** (0.134)
    - NGA: 0.1345 (0.115)
    - ZAF: 5.573*** (0.670)
    - ZMB: -0.156 (0.325)
    - ZWE: -0.207*** (0.0373)
    - AUA: 0.777*** (0.135)
    - WAM: 0.425** (0.184)
  - ex ante US real policy rate coefficient:
    - CEMAC: 0.319*** (0.0308)
    - GHA: 0.223 (0.143)
    - KEN: 0.412*** (0.111)
    - MOZ: 0.469* (0.253)
    - MUS: 0.609*** (0.105)
    - NGA: 0.487*** (0.105)
    - ZAF: 0.632*** (0.152)
    - ZMB: 0.424*** (0.151)
    - ZWE: 0.002580 (0.0297)
    - AUA: 0.334*** (0.0988)
    - WAM: 0.625*** (0.116)
  - Constant:
    - CEMAC: 0.529*** (0.129)
    - GHA: 16.58*** (2.341)
    - KEN: -3.036 (4.599)
    - MOZ: 12.29*** (1.957)
    - MUS: -0.275 (0.583)
    - NGA: 5.539*** (1.430)
    - ZAF: -25.87*** (4.131)
    - ZMB: 7.878*** (2.217)
    - ZWE: 1.620*** (0.166)
    - AUA: 1.497*** (0.548)
    - WAM: 5.011*** (1.539)
  - Observations:
    - CEMAC: 888
    - GHA: 882
    - KEN: 887
    - MOZ: 708
    - MUS: 888
    - NGA: 888
    - ZAF: 888
    - ZMB: 888
    - ZWE: 888
    - AUA: 876
    - WAM: 876
  - R-squared:
    - CEMAC: 0.757
    - GHA: 0.114
    - KEN: 0.182
    - MOZ: 0.130
    - MUS: 0.508
    - NGA: 0.336
    - ZAF: 0.492
    - ZMB: 0.086
    - ZWE: 0.272
    - AUA: 0.397
    - WAM: 0.456

### Estimates of the Neutral Real Interest Rates – International Comparison (Figure A.3)
- Sources: IMF World Economic Outlook; IMF International Financial Statistics; The Federal Reserve Bank of New York; Haver and authors’ calculations.
- Notes: The average r-star is the mean of all the estimated r-stars across methodologies, excluding the Taylor rule. US/Euro area refers to the average of r-star for the United States and the euro area, based on Holston et. al. (2017) with updated estimates from the Federal Reserve Bank of New York.
- Time series panels shown for multiple jurisdictions with labeled axes from 2003–2023 or 2005–2023 depending on country; visual series include:
  - Mauritius: range from -3 to 4, years 2007–2023.
  - Mozambique: range from 0 to 10, years 2003–2023.
  - Nigeria: range from -1 to 6, years 2003–2023.
  - Tanzania: range from 0 to 10, years 2003–2023.
  - Uganda: range from 0 to 8, years 2003–2023.
  - WAEMU: range from 0 to 3, years 2003–2023.
  - South Africa: range from 0 to 6, years 2003–2023.
  - Zambia: range from -2 to 8, years 2005–2023.
  - CEMAC: range from 0 to 3, years 2003–2023.
  - Ghana: range from 0 to 14, years 2003–2023.
  - Kenya: range from 0 to 6, years 2003–2023.
- Each panel plots:
  - Average rstar
  - US/Euro area rstar average
  - Country-specific rstar series across the indicated years.

### Cross-Correlations of Interest Rate Gaps and FCIs (Figure A.4)
- Sources: IMF World Economic Outlook; IMF International Financial Statistics; Haver and authors’ calculations.
- Notes: Blue bars represent correlation coefficients between the FCIs (excluding monetary policy rates) and several lags and leads of the average interest rate gaps.
- Panels for each jurisdiction show correlation coefficients from -1.0 to 1.0 across lags/leads -6 to 6:
  - Mauritius
  - Mozambique
  - Nigeria
  - Tanzania
  - Uganda
  - WAEMU
  - South Africa
  - Zambia
  - CEMAC
  - Ghana
  - Kenya
- Visual patterning indicates positive and negative correlations at various lags/leads for different jurisdictions (exact numeric correlation coefficients are presented visually in the figure panels).

