This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

Author's personal copy

Implementation of energy efficiency projects by Dutch industry

Christiaan Abeelen a,n, Robert Harmsen b, Ernst Worrell b

a NL Agency, Croeselaan 15, 3521 BJ Utrecht, Utrecht, The Netherlandsb Copernicus Institute of Sustainable Development, Utrecht University, Heidelberglaan 2, 3584 CS Utrecht, The Netherlands

H I G H L I G H T S

� Dutch voluntary agreements monitor all energy saving project by participants.� More than 20 000 energy saving projects have been implemented.� The reported saving effect seems to be fairly accurate.� Energy price is unlikely to be an important driver for savings.� Participation in ETS does not lead to higher investments in energy saving techniques.

a r t i c l e i n f o

Article history:Received 8 July 2013Received in revised form3 September 2013Accepted 11 September 2013Available online 3 October 2013

Keywords:Energy efficiencyIndustryVoluntary agreement

a b s t r a c t

As potentials for energy savings are huge, industry can provide a major contribution to energy savingsgoals. This paper focuses on the energy savings realized under the Dutch voluntary agreements in theperiod 2001–2011. Participants in these schemes are obliged to plan and implement all measures with apayback period of less than 5 years. This paper shows how many of these projects have beenimplemented and how much savings they generate. Our findings show that large differences exist inthe realized savings between individual companies. There is however no significant difference in savingsobserved between companies that participate in the Emission Trading System (ETS) and companies thatdo not. Although it is impossible to disentangle the drivers behind the implementation of these projects,the amount of savings suggest that at least part of them was implemented because of different energypolicy instruments.

& 2013 Elsevier Ltd. All rights reserved.

1. Introduction

As industry is responsible for 22% of final energy demand in theEuropean Union (EU)1, it is essential that efficiency improvementsalso occur in this sector to make the economy as a whole moreefficient. According to a study by Boßmann et al. (2012), realizingthe potential of industrial energy savings would result in 26%energy savings in 2030 compared to the baseline projection. TheEuropean Commission recognises the importance of energysavings and has set an economy-wide policy target of achieving20% energy savings by 2020 (European Parliament and Council,2012). The urgency to strengthen energy savings policies at theEuropean and Member State level is upfront as Europe is not ontrack in meeting its energy savings target (Harmsen et al., 2013;Wesselink et al., 2010). Actual implementation of new efficienttechnology is slow. Even technology that is profitable from an

economic view is slowly implemented (DeCanio and Watkins,1998). Much has been written about the barriers that obstructimplementation of energy efficient techniques (Fleiter et al., 2011;OECD/IEA, 2007; Johansson et al., 2007; Masselink, 2008). Differ-ent authors use different classifications of barriers, depending onthe scope of their topic. Cagno et al. (2013) developed a newtaxonomy that includes all barriers and tries to reduce overlapbetween barriers. This taxonomy has categorized barriers inbehavioural, technological, economic, organisational barriers andbarriers related to competences, information and awareness.

The Netherlands Court of Audit (in Dutch: Algemene Rekenka-mer), in an evaluation of Dutch energy policy, identifies theexistence of other investment opportunities as the most importantbarrier for firms (Netherlands Court of Audit, 2011). Especially iffinancial or other constraints play a role, investments in energysaving are likely to be pushed aside in favour of other investments.Other barriers are the resistance to replace existing machinery andthe relatively small amount of money spent on energy (De Grootet al., 1999).

Financial instruments dominate in industrial energy efficiencypolicy (ADEME, 2012). However, voluntary agreements (VA) have

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/enpol

Energy Policy

0301-4215/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.enpol.2013.09.048

n Corresponding author. Tel.: þ3188 602 2633.E-mail address: christiaan.abeelen@agentschapnl.nl (C. Abeelen).1 Source: IEA Statistics, 2009

Energy Policy 63 (2013) 408–418

Author's personal copy

also been implemented2, often in conjunction with financialinstruments. Co-operative policy instruments like voluntary agree-ments are especially popular in countries with close relationsbetween government and industry (ADEME, 2012). The establish-ment of voluntary agreements are a common approach to getstakeholders involved in energy efficiency policies (Geiss et al.,2009). Some of the implemented agreements have a special focuson energy saving projects to be implemented by the participants.The VA-programs in Australia, Sweden and the Netherlands, aswell as the Energy Efficiency Networks in Switzerland andGermany, are examples of programs that – each in their ownway – ask participants to implement and report energy savingprojects (Commonwealth of Australia, 2011; Johansson et al., 2007;NL Agency, 2008, 2009; Koewener et al., 2011).

The main purpose of the Dutch voluntary agreements is thespeeding of implementation of energy saving techniques. Thisprogram is – as we know it – one of the few which hasdocumented all types of projects that have been implemented bycompanies for more than 5 years and for a large share of industry.The aim of this paper is to analyse the actual implementation ofenergy saving technologies. We describe the savings reported bythe companies participating in the Dutch voluntary agreements.We look at the properties of the projects that have been imple-mented and the savings as a result of these projects. The questionis whether these savings actually show up in the energy use: dothese projects make the difference, or are other effects moreimportant? Are the savings as reported by companies an accurateestimate of real savings? Can we make a distinction between‘policy-induced’ savings and autonomous savings?

Therefore our first research question is:

(1) What is the impact of implemented energy efficiency projectson the actual development of energy use; in other words, arethe implemented projects a significant driver of the develop-ment of industrial energy use?As the results for different sectors and companies show rathergreat differences, we look at the characteristics of the compa-nies that yield the highest results, in an attempt to identify thepreconditions for higher results for the industry as a whole.This leads to the second research question:

(2) Which characteristics (such as size of the company, energyprice, participation in ETS or level of energy management)determine the energy savings effort of companies?The set-up of our paper is as follows: Section 2 addresses ourmethodology. Section 3 provides description of the dataset,including a brief history of industrial voluntary agreements inthe Netherlands. Section 4 shows the results of the analysis.Section 5 discusses the results and the various limitationsregarding the input data. Finally, Section 6 provides conclu-sions and recommendations.

2. Methodology

In order to answer our research questions, we analyse the datain the following way:

� We have assembled the necessary data from both the LTA andLEE covenant into a single dataset. Due to differences inreporting formats between LTA and LEE, it was necessary toadapt the data to make them comparable. These differencesregard especially the level of detail in reporting.

� The total group of companies consists of a wide variety ofsectors and fluctuates because of companies entering or leavingthe agreement. In order to produce a dataset from a moreuniform group of companies, without interference of changesin the covenant participants, we have singled out one distinc-tive sector for some analyses. The chemical industry waschosen as it is a large sector, both regarding the number ofparticipating companies and energy use, and therefore impor-tant for the result of the total VA.

For analysing the first research question, we make a compar-ison of the different factors that influence the energy use:production changes, saving projects and other effects. If theseeffects together add up to the change in energy use, it is plausiblethat the reported effects are correct. Besides, we can arrangedifferent effects by size, to define the main drivers for change inenergy use.

For analysing the second research question we define the meanyearly savings per company and compare this to several character-istics of these companies: size, participation in ETS, energymanagement and energy prices.

