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/copyright

Author's personal copy

Innovation in the U.S. building sector: An assessment of patent citations inbuilding energy control technology

Joy E. Altwies a, Gregory F. Nemet a,b,n

a Nelson Institute Center for Sustainability and the Global Environment (SAGE), University of Wisconsin-Madison, WI 53706, USAb La Follette School of Public Affairs, University of Wisconsin-Madison, 1225 Observatory Drive, Madison, WI 53706, USA

H I G H L I G H T S

c We investigate the innovation process for buildings in the U.S. using patents.c We use commercial and residential building controls technology as a case study.c Patenting peaked around 1980, declined, and then increased in the past decade.c Commercial building control patents account for most of the recent increase.c Inventions in electronics and computers have led to innovation in building controls.

a r t i c l e i n f o

Article history:

Received 7 June 2012

Accepted 21 October 2012Available online 11 November 2012

Keywords:

Energy management

Building automation

Patents

a b s t r a c t

Buildings are crucial to addressing energy problems because they are large consumers of end-use

energy, and potential exists to dramatically improve their efficiencies. However, the pace of innovation

in buildings is generally characterized as inadequate, despite the implementation of an array of policy

instruments aimed at promoting efficiency. The literature on innovation in the building industry

provides several explanations including: fragmented decision-making, principal agent problems,

inadequate information, and limited learning across heterogeneous projects. We investigate the

innovation process for buildings in the U.S. with a case study of patenting in energy management

control systems (EMCS) for commercial buildings and programmable thermostats (PT) for residential

buildings. Using U.S. patent data, we find that: (1) patenting activity peaked around 1980, subsequently

declined, and then increased considerably in the past decade; (2) commercial, rather than residential,

buildings account for the recent increase; and (3) building control technologies have benefitted from

inventions originating outside the industry, notably from electronics and computers, with a shift

toward the latter in recent years.

& 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Improving the energy-efficiency of buildings is central toefforts to address climate change and other energy relatedproblems (GEA, 2012). Building efficiency is important in partbecause such a large portion of final energy is consumed in them,and in part because the costs of energy savings there tendto be among the lowest. The IPCC comprehensively reviewedworldwide building sector mitigation opportunities and found‘‘substantial reductions in CO2 emissions from energy use inbuildings can be achieved’’ using existing technologies, and with

net benefits rather than costs (Levine et al., 2007). Urge-Vorsatzand Metz (2009) characterize building energy efficiency as the‘‘most important lever’’ available for climate stabilization in thenear term. Pacala and Socolow (2004) list emissions reductionsfrom buildings as one of the 15 climate stabilization ‘‘wedges,’’ ofsimilar magnitude to massive deployment of solar, wind, andnuclear energy.

A wide variety of public policies have been implemented toencourage innovation and adoption of energy-saving technologiesin residential and commercial buildings. While they have hadvarying degrees of success, the opportunity for further efficiencyimprovements is consistently found to be large. United NationsEnvironment Programme & CEU (2007) summarize the manypolicy options available for reducing emissions from buildingsector energy use and analyze policy effectiveness in over 80 casestudies from 52 countries. Urge-Vorsatz et al. (2007) find that it is

Contents lists available at SciVerse ScienceDirect

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

Energy Policy

0301-4215/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.enpol.2012.10.050

n Corresponding author at: La Follette School of Public Affairs, University of

Wisconsin-Madison, 1225 Observatory Drive, Madison, WI 53706 USA.

Tel.: þ1 608 265 3469; fax: þ1 608 265 3233.

E-mail address: nemet@wisc.edu (G.F. Nemet).

Energy Policy 52 (2013) 819–831

Author's personal copy

possible to achieve a 30% reduction in buildings’ emissions withcarefully designed combinations of policy measures. InternationalEnergy Agency (2008) recommends that governments takeimmediate action to implement policies related to energy use,including a list of policies specific to the building sector. Gelleret al. (2006) review the past 30 years of energy intensity datafrom OECD countries, demonstrating that energy efficiency policyefforts across all sectors of the economy resulted in 49% lessenergy use by 1998 than without those efforts. The gains have notbeen uniform, however. They note that energy intensity in theJapanese housing sector has actually increased over this time-frame, albeit from low levels relative to other countries. Ryghaugand Sorensen (2009) find that newer buildings in Norway usemore energy than older ones—a situation they do not attribute todifferences in the buildings themselves, but to a combination ofineffective policies and building industry culture. van Bueren andPriemus (2002) also find a lack of progress toward energyefficiency in the Netherlands’ building sector.

These results give rise to several questions of interest to policy-makers. If current technology can be used to reduce buildings’energy use at low or negative costs, why is the building industry notadopting it more quickly? Do particular characteristics of thebuilding industry make innovation and adoption of energy-efficient technologies different from other sectors? Resolving thesequestions would provide policy-makers with greater understandingof what motivates building industry decision-makers and wouldhelp them design and select effective policy measures that drivemore rapid energy efficiency improvements.

This study addresses these questions in two ways. First, wereview the literature related to innovation in the building indus-try. We include literature addressing the innovation process,challenges of reducing energy use in buildings, the role ofgovernment and policy, and the need for greater researchemphasis on the buildings sector. Second, we contribute to thisknowledge by investigating the innovation process of an energy-saving building technology. Using U.S. patent data, we trace thedevelopment of building control technologies known as energymanagement control systems (EMCS). We identify patents for thiscommercial building technology, along with a similar one forcontrol of the residential building environment, programmablethermostats (PT). We use the patent data to characterize trends ininventive activity over the past 40 years. We also use patentcitations to identify important patents and to assess the extent towhich these technologies benefitted from technology developedin other sectors, such as information technology.

Section 2 reviews the literature related to the process oftechnological change, the challenges of improving energy effi-ciency in the building sector, and the role of government andpolicy. Section 3 reviews the use of patents for studying innova-tion, and the selection of building controls as a case study. Section4 describes the approach to the patent analysis focusing on themethods used to define the technology and identify relevantpatents. Section 5 presents the results of the patent analysis,including descriptive statistics and trends. The paper concludeswith a discussion of the results and implications for policy andfuture research.

2. Technological change, building energy consumption, andthe role of policy

The need to reduce energy use in the building sector is arecurring finding in the literature. Buildings represent 40% ofprimary energy consumption in most countries (IEA 2008), andcontinued growth is expected in this sector (Levine et al., 2007;Energy Information Administration, 2011; International Energy

Agency, 2011a). By 2050, International Energy Agency (2011b)projects that residential households worldwide will grow by 67%,while commercial floor space increases by 195%. Improvements inenergy end-use technologies, including those in the buildingsector, are necessary in order to meet even the least stringentemissions stabilization targets (Kyle et al., 2011). Stimulating theprocess of technological change in buildings is therefore a grow-ing area of concern for policy-makers. This section reviews theliterature related to the innovation process, challenges posed bythe building sector, and the role of government in correctingmarket failures in energy efficiency.

2.1. The process of technological change

The process of technological change – commonly called ‘‘inno-vation’’ – involves invention, innovation, and diffusion (Grubleret al., 2012). Note that it is often confusing that the process of‘‘innovation’’ includes a stage called ‘‘innovation’’. The inventionand innovation stages, often studied together, involve the devel-opment of new products or processes, while diffusion describesthe adoption of a new technology by individuals or firms. Theefforts to create and continually develop new technologies –invention and innovation – are distinctly different from diffusionin the marketplace, which involves adoption decisions by end-users (Noailly, 2011; Noailly and Batrakova, 2010). However,Wilson et al. (2012) show that these processes are not indepen-dent, but in fact interconnected in an overall ‘‘innovation system.’’They offer examples of successful policy approaches thataddressed both technology development and market adoption.The authors show that successful innovation incorporates end-user feedback throughout the stages of technology development.Kiss and Neij (2011) provide a relevant example in building sectortechnologies. Their investigation into the development history ofwindow technologies in Sweden shows how policies can promotelearning processes, leading to improvement in the technology.Further, consumer acceptance is also part of this innovationsystems perspective. Darley (1978) highlighted the challenge ofgetting consumers to adopt new thermostats.

Recent work argues that systemic analysis of each phase isnecessary for the process of technological change to be under-stood, and ultimately used to inform policy (Gallagher et al.,2012). For example, a policy that promotes invention of energy-saving technologies will have little impact if the various actors inthe building industry do not adopt them. Conversely, policies thatgenerate demand ‘‘pull’’ from the marketplace can stimulateinventive activity by creating strong incentives for new technol-ogies (Nemet, 2009).

2.2. Characteristics of the building sector that affect

energy efficiency

The building sector is commonly divided into residential andcommercial categories based upon the primary usage of thebuilding. A ‘‘commercial building’’ is defined as any buildingdevoting at least 50% of its floor space to commercial activities,while a ‘‘residential building’’ includes any structure used pri-marily as a dwelling for one or more households (EnergyInformation Administration, 2012). The commercial buildingsector excludes industrial and agricultural buildings, and can befurther subdivided into many types, such as office space, hospi-tals, public institutions, retail, warehousing, and many more.Decisions regarding energy use and related technology adoptionare made by a wide variety of actors, depending on the age andtype of structure. For example, in a new commercial building, thedeveloper or owner, architect, engineers, and construction con-tractors will all have influence on the design and equipment

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831820

Author's personal copy

choices. In an existing commercial property, the facility manager,operations staff, and financial decision-makers will control thelong-term energy performance and equipment upgrade decisions.In residential properties, builders, homeowners, landlords, andtenants will impact energy performance depending on the type ofhome and its age.

