1 Joint Research Centre (JRC) * The views expressed in this paper belong to the authors and should not be attributed to the European Commission or its.

1

Joint Research Centre (JRC)

* The views expressed in this paper belong to the authors and should not be attributed to the European Commission or its services

José M. Rueda-Cantuche* ([email protected])

IPTS - Institute for Prospective Technological StudiesSeville - Spain

WIOD Conference on Industry-level Analyses of Globalization and its Consequences – Vienna (Austria) 26 - 28 May 2010

Underestimation of the performance of the EU CO2 emission reductions via external trade

2WIOD Conference on Industry-level Analyses of Globalization and its Consequences – Vienna (Austria) 26 - 28 May 2010

Main issues are related to sustainable production and consumption:

• the identification of the most polluting activities;

• the identification of which products consumed by final users are the most environmentally harmful;

• The distinction between domestic vs. embodied emissions (multiregional IOA and international databases crucial).

1. Background


(a)This paper provides an alternative (and rather more complementary than exclusive) measure of the performance of the European Union (or any country at hand) in reducing its carbon dioxide emissions via external trade.

(b)this paper will provide a ratio of performance that measures at the industry level how far the current productive structure of an economy is from its maximum polluting capacity with a given domestic technology.


(c)This approach is perfectly in line with Leontief's (1953) approach, the main purpose of which is to compute not the labour/capital consumption of the rest of the world, but the labour/capital saved by the US through imports.

(d)Our purpose will be to estimate the carbon dioxide emissions saved by the EU through imports and not to compute the emission intensity of the rest of the world.


(e)It seems sensible to take for granted the domestic technology assumption for the reduction in emissions by the EU through imports. Otherwise, imported products should have been produced domestically.

(f) This approach may be criticised for not being too realistic, but this would be true only if we wished to measure international emissions, which is not the purpose of this paper.

(g)Dietzenbacher and Mukhopadhyay (2007) and Rueda-Cantuche and Amores (2010).


(h) This paper will also prove that the standard measures available in the literature would underestimate the ratio of performance and thus, would assign the EU less amounts of exported air emissions (carbon dioxide) on average.

(i) In addition, another advantage of this approach is that all the data needed to make the calculations are ready to use worldwide at many countries’ National Statistical Institutes websites.


2. Methodological framework

To our knowledge, the IO type of analysis used so far by LCA and industrial ecology practitioners is based almost exclusively on the Leontief quantity model (Dietzenbacher, 2005) and the multipliers obtained through the so-called Leontief inverse.

→ By changing the quantities of a mixed bundle of products consumed by final users, the Leontief quantity model yields variations in industry outputs (considering industry by industry tables).

→ Emission coefficients provided generally by LCA practitioners and/or NAMEA can be eventually used to determine the change in emissions caused by the initial variation in final demand.


Emission multipliers have been reported in a number of studies for different countries (Germany, United Kingdom, Sweden, Australia, Japan, Netherlands, New Zealand and Spain, among others) either in a single- or multi-regional IO framework.

• Proops et al (1993)• Östblom (1998)• Lenzen (1998) and Lenzen et al. (2004)• Gerilla et al (2001)• Haan (2001)• Creedy and Sleeman (2005)• Serrano and Roca (2007) …


Stochastic input-output analysis

* While these studies provide environmental impact estimations, to our knowledge, IO-LCA practitioners have paid little attention to the POSITIVE and SIGNIFICANT BIAS of the multipliers derived from the Leontief inverse (Dietzenbacher, 2006)

→ Assuming stochastic technical coefficients (A) leads to a positively biased Leontief inverse (Simonovits, 1975…)

→ All separate positive biases accumulate in the projection for the output levels since the Leontief inverse is post-multiplied by an exogenously specified (positive) final demand vector.