3. Data set

3.1. History of voluntary agreements in the Netherlands

The Netherlands have a history of voluntary agreements ofmore than 20 years, covering almost the complete industry. In thisperiod, five different agreements on industrial energy efficiencyhave been implemented, each with particular characteristics. Fig. 1gives an overview of the different VAs that were in force, with arough indication of their total energy use. In all cases, the VA wasbacked up by other policy instruments, mainly financial. Thecombination of these policy instruments is sometimes complex,but also offer a large degree of flexibility for companies (Abeelenand Both, 2012).

The first Long Term Agreement on energy efficiency started in1991. The use of voluntary agreements became a quite commoninstrument in Dutch environmental policy of the nineties (DeBruijn et al., 2003). The first VA were considered very successful(Das et al., 1997) and therefore succeeded by new agreements. In2000 the larger energy-intensive industry split off and joined thenew Benchmarking covenant (Phylipsen et al., 2002). About 900smaller companies joined the second generation VA (LTA2). Bothtarget and method of the benchmark covenant and LTA2 differed.For LTA2 target and method were copied from LTA1: an annual 2%improvement in efficiency. This target was not deemed suitable forthe large industries, as some of them had already reached largeimprovements in LTA1. Therefore, the target for the Benchmarkingcovenant was to reach the top 10% of most efficient installations inthe world. However, the results of the Benchmarking covenantwere disappointing. The progress report in 2007 revealed that therelative lead in the World Benchmark ranking on efficiency at thestart of the Benchmarking covenant had dwindled: Energy effi-ciency improvement from 2000 to 2007 was little over 0.5% peryear. Short pay back periods for investment in energy savingtechniques were mentioned as a reason for the low results(Davidson et al., 2010; VBE, 2008; Netherlands Court of Audit,2011). These results led to suspension of the Benchmarkingcovenant in 2008. Negotiations about a new agreement, morealong the LTA approach, started and after lengthy negotiations twonew voluntary agreements were signed. In 2008 LTA3 was signed,more or less with the same companies that participated in LTA2.In fact, LTA3 amends LTA2 and companies that joined LTA2automatically joined LTA3, unless they chose to leave the2 e.g. in 13 EU-countries (ADEME, 2012)

C. Abeelen et al. / Energy Policy 63 (2013) 408–418 409

Author's personal copy

agreement. A few large companies from the Benchmark covenantand two new sectors also joined LTA3. Most of the companies thatformerly participated in the Benchmark covenant signed a sepa-rate agreement in 2009: the Long term agreement Energy effi-ciency for ETS-companies (LEE), which officially is an amendmentto the Benchmarking Covenant, but has different targets and rules.

3.2. Companies joining the voluntary agreements

In 2013, two different agreements on energy efficiency are inforce: LTA3 and LEE. LEE, the Long term agreement Energyefficiency for ETS-companies, was signed in 2009 as a continuationof the Benchmarking covenant and was signed by most of thecompanies that formerly participated in that covenant. AlthoughLEE is meant in particular for companies that fall under the ETS-scheme, the distinction lines are not very sharp: Not all LEE-companies actually participate in ETS, while some companies inLTA3 also fall under the ETS-scheme. In total, 114 companies in7 sectors joined the LEE-covenant, with an energy use of 602 PJ(2011), see Table 1. The LTA3 (total energy use 237 PJ) is joined byover 900 companies in 32 sectors, mostly industrial sectors, butalso some services and rail transport (Table 2). Classification ofsectors was mostly based on the processes or products within thecompanies, but also partly on the level of organisation within asector, as the agreement was officially signed with the sectororganisations. Sectors without a representative organisation couldjoin the sector ‘non-branch related industry’. The share of thecompanies joining a VA compared to the energy use of total Dutchindustry fluctuated between 80 and 95%.

Companies joining a VA endorse both rights and duties stem-ming from the text of that agreement: (NL Agency, 2009) for LEE,(NL Agency, 2008) for LTA3. For different parts of the agreementseparate guidelines have been formulated. The monitoring guide-lines are described in (NL Agency, 2012a,2012b).

The most important obligation for companies within the VAs isto plan and implement energy efficiency measures. Therefore, they

have to deliver an Energy Efficiency Plan (EEP) every four yearsand a monitoring report on an annual base. The data used in thisarticle is taken from the monitoring reports of the participatingcompanies. The differences in population and reporting require-ments between the two agreements pose some restrictions on theavailability of comparable data. The most important restriction isthe fact that hardly any data, other than publicly available reports,are available from the Benchmarking covenant, as this wassupported by a special foundation under rule of confidentiality.At the start of LEE, companies have provided information from2006, as this is the reference year for LEE. This means that for LEEcompanies, data on company level are available from 2006onwards. Companies that joined LTA2 from the start have reportedfrom 2001 onwards. However, not all companies joined from thestart and not all companies have reported every year. The numberof participating companies has risen from 500 in 2001 to almost900 in 2011, with a comparable growth in energy use. A completedataset from 2001 onwards exists for 500 companies, for 850companies from 2005 onwards. For this article we have madechoices regarding the selection of companies and time period. If alarge time period is wanted, only LTA data are used. If LEE-companies have to be included, we must limit the time period to2006–2011.

There are large differences in size of the participating compa-nies (see Fig. 2). The smallest companies have an energy use of0.001 PJ, the largest use more than 70 PJ. Smaller companies areoverrepresented in the agreements: about half of the companieshave an energy use of less than 0.1 PJ. On the other side of thespectrum: the largest 10 companies use 51% of total energy. Thisskewed distribution means there is a large difference between theaverage and median of energy use and between weighted andnon-weighted average. In this article, weighted average is usedunless otherwise stated. In general it is true that the largercompanies participate in the LEE-agreement. However, the dis-tinction is not sharp and there is an overlap: The largest companyin LTA has an energy use of 20 PJ, while the smallest LEE-companies have an energy use of 0.1 PJ.

All energy (both consumed and saved) is converted intoprimary energy using national default heating values. For fuels,these values are based on research on the most commonly usedfuels (Vreuls and Zijlema, 2012). For electricity a conversion factorof 9 MJ/kW h is used (40% conversion efficiency), based oncalculations on the production mix in the Netherlands(Harmelink et al., 2012; NL Agency, 2012a,2012b). Companiesjoining the LEE are able to deviate from these default values, forinstance when using exhaust gases with a specific composition.In these cases company-specific conversion factors are used.

3.3. Implementation of measures

Participating companies must plan and implement all profit-able measures, whereby profitable means measures with a posi-tive net cash value at an internal discount rate of 15%.Alternatively, a payback period of five years can be used. Measuresfall in two (LEE) or three (LTA) distinct categories: processefficiency and chain efficiency for LEE measures and renewableenergy as the additional measure for LTA (NL Agency, 2008, 2009).For this article, only measures in the category ‘process efficiency’are assessed. This category is directly linked to the energyuse of the company itself.3 Process efficiency measures can be

1991 2000 2008 2009

2020

Ene

rgy

use

(PJ)

1000

800

600

400

200

BenchmarkingConvenant

(650)

LTA1(ca. 730 PJ)

LTA2(160)

LTA3(220)

LEE(626)

Fig. 1. Overview of voluntary agreements in Dutch industry. (Shown in the figure isthe approximate total primary energy use (excluding non-energy use) of theparticipating companies).

Table 1Sectors participating in LEE.

