Public sector innovation in context: A comparative study of innovation types

ABSTRACT

The purpose of this article is to analyse the role of national context in public sector innovation. Whilst there is a growing literature on innovation types in the public sector, prior studies have analysed data from a single country. Consequently, there is an incomplete understanding of the national context. Our comparative study examines 108 innovations from Italy, Japan and Turkey. The first stage of our empirical evidence uncovers a divergent configuration of innovation types. Further analysis of how national context was constructed by the innovators allows us to provide evidence for a national-context framework for public sector innovation.

Introduction

Public sector innovation (PSI) has attracted a growing level of interest amongst scholars, resulting in a substantial body of knowledge accruing over the past two decades (e.g. Osborne and Brown 2011; Korac, Saliterer, and Walker 2017; Borins 2018; Qiu and Chreim 2021; Wirtz, Kubin, and Weyerer 2021). However, context, particularly national context, remains a nebulous concept in the PSI literature (Demircioglu 2020) despite its crucial role in public administration (Pollitt 2011, 2013). Significantly, De Vries, Bekkers, and Tummers’s (2016) observation in their systematic review is still valid: There are very few international comparative studies and those that exist are from a particular context, namely the European Union (e.g. Arundel, Casali, and Hollanders 2015; Lapuente and Suzuki 2020). This results in a failure to shed light on variations across different national contexts. In response to the Demircioglu’s (2020) call for cross-national comparative studies, the purpose of this article is to conceptualize the role of national context for PSI.

Innovation types now form a significant stream of research (Walker 2006; Wu, Ma, and Yang 2013; Chen, Walker, and Sawhney 2020) and can serve as a suitable proxy to analyse the role of national context in PSI. These innovation types reflect distinct dimensions of innovation, and therefore enable researchers to conceptualize a common understanding of PSI (Walker 2006). Nonetheless, the innovation types literature also suffers from contextual limitations. Firstly, a limited number of contexts, particularly the USA, EU countries and Australia, have been studied empirically (De Vries, Bekkers, and Tummers 2016). Second, a limited number of the studies have examined the influence of context on the innovation types (e.g. Walker 2006; Korac, Saliterer and Walker 2017), and those that have been conducted based their analysis on the organizational environment within a single country. Hence, these studies were unable to evaluate, compare and systematically test the influence of national context, where the investigation of the national context necessitates a cross-national comparative analysis (Pollitt 2013). Third, an emphasis on quantitative analysis of contextual indices as variables (e.g. Damanpour and Schneider 2009) forms another limitation of this research stream. The investigation of the role of context requires further qualitative analysis, since the context is interpretively constructed and subjectively situated by individuals and groups (Bevir, Rhodes, and Weller 2003). For these reasons, the role of context in PSI and innovation types has remained ambiguous and can be viewed as a missing link (Pollitt 2013, xv) within the public administration field. This informs the rationale for the present research.

Our research examines data from the United Nations Public Service Awards (UNPSA), following the approach adopted by Borins (2000b, 2014), Wu, Ma, and Yang (2013) and Chen, Walker, and Sawhney (2020). First, we employ content analysis to quantitatively study the extent to which the innovation types are divergent across the countries. Subsequently, thematic analysis is used to qualitatively identify the constructed contexts where innovation types emerged. The principles and perspectives introduced by the seminal collective work edited by Pollitt (2013) are adopted, in order to understand the role of context and how the context was articulated within the applications for the innovation awards.

This article provides three main contributions. First, we provide a rare international comparison for innovation types literature (Wu, Ma, and Yang 2013; De Vries, Bekkers, and Tummers 2016) and uncover that they are divergent in three understudied contexts of PSI: Italy, Japan and Turkey. Second, we construct a unique contextual framework to reflect the role of national context in PSI types building upon the fragmented literature on context in public management literature (e.g. Osborne and Brown 2011; Pollitt and Bouckaert 2017; Borins 2018; Scott 2021). We provide empirical evidence via identifying the elements of national context through the qualitative interpretations of the applicants, which prior single country studies have been unable to examine (e.g. Walker et al. 2010; Korac, Saliterer, and Walker 2017). Our conceptual framework provides clarity for future researchers studying PSI by identifying the role of administrative, temporal, political, social and economic & technological contexts. Third, our analysis revealed the importance of intertwining administrative, political and temporal contexts, as theoretically proposed by Pollitt (2013).

The remainder of the article proceeds as follows. First, we review the literature on innovation types in the public sector and define the analysed types. Second, we pull together the fragmented literature on context in PSI and public management (Pollitt 2013; Christensen and Lægreid 2013) to define and conceptualize context. Third, we describe our research methodology. Subsequently, we quantitatively investigate the types of award-winning PSI through content analysis. Finally, following the identification of divergent innovation types, we explore qualitatively the role of national context via thematic analysis and a comparative discussion.

Types of public sector innovation

PSI is defined in this article, as the adoption, creation or development of ideas, objects and practices that are new to the unit of adoption (Chen, Walker, and Sawhney 2020). The innovations differ in their characteristics as they provide solutions to complex problems (Korac, Saliterer, and Walker 2017). Thus, innovation types are vital to constructing a common understanding of PSI. A variety of innovation types have been proposed and studied since the 1990s. Table 1. summarizes our review of empirical studies examining innovation types and shows the literature lacks a commonly accepted classification. De Vries, Bekkers, and Tummers’s (2016) recent systematic review can serve as a useful reference point. They identified five innovation types: New service, administrative process, technological process, conceptual and governance innovation. However, systemic innovation to establish novel ways of interactions between organizations as well as citizen networks (Windrum 2008), was overlooked in their typology. Similarly, social innovation, that attempts to solve complex and wicked social problems (Mulgan 2006), was not included. Thus, we extend De Vries, Bekkers, and Tummers (2016)`s typology and include these two additional types. The following section briefly defines these seven innovation types (Appendix A provides the coding book for each of these innovation types).

Table 1. Empirical innovation types studies and contextual factors.

