19
This article was downloaded by: [Umeå University Library] On: 20 September 2013, At: 20:26 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Economics of Innovation and New Technology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gein20 Does enterprise software matter for service innovation? Standardization versus customization Benjamin Engelstätter a & Miruna Sarbu a a ZEW, ICT Research Group , Centre for European Economic Research , Mannheim , Germany Published online: 05 Feb 2013. To cite this article: Benjamin Engelstätter & Miruna Sarbu (2013) Does enterprise software matter for service innovation? Standardization versus customization, Economics of Innovation and New Technology, 22:4, 412-429, DOI: 10.1080/10438599.2012.759705 To link to this article: http://dx.doi.org/10.1080/10438599.2012.759705 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions

Does enterprise software matter for service innovation? Standardization versus customization

  • Upload
    miruna

  • View
    215

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Does enterprise software matter for service innovation? Standardization versus customization

This article was downloaded by: [Umeå University Library]On: 20 September 2013, At: 20:26Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Economics of Innovation and NewTechnologyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/gein20

Does enterprise software matter forservice innovation? Standardizationversus customizationBenjamin Engelstätter a & Miruna Sarbu aa ZEW, ICT Research Group , Centre for European EconomicResearch , Mannheim , GermanyPublished online: 05 Feb 2013.

To cite this article: Benjamin Engelstätter & Miruna Sarbu (2013) Does enterprise software matterfor service innovation? Standardization versus customization, Economics of Innovation and NewTechnology, 22:4, 412-429, DOI: 10.1080/10438599.2012.759705

To link to this article: http://dx.doi.org/10.1080/10438599.2012.759705

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Page 2: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology, 2013Vol. 22, No. 4, 412–429, http://dx.doi.org/10.1080/10438599.2012.759705

Does enterprise software matter for service innovation?Standardization versus customization

Benjamin Engelstätter* and Miruna Sarbu

ZEW, ICT Research Group, Centre for European Economic Research, Mannheim, Germany

(Received 13 July 2012; final version received 12 December 2012 )

This paper analyzes the relationship between service innovation and different types ofenterprise software systems, i.e. standardized enterprise software designed to fit onecertain business sector and enterprise software specifically customized for a single firm.Using recent firm-level data of a survey among information and communication tech-nology service providers as well as knowledge-intensive service providers in Germany,this is the first paper which empirically analyzes whether the use of sector specific orcustomized enterprise software triggers innovation. The results based on a knowledgeproduction function suggest that customized enterprise software is related to the occur-rence of service innovation. However, there is no relationship between sector specificenterprise software and innovation activity. The results stay robust to several differentspecifications and suggest that the causality runs from customized software usage toservice innovation.Keywords: enterprise systems; service innovation; customized enterprise software;sector specific enterprise software; service sector

JEL Classification: L10; M20; O31

1. IntroductionA large range of enterprise software products nowadays supports day-to-day business oper-ations, controls manufacturing plants, schedules services or facilitates decision-making.The purpose of these software applications is, in general, to automate operations reach-ing from supply management, inventory control or sales force automation to almost anyother data-oriented management process. Especially in many fields like semiconductors,biotechnology, information and telecommunication or other knowledge-intensive industrybranches, employees might be able to observe, measure or even envision certain phenomenaonly by using specific enterprise software applications. SAP, the largest global enterprisesoftware vendor, estimates the addressable market for enterprise software applications tobe roughly $110 billion in 2010.1

Overall, enterprise software can be categorized into three types, as users distinguishbetween generic applications such as an enterprise resource planning system purchasedin standardized form from the vendor,2 software systems or special modules specificallydesigned to fit one business sector3 or completely customized enterprise software packages

*Corresponding author. Email: [email protected]

© 2013 Taylor & Francis

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 3: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 413

developed for a single firm in particular and adopted to its specific needs.4 Thus, customizedenterprise software systems are usually unique.

Although the usage of information and communication technology (ICT) applicationsin general is suspected to enhance firms’ innovative activity (Hempell and Zwick 2008;Brynjolfsson and Saunders 2010; Engelstätter 2012), the potential impact of sector spe-cific or completely customized enterprise software on innovation performance in particularis still not investigated. The literature in this field is scarce, offering only a few stud-ies which examine the benefits of enterprise software for innovation activity (Shang andSeddon 2002). Empirical evidence is even scarcer, provided only by Engelstätter (2012)who outlines the impact of three common generic enterprise systems on firms’ ability torealize process and product innovations. However, being more general based on the avail-able sample, Engelstätter (2012) does not picture the impacts of enterprise software onspecific service innovations. Those service innovations might be driven by other character-istics such as enhanced knowledge handling or contact to customers compared with mainlyresearch driven innovations in manufacturing. However, to succeed in realizing innovationsin dynamic, information-rich environments like the service sector, a firm should engage ina combination of different practices like promoting information absorption and diffusion,knowledge handling or development of an extended intra- and inter-organizational network(Mendelson and Pillai 1999). Engaging in these practices, however, is expected to be facil-itated with advanced knowledge handling and storing capabilities of utilized enterprisesoftware. In light of this research gap, the current study provides the first empirical evi-dence of the impact of business sector specific or customized enterprise software on firms’innovation performance. We focus on the specific case of service firms in our analysis asbusiness sector specific software packages are incomparable between manufacturing andservice firms.

Our study relies on a unique database consisting of 336 German firms from ICT andknowledge-intensive service sectors. As analytical framework, we employ a knowledgeproduction function for our empirical analysis. Based on a probit approach, the resultsindicate that service firms relying on customized enterprise software have a higher proba-bility of realizing service innovations compared with firms not using customized softwarepackages. Concerning sector specific enterprise software, we find no evidence of a positiveimpact on service firms’ innovative activity. These results stay robust to many specificationscontrolling for different sample sizes and several sources of firm heterogeneity like size,age, workforce structure, competitive situation or former innovative experience.

The paper proceeds as follows: Section 2 presents the economic framework of ICT,innovation and service innovation our study is based upon. Section 3 explains in detailthe relationship of enterprise software and service innovation and its potential drawbacks.Section 4 introduces the dataset. The estimation procedure is derived in Section 5 andSection 6 presents the estimation results. Section 7 concludes.

2. Information and communication technology, innovation and service innovationThis study relates to the literature exploring the impacts on innovation activity of ICT ingeneral and of enterprise software on service innovation in particular. ICT is expected toenable productivity and performance gains by supporting the optimization of firms’ busi-ness processes (Brynjolfsson and Hitt 2000). A crucial prerequisite for these productivitygains, however, is the firms’ innovation activity (Hall, Lotti, and Mairesse, 2009). Never-theless, the link between ICT and innovation is not as extensively studied in the empiricalliterature as the relationship between ICT and productivity. Studies investigating the effects

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 4: Does enterprise software matter for service innovation? Standardization versus customization

414 B. Engelstätter and M. Sarbu

of ICT investments on innovative performance at the firm-level usually report a positiveand significant impact which may emerge directly (Gera and Gu 2004) or indirectly viacomplementary assets as shown in Bresnahan et al. (2002) or Hempell and Zwick (2008).