LEE Companies Energy use

Sector (NACE code)a # PJBreweries (11.05) 5 3Chemical industry (20–21) 58 319Glass industry (23.1) 9 13Metallurgical industry (24) 5 74Non-branch related industry (various) 13 18Paper and cardboard industry (17) 19 23Refineries (19.20) 5 152Total LEE 114 602

a A NACE code in this table does not imply that all companies under that codeare participating in LTA. In class 86.1 for instance, only the (bigger) academichospitals are participating, but not the other (smaller) hospitals.

3 The second category is chain efficiency. These measures improve energyefficiency not within the own company but elsewhere in the production chain fromraw material to end use and the energy supply needed for that purpose. Chainefficiency can be improved by: improving performance, reducing the amount ofmaterials needed, more efficient transport of goods and products, savings during

C. Abeelen et al. / Energy Policy 63 (2013) 408–418410

Author's personal copy

placed in four subcategories. Table 3 shows examples of the foursubcategories.

3.4. Efficiency plans and monitoring reports

Every four years, the participating companies must produce anenergy efficiency plan. These plans contain a description of theprocesses in the company, an energy balance and a list of projectsto be implemented in the next four years (NL Agency, 2008, 2009).For each project, the expected amount of energy savings is given.These deemed savings can be based on calculation, measurementsin test setting, specifications of suppliers or an educated guessusing assumptions for variables like e.g. output volume, produc-tion speed. The deemed savings are given as the saved amount ofprimary energy in Joules. To calculate the deemed saving percen-tage (SP), the absolute saving effect (S, in Joules) is divided by thetotal primary energy use (E, in Joules) in the most recent yearavailable at the time of planning (NL Agency, 2012a,2012b). So forthe efficiency plan that is made in 2012, the energy use in 2011 isthe reference energy use:

SPð%Þ ¼ S2011=E2011 ð1Þ

where SP is the saving percentage, S is absolute savings in Joule,E¼energy use in Joule.

Each year, companies must produce a monitoring report. Thisreport contains the following data:

� Use of energy carriers (electricity, heat, gas, and other fuels), inkW h, m3, tonnes etc

� Production (in physical units or an index for LEE-participants4).� Implemented saving projects� Other factors influencing energy use� Status of energy management (LTA only)

The most extensive part of the monitoring report is the part onenergy saving projects. For each project, the following data areregistered:

� Title of the project� (Sub)category (see Table 3)� Certainty level (certain, conditional, uncertain) at moment of

planning� Year of implementation� Expected saving (at the moment of planning) (in Joules)� Realised saving, in units of avoided final energy use (in Joules)� If necessary, an explanation is added

The monitoring reports are sent to NL Agency, the executivebody of the Dutch VAs, by internet application or email. All reportsare checked for completeness and correctness. These checks are infact a probability check, as a more thorough control (for instanceby a crosscheck with metering data) is not possible. Only inexceptional cases companies are visited to help them with theirmonitoring report.

Actual realised savings (AS) can differ from the savings asanticipated at the time of planning of the project. In the timebetween the planning and actual realisation of the project, whichin some cases can take years, many project variables may havechanged. After implementation of the project, companies have abetter view on the variables to calculate the realised savings. Eventhen, it is often necessary to use assumptions. The realised savingsare reported as absolute saved amount of primary energy, based

Table 2Sectors participating in LTA3.

LTA3 Companies Energy use LTA3 (cont.) Companies Energy use

Sector (NACE code)iii # PJ Sector # PJAsphalt industry (23.99) 41 3 Rubber and plastics industry (22) 98 9Carpet industry (13.93) 13 1 Potato processing industry (10.31) 15 8Chemical industry (20–21) 63 40 Rail sector (49.1–49.2) 2 14Coarse ceramics industry (23.3þ8.12) 40 8 Non-branch related industry (various) 39 14Cocoa industry (10.82) 6 2 Oil and gas-producing industry (06) 10 43Coffee roasting industry (10.83) 10 1 Refrigeration and cold storage industry (52.10) 85 3Dairy industry (10.51) 49 18 Sand limestone and cellular concrete industry (23.5þ8.12) 10 1Financial service providers (64–65) 10 3 Soft drinks, waters and juices (11.07) 9 1Fine ceramics industry (23.31) 7 1 Surface treatment industry (25.61) 61 2Flour manufacturers (10.61) 6 1 Vegetable and fruit processing industry (10.3) 20 3Foundries (24.5) 16 3 Tank storage companies (52.10) 16 2Higher vocational education (85.4) 35 2 Textile industry (13) 24 1ICT sector (63.1) 38 16 Textile service industry (96.01) 56 2Margarine, fats and oil industry (10.4) 17 8 Universities (85.42) 14 6Meat processing industry (10.1) 54 4 University medical centres (86.1) 9 5Metallurgical industry (25) 19 4 Waste water treatment district boards (37.0) 25 8

Total LTA3 917 237

Fig. 2. Distribution of total primary energy use of companies participating in LTAand LEE. (All fuels have been converted using default conversion factors. Forelectricity, a default efficiency of 40% has been assumed).

(footnote continued)the phase of product use (lower energy consumption or life cycle extension) orsavings arising from efficient and effective disposal of products (reuse, recycling,the use of material for energy generation). For LTA companies, renewable energy isalso an option as third category. These measures do not save energy use per se, butonly fossil energy use and thereby ‘green’ the energy use of the company. Althoughall three categories contribute to the result of the VA, they cannot be addedtogether.

4 The production-index is the ratio of the production in the reporting yearcompared to the reference year, weighted by the specific energy use. Theproduction level is based on one or more measures of production, in physicalunits. For reasons of confidentiality, only the resulting index is reported.

C. Abeelen et al. / Energy Policy 63 (2013) 408–418 411

Author's personal copy

on measurements or calculations. The relative effect is calculatedas:

RSxð%Þ ¼ ASx=ðExþASxÞ ð2Þ

where RS¼realised saving as a percentage of total energy use inyear x, AS is actual savings in Joule and E is primary energy usein Joule.

The denominator (ExþASx) is the energy used if the projectwould not have been implemented. The factor þASx in thedenominator is added because the energy use in the reportingyear is the energy use after implementation of the project. Forsmall projects this extra factor does not have a significant impact(being in order of 1–2%), but for large projects, or if many projectsare combined, the effect can be significant.

The calculated saving percentage is the percentage in thereporting year. To calculate the realised saving after several years,the yearly realised percentages are multiplied:

SI¼ 1�½ð1�RS1Þnð1�RS2Þn…ð1�RSxÞ� ð3Þ

where SI is the saving indicator, RS is the realised saving percen-tage in year x.

Projects are reported after realisation. An important assump-tion is that the savings will be in effect for the complete durationof the agreement (which runs to 2020). For building-relatedprojects (insulation) and projects that have to do with replace-ment of installations, this is a relatively safe assumption. Theduration of the agreement (12 years) is in the same range as thedefault project lifetimes (10–25 years) as assumed in Annex V ofthe Energy Efficiency Directive (EED) (European Parliament andCouncil, 2012). For projects under the subcategory energy man-agement, this assumption might not be true. The EED has classifiedthese measures as ‘short-term savings’. This uncertainty is dis-cussed in Section 5.