The variety of actors involved in decisions leads to problems inimplementing energy efficiency strategies. The IPCC identifiestraditional design and construction processes that inhibit holisticdesign choices, misplaced incentives for saving energy, regulatorybarriers, perceived risk, and behavioral issues, among others(Levine et al., 2007). van Bueren and De Jong (2007) find a lowlevel of R&D investment among construction firms, as well as longbuilding lifespans, principal-agent problems, late agenda-setting,and aversion to risk-taking among building sector actors.Principal-agent problems, in which the person who must investin energy-saving technology will not reap the savings, are acommon barrier reported by many studies of the building sector.Studying Norway, Ryghaug and Sorensen (2009) describe threeinterrelated problems restricting energy efficiency: principal-agent problems, reliance on ‘‘general and indirect’’ forms ofregulation of the industry, and a conservative culture amongbuilding industry actors. When reporting the experience in theNetherlands, van Bueren and Priemus (2002) indicate that thebuilding sector ‘‘is characterized by decentralized and fragmenteddecision-making processes,’’ leading to ‘‘many missed opportunities.’’

For new buildings, the literature repeatedly finds that char-acteristics of the building design and construction industry createbarriers to technological change. The industry is distinguished bya large number of small firms, working in cooperation under strictschedule and budget limitations. Each new building is a uniqueconstruction project. This project-based nature of the industrymakes it fundamentally different from other sectors of theeconomy, with relatively little profit derived from efforts atinnovation (Brown, 2001; Lim et al., 2010; Tombesi, 2006).Learning from one project to the next is limited (Salter andGann, 2003; van Bueren and Priemus, 2002), with inadequatefeedback mechanisms to support technology innovation pro-cesses. Brown (2001) also finds that this industry invests verylittle in R&D: 0.2% compared to the economy-wide average of3.5%. Even the hiring and contracting processes used to engagethese firms create impediments to innovation. Christodoulouet al. (2003) explain that low-bid selection procedures, commonlyused to hire architecture, engineering, and construction firms(A/E/C), discourage innovation. When innovation does occur, ittypically does so by ideas shared among employees withina single firm (Salter and Gann, 2003; Kale and Arditi, 2010).This tacit and internal diffusion of knowledge indicates thatdecision-makers within these firms are less likely to be influencedby external information, policy efforts, or even their ownclients (Kale and Arditi, 2010). Linear design processes, concur-rent project duties, and extreme time constraints on eachproject create further disincentives for innovation within thesefirms (Salter and Gann, 2003). Harty (2005) provides contextfor understanding innovation in the construction industry byhighlighting the importance of collaboration, project-focus, com-munication, inter-organization relationships, and distribution ofpower. The competing priorities facing construction industrydecision-makers lead to diminished focus on energy efficiency.Diakaki et al. (2008) describe how the decision-maker in questionmust balance environmental and energy factors with financialand social factors when deciding among the options to selectfor a given building. Despite these challenges, recent emergingprocesses and tools may provide opportunities for improve-ment in design and construction outcomes. Integrated whole-building design (IWBD) processes encourage collaboration among

designers, while tools like building information modeling (BIM)promote coordination between design and construction teams ona project (American Society of Heating, Refrigerating, and Air-Conditioning Engineers, 2007; National Institute of BuildingSciences, 2012). These concepts have the potential to increaseinnovation in the building industry, and should provide futureresearch opportunities as their usage becomes more widespread.These studies on innovation in building design and constructionaddress processes and procedures within firms, as well as theselection of products and technologies for the buildings theycreate. Both types of studies are included here, with recognitionthat internal innovation processes within firms will ultimatelyaffect their profitability and competitiveness. A firm’s profitabilitywill, in turn, impact its willingness to take on the risks of adoptinginnovative energy-saving technologies for its clients’ buildings.Both types of innovation should therefore be of concern to policy-makers.

The challenges of reducing energy use in buildings are notlimited to new buildings. Existing residential and commercialbuilding stocks present ample energy efficiency opportunities.Clarke et al. (2008) and Ravetz (2008) provide examples from theUK, describing existing building stocks as generally poor qualityand often lacking insulation. Yet many energy efficiency upgradescan be implemented with net cost savings for the owner. This hasled to debate in the literature about the apparent ‘‘efficiency gap,’’where building owners fail to take advantage of energy efficiencyopportunities that offer reasonable rates of return on theirinvestment (Greene, 2011). Brown (2001) provides a definitionof this phenomenon, and finds that economics alone cannotexplain the slow adoption of certain end-use technologies. Amongthe challenges facing existing buildings, the author describesorganizational barriers such as ‘‘capital rationing’’ and ‘‘lack oforganizational rewards for energy managers who reduce utilitybills.’’ Jackson (2010) also discusses the ‘‘efficiency gap,’’ whileexplaining that simple payback financial analysis is not the bestapproach for energy efficiency investment decisions. Buildingindustry engineers and financial decision-makers routinely usesimple payback analysis when considering capital investmentoptions. This method does not adequately address risk manage-ment, however, and leads to under-investment in energy-savingtechnologies (Jackson, 2010).

2.3. Market failures provide justification for policy intervention

Allcott and Greenstone (2012) distinguish between two maintypes of market failures that cause under-investment in energyefficiency: energy-use externalities, such as pollution, and invest-ment inefficiencies, such as the principle-agent problem. Energy-use externalities are costs imposed on society due to energy use,but which are not reflected in the price paid for the energy.Gillingham et al. (2009) explain the economic basis of energyefficiency policies, defining the concepts of market failures,market barriers, and behavioral failures. Jaffe et al. (2005) findthat energy-use externalities are compounded by market failuresrelated to knowledge spillovers in technology innovation, inwhich the inventor of a technology is unable to capture its fulleconomic value. van Bueren and De Jong (2007) describe barriersin the building industry that have led to poor policy outcomes,including the number and variety of firms, long lifespans ofbuildings, and principal-agent problems. They also point outthat ‘‘projects only put sustainability on the agenda oncedecision-making is well under way,’’ thereby limiting optionsfor efficiency. Tombesi (2006) finds that firms are unlikely toinnovate on construction projects due to higher marginal costs,perceived profit risk, and low valuation of energy saving benefitsin future years. The variety of market failures impact all phases of

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831 821

Author's personal copy

the technology change process, from knowledge spillovers thatdiscourage invention to principal-agent problems that inhibitadoption.

The presence of these multiple market failures provides thebasis for government intervention, which can take many forms.Examples include building energy codes, utility rebate and cashincentive programs, tax incentives, mandatory and voluntarylabeling schemes, subsidies, and information campaigns (Urge-Vorsatz et al., 2007; United Nations Environment Programme &CEU, 2007). The literature on energy efficiency policy includessummaries and analyses, with examples at both the national andsubnational levels (Byrne et al., 2007; Carley, 2011; Dixon et al.,2010; Geller et al., 2006; Lee and Yik, 2004; Lund, 2007; Mundacaet al., 2010). In the U.S., policies implemented at the federal, state,and local levels affect building technology adoption decisions. Thecharacteristics of building energy codes are particularly importantand vary across states and municipalities. These regulatoryrequirements, among other functions, set minimum energy effi-ciency levels for new commercial and residential buildings.Misuriello et al. (2010) reviewed 50 studies of code complianceefforts in the U.S. and found that enforcement of the codes varieswidely among states.

2.4. The need for more research focus on energy end-uses and

the building sector

With a large opportunity amidst multiple challenges forefficiency improvement, has the building sector, and energyend-use in general, been given research focus commensuratewith its potential for energy savings? Wilson et al. (2012) arguethat directed efforts toward energy innovation are weighteddisproportionally toward energy supply rather than end use. Theydefine supply technologies as those ‘‘used to extract, process,transport and convert energy resources’’, and show that innova-tion in end-use technologies has provided greater impact, butreceived less support in research agendas and funding. Theauthors offer several possible reasons, including the difficulty ofdata gathering and analysis for the many end-use technologies; ahistory of well-established infrastructure supporting R&D invest-ment in supply-side technology; end-use technologies being lesssalient than those for supply; and end-uses lacking the large, andconcentrated, political influence of the energy supply industry.Dieperink et al. (2004) present a framework for explaining thediffusion of energy efficiency technologies, in particular in theNetherlands. Their study highlights how the diffusion of thesetechnologies is not well understood or documented. Noailly(2011) also describes a relative lack of focus on the buildingssector, from both the environmental and innovation literature.Sexton and Barrett (2003) characterize innovation research in theconstruction industry as being in its ‘‘infancy.’’ Similarly, Gannand Salter (2000) claim research into innovation has not ade-quately investigated ‘‘project-based’’ industries such as construc-tion, in which the ‘‘discontinuous and temporary natureyofproduction creates problems for rapid assimilation of new knowl-edge.’’ While end-use technologies have received attention frompolicy-makers, these studies would seem to indicate that empiri-cal research efforts are insufficient, even if increasing.

