Structural Change and Economic Dynamics 27 (2013) 146– 159
Contents lists available at ScienceDirect
Structural Change and Economic Dynamics
journa l h omepa g e: www.elsev ier .com/ locate /sced
rom too little to too much innovation? Issues in measuringnnovation in the public sector
nthony Arundela,b,∗, Dorothea Huberc
Australian Innovation Research Centre, 1 College Road, University of Tasmania, Hobart, AustraliaUNU-MERIT, Maastricht, The NetherlandsDepartment of Innovation, Industry, Science, Research and Tertiary Education, 10 Binara Street, Canberra, Australia
a r t i c l e i n f o
rticle history:eceived January 2013eceived in revised form June 2013ccepted June 2013vailable online 2 September 2013
EL classification:3132
a b s t r a c t
Interviews with 37 branch level managers in the Australian Federal Government wereconducted to determine how managers understood the concept of innovation and theirfamiliarity with different types of innovations. A follow-on survey found that 91% ofbranches introduced an innovation in the previous two years. This high rate suggests thatmany of the innovations could be minor. Extensive cognitive testing found that public sectormanagers can provide high quality estimates of the amount of person months expended oninnovations and on other measures of the significance of an innovation. Using this informa-tion, the share of branches that introduced a significant innovation is approximately 60%.
Although suggestive, there is no statistically significant difference in the time required todevelop innovations derived from ideas provided by upper management or by lower levelstaff. These and other results are relevant to the design and interpretation of public sectorinnovation surveys.
The OECD (2012) defines the public sector as ‘the gen-ral government sector at the national, regional and localevels plus all public corporations including the central
ank’. Based on this definition, the public sector accountsor between 20% and 30% of GDP in economically developedountries. 1 This is a substantial share of economic output
∗ Corresponding author at: Australian Innovation Research Centre, 1ollege Road, University of Tasmania, Hobart, Australia. Tel.: +61 3 6226357.
E-mail address: [email protected] (A. Arundel).1 Estimates of the public sector share of GDP of up to 50% or 55%, widely
eported in popular magazines and newspapers and in some academicesearch (see Potts and Kastelle, 2010; Thenint, 2010) are due to confusingax revenues with GDP. In 2005, the tax revenue share of GDP in Denmarknd Sweden was close to or over 50%, at 50.8% and 48.9% respectivelyOECD, 2012). Tax revenue is not a measure of the economic contribution
and considerably more than the share of manufacturing inmost countries. Not surprisingly, there is growing policyinterest in how to measure and evaluate innovation in thepublic sector as part of a goal to improve the efficiencyand quality of public sector services. This has led to theEuropean Commission’s support for several large projectsof relevance to public sector innovation.2
Measurement requires agreement on how to define
innovation in the public sector. Although there is currentlya lack of agreement, with multiple definitions of differenttypes of public sector innovations, a common theme is that
of the public sector to GDP because a sizable fraction of these revenues isspent on transfer payments to individuals for pensions and welfare pay-ments or to private businesses as subsidies. The economic contribution ofthese transfer payments to GDP is based on how the recipients choose tospend or invest these payments.
2 Examples include the ServPPIN project (Rubalcaba et al., 2013) andPUBLIN (Koch and Hauknes, 2005).
istics of private businesses. Laegreid et al. (2011) is the onlystudy to examine innovation in quasi-autonomous publicorganizations.
4 The method is biased towards publications in English and excludes
A. Arundel, D. Huber / Structural Chang
public sector innovation involves novelty and the intentionof making something better, for instance through new orimproved services and processes.
It is not clear when academic researchers began to sys-tematically examine public sector innovation. Roessner’s(1977) study ‘Incentives to innovate in public and privateorganizations’ could be the first study to directly examinethe concept of innovation in the public sector, but relevantresearch from a management or entrepreneurial perspec-tive was published in the 1960s.3 Until the early 2000s,most research on public sector innovation was by aca-demics in one of these two disciplines and appears to havebeen dominated by case studies.
Management research in the 1980s and 1990s focusedon the adoption of New Public Management, whichsupported major organizational changes to reduce hier-archical structures and apply practices in use in theprivate sector, such as contracting out, performance tar-gets, internal markets to provide pressure to innovate, andincreased independence for senior management (Bartlettand Dibben, 2002; Laegreid et al., 2011). New Public Man-agement methods were viewed as a solution to a perceivedlack of innovation in the public sector, due in part to anaversion to risk and an anti-innovation public sector cul-ture (Windrum, 2008). The increased responsibility formanagers would permit them to make top-down decisionsto introduce innovations. Other factors that were seen aslimiting public sector innovation included several uniqueattributes of the public sector in comparison to privatefirms, such as deriving revenue from budgetary allocationsrather than sales (which removes a market incentive toinnovate), the need to respond to many constituencies, andbecause government must meet explicit moral imperativesas well as demands for economic efficiency (Drucker, 1985).The perception that the public sector is non-innovative dueto risk aversion, a hostile culture to innovation and a lack ofincentives has persisted over time (see Mulgan and Albury(2003) and Potts and Kastelle, 2010).
This perception could partly be due to the dominanceof case studies in research on public sector innova-tion. Although case studies provide a valuable in-depthunderstanding of how innovation can occur and are ofconsiderable value to theory building, they are not gen-eralizable to a population (in this case the population ofpublic sector organizations) and they cannot provide indi-cators that can be used for benchmarking or tracking theprevalence of innovation activity over time. A full under-standing of the types of innovations that are developed inthe public sector, how innovation occurs, and the factorsthat promote or hinder innovation requires data on a largenumber of public sector organizations or on a large numberof public sector innovations.
Larger-scale studies of public sector innovation are rel-atively recent, with almost all publications appearing after2000. Since 2005, the focus of this research has shifted
towards innovation surveys, almost all of which have beenmodelled to a greater or lesser extent on innovation sur-veys for the private sector. This is work in progress: to
3 See Windrum (2008) for a review of some of these earlier studies.
onomic Dynamics 27 (2013) 146– 159 147
date academic researchers lack an adequate theory forhow public sector innovation occurs to guide the designof innovation questionnaires for public organizations. Thisrequires a lengthy iterative process between theory gen-eration and empirical testing. As an example, the OECD’sfirst edition of the Oslo Manual (OECD, 1992), providingguidelines for how to measure innovation in the privatesector, was based on over a decade of experimental surveyresearch in Canada, Europe and the United States (Arundeland Smith, 2013).
The goal of many academic researchers is to developsufficient understanding of public sector innovation to pro-duce a manual for measuring innovation in the publicsector that is equivalent to the Oslo Manual. The purposeof this article is to use the results of a series of interviewsand a pilot survey of public sector innovation in Australia tocontribute to two issues of relevance to this goal. The firstis how public sector managers view innovation, includingtheir understanding of different types of innovations. Theresults of this research can help to address a puzzle: why dolarge-scale innovation surveys find that over 80% of pub-lic sector organizations are innovative, given the assumedbarriers to innovation such as the lack of market drivers?The second is the source of the ideas for innovation. Bothissues are relevant to not only the design of innovationquestionnaires, but also to how we interpret the resultsof public-sector innovation surveys.
The next section of the paper provides a brief summaryof large-scale research into public sector innovation. This isfollowed by a description of the methodology, a discussionof the results, and a few conclusions of relevance to futurework.
