The document discusses the regional impact of Russia's oil and gas sectors. It analyzes whether resource-rich Russian regions experience "Dutch disease" effects, where natural resource abundance negatively impacts other industries. The authors test two hypotheses - that industries with higher returns to scale and those that are less open internationally will be more disadvantaged in resource-rich regions. While initial regressions do not find supportive evidence, non-parametric estimations using original industry data suggest some industries may experience negative effects, warranting further investigation into potential Dutch disease impacts within Russia.
Presentation by Chloé Le Coq, Associate Professor at SITE-SSE
1. REGIONAL IMPACT OF
THE RUSSIAN ENERGY SECTOR
SITE-UI-FOI conference
The Russian Economy — Stagnation or Modernization?
13 June, 2017
Chloé Le Coq
SITE
Stockholm School of Economics
Elena Paltseva
SITE
Stockholm School of Economics
New Economic School, Russia
Natalya Volchkova
New Economic School, Russia
CEFIR
1
2. INTRODUCTION
Old question: Are natural resources good for a country?
Resource curse: a robust negative relationship between a country’s share of
primary exports in GDP and its subsequent economic growth
2
3. INTRODUCTION
Old question: Are natural resources good for a country?
§ Potential explanations for negative relationship between natural resources
and economic development
– Political economy reasons : resource endowments distort political incentives
– Pure economic factors :
Exposure to commodity price swings
Dutch disease (deindustrialization due to expansion of natural resource)
3
4. OIL AND GAS RESOURCES IN RUSSIAN REGIONS
NB: darker blue => larger share of oil and gas in the growth regional product4
5. PURPOSE
Our focus: the impact of hydrocarbon resources within a country
– More specifically,
• consider the regional impact of resources
• go beyond the country-wide political economy component
Research question: Regional impact of the Russian oil and gas sectors
– Well-known discussion about the importance of the energy sector for Russia
• importance for the economy correlation between oil price and GDP (Becker, 2016),
• but also as a political tool energy security for EU (Le Coq and Paltseva, 2012),….
– But must less focus on energy resources within different regions across Russia
• test the theory of Dutch disease in a cross-region study
5
6. DUTCH DISEASE IN RUSSIA: PERCEPTION
• Dutch Disease – very popular term widely used in Russia over the last decade
• By journalists…
• By authorities:
• Russia learned well a lesson of “Dutch Disease” (Vladimir Putin, July 6, 2006)
• Dutch Disease is weakening in Russia (Dmitry Medvedev, Oct 2016)
• Dutch disease is over in Russia (Anton Siluanov, Minister of Finance of RF, Feb
2015)
= > Justifies government investments in industry, despite their inefficiencies
• “Russia needs to diversify the economy, to make it more innovative.
For this we established the Investment Fund of Russian Federation” (V. Putin, 2006).
6
7. THE DUTCH DISEASE (DD) EFFECT
Spending effect: Resource windfall → extra income → non-tradable good
(services) relative prices ↑ (=appreciation of real exchange rate)! higher
wages in service sector → labor shifts from manufacturing to services and
(possibly) resource sector → de-industrialization
Factor movement effect: Resource windfall → increase in marginal product
of labor in resource sector → reallocation of labor to the resource sector
from other sectors
=> Both effects contribute to the decline of output in the Tradable sector
7
8. MIXED EMPIRICAL EVIDENCE ON DUTCH DISEASE
• Supportive evidence:
– Resource boom causes a decline in manufacturing exports and an expansion of
service sector (e.g. Harding and Venables (2010) or Sachs and Warner (1995))
– Spatafora and Warner (1995): positive effect of the terms of trade shock on domestic
investments in non-tradable sector
• Non-supportive evidence
– Resource boom does NOT cause a decline in manufacturing exports and an
expansion of service sector (e.g. Sala-i-Martin and Subramanian (2003)).
– Time series approach:
• Spatafora and Warner (1995): no significant contraction in manufacturing or
agriculture sectors (approximation of tradable sector in the model) in response to
a rise in oil-prices
• Hutchison (1994): no tradeoff between development of the energy sector and
subsequent developments in manufacturing
– Leonov and Volchkova (2013): similar methodology but cross-country data. Find no
evidence of Dutch Disease.
8
9. OUR APPROACH
Potential reasons for mixed evidence for DD:
The weak empirical evidence for DD effect may be explained by, e.g.:
• omitted variable biases (e.g., some institutional/policy/economics features
influence this correlation)
• endogeneity of resource wealth variable (e.g. exports/GDP), etc.