The literature on enterprise software applications follows a similar pattern as the lit-erature on ICT. There are many studies extensively examining the performance impactsof enterprise software5 and nearly no analysis examining potential impacts on innovationactivity. Only Engelstätter (2012) empirically shows that different types of generic enter-prise software systems enhance the innovation activity of using firms, which results inmore realized process as well as product innovations. In light of this research gap, our studyexplicitly contributes to the literature on enterprise software as we offer evidence that notonly generic but also very specific enterprise software can be associated with increasedinnovative performance.

For a particular type of innovation, i.e. the innovation in services, however, the impactsof most recent ICT like enterprise software might be even more important. In general, serviceinnovations are argued to be either driven by the adoption of ICT or by the exchangeof information which in particular is facilitated using enterprise software. Nevertheless,empirical studies exploring the impact of ICT on service innovation are even scarcer. Thereason for this lack of empirical evidence may be due to the fact that the nature of servicescomplicates the use of traditional economic measurements for innovations as this field isvery heterogeneous due to features like intangibility, interactivity and coterminality (Galloujand Weinstein 1997). In detail, services turn out to be intangible because they are hard tostore or transport, and can sometimes not even be displayed to a customer in advance (Hippand Grupp 2005). Interactivity is constituted in high communication, and coordination needsbetween client and supplier as in most cases both have to be present for the transaction.Coterminality captures the fact that services are often produced and consumed at the sameplace and time. Due to this heterogeneity several streams6 of literature emerged, each oneproclaiming different reasons for innovative activity in services.

Several studies, e.g. Pilat (2001) and Agrawal and Berg (2008), adopt the so called tech-nologist approach for explaining service innovation in their analysis. This most prominentapproach argues that service innovations might be driven by the external adoption of ICT.However, services are often co-produced and co-made by the provider and the user work-ing together (Howells 2010), exchanging information or knowledge, emotions, verbal orgestural civilities, or the performance of repair or rectification tasks (Gallouj 2002). Expect-ing these exchanges as a driver for the service innovation, the service-oriented approachemerged which is also frequently adopted in the literature (Sundbo and Gallouj 2000; denHertog 2000; Djellal and Gallouj 2001).

In line with this approach, key elements for service innovation are, in particular, internalknowledge within the firm and its employees and the external network of the firm includingcustomers and other businesses (Sundbo 1997). Human capital, especially personal skillslike experience or extensive consumer contact, knowledge about markets, and consumerhabits and tastes are also important for realizing innovations in a service company (Meyer2010). In addition, sources of information like consumers and suppliers of equipment canprovide essential clues for service enhancement and advancement. However, assumingrecent ICT as a driver for service innovation is grounded in both approaches as modernICT like enterprise software not only enhances business process management and short-ens response times but also facilitates knowledge handling and management. We provideempirical evidence for this claim as our study empirically tests for the impact of most recententerprise software applications on service innovation. In the next section, we explain indepth how enterprise software is related to both approaches.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 5: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 415

3. Enterprise software and service innovationEnterprise software may impact the firms’ innovation activity through different channels:either directly on the particular benefits provided by a specific kind of enterprise softwareor indirectly as most software packages share common features facilitating knowledgehandling which might improve the innovation process. In line with the technologistapproach, the benefits of enterprise software which directly impact innovation performanceare the result of technical features varying between the types of software used. The featuresfacilitating knowledge management most enterprise software applications share supportthe service-oriented approach claiming that information and knowledge results in serviceinnovation. We will explain both channels of benefits in detail in the following. How-ever, scholars only explore innovation in general in the context of enterprise software anddo not investigate service innovation in particular. Our study fills this research gap byanalyzing the relationship between specific enterprise software applications and serviceinnovations.

The common features of most enterprise software packages enable enhanced informa-tion handling and processing, thereby facilitating communication, knowledge transfer andcontact maintenance between employees or consumers and partners. Thereby, its businessunits reach a more centralized network position as all necessary information reach themfaster which may result in the business becoming more innovative (Tsai 2001). Offeringand presenting information in an adequate manner and providing frequently updated real-time databases where they are needed, enterprise software might function as a ‘softer’source of innovation according to Tether (2005) and can be expected to improve innovativeperformance in service firms. Following the argument that acquired information drives ser-vice innovations Criscuolo, Haskel, and Slaughter (2005) argue that firms generate moreinnovations with established upstream/downstream contacts to suppliers and customers.This relation especially holds for service innovations as customers and suppliers are oftenproviders of essential guidelines and ideas for enhancement and advancement of providedservices as proclaimed in the service-oriented approach. Roper, Du, and Love (2006) evensupport this argument as they stress the high value of backwards and horizontal knowledgelinkages for innovations.

Exploring the benefits of enterprise software it stands out that some industry brancheslike biotechnology, semiconductors or architects need particular sector specific software likecomputer aided architecture or simulation software to complete their business tasks. Forthese industries so called sector specific enterprise software represents merely a technicalworking tool which could improve process handling and management but is not necessarilyrelated to improvements in innovation performance. But, Thomke (1998) shows for theautomotive industry that the use of this working tool in form of specific computer simula-tion software is associated with overall better R & D output. However, if a firm employsenterprise software, which is specifically designed for the own company, it can incorporatelong-term experience and knowledge into the software. Thereby, a firm is able to createsoftware which perfectly fulfills all requirements for its specific business. Such a softwareenables shortened response times and fast information delivery resulting in an increasedexternal focus for the utilizing firm which is hypothesized to increase the returns on infor-mation technology (Tambe, Hitt, and Brynjolfsson 2012). Tambe, Hitt, and Brynjolfsson(2012) also show that this external focus leads to overall improved innovation performancein combination with decentralization and the use of sophisticated information technologylike enterprise software solutions. The necessary decentralization, however, is associatedwith customized enterprise software usage (Gronau 2010). Malhorta et al. (2001) offer firstevidence of these positive impacts of customized enterprise software on firms’ innovative

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 6: Does enterprise software matter for service innovation? Standardization versus customization

416 B. Engelstätter and M. Sarbu

performance as they show that externally developed customized virtual teaming softwareyields crucial innovations at Boeing-Rocketdyne.