In summary, the related literature motivating this studyindicates that multiple market failures and peculiarities of thebuilding sector require a better understanding of the process oftechnological change, specific to energy use in buildings. Withoutthis knowledge, researchers and policy-makers will continue on apath of slow progress through trial and error, heuristic inter-pretations of ambiguous outcomes, and at best learning fromtheir own experiences and those of other policy-makers.

3. Analyzing innovation in building energy technology

In this paper, we focus on the invention phase of the process ofinnovation in buildings. We use patents as a proxy for inventiveactivity and select building energy controls as a representativecase study. This approach aims to contribute to the systemic viewon innovation described above, and complements both previouswork on adoption of these technologies and our own on-goingwork to identify the effects of policy on innovation.

3.1. Patents as indicators of invention

Patents provide an appealing data set to use for severalreasons, including large and consistent data over decades,detailed information on each invention, and the ability to dis-aggregate the data for specific technologies (Griliches, 1990;Johnstone et al., 2010; Popp, 2005a). While the fundamentalsocial purpose of patents is to encourage investment in technicalprogress by granting a temporary monopoly to the inventor, theiruse as a research tool has been mainly to exploit the detailedinformation included in them. In addition to the advantagesdescribed above, disadvantages in using patent data for studyinginnovation exist. For example, patents are used differently acrossfirms and across sectors, lags in the process from invention toapproval are heterogeneous, and patents by themselves offer noinsight into the adoption of a technology and only indirect insightinto its commercial value. In part by addressing these and othershortcomings, innovation research using patents has producedimportant results: that innovation responds to incentives, thatsocial returns to R&D are higher than private returns, and thatpolicy types do affect new inventions (Popp, 2005a). Johnstoneet al. (2010) also make connections to policy using data on energysupply patents, claiming that policy is a more important driver ofinventive activity than energy prices. They also find that specificpolicy types were more effective in certain situations. For exam-ple, feed-in tariffs worked well for technologies that are not nearmarket competitiveness, while renewable energy certificate poli-cies worked better for near-market competitive technologies likewind. Nemet and Kammen (2007) also use energy supply patentdata to study the outcomes of R&D investment trends, whichshow dramatic declines from the mid-1980s to mid-2000s. Daimet al. (2006) describe the application of patent trend analysis forforecasting the emergence of new technology. Studies have alsoused patents to analyze technologies in the building sector.Noailly and Batrakova (2010) and Noailly (2011) analyze patentdata on energy end-use technologies in European buildings,including insulation, heat pumps, lighting, and climate controls.In Noailly (2011), the author concludes that insulation standards(codes) and government R&D increase the likelihood of innovativeactivity, but energy prices have no effect.

While these examples show the various ways patent countscan be analyzed, the data within patents offers even moreopportunities for research. Specifically, patent citations have beenused to trace the flow of knowledge across technology sectors andto identify important patents (Nemet, 2009). When preparing apatent application, the inventor lists any previous patents that arerelevant to the current invention; these are known as patentcitations. In some cases, the patent examiner may also add relatedcitations (Popp, 2005b). Since all patents are assigned to a specifictechnology classification, these citations can be used to tracewhether a patent references previous technology from a differentclass. For researchers, these ‘‘backward’’ citations offer insightinto flows of knowledge both within and across technologydomains. Nemet and Johnson (2012) use patent classificationsas a method of determining technology domains and use citationsto trace flows of knowledge among them. While references to

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831822

Author's personal copy

previous patents provide information on the flow of knowledge,citations received by a patent after its publication are used byresearchers to gauge the importance of the patent (Harhoff et al.,1999; Hall et al., 2005). Trajtenberg (1990) uses these ‘‘forward’’citations to examine an example technology, finding a strongcorrelation between the number of citations received and theinvention’s real-world importance. Hall et al. (2001) describe theuse of citations for both types of research; we use an updatedversion of their data for the analysis in this paper.

3.2. Building controls as a representative case study technology

A special challenge in studying energy-end use technologies isthe sheer number and variety of available technologies (Wilsonet al., 2012). Contrast the thousands of possible end-use technol-ogies with the much more limited number of energy supplytechnologies. Selection of appropriate case studies is thus impor-tant. We examine the case of building controls, which can provideinsights about innovation in buildings in general, for severalreasons. They can be used in any type of building, of any vintage,and in any location; the systems come in a variety of configura-tions with a common objective; and they affect major sources ofbuilding energy consumption, heating and cooling.

The term ‘‘building controls’’ generally describes a combinationof devices and software used to operate heating, ventilating, air-conditioning (HVAC), and related equipment. In commercial build-ings, these control systems can be extended to incorporate lightingsystem operation, security and access systems, and other buildingfunctions. The terminology related to commercial building controlsincludes several common phrases, including building automationsystems (BAS), direct digital controls (DDC), energy managementcontrol systems (EMCS), energy management systems (EMS), andseveral others (American Society of Heating, Refrigerating, and Air-Conditioning Engineers, 1991, 2007). The various terms are oftenused interchangeably to refer to modern commercial buildingcontrols. Residential systems are usually simpler, primarily consist-ing of manual or programmable thermostats (PT). ASHRAE (1991and 2007) and Energy Information Administration (2012) are usefulresources for definitions of these terms, which will be described ingreater detail in Section 4 of this paper. For the purposes of thisresearch, the terms energy management control system (EMCS) andprogrammable thermostats (PT) will be used to describe thetechnologies being investigated (Table 1). We exclude any technol-ogy that does not meet these definitions, for example, manual ortime-clock type thermostats and pneumatic controls.

Building control technologies offer several advantages forstudying innovation in the building sector. Foremost, they canbe used in any building in any climate. Andrews and Krogmann(2009) analyze data from the Commercial Building Energy Con-sumption Survey (CBECS), which has been conducted every fouryears by the U.S. Department of Energy since the late 1970s, and

find that EMCS adoption is not strongly correlated to any regionor climate zone. Building controls are also advantageous as a casestudy because they are used in both residential and commercialsectors, allowing analysis of the entire building sector. While theexact devices are different, the functions they perform achieve thesame purpose. By comparison, many other energy end-use tech-nologies used in buildings are only advantageous in certainclimates, in particular building types, or in one sector of themarket. From a research standpoint, this attribute offers greaterdata availability and greater generalizability. The use of buildingcontrols for patent research also has precedent. Noailly andBatrakova (2010) include climate controls among a group ofbuilding efficiency technologies selected for their study of patentsin the Netherlands, finding that environmental policy has directimpact on innovative activity.

While EMCS and PT offer significant advantages as a case studytechnology, they also have limitations. Meier et al. (2010, 2011)present research showing that even when a programmable thermo-stat is installed, it may not be used properly to actually achieve thepotential energy savings. The commercial building sector is alsosubject to this phenomenon. Lowry (2002) surveys building opera-tors and their usage of EMCS systems, finding that they use themprimarily for equipment control and not for energy efficiency. Theoperators are aware of the energy-saving functionality, yet do notmake use of it. While this may appear to limit the usefulness ofbuilding controls as a case study, it must be noted that other energy-saving building technologies are also subject to limitations. Forexample, insulation or high-performance windows will only saveenergy if properly installed. In this study, the selection criterion wasa technology’s capability to save energy. When combined with theother selection criteria noted earlier, building controls offered themost advantages for this investigation.

When studying innovation using patents, it is helpful to select atechnology that has a historical record to analyze but is stillevolving, so that results are directly relevant to decision making.The U.S. Patent and Trademark Office (USPTO) has made patentinformation available in digital, searchable format for patents datingback to the 1970s, making a technology that has been developedduring the last 30–40 years ideal. The EMCS and PT technologies aregood fits for this research criterion, because they are also stillevolving. EMCS is continually developing with advanced controltheory, internet connectivity, ‘‘learning’’ thermostats, and neuralnetworks (Budiardjo, 2010; Dounis and Caraiscos, 2009). Meier et al.(2011) describe opportunities for improvement of PT technologiesthrough their suggestions for new usability metrics.

4. Approach

Our ultimate objective in this line of research is to conduct aseries of empirical tests to understand the effects of incentives on

Table 1Technology definitions.

Name Definition Source

Energy management

and control system

(EMCS)

An energy conservation feature that uses mini/microcomputers, instrumentation,

control equipment, and software to manage a building’s use of energy for heating,

ventilation, air conditioning, lighting, and/or business-related processes.

These systems can also manage fire control, safety, and security.

Not included as EMCS are time-clock thermostats.

EIA Glossary

Programmable

thermostat (PT)

A device that enables the user to set one or more time periods each day when a comfort

setpoint temperature is maintained and one or more time periods each day when an

energy-saving setpoint temperature is maintained. This device enables the user to save

energy because the heating and cooling equipment is not running needlessly at a

comfort temperature setpoint 24 h per day. A programmable thermostat may be capable

of controlling one or more zones of a conditioned space.

