## _wp12115

## Source details

**Canonical URL:** [_wp12115](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12115.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12115.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12115.pdf.json)

---

### Data description (Section 4)
- Section 4 describes the salient features of the data.
- Monthly frequency sampling.
- Oil supply: world crude oil production (excluding natural gas liquids) — U.S. Energy Information Agency.
- Base metal supply: production of refined products — World Metal Bulletin Statistics.
- Apparent consumption defined as domestic production + imports − exports.
- RoW industrial production: PPP GDP-weighted aggregates based on national indexes.
- China industrial production and apparent consumption seasonally adjusted using U.S. Census Bureau’s X-12; U.S. series already seasonally adjusted.
- Ex-post real interest rate: CPI-deflated Federal Funds effective interest rate monthly average.
- U.S. dollar REER sourced from IMF.
- Commodity prices: U.S. dollar spot prices reported by IMF, deflated by U.S. CPI.
- Sample period: January 2000 to September 2011 (exception: crude oil apparent consumption begins January 2002; crude oil estimations run excluding this variable from January 2000 with quantitatively similar results).
- Unit root testing:
  - Over 2000–2011, cannot reject null that log level of each variable (except tin production) is non-stationary.
  - First differences of logs for all variables show clear evidence of stationarity.
- Lag length selection:
  - Information criteria generally selected lag lengths between 3 and 6 months.
  - Most results reported use lag = 5 months.

### China’s role in world commodity markets — Long-term structural trends (Section II.A)
- China’s consumption during 2010 accounted for about 20 percent of non-renewable energy resources, 23 percent of major agricultural crops, and 40 percent of base metals.
- Market shares have increased sharply since 2000; historical pattern described as the S-curve.
- China’s commodity intensity of demand is unusually high:
  - China consumes about 35 percent more energy than Korea and twice the level of Brazil at comparable income levels.
  - For base metals, China consumes significantly more than Korea and Brazil at the same income.
- Rapid expansion in the tradable export sector and large-scale fixed asset investment—particularly since 2000—contributed to fast-growing commodity intensity (Yu, 2011).
  - Ye (2008) estimates just over ½ of China’s copper usage is accounted for by infrastructure investment and construction, with ⅓ accounted for by consumer and industrial goods.
- Structural factors and domestic policy distortions highlighted in previous studies (IMF, 2011a) contribute to China’s growth structure and high commodity intensity.
- China’s strategic policy decisions affect global trade patterns (example: self-sufficiency in key grains vs. reliance on oilseed imports).
- China’s relative role by commodity group:
  - Dominant in base metals markets and to a lesser extent agricultural raw materials.
  - Has not yet assumed a large role in global food and energy markets, though its share of world imports is rising gradually.

### Short-term fluctuations (Section II.B)
- China’s high-frequency influence on commodity markets reflects business cycle, seasonality, and unanticipated transitory changes in supply-demand balance.
- Unanticipated shocks generally have larger market impact than anticipated fluctuations because inventories can smooth anticipated changes.
- Two types of demand shocks considered:
  1. Aggregate economic activity shocks — measured by industrial production.
  2. Commodity-specific demand shocks — measured by apparent consumption (domestic production + imports − exports).
- Important examples of commodity-specific demand shocks in China:
  - State inventory operations: China’s State Reserve Bureau believed to have purchased about 235,000 tons of copper during Q1 2009, accounting for about 65 percent of copper inventories held in London and Shanghai Exchange warehouses at end-2008 (Financial Times, 2009).
  - Temporary demand substitution (e.g., electric power shortages increasing oil and diesel demand for generators).
- Aggregate activity measured by industrial production is preferred to GDP for monthly-frequency analysis due to data availability.