Besides all benefits, firms’ relying on enterprise software may also face some seriousdrawbacks. However, as there are no procedures similar to clinical trials to assess these draw-backs as well as the benefits of enterprise software objectively beforehand (Ghosh 2006),firms can only identify potential drawbacks due to experience when using the software.We expect some drawbacks of enterprise software to occur in the long-run. Independent ofthe type of enterprise software used the increased automation and streamlining of processesmight bind the workers to certain limits restricting the workers’ freedom and creativity. Also,enterprise software only tends to store and transfer information crucial for the specific pro-cess at hand and may miss vague or tacit knowledge necessary for further innovations in thelong-run. Thus, heavily relying on software may result into very homogeneous processes,practices and solutions hampering a firms’ future innovative potential.

4. Description of dataThe data we use in this study are taken from the quarterly business survey among the‘service providers of the information society’ conducted by the Centre for European Eco-nomic Research (ZEW) in cooperation with the credit rating agency Creditreform. Thesector ‘service providers of the information society’ comprises nine industry sectors, threeof information and communication service sectors and six knowledge-intensive servicesectors.7 Every quarter, a one-page questionnaire is sent to about 4.000 mostly small-and medium-sized companies. The sample is stratified with respect to firm size, region(East/West Germany) and sector affiliation. The survey achieves a response rate of about25% for each wave and builds a representative sample of the German service sector. Theinterviewed candidates may choose between responding via pen and paper, fax or online.The questionnaire consists of two parts. In the first part of the questionnaire, companiescomplete questions on their current business situation with respect to the previous quarteras well as their expectations for the next quarter. The second part is dedicated to questionsconcerning diffusion and use of ICT and further firm characteristics like innovative activi-ties or training behavior. The questions in the second part change every quarter but mightbe repeated annually or biyearly. Details on the survey design are presented in Vanberg(2003). Overall, the survey is constructed as a panel with different waves being employedin several different empirical setting as in Bertschek and Kaiser (2004) or Meyer (2010).

In the survey, the questions covering enterprise software usage were only included in thesecond quarter of 2007. The questions about innovative activities were asked in the secondquarter of 2009. Thus, a panel data analysis cannot be provided in this paper. Accordingly,we focus on a cross-section analysis by merging the second wave of the year 2007 to coverenterprise software usage and the second wave of the year 2009 to cover firms’ innovativeactivity, thereby forming a well-defined temporal sequence. Considering item non-responsefor enterprise software and innovation, a sample consisting of 335 firms remains.

According to the OECD (2005), we define service innovations in our analysis as acompletely new service or an essential improvement8 to an existing service that has beenintroduced between June 2008 and June 2009. Service innovation performance representingthe dependent variable in our empirical analysis is accordingly measured as a dummyvariable that takes the value one for firms realizing a service innovation and zero otherwise.

In the survey, the firms were asked about using two types of enterprise software, i.e. sec-tor specific software and customized software. The variables capturing the use of enterprisesoftware are dummy variables which take the value one if a firm uses the respective type of

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 7: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 417

Figure 1. Usage of enterprise software and service innovation.Note: ‘Usage’ refers to the share of firms using the indicated software. ‘Service innovation’ refers tothe share of firms which use the indicated software and reported having realized a service innovation.

enterprise software in June 2007 and zero otherwise. Figure 1 shows that more than threequarters of the service firms use sector specific software and 38% of the firms uses cus-tomized software. However, both software types are non-exclusive. Hence, some firms havealso customized as well as sector specific software systems running (27%, not reported).

Overall, 39% of the firms in our sample reported realizing a service innovation betweenJune 2008 and June 2009.9 For a first illustration, Figure 1 also pictures the share of firmswhich realized a service innovation and also use enterprise software. Concerning sectorspecific enterprise software the according share amounts to 37%. By contrast, more than halfof the firms using customized enterprise software realized service innovations in the coveredtime period. This relatively high share yields first descriptive evidence for our hypothesis thatthe use of customized enterprise software seems to foster the innovative activity in servicefirms. However, for sector specific enterprise software there is no descriptive evidence fora positive innovative impact.

For a further overview, Table 1 provides summary statistics for our employed sample.We describe the construction of each variable and its relationship to service innovationsin the next section in detail. However, it stands out that our representative sample of theGerman information service sector contains mostly small- and medium-sized enterprises(SMEs) with 38 employees at mean. Nevertheless, it seems like bigger firms are moreeager to employ customized enterprise software solutions as the mean size amounts to 52employees for firms using customized software (not reported). The appropriate mean in sizefor firms relying on sector specific software systems, in contrast, turns out to be smaller (36,not reported). Two different reasons might explain this issue. First, large firms generally tendto have the financial infrastructure to implement big and complicated customized softwaresolutions. Secondly, the firm structure and business processes in large service firms might,in comparison to the situation in small firms, reach such a high level of complicationand integration a simple generic enterprise software solution might be unable to handle.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 8: Does enterprise software matter for service innovation? Standardization versus customization

418 B. Engelstätter and M. Sarbu

Table 1. Summary statistics.

Variable Mean Minimum Maximum N

Service innovation 0.386 0 1 336Sector specific software 0.758 0 1 336Customized software 0.380 0 1 336Number of employees 38 1 449 334Log (number of employees) 2.718 0 6.107 334Firm age 20 2 108 310Log (firm age) 2.851 0.693 4.682 3100–5 competitors 0.243 0 1 3126–20 competitors 0.304 0 1 312More than 20 competitors 0.451 0 1 312Share of employees working with PC 0.786 0.01 1 324Share of highly qualified employees 0.435 0 1 316Share of medium qualified employees 0.159 0 1 303Share of low qualified employees 0.383 0 1 311Share of employees younger than 30 years 0.194 0 1 308Share of employees between 30 and 55 years 0.656 0 1 318Share of employees older than 55 years 0.160 0 1 307Former service innovation 0.414 0 1 258Former process innovation 0.437 0 1 265East Germany 0.405 0 1 335

Source: ZEW quarterly business survey among service providers of the information society, owncalculations.

Accordingly, large firms may rely on customized software systems suited for their specificneeds more frequently. As one might argue that those large firms could possibly drive theresults of our empirical analysis, we conduct appropriate robustness checks in Section 6.