It is important to realise that the method in LTA3 and LEE –

savings as a result of implemented projects – is quite differentfrom the methods used under LTA2 and the Benchmarkingcovenant. In LTA2, the result was reported as an Energy EfficiencyIndicator (EEI), defined as the quotient of the actual energy useand a ‘frozen efficiency’ energy use. The target of LTA2 was anannual 2% improvement of the EEI. The Benchmarking covenantalso reported the EEI, although improvement of the EEI was noofficial target. As the EEI is dependent on factors not influenced bythe companies themselves, the covenant partners decided on anew method with a stronger link to their own saving efforts(which is the method described with formulas 1–3). Thereforeresults from LTA2 and the benchmarking covenant cannot beeasily compared with LTA3/LEE.

4. Results

4.1. Savings

Fig. 3 shows the relative savings of projects that have beenimplemented by LTA companies in the period 2001–2011 (asmentioned before; process efficiencies measures only). On average1.9% per year was saved (weighted average). In total about 20.000process efficiency measures have been implemented in the period2001–2011. More than half of the process efficiency projects fallunder the category ‘Adjustment in processes’, which makes senseas process installations account for the largest part of industrialenergy use (Sorrell et al., 2011). About a quarter of the savings aremeasures in utilities or buildings. The contribution of energymanagement projects is smaller, but represents a larger numberof small projects. Energy management is supposed to act as adriving force for other projects: it generates permanent attentiontowards energy use. Typically these projects require little to noinvestments. In some sectors, no or low-cost improvements inenergy management can lead to significant savings even beforeany investment is needed (IEA, 2011).There are not many strategicprojects, but these projects are typically larger. Strategic projectsare the most innovative, like the installation of a new plant or theintroduction of a new product.

As we can see, the resulting savings for the total LTA-population are between 1.5 and 2.5% per year (weighted average1.9%). There are however large differences between the companies.Some companies have implemented measures that save up to 50%over the total period, meaning savings of more than 5% per year onaverage. These large savings typically come in large steps, whenmajor installations are replaced. Therefore, there are also largevariations from year to year savings, at least for individualcompanies. On the intermediate level, the differences betweensectors are smaller, but still quite large. On average, the savings aretwice as high as the autonomous effect which, according to aliterature survey for an evaluation of Dutch climate policy, issupposed to be in the range of 0.8–1% (House of Representatives,2012, based on i.a. Rietbergen and Blok, 1999). It should be notedthat the reported savings are not isolated from the savings effectsof other policy instruments, in other words, are not fully ‘addi-tional’ to other energy policy instruments. Indeed, a sizeable partof the reported savings is also reported under the EIA-scheme, atax deduction instrument for energy efficiency investments byfirms (NL Agency, 2012b).

4.2. Impact on energy use

One of our research questions is whether implemented projectsdo actually reduce energy use. To answer this question we

Table 3Subcategories of process efficiency measures.

Subcategory Example

Energy management and goodhousekeeping

Awareness campaignsMonitoring systemImprovement of maintenance programs

Adjustment in processes New machinesDifferent process-setting (temperature,pressure etc.)New process design

Utilities and buildings Efficient lightingInsulationCHP (combined heat and power)

Strategic measures New production plantsReplacementsChange in product mix

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

% o

f prim

ary

ener

gy u

se s

aved

strategicutilities and buildings

processesenergy management

Fig. 3. Yearly savings by LTA2 and LTA3 companies as a result of implementedprojects.

C. Abeelen et al. / Energy Policy 63 (2013) 408–418412

Author's personal copy

visualise the impact of several influences on the development ofactual energy use. To minimize the impact of population changesand structural changes, we do this exercise for a single sector (i.e.the chemical industry) and for a restricted period (2006–2011).Because new companies have entered the VA and some companiesleft, the energy use in 2011 was 3.7% higher than the energy use ofthe original population only, see Fig. 4. Fig. 4 also shows thatduring 2006–2011 the energy use (without population change)has decreased with almost 7%, but that the years in between showa large variation. The strong decrease in energy use in 2008–2009is the direct result of the economic crisis.

Fig. 5 shows the differences between the energy use at the startand end of this period and the effect of several factors thatinfluenced the energy use. The change in energy use is decom-posed in four separate effects:

� Saving projects;� Changes in production volume;� Companies joining or leaving the VA;� Other influences.

This is not a formal decomposition method, but compriseselements of an additive structural decomposition method(Hoekstra and Van der Bergh, 2003), as the addition of the sectoraleffects together should match the change in energy use. The maindifference with formal decomposition methods is that it is built upfrom individual company data.

The effect of saving projects is the sum of the savings ofreported projects. The volume effect is calculated on the basis ofthe reported production figures, assuming a linear dependencybetween production and energy use. Although this assumption is

not entirely correct as it ignores the base load5, in most cases itgives a good fit. A shift in production towards companies with adifferent specific energy use (a structure effect), will be visible as avolume effect in this figure, as production figures are weighted byspecific energy consumption6. As this figure shows a relativelyshort period and only one sector, the assumption is that structuralchanges play a relatively small role. Lastly, other influences arereported by companies if special occasions have an impact onenergy use, like weather conditions or incidents.

Shown in Fig. 5 is a relatively small decrease in energy use of3%. As we can see in the figure, the effect of saving projects is quitelarge, compared to the total change in energy use. The net effect ofproduction changes is almost zero, but there were large produc-tion changes in the intermediate years. Production increases in2010 largely compensated for production decreases in 2008/2009(see also Fig. 4). The effect of decreasing and increasing productionis separated, showing that the gross volume-effect is larger thanthe effect of saving projects. Another large effect was the numberof new companies entering the VA scheme. Other influencesplayed a minor role. As not all changes could be accounted for,due to inaccuracies in energy and production figures and unac-counted other effects, a small residual remains, but together, thereported influences give a good fit to the change in energy use.This unaccounted part is an important criterion in determining thequality of the data. For large (LEE) companies, such an analysis isexecuted yearly, to explain the change in energy use. A largeunaccounted part indicates a possible error in the reported data oromitted factors that could explain the resulting difference. Severalof these possible errors are discussed in Section 5.

In the next paragraphs, we will look at the savings of thecovenant companies in more detail. By comparing realised savingson company level with characteristics of these companies, we tryto establish the characteristics of companies that realize highersavings. If we can find the drivers behind energy saving projects,we might be able to establish if these savings are autonomous orpolicy-induced.

4.3. Savings in relation to energy use

Because of the large differences from year to year we havecalculated the average yearly savings over the period 2006–2011.This period has been chosen because of availability of data for theLEE-companies.

In Fig. 6 the relationship between the realized savings and thesize of the company is shown. There are two reasons to expectlarger companies to realize lower savings: first, the assumptionthat larger companies pay less for their energy, making invest-ments in energy saving techniques relatively less profitable andsecond, the fact that larger companies are often also participatingin the ETS, as ETS-companies that have a shortage of emissionrights may opt to buy extra emission rights, which cannot beclaimed under the voluntary agreement. However, although largercompanies have realized slightly lower results, the difference isnot significant. Most companies report savings in the range of0–3% per year, but as we can see, even over a period of 6 year,several dozens of companies have reported savings in the range of

320

330

340

350

360

370

380

2006 2007 2008 2009 2010 2011

ener

gy u

se (P

J)

energy use without populationchanges

energy use

Fig. 4. Development of energy use of companies in the chemical industry (127companies in LTA and LEE).