2. Progress towards measuring public sectorinnovation
Keyword searches, citation analysis and a review oflisted references to published documents identified 18studies that used 15 surveys and other large-scale datasources to evaluate public sector innovation in developedeconomies, with the exception of service providers forhealth and education, which were specifically excluded.4
None of the large-scale studies cover all types of pub-lic sector organizations, with most focused on publicadministration. Only Palmer and Dunford (2001) examineinnovation by large state-owned companies. These typesof companies are often excluded from research on publicsector innovation because they are frequently subject tocompetition and the goal of earning a profit, both character-
relevant studies outside Europe, Australia, Canada and the United States.In addition, we do not include research in developing countries, suchas the object-based studies by Wu et al. (2010) of public sector innova-tion in China. Health and education providers such as hospitals, schoolsand universities were excluded because they have been the subject ofinnovation-related research for decades and have particular characteris-tics that may not be comparable to other public sector organizations.
148 A. Arundel, D. Huber / Structural Change and Economic Dynamics 27 (2013) 146– 159
Table 1Methodology of empirical research on public sector innovation in developed countries (excludes case studies).
Reference Date of data Data collectionmethod
Study size Country Type of Public Organization
Object-based studies (examines specific innovations introduced by public sector organizations)Borins (2000, 2001) 1990–1998 Semi-finalists to
United Kingdom Examples from 85 centraldepartments & high levelagencies including health,environment, housing,revenue, learning, defense,water services, ordnancesurvey, etc.
Business practice studies (asks public sector organizations if they had introduced each of a specified list of practices)Damanpour andSchneider (2006),Moon and de Leon(2001)
1997, 2001 List ofmunicipalitieswith > 10,000 pop
1276 citymanagers
United States Municipal governments
Palmer and Dunford(2001)
1999? List of largestpublic commercialorgs
90 humanresourcemanagers
Australia Government ownedcommercial enterprises
Lonti and Verma (2003) 1998–1999 List of public orgs 774 work unitmanagers
Canada Public administration,government ownedcommercial enterprises, andagencies
Walker (2006) 2001 List of localauthorities
120authorities, upto 7 informantsper authority
England Local authorities in education,social care, planning, wastemanagement, housing, library,leisure, benefits
Earl (2004) 2000–2002 Business register ofpublic and privateorgs
810 agencies Canada Educational services, healthcare & social assistance, publicadministration
Survey + focused innovation surveys (IS)Laegreid et al. (2011) 2002–2004 List of public orgs 121 head
managersNorway andFinland
Public agencies (excludespublic administration(ministries) and commercialgovernment organizations)
Central and local publicadministration plus hospitalsand schools in some countries
Hughes et al. (2011)(IS)
2010 List of public orgs 175 headmanagers
England Health and local governmentauthorities
EC (2010), Arundel andHollanders (2011) (IS)
2010 Business register 3699 managers EU-27 countries Local, regional and nationalpublic administration officesdefined by NACE codes.
APSC (2011) 2011 Census of agencies, 97 agencyheads,
10,000employ
Australia Public administration agencies
so(
sample ofemployees
As summarized in Table 1, large scale studies of publicector innovation have used three research methods: (1) anbject-based method that examines specific innovationsthe object), (2) business practice surveys that ask public
approx.
ees
(ministries). Employee surveylimited to ‘most importantinnovation’
sector managers about their use of specific innovative busi-ness practices and technologies, and (3) innovation surveysthat ask about a range of innovation activities and typesof innovations that were implemented over a defined time
e and Ec
came from the ‘bottom up’, including middle management(below agency heads) and front-line staff. In the UnitedStates, middle management played an initiator role in 82%
A. Arundel, D. Huber / Structural Chang
period. The methodology has shifted from object-based andbusiness practice studies to innovation surveys, with fourof the six innovation surveys influenced by the Olso Manualguidelines for measuring innovation in the private sector.5
There are several causes for this shift in methodolo-gies. First, as noted by Fuglsang (2010), theories of publicsector innovation have shifted from a focus on imposedpolicy-driven change to an interest in internal problemsolving capabilities. Second, the object method is biasedtowards the best and most successful innovations (Pottsand Kastelle, 2010) and often intentionally selects success-ful innovations as in the innovation award semi-finalistsstudied by Borins (2000, 2001, 2010) or in the study byParna and von Tunzelmann (2007). Although this is of valuefor research on the factors that are linked to high qualitysuccessful innovations, this approach cannot explore thefactors that influence the full range of innovations that areimplemented by public sector organizations. In contrast,innovation surveys can collect data on processes within thepublic sector organization to support innovation and avoidthe bias issue by collecting information on a wide range ofdifferent types of innovations.
The Oslo Manual identifies four types of innovation inthe private sector: product, process, organizational, andmarketing innovations. Currently, there is no agreed upontypology for innovation in the public sector. The MEPINstudy (Bugge et al., 2011) follows the Oslo Manual def-initions of private sector innovations more closely thanother studies, identifying product, process, organizationaland communication innovations, with the latter replacingmarketing innovations. Other studies include both severaltypes of innovation that occur in the private sector, such asservice and organizational innovation, plus specific types ofinnovations that are thought to be uniquely characteristicof the public sector. For example, the Audit Commission(2007) in the United Kingdom identifies democraticinnovation, or new practices for increasing democraticengagement with citizens. Windrum (2008) suggests twonew categories: conceptual innovation involving ‘newmissions, new world views, objectives, strategies and ratio-nales’ and policy innovations.
Most of the studies listed in Table 1 collected informa-tion on the type of innovation, such as a process, service orstrategic innovation, although Parna and von Tunzelmann(2007) only report that the innovations involved ‘techno-logical’ features. The most frequently evaluated innovationtypes are service innovations and process innovations.Several studies also collected data on organizational inno-vations, although Arundel and Hollanders (2011) combinedprocess and organizational innovations because of a con-cern that respondents might find it very difficult to separatethe two activities. Three studies also collected data onmarketing or communication innovations, which are con-
ceptually similar. Data on other types of innovations arecollected in only one or two studies: products, strategic,
5 These include the studies by Bugge et al. (2011), EC (2010), Arundeland Hollanders (2011), Hughes et al. (2011) and the Audit Commission(2007).
onomic Dynamics 27 (2013) 146– 159 149
systemic, and structural. None of the studies collected dataon policy innovations.
Data on product or goods innovations are not collectedin most studies because of the belief that the public sec-tor rarely develop and or implement new types of goods,with the focus primarily on services. However, the MEPINsurvey for Scandinavia (Bugge et al., 2011) found thatapproximately 10% of responding public sector organiza-tions reported this type of innovation.
Strategic innovation is defined differently in two stud-ies. The Audit Commission’s (2007) survey defines strategicinnovation as ‘re-positioning the authority in line with newcorporate objectives or new ‘customers’, including usingalternative service delivery models’. The NESTA surveydefines strategic innovations as ‘changes to your organiza-tion’s activities (e.g. by outsourcing services and processes,or by building new capabilities within your organization’(Hughes et al., 2011). Both definitions are problematic, theformer because service delivery models could fall underprocess innovation (it is not clear how different these newmodels would need to be to count as a strategic innovation)while the latter definition can be subsumed under organi-zational innovations. Systemic and structural innovationsare also either difficult to define or overlap with other typesof innovations.
The innovation surveys collect a wider range of othertypes of data than the object-based and business practicesurveys, including information on external informationsources, incentives, the source of innovative ideas, and out-comes. Only the study by Hughes et al. (2011) collectsinformation on investment in innovation.
An important issue in many of the large scale studies ishow innovation occurs, or where do the ideas come fromand what factors support the development of the idea intoan innovation? This question concerns the process of devel-oping and implementing an innovation and involves theactivities of different levels of staff, ranging from front-linestaff to top-level management.