Our approach:
• abstract from across-countries differences, thereby weakening the omitted
variable bias issue
• That is, look at the impact of resources above and beyond common
institutional/policy component => study at a regional level
– Institutional/political economy/policy component is (arguably) relatively similar
across Russian regions
9
10. RELATED LITERATURE
• DD literature (already mentioned)
• Alexeev and Chernyavskiy (2014): mineral-rich regions are
significantly richer than the other regions in Russia
• Significant impact of fracking technology in US on
employment and economic outcome at the county and regional
level : Maniloff and Mastromonaco (2014) and Feyrer, Mansur
and Sacerdote (2016)
=> We look at how the industrial sectors change
10
11. HYPOTHESIS 1
Following Leonov and Volchkova (2013):
Suppose there is/are both resource reallocation and/or spending effects such
that the factors should outflow from manufacturing sector.
=> The manufacturing industries with higher return to scale should suffer more
Hypothesis 1
The manufacturing industries with higher return to scale should be in
disadvantage in resource rich regions compared to resource poor regions.
11
12. HYPOTHESIS 2
Following Leonov and Volchkova (2013):
Suppose there is a spending effect. Then resources move from manufacture to
non-tradable sector because of relative price changes.
=> The manufacturing sectors with less scope for price adjustment should
suffer more.
=> Less open sectors have more room for price adjustment
Hypothesis 2
The manufacturing industries with higher degree of “openness” should be in
disadvantage in resource rich region compared to resource poor region
12
13. EMPIRICAL FRAMEWORK
Difference-and-Difference approach (as in Rajan and Zingales, 1998)
Assuming
(i) Ind A has higher return to scale than Ind B
(ii) Ind A is more open than Ind B
[Growth(Ind A) – Growth(Ind B)] in resource poor region
> [Growth(Ind A) – Growth(Ind B)] in resource rich region
13
14. MODEL
Growth i,k= Const +αk+ βi+ δ ·Xik+
+ γ·Sensitivityi·RESk+ εi,k
where
Growthi,k : average annual real growth rates of sector i in region k
βi: industry (dummy)
αk: country (dummy)
Xik : share of sector i in manufacturing production of region k at the beginning
of period
RESk : resource richness of region k
Sensitivity:
(1): Sensitivity= Degree of return to scale in sector i, if γ<0 => H1 is supported
(2): Sensitivity=Degree of openness in sector i, if γ<0 => H2 is supported
14
15. DATA: REGIONAL HYDROCARBON RESOURCE
ABUNDANCE
• Our variable:
𝑅𝑒𝑠𝑜𝑢𝑟𝑐𝑒 𝑤𝑒𝑎𝑙𝑡ℎ
𝐺𝑟𝑜𝑠𝑠 𝑅𝑒𝑔𝑖𝑜𝑛𝑎𝑙 𝑃𝑟𝑜𝑑𝑢𝑐𝑡
=
𝑅𝑒𝑠𝑜𝑢𝑟𝑐𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 ∗ 𝑃𝑟𝑖𝑐𝑒
𝐺𝑟𝑜𝑠𝑠 𝑅𝑒𝑔𝑖𝑜𝑛𝑎𝑙 𝑃𝑟𝑜𝑑𝑢𝑐𝑡
• Production: Annual oil and natural gas production of region, 2006-2014
– Source: Tax Office of RF, form 5-NDPI (Extraction Tax statistics)
• Price: world oil/gas price (source: WB commodity prices with annualized version)
• Gross regional product (source: RF Federal state statistical service)
Remark: The resulting statistics may sometimes exceed 100% (as we take world prices,
and not domestic ones, at which GRP is calculated)
• We believe this is not a problem
– Systematic bias across regions
– We check for robustness, by separating regions into resource rich (>5% of GRP from
oil/gas), and resource-poor, and using this categorical variable instead of continuous one
15
16. DATA: INDUSTRIAL CHARACTERISTICS
• Return to scale in industry:
W.E. Diewert and K.J. Fox (2004), “Returns to Scale, Technical Progress and Monopolistic Markups”
(Based on US data, 16 industries, 1950-2000)
• Openness of industry: the industry Share of Industry’s World Export in
Industry’s World Output (OECD data, 22 countries)
• Av. annual real growth rates of subsectors of manuf. sector
(at the regional level for 2006-2014, RF Federal state statistical service)