Fig. 5. Decomposition of change in energy use between 2006 and 2011 in thechemical industry (127 companies in LTA and LEE).

5 In this context, base load is the energy consumption which is independent ofthe production level (e.g. lighting, space heating). In other words, a reduction ofproduction level will not result in a reduction in base load demand. Therefore, theenergy demand is likely to decrease less than the production.

6 For each product (or group of products) the specific energy consumption(SEC) in the reference year is determined (Et0/Pt0). In the reporting year, for eachproduct (group) the reference energy use (or frozen efficiency energy use) iscalculated by multiplying this SEC by the production in the reporting year:Ereft1¼SECt0xPt.

C. Abeelen et al. / Energy Policy 63 (2013) 408–418 413

Author's personal copy

5–10% per year. This would mean that those companies haveincreased their energy efficiency over this period with about 50%.Most of these companies have implemented radical changes intheir production plants, either by building new production lines orwhole plants or by introducing totally new products with otherenergy characteristics.

About 26% of the primary energy savings are natural gassavings, 66% electricity savings, the remaining 8% are savings onother fuels. Electricity savings are somewhat higher than expectedconsidering the share of electricity in total primary energy use(being 51%), whereas gas savings are somewhat lower than theshare of gas in primary energy use (41%).

Hardly any differences have been found between companiesthat participate in the ETS and those that are not. Although there isa small tendency towards higher savings for non-ETS-companies,this difference is not significant. When looking at the chemicalindustry, the results are roughly the same. Apparently, participa-tion in the ETS does not lead to higher investments in energysavings techniques, at least not in the observed period, whichcould be explained by modest carbon prices.

4.4. Savings in relation to energy costs

Several authors claim a relation between savings and energyprices (i.a. Boonekamp, 2011; IEA, 1997; De Buck et al., 2010). Tosee if this relation is valid for the reported savings in the VA, welook at the Dutch chemical industry again.

There are two possible hypotheses regarding the relationbetween energy prices and energy savings behaviour. The firsthypothesis is that the larger the share of energy costs in totalproduction costs, the larger the incentive to save energy. The logicbehind this hypothesis is that energy saving investments have tocompete with other investments and even profitable energysavings investments will not be implemented when competingagainst other more profitable investments (Swigchem et al., 2002,De Buck et al., 2010). De Groot et al. (1999) even claim that this isthe most important barrier for investments in energy savingtechniques. If the share of energy costs in total costs is higher,energy saving investments will rank higher compared to otherinvestments. Unfortunately, the monitoring data do not containdata on total production costs, therefore it is not possible toinvestigate this hypothesis on a company level. Using the energyprices from the efficiency plans, we can calculate total costs for

electricity for the participating companies at about 400 M€, whereasthe total costs for gas amount to 1300 M€. Together, gas andelectricity count for almost 60% of energy use in these companies,the rest is divided over a wide range of fuels (gas oil, coal, residualgases) and steam. If we assume the same price level for these fuelsas for gas and electricity, total energy costs for the covenantcompanies in this sector are about 2800 M€. The Dutch StatisticalOffice (2012) estimates production costs for the sector at 53742 M€.Calculated this way, total energy costs are 5.0% of total productioncosts. This is in line with Ramirez et al. (2005), who estimates theshare of energy costs at 1.5–8% for different parts of the chemicalindustry. In comparison, labour costs are a factor 2–20 higher. Wecan therefore conclude that for the chemical sector in total, the shareof energy costs is not very high and is unlikely to rank very high as amotivation for energy savings. This could be different for some veryenergy intensive companies or companies with a high share of non-energy use, as for these companies energy prices also determine theprice of their raw materials.

The second hypothesis is ‘the higher the price paid for energy,the larger the incentive to save energy’. The logic behind thishypothesis is that a higher energy price will yield higher benefitswhen applying energy saving investments, lowering the paybackperiod for these investments. The same investment will thereforebe more likely to be implemented in a period or a company withhigher energy prices. Several authors claim an effect of energyprices on energy efficiency. Boonekamp (2011) states that higherenergy prices will increase autonomous savings (in absence ofadditional policy). IEA (1997) mentions a clear relation betweenhigher oil prices and national energy efficiency. In industrialisedcountries an increase in energy efficiency with 1.5% to 2.2% peryear was seen after the oil crises in 1973–1985. Davidson et al.(2010) on the other hand suggest a relatively low price-elasticityfor energy demand in Dutch industry of 0 to maximum �0.2 inthe long term. This argument is supported by the fact thatefficiency improvement lags behind price increases, as it takestime to replace machinery. If prices stay high for only a shortperiod or if prices are expected to decrease again, motivation forinvesting in energy saving will probably decrease.

On average, larger companies pay less for their energy. Com-panies negotiate the price of electricity and gas with their energysupplier. This results in big differences, even within sectors. Ingeneral, the higher the energy use and the more constant thedemand, the lower the price (Van Dril et al., 2012). However,within the companies in the chemical sector that have joined theLTA, this general trend is hardly visible. As part of their efficiencyplans, companies have reported their 2011 prices that they use tocalculate the payback period for investments. This might not bethe same price as the price actually paid, as the latter price canvary within a year (some companies even operate on the spotmarket). However, the price used in payback calculations is betterfitted for our purpose, as this figure indicates the ‘value’ of energyfor a company. Because of confidentiality, a few companies did notdeclare their prices, but 95% did.

There are remarkable large differences between the pricesdifferent companies have to pay for their electricity and naturalgas. The median price for natural gas is 0.26 €/m3 ranging from0.17 to 0.45 €/m3. For electricity this is 0.07 €/kW h, ranging from0.05 to 0.12 €/kW h. Earlier studies found large differences inenergy prices between countries and also between differentsectors within one country (UNIDO, 2011). Apparently, the sameholds for companies in the chemical industry in the Netherlands.One obvious reason for these large differences could be the largevariation in energy use between companies: for gas the differencebetween the largest and smallest user in this sector is a factor100 000, for electricity the difference is a factor 2000. However,from the data in this sector we learn that the level of energy use is

y = -1E-07x + 0,0154R2 = 0,0117

y = -3E-07x + 0,0182R2 = 0,0033

0%

2%

4%

6%

8%

10%

12%

0 1 10 100 1.000 10.000 100.000energy use (TJ, logarithmic scale))

mea

n ye

arly

sav

ings

(% o

f ene

rgy

use)

ETS (n=156)non-ETS (n=605)

Lineair (ETS (n=156))Lineair (non-ETS (n=605))

Fig. 6. Relation between energy use and yearly saving percentage of companiesparticipating in LTA and LEE (n¼762, only companies that participated for thewhole period).

C. Abeelen et al. / Energy Policy 63 (2013) 408–418414

Author's personal copy

not the most important variable in defining the price: there is asmall trend towards a lower energy price for large users, but notsignificant. For smaller users the trend is a bit stronger, but still notsignificant.7 Apparently, above a certain level, prices are deter-mined on other grounds than demand alone. Another possiblereason for the large differences is that companies can choose froma large variety of contracts, with prices fixed for longer or shorterperiods. Given the large variations in energy prices in the lastyears, the moment the contract is signed and the period for whichprices are fixed in that contract can have a large effect. A thirdreason is fluctuation in demand profiles. Users with large varia-tions in demand, or a high demand at peak moments usually payhigher prices (Van Dril et al., 2012).

