Introduction

In 2018, Policy Sciences celebrates its 50th anniversary and recognises the five most cited articles from the last five decades. Colin J. Bennett and Michael Howlett’s 1992 article, ‘The lessons of learning: reconciling theories of policy learning and policy change’, is one of these articles. This paper reflects on how the article has been used in learning literature.

Bennett and Howlett (1992) offer a review of the literature on policy change based on different conceptualisations of learning. Although the then existing conventional explanations of policy change explained state actions and policy change as a product of social pressures and conflicts (i.e. ‘conflict-based explanations’), they recognise the emerging importance of the role of ideas and learning in explaining policy change (i.e. ‘learning-based approaches’). They are also concerned about conceptual ambiguity, under-theorised and fragmented learning literature in the learning subjects, objects and outcomes that constrain the understanding and appreciation of cumulative knowledge and empirical evidence with lack of comparisons and integration in the literature. Specifically, based on five competing conceptualisations of learning and its role in public policy formulation, namely ‘political learning’ (Heclo 1974), ‘policy-oriented learning’ (Sabatier 1987, 1988), ‘lesson-drawing’ (Rose 1991), ‘social learning’ (Hall 1988, see also Hall 1989) and ‘government learning’ (Etheredge 1981), they identify different actors, different types of learning and different effects of learning. Thus, they aim to reconcile the theories of policy learning and policy change

…by accepting that (a) that learning is in fact a complex, multi-tiered phenomenon which can affect either decision making organizations and processes; specific programs and instruments used to implement policy; and/or the ends to which policy is developed, ‘and (b) that the agent of each type of learning will be different. (Bennett and Howlett 1992: 289)

The main contributions of Bennett and Howlett (BH) 1992 article to the literature on policy learning as a source of policy change are threefold. First, they offer one of the most widely accepted definitions of learning in the literature. They define learning as ‘the commonly described tendency for some policy decisions to be made on the basis of knowledge of past experiences and knowledge-based judgments as to future expectations’ (Bennett and Howlett 1992: 278). In doing so, they take a step forward towards overcoming the problem of definitional ambiguity of the concept. Second, they identify three components of learning, which include ‘who learns, what is learned, and what effects on resulting policies emerge as a result of learning’ (278). These analytical questions deal with ‘the complexity of the subjects, objects and effects of learning’ (290). More importantly, their policy learning framework privileges the role of actor to help explain policy change resulting from policy makers’ ‘general increase in knowledge about policies’ especially that of policy design (288–289). Third, they distinguish three types of learning, which include government learning by ‘state officials’ learning about public policy making process and generating ‘organisational change’; lesson-drawing by ‘policy networks’ about ‘policy instruments’ and generating policy ‘programme change’, and social learning by ‘policy communities’ about policy ‘ideas’ and generating policy ‘paradigm shifts’ (Bennett and Howlett 1992: 289). In doing so, they link three types of learning with three types of actors, subjects of learning and change. Thus, the principal importance of the article is in its conceptual framework that integrates the three types of learning and policy and organisational change by unpacking the different dynamics and outcomes that relate to varieties of learning. Unsurprisingly, parallel to accelerating academic interest in approaches to policy learning and change, Fig. 1 shows the increasing amount of interest in the BH article over the years. Citations increased sharply over the last few years, with peaks in 2016 and 2017.

Fig. 1
figure 1

Source Compiled from Thomson and Reuters Web of Knowledge’s SSCI

Number of citations per year (n = 242).

Full size image

If ‘citations provide a valid and reliable indication of the scientific interaction between researchers and research [organisations]’ (Kraus et al. 2014: 277), then it is legitimate to offer citation data on BH (1992) and the discussions it has contributed to. Thus, this survey presents a brief citation analysis that investigates citing journals and authors to illustrate the academic impact of the 1992 article.

The remainder of the paper first describes the methodology used in carrying out the citation analysis. It then presents the dataset, offers its analysis and discusses the findings. The conclusion summarises the main findings.

Method

The method used in this paper is a citation analysis that investigates citing journals and some of the articles that are built on BH 1992. There were three criteria employed. First, the focus has been on peer-reviewed journals because these are regarded as confirmed knowledge (Podsakoff et al. 2005). The exclusion of books and book chapters is due to variations in the peer review processes (Jones et al. 2011). Second, articles selected had ‘policy learning’ and ‘policy change’ as the topic (i.e. appearing in the title, keyword or abstract). Third, the coding criteria (see "Appendix") included author, year, title, journal, journal country and category, location of citation (abstract, introduction, literature review, body, conclusion and appendix) and type of manuscript (conceptual, empirical or literature review). In doing so, this paper aims to examine various citation features of the 1992 article, and discuss its influence on the top most cited 13 articles that cite it at least three times.