5. Analytical framework and estimation procedureIntroduced by Griliches (1979), this study will be based on a knowledge production function,following the basic assumption that the output of the innovation process represents a resultof several inputs linked to research and ongoing knowledge accumulation, such as formerinnovative experience or human capital (Vinding 2006). Following Engelstätter (2012), weinclude enterprise software in the knowledge production function, providing first insightsinto the relationship between business sector specific or completely customized enterprisesoftware usage and the firm’s innovation activity. This yields the following innovationrelation:

SIi = f (ESi, Li, Ci, FAi, FSi,−1, FPi,−1, controls),

with SIi covering service innovation for firm i, ESi the enterprise software used by firmi, Li the labor input, Ci the competitive environment and FAi the age of the firm. Formerservice and process innovations (FSi,−1 and FPi,−1) as well as controls like industry sectorclassifications and region dummy are also included. The employed explanatory variablesand their temporal sequence are explained in detail below. The endogeneous variable we useas measure for innovation contains the information whether the firms are service innovatorsor not. As this dependent variable is a dummy variable and we assume a normally distributederror term, the widely established probit model as introduced in Greene (2003) is used forinference.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 9: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 419

The labor input Li consists of firm size, qualification structure of employees, age struc-ture of employees and IT-intensity. We control for firm size by the logarithm of the numberof employees. Larger firms tend to have more lines of activity and therefore more areas inwhich they can innovate. This is valid for both the manufacturing and the service sectors(Leiponen 2005; Meyer 2010). Firm size is reported for the year 2008.

We also consider the qualification structure of the workforce by creating three controlvariables: the share of highly qualified (university or university of applied science degree),medium qualified (degree in technical college or vocational qualification) and low quali-fied (other) employees. All shares are measured in June 2009. The share of low qualifiedemployees is taken as the reference category. In general, qualification pictures the suitableknow-how and human capital which are essential for starting and enhancing innovations.Without suitable know-how, neither is successfully possible (Meyer 2010). Therefore, weassume that the higher the qualification of employees, the higher the innovative activity.

We control for the age structure of the employees with three variables. The first onerepresents the share of employees younger than 30 years and builds our reference category.The second variable captures the share of employees between 30 and 55 years, whereasthe third variable encompasses the share of employees over 55 years. Overall, the agestructure of the employees is expected to drive the firms’ innovative behavior. Börsch-Supan, Düzgün, and Weiss (2006) point out that the process of aging leads to a cutbackof fluid intelligence which is needed for new solutions and fast processing of information.Based on this cutback, older workers might be less innovative. In addition, older workersmay resist innovations as their ‘human capital’ depreciates. However, older workers grewin crystalline intelligence (experience and general knowledge) which in turn could lead tomore innovations. Thus, the effect of the age structure of employees on innovative activityis an ambiguous issue. The age structure was measured in June 2009 in our survey.

Following Engelstätter (2012), we proxy the IT-intensity of firms by the share of employ-ees working with a computer in June 2007. Licht and Moch (1999) mention that IT canimprove the quality of existing services, in particular customer service, timeliness and con-venience. Moreover, the productive use of IT is closely linked to complementary innovations(Hempell 2005).

The effect of firm age on innovation activity is still an ambiguous topic subjected todiscussion. Koch and Strotmann (2006) mention that innovative output is higher in youngerfirms than in older ones. However, it is lowest in the middle-aged (18–20 years) firms andrises again with an age of over 25 years. On the one hand, firms could lose their adaptability tothe environment with an increasing age or, on the other hand, organizational aging increasesinnovativeness due to learning processes. Firm age is also measured in the year 2008 in oursample.

The competitive situation is another relevant driver of innovative activity. We createdthree dummy variables representing the number of main competitors in June 2009 accordingto the firms’ self-assessment. The first one includes 0–5 competitors, the second one 6–20competitors which is our reference category and the last one more than 20 competitors.The relationship between innovation and competition is supposed to look like an invertedU curve (Aghion et al. 2005). A monopolist has less incentives to innovate as he alreadyenjoys a high flow of profit. In a competitive situation, there are less incentives to innovateif there is no possibility to fully reap the returns of the innovation (Gilbert 2006).

There are several reasons for taking prior innovation into account in our analysis. Oneof them is that innovative experience plays an important role in explaining innovations assuccessful innovations in the past lead to a higher probability for innovative success in thefuture (Flaig and Stadler 1994; Peters 2009). Another reason is a potential endogeneity bias

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 10: Does enterprise software matter for service innovation? Standardization versus customization

420 B. Engelstätter and M. Sarbu

our result might face, as it is not clear whether enterprise software leads to innovations orif innovative firms apply enterprise software merely as a diffusion channel for innovations.Both enterprise software variables were collected in June 2007 whereas the innovationvariable was measured between June 2008 and June 2009. Accordingly, there is actually atime shift between the dependent and independent variable already forming a well-definedtemporal sequence. Nevertheless, it is still possible that firms strategically purchased theirenterprise systems in June 2007 or earlier simply for the diffusion of new innovated servicesstarting out in June 2008. This may result in an upward bias of our estimated enterprisesoftware coefficients. However, by controlling for prior innovative activity we capture theoverall innovativeness of a firm to some extent. If enterprise software shows no significantimpact on today’s innovation any more when controlling for prior innovation, we canexpect that the software was employed only because the firm is already innovative. Asignificant impact of the enterprise software in this case, however, would point towardsa causality running from the adoption of the software to the realization of new serviceinnovations as the employed software still has an impact on recent innovations with firms’overall innovativeness controlled for.10 We use two dummy variables to control for priorinnovations. The first one is prior service innovation that takes the value one if the firmrealized at least one new or essentially improved service between March 2006 and March2007. The second dummy variable is prior process innovation that takes the value oneif the firm implemented new or essentially improved technologies during the same timeperiod. We control for both types of prior innovations as service and process innovationsare dynamically interrelated. In addition, we use nine industry sector dummies to control forindustry-specific fixed effects.11 A dummy variable for East Germany accounts for potentialregional differences.

6. Results6.1. Main resultsTable 2 shows the average marginal effects of the probit estimation of Equation (1).12 Inthe first model specification, we estimate the raw effect of both enterprise software typeson service innovation. The results indicate that sector specific software has no impact onservice innovation. Firms using customized enterprise software instead seem more likelyto innovate than firms which do not use this type of enterprise software. Based on a highsignificance level the probability to innovate is about 24.2 percentage points higher for firmsusing customized enterprise software.

In the second specification, we include firm size, firm age and IT-intensity. The impactof sector specific and customized software on service innovation remains qualitativelyunchanged in this specification suggesting that firms using customized software still face aprobability of innovating that is 22.8 percentage points higher compared with firms not usingthis type of enterprise software. Furthermore, we observe that larger firms seem to have ahigher probability of innovating as the marginal effect is significant at the 5% level. Firm ageand IT-intensity appear to have no effect on service innovation. The insignificant impact ofIT-intensity suggests that the significantly positive impact of customized enterprise softwarepictures not only an overall positive IT-effect but also the real effect of this type of enterprisesoftware.

In the third specification, further variables capturing competitive situation, qualificationstructure and age structure of employees are added. The impacts of both enterprise softwaresystems do again not change compared with former specifications indicating that the prob-ability of realizing service innovations is higher for firms utilizing customized software.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 11: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 421

Table 2. Probit estimation results: average marginal effects.

