## 27. Real non-commodity GDP Response

## Source details

**Canonical URL:** [27. Real non-commodity GDP Response](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021163-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021163-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021163-print-pdf.pdf.json)

---

### Overview
- Study focus: asymmetric response of the non-commodity economy to changes in commodity prices, with main focus on oil-exporting countries.
- Sample: 56 developing commodity exporters (of which 27 are oil-exporting) over 1970-2016.
- Key contextual numbers:
  - WTI spot oil price fell to less than $30 per barrel in February 2016, more than doubled by the end of October 2018, then slid in 2019 and more recently on the back of the COVID-19 pandemic.
  - Non-commodity, non-agricultural sectors made up an average of 58 percent of oil exporters’ real GDP in 2010, and 68 percent for non-oil commodity exporters.
  - These shares have increased 7 and 2 percentage points respectively since 1990.
  - Terms-of-trade increased 33 percent on average for oil exporters, and 3 percent for non-oil commodity exporters.

### Methodology
- Estimation approach: panel VAR following Broda (2004), Love and Zicchino (2006), Shen et al. (2015).
- Endogenous variables: changes in CTOT and changes in non-commodity output, allowing positive and negative CTOT changes (∆+ and ∆−).
- Lag order: q = 1 (selected by Bayesian Information Criterion).
- Fixed effects removal: Helmert forward-mean transformation (Arellano and Bover 1995).
- Exogenous control: global real GDP growth (preferred to time fixed effects).
- Identification assumption: exogenous movements in terms-of-trade (no single country large enough to affect world relative prices); Granger causality tests and exclusions used in robustness.
- Inference:
  - Impulse response functions with Monte Carlo simulations.
  - Confidence intervals by 5th and 95th percentiles from 500 bootstraps.
- Initial-condition interactions:
  - High public debt dummy = 1 if public debt-to-GDP > 50 percent at time of shock (27 percent of observations for oil exporters; 39 percent for the broader commodity-exporter group).
  - International reserves threshold set at the historical median.

### Key empirical findings — symmetric shocks
- A 1 percentage change in the commodity terms of trade index increases real non-commodity non-agricultural GDP by nearly 0.3 percent in the first period, with the impact fading quickly afterwards.
- Country-specific elasticity: ε_i = θ τ_i where θ is the impulse response (example numbers):
  - Gabon: net commodity exports share = 50 percent → elasticity = 0.15.
  - Nigeria: net commodity exports share = 10 percent → elasticity ≈ 0.03.
- Weaker pre-existing fundamentals reduce the response:
  - High public debt or low international reserves lower or render the response insignificant.

### Key empirical findings — asymmetric shocks
- Negative CTOT shocks have larger impacts on real non-commodity GDP than positive shocks.
  - The impact of a negative shock on non-commodity output after the first year is statistically and economically significant and nearly double the size of the (statistically insignificant) positive shock.
  - Asymmetry persists into the medium term; negative shocks show more persistence and larger cumulative effects.
- Role of public debt:
  - High-debt oil exporters exhibit much more acute asymmetry: negative shocks have deeper impacts, positive shocks remain statistically insignificant.
  - Low-debt countries: asymmetry present but less pronounced; positive shocks show a more favorable response.
- Role of international reserves:
  - High reserves can reverse asymmetry: positive shocks provoke a more impactful response in non-commodity output growth for high-reserve countries.
  - Low reserves: negative shocks are more than twice the size than for high-reserve countries, though the impact fades very quickly.

### Heterogeneity and conditional responses
- Conditional splits and comparisons:
  - Responses conditional on public debt (high debt vs low debt).
  - Responses conditional on international reserves (high reserves vs low reserves).
  - Responses after CTOT upswing vs after CTOT downswing.
  - Cross-commodity comparisons: commodity exporters, non-oil commodity exporters, diversified vs non-diversified exporters.
  - Temporal splits: pre-2000 vs post-2000; post-1990 regional subsamples.
- Regional post-1990 groups for oil exporters:
  - "SSA oil exporters post-1990": Angola, Chad, Republic of Congo, Equatorial Guinea, Gabon, Nigeria, South Sudan (also tested including Cameroon).
  - "Middle East & North Africa oil exporters".
  - "Latin America oil exporters".
  - "Eastern Europe & Central Asia oil exporters".
- Key heterogeneity results:
  - Response to positive shocks strongest following a period of CTOT increases; negative shocks more impactful after a period of CTOT decline.
  - Volatility of response greater for countries with high public debt and low international reserves.
  - Asymmetric response robust across most oil-exporting regions: Middle East & North Africa and sub-Saharan Africa exhibit the asymmetric response; Eastern Europe & Central Asia show the pattern with large standard errors; Latin America shows some indication of a slight asymmetry in the opposite direction.