Dataset

This study used Thomson and Reuters Web of Knowledge’s Social Science Citation Index (SSCI) database as the source of citation data in this paper. This is because it is one of the most comprehensive databases of peer-reviewed social sciences journals published in English. The search covered articles published between January 1992 and October 2017. As of October 2017, there were 5713 articles in English recorded in the Web of Knowledge SSCI with ‘policy learning’ and ‘policy change’ as the topic. The BH (1992) article has been the 17th most cited article (or in the top 3%) among these articles. In terms of document types, the ISI Web of Knowledge listed 242 articles, 20 proceedings papers, 15 reviews, six editorial materials and one book review. This paper focused on the content of 13 articles that cited BH at least three times. In doing so, it excludes light citations.

Descriptive data analysis

The impact of the BH article has been felt beyond the public policy field (Fig. 2). More than half of the articles were published in the public administration (116) and political science (97) categories in the SSCI. This was followed by environmental studies (50), planning and development (33), where public policy journals are listed, and environmental sciences (24).

Fig. 2
figure 2

Source Compiled from Thomson and Reuters Web of Knowledge’s SSCI

Types of academic fields (cited publications, number and per cent).

Full size image

As Fig. 3 shows, the majority of citations come from journals based in the UK (58%), the Netherlands (21%) and the USA (14%). The geographical distribution shows that almost 82% of the journals are based in Europe. This is due largely to renewed interest in European integration studies on analysing the ‘micro processes in EU decision making’ and ‘[p]ragmatic considerations following enlargement and a change in preferences … in favor of networks and related learning-driven instruments… have made learning a major theme on the EU academic and political agendas’ (Zito and Schout 2009: 1116).

Fig. 3
figure 3

Source Compiled from Thomson and Reuters Web of Knowledge’s SSCI

Geographical distribution of journals in original dataset.

Full size image

Figure 4 shows the ranks of the top ten journals with most citations of the BH article. Based on the SSCI impact factor, most of these journals are top or highly respected journals, indicating a high level of academic importance. As anticipated, most of these are European journals and, except for Forest Policy and Economics, they are listed in the research fields of political science, public administration and/or public policy.

Fig. 4
figure 4

Source Compiled from Thomson and Reuters Web of Knowledge’s SSCI

Top ten citing journals of the BH article in the original dataset.

Full size image

Discussion

Beyond the general assessment of descriptive citation data, this section focuses on the nature of these citations to gauge how the learning literature made sense of the learning type framework. For reasons of space and in order to concentrate on articles citing the BH article the most, this section excludes the bulk of articles that cited the 1992 article only once or twice. Thus, the focus has been on the 13 articles in the learning literature that cited this article at least three times (see “Appendix”). At the methodological level, all of the empirical papers used a qualitative case study method. This was because they were interested in time and context-specific who, when, how and why questions that relate to learning, rather than questions focusing on what. Whether the citations in these articles are located in the introduction, body or conclusion reveals the utilisation of the 1992 article. Most of these articles cite it in the literature review section. The four articles that incorporated the learning framework deserve closer attention.

BH’s reconceptualisation of the three forms of learning with special reference to who learns what with what effect served as a starting point for further advancement of the learning literature. Specifically, there are four conceptual and empirical articles that cited this work five times (Boomberg 2007; Borrás 2011; Leach et al. 2014; Hezri 2004) and relied heavily on Bennett and Howlett’s conceptual framework of three types of learning and policy change. Bomberg (2007: 263) expands the BH policy learning framework by focusing on ‘learners’ (‘who’s learning, what, and to what effect?’) and ‘by exploring the dynamics of policy teaching’ (‘who’s teaching and with what effect?’). In other words, she focuses on ‘how “teachers” of policy ideas and instruments encourage different types of learning, and with what effect’ (257). As an empirical context, Bomberg analyses environmental non-governmental organisations’ activities as teachers to shape the adoption and use of new environmental policy instruments in the context of European Union enlargement. This work advances our knowledge in policy learning by developing and refining what it calls ‘cautious’ or ‘negative learning’, that is, ‘learning what not to do’ (263).

