Economic drivers of greenhouse gas-emissions in small open economies: A hierarchical structural decomposition analysis

Transcription

1 MPRA Munich Personal RePEc Archive Economic rivers of greenhouse gas-emissions in small open economies: A hierarchical structural ecomposition analysis Daniel Croner an Wolfgang Koller an Bernhar Mahlberg Institute for Inustrial Research, Vienna University of Technology, Vienna University of Economics an Business 15 February 2018 Online at MPRA Paper No , poste 11 April :42 UTC

2 Economic rivers of greenhouse gas-emissions in small open economies: A hierarchical structural ecomposition analysis 1 Daniel Croner a,b, Wolfgang Koller a an Bernhar Mahlberg a,c, a Institute for Inustrial Research (IWI), Mittersteig 10, 1050 Vienna, Austria b Vienna University of Technology (TU Wien), Wiener Hauptstraße 8, 1040 Vienna, Austria c Vienna University of Economics an Business (WU Wien), Welthanelsplatz 1, 1020 Vienna, Austria Abstract The Paris agreement has prescribe strict Greenhouse Gas (GHG) reuction targets for participating countries. Implementation of climate protection policies is challenging, especially if the economy is export riven. We introuce a hierarchical structural ecomposition moel in orer to investigate the effects of exports, imports, economic structure, consumption patterns, consumption level, outsourcing an insourcing on national GHG emissions. This moel is applie to the ata of national environmental accounts an to a harmonize an price-eflate series of national input-output tables of Austria for the years 1995, 2000, 2005 an Over the whole time perio, the results inicate that the final eman effect was the main river of GHG emissions, with exports as most important factor. Surprisingly, emission intensity contribute to an increase of GHG emissions uring the perio as well, mostly ue to increasing emission intensity in the transport sector. Keywors: Leontief Moel, Emissions Emboie in Exports, Trae Integration, Economic Structure, CO 2 -Intensity, Competitiveness JEL Classification: Q53, Q56, C67, L16 1 The authors gratefully acknowlege the valuable comments receive on earlier versions of this paper from Martin Bruckner, Stefan Giljum, an Hanspeter Wielan. The authors also woul like to thank the Anniversary Fun of the Oesterreichische Nationalbank for supporting the project uner project number The usual isclaimer applies. Corresponing author: Bernhar.Mahlberg@wu.ac.at an Mahlberg@iwi.ac.at

3 1 Introuction The relevance of climate policy has change the view on a country s economic performance. While classical measures for economic well-being, like gross omestic prouct (GDP), total factor prouctivity (TFP) or unemployment, are still very important inicators, the evaluation of the economic performance of a country woul not be complete without looking at sustainable inicators like inequality, health, safety or, with respect to climate change, carbon ioxie (CO2) emissions. The Paris Agreement has prescribe specific targets for participating countries an ecision entities like the European Union. For instance, in the year 2030, there shoul be a cut of greenhouse gas emissions (GHG) of at least 40% of the 1990 level. The share of renewable energy shoul be 27% an the energy efficiency also shoul improve by at least 27%. 2 The economic challenges arising from these requirements are fairly ifferent between countries. The EU members have ifferent structures of the economy, some relying more on heavy manufacturing some more on the service sector. The energy efficiency also varies strongly, often epening on the stage of evelopment of a country. 3 Export intensive economies might face aitional challenges by proucing carbon intensive proucts for consumption elsewhere. Furthermore, small economies within the EU are highly epenent on economic conitions in other EU countries. We analyze the competitiveness of Austria as an example of a small open economy with respect to GHG emissions. A small open economy participates in international trae, but is small enough compare to its traing partners so that its policies o not alter worl prices, exchange rates, interest rates, or incomes. Such an economy is characterize by particularly pronounce trae integration an outsourcing. For such countries competitiveness is of special importance. This is also the case in terms of environmental measures: For an economy which is highly expose to international competition, environmental policy might be especially expensive ue to the threat of loss in competitiveness. However, there are two more points of views on this issue. The link between competitiveness an environmental regulation is often iscusse in the context of the Porter Hypothesis (Porter an van er Line, 1995) which states that impeing environmental amage oes not have to be a contraiction to economic competitiveness. As Pasurka (2008) highlights, pollution abatement lowers prouctivity ue to higher costs but might increase prouctivity ue to higher efficiency an research an evelopment efforts. Those stuies still take long term economic output as reference point for competitiveness. The thir perspective was evelope by Aiginger an Vogel (2015) who introuce a new notion of competitiveness by incluing social an environmental pillars which exten the traitional notion of competitiveness with inicators like resource prouctivity, CO 2 emissions an renewable energy share. However, this might be conflicting with the classical factors of competitiveness as stronger pollution abatement gives incentives to outsource pollution intensive inustries which might be competitive in a purely economic sense. Due to the consierations above we treat environmental sustainability itself as a measure for competitiveness an therefore analyze the evelopment of GHG emissions an its eterminants on a etaile level for a small economy with high trae integration. The main question aresse in this paper is which eterminant factors were crucial for the emission evelopment an to what extent? More exactly, we introuce a hierarchical structural ecomposition moel (HSDA) in orer to exam- 2 (accesse on December 15 th, 2017) 3 See Croner an Frankovic (2018) an Voigt et al. (2014) for trens of energy intensity in 40 major economies 1