### Variance decomposition and quantitative importance
- Forecast-error variance decomposition measures importance of CTOT shocks for non-commodity GDP:
  - Initial estimate: among oil exporters, 4.6 percent of non-commodity GDP variance is explained by CTOT shocks.
  - Adjusted estimate (removing years before oil discoveries): 11.8 percent.
- Comparative notes:
  - Previous literature (using real GDP including commodity sector) reported ranges from 10 percent to 40 percent depending on study.
- Cross-region variation:
  - Contribution lowest in the 1980s; largest in Latin America and sub-Saharan Africa; smallest in Middle East & North Africa.

### Robustness checks and sensitivity analyses
- Sample expansions and splits:
  - Expanding to all 56 commodity exporters preserves asymmetry but with smaller magnitude.
  - Non-oil commodity exporters: asymmetry appears to reverse, suggesting oil exporters drive main result.
  - Non-diversified commodity exporters: main result holds with smaller magnitude.
- Time periods:
  - Pre-2000 and post-2000 splits: main results hold; negative shocks in post-2000 have same magnitude but much longer-lasting impact.
- Shock-size thresholds:
  - Threshold > 2 percent: these shocks represent 51 percent of CTOT changes; median ‘big’ positive shock = 5.9% and median ‘big’ negative shock = -5.4 percent.
  - Threshold > 5 percent: these shocks represent 29 percent of CTOT changes; median ‘big’ positive shock = 8.7 percent and median ‘big’ negative shock = -10.6 percent.
  - Asymmetry persists across thresholds.
- Specification changes:
  - Using time dummies instead of global real GDP growth: results relatively insensitive.
  - Including agriculture in GDP measure: does not meaningfully alter results.
  - Excluding Libya since 2012: results very close to full-sample estimates.
  - Using fixed commodity weights over time: asymmetry slightly less pronounced but negative shocks still larger.
- Additional checks:
  - Granger causality tests: changes in real non-commodity output fail to Granger cause changes in CTOT; causation runs from CTOT to non-commodity output.
  - Appendix H: including log of real government consumption (HP-filter deviations) reduces sample size but maintains asymmetry; fiscal variable statistically insignificant and excluded from main specification.

### Data, definitions, and sample notes
- Dependent variable: non-commodity, non-agricultural GDP (UN Statistical Database).
- Commodity terms-of-trade index: weights based on prior 3-year trade patterns using COMTRADE and IMF data; time-varying country-specific weights τ.
- Oil exporters defined as countries with fuel export earnings > 50 percent of total export earnings for 2008-2012 (SITC 3).
- Public debt-to-GDP: General government gross debt, percent of fiscal year GDP (WEO).
- International reserves-to-GDP: reserves in US dollars as percent of nominal GDP in US dollars (WEO, IFS).
- Commodities included: coal, crude oil, natural gas; multiple metals, food items, and other primary products (full list in appendix).
- Diversification scoring: IMF Export Diversification Database ranges from 1 (most diversified) to 6.4 (least diversified); oil exporters score average 5.0, non-oil commodity exporters score 4.1.
- Country lists for oil and non-oil commodity exporters provided in Appendix A.

### Conclusions and policy implications
- Main findings:
  - Linkages between commodity and non-commodity sectors are important and asymmetric.
  - Negative CTOT shocks are more impactful on non-commodity output growth than positive developments, especially with weak macroeconomic fundamentals or after prolonged commodity price declines.
- Policy implications:
  - Assuming symmetric responses to busts and booms can lead to inadequate or overzealous policy responses.
  - Importance of strong pre-existing economic fundamentals: fiscal space and international reserve buffers mitigate adverse effects and can alter asymmetry.
- Directions for future work:
  - Investigate channels of asymmetric transmission between commodity and non-commodity sectors.
  - Explore policy responses — both automatic and discretionary — including fiscal policy reaction to price shocks.
  - Examine impacts of structural changes in oil- and commodity-exporting countries.