The ‘traditional’ view is that innovation in the pub-lic sector is initiated by high-level policy decisions at theministerial or political level and implemented through theactions of senior management (Hartley, 2005). The latter,if empowered through new public management strate-gies, can also decide to implement innovations, but inall cases innovation is primarily viewed as driven fromthe ‘top down’.6 This view was called into question byBorins’ (2000, 2001) results finding that the initiators forthe majority of innovations that were award semi-finalists
6 Brudney and Hebert’s (1987) analysis of a 1978 survey of 1393 headsof State agencies in the United States questions the dominance of topdown influences on decisions in the public sector, although they did notlook directly at innovation. They examined the influence of governor, leg-islature, clientele groups and professional associations on four outputs:the total agency budget, the budget for specific programmes, changes inpolicy, and changes to rules and regulations. They find that the public sec-tor is not homogenous, with differences in the influence of policy drivers(governor and legislature) on outputs.
1 e and Ec
ocat
idsiamrtfApitmm
wgdDS‘4(btfitt
Dnsdu2attd3Boaimmmtiits
i
50 A. Arundel, D. Huber / Structural Chang
f the innovations while 20% of the initiators were politi-ians and 25% front-line staff (the total sums to over 100%s it is possible to have two or more initiators per innova-ion).
Research since 2000 has supported Borins results, find-ng that innovative ideas come from both sides of theivide: from middle management and below and fromenior management up to politicians. Hughes et al. (2011),n a survey of 175 local governments and health servicegencies, finds that frontline staff account for 65% of theain idea for innovations. Managers are cited at the same
ate for health service agencies, but account for only 48% ofhe ideas for innovations by local governments. The ideasor innovation can often come from multiple sources. ThePSC (2011) survey of approximately 10,000 Australianublic sector employees at all job levels found that valuable
deas for the ‘most significant innovation’ implemented byhe respondents’ work group were obtained from other
embers of their work group (reported by 90%) and senioranagers (reported by 68%).The NAO (2006) study examined 125 innovations that
ere reported in 2006 by the senior management of 85overnment departments and agencies in the United King-om, including the Environment Agency, the Ministry ofefense, and large executive agencies such as the Ordnanceurvey and the UK Passport Service. The study found that
senior management or middle management’ accounted for8% of 216 citations for the primary source of the ideamore than one primary source per innovation was possi-le), central government or ministers accounted for 19% ofhe citations, and front-line staff or individuals accountedor only 14%. The study noted that about 25% of the reportednnovations were relatively minor, but did not correlatehe source of the idea with the novelty of the innova-ion.
Fuglsang (2010), in a case study of elderly care inenmark, reports that the source of the idea influences theovelty or ‘innovation intensity’ of innovations. Front-linetaff introduce incremental innovations while managersevelop innovations to solve larger-scale problems. Borinspdated his research on innovation award finalists in a010 paper and observed that the role of middle man-gers as the initiators of innovations had declined overime, from 43% of semi-finalists between 1995 and 1998,o 32% between 2008 and 2009. The role of front-line staffeclined even more, from 27% between 1995 and 1998 to% between 2008 and 2009. One explanation suggested byorins (2010) is that the bar for semi-finalists had increasedver time, possibly because the easier innovations hadlready been implemented. This could result in a declinen the share of semi-finalists that are front line staff and
iddle managers if these two groups tend to develop incre-ental innovations or less novel innovations than senioranagers. Thenint (2010) suggests that bottom up innova-
ions are likely to have narrower impacts than top downnnovations because they are created to solve problemsn specific locations. Consequently, they are more likely
han top down innovation to remain isolated and to diffuselowly, if at all.
Several of the large scale studies found that not only wasnnovation common among public sector organizations,
onomic Dynamics 27 (2013) 146– 159
but that it was equally or more prevalent among pub-lic sector organizations than among private sector firms.Earl (2004) found little difference in the adoption ratefor innovative business practices between public organiza-tions and private firms with over 500 full-time employeesin Canada, although a higher percentage of private firmsdeveloped new technology in-house (54% versus 45%).The 2010 Innobarometer survey of 3699 European pub-lic sector organizations found that 91.5% had introducedan innovation over a two-year period. This is considerablyhigher than innovation rates among European businessesof approximately 52.8% over a three-year period, as identi-fied in the 2010 European Community Innovation Survey.Innovation rates of over 80% for some countries and typesof public sector organizations were also reported by theMEPIN study (Bugge et al., 2011).
The observed high rates of innovation in the publicsector conflicts with many of the common perspectiveson public sector innovation, such as risk averse attitudesamong public sector managers and a lack of incentivesto innovate. An alternative explanation is that the typesof innovations reported by public sector organizations arenot comparable to those reported by private sector firms,for instance they could require substantially less effort todevelop or have only minor effects. Answering this ques-tion requires obtaining good data on the types of innovationintroduced by public sector organizations and the neces-sary inputs to develop them.
3. Methodology
For research purposes, the public sector can be subdi-vided into five categories: public administration; publicservice providers of fire protection, police, education,health, residential care, and research; quangos and quasi-autonomous organizations; publicly-owned commercialcorporations; and public–private partnerships where boththe government and businesses share in providing aservice. The current study is limited to public adminis-tration and uses interviews and a survey to assess theinnovation activities within the Australian national gov-ernment.
A major issue for questionnaire surveys of publicsector innovation is the selection of the potential pop-ulation of respondents. Previous surveys have compiledlists of public sector organizations from a range of datasources or selected readily identifiable organizations suchas municipalities or local authorities (see Table 1). A morecomprehensive alternative used by the Innobarometer (EC,2010) and MEPIN (Bugge et al., 2011) surveys is to constructa population of public sector organizations from businessregisters. This is similar to how firms are selected forinnovation surveys. In the public sector context, businessregisters provide sampling units at all levels of govern-ment; local, regional and national. As with businesses,these units can vary substantially in size, from very largenational ministries of Health or Education with tens of
thousands of employees to small municipalities or agencieswith less than ten employees. The variation in size couldcreate problems if the managers of very large organiza-tions are too far removed from many types of innovative
e and Ec
A. Arundel, D. Huber / Structural Chang
activities to have a full understanding of how they aredeveloped and implemented.
This study addresses this issue by sampling Band 1 man-agers of the Senior Executive service, using the AustralianPublic Service Commission’s database of all 167,000 FederalGovernment employees. These managers have sufficientresponsibility to develop and implement innovations, butare at a lower level in the Government hierarchy than upperlevel managers and the heads of Government Departmentsor Agencies.7 The Band 1 group consists of approximately2100 employees, almost all of which are branch managers.Those not responsible for a branch and are excluded fromthe analyses.
Cognitive testing was conducted in May and June of2012 to determine if potential respondents understood thequestions as intended and if they were able to provideaccurate answers. 8 The testing was based on face-to-faceinterviews (of about 1 h duration) and followed estab-lished methods (Collins, 2003; Willis, 1994). A total of 16women and 14 men who held SES Band 1 positions wereinterviewed. Although interviewees were not selected ran-domly, every effort was made to interview managers froma diverse range of government agencies and functions. Themanagers were on average responsible for 42 employees(after excluding an outlier with 640 employees). The inter-views were conducted in three phases of 10 interviews eachand results from the first and second phases were used tomake several adjustments to the questions in the secondand third phases respectively.
The survey was implemented in August and September2012 using an online questionnaire. The questionnaire tookapproximately 30 min to complete if all questions wereanswered. The questions covered innovative activities overthe previous two fiscal years (from 1 July 2010 to 30 June2012).
The response rate after two reminder emails wasapproximately 40%, with 367 responses. After removing3.5% of respondents with no employees, 356 respondentsonly answered questions about their branch while 344respondents answered questions on the types of innova-tions implemented by their branch.9 Due to an error in theonline filter separating the first and second sections of thequestionnaire, only 145 respondents answered questionson their branch’s most successful innovation.