4 ine the contribution of emission intensity, structural change, consumption level, consumption patterns, outsourcing, insourcing an the exports of final goos on omestic GHG emissions of a small economy. We also conuct this analysis on the commoity level in orer to get etaile insights on the most polluting commoities an its evelopment over time. In a case stuy, we apply this moel to Austria for the years 1995 to 2010, using ata of national environmental accounts an a harmonize an price-eflate time series of national input-output tables. We ivie the sample perio in the three subperios 1995 to 2000, 2000 to 2005, an 2005 to This paper is mainly base on the growing literature of structural ecomposition analysis an CO 2 emissions. The methoological prerequisites are evelope in Koller an Stehrer (2010). They introuce the HSDA an apply it to examine the influence of trae integration an outsourcing on the Austrian labour market. The effect of increasing international trae linkage on CO 2 emissions has been stuie by various research works, like e.g. Wiemann et al. (2007), Xu an Dietzenbacher (2014), Hoekstra et al. (2016), Malik an Lan (2016), an Kaltenegger et al. (2018). 4 Most of these stuies use a Multi Region Input-Output Moel to obtain an overview of global CO 2 trens. Another stran of the literature analyzes only country specific CO 2 emissions, like e.g. Chang an Lahr (2016) or Guevara an Rorigues (2016). The reason for this is often a more etaile isaggregation of the sectors or commoities of an economy which allows for more etaile examinations of the origin of emission. Also notably, we apply a prouction-base approach to the Austrian economy (as oppose to a consumption-base approach). There are several avantages an isavantages of prouctionbase accounting (Peters, 2008). As emission ata are collecte irectly at the place of prouction, this approach requires consierably less assumptions an uncertainty. More important, the prouction-base approach is the preferre metho when the focus is on inicating possible areas of mitigation efforts an national policy action rather than the allocation of responsibility towars consuming countries. Our contribution to the literature is threefol: First, the HSDA moel provies a fairly etaile view on a country s GHG emission. We are able to ecompose the changes in emissions over time into the typical factors of emission intensity, technology an final eman on the first level of the moel hierarchy. On the secon level we further ecompose these factors into omestic an export factors while on the thir level we account for outsourcing, insourcing, export- an omestic eman structure an level. Secon, the analysis of the Austrian economy offers a etaile view on the performance of a small open economy with respect to GHG emissions. Thir, the analysis is base on a commoity-by-commoity input-output table, not, like many other stuies, on an inustry-byinustry table. This gives a more accurate view on where emissions really come from as sectors often prouce various, an sometimes very ifferent, goos. Our results for Austria inicate that the highest increase in GHG emissions occurre in the subperio 2000 to The main river in all perios is the final eman level where especially the increase of GHG emissions in exports contribute strongly. Surprisingly, the intensity effect contribute to an increase of GHG emissions between 2000 an The commoity level analysis reveals that this is mainly ue to transport services. 4 See Lenzen (2016) for a more complete overview of the recent literature on this issue. 2