*Source: wpiea2021163-print-pdf.*

### 1. Oil Exporters: Cumulative Change in Annual Real Non-Commodity, Non-Agricultural GDP

### 1. Oil Exporters: Cumulative Change in Annual Real Non-Commodity, Non-Agricultural GDP

### Cumulative change overview (1970-2016)
- Figure titled "Growth, 1970-2016" presents cumulative change in annual real non-commodity, non-agricultural GDP for oil exporters covering 1970-2016.
- A companion figure isolates "Large shocks" for the same period, indicating analyses that distinguish typical episodes from those with large commodity terms-of-trade or price shocks.

### Dynamic responses of real non-commodity, non-agricultural GDP
- Multiple impulse-response–style figures labeled "Real non-commodity, non-agricultural GDP Response" present the dynamic path of non-commodity, non-agricultural GDP following shocks.
- Variants include:
  - Responses conditional on the level of public debt.
  - Responses conditional on the level of international reserves.
  - Cumulative response plots aggregating period-by-period impacts.

### Heterogeneity by fiscal and external positions
- Separate response figures assess contrasts by:
  - "high debt" versus "low debt".
  - "high reserves" versus "low reserves".
- These figures enable comparison of the GDP response to shocks contingent on fiscal space and reserve buffers.

### Responses around commodity terms-of-trade (CTOT) cycles
- Figures labeled "after CTOT upswing" and "after CTOT downswing" capture how the non-commodity, non-agricultural GDP response differs depending on whether the economy is in the upswing or downswing phase of commodity terms-of-trade.

### Cross-commodity and diversification comparisons
- Response figures compare:
  - "commodity exporters" broadly.
  - "non-oil commodity exporters".
  - "diversified commodity exporters".
  - "non-diversified commodity exporters".
- These allow assessment of whether oil-exporter dynamics differ from other commodity-exporting country groups and by export concentration.

### Period splits and shock magnitudes
- Temporal subsamples:
  - "pre-2000" and "post-2000" response figures examine whether the relationship changed across these periods.
  - Regional and post-1990 subsamples are provided for several regions (see Regional section).
- Shock-size splits:
  - Responses to "large CTOT shocks (>2%)" and to "large CTOT shocks (>5%)" are shown separately, facilitating comparison of medium versus very large shocks.

### Regional and subgroup analyses (post-1990)
- Country-group specific response figures for oil exporters post-1990:
  - "SSA oil exporters post-1990"
  - "Middle East & North Africa oil exporters post-1990"
  - "Latin America oil exporters post-1990"
  - "Eastern Europe & Central Asia oil exporters post-1990"
- These figures support assessment of regional heterogeneity in post-1990 dynamics.

### Robustness checks
- A figure "using year fixed effects" presents a specification that controls for common time shocks, serving as a robustness check on baseline response estimates.

*Source: wpiea2021163-print-pdf - 1. Oil Exporters: Cumulative Change in Annual Real Non-Commodity, Non-Agricultural GDP*

### 27. Real non-commodity GDP Response.....................................................................................

### 27. Real non-commodity GDP Response

### Introduction
- Study focus: asymmetric response of the non-commodity economy to changes in commodity prices, with main focus on oil-exporting countries.
- Sample: 56 developing commodity exporters (of which 27 are oil-exporting) over 1970-2016.
- Contextual facts:
  - WTI spot oil price fell to less than $30 per barrel in February 2016, more than doubled by the end of October 2018, then slid in 2019 and more recently on the back of the COVID-19 pandemic.
  - Question: how quickly will oil price recovery translate into growth in non-commodity sectors?