### Econometric methodology (Section III)
- Baseline reduced-form VAR (recursive identification) follows Kilian (2009) and Helbling (2012).
- Seven endogenous variables in the baseline VAR (first differences of logs except real interest rate which is first-differenced):
  - world primary production of commodity (QW)
  - world excluding country i industrial production (XRoW)
  - country i industrial production (Xi)
  - country i apparent consumption (Ci)
  - the real short-term U.S. interest rate (R)
  - the U.S. dollar real effective exchange rate (REER)
  - the real price of the commodity (P/PUS)
- Vector Z: (ln QW, ln XRoW, ln Xi, ln Ci, R, ln REER, ln P/PUS)′ (first differences applied).
- Recursive ordering identifies structural shocks: supply shock; non-China aggregate activity shock (RoW); China aggregate activity shock; China commodity-specific demand shock; exchange rate shock; real commodity price shock (interpreted as commodity-specific demand shock in rest of world).
- Short-run identification restrictions include:
  - commodity supply curve vertical in the very short run;
  - RoW industrial production unaffected contemporaneously by China output shocks;
  - China industrial production responds with a lag to REER and real commodity price shocks;
  - aggregate activity shocks affect apparent consumption contemporaneously, but not vice versa within the same month;
  - REER does not respond contemporaneously to commodity price shocks.
- Choice of variables rationale and alternatives:
  - Industrial production used as measure of “real” aggregate demand; alternatives exist (e.g., Kilian’s freight-cost index).
  - U.S. dollar exchange rate included to control purchasing power and hedging; interest rate can affect inventory demand.
  - Alternative specification used bilateral renminbi-U.S. dollar exchange rate in place of U.S. dollar REER but had little explanatory power.
  - Reasons to explicitly identify a China demand shock include different commodity intensity of similar output shocks across countries and different dynamics/persistence and spillovers.

### Results (Section V)
- Estimation details:
  - VAR with vector Z, 5 lags, recursive ordering, sample January 2000–September 2011 (January 2002 start for crude oil estimations).
- A. Block Exogeneity (Granger causality) tests
  - For N=7 variables with P=5 lags, imposed 30 restrictions; log-likelihood chi-squared tests used.
  - Evidence is mixed for China:
    - China demand variables useful for forecasting some commodities (including crude oil) but not others (including copper).
    - United States variables showed stronger Granger causality in almost every case.
- B. Effects of real activity growth rate shocks (China)
  - A one-time 1 percentage point shock to the real month-on-month growth rate of China’s industrial production leads to:
    - an increase in the real price of oil by about 2½ percent after 4 quarters, with slight moderation thereafter.
    - an increase in the real price of copper by almost 2¼ percent after 4 quarters, with slight moderation thereafter.
  - Price responses of other base metals to the China activity shock were generally smaller and statistically insignificant (tin borderline significant in some specifications).
  - For some commodities with higher short-run supply elasticities, production responses offset price effects; example: aluminum production increases by a statistically significant amount.
- Impact of China activity growth shocks on global commodity prices — selected cumulative responses to a 1 percentage point China industrial production growth shock (real price level responses and standard errors):
  - Oil:
    - 1 quarter: 2.2 (standard error (1.06))
    - 2 quarters: 2.37 (1.50)
    - 4 quarters: 1.90 (1.06)
    - 8 quarters: 2.03 (1.01)
  - Aluminum:
    - 1 quarter: 0.76 (0.49)
    - 2 quarters: 0.98 (0.73)
    - 4 quarters: 0.89 (0.71)
    - 8 quarters: 0.72 (0.59)
  - Copper:
    - 1 quarter: 1.87 (0.78)
    - 2 quarters: 2.54 (1.12)
    - 4 quarters: 2.27 (1.07)
    - 8 quarters: 2.05 (0.93)
- Variance decomposition (China demand shocks contribution to total price variance, selected):
  - Oil: 4 quarters: 6.85 (standard error (3.48))
  - Copper: 4 quarters: 7.24 (3.94)