Study	Theory and Methods	National Context	Innovation Types	Context Factors
Walker (2006)	Adoptionof Innovation Innovation typology Quantitative methods Local governments	ENGLAND	Product Process Ancillary	Political disposition (proportion of Labour votes)	Changes in the social, political and economic context • Service need	Diffusion determinants: Public pressures: external sources Diversity of service need
Damanpour and Schneider (2009)	Adoptionof innovation Innovation typology Quantitative methods Local governments	USA	Administrative	Urbanization Community wealth	Population Growth Unemployment rate
Walker (2007)	Adoptionof Innovation Innovation typology Quantitative methods Local governments	ENGLAND	Product Process Marketization Ancillary	Service need Diversity of need Increasesin population	Turbulent external political climates	An organization’s perceived ability to influence the environment
Damanpour, Walker, and Avellaneda (2009)	Resource-based view Innovation typology Quantitative methods Local governments	ENGLAND	Service Technological process Administrative process	Urbanization	Service need Indexof Multiple Deprivation	Service diversity-Ethnic Diversity
Wu, Ma, and Yang (2013)	Innovation typology Content analysis Bestpractice research Provincialand local governments	CHINA	Service Technological Management Collaborative Governance	Level of economic development	Level of administrative unit
Arundel and Huber (2013)	Innovation measures Innovation typology Quantitative methods Federal agencies	AUSTRALIA	Product-service Process Communication Policy	The study did not include any contextual variables
Gonzalez, Llopis, and Gasco (2013)	Strategic management Innovation typology Quantitative methods Local governments	SPAIN	Product Process Collaborative	The study did not include any contextual variables
Bloch and Bugge (2013)	Innovation measures Innovation typology Quantitative methods Central, regional, local govs	DENMARK, NORWAY, SWEDEN, FINLAND, ICELAND	Product Process Organizational Communication	The study did not include any contextual variables
Arundel, Casali, and Hollanders (2015	Innovation measures Innovation typology Quantitative methods Central, regional, local govs	EU COUNTRIES, NORWAY, AND SWITZERLAND	Service Organizational Process	Economic context
Walker et al. (2015)	Resource-based view Innovation typology Quantitative methods Local Governments	ENGLAND	Service Process: marketization, organizational, and technological Ancillary	Capacity: ○Political ○Economic ○Social variables ○Community	Complexity: ○Concentration ○Heterogeneity diversity	Dynamism ○Uncertain socioeconomic context, ○Uncertain political context
Bugge and Bloch (2016)	Innovation typology Content Analysis Central, regional, local governments	DENMARK, NORWAY, SWEDEN, FINLAND, ICELAND, UK	Product Process Organizational Communication Bricolage Systemic	The study did not include any contextual variables
Torugsa and Arundel (2016)	Total innovativeness Innovation typology Quantitative methods Federal government	AUSTRALIA	Policy Service Service Delivery Administrative Conceptual	The study did not include any contextual variables
Korac, Saliterer, and Walker (2017)	Resource dependency Innovation typology Quantitative methods Local governments	AUSTRIA	Organizational Service Marketization Ancillary	Public pressures: ○Citizen demands ○Media pressures	Vertical pressures: central government mandate	Contextual Pressures: social, economic, and political changes competition
Chen, Walker, and Sawhney (2020)	Innovation Typology Public Value Content Analysis Best Practice Research Federal, State, Local govs	USA	Mission Policy Management Partner Service Citizen	The study did not include any contextual variables

(i) Service innovations form the first type and are closely related to service provision to public users. A novel service can be delivered to a group of existing users (Walker 2008). The provision of existing services to a new user group is also regarded as a new service innovation (Osborne 1998).

Process innovations emerge in two forms. Firstly, (ii) Administrative process innovations refer to the creation of new ways, methods and forms of undertaking tasks within the organization. These innovations are closely related to redesigning operational routines (De Vries, Bekkers, and Tummers 2016). Secondly, (iii) Technological process innovations involve the application of technology to operational activities and service delivery mechanisms (Walker 2008), which can range from digital forms to automated decision-making by algorithms (Cinar 2021).

(iv) Conceptual innovation is `the development of new world views that challenge assumptions that underpin existing service products, processes and organizational forms` (Windrum 2008, 9).

(v) Governance innovations introduce new participation mechanisms for citizens and novel ways to increase transparency and accountability within the public sector (De Vries et al. 2015). They are distinct from systemic innovations because of their political nature to democratize the public sector (Wu, Ma, and Yang 2013).

(vi) Systemic innovations capture `new or improved ways of interacting with other organizations and knowledge bases` to co-deliver public services and is a consequence of increasing interaction between PSOs, civil society and businesses (Windrum 2008, 9). This definition is similar to ancillary innovations (Korac, Saliterer, and Walker 2017) or collaborative innovation (Wu, Ma, and Yang 2013), which require cross-boundary activities outside the PSO.

(vii) Social innovations, introduced by Mulgan (2006) into the field of PSI, as a broad and cross-sectoral concept, aim to meet social needs of disadvantaged groups and target underlying reasons of social problems (Voorberg, Bekkers, and Tummers 2015).

It is worth noting that a single innovation may be categorized into more than one innovation type as a novel initiative can introduce a hybrid combination of different types (De Vries, Bekkers, and Tummers 2016; Torugsa and Arundel 2016). This is critical to capture the multidimensional and complex nature of PSI.

Context in public sector innovation

Context is defined as ‘broader social or normative environments (including various dimensions of organizational or national culture, or unit or organizational climate) … industry-, sector-, or economy-wide characteristics, as well as other normative and institutional structures and regimes’ (Bamberger 2008, 840). Following Christensen and Lægreid (2013) our study conceptualizes context as the circumstances, background or settings, which have the potential to clarify, or influence a phenomenon. Pollitt (2013, xviii) positioned context as ‘a missing link’, in public administration, that is, ‘something that enables us to understand the different evolutions of public policy and management in different habitats’.

The context-related factors in PSI literature have been referred to as environmental (Bernier, Hafsi, and Deschamps 2015), contextual (Korac, Saliterer, and Walker 2017), and external (Walker et al. 2015). Those studies (see Table 1) that analysed a limited number of contextual variables focused on the organizational environment. Thus, the existing literature provides limited understanding of the role of national context in the configuration of innovation types in the public sector. We construct a conceptual framework (Figure 1) from the broader and fragmented public management, PSI and innovation management literatures to understand the dimensions of national context:

Figure 1. Conceptual framework to study the role of national context on PSI.