### Literature survey — key channels and initial conditions
- Initial conditions that mitigate/amplify shocks:
  - International reserve buffers can smooth external adjustment to commodity terms-of-trade shocks (e.g., Grigoli 2017).
  - Fiscal space (low public debt-to-GDP) allows countercyclical policy; evidence that greater fiscal space dampens shock effects (Roch 2017; Kinda 2018; Grigoli 2017).
  - Flexible exchange rate regimes and institutional quality may smooth shocks; lower financial development and procyclical investment may exacerbate volatility.
- Transmission mechanisms from commodity sector to real economy:
  - Fiscal channel: volatile commodity revenues affect government revenues and fiscal policy (Gelb et al. 1998; Van der Ploeg and Poelhekke 2008; Spatafora 2012).
  - Financial sector: booms lower risk premia and can lead to excessive credit growth; busts raise non-performing loans and banking risks (Kinda 2010; Eberhardt and Presiberto 2018).
  - Demand effects and spillovers to transportation, logistics, and investment (Christensen 2016; Gruss 2014; Fornero et al. 2016).
- Asymmetry evidence in literature:
  - Mixed findings: some studies find larger impacts from negative shocks (Dehn 2000; Edwards and Levy-Yeyati 2004), others find larger positive-shock impacts or no asymmetry depending on sample and regime (Camacho and Perez-Quiros 2013; Broda 2004; Herrera et al. 2015).

### Contribution of this paper
- Uses Net Commodity Price Index (NCPI) / country-specific Commodity Terms of Trade (CTOT) to capture country-specific net exports share and changing price baskets.
- Focuses on non-commodity, non-agricultural GDP to isolate responses in remainder of economy.
- Examines role of initial conditions: public debt (stock preferred over fiscal balance) and international reserves.
- Investigates cyclical variation: allows different responses depending on prior period being an upswing or downswing in CTOT.

### Data & stylized facts
- Data: annual data, UN classification of primary commodity exporters; oil exporters defined as countries with fuel export earnings > 50 percent of total export earnings for 2008-2012 (SITC 3).
- Dependent variable: non-commodity, non-agriculture GDP (UN Statistical Database).
- Shares and trends:
  - Non-commodity, non-agricultural sectors made up an average of 58 percent of oil exporters’ real GDP in 2010, and 68 percent for non-oil commodity exporters.
  - These shares have increased 7 and 2 percentage points respectively since 1990.
  - Terms-of-trade increased 33 percent on average for oil exporters, and 3 percent for non-oil commodity exporters.
- Stylized event-study evidence:
  - For oil exporters, a 1 percent increase in the index appears to decrease non-commodity non-agricultural real GDP by 0.6 percent for negative shocks, three times more than for a positive shock (figures cited in text).
  - Asymmetry disappears and slightly reverses for non-oil commodity exporters (explored in robustness checks).

### Methodology
- Estimation approach: panel VAR following Broda (2004), Love and Zicchino (2006), Shen et al. (2015).
- Endogenous variables: vector including changes in CTOT and changes in non-commodity output; specification allows splitting positive and negative CTOT changes (∆+ and ∆− components) with indicator functions.
- Lag order: q = 1 (selected by Bayesian Information Criterion).
- Fixed effects: country-specific fixed effects removed via Helmert forward-mean transformation (Arellano and Bover 1995) to preserve orthogonality and allow system GMM.
- Exogenous control: global real GDP growth used to capture common macroeconomic shocks (preferred to time fixed effects).
- Identification: assumes exogenous movements in terms-of-trade (no single country large enough to affect world relative prices); Granger causality tests and robustness excluding large oil exporters discussed in appendices.
- Inference:
  - Impulse response functions with Monte Carlo simulations based on estimated coefficients and standard errors.
  - Confidence intervals defined by the 5th and 95th percentiles of distributions generated by 500 bootstraps.
  - Variance decompositions used to assess percent of non-commodity GDP variation explained by CTOT shocks.
- Initial-condition interactions:
  - High public debt dummy = 1 if public debt-to-GDP > 50 percent at time of shock (27 percent of observations for oil exporters; 39 percent for the broader commodity-exporter group).
  - Low debt dummy analogous for debt < 50 percent.
  - International reserves threshold set at the historical median (same threshold as IMF Regional Economic Outlook: Sub-Saharan Africa 2015).

### Empirical findings — symmetric shocks
- Main symmetric-result: a 1 percentage change in the commodity terms of trade index increases real non-commodity non-agricultural GDP by nearly 0.3 percent in the first period, with the impact fading quickly afterwards.
- Country-specific elasticity: multiply the 0.3 percent response by the country-specific net export share of GDP (τ_i,j,t). Examples:
  - Gabon: net commodity exports share = 50 percent → elasticity = 0.15.
  - Nigeria: net commodity exports share = 10 percent → elasticity ≈ 0.03.
- Weaker pre-existing fundamentals reduce the response:
  - High public debt or low international reserves lower or render the response insignificant, suggesting reserves and fiscal space help smooth shocks.