U.S. Environmental Protection Agency. Energy

Star Programmable Thermostat

MOU—Version 1.1

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831 823

Author's personal copy

innovation in the building sector, as discussed in Sections 2 and 3.As an intermediate step toward that end, this study provides anassessment of the characteristics and trends in inventive activityfor building control technology. We use data on patenting andpatent citations to represent efforts at innovation, the quality ofinventions generated, and the sources of knowledge on whichthese inventions are built. We include both commercial andresidential controls, using the terms EMCS and PT, respectively.We first define these technology categories more specifically, andthen describe the process we used to identify relevant patents.

4.1. Technology category definitions for EMCS and PT

For commercial buildings, EMCS consists of both hardware andsoftware components to operate heating, ventilating, air-conditioning (HVAC), lighting systems, and other building-widefunctions. In the residential buildings, a programmable thermo-stat can be used to maintain comfort and automatically adjust thecontrol of equipment to achieve energy savings. Table 2 providesdetail on how the technology works and what it controls inbuildings. We use these definitions to identify relevant patents.

4.2. Selection of relevant EMCS and PT patents

We identify patents for these technologies using several steps.The U.S. Patent and Trademark Office categorizes all patents intotens of thousands of classes and subclasses of technology areas.Because there is no class or subclass devoted to EMCS or PTtechnology alone, the first step was to find existing classes andsubclasses that contained relevant patents. A search for ‘‘HVACcontrols’’ and ‘‘thermostats’’ identified a large number of potentialpatents in various classifications. Classes 236 and 700 containedthe greatest number of relevant patents, which were scatteredamong several subclasses.

We reviewed each subclass for relevance to narrow down thelist. We then identified additional subclasses through consulta-tion with a USPTO patent examiner responsible for applications inthis area. Appendix contains a listing of these class/subclasscombinations, with supplementary information on each (Table 9).

Not all of the patents in these class/subclass combinations arerelevant to EMCS and PT technology. To filter out as manyirrelevant patents as possible, a Boolean search string was createdthat excluded terms such as ‘‘vehicle’’ and ‘‘refrigeration’’. Thesearch string was also amended to include the term ‘‘BACnet,’’ acommunication protocol for building automation and controlnetworks that is important to current EMCS technology trends.The Boolean search string is shown in Appendix.

These search parameters identified 2605 patents granted fromthe beginning of the U.S. patent system in 1790 until mid-2011.The first identified was granted in 1920. Even with the carefullycrafted search string, the resulting list of patents included many

that were not relevant to the building industry. We read thepatent title for each of these 2605 and coded each as eitherrelevant or not relevant, checking the patent abstract whennecessary to make a determination.

We also coded each relevant patent as either EMCS, PT, orapplicable to both. Most patents were initially marked as‘‘unknown’’ unless the title clearly indicated otherwise. Forexample, if the title included ‘‘building automation,’’ the patentwas marked with a ‘‘1’’ for EMCS. Any title that included ‘‘home’’or ‘‘home automation’’ was marked with a ‘‘2’’ for PT, even if thatpatent did not obviously relate to the thermostat. For any patentsthat could not be coded for relevance or EMCS/PT/Both based ontheir title and abstract (coded as unknown), the full patentpublication was reviewed. The list of relevant patents thusincludes not just devices, but also systems, control strategies,and methods related to energy management in commercial andresidential buildings. Table 3 summarizes the final results of thesearch and coding showing 1505 relevant EMCS and PT patents.Fig. 1 and Fig. 2 in Section 5 show time series of these patents byissue date. Table 11 in Appendix lists the most frequent patentrecipients. Honeywell, Carrier, Emerson, Siemens, and LG Electro-nics are among the top recipients.

4.3. Counts of forward and backward patent citations

We also analyze the citations made and received by each EMCSand PT patent. We use citations to previous patents, backwardcitations, to identify the knowledge on which EMCS and PTpatents are based. We use the citations received by these patents,forward citations, to identify which of these patents are impor-tant. To perform this analysis we merged the list of EMCS and PTpatents with patent citation analysis records published by theNational Bureau of Economic Research (NBER), as described inHall et al. (2001) and updated to include patents granted between1976 and 2006. By restricting the time frame to those grantedfrom 1976 to 2006, the number of EMCS and PT patents wasreduced from 1505 to 1104 for this portion of the research. These1104 patents made a total of 8544 backward citations, andreceived a total of 9819 forward citations. Of the 1104, 13.5%

Table 2Technology functions.

Technology Functions Components controlled

Energy managementand control system(EMCS)

Sensors measure temperature, pressure, humidity, airflow, etc., then controllers

adjust valves, dampers, actuators to maintain desired settings. Building operator

uses a computer workstation to create and change settings, respond to alarms,

perform scheduled maintenance (centralized control). Settings for individual

rooms and zones (groups of rooms) may or may not be controlled by occupants.

Computer software can be programmed for energy-saving functions (night

temperature set back, 100% outside air, demand-based ventilation, daylight-

based dimming, etc.)

Chillers, boilers, air handling devices, (all HVAC

components), lighting, security, others

Programmablethermostat (PT)

Occupant chooses desired temperature for each time and day of week,

thermostat controls automatically

Heating (furnace, electric heat, boiler for baseboard or

underfloor heat with associated pumps), heat pump, fans,

air conditioner

Table 3Final classification results (n¼2605 patents).

Classification Number ofpatents

% of 2605patents

% of relevantpatents

Relevant, Y¼1 1505 57.8 NA

Not relevant, N¼2 1100 42.2 NA

EMCS¼1 672 25.8 44.7

PT¼2 401 15.4 26.6

Both¼3 432 16.6 28.7

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831824

Author's personal copy

received no citations. We apply a definition of highly-cited toidentify important patents. To account for changes in citationpotential over time – newer patents have fewer chances of beingcited – we compare each patent’s forward citations to those of otherpatents in the same year. We select the 75th percentile as thethreshold and define highly-cited as any patent with forwardcitations Z75th percentile of all patents granted in the same year(Nemet, 2012). For recent years in which the 75th percentile is zerocitations, we use the definition 475th percentile. Percentiles arebased on forward citations for all patents in each year, not just theEMCS & PT patents in that year. For example, a patent granted in1976 would have to receive 11 or more citations to be consideredhighly-cited. The number of citations received includes all citationsreceived by the last year of data available, 2006.

5. Results

In this section, we describe the EMCS and PT patent character-istics and trends for patent counts, citations made and received,and highly-cited patents. We use these important patents tofurther identify flows of knowledge from other technology areas.

5.1. Descriptive statistics and trends

Fig. 1 shows combined EMCS & PT patents granted in each yearfrom 1970 to 2010. The time series is shown as a percentage of allpatents in order to eliminate the effect of the substantial increasein total patenting activity over time. Note that 0.08 refers to 0.08%not 8%. The series by application year exhibits a similar shapewith a typical 2–3 year negative lag compared to grant year (seeSupporting information Appendix B).

Fig. 2 shows the time series of 672 EMCS-only patents by yearof publication and the 401 PT-only patents, as a percentage of allpatents each year. Compared to PT patents, the publications forEMCS patents show a stronger increasing trend beginning in themid-1990s. Fitted lines are shown for the 1995–2010 range, withestimated slopes of 0.000670 for EMCS and 0.000068 for PT.While not shown on the figure because they only cover a partialyear, patents granted through mid-2011 are included in thedescriptive statistics as noted.

Table 4 includes descriptive statistics about citations and othercharacteristics for the 736 EMCS and PT patents granted from1981 to 2001. The patents are restricted to these years in order touse a 10-year forward citation window, so that more recent

Table 4Descriptive statistics for 1981–2001 EMCS and PT patents (n¼736 observationsa).

Variable Mean Std. Dev. Min Max

Citations received (within next 10 years) 7.125 9.230 0 101

Citations made (previous 10 years) 5.099 5.501 0 43

Citation laga 4.851 1.740 0.667 10

Claims made 14.47 13.07 1 126

Grant year 1991.1 6.261 1981 2001

U.S. corp. 0.507 0.500 0 1

U.S. govt. 0.001 0.036 0 1

EMCS 0.389 0.488 0 1

PT 0.318 0.466 0 1

Both 0.293 0.456 0 1

a For the citation lag variable, n¼681.

Table 5Inter-domain knowledge flows from previous patents to highly-cited EMCS and PT patents, excluding EMCS–EMCS pairs (n¼858 ‘‘far external’’ pairs).

Source (‘‘cited’’ patent) Destination (category of EMCS/PT ‘‘citing’’ patent)

Chem (%) Comp (%) Med (%) Elec (%) Mech (%) Other (%) Total (%)

Chem – 1.0 0 0 0.1 2.1 3.2

Comp 0 – 0 0.9 0.1 35.0 36.0

Med 0 1.0 – 0 0 0.1 1.1

Elec 0 15.4 0 – 0 14.5 29.9

Mech 0.2 2.7 0 0 – 4.4 7.3

Other 0.1 20.6 0 0.7 0.9 – 22.3

Total 0.3 40.7 0.0 1.6 1.1 56.1 100

Table 6Inter-domain knowledge flows from previous patents to highly-cited EMCS and PT patents, change from 1980 to 2006, excluding EMCS–EMCS pairs.