Dependent variable: dummy for service innovation(1) (2) (3) (4)

Sector specific software −0.055 0.026 −0.011 −0.003(0.060) (0.064) (0.073) (0.081)

Customized software 0.242∗∗∗ 0.228∗∗∗ 0.264∗∗∗ 0.182∗∗(0.054) (0.060) (0.065) (0.075)

Logarithmic firm size 0.047∗∗ 0.026 0.020(0.020) (0.023) (0.025)

Logarithmic firm age −0.065 −0.150∗∗ −0.098(0.055) (0.063) (0.072)

IT-intensity 0.056 −0.065 0.063(0.110) (0.131) (0.155)

Competitors 0–5 −0.028 0.018(0.078) (0.089)

Competitors >20 −0.055 −0.040(0.071) (0.077)

Highly qualified employees 0.026 −0.002(0.124) (0.137)

Medium qualified employees 0.018 −0.128(0.165) (0.178)

Employees 30–55 years −0.468∗∗∗ −0.184(0.172) (0.194)

Employees >55 years −0.368∗ −0.189(0.221) (0.244)

Prior service innovation 0.259∗∗∗(0.080)

Prior process innovation −0.033(0.074)

Additional controls Industry Industry IndustrySector East Sector East Sector East

Observations 336 298 240 179Pseudo R2 0.046 0.103 0.147 0.206

Note: Reference categories: competitors 6–20, unqualified employees, employees <30 years.∗10% significance level.∗∗5% significance level.∗∗∗1% significance level.

Older firms seem less likely to innovate, based on an estimated marginal effect significant atthe 5% level. The age structure of the workforce reveals some interesting results. Firms witha higher share of employees between 30 and 55 years as well as employees over 55 yearsare less likely to innovate compared with firms with a higher share of younger employees.The impact of employees between 30 and 55 years is significant at 1% while the impact ofemployees over 55 years is only significant at 10%.

In the fourth specification, we include dummy variables measuring prior service andprocess innovations in our analysis. Based on a high significance level, the average marginaleffect suggests that the probability to innovate is larger for firms which have already realizedservice innovations in the past. The average marginal effect of customized software remainspositive and significant proposing that customized software could indeed lead to serviceinnovation instead of being employed simply because utilizing firms are already innovativeas argued in Section 5. However, the incorporation of former service innovation weakensthe impact of customized software by reducing its significance level from 1% to 5%. Incontrast to prior service innovations, prior process innovations seem to have no impact oncurrent service innovations. The impact of firm age, employees between 30 and 55 years

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 12: Does enterprise software matter for service innovation? Standardization versus customization

422 B. Engelstätter and M. Sarbu

and employees over 55 years turns insignificant once we include former innovations intothe estimation specification.

In summary, our results suggest that firms using customized enterprise software experi-ence a higher probability of innovating compared with firms without this type of enterprisesoftware or sector specific enterprise software. This result stays robust across all specifica-tions and supports our hypothesis that customized enterprise software applications tailoredto specific firms’ need helps to enhance service innovation activity. Additionally, our resultsalso indicate that sector specific enterprise software solutions seem to have no impact onservice firms’ innovative performance.

6.2. Robustness checksTo ensure the validity of our results obtained, we also conduct several robustness checks.13

First, as firms could adopt sector specific in conjunction with customized enterprise soft-ware, we also estimate the model with an interaction term of the two enterprise softwaresystems added. However, the interaction effect is not significant in all specifications anddoes not change the other results qualitatively.

The consideration of prior innovations reduces our sample to a very low size of 179observations. Due to the insufficient panel structure, we decide to estimate all specificationswith this reduced sample size as another robustness check to ensure that our results are notdriven by observation loss.14 As a further robustness check, we also estimate all specifica-tions without the industry sector and regional fixed effects. The results regarding the use ofsector specific and customized enterprise software do not change qualitatively in all theserobustness checks.

In order to further investigate the population of the reduced sample with 179 observa-tions, we analyze some descriptive statistics and the distribution of the industry sectors.Tables A4 and A5 in the appendix picture the corresponding results. The descriptive statis-tics in Table A4 shows that there are no big differences in the population of the reducedsample compared with the whole sample. The mean values are very similar for all variables.Especially, firm size and firm age deliver almost the same values. The most important vari-ables which lead to slightly different mean values are the share of firms realizing serviceinnovations, the share of firms using customized enterprise software and the share of firmshaving realized process innovations in the past. For service innovations, the mean valueof the reduced sample is about 0.36 compared with the mean value of the whole sampleof 0.38. The share of firms using customized enterprise software is 0.42 in the reducedsample and 0.38 in the whole sample. The mean value of prior process innovations is 0.42in the reduced sample in contrast to 0.44 in the whole sample. Nevertheless, these smalldifferences do not affect the estimation results with the reduced sample qualitatively. Thedistribution of industry sectors in the reduced sample (see Table A5) does not show any bigdifferences either compared with the distribution of the whole sample pictured in Table A1.

Our last robustness check covers the firm size in our sample. As it could be the casethat especially some big enterprises drive the results we decide to restrict our sample tothose enterprises with a number of employees at or below the mean in size for the completesample, i.e. 37 employees or less. As this robustness check reduces our sample to 262 SMEs,we estimate all specifications as before except the last one controlling for prior innovations.Adding prior innovations to this reduced sample results in a number of observations toosmall for the model to achieve convergence. The results generated from analyzing our SMEssample do not change qualitatively compared with the results obtained before. Accordingly,

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 13: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 423

we suspect that big enterprises do not drive our empirical results indicating that customizedenterprise software is useful for innovative activity in firms of any size class.

7. ConclusionThis paper analyzes the relationship between different types of enterprise software systemsand innovation in services. In the service sector, enterprise software is an essential tool forproviding services. Therefore, it may represent a crucial contribution to a firm’s innovationperformance. We analyze the innovative impact of two different kinds of enterprise software,i.e. business sector specific and completely customized enterprise software. In essence,sector specific enterprise software is off-the-shelf software designed and standardized forcertain industries, whereas customized software is designed and adopted to the needs ofa single firm thereby implying unique features. The analysis is based on a knowledgeproduction function constituted by an innovation equation with data of German firms inICT- and knowledge-intensive service sectors.

Our results suggest that ICT- and knowledge-intensive service firms using customizedenterprise software that fulfills their specific requirements realize positive impacts in inno-vative activity. Firms using business sector specific software on the other hand do notreceive any positive impacts. The results stay robust to several specifications and robust-ness checks proposing that not only big enterprises drive this positive impact but also SMEsmay realize gains in innovative performance. With customized enterprise software posi-tively impacting firms’ innovative performance, our results support the technologist as wellas the service-oriented approach of explaining service innovation as adopting recent ICT aswell as exchanging information paired with improved knowledge handling drives serviceinnovation.