### Empirical findings — asymmetric shocks
- Asymmetry documented: negative CTOT shocks have larger impacts on real non-commodity GDP than positive shocks.
  - The impact of a negative commodity terms-of-trade shock on non-commodity output after the first year is statistically and economically significant and nearly double the size of the (statistically insignificant) positive shock.
  - This asymmetry persists into the medium term; negative shocks show more persistence and produce larger cumulative effects.
- Role of public debt:
  - High-debt oil exporters exhibit a much more acute asymmetry: negative shocks have deeper impacts, while positive shocks remain statistically insignificant.
  - In low-debt countries the asymmetry is present but less pronounced, and positive shocks show a more favorable response.
  - Interpretation: high-debt countries have less fiscal space to pursue countercyclical policies or to borrow on favorable terms when faced with negative shocks.
- Variance decomposition:
  - Among oil exporters, CTOT shocks have historically explained up to 12 percent of non-commodity GDP variance for years after oil discoveries.
  - Contribution of CTOT shocks lowest in the 1980s (possibly reflecting reduced integration between sectors), largest in Latin America and sub-Saharan Africa, smallest in the Middle East and North Africa.

### Additional empirical nuances
- Preceding CTOT dynamics matter:
  - Response to positive shocks is strongest following a period of CTOT increases rather than after a period of decline.
  - Negative shocks are more impactful after a decline in CTOT.
- Volatility of response is greater for countries with high public debt and low international reserves.
- Robustness: results for oil exporters are supported by various robustness tests (appendices referenced).

*Source: wpiea2021163-print-pdf - 27. Real non-commodity GDP Response.*

### Appendix H tests for the significance of real government consumption directly but the approach reduced data

### wpiea2021163-print-pdf - Appendix H tests for the significance of real government consumption directly but the approach reduced data availability and sample size, so that the paper proceeds without the statistically insignificant variable.

### Asymmetric impact of commodity terms-of-trade shocks
- Negative commodity terms-of-trade shocks are more impactful on non-commodity output growth than positive shocks.
- For countries with relatively high levels of reserves:
  - There is a reversal in the asymmetry: positive shocks provoke a more impactful response in non-commodity output growth (Figure 10).
- For countries with low reserves:
  - Negative shocks are more than twice the size than for countries with high reserves, though the impact fades very quickly (Figure 11).
- Interpretation:
  - International reserve buffers allow countries to respond to negative shocks by damping direct transmission and smoothing adjustment of external relative prices.
  - The reserves dummy may capture a broader sense of macroeconomic stability; stronger fundamentals could explain why positive shocks yield larger benefits in high-reserve environments.

### Effect of prior commodity terms-of-trade developments
- The impact on output growth depends on the development of commodity terms-of-trade (CTOT) prior to the shock.
- When shocks follow a period of increasing CTOT (aggregate increase in the CTOT index over the prior 5-year period):
  - Response function is relatively symmetric (Figure 12).
- When shocks follow a 5-year aggregate decrease in CTOT:
  - Asymmetry prevails, with negative shocks producing larger adverse effects (Figure 13).
- Interpretation:
  - A long period of commodity price decline weakens an oil-exporting economy beyond the direct impact on the commodity sector.

### Robustness checks — summary of sensitivity analyses
- Core result: output responses to negative shocks are larger than to positive shocks; this holds across multiple specifications, samples, and definitions of shocks.

A. Commodity exporters
- Expanding the sample to all 56 commodity exporters (including 29 non-oil commodity exporters) preserves the asymmetry, though the magnitude and extent are smaller than for oil exporters (Figure 14).
- When testing only non-oil commodity exporters, the asymmetry appears to reverse, suggesting oil exporters drive the main result (Figure 15).
- Sample split by diversification (IMF Export Diversification Index):
  - Non-diversified commodity exporters: main result holds but with smaller magnitude, suggesting oil dependence is a primary driver.
  - Of the 27 non-diversified commodity exporters, only one third are not oil exporters (Burundi, Central African Republic, Guinea, Liberia, Malawi, Mali, Mauritania, Suriname, Zambia).