1. The administrative context is related to the structure of public administration in the country and machinery of central, regional and local government (Hantrais 1999). We discuss three. First, how the power and resources are shared among the level of governments describes centralized or decentralized administrative context (Pollitt and Bouckaert 2017). Decentralized administrative settings are considered more suitable for co-creation oriented social and governance innovations (Massey and Johnston 2016; Ferlie 2021). Second, the coordination mechanisms between level of governments range from highly coordinated to highly fragmented at each extreme. The variety of institutional arrangements for sub-national governments (e.g. regions, prefectures, municipalities, metropolitan areas, etc.) result in highly fragmented contexts (Pollitt and Bouckaert 2017). Innovations to improve horizontal and vertical coordination can be impacted by this administrative context dimension (Casula 2019). Third, a specialized central institution to coordinate the innovation activity in the government (Mele 2010) forms an enabling dimension of the administrative context. Finally, models of public management (Traditional Weberian Model, New Weberian State, New Public Management and New Public Governance) (McMullin 2021) are elements of administrative context to enable PSI (Hartley, Sørensen and Torfing 2013; Pollitt and Bouckaert 2017). These models prescribe how public services should be delivered and prioritize different mechanisms (e.g. bureaucracy and rule of law vs. modernization and legitimacy vs. markets and businesses principles vs. collaboration and open networks). These models are also conceptualized in a variety of ways such as public governance paradigms (Torfing et al. 2020) or public management reform models (Pollitt and Bouckaert 2017) which indicates their interplay with political and temporal contexts.

2. Economic and technological context is related to macroeconomic features and technological capabilities of the country. Innovation management literature and global innovation indexes both identify a significant relationship between national economic development and national technological capabilities and innovative outputs (e.g. Freeman 1995, WIPO 2021). The PSI literature notes different mechanisms for organizational resource availability, which can have a direct link with macroeconomic conditions: An economic recession imposes challenging conditions addressed by the innovation such as cutbacks and austerity (Pollitt and Bouckaert 2017; Van der Voet 2019) or economic growth provides enabling conditions for the innovation such as slack resources (Salge 2011; Bernier, Hafsi, and Deschamps 2015). The level of technological advancement in a country also constitutes a context factor for PSI. For instance, technology readiness in Japan, Singapore and New Zealand facilitated cloud-based health innovations (Raghavan, Demircioglu, and Taeihagh 2021). Borins (2001, 729) found that developing Commonwealth countries were “concentrating on leapfrog, rather than leading edge, information technology” to fill the technological gap in public services. However, this context can interplay with the administrative and political contexts to justify the status quo in traditional bureaucracies (Scott 2021, 20). Finally, considering the collaboration between private sector and PSOs in innovative e-government projects (Szkuta, Pizzicannella, and Osimo 2014), we argue the national technological context for PSI includes the technological capability of private firms as well.

3. The political context reflects the party political nature of the public services, which necessitates decisions to allocate resources (Mergel 2019). These political decisions bounded by power relations (Osborne and Brown 2011) form the authorizing environment of the public value triangle where politicians can facilitate innovation at different levels of government (Sorensen et al. 2021). Whilst the recent literature focuses on the role of local political contexts (e.g. Torfing and Ansell 2017), there has been evidence for the meso-level political context such as U.S. states (Roberts and King 1991). Further, the national governing party and its political ideology forms a context for PSI. On the one hand, the political context where a strong leader has a political change agenda can facilitate innovation. For instance, Scott (2021) pointed out that the political context during the term of the first selected head of Hong Kong government, Tung Chee-hwa, drove PSI. On the other hand, Borins (2018) argues that the radical populism by the U.S. President Trump constitutes an hostile political context for PSI. We should note that the scope of these political context interplays with the administrative context as it is closely related to the level of administrative centralization. Finally, the international political relations is an element of the political context. It is evident that knowledge scanning PSOs benefit from international networks, visits and conferences (Arundel, Casali, and Hollanders 2015) which necessitates enabling good international political relationships. In addition, international institutions such as OECD and UN as well as supranational governments such as European Commission (that has more authoritative powers than the formers) have been facilitating PSI and novel public management ideas (Arundel, Bloch, and Ferguson 2019). The political willingness to engage with these international institutions can serve as enabling condition for PSI.

4. The social context consists of socio-demographic factors, such as ageing or booming population, urbanization, crime, migration, and social inequalities (Pollitt and Bouckaert 2017). First, on the one hand, the ageing population in developed countries led to rising social expenditure as well as immigration from developing countries (Acik, Trott, and Cinar 2021) all of which compose a challenging social context for PSI to address (Demircioglu and Vivona 2020). On the other hand, the booming young population and the social inequalities are the parts of the challenging social context in developing countries that PSI attempts to respond to (Cornet and Barpanda 2020). Secondly, the population density in cities has been regarded as an external antecedent driving PSI (Walker 2014). Urban cities and local areas constitute diverse social contexts for PSI (Cornet and Barpanda 2020; Ye et al. 2021). Thus, the level of national urbanization can influence PSI as a social context. Finally, complex social problems embedded in the wider society such as juvenile delinquency, organized crime and corruption are challenging social context conditions for PSI (Aagaard 2012).

5. The temporal context forms the dynamic intertwining time dimension of all the contexts above (Ongaro, Gong, and Jing 2021) as these contexts are typically path-dependent (Bamberger 2008; Pollitt 2013). The majority of the organizational-level PSI studies reviewed in Table 1. adopted cross-sectional analysis, which considers the context to be stable at a single point in time. Walker (2007) is a notable exception, having studied innovation types longitudinally and found that previous innovations can lead to different types of innovation over time. This reveals both the generative and complementary nature of innovation types, and the influence of the temporal context. There is additional evidence that PSOs follow temporal trajectories in innovative search (Salge 2012). We argue this is valid for the temporality of national context as well. Both (i) Long periods of reform (e.g. innovation policy) and (ii) significant events (e.g. financial crisis, natural disasters and public scandals) can trigger a variety of innovations. Whilst the first as an enabling condition is related to “path-dependency” (Pollitt and Bouckaert 2017), “evolutionary changes” and learning (Autio et al. 2014), the latter “chance events” (Pollitt and Bouckaert 2017, 239) challenge “deeply embedded … assumptions, expectations and values” (Pollitt 2013, xvii) where innovation can solve the policy gridlock.