Borrás (2011: 730) focuses on cross-national diversity in learning: ‘Why are some countries learning more than others?’ She argues that ‘the answers are linked to organisational capacities at the national level in all three levels (governmental, policy network and social learning)’ with special reference to transnational learning mechanisms of innovation policy in advanced European countries. For Borrás, ‘government learning relates to the “administrative capacity” of the government itself, or the set of organizational practices and structures by which the administration manages tangible or intangible resources’ generating ‘innovation policy management change’ (729, emphases in original). ‘Policy network learning requires “analytical capacity”, or …the set of competences that deals with use and command of techniques of policy analysis [that is] the innovation-related techniques such as research programme evaluation, constructive technological assessment, technological forecasting, etc.’ generating ‘innovation policy programme change’ (729). Finally, ‘[t]he organizational capacity required in [social learning] is more diffused than the previous two, as it entails a certain degree of reflexive skills in a widely dispersed set of organizational actors, and their ability to communicate and create a sense of collective understanding’ generating an ‘innovation policy paradigm shift’ (729, emphases in original). This approach makes a valuable conceptual contribution for advancing the BH framework through incorporating a context-specific organisational capacity that filters the learning process and generates divergent outcomes across European countries.

Leach et al. (2014: 593) add two new dimensions to BH’s seminal work: (1) ‘Where learning occurs. Under what circumstances does learning occur within collaborative partnerships that address a technically complex environmental policy issue?’ (2) ‘How learning occurs. How does the design of the collaborative process influence learning? and how do the traits of the participating stakeholders influence learning?’ They analyse ten collaborative partnerships in marine aquaculture in the USA, utilising case study and survey methods.

Based on a descriptive case study of national sustainability indicators development with particular emphasis on their penetration of policy processes in Malaysia, Hezri (2004: 365) incorporates BH’s policy learning framework to knowledge utilisation activities (i.e. creation, dissemination, diffusion and utilisation) ‘to complete the feedback loop of an information cycle in a policy process’.

Among the remaining nine articles, there are two articles (Dale 1999; Huntjens et al. 2011) that cite BH four times. Dale cites the 1992 article in the literature review section and benefits from this work in clarifying eight mechanisms of policy transfer and influence (i.e. borrowing, learning, teaching, harmonisation, dissemination, standardisation, installing interdependence and imposition) through which the external effects on national education systems are carried and delivered. In addition to the types of learning, he specifically benefits from the insight of the 1992 article that ‘some form of policy learning is likely to be present in any mechanism of policy transfer’ (Dale 1999: 10). Dale argues that in education, ‘the effects of globalization are largely indirect, the result of the stances adopted by nation-states in response to globalization, rather than a direct effect of globalization’ (17).

Huntjens et al. (2011: 146–149) cite the 1992 article when they refer to policy learning approaches, experience-induced policy change and different levels of policy learning in the introduction section. They aim to explore

whether there is a link between regime characteristics and policy learning (as an output of the regime at play). For this purpose we needed to develop two different and independent analytical frameworks: (1) A framework for assessing the characteristics of a water management regime (2) A framework for assessing different types or levels of policy learning. (148)

The BH article provides one of the key conceptual backgrounds for both frameworks. In a case study, they aim to contribute to understanding the policy learning process and ‘its structural constraints in the field of river basin management, in particular related to coping with current and future climatic hazards such as floods and droughts’ (145).

Hoppmann et al. (2014) cite the BH article three times when they refer to the policy learning literature (1423), the ‘process of policy learning’ and ‘different forms of policy learning’ (1424). They aim to ‘address the question of how the complex dynamics of innovation systems shape the process of policy interventions targeted at inducing technological change’ in a case study on ‘the evolution of the widely copied German feed-in tariff … system for solar photovoltaic (PV) power as an instrument that has been highly effective in driving the development and diffusion of PV technology’ (1423).

Lee and Menee (2012) adopt the policy learning definition of BH and refer to types of learning (202) and actors involved in ‘learning about policy goals and scope’ (203). Their work analyses the network structure of policy learning in the C40 Cities Climate Leadership Group. They argue that ‘policy learning is inherently relational; it is dependent on the interactions among different actors’ (203).

