1. Do Heterogeneous Countries Respond Differently to Oil Price Shocks?
Santiago Guerrero-Escobar
(DINAMA, Uruguay)
Gerardo Hernandez-del-Valle
(Actinver, México)
Marco Hernandez-Vega
(Banco de México)

2. Introduction
How a rise in oil prices affects economic variables, particularly output, has been a topic of major interest in the literature.
A generally accepted result is that an increase in oil prices has an adverse effect on output and rises prices leading to tighter monetary policy, especially in developed non-oil producer countries.
Hamilton (1983) shows that after WWII and up to late 70s, most US recessions episodes were preceded by an increase in oil prices.
But evidence of such relationship weakened after the collapse of oil prices in the 1980s.

3. Introduction
The literature that followed focused on testing nonlinear specifications of the effect of oil prices in economic activity (Mork 1989, Lee et al. 1995, Hamilton 1996 and Hamilton 2003).
The rationale for testing nonlinear specifications is that oil prices are endogenous to economic activity, especially in the US, making it necessary to transform oil price data to more properly reflect oil price shocks.
More recently, several articles suggest that such nonlinear models were either misspecified or failed to avoid the endogeneity of oil prices (Herrera et al and Barsky and Killian 2004).

4. Introduction
Killian (2009a) shows that for the US economy the impact of oil price shocks may be different depending on their source:
Oil supply: causes a temporary decrease in GDP with no effect on inflation.
Oil demand driven by global economic growth: GDP rises in the short run but falls in the long run, while inflation increases.
Oil-specific demand: GDP falls and inflation increases.

5. Introduction
Several authors have used similar empirical specifications to test such results looking at more countries.
Peersman and Van Robays (2012) found that oil shocks do not have the same impact on a set of developed oil exporters and importers:
Oil supply shocks tend to lower GDP in oil importers, while oil exporters tend to benefit from them.
Oil supply shocks driven by global demand rise GDP and prices for most countries in their sample.
Oil-specific demand shocks reduce GDP and rise prices for all countries, except a couple oil exporters.

6. Introduction
Aastveit et al. (2015) is the first paper to include developing economies in the analysis of oil price shocks.
They find that more than 50% of the fluctuation in oil prices are driven by demand shocks being those from developing countries the most important.
Their results show that the US and EU countries experience stronger declines in their economic activity after an oil-specific demand and oil supply shocks, compared to Brazil, Peru and some Asian economies where activity actually increases.

7. Introduction
Finally, Cunado et al. (2015) find that Asian economies do not respond to oil supply disruptions between
When the shock is global demand driven, GDP rises and the opposite happens after an oil-specific demand shock.
They also find that Japan and Korea's monetary policy are effective at controlling oil supply shocks.

8. Contribution
We contribute to the literature of macroeconomic impacts of oil shocks in several fronts:
Analyzing 17 highly heterogeneous countries, not only in terms of their development stage but also in terms of whether they are oil importers or exporters, or have implemented gasoline price controls.
Focusing our analysis in the period when analyzed countries have experienced drastic reductions in terms of their net oil imports and energy intensities.
Performing statistical tests on the differences in responses across country groups (5, 15 and 24 months).

9. PREVIEW OF RESULTS
Our results show that responses of economic activity to oil shocks are less differentiated across countries that what the literature has reported for previous decades.
Oil supply shocks have contractionary impacts on advanced economies but are less intense and contrary to what occurred in the 80s, they negatively affect interest rates.
Remarkably, emerging economies react in a very similar way.
Differences across country groups in the way oil shocks affect industrial activity tend to disappear over time and are not statistically different after 2 years of the shock.
In sharp contrast with other studies findings, economic activity in oil producer countries does not increase after an oil supply shock.

10. DATA Variable Acronym Description Units Source
Oil Market Conditions Variables
Oil World Supply
𝑄 𝑡 𝑜𝑖𝑙
Total World Oil Supply
Millions of barrels per day
US EIA
Oil Prices
𝑃 𝑡 𝑜𝑖𝑙
US Crude Oil Imported Acquisition Cost by Refiners
US dollars per barrel
World Economic Activity
𝑌 𝑡 𝑤
Total Industry Production Index s.a.
Index
OECD
Country-Specific Macroeconomic Variables
Prices
𝐶𝑃𝐼 𝑖,𝑡
Consumer Price Index
Haver Analytics
Economic Activity
𝐼𝑃 𝑖,𝑡
Industrial Production s.a.
Haver Analytics and Bloomberg
Interest Rates
𝑖 𝑖,𝑡
Nominal Short-Term Interest Rates
Percent
Haver Analytics and OECD
Exchange Rates
𝐸 𝑖,𝑡
Nominal Effective Exchange Rates
IMF

11. Emerging Market Economies
DATA
All data is at monthly frequency from January 2000 to September The country sample is:
Advanced Economies
Emerging Market Economies
Canada
Brazil
France
Chile
Germany
Czech Republic
Italy
Ireland
Japan
Israel
Norway
Mexico
Spain
Poland
United kingdom
Russia
United States