Methodology

Data

The data for this study is retrieved from the applications submitted to UN Public Service Awards (UNPSA). United Nations initiated Public Service Awards in 2003 and it has been the sole global award scheme. The scheme aims to reward innovative public sector projects worldwide. UNPSA accepts broad categories of innovation initiatives, from improving the delivery of public services and knowledge management to gender responsive delivery of public services and preventing corruption (United Nations 2015). Thus, the application forms can serve a suitable proxy to study PSI countries from different regions of the world.

The UN used a nomination form particularly focused on the innovation process in the period of 2009 to 2017. After 2017, the UN changed the nomination form significantly and directed the focus of the competition to Sustainable Development Goals. To study cases submitted in a stable procedure across the years we focused on the period of 2009–2017.¹ These applications include rich qualitative data and form the basis of our study. The nominations were submitted via an online structured form in English to respond to questions qualitatively about the innovation process: Problem, solution, idea sources, innovation strategy, stakeholders, resources, obstacles and outputs (see Appendix B). Applications were evaluated through three rounds. In the first round, nominations were assessed and ranked by UN experts. Applications above a certain threshold passed to the second round, where additional supporting documents were assessed. In the third round, a United Nations Committee of Experts in Public Administration decided the first and second place winners for each region (United Nations 2015). The UN openly provides the application forms short-listed for the second round of evaluation between 2007 and 2021² for the benefits of practitioners and academicians.³ We studied the complete population of application forms selected for the second round between the years of 2009 and 2017 (108) from: (i) Italy (40 innovations), (ii) Japan (27 innovations) and (iii) Turkey (41 innovations).⁴ De Vries, Bekkers, and Tummers (2016) identified that most PSI studies have been conducted in USA and EU and that cross-country data collection was also very scarce. This informed our rationale to select three different settings in Europe, Eurasia and East Asia. Our unit of analysis is a single innovation rather than organizational innovativeness, as suggested by Arundel, Bloch, and Ferguson (2019). Each studied case is enumerated with the country abbreviations (e.g. TR01, IT01, JP01)

Methods

We utilized a sequential approach for the analysis of 108 open-questionnaire forms. As a first step, we coded the types of innovation through the pre-defined coding book (see Appendix A) in accordance with quantitative content analysis adopted by prior studies examining innovation types (e.g. Wu, Ma, and Yang 2013; Chen, Walker, and Sawhney 2020). To test the reliability of the coding, four cases from each country were randomly selected, with a total of 12 cases being coded independently by two researchers. The inter-coder agreement through the Holsti co-efficient was over 0.9 which is a commonly accepted measure of reliability within the literature (Neuendorf 2002, 149). Subsequently, the lead author coded each of the applications and the results were discussed in regular meetings.

After the identification of the divergent innovation typology (see Table 3), we attempted to explore the role of national context. The emphasis of a number of prior studies has been on the success of New Public Management (De Vries, Bekkers, and Tummers 2016). Consequently, internal and managerial variables have largely been measured through perception questions, where respondents can link their success to themselves and the organization, by contrast, contextual variables have primarily been explored using quantitative indices from secondary data (e.g. Demircioglu 2020; Torugsa and Arundel 2016, 2017). Hence, these studies suggest that contextual factors have a weaker influence (e.g. Walker 2006; Damanpour and Schneider 2006; Damanpour, Walker, and Avellaneda 2009). We argue that the influence of context should be investigated qualitatively because the meaning and influence of context is constructed interpretively and situated subjectively by individuals and groups. Thus, their qualitative articulations can uncover the role of context (Bevir, Rhodes, and Weller 2003). We coded openly any text that referred to, and provided information about, the pre-conditions leading to an innovation. Most of this information was contained within the first question: `What was the problem before the implementation of the initiative?`. Our coding of the applications also identified pertinent evidence for context within Questions Four and Six, which addressed the initiation of the innovation and funding, respectively (see Appendix B). A qualitative thematic analysis approach was adopted, following Braun and Clarke (2006) and Xing, Liu, and Cooper (2018). We identified first-order concepts in detail and this subsequently led to second-order themes. We aggregated the themes, which resulted in categories as dimensions of national context (see Table 5 and Table C1).

Methodological criteria to study contexts

Pollitt (2013, 94) underscored the need for contextual analysis in public administration literature, yet pointed out a methodological concern to view context as ‘a kitchen sink into which any old thing, big or small, could be thrown’. Thus, he proposed methodological insights to study contexts: First, context(s) should be defined, theorized and operationalized. Thus, we develop a conceptual framework and provide empirical evidence to theorize context. Our analysis is operationalized through a macro level analysis of country contexts. Second, contexts are multiple and intersecting. The national context on its own is oversimplified and to avoid this we conducted a detailed coding of the context, whilst paying particular attention to identifying interlayering contexts. Third, contexts can be restrictive where these conditions can either lead PSI to address the challenges, or provide enabling conditions to drive them (Ongaro, Gong, and Jing 2021). Therefore, we analyse how applicants articulated the problem that led to each innovation. Fourth, there is a need to consider the mechanisms and processes that animate contexts: According to Pollitt (2013, 420), ‘these process and mechanisms … occur in sufficiently similar forms to facilitate comparisons across a range of local contexts’. It is a well-accepted mechanism that faced with complex problems governments seek to innovate (Torfing and Triantafillou 2016). Thus, our analysis can animate the role of context in PSI classification. Finally, comparison can play a valuable role in the analysis of context. Yet the compared objects require justification, which we will explain in detail within characteristics of the countries below.

Limitations

Whilst our data and analysis enabled us to design an international comparison to understand the role of national context, we acknowledge three main limitations that are worthy of note: First, our analysis is context dependent and limited by the information provided by the applicants. Thus, we do not claim generalizability across other countries. Second, potential problems of selection bias should be acknowledged. It has been debated that applicants may attempt to report minor changes and improvements as innovation (see Bretschneider, Marc-Aurele, and Wu 2004). To avoid this, the Oslo Manual (OECD 2018) suggests an object-oriented approach (data collection for single innovations) rather than subject-oriented approach (data for organizational innovativeness), which our study follows. UNPSA applicants were also made aware of inspections by independent committees, judges as well as requests for additional evidence. In addition, self-nominations were not permitted over the studied years. Each challenged their ability to portray minor changes as innovation in order to be selected.