However, it is important to mention here that customized enterprise software can onlysupport service innovation if it is developed and applied properly, if the firm has completeknowledge of its organizational structure and processes and is aware of the goals it wants toachieve by using customized enterprise software. These facts ensure an enterprise softwaresystem that is perfectly suitable for all business requirements. Only given these circum-stances, service firms are able to profit from the quick delivery of information, enhancedknowledge processing, the strengthened connections of information sources or reflection ofall needed business processes customized enterprise software is linked to. Another benefitthat arises for firms using customized enterprise software is the increased IT know-how,especially when developing the software themselves. This know-how is an essential toolfor innovation, which is especially useful to ICT-intensive service providers, as these firmscould generate benefits out of it given that software development, for instance, is a majortask in these industries.

By contrast, firms that use sector specific enterprise software cannot exploit thebenefits outlined. Although this type of enterprise software is very supportive by provid-ing frequently updated databases or presenting information in an adequate manner, theseadvantages by themselves seem, based on our results, insufficient to support service inno-vation. Accordingly, relying on off-the-shelf software applications seems to be no adequatestrategy when aiming for innovations.

However, the current study is not without its limitations, one of these being the estab-lishment of a causality relationship as already mentioned above. Restrained to the availabledata, we also have no information about other unobserved factors potentially influencing thesoftware adoption decision like management quality or implementation problems and costsof the enterprise systems installed. Accordingly, we have to use dummy proxy variables

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 14: Does enterprise software matter for service innovation? Standardization versus customization

424 B. Engelstätter and M. Sarbu

instead, possibly resulting in unobserved heterogeneity the estimation procedure cannotcompletely account for. Another aspect one should be aware of is the timing dimension. Asour analysis is based on a lagged cross-section, we are unable to picture long-term impactsof enterprise software on firms’ innovation activity. However, as we argue in Section 3,these long-term impacts may possibly be negative as relying on the software might generateprocesses and solutions to homogeneous to allow for worker creativity, thereby hamperinginnovation activity. Future availability of new data might offer the opportunity to controleven for the mentioned special cases of firm heterogeneity and allow to picture long-termimpacts on innovation activity. Exploring in detail the determinants driving the adoption ofdifferent types of enterprise software forms another valid aim for future research.

AcknowledgementsWe would like to thank Irene Bertschek, Daniel Cerquera, Francois Laisney, Pierre A. Mohnen, AshokKaul and Michael R. Ward and anonymous referees for helpful comments and suggestions. All errorsare our own.

Notes1. See SAP presentation, available at http://www.sap.com/about/investor/presentations/pdf/

WB_DB_London_8Sep2010.pdf, last visited 29 June 2012.2. E.g. SAP Business One or Oracle E-Business Suite.3. Examples for sector specific enterprise software contain computer aided design or manufacturing

programs, e.g. solutions offered from Sage.4. Examples here are completely designed software solutions like applications from firms as, for

instance, Supremistic or Jay Technologies Inc. In addition, firms could also augment genericsolutions like SAP packages with custom-made modules.

5. See, e.g. among others, Hunton, Lippincott, and Reck (2003), Wier, Hunton, and HassabElnaby(2007) or Hendricks, Singhal, and Stratman (2007) for analyzing the impacts of enterprise soft-ware on several return measures. Hitt, Wou, and Zhou (2002), Shin (2006), Wieder et al. (2006)are some examples for studies exploring the impacts of enterprise software on labor productivity,net profit margin and value added.

6. See Howells (2010) for a comprehensive overview of all streams.7. The industry sectors are listed in Table A1 in the appendix.8. Based on such improvements, firms might also realize productivity gains placing this study also

in the literature of IT and productivity. However, as these improvements are determined as serviceinnovations (OECD 2005), we refrain from the IT and productivity literature branch and focusonly on the innovative impacts of IT.

9. See the summary statistics pictured in Table 1.10. The endogeneity bias due to commonly unobserved shocks affecting our sample as well as unob-

served heterogeneity should also be reduced due to the large set of control variables we employ.However, without an available instrument we cannot exclude a potential bias completely.

11. We picture the share of the industry classifications represented in our sample in detail in Table A1in the appendix. The most prominent industry branches in our sample are tax consultancies (17%)and architecture (16%).

12. Sample averages of the changes in the quantities of interest evaluated for each observation.Table A2 in the appendix contains the coefficient estimates.

13. All tables of the regressions performed as robustness checks are available from the authors uponrequest.

14. Table A3 in the appendix pictures the results of this robustness check.

ReferencesAghion, P., N. Bloom, R. Blundell, R. Griffith, and P. Howitt. 2005. “Competition and Innovation:

An Inverted U-relationship.” Quarterly Journal of Economics 120 (2): 701–728.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 15: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 425

Agrawal, G. K., and D. Berg. 2008. “Role and Impact of ‘Technology’ in the Service DevelopmentProcess: A Research Study.” International Journal of Services Technology and Management 9(2): 103–121.

Bertschek, I., and U. Kaiser. 2004. “Productivity Effects of Organizational Change: MicroeconometricEvidence.” Management Science 50 (3): 394–404.

Börsch-Supan, A., I. Düzgün, and M. Weiss. 2006. Sinkende Produktivität alternder Belegschaften?Zum Stand der Forschung. In Länger leben, arbeiten und sich engagieren. Chancen wertschaf-fender Beschäftigung bis ins Alter, edited by J. U. Prager and A. Schleiter, 85–102. Gütersloh,Germany: Bertelsmann Stiftung.

Bresnahan, T., E. Brynjolfsson, and L. M. Hitt. 2002. “Information Technology, Workplace Organi-zation and the Demand for Skilled Labor: Firm-level Evidence.” Quarterly Journal of Economics117 (1): 339–376.

Brynjolfsson, E., and L. M. Hitt. 2000. “Beyond Computation: Information Technology, Organiza-tional Transformation and Business Performance.” The Journal of Economic Perspectives 14 (4):23–48.

Brynjolfsson, E., and A. S. Saunders. 2010. Wired for Innovation: How it is Reshaping the Economy.1st ed. Cambridge, MA: MIT Press.

Criscuolo, C., J. Haskel, and M. Slaughter. 2005. Why are Some firms More Innovative? KnowledgeInputs, Knowledge Stocks and the Role of Global Engagement. NBER Working Paper no. 11479,National Bureau of Economic Research, Cambridge, MA.

den Hertog, P. 2000. “Knowledge-intensive Business Sectors as Co-producers of Innovation.”International Journal of Innovation Management 4 (4): 491–528.