*IMF WORKING PAPERS Measuring Monetary Policy Stance in Sub-Saharan African Emerging and Frontier Markets — INTERNATIONAL MONETARY FUND*

### Annex B. Financial Conditions Indices for SSA

### Annex B. Financial Conditions Indices for SSA

### B.1. The selected financial variables
- Objective: construct FCIs that reflect the availability and terms of financing for economic agents in SSA by aggregating a broad set of indicators across key segments of SSA economies’ financial systems.
- Variable categories (Table B.1):
  - Price indicators (interest rates, yields): policy rate, lending rate, deposit rate, government treasury bill yield, government bond yield.
  - Quantity indicators: monetary base, broad money, credit to the private sector, credit to government.
  - Foreign exchange market: nominal effective exchange rate.
  - Global financial factors: Volatility index, VIX; Emerging Market Bond Index (EMBI) spread (aggregate EMBI sovereign spread index).
- Availability of specific indicators by country/area (as presented in Table B.1):
  - Policy rate: CEMAC, GHA, KEN, MUS, MOZ, NGA, TZA, UGA, WAEMU, ZAF, ZMB
  - Lending rate: KEN, MUS, MOZ, NGA, TZA, UGA, WAEMU, ZAF, ZMB
  - Deposit rate: GHA, KEN, MUS, MOZ, NGA, TZA, UGA, WAEMU, ZAF, ZMB
  - Government treasury bill yield: GHA, KEN, MUS, MOZ, NGA, TZA, UGA, ZAF, ZMB
  - Government bond yield: GHA, KEN, MUS, NGA, TZA, UGA, ZAF, ZMB
  - Monetary base: CEMAC, GHA, KEN, MUS, MOZ, NGA, TZA, UGA, WAEMU, ZAF, ZMB
  - Nominal effective exchange rate: CEMAC, GHA, KEN, MUS, MOZ, NGA, TZA, UGA, WAEMU, ZAF, ZMB
  - Volatility index, VIX: CEMAC, GHA, KEN, MUS, MOZ, NGA, TZA, UGA, WAEMU, ZAF, ZMB
  - Emerging market bond spreads, EMBI: (aggregate EMBI sovereign spread index included where available)
- Notes on interpretation of signs:
  - A positive sign indicates a tightening in the FCI; a negative sign indicates an easing.
  - An increase in the nominal effective exchange rate indicates an appreciation of the national currency.

### B.2. An index of financial conditions — construction methods
- Methods considered:
  - Weighted-sum approach: weights based on estimated relative impacts on real GDP (reduced-form aggregate demand, VAR impulse responses, simulations).
  - Principal-components analysis (PCA): extracts common factor(s) from many financial variables; first principal component interpreted as FCI-PCA.
- Baseline method used: Equal-weights FCI (FCI-EW)
  - Rationale: simplicity, ease of computation, interpretability; robust in contexts with limited/high-frequency data irregularities.
  - Data transformations:
    - Variables transformed so higher values indicate tighter financial conditions (quantity indicators reversed with negative sign).
    - Price indicators included in levels; quantity indicators expressed in growth rates (YoY percent change for monetary and credit aggregates).
    - Each variable standardized (subtract mean, divide by standard deviation) to yield z-scores.
    - Overall FCI standardized: zero reflects average/“normal” financial conditions over the sample period; positive = tighter-than-average; negative = looser-than-average.
  - Dataset contains 12 variables; number included per country varies with availability.
  - Supporting evidence: Arrigoni et al. (2022) show equal-weight averages can perform as well as PCA in large AE/EM samples.
- Robustness check: FCI-PCA
  - PCA extracts the first principal component (PC1) as FCI-PCA.
  - PC1 explains approximately 23 to 40 percent of the total variance across different economies (text statement).
  - In practice, the first principal component accounts for the following shares of total variance (Table B.2):
    - CEMAC: 34.05
    - GHA: 32.64
    - KEN: 32.71
    - MOZ: 34.14
    - MUS: 39.63
    - NGA: 26.20
    - TZA: 27.96
    - UGA: 33.42
    - WAEMU: 26.47
    - ZAF: 38.41
    - ZMB: 24.37