Source (‘‘cited’’ patent) Destination (category of EMCS/PT ‘‘citing’’ patent)

Chem Comp (%) Med Elec Mech Other (%) Total (%)

Chem – þ2.0 – – – þ2.0 þ4.0

Comp – – – – – þ65.2 þ65.2

Med – – – – – – –

Elec – �27.4 – – – �4.2 �31.6

Mech – þ2.0 – – – þ2.0 þ4.0

Other – �41.7 – – – – �41.7

Total – �65.1 – – – þ65.0 –

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831 825

Author's personal copy

patents have a similar opportunity to get cited. The table includescitations received as late as 2011 for patents granted in 2001. Oneinteresting observation is a near total lack of patents assigned tothe U.S. government in this technology: only 1 patent out of 736(0.14%). By contrast, Hall et al. (2001) find that for all patentsgranted through 1999, U.S. government patents make up 1.7% ofthe total. The relevant patents include both U.S. and foreignapplicants, with approximately 51% awarded to U.S. entities.

5.2. Highly-cited patents

In this section of results, to address the wide dispersion invalue across patents, we analyze the subset of highly-cited EMCSand PT patents using the definition above (Z75th percentile ineach year). For the dates 1976–2006 available in the NBER data,this definition includes 396 patents for EMCS & PT technology. Toqualify as the 75th percentile or higher, the number of citationsreceived ranges from 1 for grant years 2004, 2005, and 2006, to 14citations for grant year 1989. Fig. 3 shows these 396 highly-citedEMCS & PT patents by year of publication. Data truncation issuesduring the last few years affect the time series. Because the dataare counts, exactly 1 citation was required for all patents pub-lished in 2004, 2005, or 2006 to be considered highly-cited.However, patents in 2004 had more time to receive citations thanthose in 2005 and 2006. The decreasing trend in publications from2005 to 2006 is at least partially due to the limited time availableto receive a citation. Table 10 in Appendix lists the top 15 highly-

cited EMCS & PT patents, along with details on each patent. Themost highly cited patents received more than 100 citations andmore than 10 times the median for their grant year cohort.

5.3. Flows of knowledge to EMCS and PT patents

Whereas forward citations measure value, analysis of backwardcitations provides insight into the sources of knowledge for thesepatents. Hall et al. (2001) (HJT) aggregate patent classes to code allpatents into one of 6 HJT categories: (1) Chemical, (2) Computersand Communication, (3) Drugs & Medical, (4) Electrical & Electronic,(5) Mechanical, and (6) Others. Since the USPTO class 236 iscategorized as 6 Other, and USPTO class 700 falls under category 2Computers and Communication, the majority of relevant EMCS & PTpatents are in these two categories. If a patent cites a previouspatent from a different HJT category, this is defined as an ‘‘external’’pair. This coding of citation pairs is useful in determining the extentto which knowledge is crossing boundaries by flowing from onedomain of technology to another (Nemet and Johnson, 2012).

Table 5 shows sources of knowledge for the 396 highly-citedEMCS & PT patents. We exclude any pairs where both the citing andcited patents are EMCS or PT. We refer to these pairs as EMCS–EMCSpairs. We also drop all pairs for patents granted in 1976 (32 patents),due to lack of data on the characteristics of patents granted before1976. The remaining 325 patents made 858 citations to prior artoutside of their HJT categories. These external citation pairs areshown in the table, with the columns representing the categories ofthe EMCS & PT patents, and the rows representing the categories ofthe previous patents they cite. In essence, knowledge is flowing fromthe categories in each row to the categories in each column. Thevalues shown are percentages for each category combination of thetotal 858 external citation pairs. To clarify interpretation of Table 5the following observations can be made: 15.4% of external knowledgeflows among highly-cited EMCS/PT patents were from electronics tocomputers; 40.7% of highly-cited EMCS/PT patents are categorized ascomputers; 7.3% of external knowledge used by highly-cited EMCS/PTpatents came from the mechanical category.

Table 7Most frequently referenced classes by highly-cited EMCS & PT patents granted 1976-2006.

USPTO class ofprevious patent

Class name Total number of references toprevious patents in this class

236 Automatic temperature and humidity regulation 894

700 Data processing: generic control systems or specific applications 674

165 Heat exchange 445

62 Refrigeration 295

340 Communications: electrical 242

702 Data processing: measuring, calibrating, or testing 90

307 Electrical transmission or interconnection systems 80

709 Electrical computers and digital processing systems:

multicomputer data transferring

66

315 Electric lamp and discharge devices: systems 56

454 Ventilation 52

379 Telephonic communications 48

375 Pulse or digital communications 32

318 Electricity: motive power systems 30

374 Thermal measuring and testing 30

219 Electric heating 29

715 Data processing: presentation processing of document, operator interface

processing, and screen saver display processing

27

237 Heating systems 26

392 Electric resistance heating devices 26

705 Data processing: financial, business practice, management, or cost/price determination 25

370 Multiplex communications 24

707 Data processing: database and file management or data structures 23

713 Electrical computers and digital processing systems: support 23

47 Plant husbandry 21

345 Computer graphics processing and selective visual display systems 20

Table 8Composition of citations received for highly-cited EMCS and PT patents citing

some external prior art and for those citing no external prior art.

Citations receivedfrom all patents

Citations received fromother EMCS & PT patents

Mean n Mean n

Patents citing Z1 external 19.5 341 7.5 318

Patents citing 0 external 6.0 55 2.8 34

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831826

Author's personal copy

The results show a strong connection between the Computer &Communication category and the Other category. Some of thisconnection is possibly due to inconsistent categorization by patent

examiners of EMCS and PT into 236 (Computers) or 700 (Other).There is also a notable flow of knowledge from Electrical andElectronics to both Computers and Other. This seems a reasonableresult, as the EMCS and PT technologies can be described as digitalor electronic control systems. Compared to a similar analysis ofenergy technology in general by Nemet (2012), EMCS and PT aremuch more likely to cite Computer and Communication and aremuch less likely to cite Mechanical and Electrical patents.

Comparing knowledge flows for individual years 1980 and 2006shows a distinct shift in the citation patterns over time. In 1980, mostexternal knowledge was coming from the Electrical and Electroniccategory. By 2006, the majority of knowledge comes from theComputer and Communications category. Table 6 shows the changefrom 1980 to 2006, with the largest increase (þ65.2%) in knowledgecoming from the Computer and Communications category.

Table 7 lists the specific USPTO classes that highly-cited EMCS &PT patents reference most often. In contrast to the focus on externalcitations above, this list covers all backward citations made by theseimportant EMCS and PT patents, and includes EMCS–EMCS pairs.The third column lists the total number of references to that class,and includes multiple references to the same class by a singlepatent. For example, if an EMCS patent lists 10 references, and eightare in class 700, then eight are counted toward the total shown inthe table for class 700. Hence, the totals shown are higher than thenumber of highly-cited patents. The list includes any class cited atleast 20 times. One can see classes representing the control systems,communications, as well as the heating and cooling equipment.

Table 8 identifies what types of future patents are citing the396 highly-cited EMCS & PT patents. This information describeswhether EMCS & PT knowledge is flowing primarily to otherEMCS & PT patents or elsewhere. The table includes patentsgranted between 1976 and 2006, and is not adjusted for trunca-tion effects. Since only 352 out of 396 highly-cited patentsreceived citations from future EMCS and PT patents, the countsin the last column only add to 352.

6. Discussion

The number of EMCS and PT patents over time reveal sometrends. As a percentage of all patents, this technology shows anincrease beginning in the mid-1970s and tapering off in the late1980s, as seen in Fig. 1. This time series is similar to patenting

Fig. 2. EMCS and PT patents separately, as a percentage of all patents granted

between 1970 and 2010.

Table 9USPTO class/subclasses searched.

Classno.

Class name Sub-classno.

Subclass name No. of patentssince 1976

No. ofpatents

236 Automatic temperature and humidity regulation 1R Miscellaneous 33 733

1B Zone control for heating and cooling medium 140 291

1C Heating and cooling controls 158 499

1E Multistage controls 61 465

9R Combined Heater and apartment controlled 11 608

9A Combined heating and apartment control with heating

medium circulation control

30 199

44R Humidity control 91 1061

44C Humidity and temperature control 249 412

44A Humidity control per se 126 376

46R With timing element 420 1074

46A Diverse sensor 12 59

46C Controlled diverse means 25 77

46F Timer other than clock 87 133

47 High and low temperature alternate 153 225

51 Distance-adjusted 397 497

700 Data processing: generic control systems or

specific applications

276 HVAC control 432 433

277 Multiple zones 144 145

278 Specific thermally responsive controller 165 167

Fig. 1. EMCS & PT patents as a percentage of all patents by year, for all patents

granted between 1970 and 2010.

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831 827

Author's personal copy

patterns in many energy technologies, as well as to energy prices.Whereas EMCS and PT account for an equal share of patents for mostof their history, EMCS patents show a relative increase beginning inthe late 1990s (Fig. 2). Looking at highly-cited patents alone in Fig. 3,one can see a cluster of valuable patents in the early 1980s and whatappears to be a second one in the mid-2000s.