5 The structure of the paper is as follows: In section 2 we introuce the hierarchical structural ecomposition moel applie on GHG emissions. In section 3 we explain the ata preparation steps for the case of Austria. In section 4 we present the results of the analysis. Section 5 conclues. 2 Metho The methoological approach of this paper relies mainly on a hierarchical ecomposition (HDSA) moel introuce by Koller an Stehrer (2010). We aapt this moel to analyze the evelopment of GHG emissions, whereby we leave the basic hierarchical structure of the moel unchange. The HSDA buils on the classical Leontief input-output moel which is our starting point: q A q f L f (1) where q is the vector of gross outputs of the goos prouce in the economy, omestic input coefficients (i.e., incluing imports), an The omestic Leontief inverse is efine as: I D A 1 A the matrix of f the final eman for omestic proucts. L (2) A where A is the technical coefficient matrix an D A = A A the matrix of omestic shares. Hereby, an enote element wise ivision an multiplication. Equation (1) gives the require gross output of all goos for a given final eman for goos prouce in the omestic economy. By aing the emission coefficient c we obtain the GHG emissions E which are associate with final eman: T E c L f (3) The emission coefficient c is efine as c = m q where m is the emission vector which inicates the irect emissions occurring in prouction of a certain goo an q the gross output of this goo. The superscript T is use for transposing the vector. While E is a scalar which gives the overall emissions associate with omestic prouction the formula e cl ˆ f (4) gives the vector of emissions associate with prouction of each specific goo. ĉ is the matrix resulting from the iagonalization of vector c. 5 In our analysis we are intereste in the etails of the structure of intermeiate prouction an final eman which etermines the evelopment of E an e. Therefore we analyze etail. L an f in more Throughout the paper we enote the change in a variable between two time perios with x x 1 x 0. On the first level of the hierarchy we ecompose e (or E respectively) in the factors in (4): 5 Note that this approach is prouction-base an not consumption-base as woul be achieve with a iagonalization of f. 3

6 C e L e e e ˆ (5) f where e x enotes the change of e ue to the variation of x when all other factors of e remains unchange. Those ecomposition factors are again ecompose in its elements further own the hierarchy. Figure 1: Hierarchical Decomposition Figure 1 gives an overview of the complete moel. The share of omestic input coefficients ecompose into outsourcing D A, o an insourcing A i D A is D,. More exactly, if the omestic share of an intermeiate input increases (ecreases), we observe insourcing (outsourcing). Formally we write, DA; m, n, if DA ; m, n 0 DA, o (6) 0, otherwise DA; m, n, if DA ; m, n 0 DA, i (7) 0, otherwise Final eman for goos of omestic origin f can be ecompose into final eman (excluing exports) of omestically prouce goos h an exports x. In the last step, we istinguish between final eman excluing exports h an the share of omestic eman h. We use the relationship ~ x sx f to further ecompose (changes in) x in (changes in) the shares of export in final eman an (changes in) the level of final eman ( f ~ an enote the same varia le we use the sign to enote that in the ecomposition moel their time stamp is varie inepenently of each other). This ecomposition analysis gives a clear picture of the effects of trae integration on the omestic prouction-base emissions. For example, those omestic emissions epen on whether the supply chain of omestic prouction is also omestic or prouction relies more on foreign intermeiates. Uner strict environmental policies in the European Union, countries might have an incentive to outsource emission intensive prouction in orer to improve its balance. This can be backtrace with the variable D A, o in our moel. f 4

7 Another aspect of trae in our moel is the export of goos. The export share effect s, E x, shows the change in GHG emissions that are ue to a change in the shares of export eman in overall final eman for omestic goos. If this effect rises, an economy is confronte with a higher challenge in achieving its environmental goals. Especially economies which are specialize on carbon intensive sectors shoul carefully monitor the evelopment of this effect. 3 Data s x For our analysis we obtain the GHG ata for Austria from the air emission accounts of Statistics Austria. We calculate CO 2 equivalents for a global warming potential (GWP) of a 100 year time horizon. GWP factors are taken from the UNFCCC (2012). Besie CO 2, we use methane (CH 4 ) with a GWP factor of 25 an Nitrogen oxie (NO x ) with GWP 298. We focus on GHG emissions in the prouction process, an hence we o not account for the emissions which occur by private persons, e.g. use of private cars. The ata are available for 55 ifferent sectors, accoring to the ÖNACE 6 segmentation. We use input-output tables from Statistics Austria for the years 1995, 2000, 2005 an 2010 an choose a prouct-by-prouct rather than an inustry-by-inustry approach, because goos prouce in an inustry might be very heterogeneous with respect to GHG emissions an therefore the inustry-by-inustry results coul be istorte. However, CO 2 emission ata an equivalents are only available in the imension of inustries (ÖNACE). In orer to be able to use them in the framework of an input-output moel that is formulate in the prouct-by-prouct imension, the emission ata also have to be transforme into the prouct imension. Usually, for that purpose two ifferent basic approaches are available, the commoity technology assumption an the inustry technology assumption (ITA). Which one is to be preferre is, ultimately, an empirical question. For most types of variables, as also for energy inputs an GHG emissions, the commoity technology assumption (CTA) is consiere as more realistic. While GHG emissions by unit of prouction are expecte to vary over ifferent inustries that prouce the same commoity, they will not vary as much as over ifferent commoities, even when prouce by the same inustry. In the present application a further avantage of the CTA is its logical consistency with an input-output moel specifie in the prouct-byprouct imension. However, ue to the well known problem of negatives in the CTA, various approaches have been evise to allow for (small) eviations from the CTA, one of them being the algorithm by Almon (2000). Elements of the transforme ata vectors an matrices that are negative with plain CTA come out as zero when the algorithm by Almon is use. Since this (zero emissions of GHG) is still not consiere as plausible in many applications such as the present one, some approaches combine the CTA an the ITA in a heuristic way or efine lower bouns for those elements. In this paper we apply a hybri approach where we use the well establishe commoity technology assumption to 75% an the inustry technology assumption to 25%. Aitionally, in orer to obtain consistent input-output tables we reclassify the table of 2010 from ÖCPA 2008 to ÖCPA All tables are eflate an expresse in constant prices of 1995 using price information from Austrian national accounts (Statistics Austria). We aggregate all tables to 50 commoity groups in orer to reach consistency over the whole time perio. For our analysis, ata 6 ÖNACE is the Austrian version of NACE Rev. 2, the classification of economic activities. For etails see evelopment _of_the_onace2008/inex.html 7 ÖCPA is the Austrian version of CPA, the classification of proucts by activity. 5