### B.3. Interpreting FCIs for SSA — dynamics, episodes, and robustness
- Key historical episodes captured by FCIs (quarterly, 2003Q1–2023Q4):
  - Global Financial Crisis (2008–09): sharp tightening in financial conditions across SSA driven by a rise in global financial risk indicators (VIX and EMBI spreads).
  - Post-GFC: generally more expansionary financial conditions across most countries.
  - Euro area sovereign debt crisis (2011–13): tightening in some countries (Kenya, Mozambique, Mauritius, Tanzania, Uganda) via higher interest rates, weaker growth in monetary and credit aggregates, and appreciating exchange rates.
  - 2016–17: spikes in FCIs due to China’s slowdown and the slump in commodity prices.
  - COVID-19 pandemic (2020–21): initial loosening in financial conditions as lower interest rates, depreciating exchange rates, and stronger growth in monetary and credit aggregates offset elevated global financial risks.
  - Post-pandemic (notably 2023): broad tightening as surging inflation prompted monetary tightening and slower growth in monetary aggregates; some easing observed in Mozambique, Tanzania, and Zambia.
- Robustness: comparison FCI-EW vs FCI-PCA
  - FCI-PCA and FCI-EW exhibit high correlation across economies and consistently capture the main episodes of loosening and tightening.
  - Marginal divergences arise from differences in PC factor loadings versus equal weights, affecting magnitude and occasionally sign.
- Comparison with other EMDE measures
  - SSA FCIs are relatively correlated with IMF GFSR Emerging Market Excluding China Index and Goldman Sachs Emerging Market Index, though magnitudes differ.
  - Other EMDEs tightened more sharply than SSA during the GFC — likely reflecting greater financial integration of other EMDEs.
  - EMDE indices show pandemic tightening followed by loosening; recent monetary tightening produced more pronounced tightening in other EMDEs than in SSA.

### Key quantitative details and technical notes
- Number of variables in dataset: 12 (subject to country availability)
- Variable transformations:
  - Price indicators: levels
  - Quantity indicators (monetary base, broad money, credit aggregates): YoY percent change
  - Quantity indicators entered with negative sign to align direction with tighter = higher FCI
- Standardization:
  - Individual variables standardized to zero mean and unit standard deviation over sample period (z-scores).
  - Overall FCI standardized (zero = average historical financial conditions).
- Share of total variance explained by first principal component (PC1) by economy (Table B.2 repeats):
  - CEMAC: 34.05
  - GHA: 32.64
  - KEN: 32.71
  - MOZ: 34.14
  - MUS: 39.63
  - NGA: 26.20
  - TZA: 27.96
  - UGA: 33.42
  - WAEMU: 26.47
  - ZAF: 38.41
  - ZMB: 24.37

*Source: Annex B. Financial Conditions Indices for SSA (from the provided IMF working paper content).*

### Annex C. Data Sources

### Annex C. Data Sources

### Table C.1: List of Countries
- Emerging markets
  - South Africa
  - Ghana
  - Kenya
  - Mauritius
  - Mozambique
  - Nigeria
  - Tanzania
  - Uganda
  - Zambia
- Frontier markets
  - Cameroon
  - Central African Rep.
  - Chad
  - Rep. of Congo
  - Equatorial Guinea
  - Gabon
- Monetary Unions
  - CEMAC
    - Cameroon
    - Central African Rep.
    - Chad
    - Republic of Congo
    - Equatorial Guinea
    - Gabon
  - WAEMU
    - Benin
    - Burkina Faso
    - Côte d’Ivoire
    - Guinea-Bissau
    - Mali
    - Niger
    - Senegal
    - Togo

### Table C.2: Classification of Exchange Rate Arrangements and Monetary Policy Frameworks in SSA Emerging and Frontier Markets
Sources: AREAER database 2023; and websites of central banks
Notes:
- 1/ Includes countries that have no explicitly stated nominal anchor, but rather monitor various indicators in conducting monetary policy.
- 2/ *: The policy rate is the main instrument to signal the stance of monetary policy. +: The central bank uses a variety of instruments to align the operating target with the policy rate.
- 3/ In percentage.
- 4/ The central bank is in transition toward inflation-targeting.

- Exchange rate anchors listed in the source:
  - US dollar
  - Euro
  - Composite
  - Other
  - Other 1/
- Monetary policy frameworks and associated annotations (as presented in source):
  - Inflation target
  - Price stability objective
  - Monetary aggregate target
  - Inflation-targeting framework (including transitional status 4/)
  - Conventional peg (14)
  - Central African Economic and Monetary Union (CEMAC) (6)
  - West African Economic and Monetary Union (WAEMU) (8)
- Selected country-level entries and figures (verbatim from source table rows):
  - Mozambique 4/ ** 7.9
  - Nigeria *+* 13.2
  - Tanzania *+* 5 6.6
  - Ghana ** 8 15.5
  - Kenya ** 5 7.7
  - Mauritius ** 2-5 4.8
  - South Africa ** 3-6 5.3
  - Uganda ** 5 6.4
  - Zambia ** 6-8 12.4
- Additional table fragments included in source:
  - Floating (3)
  - Crawl-like arrangement (3)
  - Stabilized arrangement (3)
  - Open market operations
  - Main instruments
  - Policy interest rate 2/
  - Exchange rate arrangement
  - Exchange rate anchor
  - Average inflation (2002-2023)
  - Price stability 3/ 2.72.3
  - 1-3