Composing a consistent time series of energy prices for Dutchindustry is difficult, as Statistics Netherlands introduced a newmethodology in 2007 with a different classification of users andprice components (Denneman, 2012). In general, both electricityand gas prices rose very fast in the period 2000–2007 andstabilised or even decreased in the period 2007–2001 (Eurostat,2012; Statistics Netherlands, 2012), see Fig. 7. For industry as awhole, energy costs rose much faster than value added (whichgrew 11% between 2000 and 2010). To a lesser extent, this is alsothe case for the chemical sector (value added þ47%) Therefore, therelative importance of energy costs has grown.

Fig. 8 shows the development of annual savings of LTAcompanies. By comparing this with the development of energyprices a relation can be found. Especially the period 2004–2008shows a high increase in prices, but no increase in savings. Theyears 2008–2011 show relative constant prices, but an increase insavings. If these two are correlated, there is a lag of four years. Thisfinding is in line with Saygin (2012), who compared energyefficiency improvements and energy prices over the period1993–2008 for 6 industrial sectors in the Netherlands, but couldnot find a clear relation. Literature suggests three possible reasonswhy energy efficiency is not directly related to energy prices:

(i). The relative low share of energy costs in most sectors(Ramirez et al., 2005), as is the case in for instance thechemical industry in the Netherlands

(ii). Unclear long-term development of energy prices (e.g.Boonekamp et al., 2002). The large changes in energy prices,especially in the period 2000–2006, could indeed add to theuncertainty of expected future price levels. This uncertaintyin itself could pose a barrier to investments in energyefficiency (Sorrell et al., 2011)

(iii). Possible delay between the increase in energy prices andinvestments in energy saving projects (Saygin, 2012).

We conclude that for the studied period a relation betweenenergy prices and investments does not exist in the case of theDutch industry as a whole, but could be true in individualcompanies with a relatively high share of energy costs. Thereforeit is not possible to establish if prices are the driving force behindpart of the investments in energy efficiency. It is plausible that partof the investments is driven by energy prices and thereforeautonomous, but which share of the investments is unknown.

4.5. Savings in relation to energy managements systems

In theory, an energy management system (EMS) is an incentivefor energy savings. An EMS is meant to assure constant attention

towards energy use and alert companies to new possible savings.Therefore, the Energy Efficiency Directive stimulates the adoptionof management systems (European Parliament and Council, 2012).

In LTA3, companies are obliged to have an EMS and arestimulated to certify their system with an ISO-certificate. If anISO system is in place, energy management should be integrated inthe energy paragraph. Fig. 9 shows the difference in energy usebetween companies that have an EMS with an ISO certificate andthose that have no certificate. As is to be expected, companies withan ISO-certificate are in general larger than those without acertificate. Companies with a certificate score somewhat better,but the difference is hardly significant. Within the two separategroups of companies with or without an ISO-certificate, there is nosignificant difference between larger or smaller companies. In away this small difference is not surprising, because an EMS is acondition for an ISO-certificate, but not the other way round. Inother words, companies without an ISO-certificate can have aperfectly well organised EMS. From the monitoring reports it isclear that almost all companies have at least some form of EMS,even given the fact that only 44% has an ISO-certificate. Therefore,the distinction between companies with or without an ISO-certificate is in practice not very meaningful.

5. Discussion

Although the monitoring reports from the participating com-panies form a valuable source of information on implementedenergy savings projects in the Dutch industry, this data source alsohas its limitations. The reporting format was drafted to satisfy thereporting requirements of a consensus agreement. There was nointention to use it for scientific analysis. Therefore the reporteddata does not always contain the information needed to answer allresearch questions. Another data complication is that not allcompanies participated in the agreements during the wholeperiod. Therefore choices had to be made regarding the periodand population under investigation. Finally, the data containssome uncertainty due to the fact that data were reported bycompanies themselves.

An obvious question regarding the new monitoring method isthe accuracy of the reported data. On a company level, productionand energy data will be easier to collect than data on realisedprojects. For many projects, it is only possible to estimate therealised savings. Therefore, the results obtained with the newmethod, which relies on project data, will probably have a lowerlevel of accuracy than those obtained with the old LTA2 methodthat used only production and energy data. Still, there is no reasonto expect that the level of inaccuracy is too high. Large mistakesare detected in the data audits that are structurally executed bythe executive body. Smaller errors are likely, but because of thelarge number of implemented projects it is likely that positiveerrors in one project will be compensated by negative errors in

60

80

100

120

140

160

180

200

220

240

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

inde

x (2

000=

100)

Fig. 7. Development of energy prices in Dutch industry (index 2000¼100).

7 y¼�5E�07xþ0,2723, R2¼0,0015. A more in-depth analysis to furtherexplore the relation between energy prices and investments would be interestingbut is beyond the scope of this paper.

C. Abeelen et al. / Energy Policy 63 (2013) 408–418 415

Author's personal copy

another project, therefore ruling each other out on an aggregatelevel.

There are reasons to expect companies to over-report realisedsavings. Especially companies that score lower results might feelstimulated to report more savings than actually been realized. Anevaluation study of the Danish energy efficiency obligation found arelatively small overestimation of projects reported by participantsof 6% (Bundgaard et al., 2013) In our study we found very fewexamples of projects that have been reported but turned out not tobe implemented. There are also many known cases of projects thathave been implemented, but have not been reported. Sometimescompanies forget to report projects, for instance in the case ofsmaller projects, sometimes companies report that a project hasbeen implemented, but cannot quantify the savings and thereforereport a zero saving. Knowing this, it seems reasonable to expectover- and underreporting to compensate each other, meaning thatthe inaccuracy on an aggregated level will be small. Bottom-updecomposition analysis of the changes in energy use for manycompanies and sectors show that the reported saving effect,volume effect and other influences together give a good fit forthe change in reported energy use, making it likely that the savingeffect is fairly accurate.

There is a large time gap between the planning and imple-mentation of projects, sometimes years. Therefore, actual realisedsavings can differ from the savings as anticipated at the time ofplanning of the project. In the time between the planning andactual realisation of the project, many project variables may havechanged. After implementation of the project companies have abetter view on the variables with which to calculate the realisedsaving. Even then, it is often necessary to use assumptions.A possible error is introduced if companies just copy the plannedsavings from their energy efficiency plan without checking ifproject variables have changed. For companies this is the easiestway, not only because they do not need to recalculate, but alsobecause it might raise questions if realised savings differ signifi-cantly from planned savings. In practice, this error seems small(o1%). Especially for bigger projects, companies tend to calculatethe actual realised savings at the time of realisation.