8 m tons from Statistics Austria provies more etaile an reliable information for the Austrian economy an therefore we prefer this ata source to other commonly use international input-output tables such as Worl Input Output Database (WIOD) or EORA MRIO atabase. Data inclue intermeiate eliveries of each omestic sector, imports, exports an final eman. After eflating the Input- Output tables we correct the horizontal accounting equation in orer to ensure the sum of intermeiate eliveries an final eman to be equal to overall output of a commoity. 8 In the Appenix the intereste reaer can fin more etails on the rationale an proceures applie in the preparation of the eflate input-output tables. 4 Results 4.1 Results for the whole economy Greenhouse gas emissions in Austria exhibite a sharp increase from about 87 million tons in 1995 to 118 million tons in 2008 (Figure 2). After the global financial crisis emissions ecrease probably ue to reuce economic activity. During the time of economic recovery, the rop was partly offset in 2010 but the evelopment shows a persistent negative tren afterwars. Figure 2: GHG Emissions in Austria from 1995 to 2014 excluing househol emissions We are intereste in the reasons of the increase of GHG emissions in the perio between 1995 an Emissions increase by 29.2 million tons in this time perio which makes an increase of 33.4% compare to the 1995 level. Figure 3 provies an overview of the results for every ecomposition variable from 1995 to The number in brackets gives the percentage of factor contribution relative to the overall emissions in Main reason of the increase was the final eman ( f ) which contribute 35.8 million tons to the rise of emissions. Interestingly, most of this effect was ue to GHG emboie in exports ( x ) 8 The ata for 1995 are converte from Austrian Schilling, which was the national currency of that time, to EURO. 9 The choice of this perio is motivate by the availability to ata. It epicts the perio of increasing emissions in Austria. More recent trens will be analyse in future. 6