B. Time periods
- Splitting sample between pre-2000 and post-2000:
  - Main results hold.
  - Negative shocks in the post-2000 period have the same magnitude but a much longer-lasting impact, possibly reflecting stronger transmission channels including through the financial sector (Figures 18 and 19).

C. Size of shocks
- Redefining shocks to exclude small changes (to reduce noise) does not alter results significantly:
  - Threshold greater than 2 percent:
    - These shocks represent 51 percent of the CTOT changes.
    - Median ‘big’ positive shock is 5.9% and median ‘big’ negative shock is -5.4 percent.
    - Outcome similar to baseline with slightly greater statistical significance (Figure 20).
  - Threshold greater than 5 percent:
    - The ‘big’ shocks represent 29 percent of all CTOT changes.
    - Median ‘big’ positive shock is 8.7 percent and median ‘big’ negative shock is -10.6 percent.
    - Asymmetry persists but to a narrower extent (Figure 21).

D. Geographical analysis
- Asymmetric response robust across most oil-exporting regions:
  - Middle East and North Africa and sub-Saharan Africa exhibit the asymmetric response; together they make up two-thirds of the oil exporters in the sample.
  - Eastern Europe and Central Asia show the pattern but with large standard errors.
  - Latin America shows some indication of a slight asymmetry in the opposite direction, consistent with Camacho and Perez (2014).
- Note on Sub-Saharan Africa sample: taken from post-1990 as prior to this period oil production was very low in many component countries.

E. Specification changes
- Using time dummies instead of global real GDP growth as an exogenous variable:
  - Results are relatively insensitive to this change (Figure 26).
- Including agriculture in the GDP measure:
  - Does not meaningfully alter results (Figure 27).
- Excluding Libya due to data concerns since 2012:
  - Results are very close to the full-sample estimates.
- Using an index with fixed country commodity weights over time (instead of time-varying weights):
  - Asymmetry is slightly less pronounced but negative price shocks still have larger impact.
  - Authors prefer time-varying weights because fixed weights risk misrepresenting changing commodity importance over time.

### Conclusion and implications
- Main findings:
  - Linkages between commodity and non-commodity sectors are important and asymmetric.
  - Negative commodity terms-of-trade shocks are more impactful on non-commodity output growth than positive developments, especially with weak macroeconomic fundamentals or after prolonged commodity price declines.
  - Spillovers from the oil economy to the non-oil side must be carefully understood to design appropriate policy responses.
- Policy implications:
  - Assuming symmetric responses to busts and booms can lead to inadequate or overzealous policy responses.
  - Emphasizes the importance of strong pre-existing economic fundamentals.
- Directions for future work:
  - Investigate channels of asymmetric transmission between commodity and non-commodity sectors.
  - Explore policy responses — both automatic and discretionary — for example, how fiscal policy responds to price shocks.
  - Examine the impact of structural changes in oil- and commodity-exporting countries.

### Appendix A: Data, definitions, and key sample characteristics
- Construction of non-commodity, non-agricultural GDP:
  - Real GDP 2010 constant prices, local currency (IMF WEO database), multiplied by share of non-commodity, non-agricultural GDP.
  - Excludes accompanying categories “Agriculture, hunting, forestry, fishing” and “Mining and utilities” where Mining includes fuel and other primary products.
  - Robustness checks using non-commodity GDP (including agriculture) give qualitatively similar results.
- Mining and utilities share in resource-poor countries typically 1.5-2.5 percent of GDP over time and very stable.
- Data definitions:
  - Commodity terms-of-trade index: weights based on prior 3-year trade patterns using COMTRADE and IMF data (Gruss and Kebhaj (2019) methodology).
  - Exchange rate classification: Ilzetzki, Reinhart and Rogoff (2017) de facto classification; fixed regime if score 1 or 2 on “coarse” classification.
  - Public debt-to-GDP ratio: General government gross debt, percent of fiscal year GDP (WEO).
  - International reserves-to-GDP ratio: International reserves in US dollars taken as percent of nominal GDP in US dollars (WEO, IFS).
  - Commodity revenue/GDP: General government commodity-related revenues (WEO).
- Commodities included in CTOT index:
  - Coal, crude oil, natural gas; aluminum, copper, iron ore, lead, nickel, tin, and zinc; bananas, barley, beef, cocoa, coconut oil, coffee, corn, fish, fish meal, groundnuts, lamb, oranges, palm oil, poultry, rice, shrimp, soybean meal, soybean oil, soybeans, sugar, sunflower oil, tea, wheat; cotton, hardwood logs and sawn wood, hides, rubber, softwood logs and sawn wood, soybean meal, wool.
- Diversification scores:
  - IMF Export Diversification Database scoring ranges from 1 (most diversified) to 6.4 (least diversified).
  - Oil exporters score an average 5.0 over the sample period whereas non-oil commodity exporters score 4.1.