What information can be gained from these trends? The purposeof studying patents for this technology is to gain insight into thefactors that stimulate innovation in the building sector. It appearsthat something was driving activity in both EMCS and PT technologyduring the 1970s and 1980s. The energy price shocks of the 1970swould be a logical driving factor, as would associated investments in

Table 10Top 15 highly-cited EMCS & PT patents, 1976–2006 publication range. A¼citations received/median of grant year, B¼count of citations received.

Patent number, title (A) (B) Grant year, assignee Abstract excerpt or description

5621662, Home automation system 14.6 102 1997, IntelliNet, Inc. ‘‘A home automation system comprises a number of sub-systems for

controlling various aspects of a house, such as a security sub-system, an

HVAC sub-system, a lighting control sub-system, and an entertainment

sub-system. The network comprises a host computer connected through a

host interface to a plurality of nodes.’’

5311451, Reconfigurable controller for

monitoring and controlling environmental

conditions

11.9 107 1994, M. T. McBrian Company,

Inc.

‘‘A highly distributed direct digital process control system for use in

controlling a fully distributed process includes at least one device

controller independently monitoring and controlling a plurality of

external devices for performing a complete process.’’

5103391, Control system for controlling

environmental conditions in a closed

building or other conditions

11.7 88 1992, M. T. McBrian Inc. ‘‘A highly distributed direct digital process control system for use in

controlling a fully distributed processy The central information

processing means is capable of updating control information used by

specific controllers.’’

5622221, Integrated zoning circulator with

priority controller

11.3 79 1997, Taco, Inc. ‘‘An integrated zoning circulator includes a priority controller, integral

with the zoning circulator, for determining the priority given to the

circulator in an individual zone of a multi-zone hydronic systemy the

zoning circulator also includesy a plurality of control boxes each having

a thermostaty’’

6192282, Method and apparatus for improved

building automation

11.1 50 2001, Intelihome, Inc. ‘‘An improved building automation system is provided which is modular

in design thus minimizing the amount of instruction necessary to affect

control of a particular building system.’’

4071745, Programmable time varying control

system and method

9.3 74 1978, Hall B C ‘‘An electronic system and method are disclosed for controlling a

measured temperature during a sequence of time intervalsy the values

of the reference temperatures are programmable and changeable by the

user.’’

5602758, Installation link-up procedure 9 63 1997, Gas Research Institute ‘‘An apparatus and method for preparing an indoor environmental

conditioning system for spaces, for operationy the indoor environmental

conditioning inputs are actuated one at a time, and operably linked

thereafter with the indoor environmental conditioning sensor which

reports a change in environmental conditioning status.’’

5924486, Environmental condition control and

energy management system and method

8.5 51 1999, Tecom, Inc. ‘‘An indoor environmental condition control and energy management

system includes a plurality of inputsy a processor, coupled to the inputs,

computes an environmental condition deadband range for multiple

energy unit price points based on the user input parameters and controls

at least one energy-consuming load devicey’’

5544036, Energy management and home

automation system

8.4 59 1996, Brown, Jr., Robert J.;

Romanowiz, James D.; Staples,

Charles W.

‘‘yincludes one or more controllers in each facility being managed and

one or more energy consuming devices attached to each controller. Each

controller responds to digital paging signals from a central command

centery’’

6216956, Environmental condition control and

energy management system and method

7.1 32 2001, Tocom, Inc. ‘‘An indoor environmental condition control and energy management

system includes a plurality of inputs. A user input receives user input

parameters including a desired indoor environmental condition range for

at least one energy unit price pointy’’

4206872, Electronic thermostat 6.8 68 1980, Levine Michael R ‘‘A thermostat for generating control signals for a furnace, air conditioner,

or the likey the stable output is also provided to dividing circuits to

generate a signal representative of the time that is used to interrogate a

digital memory storing a desired temperature-time program for the

thermostat for a repetitive period, such as a week or month.’’

4217646, Automatic control system for a

building

6.8 68 1980, The Singer Company ‘‘An automatic system for controlling the environmental conditions in a

building divided into a plurality of zones includes a control station that

sends control signals over the power lines of the building to control

power consuming appliances in the various zones. The control station

includes a programmable computery’’

4173754, Distributed control system 6.8 54 1979, General Electric

Company

‘‘A redundant digital address and function code is generated at a central

location and transmitted through a signal circuity at each addressed

receiver location, the encoded function is performed to control selected

loads in a predetermined manner.’’

5390206, Wireless communication system for

air distribution system

6.7 47 1995, American Standard Inc. ‘‘The present invention is directed to an air distribution system for an air

conditioning system, and more particularly, to a wireless communication

system between the air distribution controllers and the zone temperature

sensors in the zone to be controlled.’’

6487457, Database for a remotely accessible

building information system

6.4 16 2002, Honeywell International,

Inc.

‘‘A building information system stores and processes both building

management system (BMS) data and building-related configuration

information, such as building and equipment descriptions and locations,

point and meter descriptions, contact information for alarms, utility rates,

and manufacturer and vendor information.’’

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831828

Author's personal copy

R&D. Anecdotally, the background sections of many EMCS and PTpatents reviewed for this research mentioned the need to reduceexposure to volatile energy prices. These years also saw thedevelopment of the first state and local energy codes in many U.S.locations. Several studies have investigated the development andimpact of these mandatory minimum energy codes for buildings(Aroonruengsawat et al., 2012; Deason and Hobbs, 2011; Jacobsenand Kotchen, 2010). We can see from the results of this analysis thatpatent activity in this technology increased in the same timeframe.Demand from the market, driven by energy codes or increasedawareness of energy issues, may have also created increased incen-tives for inventive activity during this period.

The literature offers some clues as to why EMCS has more recentlybecome the dominant driver of patenting, rather than PT. Severalstudies discuss the use of internet protocol in commercial buildingcontrols (Brambley et al., 2000; Budiardjo, 2010; Dounis andCaraiscos, 2009). This trend is likely captured in the growing knowl-edge flow from Computer and Communication technology seen in the

patent analysis (Table 6). However, the increase in EMCS relative toPT is not explained by the general increase of Computer patents overtime since the shares of total patents for computers and EMCS & PTare uncorrelated. Further, computer technology is inherent in bothEMCS and PT. Other market influences during this timeframe mayalso be contributing factors. These include the development ofstandard communication protocols for EMCS devices, such as BACnet,changes in government procurement from pneumatic to digitalcontrols, and growing use of labeling schemes like Energy Star andLEED. Kok et al. (2012) examine the spread of energy efficientbuilding practices in commercial properties using Energy Star andLEED certification as a measurement tool. They analyze 48 metropo-litan areas in the U.S. from 1995 to 2010, and determine that thespread has been relatively rapid. This growing use of building labelingcould also be driving innovation in energy end-use technology. Therelatively recent rise in patent activity specifically for commercialbuilding EMCS technologies does occur over the same timeframe.Identification of the effects of each of these possible factors oninnovation is an important issue in our current research.

The combination of the very strong flow of knowledge from theinformation and communication technology (ICT) sector to EMCS& PT with the relative increase in that flow over time suggestssome special considerations for policy makers, especially in termsof demand pull and technology push incentives (Nemet, 2009).From a demand pull perspective, it may be important to realizethat the incentives of interest for innovation in these technologiesare those that affect new types of entities that have not tradition-ally been central to innovation in energy technologies. We mayneed to be concerned with how ICT companies, rather thantraditional energy equipment suppliers, respond to building codes,energy prices, pollution pricing and other incentives. We may alsoneed to consider how these types of companies have peculiaritiesin how they address risk, uncertainty, and market failures. From atechnology push perspective, it seems clear that technologies suchas these may require support for R&D funding that goes wellbeyond traditional energy technologies and into information andcommunications. Funding decisions need to consider the extent towhich knowledge spillovers exist in this area since they may bedifferent from traditional energy technologies. It may also beimportant to examine whether the technological distance betweenICT and buildings requires any additional support for cross-cuttingresearch and boundary spanning firms.

7. Conclusion

Invention and innovation, along with adoption, are importantparts of the process of technological change. Understanding thedrivers of this process will allow policy-makers to design andselect solutions that accelerate energy efficiency in the buildingsector. In this research, we selected building controls as a casestudy to investigate inventive activity in energy end-use technol-ogies for buildings. Using energy management control systems(EMCS) and programmable thermostats (PT) to represent both thecommercial and residential building sectors, we identify patentsgranted in the U.S. for these technologies and analyze trends overthe last four decades. We find that patent activity for both EMCSand PT increased during the mid-1970s and peaked around 1980.In addition, commercial building (EMCS) patents exhibit a notableincrease relative to residential building (PT) beginning in the mid-1990s. Using patent citation data, we identify important patentsin these technologies, and find that they show a growing use ofexternal knowledge from the Computer and Communicationindustry domain.

Possible explanations for these trends include energy pricevolatility, implementation of building codes, growing use of

Table 11Top 20 assignees among 1505 EMCS and PT patents, by number of patents

granted.