The third limitation is the extent to which the awards applications can reflect PSI in the relevant context (Borins 2014). We identified several procedures used by the UN, which can address this. The categories of the UNPSA have been coherent with the broad variety of the recent global public management trends, technological change and social challenges, which are closely related to diverse PSI activity. UNPSA also calls for all levels of government for nominations and we checked that the coded cases for this research can reflect the historical innovation activity by the respective level of governments, as discussed below. The UN also collaborated closely with national governments to organize the nominations (see Formez PA News 2010). Further, we identified that UN and national governments announced the competition through official channels to make organizations aware (see TRT News 2015; Formez PA News 2010; Ministry of Foreign Affairs of Japan 2010). We argue that these factors enable two necessary conditions for national level cross-country comparison, considering the governments in each country could prioritize distinct combinations of PSIs (Pollitt and Bouckaert 2017): (i) The nominations have the potential to reflect the country-level PSI activity, (ii) They have the potential to reflect the divergence of cross-country innovation agenda. Considering these limitations and our checks to remedy them, our findings should be regarded as exploratory and thus need to be tested further via different data and contexts.

Presentation of results

This section will first provide a discussion on the characteristics of Italy, Japan and Turkey. Subsequently, the configuration of the innovation types based on our content analysis will be presented. Then, the cross-country analysis based on the typology configuration and thematic analysis will provide evidence for the five concurrent national context elements.

Characteristics of Italy, Japan, and Turkey

As Pollitt (2013) advised, a cross-country analysis requires similarities as a base for comparison, as well as differences, to identify contextual divergences. The three countries under analysis possess similarities on the one hand: Each country is a well-established member of the OECD with large economies, mature public administration structures and a relatively large population (OECD 2020) and has not frequently been studied for PSI research. Secondly, the model of public management in all countries has been based on Weberian bureaucracies and put the state at the centre (see Lapuente and Suzuki 2020). However, all three have introduced significant modernization and e-government oriented public sector reforms over the last three decades (Furukawa 1999; Mele 2008; Natalini and Stolfi 2012; Sezen 2016; Kim 2017), which has the potential to develop innovation capacity in some parts of the central and local governments (Pollitt and Bouckaert 2017). On the other hand, each country shows variations in terms of economic development, administrative structures, societal problems and reform agenda (See Table 2). Italy is an EU member and has a developed economy but has suffered from economic and social problems in recent decades. However, the administrative tradition is different from continental Europe and Italy represents Southern European public administration. Italy has aimed to decentralize its unitary state for along time (Ongaro 2009).

Table 2. Characteristics of Italy, Japan and Turkey.

	Italy	Japan	Turkey
Region	Southern Europe	Asia	Eurasia
OECD Membership Year	1962	1964	1961
Administrative Context	Path-dependently centralistic and unitary. Fragmented administrative structure, due to the recent administrative reforms. A traditional Weberian approach which puts the state and the bureaucracy at the centre. The recent modernization efforts were regarded as a New-Weberian State model.	Path-dependently centralistic and unitary. The boundaries between central and local governments have become unclear due to complex tax and employment relationships. Path dependent attempts to shift from Weberian governance model to NPM and then to NPG.	Strongly Centralized. A traditional Weberian approach which puts the state and the bureaucracy at the centre, although the bureaucracy is highly politized and clientelist. There has been attempts to introduce NPM reforms due to EU membership process. The e-government reforms aim to modernize the state.
Political Context	Frequently changing prime ministers and governing party. Europeanization politics.	Strong prime ministers governing for long periods. The political nature of strong prefecture governors. Asian and Pacifist regional politics based on trade.	Strong Prime Minister and governing party engaged with the EU membership process between 2004 and 2012.
Economic andTechnological Context	High income group country but highly impacted by the 2008 crisis. Historically high-level of knowledge and technology outputs by SMEs and corporations.	High income group country but highly impacted by the 2008 crisis. Relatively highest level of knowledge and technology outputs by large corporations.	Upper middle-income group country. Sustained economic growth since 2000s. Low level of knowledge and technology outputs. Dependent on technology transfer.
Social Context	Social inequalities across regions. Ageing population. Immigration from developing countries. Historical organized crime problem.	Ageing population. Domestic population movements to industrialized cities and urbanization.	Social inequalities across regions. Relatively young population. Domestic population movements to industrialized cities and urbanization.
Temporal Context	Top-down policy based e-government and NPM reforms from 1990s initiated by the central government.	E-government and NPM reforms were adopted by governors and mayors from 1990s and diffused by administrative guidance by central government. A recent shift to NPG through the same mechanism.	Top-down policy based e-government and modernization reform from 2000s initiated by the governing party.

Source: OECD (2020) and Literature.

Japan is the most developed and technologically advanced economy among our three countries, yet has faced economic challenges over recent years. Japan differs significantly to Italy and Turkey concerning its e-government advancement (The World Bank Group 2021) as of 2011. Whilst Japan has had a centralized state structure, the role of local governments has changed significantly in recent years with 70% of expenditure now made by local governments (Suzuki, Dollery, and Kortt 2021). Turkey has been an EU membership candidate with a strong central government tradition and introduced connected reforms since 2000. Economic development is comparatively the poorest in Turkey, but it has shown significant economic growth since 2002 (Sezen 2016), which has made Turkey a distinctive context in Eurasia.

Our initial coding revealed that the central-local government composition of submitted applications can reflect PSI activity in each country considering the context discussion above. Japanese applications were submitted mainly by prefectures and municipalities (70%), which can reflect its changing administrative structure. Conversely, 78% of Turkish cases were from central PSOs, which serves as a suitable proxy to represent the strong unitary characteristic of public administration. Finally, innovations from central governments accounted for 56% of Italian cases. This more balanced composition can reflect the centralistic feature of Italian public administration, as well as the fragmented administrative structure because local government innovations were diverse from municipalities, regions and municipality unions. Thus, following prior research utilizing awards application forms (e.g. Borins 2014; Chen, Walker, and Sawhney 2020) we argue that the applications from Italy, Japan and Turkey can serve as a suitable data source to explore PSI and reflect the PSI activity in the selected countries.