9 with 29.7 million tons. The emission weighte increase of omestic eman ( h ) on the other han was quite moerate with only 6.1 million tons. On the thir level, we can reveal the effects of change in trae integration of Austria on emissions. The increasing share of exports ( s x ) in final eman le to an 11.8% increase in omestic emissions. The reverse pattern can be observe for the share of omestic eman for goos of omestic origin ( ) which contribute to a 5.8% ecline of emissions h uring that perio. More important was the overall eman level effect which amounte to 19.4 ~ million tons for exports ( ) an 11.2 million tons for omestic eman ( h ). f Figure 3: Overall Results for the perio from 1995 to 2010 Oppose to this, the intensity effect ( ĉ ) ecreases overall emissions. Technology an efficiency improvements le to a ecline of 10.4% relative to the 1995 level. Changes in the omestic Leontief inverse i not alter the emissions much at the first view. The overall increase ue to a changing input structure of the economy ( L ) was 2.5 million tons which accounte for only 2.8% percent. How- ever, by looking eeper into the reasons for that, we can observe strong heterogeneity. First, there was an opposite effect between the omestic share matrix ( D ) an the matrix of technical input coefficients ( A ). While there was a ecline of emissions emboie in omestic inputs, which le to a 6.9% rop of emissions, the overall structure of technology inputs ( A ) i not change in favor of the environment an increase by 9.7%. The ecline in the emissions ue to omestic emissions was cause by a relatively strong tren of outsourcing. In fact, leaving everything else in the economy fixe, outsourcing ( D A, o ) woul have ecrease omestic emissions at 21.5%. Thus, Austria tens to outsource carbon intensive inustries. The reverse factor, insourcing ( D, ), augmente emissions at 14.6%. The strong peculiarity of both trens is an inicator of increasing trae integration. Table 1 shows all these trens ivie in the three subperios. Remarkably, the intensity effect shows a fairly ifferent pattern across the subperios. While in the first subperio, intensity ecline strongly (8.9% relative to 1995 level), it even increase in the secon subperio with 2.5% relative to the 2000 level an finally slightly ecrease by 3.3% compare to We will go more into etail in the analysis of iniviual activities in the next section. Leontief Inverse an final eman showe both a continuous tren. While the changes in the input structure le to a ecrease in emissions in the first subperio, this change for the last two subpe- A A i 7

10 rios. In the last subperio it even was the strongest factor towars a more polluting economy. The effect of final eman ecline over time, although it was always positive. The rop in the final eman effect was most importantly ue to the eclining emissions emboie in exports of final goos, which roppe from an 18.5% increase in the first interval to 2.2% in the last. The omestic eman effect was eclining as well, even resulting in a lowering of GHG emissions in the last subperio. This was ue to an overall eclining omestic eman an also because the share of importe final eman goos increase (inicate by the eclining share of omestic eman). Trae integration of intermeiate goos tens to reverse its tenency uring the time uner consieration. In the first subperio, the carbon content of outsourcing was ominating that of insourcing resulting in a 5% rop of CO 2 equivalents ue to the omestic share matrix. However, the results show that this tren weakene in the next subperio an even reverse in the last subperio. Tab. 1: Overall Results for , , an Effect Variable Percentage Percentage Percentage Intensity c -7,800, ,387, ,669, Leontief Inv. L -5,637, ,142, ,970, Domestic Share D A -4,403, ,823, ,204, Outsourcing D Ao -6,499, ,443, ,902, Insourcing D Ai 2,096, ,619, ,106, Structure A -1,234, ,965, ,766, Final Deman f 21,140, ,071, , Domestic Deman h 4,932, ,038, ,852, Share of om. Deman h -790, , ,370, Domestic Deman h 5,722, ,958, ,517, Export of final goos x 16,207, ,033, ,469, Share of Exports s x 3,675, ,896, ,778, Final Deman Export f x 12,532, ,136, , Results on commoity level First, we present results on an aggregate activity level for the whole sample perio 1995 to We istinguish between transport, service, construction, electricity, manufacturing with low an high carbon intensity, mining an agriculture. As Figure 4 shows, the intensity effect in the prouction of the most commoities contribute to a ecline of GHG emissions uring that perio. Figure 4: Intensity Effect Transport Service Construction Electricity Manufact. Low Manufact. High Mining Agriculture Million Tons 8

11 Exceptions are transport an construction. Especially the latter one ha a remarkable ba performance, a pattern which is persistent across all subperios. Most improvements were one in the electricity sector where the change of intensity in prouction almost save 7 million tons of GHG emissions. This effect was more than offset by the change in the input structure which inicates changes in the intermeiate input structure (Figure 5). Figure 5: Effects of the Input Structure Transport Service Construction Electricity Manufact. Low Manufact. High Mining Agriculture Million Tons As a more etaile analysis reveals (see supplementary material for the tables with all results on a commoity level), this was ue to the matrix of technological input coefficients, an, more in etail, it was ue to the higher requirements of electricity in the prouction of electricity itself. This effect might be parallel to improving intensity ue to accounting specifics. Figure 6: Final Deman Effect Transport Service Construction Electricity Manufact. Low Manufact. High Mining Agriculture Million Tons For services, construction an mining, the input structure was also positive. The other activities improve their emission balance ue to the intermeiate input structure. Final eman, the main river of increasing GHG emissions, was positive for all activities, whereby highly GHG intensive manufacturing contribute most (Figure 6). Everything else fixe, the growing eman of such goos woul have cause an increase of more than 15 million tons of GHG emissions. Electricity an Transport, the secon an thir most eman riven areas, raise emissions by 6.9 an 5.6 million tons, respectively. 9