– Constructed from monthly indexes of sectoral production
– Some data missing => “reconstruct” the data for narrower subsectors
• Share of sector in total manufacturing production in 2006
(RF Federal state statistical service)
– volume of own shipped production
16
17. RESULTS 1: INTERACTION WITH RETURNS TO SCALE
(1) (2) (3) (4) (5) (6)
Dependent variable: average annual index of sectoral output
VARIABLES
Resource wealth - continous
variable
Resource wealth - dummy
(rich >5% of GRP)
Initial share of sectoral output -0.0400 -0.0396 -0.0399 -0.0417 -0.0392 -0.0409
(0.0686) (0.0690) (0.0689) (0.0693) (0.0686) (0.0692)
Returns to scale*Oil wealth/GRP 0.0707 0.0386
(0.0717) (0.0504)
Returns to scale*Natural gas
wealth/GRP 0.0221*** -0.0046
(0.0045) (0.0595)
Returns to scale*Oil + gas
wealth/GRP 0.0272** 0.0385
(0.0129) (0.0476)
Constant -0.1085*** -0.1088*** -0.1089*** -0.1083*** -0.1084*** -0.1084***
(0.0273) (0.0273) (0.0273) (0.0273) (0.0273) (0.0273)
Sector fixed effect YES YES YES YES YES YES
Region fixed effect YES YES YES YES YES YES
Observations 1,224 1,224 1,224 1,224 1,224 1,224
R-squared 0.1626 0.1623 0.1625 0.1623 0.1621 0.1623
Standard errors clustered at regional level in parentheses
*** p<0.01, ** p<0.05, * p<0.1
17
18. RESULTS 1: SUMMARY AND NEXT STEP
• Hypothesis is not supported by data: no indication of Dutch Disease
(if anything, the opposite)
• Could it be the problem of our data modification?
– Recall that we “recreated” data for some of the subindustries
• Let’s do the non-parametric test on the original data and compare the
outcomes
Growth i,k=Const +αk+ βi+ δ ·Xik+ ζ·Sectori·RESk+ εi,k
where Sectori is dummy variable for sector I
– For the sake of time we show graphical evidence only
18
19. RESULTS 1A: NON-PARAMETRIC ESTIMATION AS A
ROBUSTNESS CHECK
food products
wood products, except furniture
textiles and apparel
other manufacturing
other chemicals
electric machinery and professional equipment
rubber and plastic
primary and fabricated metals
transport equipment
paper and publishing
class and products
-.06-.04-.020.02
Non-parametricsensitivitytoOil&Gas
.4 .6 .8 1 1.2
Return to scale
19
21. RESULTS 2A: NON-PARAMETRIC ESTIMATION
rubber and plastic
wood products
other manufacturing
class and products
textiles and apparel
food products
paper and publishing
other chemicals
primary and fabricated metals
transport equipment
electric machinery and professional equipment
-.06-.04-.020.02
Non-parametricsensitivitytoOil&Gas
.4 .6 .8 1 1.2 1.4
Openness
21
22. CONCLUSIONS FROM EMPIRICAL ANALYSIS
• There is no indication of Dutch Disease for Russian regions:
Neither of our hypotheses is supported by data
– no matter what kind of resource wealth variable we use, continuous or
categorical
– non-parametric robustness check is in line with the main test for returns-
to-scale, while less so for the openness…
• Why?
– Poor data?
• Data quality?
• Time period?
– Theory is wrong?
– Theory is correct but additional mechanisms in place?
22
23. POLICY IMPLICATION OF OUR RESULTS
• More generally, is there a resource curse in Russian regions?
– Our study: no Dutch disease
– Alexeev and Chernyavskiy (2014): mineral wealth has not significantly affected
regional economic growth
– Becker (2016) : statistical relationship between Russian GDP and international oil
prices.
• Our study suggests: more emphasis on policy measures aimed at other
reasons for Resource Curse besides Dutch Disease
– Political economy of rent redistribution – fight against corruption (Gylfason 2001),
competition enhancement (Auty 2001), property rights protection (Guriev,
Kolotilin and Sonin 2007)
– Timely reform implementation (Sachs and Warner 1999)
– Human capital development (Suslova and Volchkova 2007)
– Transparency, check & balances (Egorov, Guriev and Sonin, 2007)
– Volatility of government expenditures (Ramey and Ramey 1994)
23