  - Aluminum: 4 quarters: 4.43 (3.70)
  - Interpretation: China demand shocks explain about 7 percent total price variance over 4 quarters for crude oil and copper.
- Comparison with U.S. activity shocks
  - Impacts of real activity growth rate shocks in China on global commodity prices are somewhat lower than those for the United States.
  - Selected cumulative responses to a 1 percentage point shock to the growth rate of U.S. industrial production (real price level responses and standard errors):
    - Oil:
      - 1 quarter: 4.91 (2.32)
      - 2 quarters: 9.34 (3.24)
      - 4 quarters: 9.88 (3.85)
      - 8 quarters: 11.00 (4.60)
    - Copper:
      - 1 quarter: 4.29 (2.05)
      - 2 quarters: 7.88 (3.26)
      - 4 quarters: 7.63 (4.42)
      - 8 quarters: 6.18 (5.15)
  - Variance decomposition (U.S. demand shocks contribution, selected):
    - Oil: 4 quarters: 9.45 (4.43)
    - Copper: 4 quarters: 11.13 (5.16)
    - Aluminum: 4 quarters: 7.83 (4.60)
  - Key reasons for larger U.S. impact:
    - U.S. shocks are more persistent: an initial unit shock leads to a rise in the level of U.S. activity by about 2¼ percent after 4 quarters.
    - Larger spillovers from U.S. shocks to the rest of the world: an initial activity shock in the U.S. boosts activity in the rest of the world about 1½ percentage points after 4 quarters.
    - By contrast, China shocks exhibit weak persistence and small, statistically insignificant spillovers to RoW activity.
- Commodity-specific demand shocks and inventories
  - Commodity-specific country demand shocks (controlling for economic activity) have no major effect on commodity prices for both China and the United States.
  - Time-series properties: commodity-specific demand shocks are partially unwound quickly; the cumulative effect on the level of commodity-specific demand more than halves after two quarters.
  - China inventory behavior:
    - Causality appears to run from prices to China’s apparent consumption rather than the reverse (no such effects for the United States).
    - A unit shock to the real commodity price leads to a decline in apparent consumption of about ¼ to ½ percent for some base metals (statistical significance for copper and lead).
    - Crude oil is an exception: no meaningful effect of price shocks on apparent consumption.
    - Large commodity price declines have been associated with a subsequent rise in both imports and inventories in China, consistent with price-sensitive inventory accumulation.
- Effects of monetary policy-related shocks
  - One-time shocks examined:
    - A 100 basis points increase in the average monthly real Federal Funds effective interest rate (one-time shock to first difference).
    - A 1 percentage point appreciation in the U.S. dollar real exchange rate (one-time shock to first difference).
  - Findings:
    - Interest rate shock: small, negative, and short-lived effect on prices, mainly for crude oil and, to a lesser extent, aluminum.
    - U.S. dollar appreciation: large, negative, and persistent effect on the prices of all commodities; impact peaks around 4 quarters and cumulative impacts remain above the 1 percentage point anticipated by static demand models.
  - Selected cumulative responses to U.S. real effective exchange rate shock (1 percentage point appreciation):
    - Oil:
      - 1 quarter: -3.99 (1.06)
      - 2 quarters: -4.52 (1.72)
      - 4 quarters: -3.45 (2.04)
      - 8 quarters: -3.21 (1.52)
    - Copper:
      - 1 quarter: -4.10 (1.02)
      - 2 quarters: -4.83 (1.70)
      - 4 quarters: -2.70 (2.17)
      - 8 quarters: -2.58 (1.87)
- Robustness and stability checks
  - Recursive estimations (VARs with expanding sample end dates, samples begin Jan-2000) show the impact of China on global commodity prices has been steadily increasing over the last 5 years.
  - A spike in estimated impulse responses in late 2008 driven by large falls in commodity prices and industrial production during the post-Lehman Brothers period; including dummies for Oct, Nov, Dec 2008 smoothed the evolution but results remained similar.
  - Reversing the ordering of country industrial activity and apparent commodity consumption variables in the VAR produced impulse responses very close to the original specification.