12 Figure 7: Total Effect Transport Service Construction Electricity Manufact. Low Manufact. High Mining Agriculture Million Tons All effects together, electricity was the main river of the rise of CO 2 equivalents between 1995 an Carbon intensive manufacturing, transport an construction were also strongly responsible for increasing emissions. Prouction of low carbon intensive manufacturing goos an mining were the only activity which i not increase its emissions but exhibite a rise in final eman. Therefore, in those fiels, the technological improvements were sufficient to offset raise of eman. The trens show some heterogeneity across the three subperios. As shown in the etaile breakown across the three subperios in the supplementary material, especially the secon subperio exhibits a ecline of emissions efficiency in many sectors. This is particularly striking for transport activities, where emissions increase more than 2 million tons ue to poorer emission intensity. This pattern reverse in the last subperio where transport activity showe consierably improve emission intensity. Table 2 shows a more etaile pattern of emissions on activity level. It epicts the five worst performing an the five best performing commoities accoring to the intensity an the total effect in each subperio. As alreay seen in the aggregate view, construction an transport contribute strongly towar increasing emissions in the first two subperios. Especially lan transport raise carbon intensity until 2005, where it was the best performing sector between 2005 an 2010 an totally offsets the intensity effect of the former subperios. Lan transport contributes most to the overall raise of emissions in Austria in the first two subperios. Between 2000 an 2010, the prouction of electricity contribute most to the increase in emissions. Besies the afore-mentione commoities, air transport, basic metals an the prouction of woo an wooen proucts strongly contribute to emission increase. In the last subperio, agriculture exhibite a strong increase of CO 2 equivalent emissions as well, mainly ue to poorer emission intensity. The importance of the electricity sector for overall emissions motivates a closer look. Figure 7 shows the etaile ecomposition result for the electricity prouction for the whole time perio 1995 to Overall, in supply of electricity there was an increase of 9.3 million tons of CO 2 equivalents. This was a 71.6% growth compare to the 1995 level an makes 31.9% of the total emission increase of the economy. The growth of emissions in the electricity sector was consierably faster than the average growth of emissions. 10

13 Tab. 2: Intensity effect an total effect on the five worst an the five best performing sectors Intensity Intensity Intensity Construction 1,377,408 Lan Transport 2,347,372 Construction 1,831,168 Lan Transport 1,260,626 Other non-metallic mineral 1,217,943 Woo 1,797,706 Air Transport 666,138 Pulp, paper an paper 984,013 Other non-metallic minerals 1,191,275 Foo+ beverages + tobacco 372,972 Construction 938,288 Agriculture 1,018,045 Woo 178,870 Coke, refine petroleum 912,908 Supporting Transp. Activity 276, Recovery of Seconary -1,033,717 Chemicals -223,022 Coal an lignite etc. -398,452 Electricity, Water -1,242,907 Foo+ beverages + tobacco -267,399 Agriculture -1,499,579 Recovery of Seconary -387,936 Recovery of sec. raw material Coke, refine petroleum -747,871-1,197,625 Pulp, paper an paper -1,670,710 Agriculture -658,924 Electricity -2,392,096 Basic metals -2,249,058 Electricity, Water -3,419,583 Transport -3,659, Total Total Total Lan Transport 2,936,479 Electricity, Water 5,915,148 Electricity 3,151,804 Air transport 2,918,295 Lan Transport 3,152,912 Construction 2,655,576 Construction 1,644,503 Basic metals 2,488,594 Woo 1,654,834 Chemicals 993,466 Other non-metallic mineral 1,193,098 Agriculture 1,474,970 Basic metals 771,684 Construction 1,079,630 Chemicals 395, Agriculture -409,861 Membership organisation n.e.c. -17,998 Eucation -233,275 Coal an lignite etc. -413,535 Renting services of Recovery of sec. raw -26,062 machinery mat. -377,481 Pulp, paper an paper -415,094 Agriculture -48,579 Coal an lignite etc. -383,841 Recovery of Seconary -685,716 Recovery of Seconary -109,969 Coke, refine petroleum -724,186 Coke, refine petroleum -711,072 Chemicals -156,382 Lan Transport -5,004,275 The secon layer of Figure 8 reveals the reasons for that pattern in more etail. While there was a strong improvement in emission intensity, the intermeiate input structure an final eman le to more pollution. By analyzing the Leontief inverse it becomes clear that the major increase steme from the matrix of technical input coefficients. However, as mentione before, change in the intensity an in the matrix of technical input coefficients might be complementary as we observe the major change in A ue to changes in the self eman of electricity. Therefore, accounting issues coul lea to this strong manifestation of intensity an input structure. Final eman of electricity le to an increase of 52.6%, where the main part was ue to exports (31.7%). Finally, we can investigate the role of trae integration of the Austrian economy for the GHG emissions of the electricity sector. Austrian firms which rely more on omestic intermeiate inputs cause higher prouction in the omestic electricity sector an therefore increase its emissions. Due to insourcing, emissions in the prouction of electricity increase with 1.9 million tons. Outsourcing works in the opposite irection an cause a ecline of 1.3 million tons. On the final eman sie, the level of omestic eman was the most important river, causing an increase in electricity relate emissions of 22.5%. However, 11