12. Methodology
To disentangle the impact of oil market shocks across countries, we use an Panel SVAR approach adopting Pesaran and Smith (1995) “Mean Group Estimator” (MGE).
First, we model Impulse-Response Functions (IRF) for each country using a VAR of the form:
𝑌 𝑡 = 𝐵 0 + 𝐵 1 𝑌 𝑡−1 + 𝐵 2 𝑌 𝑡−2 + 𝐵 3 𝑌 𝑡−3 + 𝜖 𝑡 ⋯⋯⋯(1)
where
𝑌 𝑡 ′ = 𝑄 𝑡 𝑜𝑖𝑙 𝑃 𝑡 𝑜𝑖𝑙 𝑌 𝑡 𝑜𝑖𝑙 𝐼𝑃 𝑖,𝑡 𝐶𝑃𝐼 𝑖,𝑡 𝑖 𝑖,𝑡 𝐸 𝑖,𝑡
𝜖 𝑡 = Structural shocks in the oil market
The number of lags included in the estimation was obtained according to the Akaike information criterion.

13. Methodology
Let Φ 𝑖,𝑗,𝑘,𝑡 𝑖𝑟𝑓 be the IRF of variable “i” for country “j” in group “k” at time “t”, and let 𝜔 𝑗,𝑘 be the average participation of country j’s GDP in the aggregate group’s GDP.
Then, variable i’s IRF for group k is:
Φ 𝑖,𝑘,𝑡 𝑖𝑟𝑓 = 𝑗=1 𝐽 𝜔 𝑗,𝑘 Φ 𝑖,𝑗,𝑘,𝑡 𝑖𝑟𝑓 ⋯⋯⋯⋯⋯⋯(2)
where
𝜔 𝑗,𝑘 = 1 𝑇 𝑡=1 𝑇 𝐺𝐷𝑃 𝑗,𝑘,𝑡 𝐺𝐷𝑃 𝑘,𝑡

14. Methodology
We identify oil shocks following a sings restriction convention as in Killian (2009) and Peersman and Van Robeyes (2012):
Structural Shocks
𝑸 𝒕 𝒐𝒊𝒍
𝑷 𝒕 𝒐𝒊𝒍
𝒀 𝒕 𝒘
Oil supply
<0
>0 ≤0
Oil demand driven by economic activity
Oil-specific demand

15. METHODOLOGY
Before estimating the model, we sort the countries in our sample into groups that reflect some common characteristics:
Groups
Countries Included 1
All
Canada, France, Germany, Italy, Japan, Norway, Spain, UK, US, Brazil, Chile, Czech Rep., Ireland, Israel, Mexico, Poland, Russia 2
Advanced
Canada, France, Germany, Italy, Japan, Norway, Spain, UK, US 3
Emerging
Brazil, Chile, Czech Rep., Ireland, Israel, Mexico, Poland, Russia 4
Oil Producer
Brazil, Canada, Mexico, Norway, Russia 5
Non-Oil Producer
Chile, Czech Rep., France, Germany, Ireland, Israel, Italy, Japan, Poland, Spain, UK, US 6
Controlled Energy Prices
Brazil, Chile, Mexico, Russia 7
Non-Controlled Energy Prices
Canada, Czech Rep., France, Germany, Ireland, Israel, Italy, Japan, Norway, Poland, Spain, UK, US

16. Results: oil shocks have SOMEWHAT larger effects in advanced economies than in EMEs

17. Results: oil producers DO NOT BENEFIT from oil supply shocks

18. Results: Prices are more responsive for oil producers

19. Results: interest rates decrease in non-oil producer countries after an oil supply shock

20. RESULTS: prices tend to increase more in countries with price controls than in countries with no controls

21. Results: exchange rates Appreciate for oil producers

22. Exports excluding oil as percentage of GDP have increased

23. Oil exports as percentage of GDP have decreased even for oil exporters countries

24. Energy consumption to GDP has fallen dramatically

25. Conclusion
The results suggest that oil shocks do not boost economic activity in oil producer countries as used to be the case before.
Most of the differences in the respond to oil shocks across countries arise from the source of the shock rather than from country differences.
This may be due to the recent developments in global markets:
Countries are more integrated
Growth in exports excluding oil has been constant while oil exports have fallen.
World wide energy consumption ratios to GDP have also fall constantly since the 2000s.
The above highlights the need for research on how oil exporting economies are transforming and adapting to new oil market conditions.

26. Appendix

27. Before estimating the model we validate the previous country groups via a simple statistical analysis:
Compute the overall 𝜇 and 𝜎 within group for each variable.
For each country “j” and variable “i” we standardize the corresponding IRF with 𝜇 and 𝜎.
For all “j”, find the maximum standardized value of each of the 12 IRFs (within group).
Add up these 12 values for every group; this defines a metric.
Contrast each country IRF to this metric.