Following the survey, seven post-survey interviews(average length of 50 min) were conducted in midNovember, 2012. Respondents were asked about their gen-eral impression of the survey and their interpretation of
questions on different types of innovation and the role ofhigh level management in innovation.
7 There are approximately 20 departments (such as the Department ofDefence and the Department of Health and Ageing) and over 100 agencies(such as Medicare Australia and the Bureau of Meteorology).
8 One purpose of the interviews was to improve the design of specificquestions. This resulted in many minor changes to question wording, theexclusion of several questions, and major redesigns of some questions.A summary report on these technical changes to the questionnaire isavailable from the authors on request.
9 As some respondents did not answer all questions, the number ofresponses can be less than 344.
onomic Dynamics 27 (2013) 146– 159 151
3.1. Defining innovation
The questionnaire followed the Oslo Manual(OECD/Eurostat, 2005) guidelines for measuring inno-vation in the private sector by defining an innovationas either “new or substantially changed”, although itneed not be developed by the branch in question. Surveyrespondents were also informed that innovations “needonly to be new or substantially changed for your branch.It may have already been in use by other governmentorganizations or by businesses”. The adoption of methodsor technologies that are in use elsewhere are included asinnovations because of their importance to productivityimprovements. The MEPIN and Innobarometer surveysalso included the adoption of innovations developedelsewhere.
The questionnaire was divided into two sections. Thefirst section included questions on the respondent andhis or her branch10 and on each of four types of innova-tions: processes, products (including goods) and services,communication, and policy innovations. Other varieties ofinnovations that had been discussed in the literature as ofpotential importance in the public sector, such as strate-gic, systemic or conceptual innovations, were not includedbecause they were difficult to define or likely to be veryrare and consequently not suitable for a survey.
The questionnaire asked about different varieties ofeach innovation type. For instance, the section on processinnovations included five varieties of process innovations.Annex A provides a full description of all innovation vari-eties included in the survey questionnaire: five for process,four for product and service, four for communication, andthree for policy innovations. In addition, respondents wereasked, for each innovation type, if activities to develop aninnovation were underway in the last year and they werealso given an open question that allowed them to describeother varieties of each innovation type.
As a result of the very high innovation rates reportedfor the MEPIN and Innobarometer surveys, several ques-tions were included in the survey to differentiate betweenminor changes and more significant innovations. As a firststep, the questionnaire asks specifically about ‘new orsubstantially changed’ innovations, but the word ‘substan-tial’ is subjective and can be interpreted very differentlyby different respondents. Consequently, additional ques-tions were included that were inspired by the EuropeanUnion’s Community Innovation Survey for the private sec-tor, which includes questions on the amount of investmentin innovation and if the firm had introduced a ‘coun-try first’ innovation. The cognitive testing showed thatonly a small percentage of public sector managers couldprovide a dollar figure for investment, but almost all couldprovide a good estimate of the amount of person monthsexpended on innovation. In addition, the questionnaire
included questions on ‘Australia first’ innovations and ifany Australia-first innovations were later adopted by other
10 Questions related mainly to branch function, number of employees,years the respondent has managed the branch, etc.
1 e and Economic Dynamics 27 (2013) 146– 159
Al
vtmebbofir
qqcite
4
4
bfstton
lAfatmm
17.215.5
23.9
29.3
14.1
0
5
10
15
20
25
30
35
Less than sixmonths
Six months toless than 1
year
1 year to lessthan 2 years
2 years to lessthan 4 years
4 years ormore
52 A. Arundel, D. Huber / Structural Chang
ustralian government organizations. The latter is onlyikely to occur for innovations of significant value.
Of note, these measures of the ‘significance’ of an inno-ation in the public sector are imperfect. For instance, theime expended to develop an innovation is not a direct
easure of novelty or of the effect of an innovation on thefficiency of processes or the quality of services. It is possi-le that a process or service innovation that was developedy a different organization and quickly adopted with onlyne or two person months of effort could have larger bene-ts than an entirely novel or ‘Australia first’ innovation thatequired several years and many person months to develop.
The second section of the questionnaire includeduestions on innovation investments and strategies anduestions on the branch’s most successful and least suc-essful innovations. A question on the most successfulnnovation asked the respondent to give a short writ-en description. This information was used to assign eachxample to one of the four main innovation types.
. Results and discussion
.1. Characteristics of the respondent and branch
Fig. 1 gives the distribution of branches in the surveyy nine main functions, plus an ‘other’ category. The mostrequent type of function is policy, advisory and strategicervices, reported by 32.1% of the branches. This highlightshe importance of including questions on policy innova-ion so that all respondents can find relevant questionsn their branch’s innovative activities in the question-aire.
Fig. 2 gives the distribution of branch managers by theength of time they have been responsible for their branch.lmost a third, 32.7%, had been responsible for their branch
or less than one year. Given that the questionnaire asks
bout innovations that were introduced in the precedingwo years, this could bias innovation rates downwards if
anagers are unaware of innovations that were imple-ented before they arrived.
Fig. 1. Distribution of 349 respondin
Fig. 2. Percent of 355 branches by the time the manager has been respon-sible for the branch.
The effect of time on the share of branches that reportedan innovation can be examined by comparing the per-cent of managers that reported that their branch hadimplemented an innovation in the preceding two fiscalyears and the percent of managers that reported that theirbranch had either implemented an innovation or had activ-ities underway to develop an innovation during this timeperiod. The Oslo Manual defines the combination of imple-mented innovations and innovations under developmentas ‘innovation active’. Table 2 provides the results for pro-cess and communication innovations (the results are notshown, but similar patterns occur for service and policyinnovations). Notably fewer managers with less than sixmonths of responsibility for their branch report imple-mented process innovations (67.2%) than the average of82.3%. For innovation active branches, the difference fallsfrom 15.1% percentage points (82.3–67.2%) to 6.5% percent-age points (80.3–86.8%). There is almost no difference forimplemented communication innovations between man-agers with less than six months in their position comparedto the average, but conversely a slightly higher percentage
of managers with less than six months in their position areinnovative active compared to the average (54.1% versus48.7%).
g branches by main function.
A. Arundel, D. Huber / Structural Change and Economic Dynamics 27 (2013) 146– 159 153
Table 2Percent of innovative branches, by length of time the manager has been responsible for the branch.
Process innovations Communication innovations
Time responsible for branch Implemented Implemented or underway Implemented Implemented or underway
Less than six months 67.2 80.3 39.3 54.1Six months to less than 1 year 80.0 81.8 45.5 60.01 year to less than 2 years 88.2 90.6 43.5 45.92 years to less than 4 years 88.5 92.3 42.3 49.04 years or more 80.0 82.0 28.0 34.0Average 82.3 86.8 40.6 48.7P (difference in distribution) 0.005 0.084 ns 0.087
ced an i
we assume that all non-respondents were non-innovators,the share of innovators would fall to 36.5%.11 The sameassumption would also reduce the share of innovative
Fig. 3. Percent of branches that had introdu
These results indicate that surveys of innovation in thepublic sector should obtain information on the time thatmanagers have been in their current position as well asinformation on innovations that have not yet been com-pleted.
Of interest, the results in Table 2 also suggest that inno-vative activity reaches a peak among managers who havebeen in their current position for two to four years, witha decline in innovative activity for managers with four ormore years of responsibility. This effect was discussed inthe post-survey interviews. The interviewees’ perspectivewas that innovative activity is partly linked to the life cycleof a branch, with newly established branches consideredto be inherently more innovative, with the act of estab-lishing a new branch often viewed as an innovation of itsown. The innovative activity of a new branch is thought togradually decline over time as the branch goals are met.The survey results partially support the interview results,with significantly higher rates of process and communi-cation innovations for branches that have existed for lessthan one year compared to older branches. For example(no table provided), 66.7% of branches in existence for lessthan one year were innovation active for communicationinnovations compared to 42.8% of branches that existed forfour years or more (p = 0.01). A policy innovation was eitherintroduced or underway by 64.3% of branches in existence
for less than one year, versus only 48.3% of branches inexistence for four years or more (p = 0.09). However, thereis no difference in innovation rates by the age of the branchfor service and process innovations.
nnovation in the preceding two fiscal years.