How long will a project have an effect? Projects are reportedonly once, after implementation. An important assumption is thatthe implemented projects will have an effect during the totalduration of the VA (2008–2020). For projects related to buildingsor process installations this is a safe assumption, as it is unlikelythat installed equipment is decommissioned. However, two impor-tant observations must be made. First, part of the reported projectsfall under the category ‘energy management and good house-keeping’. Part of these projects are behavioural measures (e.g.personnel training). Typically, these projects have only a tempor-ary effect and a lifetime of 12 years is unlikely, unless wellmaintained by the company. Companies are requested to reportthese kind of measures only once, but this rule is difficult touphold. Therefore, it is likely that the reported savings under thiscategory are an overestimation. As this is the smallest category, thetotal effect is not big, but could be 0.1–0.2% annually. A secondremark that could be made is whether it is justifiable to count thesavings of all projects that have been implemented during thisperiod. In case of replacement, one could argue that this replace-ment would have happened anyhow, only maybe some years later.Staniaszek and Lees (2012) argue that even with acceleratedinvestment, savings should be counted only during the remaininglifetime of original equipment. But Worrell and Biermans (2005)claim that it is ambiguous to determine ‘equipment lifetime’. Touse a project-specific lifetime would be difficult and errors arelikely, as one should make a distinction between replacementinvestments and other projects. It is easier to use the energy usebefore project implementation (this has the side benefit of makingexplanation of changes easier) and assume a certain share of‘autonomous’ improvements.

A very important discussion point relates to the additionalityof the presented results. Participating companies report imple-mented projects, but do not report the reason for implementation.Therefore, we cannot establish if a project is implemented becausethis was stimulated by participation in the VA or for anotherreason. In practice, several reasons will play a role. Therefore, thepresented savings cannot be added to the results of other policyinstruments. Also, part of the savings would have been realisedwithout any policy at all, as some of the implemented projects areregular replacements. This part would therefore be autonomous.How large this part is, cannot be established. In a literature surveyfor an evaluation of Dutch climate policy, the autonomous savingis supposed to be in the range of 0.8–1% (House of Representatives,2012), which would imply that about half of the reported savingsare autonomous and the rest is a combined effect of several policyinstruments. A special case is the additionality of the voluntaryagreements to the EU ETS. ETS-companies consume 2/3 of totalenergy use covered by the agreements. As the ETS cap onemissions should ensure a 20% reduction at the EU level, onecould argue that energy savings of these companies that avoiddirect CO2 emissions (from natural gas and other fuels) do notreduce emissions on a net base as these energy savings allow for

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

savi

ngs

per y

ear

Fig. 8. Savings per year of companies participating in the LTA covenant (n¼600–900, total energy use 160–230 PJ).

Fig. 9. Comparison of median energy use and mean savings of LTA-companies with or without ISO-certificate þ90% percentile (n¼654).

C. Abeelen et al. / Energy Policy 63 (2013) 408–418416

Author's personal copy

increased emissions elsewhere in the scheme outside the Nether-lands. Another special case of additionality is savings due to theimplementation of new products which are more efficient thanexisting products. One could treat these as a structure effectinstead of a savings effect. However, differentiation betweenincremental improvements of existing products and totally newproducts would be a research in itself, this discussion is beyondthe scope of this paper.

6. Conclusions and recommendations

The Dutch voluntary agreements LTA and LEE have system-atically promoted the implementation of energy saving projects.The LTA-program has monitored the implementation of theseprojects since 2001, the LEE-program since 2006. The characterof these measures ranges from motivational campaigns to thereplacement of complete production lines. On average, the pro-jects in the LTA2-program have realised savings of 1.9% per year,but there are large variations between sectors and companies.Individual companies have realised average annual savings over5%. It is not possible to establish the level of additionality of thesesavings, as part of these savings could be counted towardsautonomous savings and part towards other policy instruments.

To establish the accuracy of the reported savings, we havemade an analysis of the different factors that influence the use ofenergy: production, implemented savings projects and otherinfluences, all based on data reported by the companies joiningthe VA. Together, these influences give a good fit for the change inenergy use, making it likely that the savings effect calculated isfairly accurate. Although savings have a significant effect on energyuse, in most cases other effects are larger.

Our findings show that there is no correlation between the sizeof the company (in terms of energy use) and realised savings.Besides, realised savings by firms do not match the developmentof energy prices, because of the large time gap between planningand implementation of projects, the large fluctuations in prices inthe years 2008–2011 and the large uncertainty concerning futureprices. As the share of energy costs in total production costs inmost companies is quite low (o10%), the energy price is unlikelyto rank very high as a driver for energy savings.

According to our research, participation in the ETS does notlead to higher investments in energy savings techniques, at leastnot in the observed period, which could be well explained by themodest carbon prices in this period.

For further research it is recommended to look closer into thereasons behind implementation of savings, in order to develop abetter understanding of the contribution of autonomous savings.Related to this, is a more thorough exploration of the differencebetween energy savings and structure effects. Lastly, a study of thedelay between planning and realisation of projects, could shedmore light on a possible effect of energy prices.

References

ADEME, 2012. Energy Efficiency Policies in Industry. Lessons Learned from theODYSSEE-MURE Project.

Abeelen, C.J., Both, D. 2012. Energy Efficiency Measures: The Next Generation.ECEEE Summer Study Paper 1-065-12. 11-14 September 2012, Arnhem.

Boonekamp, P.G.M., 2011. How much will the energy services directive contributeto EU energy and emissions goal? Energy Efficiency 4 (2), 285–301.

Boonekamp, P.G.M., Harmsen, R., Kets, A., Menkveld, M., 2002. Besparingstrends1990–2000: Besparing, Instrumenten en Effectiviteit. ECN, Petten.

Boßmann, T., Eichhammer, W., Elsland, R., 2012. Policy Report: Contribution ofEnergy Efficiency Measures to Climate Protection within the European UnionUntil 2050. Report for the German Federal Ministry for the Environment,Nature Conservation and Nuclear Safety (BMU). Fraunhofer Institute forSystems and Innovation Research ISI, Berlin/Karlsruhe.

Bundgaard, S.S., Mikkelsen, K.D., Hansen Kjærbye, V., Togeby, M., Sommer, T.,Larsen, A.E., 2013. Spending to Save: Evaluation of the Energy EfficiencyObligation in Denmark. ECEEE Summer Study Paper 7-131-13. 3–8 June 2013,Toulon/Hyères, France.

Cagno, E., Worrell, E., Trianni, A., Pugliese, G., 2013. A novel approach for barriers toindustrial energy efficiency. Renewable and Sustainable Energy Reviews 19(2013), p290–p308. (March).

Commonwealth of Australia, 2011. Energy Efficiency Opportunities—ContinuingOpportunities 2011. Results of EEO Assessments Reported by ParticipatingCorporations. Commonwealth of Australia, Department of Resources, Energyand Tourism, Barton ACT, Australia.

Das, M.C., Driessen, P.P.J., Glasbergen, P., Habermehl, N., Vermeulen, W.J.V., Blok, K.,Farla, J.C.M., Korevaar, E.M., 1997. Evaluatie Meerjarenafspraken Over Energie-efficiency. Universiteit Utrecht, Utrecht.

Davidson, M., De Buck, A., Blom, M., Korteland, M., 2010. Evaluatie energiebespar-ingsbeleid in de industrie. Kosten en baten in de periode, 1995–2008(CE, Delft).

De Bruijn, T., Bressers, H., Lulofs, K., Van der Veer, A., 2003. Evaluatie Milieuconve-nanten. Eindrapportage. CSTM Universiteit Twente. Enschede.