Djellal, F., and F. Gallouj. 2001. “Patterns of Innovation Organisation in Service Firms: Postal SurveyResults and Theoretical Models.” Science and Public Policy 28 (1): 57–67.

Engelstätter, B. 2012. “It’s not all about Performance Gains – Enterprise Software and Innovations.”Economics of Innovation and New Technology 21 (3): 223–245.

Flaig, G., and M. Stadler. 1994. “Success Breeds Success: The Dynamics of the Innovation Process.”Empirical Economics 19 (1): 55–68.

Gallouj, F. 2002. “Innovation in Services and the Attendant old and New Myths.” The Journal ofSocio-Economics 31 (2): 137–154.

Gallouj, F., and O. Weinstein. 1997. “Innovation in Services.” Research Policy 26 (4–5): 537–556.Gera, S., and W. Gu. 2004. “The Effect of Organizational Innovation and Information and

Communications Technologies on Firm Performance.” International Productivity Monitor 9:37–51.

Ghosh, R. A. 2006. Study on the: Economic Impact of Open Source Software on Innovationand the Competitiveness of the Information and Communication Technologies (ICT) Sector inthe EU. Accessed December 11, 2012. http://ec.europa.eu/enterprise/ict/policy/doc/2006-11-20-flossimpact.pdf. Prepared by UNUMERIT for the European Commission under the contractENTR/04/112.

Gilbert, R. 2006. Looking for Mr. Schumpeter: Where are we in the Competition-innovation debate?In Innovation Policy and the Economy, vol. 6 of NBER Book Series Innovation Policy and theEconomy, edited by A. D. Jaffe, J. Lerner, and S. Stern, 159–215. Cambridge, MA: MIT Press.

Greene, W. H. 2003. Econometric Analysis. 5th ed. Upper Saddle River, NJ: Prentice Hall Inc.Griliches, Z. 1979. “Issues in Assessing the Contribution of R&D to Productivity Growth.” Bell

Journal of Economics 10 (1): 92–116.Gronau, N. 2010. Enterprise Resource Planning. 2nd extended ed. München: Oldenbourg Verlag.Hall, B. H., F. Lotti, and J. Mairesse. 2009. “Innovation and Productivity in SMEs: Empirical Evidence

for Italy.” Small Business Economics 33 (1): 13–33.Hempell, T. 2005. “Does Experience Matter? Innovations and the Productivity of Information

and Communication Technologies in German services.” Economics of Innovation and NewTechnology 14 (4): 277–303.

Hempell, T., and T. Zwick. 2008. “New Technology, Work Organisation and Innovation.” Economicsof Innovation and New Technology 17 (4): 331–354.

Hendricks, K. B., V. R. Singhal, and J. K. Stratman. 2007. “The Impact of Enterprise Systems onCorporate Performance: A Study of ERP, SCM and CRM System Implementations.” Journal ofOperations Management 25 (1): 65–82.

Hipp, C., and H. Grupp. 2005. “Innovation in the Service Sector: The Demand for Service-SpecificInnovation Measurement Concepts and Typologies.” Research Policy 34 (1): 517–535.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 16: Does enterprise software matter for service innovation? Standardization versus customization

426 B. Engelstätter and M. Sarbu

Hitt, L. M., D. J. Wou, and X. Zhou. 2002. “Investment in Enterprise Resource Planning: Busi-ness Impact and Productivity Measures.” Journal of Management Information Systems 19 (1):78–98.

Howells, J. 2010. “Services and Innovation and Service Innovation: New Theoretical Directions.” InHandbook of Innovation and Services, edited by F. Gallouj and F. Djellal, 68–83. Cheltenham,UK: Edward Elgar.

Hunton, J. E., N. Lippincott, and J. L. Reck. 2003. “Enterprise Resource Planning Systems: Com-paring Firm Performance of Adopters and Non-adopters.” International Journal of AccountingInformation Systems 4 (3): 165–184.

Koch, A., and H. Strotmann. 2006. “Determinants of Innovative Activity in Newly FoundedKnowledgeintensive Business Service Firms.” In Entrepreneurship in the Region, vol. 14 ofInternational Studies in Entrepreneurship, edited by M. Fritsch and J. Schmude, 195–224. NewYork: Springer.

Leiponen, A. 2005. “Organisation of Knowledge and Innovation: The Case of Finnish BusinessServices.” Industry and Innovation 12 (2): 185–203.

Licht, G., and D. Moch. 1999. “Innovation and Information Technology in Services.” CanadianJournal of Economics 32 (2): 363–383.

Malhorta, A., A. Majchrzak, R. Carman, and V. Lott. 2001. “Radical Innovation without Collocation:A Case Study at Boeing-Rocketdyne.” MIS Quarterly 25 (2): 225–249.

Mendelson, H., and R. Pillai. 1999. “Information Age Organizations, Dynamics and Performance.”Journal of Economic Behaviour and Organization 38 (1): 253–281.

Meyer, J. 2010. “Does Social Software Support Service Innovation.” International Journal of theEconomics of Business 17 (3): 289–311.

OECD. 2005. Oslo manual: Guidelines for Collecting and Interpreting Innovation Data. 3rd ed. Paris:OECD.

Peters, B. 2009. “Persistence of Innovation: Stylized Facts and Panel Data Evidence.” Journal ofTechnology Transfer 34 (2): 226–243.

Pilat, D. 2001. “Innovation and Productivity in Services: State of the Art.” In Services and Productivity,edited by OECD, 17–54. Paris: OECD.

Roper, S., J. Du, and J. H. Love. 2006. Knowledge Sourcing and Innovation. Aston Business SchoolResearch Paper 0605. Birmingham, UK: Aston Business School.

Shang, S., and P. B. Seddon. 2002. “Assessing and Managing the Benefits of Enterprise Systems: TheBusiness Manager’s Perspective.” Information Systems Journal 12: 271–299.

Shin, I. 2006. “Adoption of Enterprise Application Software and Firm Performance.” Small BusinessEconomics 26 (3): 241–256.

Sundbo, J. 1997. “Management of Innovation in Services.” The Service Industries Journal 17 (3):432–455.

Sundbo, J., and F. Gallouj. 2000. “Innovation as a Loosely Coupled System in Services.” InternationalJournal of Services Technology and Management 1 (1): 15–36.

Tambe, P., L. M. Hitt, and E. Brynjolfsson. 2012. “The Extroverted Firm: How External InformationPractices Affect Innovation and Productivity.” Management Science 58 (5): 843–859.

Tether, B. S. 2005. “Do Services Innovate (differently)? Insights from the European InnobarometerSurvey.” Industry and Innovation 12 (2): 153–184.