4.2. Innovation rates
Fig. 3 provides the percentage of 344 branches that hadintroduced each of the four types of innovation in the pre-ceding two fiscal years, plus the percent of branches thathad introduced any type of innovation. Process innovationsare the most common, reported by 84.9% of branches, withthe lowest rate for communication innovations, reportedby 41.9% of branches. The share of branches that reportedat least one type of innovation in the preceding two yearsis 91.3%, which is very similar to the rate of 91.5% from theEuropean Innobarometer survey. The innovation rate forAustralian businesses of 39.1% for the 2010/11 fiscal year isnot fully comparable as it is based on a one-year referenceperiod (ABS, 2012).
A possible cause for the high innovation rate in Australiais a difference in the survey response rate betweenthe managers of innovative branches and non-innovativebranches. Managers who did not innovate could have beenless likely to respond, for instance if they did not find thequestionnaire of relevance to their branch’s activities. If
11 Using this assumption, the innovation rate equals the innovation rateamong respondents times the share of the sample that responded to thesurvey (0.913 × 0.40) plus the innovation rate among non-respondentstime the share of the sample that did not respond to the survey (0.0 × 0.60).
154 A. Arundel, D. Huber / Structural Change and Economic Dynamics 27 (2013) 146– 159
enting the branch’s most successful innovation, percent of 134 respondents.
b2
pvtmfioitavic
stihnbwmdvgm
rtiTrie
hne
Fig. 4. Distribution of person months spent on developing and implem
usinesses in Europe over the three years from 2008 to010, from 52.8% to 40.0% or less.12
The alternative estimates of the share of Australianublic sector branches that introduced a significant inno-ation are considerably lower than 91.3%. Fig. 4 provideshe distribution of person months to develop the branch’s
ost successful innovation. Sixty percent of managersrom innovative branches reported that the most successfulnnovation required six months or longer to develop, whilenly 10% reported spending less than one month on thisnnovation. The former are likely to be substantial innova-ions, but some of the latter may be marginal changes thatre not innovations. However, since branches can inno-ate by implementing new ideas developed by others, evennnovations that require less than one month to developould involve more than minor changes.
The question for Australia-first innovations was askedeparately for process, product and service, communica-ion and policy innovations. An Australia-first innovationncludes innovations that already existed, as long as theyad not previously been used by any government orga-ization within Australia. As shown in Fig. 5, 40.7% ofranches reported at least one Australia-first innovation,ith process innovations the most common type and com-unication innovations the least common. Discussions
uring the cognitive interviews indicated that the inter-
iewees were not always aware of innovations in otherovernment agencies and consequently the above figureay be an overestimation.13
12 The country average response rate is 75.8%, based on reportedesponse rates in Eurostat’s national CIS 2010 quality reports for 22 ofhe 27 EU member states and other data for the UK. A better comparisons to use the average response rate for enterprises, which is unavailable.his will be lower than the country average of 75.8% because the averageesponse rate for three large EU member states (UK, Italy and Germany)s 42.7%. These three countries will account for a large share of all EUnterprises.13 The question instructions state that an Australia first innovation couldave already been in use by governments outside Australia or by busi-esses inside or outside Australia. ‘Don’t know’ responses are assumed toqual a ‘no’. We only expect the answers to this question to be suggestive
Fig. 5. Percent of 344 branches reporting ‘first to Australia’ innovations.
Another measure of a significant innovation rate is thepercentage of branches that introduced an Australia-firstinnovation that was later adopted by a different Aus-tralian government organization. This type of innovationwas reported by 13.7% of the branch managers. This is onlylikely to occur if the innovation offered a substantial ben-efit. Furthermore, the respondent would need to be awareof the interest of a different Government organization inorder to give a positive answer to the question, reducingthe likelihood of overestimating these innovations.
A best estimate of substantial innovations can be cal-culated from the percentage of branches that either spentsix months or more on their most successful innovationor introduced an Australia-first innovation that was later
adopted by another government organization. The resultis 67.1%. Unfortunately, this indicator can only be calcu-lated for respondents who answered the question on the
of the actual rate of Australia first innovations (with a random distributionof the error rates), since respondents may not have full knowledge of theactivities of other government organizations. For instance, cognitive test-ing identified one case where the interviewee was unaware that one oftheir innovations had been trialled by a different government departmenteight years previously.
e and Ec
A. Arundel, D. Huber / Structural Chang
time spent on their most successful innovation, which maybias the result upwards if the managers of more innovativebranches were more likely to answer the second sectionof the questionnaire. The ratio of 67.1% when applied tothe 91.3% of all branch managers that innovate gives a sub-stantial innovation rate of 61.2%. This result indicates thatmost of the branches implemented more than minor inno-vations, although the true rate of ‘substantial innovators’could lie closer to 61.2% than 91.3%. The cognitive testingresults discussed below indicate that the higher figure of91.3% is likely to include a number of branches with onlyvery minor innovations.
4.3. Cognitive interview results
The cognitive interviews started by asking the inter-viewees to give their own definition of an innovation. Themajority of interviewees stated that an innovation had twocomponents: it had to be novel and secondly it had to bea better, more effective or more efficient way of obtain-ing an outcome. Several interviewees offered definitions ofinnovation that followed the Oslo Manual approach, notingthat innovation covers a spectrum, from ‘small continu-ous improvements to large quantitative jumps in improvedperformance’ or ‘from marginal improvements to light bulbmoments’. Of interest, the definitions offered by the inter-viewees were rarely in terms of innovation types, such asprocesses or services.
The interviewees were next shown statements aboutthree characteristics of an innovation: it can fail, it needonly be new to their own branch, and it must be imple-mented. The first two characteristics were well received,but the last was frequently not understood and oftenforgotten, even though the questions for each type of inno-vation stated that an innovation had to be ‘introduced’ or‘implemented’.
Although there was consensus that innovations hadto be novel, there was less agreement on the amount ofnovelty or ‘innovativeness’ that was required. To explorethis issue, interviewees were asked to give an exampleof a change that was not an innovation. The intervieweesoften had difficulties in differentiating between a changethat was not an innovation and a change that was suffi-ciently large to count as an innovation. Similar problemswere reported in the pre-survey testing conducted forthe MEPIN survey.14 Examples given by the intervieweesduring the cognitive testing of changes that were not inno-vations included minor changes such as an upgrade to awebsite; changes to routine activities that are part of thebranch’s usual business such as producing reports, incre-mental changes, branch restructuring, and the introduction
of new business systems. However, other intervieweesreferred to each of these examples as an innovation or citedother very minor changes as innovations, such as revisionsto the content of a newsletter. One interviewee commented
14 The MEPIN survey conducted focus groups and interviews to deter-mine how potential respondents understood innovation concepts. Theresults are provided by Annerstedt and Bjorkbacka (2010), Gunnarssonand Oddsdottir (2009) and Mortensen and Bloch (2009).
onomic Dynamics 27 (2013) 146– 159 155
that it was particularly difficult to differentiate betweenchange, which was a constant feature of Government, andinnovative change.
Several managers, particularly those responsible forresearch or policy development, had a difficult time dif-ferentiating between a change that is part of the branch’score business and an innovation. For example, some inter-viewees that were responsible for a branch that producesreports on legal or research topics believed that everyreport is an innovation.