### Appendix A: Country group lists (as provided)
- Oil exporters:
  - Algeria, Angola, Azerbaijan, Bahrain, Bolivia, Brunei, Chad, Republic of Congo, Ecuador, Equatorial Guinea, Gabon, Iran, Iraq, Kazakhstan, Kuwait, Libya, Nigeria, Oman, Qatar, Russia, Saudi Arabia, South Sudan, Trinidad and Tobago, Turkmenistan, United Arab Emirates, Venezuela, Yemen.
- Non-oil commodity exporters:
  - Afghanistan, Argentina, Burkina Faso, Burundi, Central African Republic, Chile, Côte d’Ivoire, Democratic Republic of Congo, Eritrea, Guinea-Bissau, Guinea, Guyana, Lao P. D. R, Liberia, Malawi, Mali, Mauritania, Mongolia, Morocco, Papua New Guinea, Paraguay, Sierra Leone, Solomon Islands, South Africa, Sudan, Suriname, Tuvalu, Uruguay, Zambia.
- Sub-Saharan African (SSA) oil exporters:
  - Angola, Chad, Republic of Congo, Equatorial Guinea, Gabon, Nigeria, South Sudan (also tested including Cameroon).
- Diversified commodity exporters (per IMF Export Diversification Database):
  - Afghanistan, Argentina, Bahrain, Bolivia, Burkina Faso, Chile, Côte d’Ivoire, Democratic Republic of Congo, Ecuador, Eritrea, Guinea-Bissau, Guyana, Lao P. D. R., Mongolia, Morocco, Papua New Guinea, Paraguay, Sierra Leone, Solomon Islands, South Africa, Sudan, Tuvalu, Uruguay.

*Source: Authors’ estimates drawn from the referenced IMF working paper content.*

### Appendix B. Asymmetric Response to Commodity Price Shocks by

### Appendix B. Asymmetric Response to Commodity Price Shocks by Commodity Exporters

### Commodity exporters — estimated GDP responses
- Figures present cumulative change in annual real non-commodity, non-agricultural GDP growth relative to the three-year period prior to:
  - a normalized 1 percent shock.
  - a normalized 5 percent shock (figures report only shocks larger than 5 percent).
- Source for figures: Authors’ estimates.

### Non-oil commodity exporters — estimated GDP responses
- Figures present cumulative change in annual real non-commodity, non-agricultural GDP growth relative to the three-year period prior to:
  - a normalized 1 percent shock.
  - a normalized 5 percent shock (figures report only shocks larger than 5 percent).
- Source for figures: Authors’ estimates.

### Stationarity and econometric specification (Appendix D)
- Unit-root testing approach:
  - Pesaran (2007) panel unit-root test used because it accommodates heterogeneous panels with cross-sectional dependence; Pesaran (2004) cross-sectional dependence test confirms cross-sectional dependence.
  - Augments Augmented Dickey-Fuller tests with cross-sectional averages of lagged levels and first differences.
- Findings:
  - For the broader commodity sample, the panel test does not reject the null hypothesis of non-stationarity for levels of real non-commodity, non-agricultural GDP, but does reject non-stationarity for commodity terms of trade.
  - For oil exporters, the reverse holds: non-stationarity of the commodity terms-of-trade index cannot be rejected.
  - Panel cointegration tests are not fully conclusive for the null hypothesis of zero cointegration.
- Implementation decision:
  - Main specifications use first differences (stationary) of the endogenous variables.
  - Table A4 indicates first differences of the endogenous variables are stationary.