### Caveats, interpretation, and research implications
- Supply shocks may be understated by the VAR techniques used:
  - The finding that supply shocks have little price effects is counterintuitive and at odds with other empirical approaches.
  - Periods of long-lasting supply shortfalls can have large effects on real commodity prices, but these effects may not be well picked up by the methods used.
  - Precautionary demand and anticipated supply effects may soak up some supply impacts.
- Forward-looking challenge:
  - Understanding how Chinese demand for commodities will change if and when its economy rebalances (away from investment and exports towards consumption) remains the biggest challenge for future research in this area.

*Source: _wp12115 - Section 4 describes the salient features of the data. Section 5 will set out the main results.*

### Section 4 describes the salient features of the data. Section 5 will set out the main results.

### _wp12115 - Section 4 describes the salient features of the data. Section 5 will set out the main results.

### Data description (Section 4)
- Section 4 describes the salient features of the data.

### Main results (Section 5)
- Section 5 will set out the main results.

*Source: _wp12115 - Section 4 describes the salient features of the data. Section 5 will set out the main results.*

### Section 6 will provide brief concluding remarks.

### _wp12115 - Section 6 will provide brief concluding remarks.

### II. A BRIEF OVERVIEW OF CHINA’S ROLE IN WORLD COMMODITY MARKETS — Long-Term Structural Trends
- China’s consumption during 2010 accounted for about 20 percent of non-renewable energy resources, 23 percent of major agricultural crops, and 40 percent of base metals.
- Market shares have increased sharply since 2000, mainly reflecting China’s rapid economic growth; historical pattern described as the S-curve.
- China’s commodity intensity of demand is unusually high:
  - China consumes about 35 percent more energy than Korea and twice the level of Brazil at comparable income levels.
  - For base metals, China consumes significantly more than Korea and Brazil at the same income.
- Rapid expansion in the tradable export sector and large-scale fixed asset investment—particularly since 2000—contributed to fast-growing commodity intensity (Yu, 2011).
  - Ye (2008) estimates just over ½ of China’s copper usage is accounted for by infrastructure investment and construction, with ⅓ accounted for by consumer and industrial goods.
- Structural factors and domestic policy distortions have been highlighted in previous studies (IMF, 2011a) as contributors to China’s growth structure and high commodity intensity.
- China’s strategic policy decisions have affected global trade patterns (example: self-sufficiency in key grains vs. reliance on oilseed imports).
- China’s relative role by commodity group:
  - Dominant in base metals markets and to a lesser extent agricultural raw materials.
  - Has not yet assumed a large role in global food and energy markets, though its share of world imports is rising gradually.

### II. B BRIEF OVERVIEW — Short-term Fluctuations
- At higher frequencies, China’s influence on commodity markets reflects:
  - Business cycle, seasonality, and unanticipated transitory changes in supply-demand balance.
- Unanticipated shocks generally have larger market impact than anticipated fluctuations because inventories can smooth anticipated changes.
- Two types of demand shocks considered:
  1. Aggregate economic activity shocks — measured by industrial production.
  2. Commodity-specific demand shocks — measured by apparent consumption (domestic production + imports − exports).
- Important examples of commodity-specific demand shocks in China:
  - State inventory operations: China’s State Reserve Bureau believed to have purchased about 235,000 tons of copper during Q1 2009, accounting for about 65 percent of copper inventories held in London and Shanghai Exchange warehouses at end-2008 (Financial Times, 2009).
  - Temporary demand substitution (e.g., electric power shortages increasing oil and diesel demand for generators).
- Aggregate activity measured by industrial production is preferred to GDP for monthly-frequency analysis due to data availability.

### III. ECONOMETRIC METHODOLOGY: A STRUCTURAL VAR APPROACH
- Baseline reduced-form VAR (recursive identification) follows Kilian (2009) and Helbling (2012).
- Seven endogenous variables in the baseline VAR (first differences of logs except real interest rate which is first-differenced):
  - world primary production of commodity (QW)
  - world excluding country i industrial production (XRoW)
  - country i industrial production (Xi)
  - country i apparent consumption (Ci)
  - the real short-term U.S. interest rate (R)
  - the U.S. dollar real effective exchange rate (REER)
  - the real price of the commodity (P/PUS)
- Vector of endogenous variables Z written as:
  - (ln QW, ln XRoW, ln Xi, ln Ci, R, ln REER, ln P/PUS)′ (first differences applied)
- Recursive ordering identifies structural shocks:
  - supply shock
  - non-China aggregate activity shock (RoW)
  - China aggregate activity shock
  - China commodity-specific demand shock
  - exchange rate shock
  - real commodity price shock (interpreted as commodity-specific demand shock in rest of world)
- Short-run identification restrictions:
  - commodity supply curve vertical in the very short run (no contemporaneous supply response to demand shifts)
  - RoW industrial production unaffected contemporaneously by China output shocks
  - China industrial production responds with a lag to REER and real commodity price shocks
  - aggregate activity shocks affect apparent consumption contemporaneously, but not vice versa within the same month
  - REER does not respond contemporaneously to commodity price shocks
- Sensitivity of results depends on contemporaneous correlations of reduced-form residuals.

### III. B Choice of Variables — Rationale and alternatives
- Industrial production used as measure of “real” aggregate demand; alternatives exist (e.g., Kilian’s freight-cost index).
- U.S. dollar exchange rate included to control purchasing power and hedging; interest rate can affect inventory demand.
- Alternative specification used bilateral renminbi-U.S. dollar exchange rate in place of U.S. dollar REER but had little explanatory power.
- Reasons to explicitly identify a China demand shock:
  - Commodity intensity of industrial production, investment, and consumption differs across countries, so identical output shocks have different commodity input implications.
  - Dynamics and persistence of shocks differ by country.
  - Spillovers to RoW activity and demand vary with trade and financial linkages.