Meadowcroft (1997) benefits from BH’s review of the learning literature, their synthesis of divergent learning perspectives in particular, and the three forms of learning and their objects, subjects and effects (446). In a review article, he surveys on ‘metacritiques of planning’, ‘new governance debates’ and ‘discussion of policy-related learning’ (428) in the context of public planning in sustainable development.

Real-Dato (2009) offers a critical review of the theoretical approaches to policy change in the public policy field, with special reference to building bridges among the advocacy coalition framework, the punctuated-equilibrium theory and the multiple streams approach. He acknowledges three important insights from the 1992 article: (1) it distinguishes policy learning from policy change in that ‘learning may take place (new policy-relevant usable knowledge may be produced) without lessons being implemented. Indeed, learning may be used to reinforce the existing status quo’ (127); (2) policy learning may take place in ‘the policy subsystem’ (ibid.); and (3) ‘[p]olicy usable knowledge is socially constructed and, thus, it occupies a subordinate position regarding underlying social and political processes’ (129).

In the context of policy transfer, Stone (2000) explores how think tanks promote the spread of policy ideas about privatisation. She highlights BH’s emphasis on the common view that learning takes place ‘in complex arrangements of state and societal actors in various types of domestic and transnational policy networks and policy communities’ (2000: 60).

Wolman and Page (2002), in a case study on ‘how British local authority officials involved in the policy area of urban regeneration learn from each other’s experience’ (482), argue that ‘policy transfer does require utilization of knowledge drawn from the experience of others, although it does not require actual adoption’ (480). Their emphasis on the utilisation of knowledge is informed by BH in that ‘learning through policy transfer can occur in three different forms: process-related or institutional design; instruments or tools; and ideas or goals’ (ibid).

Zito and Schout (2009) in their introduction to a themed issue of the Journal of European Public Policy on ‘Learning and Governance in the EU Policymaking Process’ offer ‘learning modes’ ‘to give the authors some frame of reference’ (1111), which is ‘partly adapted from Bennett and Howlett’ (1110). This article (and the special issue in general) highlights contextualising the learning theory within European integration studies. It reflects the contributions of the 1992 article that, for example, emphasise learning processes arising from interactions among key individual actors such as politicians and civil servants.

Conclusion

The last 25 years have witnessed a growing academic interest in policy learning and policy change. Specifically, who learns what, from whom, when, how, why and with what effect have been explained and explored extensively. To date, various conceptual and theoretical frameworks as useful approaches have helped facilitate understanding of the various definitions, forms, processes, effects and limitations of learning with particular emphasis on organisational, institutional and policy change as well as its relationship with policy transfer. The study by Bennett and Howlett (1992), drawing on insights from the existing literature, is prominent theoretical perspectives that has proliferated and gained prominence.

The purpose of this paper was to systematically identify the academic impact of the 1992 article through citation analysis. What can we learn from research citations about its academic impact? There are three specific contributions of Bennett and Howlett (1992) through which scholars in social sciences, politics, public policy and administration in particular have explained and explored policy learning. First, they offer one of the earliest efforts to identify the academic field that follows a learning perspective in studying institutional, organisational and policy change. Second, they again offer one of the first attempts to reconcile theoretical approaches to policy learning and policy change by identifying and reconciling overarching conceptual, theoretical and methodological insights into various forms of learning and change. In doing so, they contribute to conceptual consolidation and bridge building across under-theorised and fragmented policy learning and change literatures by highlighting their unique contributions and limitations, as well as their reconciliation through interrelated multi-level (micro- and macrolevel) theoretical approaches. Third, they do not just offer a learning framework or model which only consists of a set of concepts that relate to object, subject and effect of learning, but they also take a modest step forward towards a theory of policy learning that explains how, why and when these concepts are related. In doing so, they have generated further academic interest in this topic and inspired various theoretical models of policy learning and change as well as policy transfer. As such, their 1992 seminal article on policy learning outlined a comprehensive research programme on policy learning and policy change.

Given the diversity of conceptual, theoretical and methodological approaches arising from diverse temporal and contextual as well as agency-level conditions, this brief review does not seek to imply that the work of these prolific authors has had the final word on learning and change research. Given the space and scope limitations, my paper has its own limitations. For example, it excludes books and edited book chapters; elements of subjectivity involved in the selection of papers discussed; the review is based on one database, albeit the most widely used and comprehensive, SSCI; and it covers articles published in English only. However, given the rigorous procedure of the review, I believe that none of the omitted research would have contained information that would critically alter my conclusions.