### Exogeneity of terms-of-trade and Granger causality (Appendix E)
- Heuristic for influence on commodity price through own production: country must exceed 15 percent of global export share.
  - Table 2 (2017) shows Saudi Arabia at 15.9 percent — the only country breaching the 15 percent threshold.
  - OPEC responsible for 44 percent of global oil production in 2016 (US Energy Information Administration data cited in text).
- Robustness checks for exogeneity concerns:
  - Results robust to (i) exclusion of Saudi Arabia from the sample; and (ii) lowering production threshold to 10 percent (which would exclude Russia).
- Construction of commodity terms-of-trade index mitigates endogeneity concerns:
  - Weights τ use three-year rolling averages of trade values, lagged by one year.
  - Changes in commodity price index reflect contemporaneous price variations rather than volume changes.
- Measuring output in non-commodity GDP reduces likelihood of endogeneity:
  - Commodity sector excluded from non-commodity GDP for oil exporters.
  - Granger causality tests (Table A5): changes in real non-commodity output fail to Granger cause changes in the terms-of-trade; Granger causation runs from terms-of-trade to non-commodity output.
- Source for exogeneity tests and tables: Authors’ estimates.

### Elasticity derivation (Appendix F)
- Definition:
  - Let θ be the output of the impulse response function: θ = ∂ l / ∂ l_cTOT (notation as in source).
  - All variables are in logs; using CTOT formula yields country-specific elasticity ε_i:
    - ε = ∂ l / ∂ P = (∂ l / ∂ l_cTOT) * (∂ l_cTOT / ∂ P) = θ τ
    - Therefore ε_i = θ τ_i
- Numerical example from text:
  - If the impulse response shows a 0.3 response in year 1, and a country has a net commodity export share of GDP of 0.2, then a 10 percent increase in commodity prices leads to 10%*0.3*0.2 = 0.6 percent increase in non-commodity real GDP growth in year 1.

### Variance decomposition of non-commodity GDP (Appendix G)
- Purpose:
  - Forecast-error variance decomposition to measure importance of commodity terms-of-trade shocks for non-commodity economy of oil-exporting countries.
- Key quantitative findings:
  - Initial estimates: among oil exporters, 4.6 percent of non-commodity GDP variance is explained by commodity terms-of-trade shocks.
  - Adjusted estimate (removing years before oil discoveries in country histories): 11.8 percent.
    - Example adjustment: Equatorial Guinea not classified as oil exporter prior to 1995 discovery.
- Comparison to previous literature (using real GDP, which includes commodity sector):
  - Previous estimates range from 10 percent (Schmitt-Grohe, 2015) to 20-30 percent (Kose, 2002; Broda, 2004) and as high as 40 percent (Drechsel and Tenreyro, 2017; Fernandez et al., 2015).
- Methodological notes:
  - Table G1: each figure represents percentage of total short- or long-run variance of real non-commodity, non-agriculture GDP explained by commodity terms-of-trade shocks.
  - Decompositions based on Cholesky forecast-error variance decomposition using 100 Monte Carlo simulations.
  - Short-run decompositions: two-period-ahead forecast.
  - Long-run decompositions: ten-period-ahead forecast.
  - Adjusted full sample excludes country-years where commodity revenues-to-GDP is below 10 percent, or commodity revenue data is missing.
- Cross-region variation:
  - Contribution of commodity terms-of-trade shocks lowest in early sample years; increases over time possibly due to increased global integration and growth of banking sector linking commodity and non-commodity sectors.
  - Largest contributions observed in Latin America (small sub-group) and sub-Saharan Africa.
  - Smallest contribution observed in the Middle East and North Africa; possible explanation: large sovereign wealth funds acting as stabilizers.
- Source for variance decomposition and notes: Authors’ estimates.

### Robustness check including real government consumption (Appendix H)
- Purpose:
  - Test whether terms-of-trade shocks are confounded with fiscal shocks by including a fiscal variable in the panel VAR.
- Specification:
  - Include log of real government consumption, defining fiscal shocks as deviations from trend using Hodrick-Prescott filter.
- Data considerations and outcome:
  - Sample size reduced due to data availability.
  - Asymmetry finding and magnitude of response are maintained (results shown in Appendix E of source).
  - Given invariance and to preserve full country set, the statistically insignificant fiscal variable is excluded from the main specification.
- Visualization note:
  - Figures include light lines highlighting 95 percent confidence interval.
- Source for robustness check: Authors’ estimates.

*Source: Authors’ estimates (contents of wpiea2021163-print-pdf).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021163-print-pdf.pdf_