Innovation types configuration and qualitative cross-country comparison

Table 3 provides detailed information on each innovation type in the divergent innovation configuration, as well as providing a comparison with Wu, Ma, and Yang (2013) study and De Vries, Bekkers, and Tummers’s (2016) systematic review. First, our coding of application forms reveals divergent configurations in the innovation types across the three countries studied. Technological process innovations, such as the Biomedical Research Workflow Management System in Italy and UYAP (National Judiciary Informatics System) in Turkey, were the most frequent innovation in both of these two countries. In Japan, a broader variety of innovation types was evident. Governance innovations aimed at citizen participation, such as The Green Toshima Revival Project, formed the most common type. In comparison to Wu, Ma, and Yang (2013) and De Vries, Bekkers, and Tummers (2016), the findings uncovered more technological process innovations and fewer administrative process innovations in all countries, but more governance innovations in Japanese cases.

Table 3. Typology of innovations as a percentage of total cases and comparison with previous studies.

Innovation Types	Italy (40)		Japan (27)		Turkey (41)
	Number of Occurrences	%	Number of Occurrences	%	Number of Occurrences	%
PROCESS	27	67.5%	9	33.3%	34	82.9%
Technological Process	21	52.5%	9	33.3%	29	70.7%
Administrative Process	7	17.5%	0	0.0%	5	12.2%
Systemic	12	30.0%	9	33.3%	21	51.2%
Governance	14	35.0%	14	51.9%	2	4.9%
New service	5	12.5%	6	22.2%	7	17.1%
Social	5	12.5%	7	25.9%	3	7.3%
Conceptual	3	7.5%	6	22.2%	4	9.8%

The identification of this divergence further contributes to our rationale for investigating the role of the national context. To accomplish this, as a further level of analysis, we explored the articulation of contextual factors prior to the initiation of innovation through thematic analysis. Table 4 shows the first-order concepts, second-order themes and aggregate dimensions. Table C1 provides evidence quotes, reflecting each of the context-related articulations. Table 5 summarizes comparatively how they can influence the configuration of innovation types. The following section provides a cross-country analysis and detailed information about the content of innovation types and the dimensions of each context in the relevant countries.

Table 4. Data structure.

Table 5. The cross-country configuration of contexts and most common innovation types.

	Contexts					Innovation Types
	Administrative	Economic & Technological	Political	Social	Temporal
ITALY	Fragmented by large central ministries, new regions and small municipalities. A combination of NPM and NWS to modernize the bureaucracy and to gain legitimacy and citizen trust.	Low level of technology, economic recession and austerity	Transition of central governments. Local councils	Immigration and crime	2008 financial crisis and reform period since 1990s	1. High frequency of technological process innovations to introduce first generation of e-government reforms (From manual to digital) 2. Governance innovations to gain the destroyed citizen trust to governments. 3. Systemic innovations to coordinate fragmented central, regional, municipal governments.
JAPAN	Central governments, prefectures and small municipalities under a chain of unitary administrative control. A combination of NWS and NPG for citizen participation and legitimacy.	High level of technology, economic recession despite the past economic growth, Importance of highly skilled businesses	Strong elected political executives and central government audit	Ageing population andurbanization	2011Earthquake, 2008 financial crisis	1. Governance innovations to ensure citizen participation along with the legitimacy of elected local political executives. 2. Technological process innovations to deal with big data, social media and coordination of first generation e-government reforms 3. Systemic innovations to coordinate small municipalities under the prefectures and integrate businesses/citizens to the first generation e-government reforms 4. Local social innovations to deal with unemployment, urbanization and economic recession
TURKEY	Highly centralized. NPM to modernize bureaucracy, economic gains for businesses, clientelist,citizen satisfaction but strongly path-dependent to Weberian.	Low technology level and income, but sustained periodic economic growth	Strong political party and prime minister	Immigration and social inequalities	Unique reform period since 2002	1. The dominance of technological process innovations to introduce first generation of e-government reforms. (From manual to digital) 2. Systemic innovations to coordinate periphery organizations of central large organizations

Italy

In Italian cases, our coding revealed that process innovations formed significantly the most common type in the studied years. This reflects the enabling conditions of a long period of e-government reforms that began more than 25 years ago, which included the transfer of paper-based tasks to digital ones plus second-generation reforms such as digital data exchanges. This was an interplay between administrative, political and temporal contexts in the e-government reform inspired by the NPM model to modernize the bureaucracy that was path-dependent to the classical Weberian model:

‘According to the fundamental principle of the Italian National e-Government Plan, Municipalities since 1997 are devoted to support’ (CaseIT29). In Italy, during that period, public administration branches were increasingly encouraged and supported to implement digital data exchanges thanks to specific legislation on the subject (CaseIT12).

The second most frequent type was governance innovations that were facilitated by the administrative, political and temporal contexts: This was a subsequent temporal reform agenda for transparency in 2000s that Pollitt and Bouckaert (2017) labelled the NWS public management reform model. This agenda was frequently promised by changing central governments under the EU`s political influence (Ongaro 2011). Within the administrative context, the Ministry for Public Administration emerged as the institution facilitating these transparency reform activities as proposed by Mele (2010). The articulation below illustrates these clearly:

The Prime Minister Decree of 2 October 2008 transferred the functions to the Department for Public Administration (DPA) within the Presidency of the Council of Ministers … . Transparency, prevention of corruption, personal/collective responsibility have been seen as tools to enhance democracy, participation and efficiency in the public administration and to ensure greater trust from citizens and businesses … The initiative is part of the overall Italian reform of the public administration designed by the Minister for Public Administration and Innovation since May 2008 (CaseIT28).

In addition, the deteriorating economic context was frequently revealed in governance innovations as it interacted with other contexts. Similarly, organized crime and social inequalities in Southern Italy, which represent significant aspects of social context were also reported within governance innovations as challenging conditions addressed by the innovation. The following articulation illustrates this:

the particular vulnerability of local governments with respect to the mafia’s attempts to penetrate the economy and particularly public procurement, due to the availability of significant liquidity and the companies under mafia control, as well as attempts to influence sectors of the legal economy which until then had not been involved in criminal interests (CaseIT18).