→ Hence, overestimation is not a negligible issue at all (Dietzenbacher, 2006)


* Might be more economically plausible to assume stochastics on the IOT values rather than on the input matrix A? (Dietzenbacher, 2006)

→ Yes, but it is rarely adopted (Gerking, 1976, 1979; and Dietzenbacher, 1988) due to the rather complex stochastic nature of the input coefficients induced by the transformation from IOT to A.

* But IOTs are derived from SUTs! So why stop there at IOTs and not use SUTs ?


* A similar exercise was already carried out by ten Raa and Rueda-Cantuche (2007) but with the supply and use (firms) data used to compile official SUTs.

→ Regression coefficients of a single-equation econometric model turned out to be (output, employment, emission…) input-output multipliers under the product technology assumption.

→ Consistent and unbiased employment/output multipliers; confidence intervals; hypotheses testing. Positive bias confirmed for the majority of multipliers.

→ Limited data availability.


Contributions

* To our knowledge, we present the first combined use of econometric modelling tools within a supply-use system to address environmental repercussions (CO2 emissions) from changes in quantities consumed by final users.

→ This approach provides one-shot unbiased and consistent estimates of CO2 emission multipliers on the basis of official SUTs (note that this differs slightly from the approach based on firms data).

→ Minimal requirements for IO-LCA practitioners: SUTs at basic prices; data on direct emissions; and standard econometrics.


EEIOA

SUT

Econometrics

Put together for the very first time to analyse CO2

emission coefficients

Performance of CO2 emission reduction via external trade

Confidence

intervals

Empirical

application


(a) Following ten Raa (2005) and ten Raa and Rueda-Cantuche (2007) the row vector of CO2 emission multipliers per commodity

output (γ) is defined as usual:

3. Econometric CO2 emission multipliers

1)( AIc

where c = row vector of CO2 emission per commodity; and (I-A)-1 is the standard Leontief inverse.


(b) Next by assuming the commodity technology assumption in the sense that carbon dioxide emission levels of a commodity are independent of the producing industry, we denote:

T

T

CVc

cVC

being C = row vector of CO2 emission levels by industries; and V T, the production matrix of the supply table at basic prices (commodity by industry)


(c) Similarly, the construction of the A matrix under the product technology assumption is given by:


T

T

UVA

AVU

being A = matrix of technical coefficients (product by product); U, the intermediate part of the domestic use table at basic prices (product by industry); and V T , the production matrix of the supply table at basic prices (product by industry)



1)( AIc

111 UVCVUVICUVICV TTTTT

Bearing in mind the former assumptions,

becomes into:

which can be expressed as:

)( UVC T


Now, it’s time to introduce econometrics by adding a random disturbance error.


)( UVC T

Requirement: enough degrees of freedom for estimationNr explanatory variables (n) = Nr rows in (V T-U) → Nr productsNr observations (m) = Nr elements in row vector C → Nr industries Nr degrees of freedom = m - n Rectangular oriented approach!!!

11 12 1

21 22 21 2 1 2 1 2

1 2

T

m

mm n m

n n nm

C V U

x x x

x x xC C C e e e

x x x

Directemissions

Multipliers

Net output

Errors

11 12 1

21 22 21 2 1 2 1 2

1 2

T

m

mm n m

n n nm

C V U

x x x

x x xC C C e e e

x x x

Directemissions

Multipliers

Net output

Errors


yzzx md

)1()1()1( dd

m

d

m

a

y

a

z

a

yzx

x

zayzxax ddmd ,

Suppose the simplest case of one economy with just one single sector,

)1(

1

day

x


Suppose c0 the amount of CO2 emitted per unit of product output,

)1(0

0da

c

y

xc

1 dd AIc


Suppose zm = 0, hence zd = z; all imports were to be produced domestically,

)1(0

0 a

c

y

xc

1 AIc


Since a is expected to be always greater

or equal than ad, then:

)1(

1

)1(

1

daa

d

dAIcAIc

11


= maximum polluting capacity (per one-unit increase in final demand quantities) of an economy. If all imports were to be produced domestically, it yields how much CO2 emissions were to be increased to reach the maximum level of emissions for a given domestic technology !!


d = measure of the actual polluting capacity (per one-unit increase in final demand quantities) of an economy. Not all imports were to be produced domestically!! It is expected to be benchmarked by


Subsequently, we propose the following eco-performance ratio (P):

dP 1

where P gives an idea of how far is the current productive structure of an economy from the maximum emission levels per unit of output that can be achieved (P=1) with the current domestic technology. As long as:

• P ≈ 1: the corresponding industry is benefitting from the external trade to reduce its emissions.