### IV. DATA
- Monthly frequency sampling.
- Oil supply: world crude oil production (excluding natural gas liquids) — U.S. Energy Information Agency.
- Base metal supply: production of refined products — World Metal Bulletin Statistics.
- Apparent consumption defined as domestic production + imports − exports.
- RoW industrial production: PPP GDP-weighted aggregates based on national indexes.
- China industrial production and apparent consumption seasonally adjusted using U.S. Census Bureau’s X-12; U.S. series already seasonally adjusted.
- Ex-post real interest rate: CPI-deflated Federal Funds effective interest rate monthly average.
- U.S. dollar REER sourced from IMF.
- Commodity prices: U.S. dollar spot prices reported by IMF, deflated by U.S. CPI.
- Sample period: January 2000 to September 2011 (exception: crude oil apparent consumption begins January 2002; crude oil estimations run excluding this variable from January 2000 with quantitatively similar results).
- Rationale for sample start: close to WTO entry and point at which GDP growth rose after late-1990s Asian crisis.
- Unit root testing:
  - Over 2000–2011, cannot reject null that log level of each variable (except tin production) is non-stationary.
  - First differences of logs for all variables show clear evidence of stationarity.
- Lag length selection:
  - Information criteria generally selected lag lengths between 3 and 6 months.
  - Most results reported use lag = 5 months.

### V. RESULTS
- Estimation details:
  - Results based on VAR with vector Z, 5 lags, recursive ordering, sample January 2000–September 2011 (January 2002 start for crude oil estimations).
- A. Block Exogeneity (Granger causality) tests
  - Tests drop lagged values of one variable from other equations and test null that other variables are block exogenous to that variable.
  - For N=7 variables with P=5 lags, imposed 30 restrictions; log-likelihood chi-squared tests used.
  - Evidence is mixed for China:
    - China demand variables useful for forecasting some commodities (including crude oil) but not others (including copper).
    - United States variables showed stronger Granger causality in almost every case.
  - Test statistics reported in table form in the source for Oil, Aluminum, Copper, Lead, Nickel, Tin, Zinc for both China VAR and United States VAR.
- B. Effects of real activity growth rate shocks
  - A one-time 1 percentage point shock to the real month-on-month growth rate of China’s industrial production leads to:
    - an increase in the real price of oil by about 2½ percent after 4 quarters, with slight moderation thereafter.
    - an increase in the real price of copper by almost 2¼ percent after 4 quarters, with slight moderation thereafter.
  - Price responses of other base metals to the China activity shock were generally smaller and statistically insignificant (tin borderline significant in some specifications).
  - For some commodities with higher short-run supply elasticities, production responses offset price effects; example: aluminum production increases by a statistically significant amount (details and magnitudes provided in source tables and figures).

*Source: _wp12115 - Section 6 will provide brief concluding remarks.*

### 0.4 percentage points 4 quarters after an initial activity growth rate shock in China. In part,

### _wp12115 - 0.4 percentage points 4 quarters after an initial activity growth rate shock in China. In part,

### Impact of China activity growth shocks on global commodity prices
- Shocks to aggregate activity in China have a significant and persistent short-run impact on the price of oil and some base metals.
- Example impulse responses to a 1 percentage point shock to the growth rate of China’s industrial production (selected cumulative responses):
  - Oil real price level response:
    - 1 quarter: 2.2 (standard error (1.06))
    - 2 quarters: 2.37 (1.50)
    - 4 quarters: 1.90 (1.06)
    - 8 quarters: 2.03 (1.01)
  - Aluminum real price level response:
    - 1 quarter: 0.76 (0.49)
    - 2 quarters: 0.98 (0.73)
    - 4 quarters: 0.89 (0.71)
    - 8 quarters: 0.72 (0.59)
  - Copper real price level response:
    - 1 quarter: 1.87 (0.78)
    - 2 quarters: 2.54 (1.12)
    - 4 quarters: 2.27 (1.07)
    - 8 quarters: 2.05 (0.93)
- Variance decomposition (China demand shocks contribution to total price variance, selected):
  - Oil: 4 quarters: 6.85 (standard error (3.48))
  - Copper: 4 quarters: 7.24 (3.94)
  - Aluminum: 4 quarters: 4.43 (3.70)
- Interpretation: China demand shocks explain about 7 percent total price variance over 4 quarters for crude oil and copper.