Several interviewees commented that we needed toask a question that differentiated between incremen-tal and ‘transformative’ innovations. Their definition ofa transformative innovation was one with far-reachingeffects that completely changed processes or a service. Forinstance, interviewees described ‘transformational’ changeas ground-breaking ways of doing things, or ‘new ideas thatpush the boundaries’. A question on transformative changewas introduced in the phase 3 cognitive interviews. Onlyone of the ten interviewees in phase 3 believed that theirbranch had introduced a transformative innovation. Thisquestion was not included in the survey because of con-cerns that it had not been adequately tested, but it doesprovide an option for further research and testing.
After asking each person to give their own definitionof an innovation in the public sector, the intervieweeswere shown definitions for process, service, communica-tion, policy and goods innovations, with between five andtwo different options listed under each type. The questionon goods was combined with the question on services inphase 3.
Many of the interviewees reported several types ofinnovations and multiple options within each of the fiveinnovation types. Although based on a small sample,male interviewees reported more types of innovation thanwomen. After excluding innovation types that were onlyused in one phase of the cognitive interviews, questions on16 types of innovations were included in all three phases.Men reported an average of 4.5 innovations comparedto 3.6 reported by women. This difference is not due tomale interviewees being responsible for larger branches,with more numerous opportunities for innovation. In fact,after excluding the outlier branch with 640 employees, thewomen interviewees were responsible for an average of51.1 employees compared to 31.4 employees for men. Thedifference could be a chance effect of the small samplesize, but the interview notes also suggest that women gavemore thought to the issue of whether or not a ‘change’ wassufficiently large to count as an innovation.
All interviewees were asked to describe several of theirreported innovations. Their descriptions show that theirselection of multiple sub-types and even of different maintypes of innovations often refers to different aspects ofthe same innovation. This is partly because they find itdifficult to differentiate between innovation types. Thisissue was noted by Hartley (2005) and found in the MEPINinterviews, with respondents finding that ‘many innova-
tions contained more than one element’. In Scandinavia, theproblem was particularly frequent for product and processinnovations and for communication and product innova-tions.
1 e and Ec
mpcpano
oieiiiivbi‘tTticic
tvddbwoiiibeps
sttcbrnotoi
oigftr
chy, including elected Ministers. This figure is considerablylower than the rate of 43% reported in the APSC (2011)employee survey, suggesting that employees at lowerjob levels within the public service hierarchy credit the
56 A. Arundel, D. Huber / Structural Chang
Process, service and communication innovations sharedany similarities in the minds of the interviewees. Exam-
les of policy innovations that were internal to the agencyould also be described as processes. For instance, a newolicy to permit working from home could be describeds an innovation in how work responsibilities are orga-ized. Goods innovations were frequently viewed as a typef service.
These results suggest that survey questions on the ‘typef innovation’ can tell us if a public sector branch or agencynnovates on a ‘yes’ or ‘no’ basis and the diversity of theffects of each innovation type, but these questions arenaccurate measures of the prevalence of discrete types ofnnovation. Based on our results so far, it would be mislead-ng to classify branches of the Australian public service bynnovation type. For example, a ‘service and process inno-ator’ category would not imply that the branch introducedoth a service and a process innovation. The cognitive test-
ng results suggest that the correct interpretation is that aservice and process innovator’ has introduced innovationshat encompass both service and process characteristics.hese could be different innovations or the same innova-ion. The same interpretation applies to Fig. 3. The correctnterpretation is not that 40.4% of branches introduced aommunications innovation, but that 40.4% of branchesntroduced an innovation that had the characteristics of aommunication innovation.
One option to obtain more accurate data on specificypes of innovations is to ask respondents to assign inno-ations to only one of the five main innovation types. Thisoes not appear to be a tenable solution because respon-ents already find it difficult to classify their innovationsy type. Furthermore, some examples of innovations thatere provided during the interviews clearly spanned two
r more categories. For instance, a process innovation tomprove the efficiency of data collection ended with a usernterface on a government website, with the users consist-ng of businesses and citizens. From the perspective of theranch, this innovation was a process that improved thefficiency of data collection and reduced costs. From theerspective of a business or citizen, it was an improvedervice that was simpler and faster to use.
The cognitive testing results do not suggest that wehould abandon questions on different types of innova-ions, since both the interviewees and the respondentso the survey gave clear descriptions of process, service,ommunication and policy innovations with substantialenefits. Also, interviewees want the questionnaire toeflect their own innovative activities, as noted above. Ifot, they seek the closest option, for instance a managerf a branch that specializes in policy development is likelyo attempt to fit its policy innovations into the questionsn process innovations if there are no questions on policynnovations.
Goods innovations were the least frequently cited typef innovation in the interviews, with only a third of thenterviewees reporting one. None gave an example of a
ood as their most important innovation. The survey resultsound that 21.3% of branch managers reported an innova-ion involving a ‘tangible good or software’, but only 2.3%,eported earning revenue from sales, patents or licenses to
onomic Dynamics 27 (2013) 146– 159
products and services. None of the interviewees or surveyrespondents described a goods innovation as their branch’smost important or successful innovation. The most fre-quently reported most successful innovations in the surveywere processes (cited by 64% of respondents), followed bycommunication (13%), policy (12%) and product and service(4%) innovations.
Finally, many of the interviewees were confused overthe meaning of ‘implementation’, in the sense that an inno-vation must have been implemented or introduced. This isbecause many of the innovations that were described bythe interviewees had a long implementation period withno clearly defined date of introduction. In addition, some ofthe innovations that undergo a long period of incrementalimprovements and implementation are shared across oneor more branches or agencies, thus increasing developmenttime. Similar issues were raised in the MEPIN interviews.
Several interviewees were also reluctant to excludeinnovations that were under development but not yetimplemented, since they were often the innovations thatwere foremost in their minds. Seven of 54 respondentsthat provided written comments at the end of the surveystated that the questionnaire should include innovationsthat were under development.
4.4. The source of innovations
The issue of who provides the idea for an innovationis relevant to the innovation rate, since an increase in thenumber of potential sources of ideas could increase thenumber of good ideas and consequently the share of inno-vative branches. The innovation rate could be lower if ideasare largely sourced from the limited number of upper man-agement personnel and ministers than if ideas are sourcedfrom the larger number of lower level staff and middlemanagement.
Fig. 6 gives the source of the idea for the branch’s mostsuccessful innovation. The most common source for thisinnovation was the branch’s ‘leadership group’, cited by55.2% of the branch managers. Only 13.3% of managers citedsources that were higher up in the public service hierar-
Fig. 6. Source of the idea for the branch’s most successful innovation,percent of 145 responses.
A. Arundel, D. Huber / Structural Change and Economic Dynamics 27 (2013) 146– 159 157
Table 3Distribution of person months to develop the most successful innovation by the source of the idea.
Idea source N Person months to develop the innovation
Less than 6 months 6 months to 1 year 1 year or longer
or more person-months to develop, while 40.7% reportintroducing a ‘first to Australia’ innovation. A third mea-sure of a substantive innovation, for a ‘first to Australia’innovation that was of sufficient value to be later adopted
Higher management and ministers 17 29.4%
Middle and upper-middle management 75 40.0%
Lower level staff 26 57.7%
Minister’s Office with greater innovation influence thansenior management.
Fuglsang (2010) suggests that innovations sourced fromupper level management are likely to be more substantivethan ideas sources from front line staff. Table 3 comparesthe time required to develop the most successful inno-vation by three sources of the idea for the innovation.Although the differences by the source of the idea are notstatistically significant, possibly due to the small samplesize, they indicate that innovations suggested by upperlevel managers require more time to develop than innova-tions suggested by lower level staff (p = 0.146). For example,41.2% of the innovations for which the idea came fromMinisters or higher level managers required 12 or moremonths to develop, compared to 37.3% of innovations forwhich the idea came from the leadership group and 30.8%of innovations where the idea came from lower level staff.Conversely, 57.7% of the innovations proposed by lowerlevel staff required less than 6 months to develop, versus40.0% of ideas from the leadership group and 29.4% of theideas from higher management.