Finally, our coding revealed systemic innovations as the third most common innovation type to coordinate larger central ministries, fragmented regions and municipalities. The changing nature of unitary administration, and the shift from centralization to decentralization, were expressed as the challenging administrative context of these systemic innovations, as articulated below:

the great number and heterogeneity of Italian administrations (more than 20,000, spanning from small local authorities to big ministries with more than 20,000 employees), that together with the amount of different transparency obligations, represented a sort of barrier for the effective use of the published information by the final users (CaseIT17).

Japan

The configuration of innovation types in Japanese cases is more balanced than for Turkey and Italy. Governance innovations were the most frequent type and focused on citizen participation. These emerged within the enabling NPG model administrative context following partly adopted NPM reforms and changing role of local governments, which interplayed with the challenging economic and temporal context of the financial crisis and austerity. This is consistent with the findings of Kudo (2015). The following quotes illustrates this well:

Those problems have been the resident’s needs which cannot be satisfied by only the existing administrative services and structures … we could establish systems that encourage continued participation of the residents (JP21) With the further degeneration of the financial situation of the central government and the municipalities, the point has been reached where approaches with originality and ingenuity were necessary to help create a local, community-based society conductive to raising children; it was under this background that the Baby Stations were formed. (JP17).

Furthermore, the interplay of administrative and political context of powerful elected governors and mayors also enabled governance innovations while they were seeking legitimacy for these initiatives. The sub-national governments are unique in Japan (Jain 2011; Ishihara 2021) and they sought opportunities for legitimacy & power through novel initiatives. This is captured in the quote below:

Governor X who was a well-known TV performer before taking office as a governor, frequently emphasized the importance of the policy-market research at press briefings, executive meetings and budget negotiation process in the government (CaseJP14).

Second, lower frequencies of technological process innovations can be attributed to the economic and technological, as well as the political and temporal contexts. The e-government transformation had started almost 30 years ago in 1994 (Masujima 2005). Despite the slow start, it grew rapidly and the historical enabling economic and technological context of influential corporations, such as Sony, played a significant role in this catch-up (Jain 2002). Thus, in contrast to Turkey and Italy, Japan had progressed first-generation e-government reforms prior to the studied period. Indeed, Japan was dealing with the integration of the databases, big data and social media analysis during the studied period. The well-known Ministry of Economy, Trade and Industry (METI) provided the enabling administrative context conditions for this new generation of technological process innovations. The latent content of the quote below provides evidence for these enabling contexts:

Simultaneously, the Japanese government (METI) will create the information exchange framework at a national level and implement a standardization approach.(JP03) METI E-Government team was implementing the text analysis trial project at the time.(CaseJP11)

In addition, systemic innovations surfaced within the increasingly decentralized administrative context, with a frequent aim of integrating services of local governments. This is coherent with the strong administrative control by upper governments in Japan (Masujima 2005). The quote below can illustrate this administrative context:

Approximately 200 local agencies provide support service for SMEs including consultation, sessions, and business matching etc. (CaseJP01) it is essential to set up the standardized format of public service information at both schema level and contents level. The latter is required to develop standardized format for detailed information on public services which are provided by municipal governments (CaseJP03)

Finally, the Japanese cases reported the most frequent number of social innovations, where the social context of urbanization and the ageing population were articulated frequently as challenging conditions, which is consistent with recent findings (Suzuki and Avellaneda 2020). Apart from this social context, the economic recession as the temporal context and the local growing responsibilities of municipalities were manifested as the challenging administrative context producing social innovations as illustrated below:

Due to the economic recession .in 2008, the number of unemployed families on welfare in X Prefecture rose .a 270% increase … The Prefectural Social Welfare Council declared the necessity of the prefectural government to implement a self-support project for assisting welfare recipients and provide support for children from poverty-stricken families (CaseJP07).

Turkey

In Turkey, our coding revealed mainly process innovations and systemic innovations. Technological process innovations formed the most frequent innovation type. This is consistent with the temporal, political and administrative context of the government’s enabling e-Transformation modernization agenda since 2003. The coded cases revealed a concentrated attempt to fill the technological gap in public services by introducing first-generation e-government reforms to transform manual jobs into digital ones in a short period of time. This shows also the role of the economic & technological context: The enabling high economic growth rate, alongside a challenging low level of technological capability in Turkey, compared to Japan and Italy. The articulations below exemplify these influences:

The Government of Turkey is aware of the outstanding needs and decided to modernize the X … . The E-government initiative, of which the X is a central part, is one of key government priorities … . have moved from paper-based to computer-based X and X systems. (CaseTR22) The Office of the Prime Minster started an undertaking to select E-Government projects that could be finished within one year without any additional cost. (CaseTR35)

Another enabling role of the temporal context surfaced when previously implemented successful innovations were articulated as an explanation for further technological innovations in a path-dependent way:

After the nationwide implementation of X System … , the IT Department started working on this project for an online auctioning system for the sales of seized property (CaseTR17).

The second most common type, systemic innovations to interact with other organizations, emerged as the result of the challenging administrative context of strong central-unitary government in need of coordination between central and periphery organizations. A significant number of these were supported by technological process innovations. The discourse for these innovations was characterized by coordination and a national scope. State Planning Organization (DPT) was articulated within the enabling administrative context to select and fund these extensive innovations in Turkey.

As Sezen (2016) noted the EU membership process (as a political and temporal context) played an enabling role in reforms. These were frequently articulated in the applications:

The overall objective was to improve the administrative capacity and efficiency of the tax administration in line with the fiscal blueprints and to align the Turkish tax system with the EU directives (CaseTR24).

However, Turkey is seemingly far from EU membership and related reforms frozen (Ciddi 2018). Thus, we can clearly understand how the intertwining political context (ruling populist government) and administrative context (the pragmatic utilization of NPM inspired public management model that is path-dependent to centralized Weberian model) triggered technological process innovations and systemic innovations to increase efficiency, service quality and economic gains to maintain status-quo instead of increasing the transparency and citizen participation. This differentiates Turkey from Italy and points to the variation of the political, administrative and temporal context in two countries and justifies the different configuration of innovation types.