Assignee Number ofEMCS &PT patents

‘‘pdpass’’ variablenumber

Honeywell Inc. 97 10844670

Honeywell International Inc. 78 12346559

Carrier Corporation 68 10220170

Emerson Electric Co. 50 10105804

Siemens Building Technologies, Inc. 35 11987050

LG Electronics Inc. 34 11727564

Robertshaw Controls Company 30 10107775

Toshiba Corporation 25 10040685

Daikin Industries, Ltd. 24 10146530

Johnson Controls Technology

Company

24 12041371

Sanyo Electric Co., Ltd. 21 10480018

Hitachi, Ltd. 19 10180289

American Standard Inc. 18 10493524

Johnson Service Company 17 11087528

Mitsubishi Electric Corporation 17 10067050

International Business Machines

Corporation

13 10083419

Samsung Electronics Co., Ltd. 13 10842508

General Electric Company 11 10059719

Danfoss A/S 9 10442256

Levine, Michael R 9 NA

Rosen, Howard 9 NA

Fig. 3. Highly-cited EMCS & PT patents granted 1976–2006 (n¼396).

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831 829

Author's personal copy

building labeling in the U.S., and the development of newcapabilities in the computer industry. This patent data setprovides a basis for further investigation into the effects of thesefactors. Even these descriptive results make clear that – at least inthis set of building technologies – policy makers need to considerthe effects on incentives for new types of actors that would nottraditionally be considered part of the building or energy sectors.

Appendix A

Table 9 lists the names for each patent class/subclass in whichwe searched. The following Boolean text search string was used togenerate the initial group of 2605 patents:

ccl/236/1R or ccl/236/1B or ccl/236/1C or ccl/236/1E orccl/236/9R or ccl/236/9A or ccl/236/44R or ccl/236/44C orccl/236/44A or ccl/236/46R or ccl/236/46A or ccl/236/46Cor ccl/236/46F or ccl/236/47 or ccl/236/51 or ccl/700/276or ccl/700/277 or ccl/700/278 or BACnet or aclm/‘‘bhuilding automation’’ andnot ttl/vehic$ andnot ttl/automo$ andnot ttl/refrig$

Patents were marked as ‘‘Not Relevant’’ if any of the followingcriteria were true:

� Title had ‘‘car’’ in it, or was granted to an automotive company;� Title indicated it was related to heating/cooling/humidity

control of non-building items or unoccupied spaces, such asovens, industrial chambers, aircraft, semiconductors, animalenclosures, crawl spaces, attics, etc.;� Title indicated that pneumatic control was involved, except

when digital was also indicated;� Patent had no title at all;� Title related specifically to indoor air quality testing or

measurement (not related to energy);� Title was focused on fire or smoke detection;� Title related only to water flow or consumption. However,

water heating patents were included, as these have thepotential to be used in comfort control applications;� Title was related only to building automation components that

did not affect energy management, such as building accesscontrols or security.

All patents granted before 1970 were also rejected as ‘‘notrelevant’’ because they were not digital or programmable technol-ogies, even if they dealt with thermostats or controls. These wereoften prior art, such as pneumatic controls, or improvements incomfort or cost instead of energy management.

Fig. 4 shows the distribution of the 6968 citations receivedamong the 396 highly-cited EMCS & PT patents. As in previousstudies, the distribution is strongly positive skewed (Hall et al.2001); two patents received over 100 citations, and 12 receivedmore than 50. The bimodal distribution is an artifact of thedefinition that highly-cited Z75th percentile. A large number ofpatents were granted in the last three years of the data set inwhich the 75th percentile was at or close to zero. For example, apatent issued in 2005 with 1 citation is considered highly-cited.

Table 10 lists the top 15 highly-cited EMCS & PT patents, alongwith details on each patent. The top 15 are selected by dividing thetotal citations received by the median citations received by all patentsin the same grant year (A). We also show raw counts of citations (B).

Table 11 lists the top 20 assignees for 1505 relevant EMCS & PTpatents, by number of patents granted. The table includes the‘‘pdpass’’ variable used to standardize the assignees in the NBERdata set. Patents assigned to individuals do not have ‘‘pdpass’’numbers in the data, as noted in the table.

Appendix B. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.enpol.2012.10.050.

References

Allcott, H., Greenstone, M., 2012. Is there an energy efficiency gap? The Journal ofEconomic Perspectives 26, 3 -3.

American Society of Heating, Refrigerating, and Air-Conditioning Engineers, 2007.ASHRAE handbook: Heating, Ventilating, and Air-Conditioning 0041pplica-tions (Inch-Pound Ed.). ASHRAE, Atlanta, Georgia.

American Society of Heating, Refrigerating, and Air-Conditioning Engineers, 1991.ASHRAE Terminology of Heating, Ventilation, Air-Conditioning, and Refrigera-tion, second ed. ASHRAE, Atlanta, Georgia.

Andrews, C.J., Krogmann, U., 2009. Explaining the adoption of energy-efficienttechnologies in U.S. commercial buildings. Energy and Buildings 41 (3),287–294.

Aroonruengsawat, A., Auffhammer, M., Sanstad, A., 2012. The impact of state levelbuilding codes on residential electricity consumption. The Energy Journal 33,1.

Brambley, M.R., Chassin, D.P., Gowri, K., 2000. DDC and the web. ASHRAE Journal42 (12), 38–50.

Brown, M.A., 2001. Market failures and barriers as a basis for clean energy policies.Energy Policy 29 (14), 1197–1207.

Budiardjo, A., 2010. Returning to the roots of building automation. Heating/Piping/Air Conditioning Engineering 82 (4), 36–39.

Byrne, J., Hughes, K., Rickerson, W., Kurdgelashvili, L., 2007. American policyconflict in the greenhouse: divergent trends in federal, regional, state, andlocal green energy and climate change policy. Energy Policy 35 (9),4555–4573.

Carley, S., 2011. The era of state energy policy innovation: a review of policyinstruments. Review of Policy Research 28 (3), 265–294.

Christodoulou, S., Griffis, F.H., Barrett, L.P., Okungbowa, M., 2003. In pursuit ofcatalysts for new technology—a case for qualifications-based selection (QBS)of professional A/E services. Architectural Engineering 2003—Building Inte-gration Solutions.

Clarke, J.A., Johnstone, C.M., Kelly, N.J., Strachan, P.A., Tuohy, P., 2008. The role ofbuilt environment energy efficiency in a sustainable UK energy economy.Energy Policy 36 (12), 4605–4609.

Daim, T.U., Rueda, G., Martin, H., Gerdsri, P., 2006. Forecasting emerging technol-ogies: use of bibliometrics and patent analysis. Technological Forecasting andSocial Change 73 (8), 981–1012.

Darley, J.M., 1978. Energy conservation techniques as innovations, and theirdiffusion. Energy and Buildings 1 (3), 339–343.

Deason, J., & Hobbs, A., 2011. Codes to Cleaner Buildings: Effectiveness of U.S.Building Energy Codes. Climate Policy Initiative: Available from /http://climatepolicyinitiative.org/wp-content/uploads/2011/12/Codes-to-Cleaner-Buildings-Effectiveness-of-US-Building-Energy-Codes.pdfS.Fig. 4. Number of citations received by highly-cited EMCS & PT Patents (n¼396).

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831830

Author's personal copy

Diakaki, C., Grigoroudis, E., Kolokotsa, D., 2008. Towards a multi-objectiveoptimization approach for improving energy efficiency in buildings. Energyand Buildings 40 (9), 1747–1754.

Dieperink, C., Brand, I., Vermeulen, W., 2004. Diffusion of energy-saving innova-tions in industry and the built environment: Dutch studies as inputs for amore integrated analytical framework. Energy Policy 32 (6), 773–784.

Dixon, R.K., McGowan, E., Onysko, G., Scheer, R.M., 2010. U.S. energy conservation andefficiency policies: challenges and opportunities. Energy Policy 38 (11), 6398.

Dounis, A.I., Caraiscos, C., 2009. Advanced control systems engineering for energyand comfort management in a building environment—a review. Renewableand Sustainable Energy Reviews 13 (6-7), 1246–1261.

Energy Information Administration, 2011. International energy outlook 2011.DOE/EIA-0484(2011). U.S. Energy Information Administration.

Energy Information Administration, 2012. U.S. Energy Information Administrationglossary. Available from /http://www.eia.gov/tools/glossary/index.cfmS.

Gallagher, K.S., Grubler, A., Kuhl, L., Nemet, G., Wilson, C., 2012. The energytechnology innovation system. Annual Review of Environment and Resources37, 137–162.

Gann, D.M., Salter, A.J., 2000. Innovation in project-based, service-enhanced firms:the construction of complex products and systems. Research Policy 29 (7–8),955–972.

GEA, 2012. Global Energy Assessment—Toward a Sustainable Future. CambridgeUniversity Press, Cambridge, UK and New York, NY, USA, and the InternationalInstitute for Applied Systems Analysis, Laxenburg, Austria.

Geller, H., Harrington, P., Rosenfeld, A.H., Tanishima, S., Unander, F., 2006. Policiesfor increasing energy efficiency: thirty years of experience in OECD countries.Energy Policy 34 (5), 556–573.

Gillingham, K., Newell, R.G., & Palmer, K., 2009. Energy efficiency economics andpolicy. National Bureau of Economic Research, NBER Working Papers: 15031.

Greene, D.L., 2011. Uncertainty, loss aversion, and markets for energy efficiency.Energy Economics 33, 608–616.