14 the share of omestic eman for goos with omestic origin i not alter the emission balance much. The opposite was the case for the share of exports in final eman. The share of electricity prouce for foreign eman increase an ue to those changes, emissions associate with the prouction of electricity increase with 15.1% while it rose with 16.5% ue to the overall eman level. Figure 8: Decomposition Results for Electricity between 1995 an Conclusion We analyse the evelopment of greenhouse gas (GHG) emissions an its rivers by applying a hierarchical structural ecomposition (HDSA) analysis. Our moel is particularly suitable for investigating small open economies which are characterize inter alia by pronounce trae integration an outsourcing. The approach is applie to analyse the evelopment for the Austrian economy in the perio 1995 to We ivie the sample perio in the subperios 1995 to 2000, 2000 to 2005, an 2005 to Our results inicate that the main river in the first two subperios is the change in final eman level an in the last subperio changes in the structure of intermeiate eman, mainly in the electricity prouction. In the subperio from 1995 to 2000 the intensity effect an the structural effect were slightly absorbing the emission increase. A pattern change in the subperio from 2000 to 2005 where all factors are responsible for some increase in GHG emission. In the last perio, we observe a better performance in terms of emission intensity, but still have positive emission growth for the whole economy. Final eman is strongly riven by the prouction of export goos. In the structural linkage of the economy, trae integration ha a negative effect on emissions ue to an overweight of outsourcing compare to insourcing, a pattern which reverse in the last subperio. Prouction of electricity is the main river of increasing GHG emissions in the Austrian economy. For the first two subperios it was closely followe by lan transport, however, this service strongly improve in the last perio, leaing to a moerate negative balance over the whole time interval uner consieration. In general, it becomes obvious, that economic growth has so far not ecouple from the use of electricity an the efficiency gains are not yet sufficient to offset the eman effect. Therefore, future policy shoul aim at both, improvement of technology an reuction of carbon intensive eman. The overall performance of the Austrian economy with respect to GHG emissions has to be strongly 12

15 improve compare to the perio uner consieration in orer to achieve the emission targets of the Paris Agreement. Policy must further aim at improving the GHG balance of electricity. The pattern reveale in our stuy shows, that technological improvements have not offset the higher eman for electricity. The scope of omestic policy in changing eman patterns is rather small, especially as exporte goos play an important role in increasing emissions. Therefore, it is questionable whether European countries shoul focus on the prouction-base emissions as it is currently the case in the emission targets of the European Union. Small open economies can only influence own prouction, not the consumption patterns in other countries. As both, improvements in prouction technology an reuction of eman for emission intensive goos might be necessary, a consumption-base approach shoul be aopte as well. These policy actions woul improve competitiveness efine as the ability of a country to achieve Beyon GDP goals an woul be in line with a high-roa strategy for high-income countries suggeste in by Aiginger an Vogel (2015). A high-roa strategy in this sense inclues inter alia environmental ambition. This requires, in particular, shifting technical progress towar resource an energy saving, reucing subsiies for fossil energy an encouraging renewable energy. Future research shoul focus on more etails at the reasons of sector specific fluctuations. For instance, one surprising result in our analysis is, that services of lan transport performe quite poor in the secon perio with respect to emissions intensity but improve consierably in the last perio. Another important fiel of research woul be the aoption of a consumption-base accounting in the framework of a HDSA. References Aiginger, K., Vogel., J., Competitiveness: from a misleaing concept to a strategy supporting Beyon GDP goals, Competitiveness Review 25(5), ( 0052) Almon, C., Prouct-to-Prouct Tables via Prouct-Technology with No Negative Flows. Economic Systems Research 12 (1), ( Chang, N., Lahr, M. L., Changes in chinas prouction-source co2 emissions: insights from structural ecomposition analysis an linkage analysis. Economic Systems Research 28(2), ( Croner, D., Frankovic, I., A Structural Decomposition Analysis of Global an National Energy Intensity Trens. The Energy Journal 39(2), ( Dietzenbacher, E., Hoen, A.R., Deflation of input-output ta les from the user s point of view: A heuristic approach, Review of Income an Wealth 44(1), ( Guevara, Z., Rorigues, J. F. D., Structural transitions an energy use: A ecomposition analysis of Portugal. Economic Systems Research 28(2), ( 13