Similarly, citizen participation was constructed from a clientelist view to receive votes from the citizens. This acted as a challenging political context for social and governance innovations. The sole initiative presented as a citizen participation mechanism by a municipality was essentially giving flowers, tickets and other pragmatic benefits to citizens and this approach was expressed honestly in the application:

After home visits, 1.800 families declaring their desire to take their children to theatre have been gifted with tickets to a children’s theatre play. … We have organized a yacht trip to 1.180 persons that could not enjoy the Bosporus joy so far”(CaseTR28).

Conclusions

This article is the first to examine comparatively the role of national context on PSI types. We synthesized fragmented theoretical insights on context in public administration and management (Osborne and Brown 2011; Pollitt 2013; Virtanen 2013; Pollitt and Bouckaert 2017; Borins 2018; Scott 2021) and we constructed a framework to conceptualize national contexts. Our research studied 108 innovations from Italy, Japan, and Turkey to provide evidence for the conceptual framework. Our findings provide three main contributions to research and practice. First, we provided a rare cross-country comparison of innovation types and have identified a divergent PSI configuration in three under studied countries. Specifically, Turkish and Italian cases reported a dominance of process innovations, whilst Japanese cases showed a more balanced distribution with an emphasis on governance innovations. These findings differ from recent studies in other contexts (e.g. Wu, Ma, and Yang 2013; De Vries, Bekkers, and Tummers 2016)

Second, we analysed qualitative articulations of conditions leading to innovation to identify the role of national context on innovation types. This provides a significant methodological and theoretical contribution to the literature (e.g. Pollitt and Bouckaert 2017; Lapuente and Suzuki 2020; Ongaro, Gong, and Jing 2021) as it allowed us to analyse how the role of context was constructed by the applicants. This contrasts with prior studies, which have measured context via indexes, or limited perception-based quantitative survey questionnaires (e.g. Korac, Saliterer, and Walker 2017; Walker et al. 2015). The empirical insights revealed detailed five national contexts as challenging conditions addressed by the innovation or enabling conditions for the innovation: (i) administrative context, (ii) economic and technological context, (iii) political context, (iv) social context and (v) temporal context. This conceptual framework is noteworthy as it prevents the context from being ‘a kitchen sink into which any old thing, big or small, could be thrown’ (Pollitt 2013, 94). This also builds upon the De Vries, Bekkers, and Tummers (2016) and Demircioglu (2020)’s call for the investigation of the effects of the national context on PSI.

Third, the analysis also revealed the intertwining nature of contexts as proposed theoretically by Pollitt (2013). It uncovered the role of temporal, administrative and political contexts interlayering on a temporal political reform path, which is closely related to models of public management (Hartley et al. 2013; Pollitt and Bouckaert 2017). Specifically, technological process innovations emerged within the NPM centred e-government and modernization reform period in Turkey and Italy. Similarly, governance innovations in Italy surfaced on the transparency reforms paths of NWS model. These articulations also indicated the role of the long-term reform history. Significantly, survey questions (e.g. Arundel, Casali, and Hollanders 2015; Clausen, Demircioglu, and Alsos 2020), solely addressing the influence of the short-term political mandate within the last year or the last 3 years to measure the role of top-down influence were not able to capture this contextual influence.

Future research directions

The contributions and limitations of this study lead to four areas for future research. First, PSI literature in general and innovation types in particular needs more cross-country research. In addition, single country studies also need to elaborate the national context factors. Second, the role of temporal context in PSI should be investigated in detail and for different periods. Historical institutionalism (Kuhlmann 2010) can serve as a theoretical base to investigate the mechanisms of how PSI can emerge in the path dependent temporal context while it intersects with other contexts. Third, our research provides exploratory findings on the differences between centralized and decentralized state structures. Centralized national contexts may produce different innovation types, as identified in the Turkish cases. Thus, this difference should be operationalized and investigated in future studies. Related to this, the role of the political context needs more academic attention. Unfortunately, PSI faces the realities of populism and authoritarian pressures on democratic values (Borins 2018). How this political context can influence PSI requires further attention. Finally, the purpose of our analysis was to explain the role of context in innovation types, and our findings did not identify any evidence for the national cultural context (e.g. Hofstede cultural dimensions). We propose that future studies on the innovation process dynamics, such as innovation barriers or individual innovative behaviour, should investigate the role of national cultural dimensions on PSI.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental data

Supplemental data for this article can be accessed at https://doi.org/10.1080/14719037.2022.2080860.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Emre Cinar

Emre Cinar (Ph.D., University of Portsmouth is a Senior Lecturer in Innovation Management at the Business School, University of Portsmouth. His research interest includes the barriers to public sector innovation, collaborative innovation and the impact of the context on public sector innovation.

Christopher Simms

Christopher Simms (Ph.D., University of Portsmouth) is a Professor in Innovation Management. He completed his PhD, examining the development of new products and packaging within the FMCG Industry. Chris has had a number of articles published in Research Policy, European Journal of Marketing and R&D Management.

Paul Trott

Paul Trott is Professor of Innovation Management at the Business School, University of Portsmouth. He received his Ph.D. from Cranfield University. He has published over fifty articles on innovation management; these include articles in R&D Management, Research Policy, Technology Analysis & Strategic Management, Technovation, International Journal of Innovation Management.

Mehmet Akif Demircioglu

Mehmet Akif Demircioglu is an assistant professor at the Lee Kuan Yew School of Public Policy, the National University of Singapore. He is also a research fellow at the Institute for Development Strategies at Indiana University and the Center for Organization Research and Design (CORD) at Arizona State University. His research interests are public sector innovation, employee attitudes, social media, and public sector reform. His work has published or forthcoming in several journals including Research Policy, Public Administration, Public Management Review, Governance, The American Review of Public Administration, Government Information Quarterly, Industrial and Corporate Change.

Notes

1. 2691 applications were selected to the second round from over 108 countries between 2009 and 2017.

2. UNPSA was not organized in 2016.

3. Please see https://publicadministration.un.org/en/Research/Case-Studies/unpsacases.

4. The most recent application was analysed if an innovation was resubmitted over years.