In the interviews, only two of the 27 examples of mostimportant innovations were based on legislative decisionsor ideas suggested from higher up in the hierarchy. Thepost-survey interviews delved more deeply into why theideas for most innovations appeared to come from middlemanagement or non-managerial staff. The answer is due toa combination of how government policy is implementedand how managers interpret the concept of innovation.
The post-survey interviewees agreed that the Govern-ment, Ministers or their Department head create the needfor innovation and sometimes provide ideas for an inno-vation. Yet, they consistently did not view a Ministerialdirective as an innovation because these instructions weregenerally seen as insufficient. As one interviewee stated,“we get ideas and instructions for innovations from above,but we do not see this as an innovation, partly because thesechanges are discussed in terms of ‘policy development’ or‘policy reform’. The branch’s task is to work out how toimplement these innovations.” Another interviewee notedthat “the requests for change through government direc-tives come from higher up, but . . . the innovation comesthrough the methods developed to meet them”. The con-sistent theme from the post-survey interviews is that thebranch is responsible for developing an implementablesolution in response to the requirements of legislation or aMinisterial directive. In the words of another interviewee,“top down ideas could be an innovation, but how the ideas
are applied is also an innovation”.
The interviewees appear to give priority to ideas thatlead to solutions, rather than general ideas that providea guideline for the types of innovations that are required.
29.4% 41.2 100%22.7% 37.3 100%11.5% 30.8 100%
This could underestimate the role of elected Ministers andParliament. However, the branch managers did not dis-miss the role of higher levels in the public sector hierarchyin innovation and often actively welcomed it because itsolved the problem of taking responsibility in what wasoften perceived as an environment with a ‘low tolerancefor mistakes’. An interviewee noted that if an idea for aninnovation ‘comes from above it provides an immediatemotivation – if from below there is a fear of failure.’
5. Conclusions
The results of the interviews and the survey haveimplications both for interpreting surveys of public sec-tor innovation and for the design of future internationallycomparable surveys on this topic.
Innovation in the public sector has often been viewedas an oxymoron, with many scholars assuming that it israre, due to a lack of incentives and a risk-averse attitude ofsenior managers in public organizations. Recent large-scalesurveys in Europe (Arundel and Hollanders, 2011; Buggeet al., 2011) have shown that contrary to expectations,innovation in the public sector is widespread, with over80% of public sector managers reporting at least one inno-vation in the preceding two years. The current study findsan innovation rate of 91.3% for Australian public adminis-tration branches. As these innovation rates are higher thanthose in the private sector, they raise concerns that publicsector managers may not understand the concept of inno-vation, leading them to report many minor changes thatare neither novel nor substantive.15
The results of 37 interviews with public managers inAustralia and a survey with 344 respondents show thatAustralian public sector managers largely understand theconcept of innovation, although some of the interview-ees are confused over the difference between a changethat is substantive enough to count as an innovation andone that is not. This confusion may result in some over-reporting of innovation, but two measures of substantiveinnovations show that there is still a considerable amountof innovation activity. For instance, 60% of branch man-agers report implementing an innovation that required six
15 An alternative explanation is that it is not public sector managersthat are over-reporting innovation, but private sector managers areunder-reporting innovation. Empirical evidence for under-reporting inthe private sector is provided by Arundel et al. (2013).
1 e and Ec
brcvo
oiseclipmoie
mdatsmssdapciTfifo
mapbrrtttuce
cawifisbifii
58 A. Arundel, D. Huber / Structural Chang
y a different Australian government organization, waseported by 13.7% of branch managers. Although much lessommon, these types of innovations are likely to be veryaluable or they would not have attracted the interest ofther managers.
These results show that concerns over the reportingf minor changes as innovations can be addressed byncluding questions in innovation surveys to identify sub-tantive innovations, either in terms of the amount offfort expended on innovations, the development of uniqueountry-first innovations, or valuable innovations that areater adopted by other government organizations. Thenterviews found that public sector managers are capable ofroviding high quality answers to questions on the persononths expended on innovation activities. An additional
ption that is worth exploring is to develop questions fordentifying ‘transformative’ innovations with far-reachingffects.
The high innovation rates indicate that public sectoranagers are capable of innovating in what appears to be
ifficult conditions, such as a risk-averse environment and lack of market-mediated financial incentives. Assump-ions about the public sector being risk averse and lackinguitable incentives are either misleading or public sectoranagers are able to innovate within these constraints. We
uspect that the latter may be the case, since the interviewshow that public sector managers that face such constraintso innovate. Kay and Goldspink (2012) present data from
separate interview study in Australia that suggest thatublic managers solve the problem by working within theonstraints of a ‘no-failure’ environment by developingnnovations more carefully than their private sector peers.his may explain why our interviewees often have a dif-cult time determining when the implementation phase
or an innovation ends – it could be a long slow process inrder to allow time to identify and correct problems.
Branch managers report that 13.3% of the ideas for theost important innovations come from higher up the hier-
rchy – either high level managers or Ministers. This isartly due to the perspective of the respondents, withranch managers interpreting an innovation as the workequired to solve a problem or implement a new policy,ather than the directive for a new policy. Surveying minis-ers or agency heads would probably give more importanceo the role of ministers and high level management. Never-heless, as the work to solve problems largely comes frompper-middle management and staff lower in the hierar-hy, the public sector is likely to benefit from policies thatnsure the collection and evaluation of staff ideas.
To the best of our knowledge, this is the first survey toollect information from upper-middle level branch man-gers that are responsible for a functionally coherent unitithin a larger Agency or Department. This level could be
deal for innovation surveys because the manager has suf-cient responsibility to develop innovations plus he or shehould be aware of both ideas for innovations that mightubble up from below and ideas that are suggested or
mposed from above. This level may be particularly suitableor research to address the growing interest in the role ofncentives and how to encourage innovation ‘from within’n the public sector (Windrum, 2008).
onomic Dynamics 27 (2013) 146– 159
We also find that the characteristics of the respondent(time in their current position and possibly gender) andthe branch (years since established and function) influenceinnovation outcomes and consequently this informationshould be collected in future surveys. In addition, toimprove comparability and increase the relevance of thesurvey to respondents, public sector questionnaires needto include questions on innovations that are underwaybut not yet completed and on a range of different typesof innovations, including policy innovations. The findingthat respondents often cite different types of innovationsfor the same innovation influences how results should bepresented and interpreted.
Finally, this research is relevant to ongoing researchby the OECD and others to develop guidelines for col-lecting internationally comparable data on public sectorinnovation. Although our data are only for Australia, wefind it encouraging that many of our interview results,such as difficulties in determining when a change isand is not an innovation or how to interpret the con-cept of implementation, were also found by the MEPINresearchers in their interviews of public sector managersin Scandinavia. This suggests that there may be strongsimilarities in the perspectives of public sector managers,at least for those drawn from high-income developedeconomies.
Acknowledgements
The research discussed in this paper was funded bythe Australian Department of Industry, Innovation, Sci-ence, Research and Tertiary Education (DIISRTE) as partof the Australian Public Sector Innovation Indicators(APSII) project. We would like to thank Antonio Balaguer,Louise Talbot, and Nick Yazidjoglou for their commentsand support during this project, as well as the supportand suggestions of Tony Cotton of the Australian PublicService Commission and Diane Braskic of the AustralianBureau of Statistics. The opinions expressed in this paperare solely those of the authors and do not necessarilyrepresent the views of the Australian Government or DIIS-RTE.