16 Hoekstra, R., Michel, B., Suh., S., The emission cost of international sourcing: Using structural ecomposition analysis to calculate the contribution of international sourcing to CO2-emission growth. Economic Systems Research 28(2), ( Kaltenegger, O., Löschel, A., Pothen, F., The Effect of Globalisation on Energy Footprints: Disentangling the Links of Global Value Chains. Energy Economics, in print ( Koller, W., Stehrer, R., Trae integration, outsourcing an employment in Austria: A ecomposition approach. Economic System Research 22(3), ( Lenzen, M., Structural analyses of energy use an carbon emissions an overview. Economic Systems Research 28(2), ( Malik, A., Lan, J., The role of outsourcing in riving global carbon emissions. Economic Systems Research 28(2), ( Pasurka, C., Perspectives on pollution abatement an competitiveness: Theory, ata, an analyses. Review of Environmental Economics an Policy 2(2), ( Peters, G. P., From prouction-base to consumption-base national emission inventories. Ecological Economics 65, ( Porter, M. E., van er Line, C., Towars a new conception of the environmentcompetitiveness relationship. The Journal of Economic Perspectives 9(4), ( UNFCCC (2012). Report of the conference of the parties on its seventeenth session, hel in Durban from 28 November to 11 December 2011 aenum. Part two: Action taken by the conference of the parties at its seventeenth session ( Voigt, S., Cian, E. D., Schymura, M., Verolini, E., Energy intensity evelopments in 40 major economies: Structural change or technology improvement? Energy Economics 41, ( Wiemann, T., Lenzen, M., Turner, K., Barett, J., Examining the global environmental impact of regional consumption activities Part2: Review of input-output moels for the assessment of environmental impacts emboie in trae. Ecological Economics 69, ( Xu, Y., Dietzenbacher, E., A structural ecomposition analysis of the emissions emboie in trae. Ecological Economics 101, ( 14

17 Appenix This appenix provies some backgroun an more etails concerning the construction of the inputoutput tables eflate to a common price basis, i.e. prices of In its core, the approach follows the one of Dietzenbacher an Hoen (1998) an has been evise in a similar form for the preparation of eflate input-output tables in previous work by Koller an Stehrer (2010). It is base on the construction of sectoral price inex vectors that are applie in the eflation process. Apart from sectoral price inices for omestic output an for imports, ifferent price inices were use for ifferent use categories (intermeiate input use, final consumption, fixe capital formation, exports). All price information employe has been taken from the official Austrian national accounts. The proceure proceee top-own an involve the application of biproportional fitting algorithm (RAS) in orer to ascertain the valiity of the horizontal balance equation. However, the vertical balance equation was not consiere in the proceure since eflation of value ae or its components was not an aim of the proceure. It has to be amitte that the proceure is of approximative nature for many reasons, one of them beeing that original price information is available only in the form of previous year prices. The eflation proceure can be characterize by the following steps: 1. Calculation of vectors of sectoral price inices for omestic output an imports, base on volumes an nominal values from national accounts 2. Construction of a 2 times 4 table of aggregate price inices for four ifferent use categories an separate for omestic an importe origin. This step involves RAS to make sure the restictions given by known margins are met. 3. Disaggregating price information for all 8 use categories to construct sectoral price inices. 4. Provisional eflation of the input-output table using the 8 ifferent price inices. 5. Application of a proportional correction to each row of the input-output table to make sure that the sum over all uses of commoity i equals total output or, respectively, total imports of commoity i. Furthermore, the ata proceure involve also the reclassification of the input-output table for the year 2010 from the ÖCPA 2008 to the ÖCPA 2003 classification, base on brige matrices provie by Statistics Austria. It shoul be note, that for the 2010 input-output table the reclassification proceure was applie after eflating it to the price basis of 2005 but before eflating from prices of 2005 to prices of