### Table C.3: Data Sources and Description
- Real GDP
  - Description: Real GDP in local currency units interpolated to quarterly frequency using the quadratic match method if quarterly series not available.
  - Source: Haver Analytics
- Nominal GDP
  - Description: Nominal GDP in local currency units interpolated to quarterly frequency using the quadratic match method if quarterly series not available.
  - Source: Haver Analytics
- Output gap
  - Description: Measured by the Hodrick-Prescott (HP)-filtered quarterly real GDP with a smoothing parameter λ=1600.
  - Source: IMF WEO
- Policy rate
  - Description: Nominal official/policy interest rate. The monetary policy instrument varies by country
  - Source: Haver Analytics
- NEER
  - Description: Nominal effective exchange rate
  - Source: IMF, IFS
- REER
  - Description: Real effective exchange rate, CPI based.
  - Source: IMF, IFS
- USD exchange rate
  - Description: Exchange rate; domestic currency per USD
  - Source: IMF, IFS
- CPI Inflation
  - Description: Consumer price index
  - Source: Haver Analytics
- Inflation expectations
  - Description: 1-year-ahead, 3-year-ahead and 5-year-ahead forecasts for CPI inflation. Data was interpolated to quarterly frequency using the quadratic match method.
  - Source: IMF WEO
- GDP growth expectations
  - Description: 1-year-ahead, 3-year-ahead and 5-year-ahead forecasts for GDP growth. Data was interpolated to quarterly frequency using the quadratic match method.
  - Source: IMF WEO
- Inflation target
  - Description: Inflation target/Historical average
  - Source: IMF AREAER Database
- Commodity prices, all index
  - Description: World commodity price index (2016=100), USD, includes both fuel and non-fuel price indices
  - Source: Datastream, IMF
- Commodity prices, exc. Fuel
  - Description: World commodity price index, excluding fuel (2016=100), USD, includes precious metal, food and beverages and industrial inputs price indices.
  - Source: Datastream, IMF
- Commodity food prices
  - Description: World food index, (2016=100), USD, includes cereal, vegetable oils, meat, seafood, sugar, and other food (apple (non-citrus fruit), bananas, chana (legumes), Fishmeal groundnuts, milk (dairy), tomato (veg) price indices.
  - Source: Datastream, IMF
- Oil prices
  - Description: Brent crude oil price in USD
  - Source: Datastream, IMF
- Bank lending rates
  - Description: The other depository corporations’ rate that usually meets the short- and medium-term financing needs of the private sector.
  - Source: IMF, IFS
- Bank deposit rates
  - Description: The rates offered to resident customers for demand, time, or savings deposits.
  - Source: IMF, IFS
- Money market interest rates
  - Description: The rate on short-term lending between financial institutions.
  - Source: IMF, IFS
- Government treasury bills
  - Description: The rate at which short-term government debt securities are issued or traded in the market.
  - Source: IMF, IFS, Datastream, Haver Analytics
- Government treasury bonds
  - Description: One or more series representing yields to maturity of government bonds or other bonds that would indicate longer term rates.
  - Source: IMF, IFS, Datastream, Haver Analytics
- Private sector credit
  - Description: Claims on private sector
  - Source: IMF, IFS
- Public sector credit
  - Description: Claims on central government
  - Source: IMF, IFS
- Monetary base
  - Description: Currency in circulation; Deposits with the central bank
  - Source: IMF, IFS
- Broad money
  - Description: M2
  - Source: IMF, IFS
- US Policy Rate
  - Description: Fed Funds Target Rate
  - Source: Haver Analytics
- Euro area Policy Rate
  - Description: Main refinancing rate
  - Source: Haver Analytics
- U.S. long-term interest rates
  - Description: Ten-year U.S. Treasury yield
  - Source: Haver Analytics
- Euro area long-term interest rates
  - Description: Ten-year German bund yield
  - Source: Haver Analytics
- VIX
  - Description: Chicago Board Options Exchange volatility index
  - Source: Haver Analytics
- EMBI
  - Description: Emerging Market Bond Index (JPM EMBI Global)
  - Source: Haver Analytics
- U.S., Euro area r-star
  - Description: Estimates based on Holston et. al. (2017)
  - Source: Federal Reserve Bank of New York
- FCI EM exc. China (GS)
  - Description: FCI emerging market excluding China
  - Source: Goldman Sachs
- FCI EM (GFSR)
  - Description: FCI emerging market
  - Source: IMF, Global Financial Stability Report

*Annex C. Data Sources — IMF Working Paper No. WP/2025/160*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025160-source-pdf.pdf_