### Comparison with U.S. activity shocks
- The impacts of real activity growth rate shocks in China on global commodity prices are somewhat lower than those found for the United States.
- Example impulse responses to a 1 percentage point shock to the growth rate of U.S. industrial production (selected cumulative responses):
  - Oil real price level response:
    - 1 quarter: 4.91 (2.32)
    - 2 quarters: 9.34 (3.24)
    - 4 quarters: 9.88 (3.85)
    - 8 quarters: 11.00 (4.60)
  - Copper real price level response:
    - 1 quarter: 4.29 (2.05)
    - 2 quarters: 7.88 (3.26)
    - 4 quarters: 7.63 (4.42)
    - 8 quarters: 6.18 (5.15)
- Variance decomposition (U.S. demand shocks contribution, selected):
  - Oil: 4 quarters: 9.45 (4.43)
  - Copper: 4 quarters: 11.13 (5.16)
  - Aluminum: 4 quarters: 7.83 (4.60)
- Key reasons for larger U.S. impact:
  - U.S. shocks are more persistent: an initial unit shock leads to a rise in the level of U.S. activity by about 2¼ percent after 4 quarters.
  - Larger spillovers from U.S. shocks to the rest of the world: an initial activity shock in the U.S. boosts activity in the rest of the world about 1½ percentage points after 4 quarters.
  - By contrast, China shocks exhibit weak persistence and small, statistically insignificant spillovers to RoW activity.

### Commodity-specific demand shocks and inventories
- Commodity-specific country demand shocks (controlling for economic activity) have no major effect on commodity prices for both China and the United States; results not shown.
- Time-series properties:
  - Commodity-specific demand shocks are partially unwound quickly; the cumulative effect on the level of commodity-specific demand more than halves after two quarters (i.e., for a unit shock, the cumulative change in demand is less than 0.5 percent higher).
- China inventory behavior:
  - Causality appears to run from prices to China’s apparent consumption rather than the reverse (no such effects for the United States).
  - A unit shock to the real commodity price leads to a decline in apparent consumption of about ¼ to ½ percent for some base metals (statistical significance for copper and lead).
  - Crude oil is an exception: no meaningful effect of price shocks on apparent consumption.
  - Large commodity price declines have been associated with a subsequent rise in both imports and inventories in China, consistent with price-sensitive inventory accumulation.

### Effects of monetary policy-related shocks
- Responses to one-time shocks:
  - A 100 basis points increase in the average monthly real Federal Funds effective interest rate (one-time shock to first difference).
  - A 1 percentage point appreciation in the U.S. dollar real exchange rate (one-time shock to first difference).
- Findings:
  - Interest rate shock: small, negative, and short-lived effect on prices, mainly for crude oil and, to a lesser extent, aluminum.
  - U.S. dollar appreciation: large, negative, and persistent effect on the prices of all commodities; impact peaks around 4 quarters and cumulative impacts remain above the 1 percentage point anticipated by static demand models.
- Selected cumulative responses to U.S. real effective exchange rate shock (1 percentage point appreciation):
  - Oil:
    - 1 quarter: -3.99 (1.06)
    - 2 quarters: -4.52 (1.72)
    - 4 quarters: -3.45 (2.04)
    - 8 quarters: -3.21 (1.52)
  - Copper:
    - 1 quarter: -4.10 (1.02)
    - 2 quarters: -4.83 (1.70)
    - 4 quarters: -2.70 (2.17)
    - 8 quarters: -2.58 (1.87)

### Robustness and stability checks
- Recursive estimations (VARs with expanding sample end dates, samples begin Jan-2000) show:
  - The impact of China on global commodity prices has been steadily increasing over the last 5 years.
  - A spike in estimated impulse responses in late 2008 driven by large falls in commodity prices and industrial production during the post-Lehman Brothers period; including dummies for Oct, Nov, Dec 2008 smoothed the evolution but results remained similar.
- Reversing the ordering of country industrial activity and apparent commodity consumption variables in the VAR produced impulse responses very close to the original specification, indicating results are not sensitive to variable ordering.

### Caveats, interpretation, and research implications
- Supply shocks may be understated by the VAR techniques used:
  - The finding that supply shocks have little price effects is counterintuitive and at odds with other empirical approaches.
  - Periods of long-lasting supply shortfalls can have large effects on real commodity prices, but these effects may not be well picked up by the methods used.
  - Precautionary demand and anticipated supply effects may soak up some supply impacts.
- Forward-looking challenge:
  - Understanding how Chinese demand for commodities will change if and when its economy rebalances (away from investment and exports towards consumption) remains the biggest challenge for future research in this area.

*Source: author’s calculations and analysis in the provided IMF working paper excerpt.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12115.pdf_