• P ≈ 0: the contrary applies; (e.g. no external trade).

4. Performance of emission reductions…

For example, let γd be equal to 5 tonnes and γ be equal to 10 tonnes (P=1/2), then EU imports of such commodities allow EU to be at the 50% of its maximum polluting capacity per unit of output.


)( UVC T

)( dT

d UVC

dP 1

1

1

dd AIc

AIc

Tdd

T

VUA

UVA


5. Data and results

Empirical application to EU27 (2000)

(a)Data sources: direct CO2 emissions (Eurostat); first EU27 aggregated IOT (dom + imp) for 59 industries and 59 products (Rueda-Cantuche et al., 2009) and an aggregated EU27 supply table at basic prices (Eurostat).


5. Data and results

Empirical application to EU27 (2000)

(b) Construction of the model:

* Calculation of the row vector (C) of CO2 emission levels by industries (1 x 59) and the use table at basic prices assuming the PTM for domestic and total uses.

* Aggregation of the 59 columns (regarding products) of the supply and use tables into 21 ad-hoc pollutant-wise relevant groups of commodities (according to their shares over the total European emissions of CO2)


Domestic direct emissions (percentage)

0,0%

5,0%

10,0%

15,0%

20,0%

25,0%

30,0%

35,0%

40,0%

45,0%

Domestic direct emissions (percentage)







Comparison P(L) vs P(E):

14.0

09.0

E

L

P

P

%35(a)By using the Leontief inverse, one is closer to the

maximum polluting capacity (P is lower) than it should be.

(b)Hence, we are underestimating the actual performance of the EU in reducing its CO2 emissions via external trade.


d

s

d t

0s

1s

0d

0s s

AB

C

1s s

C

AB

* The role of import shares in the eco-performance ratio (P):


(a)Sectors with large import shares but low value of P (when large Ps are expected):

• Fishing

• Textiles

(b) Sectors with small import shares but large value of P (when small Ps are expected):

• Other services

• Construction


6. Conclusions

(a) IO economics and industrial ecology (LCA) are increasingly becoming a part of government efforts to quantify national and global environmental impacts of sustainable production and consumption strategies (e.g. EU).

(b) Hence, the extensive use of the Leontief quantity model and the Leontief inverse deserves some further thinking on the stochastic limitations (positive bias) derived from the IO literature.

(c) This paper proposes therefore a new combined approach with three main advantages:

→ Improvement of the accuracy of the environmental impacts assessed by industrial ecologists;

→ Finding of a way to compute unbiased and consistent IO multipliers for the IOA community;

→ Relegate the use of IOTs in favour of SUTs (easily available), thus avoiding the use of the Leontief inverse.

39

Joint Research Centre (JRC)

* The views expressed in this paper belong to the authors and should not be attributed to the European Commission or its services

José M. Rueda-Cantuche* ([email protected])

IPTS - Institute for Prospective Technological StudiesSeville - Spain

WIOD Conference on Industry-level Analyses of Globalization and its Consequences – Vienna (Austria) 26 - 28 May 2010

Comments, more than welcome!!

Underestimation of the performance of the EU CO2 emission reductions via external trade

1 Joint Research Centre (JRC) * The views expressed in this paper belong to the authors and should not be attributed to the European Commission or its.

Documents

industry level

consequences vienna

industry outputs

industry tables

background slide

international emissions

carbon dioxide emissions

external trade slide