Thomke, S. H. 1998. “Simulation, Learning and R&D Performance: Evidence from AutomotiveDevelopment.” Research Policy 27 (1): 55–74.

Tsai, W. 2001. Knowledge Transfer in Intraorganisational Networks: Effects of Network Positionand Absorptive Capacity on Business Unit Innovation Performance.” Academy of ManagementJournal 44 (1): 996–1004.

Vanberg, M. 2003. Die Konjunkturumfrage bei Dienstleistern der Informationsgesellschaft. ZEWDokumenation 03–09. Mannheim: Centre for European Economic Research.

Vinding, A. L. 2006. “Absorptive Capacity and Innovative Performance: A Human CapitalApproach.” Economics of Innovation and New Technology 15 (4–5): 507–517.

Wieder B., P. Booth, Z. P. Matolcsy, and M.-L. Ossimitz. 2006. “The Impact of ERP Systems onFirm and Business Process Performance.” Journal of Enterprise Information Management 19(1): 13–29.

Wier, B., J. E. Hunton, and H. R. HassabElnaby. 2007. “Enterprise Resource Planning Sys-tems and Non-financial Performance Incentives: The Joint Impact on Corporate Performance.”International Journal of Accounting Information Systems 8 (3): 165–190.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 17: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 427

Appendix

Table A1. Distribution of industries in the sample.

Industry Observations Percentage

Software and IT services 43 12.80ICT-specialized trade 33 9.82Telecommunication services 13 3.87Tax consultancy and accounting 56 16.67Management consultancy 37 11.01Architecture 54 16.07Technical consultancy and planning 34 10.12Research and development 38 11.31Advertising 28 8.33Sum 336 100

Source: ZEW quarterly business survey among service providers of theinformation society, own calculations.

Table A2. Probit estimation results: coefficient estimates.

Dependent variable: dummy for service innovation(1) (2) (3) (5)

Sector specific software −0.151 0.078 −0.036 −0.022(0.164) (0.191) (0.226) (0.277)

Customized software 0.633∗∗∗ 0.630∗∗∗ 0.767∗∗∗ 0.586∗∗(0.144) (0.166) (0.194) (0.234)

Logarithmic firm size 0.139∗∗ 0.082 0.071(0.060) (0.072) (0.087)

Logarithmic firm age −0.192 −0.463∗∗ −0.337(0.164) (0.201) (0.251)

IT-intensity 0.166 −0.203 0.216(0.325) (0.405) (0.531)

Competitors 0–5 −0.088 0.063(0.245) (0.300)

Competitors >20 −0.170 −0.136(0.219) (0.263)

Highly qualified employees 0.082 −0.008(0.383) (0.468)

Medium qualified employees 0.058 −0.438(0.512) (0.610)

Employees 30–55 years −1.445∗∗∗ −0.629(0.554) (0.667)

Employees >55 years −1.136 −0.646(0.692) (0.839)

Prior service innovation 0.816∗∗∗(0.250)

Prior process innovation −0.114(0.259)

Constant term −0.426∗∗∗ −0.410 1.794∗ 0.520(0.156) (0.667) (0.929) (1.119)

Observations 336 298 240 197Pseudo R2 0.046 0.103 0.147 0.206

Note: Reference categories: competitors 6–20, unqualified employees, employees <30 years.∗10% significance level.∗∗5% significance level.∗∗∗1% significance level.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 18: Does enterprise software matter for service innovation? Standardization versus customization

428 B. Engelstätter and M. Sarbu

Table A3. Probit estimation results: average marginal effects, reduced sample.

Dependent variable: dummy for service innovation(1) (2) (3) (4)

Sector specific software −0.040 0.007 −0.001 0.003(0.079) (0.083) (0.084) (0.081)

Customized software 0.290∗∗∗ 0.250∗∗∗ 0.239∗∗∗ 0.182∗∗(0.071) (0.074) (0.074) (0.075)

Logarithmic firm size 0.030 0.026 0.020(0.025) (0.026) (0.025)

Logarithmic firm age −0.167∗∗ −0.158∗∗ −0.098(0.069) (0.072) (0.072)

IT-intensity 0.108 −0.096 0.063(0.139) (0.157) (0.155)

Competitors 0–5 −0.048 0.018(0.094) (0.089)

Competitors >20 −0.014 −0.040(0.078) (0.077)

Highly qualified employees 0.007 −0.002(0.142) (0.137)

Medium qualified employees 0.065 −0.128(0.181) (0.178)

Employees 30–55 years −0.285 −0.184(0.199) (0.194)

Employees >55 years −0.279 −0.189(0.251) (0.244)

Prior service innovation 0.259∗∗∗(0.080)

Prior process innovation −0.033(0.074)

Dummies Sector East Sector East Sector EastObservations 179 179 179 179Pseudo R2 0.071 0.145 0.157 0.206

Note: Reference categories: competitors 6–20, unqualified employees, employees <30 years.∗∗5% significance level.∗∗∗1% significance level.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

20:

26 2

0 Se

ptem

ber

2013

Page 19: Does enterprise software matter for service innovation? Standardization versus customization

Economics of Innovation and New Technology 429

Table A4. Descriptive statistics of reduced sample.

Variable Mean Minimum Maximum N

Service innovation 0.357 0 1 179Sector specific software 0.754 0 1 179Customized software 0.424 0 1 179Number of employees 39 1 449 179Log (number of employees) 2.672 0 6.107 179Firm age 20 4 108 179Log (firm age) 2.849 1.386 4.682 1790–5 competitors 0.229 0 1 1796–20 competitors 0.329 0 1 179More than 20 competitors 0.441 0 1 179Share of employees working with PC 0.799 0.01 1 179Share of highly qualified employees 0.424 0 1 179Share of medium qualified employees 0.153 0 1 179Share of low qualified employees 0.373 0 1 179Share of employees younger than 30 years 0.181 0 1 179Share of employees between 30 and 55 years 0.652 0 1 179Share of employees older than 55 years 0.159 0 1 179Prior service innovation 0.418 0 1 179Prior process innovation 0.418 0 1 179East Germany 0.376 0 1 179

Source: ZEW quarterly business survey among service providers of the information society, owncalculations.

Table A5. Distribution of industries in the reduced sample.

Industry Observations Percentage

Software and IT services 24 13.41ICT-specialized trade 16 8.94Telecommunication services 4 2.23Tax consultancy and accounting 34 18.99Management consultancy 18 10.06Architecture 36 20.11Technical consultancy and planning 16 8.94Research and development 16 8.94Advertising 15 8.38Sum 179 100

Source: ZEW quarterly business survey among service providers of theinformation society, own calculations.D

ownl

oade

d by

[U

meå

Uni

vers

ity L

ibra

ry]

at 2

0:26

20

Sept

embe

r 20

13