De Buck, A., Blom, M.J., Smit, M., Wielders, L.M.L., 2010. Convenant BenchmarkingEnergie-efficiency: resultaten en vrijstellingen energiebelasting. CE, Delft.

De Groot, H.L.F., Verhoef, E.T., Nijkamp, P., 1999. Energy Saving by Firms: DecisionMaking, Barriers and Policies. Department of Spatial Economics. Vrije Uni-versiteit, Amsterdam.

DeCanio, S.J., Watkins, W.E., 1998. Investment in energy efficiency: do the character-istics of firms matter? The Review of Economics and Statistics 80 (1), 95–107.

Denneman, A., 2012. Toelichting statlinetabel eindverbruikersprijzen aardgas enelektriciteit. Statistics Netherlands The Hague/Heerlen.

Dutch Statistical Office, 2012 (http://www.cbs.nl), Accessed 30-11-2013.European Parliament & Council, 2012. Directive 2012/27/EU of the European

Parliament and Council of 25 October 2012 on Energy Efficiency, AmendingDirectives 2009/125/EC and 2010/30/EU and repealing Directive 2004/8/EC and2006/32/EC. OJ L. 14.11.2012, p.1.

Eurostat, 2012. ⟨http://epp.eurostat.ec.europa.eu/⟩, Accessed 07-12-2012.Fleiter, T., Worell, E., Eichhammer, W., 2011. Barriers to energy efficiency in

industrial bottom-up energy demand models a review. Renewable and Sustain-able Energy Reviews 15 (2011), 3099–3111.

Geiss J., Müller, A., Schüle, R., Arens, C., Höfele, V., Becker, D., Boermans, T.,Bettgenhäuser, K., Harnisch, J., Jaworski, P., 2009. Promoting Energy Efficiencyin Europe. Insights, Experiences and Lessons Learnt from the National EnergyEfficiency Action Plans. Brochure Presented by the Energy Efficiency WatchProject. Wuppertal, Cologne, Berlin, Brussels, January 2009.

Harmelink, M., Bosselaar L., Gerdes, J., Boonekamp, P., Segers, R., Pouwelse, H., Verdonk,M., 2012. Berekening van CO2-emissies, het primair fossiel energiegebruik en hetrendement van elektriciteit in Nederland. Report for NL Agency, Utrecht.

Harmsen, R., Eichhammer, W., Wesselink, B., 2013. An exploration of possibledesign options for a binding energy savings target in Europe. Energy Efficiency ,http://dx.doi.org/10.1007/s12053-013-9202-1.

Hoekstra, H., Van der Bergh, J.C.J.M., 2003. Comparing structural and indexdecomposition analysis. Energy Economics 25 (2003), 39–64.

House of Representatives, 2012. Parlementair onderzoek Kosten en effectenklimaat- en energiebeleid. kst-33193-2. ISSN 0921–7371, The Hague.

IEA, 1997. Indicators of Energy Use and Energy Efficiency. IEA/OEDC, Paris.IEA, 2011. The Boardroom Perspective: How Does Energy Efficiency Policy Influence

Decision Making in Industry?. Institute for Industrial Productivity, Paris.Johansson, B., Modig, G., Nilsson, L.J., 2007. Policy Instruments and Industrial

Responses-experiences from Sweden. ECEEE Summer Study Paper 7,048. 4–9June 2007 La Colle sur Loup, France.

Koewener, D., Mielicke, U., Jochem E., 2011. Energy Efficiency Networks forCompanies-Concept, Achievements and Prospects. ECEEE Summer Study Paper3-325. 6–11 June 2011, Presqu’île de Giens, France.

Masselink, D.J., 2008. Barriers to Investments in Energy Saving Technologies: CaseStudy for the Industry. ECN, Petten.

NL Agency, 2008. Long-term Agreement on Energy Efficiency 2001–2020. FinalVersion. 13 June 2008, The Hague.

NL Agency, 2009. Long-term Agreement on Energy Efficiency for ETS Enterprises(LEE). 2 October 2009, The Hague.

NL Agency, 2012a. Handreiking Monitoring, Version 2.2, 4 January 2012. NL Agency,Utrecht.

NL Agency, 2012b. Jaarverslag 2011 Energie Investerings aftrek (EIA). July 2012. NLAgency, Zwolle.

Netherlands Court of Audit, 2011. Energiebesparing: ambities en resultaten. kst-33016-1. ISSN 0921–7371. 5 October 2011, Den Haag.

OECD/IEA, 2007. Tracking Industrial Energy Efficiency and CO2 Emissions. OECD/IEA, Paris.

Phylipsen, G.J.M., Blok, K., Worrell, E., de Beer, J., 2002. Benchmarking the energyefficiency of the Dutch industry: an assessment of the expected effect onenergy consumption and CO2 emissions. Energy Policy 30, 663–679.

Ramirez, C.A., Patel, M., Blok, K., 2005. The non-energy intensive manufacturingsector. An energy analysis relating to the Netherlands. Energy 30 (5), 749–767.

Rietbergen, M., Blok, K., 1999. The Environmental Performance of VoluntaryAgreements on Industrial Energy Efficiency Improvement, NWS–99068.Utrecht University, Netherlands.

Saygin, D., 2012. Assessing Industrial Energy Use and CO2 Emissions; Opportunitiesfor Energy Efficiency, Biomass and CCS. Ph.D. Thesis. Utrecht University, theNetherlands.

C. Abeelen et al. / Energy Policy 63 (2013) 408–418 417

Author's personal copy

Sorrell, S., Mallett, A., Nye, S., 2011. Barriers to Industrial Energy Efficiency: ALiterature Review. UNIDO Working Paper 10/2011.

Staniaszek, D., Lees, E., 2012. Determining energy savings for energy efficiencyobligation schemes. ECEEE, RAP.

Statistics Netherlands, 2012. Monitor Topsectoren. Statistics. the Hague/Heerlen,Netherlands.

Swigchem, J., Keizer, I., Roos, J., Rooijers, F., Klaver, M., Strating, S., Klein Teeselink,H., 2002. Energie-efficiency in de industrie, ratio achter investeringsbeslissin-gen. CE, Delft.

UNIDO, 2011. Industrial Development Report 2011. Industrial Energy Efficiency forSustainable Wealth Creation. United Nations Industrial Development Organiza-tion, Vienna Austria.

VBE (Verificatiebureau Benchmarking), 2008. Monitoringrapport 1999–2007. Rap-portage Monitoring Resultaten van het Convenant Benchmarking., Utrecht.

Van Dril, T., Gerdes, J., Marbus, S., Boelhouwer, M., 2012. Energietrends 2012. ECNand Netbeheer Nederland. November 2012, Petten.

Vreuls, H., Zijlema, P., 2012. Nederlandse lijst van energiedragers en standaard CO2-emissiefactoren, version January 2012. NL Agency, Utrecht/Sittard.

Wesselink, B., Harmsen, R., Eichhammer, W., 2010. Energy Savings 2020: How toTriple the Impact of Energy Saving Policies in Europe Report for the EuropeanClimate Foundation.

Worrell, E., Biermans, G., 2005. Move over! Stock turnover, retrofit and industrialenergy efficiency. Energy Policy 33, 949–962.

C. Abeelen et al. / Energy Policy 63 (2013) 408–418418