Griliches, Z., 1990. Patent statistics as economic indicators: a survey. Journal ofEconomic Literature 28 (4), 1661–1707.

Grubler, A., Aguayo, F., Gallagher, K., Hekkert, M., Jiang, K., Mytelka, L., et al., 2012.Chapter 24—Policies for the Energy Technology Innovation System (ETIS).Cambridge University Press, Cambridge, UK and New York, NY, USA, and theInternational Institute for Applied Systems Analysis, Laxenburg, Austria,pp. 1665-1744.

Hall, B., Jaffe, A., & Trajtenberg, M., 2001. The NBER patent citations data file:Lessons, insights and methodological tools. National Bureau of EconomicResearch, NBER Working Papers: 8498.

Hall, B., Jaffe, A., Trajtenberg, M., 2005. Market value and patent citations.The Rand Journal of Economics 36, 16 -16.

Harhoff, D., Narin, F., Scherer, F.M., Vopel, K., 1999. Citation frequency and thevalue of patented inventions. The Review of Economics and Statistics 81,511–515.

Harty, C., 2005. Innovation in construction: a sociology of technology approach.Building Research & Information 33 (6), 512–522.

International Energy Agency, 2008. In support of the G8 plan of action: energyefficiency policy recommendations. OECD/International Energy Agency, Avail-able from.

International Energy Agency, 2011a. G20 Clean Energy, and Energy EfficiencyDeployment and Policy Progress. OECD/International Energy Agency. Availablefrom /http://www.iea.org/publications/freepublications/publication/G20_paper.pdfS.

International Energy Agency, 2011b. Technology Roadmap: Energy-Efficient Build-ings: Heating and Cooling Equipment. OECD/International Energy Agency.Available from /http://www.iea.org/publications/freepublications/publica-tion/buildings_roadmap.pdfS.

Jackson, J., 2010. Promoting energy efficiency investments with risk managementdecision tools. Energy Policy 38 (8), 3865–3873.

Jacobsen, G.D., & Kotchen, M.J., 2010. Are building codes effective at savingenergy? Evidence from Residential Billing Data in Florida. National Bureau ofEconomic Research, NBER Working Papers: 16194.

Jaffe, A.B., Newell, R.G., Stavins, R.N., 2005. A tale of two market failures:technology and environmental policy. Ecological Economics 54 (2-3), 164 -163.

Johnstone, N., Hascic, I., Popp, D., 2010. Renewable energy policies and technolo-gical innovation: evidence based on patent counts. Environmental & ResourceEconomics 45 (1), 133–155.

Kale, S., Arditi, D., 2010. Innovation diffusion modeling in the construction industry.Journal of Construction Engineering & Management 136 (3), 329–340.

Kiss, B., Neij, L., 2011. The importance of learning when supporting emergenttechnologies for energy efficiency—a case study on policy intervention forlearning for the development of energy efficient windows in Sweden. EnergyPolicy 39 (10), 6514–6524.

Kok, N., McGraw, M., Quigley, J.M., 2012. The diffusion over time and space ofenergy efficiency in building. Annals of Regional Science 48 (2), 541–564.

Kyle, P., Clarke, L., Smith, S.J., Kim, S., Nathan, M., Wise, M., 2011. The value ofadvanced end-use energy technologies in meeting U.S. climate policy goals.The Energy Journal 32, 61–88. (Special Issue. Strategies for Mitigating ClimateChange Through Energy Efficiency: A Multi-Model Perspective.).

Lee, W.L., Yik, F.W.H., 2004. Regulatory and voluntary approaches for enhancingbuilding energy efficiency. Progress in Energy and Combustion Science 30 (5),477–499.

Levine, M., Urge-Vorsatz, D., Blok, K., Geng, L., Harvey, D., Lang, S., et al., 2007.Residential and Commercial Buildings. In Climate Change 2007: Mitigation.Contribution of Working Group III to the Fourth Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge University Press,Cambridge, United Kingdom and New York, NY, USA.

Lim, J.N., Schultmann, F., Ofori, G., 2010. Tailoring competitive advantages derivedfrom innovation to the needs of construction firms. Journal of ConstructionEngineering & Management 136 (5), 568–580.

Lowry, G., 2002. Technical note: factors affecting the success of building manage-ment system installations. Building Services Engineering Research & Technol-ogy 23 (1), 57.

Lund, P.D., 2007. Effectiveness of policy measures in transforming the energysystem. Energy Policy 35 (1), 627–639.

Meier, A., Aragon, C., Hurwitz, B., Peffer, T., & Pritoni, M., 2010. How PeopleActually use Thermostats. Washington, D.C.: American Council for an Energy-Efficient Economy.

Meier, A., Aragon, C., Peffer, T., Perry, D., Pritoni, M., 2011. Making energy savingseasier: usability metrics for thermostats. Journal of Usability Studies 6 (4),226–244.

Misuriello, H., Penney, S., Eldridge, M., & Foster, B., 2010. Lessons Learned fromBuilding Energy Code Compliance and Enforcement Evaluation Studies.Washington, D.C.: American Council for an Energy-Efficient Economy.

Mundaca, L., Neij, L., Worrell, E., McNeil, M., 2010. Evaluating energy efficiencypolicies with energy-economy models. Annual Review of Environment andResources 35, 305–344.

National Institute of Building Sciences, 2012. Whole Building Design Guide.Available from /www.wbdg.orgS.

Nemet, G.F., 2009. Demand-pull, technology-push, and government-led incentivesfor non-incremental technical change. Research Policy 38 (5), 700–709.

Nemet, G.F., 2012. Inter-technology knowledge spillovers for energy technologies.Energy Economics 34, 1259–1270.

Nemet, G.F., Johnson, E., 2012. Do important inventions benefit from knowledgeoriginating in other technological domains? Research Policy 41 (1), 190–200.

Nemet, G.F., Kammen, D.M., 2007. U.S. energy research and development: declin-ing investment, increasing need, and the feasibility of expansion. Energy Policy35 (1), 746.

Noailly, J., 2011. Improving the energy efficiency of buildings: the impact ofenvironmental policy on technological innovation. Energy Economics.

Noailly, J., Batrakova, S., 2010. Stimulating energy-efficient innovations in theDutch building sector: empirical evidence available from patent counts andpolicy lessons. Energy Policy 38 (12), 7803–7817.

Pacala, S., Socolow, R., 2004. Stabilization wedges: solving the climate problem forthe next 50 years with current technologies. Science 305 (5686), 968.

Popp, D., 2005a. Lessons from patents: using patents to measure technological changein environmental models. Ecological Economics 54 (2-3), 209–226.

Popp, D., 2005b. They don’t invent them like they used to: an examination ofenergy patent citations over time. National Bureau of Economic Research,NBER Working Papers: 11415.

Ravetz, J., 2008. State of the stock—what do we know about existing buildings andtheir future prospects? Energy Policy 36 (12), 4462–4470.

Ryghaug, M., Sorensen, K.H., 2009. How energy efficiency fails in the buildingindustry. Energy Policy 37 (3), 984–991.

Salter, A., Gann, D., 2003. Sources of ideas for innovation in engineering design.Research Policy 32 (8), 1309.

Sexton, M., Barrett, P., 2003. A literature synthesis of innovation in smallconstruction firms: insights, ambiguities and questions. Construction Manage-

ment & Economics 21 (6), 613–622.Tombesi, P., 2006. Good thinking and poor value: on the socialization of knowl-

edge in construction. Building Research & Information 34 (3), 272–286.Trajtenberg, M., 1990. A penny for your quotes: patent citations and the value of

innovations. RAND Journal of Economics 21 (1), 172–187.U.S. Environmental Protection Agency. Energy Star Programmable Thermostat

MOU—Version 1.1. United States Environmental Protection Agency. Availablefrom /http://www.energystar.gov/ia/partners/product_specs/program_reqs/archive/Programmable%20Thermostat%20MOU%20V1.1.pdfS.

United Nations Environment Programme & CEU, 2007. Assessment of PolicyInstruments for Reducing Greenhouse Gas Emissions from Buildings:Summary and Recommendations. Available from /http://www.unep.org/pdf/policytoolbox.pdfS.

Urge-Vorsatz, D., Koeppel, S., Mirasgedis, S., 2007. Appraisal of policy instrumentsfor reducing buildings’ CO2 emissions. Building Research and Information35 (4), 458–477.

Urge-Vorsatz, D., Metz, B., 2009. Energy efficiency: how far does it get us incontrolling climate change? Energy Efficiency 2 (2), 87–94.

van Bueren, E., De Jong, J., 2007. Establishing sustainability: policy successes andfailures. Building Research and Information 35 (5), 543–556.

van Bueren, E.M., Priemus, H., 2002. Institutional barriers to sustainable construc-tion. Environment and Planning B: Planning and Design 29 (1), 75–86.

Wilson, C., Grubler, A., Gallagher, K., & Nemet, G.F., 2012. Marginalization ofenergy end-use technologies in innovations for climate protection. NatureClimate Change 2 (11), 780–788.

J.E. Altwies, G.F. Nemet / Energy Policy 52 (2013) 819–831 831