Appendix A. Supplementary data
Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.strueco.2013.06.009.
References
ABS (Australian Bureau of Statistics), 2012. 8167.0 – Selected Character-istics of Australian Business, 2010–2011.
Annerstedt, P., Bjorkbacka, R., 2010. Measuring Public Innovation in NordicCountries: Toward a common statistical approach. Feasibility study ofPublic Sector Organizations. Statistics Sweden, February.
APSC (Australian Public Service Commission), 2011. State of the Service
Report: State of the Service Series 2010–2011. Commonwealth ofAustralia, Canberra.
Arundel, A., Hollanders, H., 2011. A Taxonomy of Innovation: How do Pub-lic Sector Agencies Innovate? Inno-Metrics. European Commission,October.
Arundel, A., Smith, K., 2013. History of the community innovation survey.In: Gault, F. (Ed.), Handbook of Innovation Indicators and Measure-ment. Edward Elgar, Cheltenham, pp. 60–86.
Arundel, A., O’Brien, K., Torugsa, A., 2013. How firm managers under-stand innovation: implications for the design of innovation surveys.In: Gault, F. (Ed.), Handbook of Innovation Indicators and Measure-ment,. Edward Elgar, Cheltenham, pp. 88–108.
Audit Commission, 2007. Seeing the light. In: Innovation in local publicservices. Local government national report, London.
Bartlett, D., Dibben, P., 2002. Public sector innovation and entrepreneur-ship: case studies from local government. Local Government Studies28, 107–121.
Borins, S., 2000. Loose Cannons and rule breakers, or enterprising leaders?Some evidence about innovative public manager. Public Administra-tion Review 60, 498–507.
Borins, S., 2001. The Challenge of Innovating in Government. Innovationin Management Series. PricewaterhouseCoopers Endowment for TheBusiness of Government, Arlington.
Borins, S., 2010. Innovation as Narrative. Ash Center for Democratic Gov-ernance and Innovation. Harvard Kennedy School, Cambridge.
Brudney, J.L., Hebert, F.T., 1987. State agencies and their environments:examining the influence of important external actors. Journal of Poli-tics 49, 186–206.
Bugge, M., Mortensen, P.S., Bloch, C., 2011. Measuring Public Innovationin Nordic Countries: Report on the Nordic Pilot Studies. Analyses ofMethodology and Results. MEPIN/NIFU, Oslo.
Collins, D., 2003. Pretesting survey instruments: an overview of cognitivemethods. Quality of Life Research 12, 229–238.
Damanpour, F., Schneider, M., 2006. Phases of the adoption of innova-tion in organizations: effects of environment, organization, and topmanagers. British Journal of Management 17, 215–236.
Damanpour, F., Walker, R.M., Avellandeda, C.N., 2009. Combinative effectsof innovation types and organizational performance: a longitudinalstudy of service organizations. Journal of Management Studies 46,650–675.
Drucker, P.F., 1985. Innovation and Entrepreneurship. Harper and Row,New York.
Earl, L., 2004. Technological Change in the Public Sector, 2000–2002.Statistics Canada, catalogue no. 88F0006XIE, no. 008. StatisticsCanada, Ottawa.
Fuglsang, L., 2010. Bricolage and invisible innovation in public serviceinnovation. Journal of Innovation Economics 1, 67–87.
Gunnarsson, T., Oddsdottir, E.E., 2009. National Feasibility Study inIceland. RANNIS, Rejkjavik.
Hartley, J., 2005. Innovation in governance and public services: past andpresent. Public Money and Management 25, 27–34.
Hughes, A., Moore, K., Kataria, N., 2011. Innovation in Public Sector Orga-nizations: A Pilot Survey for Measuring Innovation Across the PublicSector. NESTA, London, March.
Kay, R., Goldspink, C., 2012. What Public Sector Leaders Mean When They
Say They Want to Innovate. Incept Labs, Sydney.
Koch, P., Hauknes, J., 2005. On Innovation in the Public Sector, PublinReport No. D20. NIFU-STEP, Oslo, http://thelearningnetwork.net/Downloads/Library/PUBLIN-publicsectorinnovation.pdf (accessed3.6.2013).
onomic Dynamics 27 (2013) 146– 159 159
Laegreid, P., Roness, P.G., Verhoest, K., 2011. Explaining the innova-tive culture and activities of state agencies. Organization Studies 32,1321–1347.
Lonti, Z., Verma, V., 2003. The determinants of flexibility and innovationin the government workplace: recent evidence from Canada. Journalof Public Administration Research and Theory 13, 283–310.
Moon, M.J., de Leon, P., 2001. Municipal reinvention: Managerial valuesand diffusion among municipalities. Journal of Public AdministrationResearch and Theory 11, 327–351.
Mortensen, P.S., Bloch, C., 2009. Results of Part 2 of the National FeasibilityStudy in Denmark, December.
Mulgan, G., Albury, D., 2003. Innovation in the Public Sector, http://webarchive.nationalarchives.gov.uk/+/http://www.cabinetoffice.gov.uk/upload/assets/www.cabinetoffice.gov.uk/strategy/pubinov2.pdf
National Accounting Office (NAO), 2006. Achieving innovation in centralgovernment organizations. National Accounting Office (NAO), London.
OECD, 1992. OECD Proposed Guidelines for Collecting and InterpretingTechnological Innovation Data: Oslo Manual. OECD, Paris.
OECD, 2012. http://stats.oecd.org/glossary/detail.asp?ID=2199OECD/Eurostat, 2005. Oslo Manual: Guidelines for Collecting and Inter-
preting Innovation Data. OECD, Paris.Palmer, I., Dunford, R., 2001. The diffusion of managerial innovations: a
comparison of Australian public and private sector take-up rates ofnew organizational practices. International Public Management Jour-nal 4, 49–64.
Parna, O., von Tunzelmann, N., 2007. Innovation in the public sector: keyfeatures influencing the development and implementation of techno-logically innovative public sector services in the UK, Denmark, Finlandand Estonia. Information Polity 12, 109–125.
Potts, J., Kastelle, T., 2010. Public sector innovation research: what’s next?Innovation: Management, Policy and Practice 12, 122–137.
Roessner, J.D., 1977. Incentives to innovate in public and private organi-zations. Administration and Society 9, 341–365.
Rubalcaba, L., Windrum, P., Gallouj, F., Di Meglio, G., Pyka, A., Sundbo,J., Webber, M., 2013. The Contribution of Public and Private Ser-vices to European Growth and Welfare and the Role of Public-privateInnovation Networks, ServPPIN Final Published Summary Report.www.servPPIN.com
Thenint, H., 2010. Innovation in the Public Sector, Global Review of Inno-vation Intelligence and Policy Studies, PRO INNO Europe. EuropeanCommission, Brussels.
Vanagunas, S., Webb, J., 1994. Administrative innovation and thetraining of public managers. Public Personnel Management 23,437–445.
Walker, R.M., 2006. Innovation type and diffusion: an empirical analysisof local government. Public Administration 84, 311–335.
Willis, G., 1994. Cognitive Interviewing and Questionnaire Design: ATraining Manual. National Center for Health Statistics: CognitiveMethods Staff.(Working Paper no. 7).
Windrum, P., 2008. Innovation and entrepreneurship in public services.In: Windrum, P., Koch, P. (Eds.), Innovation in Public Sector Services,.
Edward Elgar, Cheltenham.
Wu, J., Ma, L., Yang, Y., 2010. Innovations in contemporary chinese publicsectors: myths and realities. In: Symposium Edition on Reform andTransition in Public Administration Theory and Practice in GreaterChina, University of Hong Kong, 4–6th February 2010.
