When, why and how institutional change takes place: a systematic review and a future research agenda on the importance of policy entrepreneurship in macroeconomic bureaucracies

Abstract Political science and public policy scholars have long emphasised the importance of understanding institutional change and policy entrepreneurship. This review article is a response to this call in the context of reform in macroeconomic bureaucracies. Adopting a ‘systematic’ approach to reviewing the literature, this paper investigates when, why and how institutional reform in key monetary and fiscal macroeconomic bureaucracies (i.e. central banks, treasuries, and ministries of finance) takes place. It reviews 29 selected articles on reforms in these bureaucracies published in Thomson & Reuters Web of Knowledge’s Social Science Citation Index, JSTOR, Sage and Wiley databases from 1980 to 2015. It shows that the current state of knowledge about institutional change in key macroeconomic bureaucracies is characterised by a lack of sufficient bridge-building among variants of institutional approaches as well as between institutional theory and public policy theory, resulting in persistent knowledge gaps. Against this background, the present review contributes to the body of knowledge on this topic in two main areas. First, it reviews the literature systematically to provide an overview of the key theoretical and empirical characteristics of when, how and why institutional reform takes place in these bureaucracies. Second, it identifies gaps and future avenues of research to stimulate progress in this important area of study.


Introduction
There has been a global move over the last two decades towards macroeconomic policies informed by neoliberalism (Simmons, Dobbin, & Garrett, 2008). With the goal of stable economic growth, these policies aim at achieving and maintaining price stability and a balanced budget, while affecting decisions and actions that relate to credit and money (monetary policy) as well as public revenue and expenditures (fiscal policy). Not surprisingly, reforms have been taking place in central banks, treasuries and finance ministries, which are the bureaucracies most responsible for the design and implementation of such macroeconomic policies. Specifically, there has been a significant worldwide trend towards the increased legal independence of central banks (Marcussen, 2005) and budget reforms in treasuries and ministries of finance (MoF), including expenditure controls and cash management improvements to achieve fiscal consolidation (Blöndal, 2003;Wanna, Jensen, & De Vries, 2010).
At the same time academic interest in public sector reform has also increased, generating a large number of empirical studies on when, how, and why these reforms have occurred. We found 9915 publications in English dated between 1 January 1980 and 31 December 2015 in the fields of political science or public administration, recorded in Web of Knowledge (Social Sciences Citation Index, SSCI) with 'reform' as the topic (i.e. appearing in the title, keyword, or abstract) in the subject category field. One hundred and one articles had 'macroeconomic bureaucracies' as the topic. 'Central bank' was the topic in 55 papers, whilst 'MoF' and 'treasury' were the topics of 16 and 30 papers, respectively. However, no comprehensive review of the timing or the processes and causes of such reforms in macroeconomic bureaucracies have ever been published. Thereupon, this paper's objective is to conduct a systematic review of when, how and why reforms in these key monetary and fiscal bureaucracies take place. In this context, it will pay special attention to the role of agency, of policy entrepreneurs in particular, in policy and institutional change processes.
Our key contributions to the literature are to provide the first comprehensive review of institutional change and policy entrepreneurship in macroeconomic bureaucracies, consolidate existing research in political science and public administration, establish connections between fragmented literatures, identify gaps between different research streams and suggest promising paths for future research on institutional change. To this end, we conducted a systematic literature review to analyse how the scholarship on reform in macroeconomic bureaucracies has built-up over the last three decades.
First, we will describe our research methodology, including a review and classification of the results. Following this, we report, synthesise, and discuss our findings. Finally, we discuss limitations of the current literature and their implications for further research.

Methodology
Despite its widespread use in medicine, health care and, more recently, management, a systematic review is an uncommon practice in the fields of political science and public administration. Traditionally, reviews in these fields have been narrative. Although a narrative approach can be valuable, it has been criticised due to a high degree of subjectivity and lack of generalisability (Mulrow, 1994). A systematic literature review combines qualitative and quantitative methods to map the fields of study in a transparent manner. In contrast to a narrative review, a systematic review employs an analytical review scheme to perform a comprehensive and critical appraisal of literature. As Tranfield, Denyer, and Smart, note Systematic reviews differ from traditional narrative reviews by adopting a replicable, scientific and transparent process, in other words a detailed technology, that aims to minimise bias through exhaustive literature searches of published and unpublished studies and by providing an audit trail of the reviewers ' decisions, procedures and conclusions. (2003, p. 209) We adopted a systematic literature review consistent with these aims. The review selects on double-blind-reviewed journal articles and focuses on when, how and why such reforms take place as described further below.
Our key data source has been Thomson & Reuters Web of Knowledge's Social Science Citation Index (SSCI) database, one of the most comprehensive databases of peer-reviewed journals in social sciences. We limited our sources to peer-reviewed journals because these are regarded as confirmed knowledge (Podsakoff, MacKenzie, Bachrach, & Podsakoff, 2005). We excluded books and book chapters due to variations in the peer review processes (Jones, Coviello, & Tang, 2011).
We adopted a three-step selection process. The first step involved a search of SSCI using a keyword combination enabling us to arrive at a pool of articles from which the most relevant contributions could be selected (number of articles: 101). The second step consisted of eliminating irrelevant articles by reading the titles, abstracts and full texts to achieve a sample of substantively relevant articles (number of articles: 21). The third step included the incorporation of eight additional articles which were either identified by cross-referencing during analysis of bibliographies, or detected in a previously performed preliminary systematic search through other major social sciences databases such as JSTOR, Sage and Wiley (number of articles: 29).
In the first step, we performed a web-based selection of all possibly relevant academic publications in English in the SSCI archive, introducing several filters. Our review was limited to peer-reviewed papers and excluded books, book chapters and non-refereed publications within the research disciplines of political science and public administration. The search process resulted in an up-to-date sample comprised of articles dated from 1 January 1980 -the earliest date that SSCI covers -until 31 December 2015. Instead of limiting the article search to 'institutional change' or 'policy entrepreneurship' , we preferred to use the broader term, 'reform' . The rationale behind our preference is the plurality of conceptual schemes adopted by the researchers. For example, we observed that many scholars have focused on 'institutional change' as a phenomenon although they do not label their work as institutional analysis. Therefore, we extended our conceptual scope beyond the confines of 'institutional change' and used 'reform' as the major search phrase to cover the broad range of definitions, concepts, methodologies and theories in this research domain.
As shown in Table 1, 'reform' is a popular topic; it has appeared in 58,214 articles published in SSCI-indexed journals and 9915 of these articles were published in political science and public administration journals. Our research yielded 101 scientific articles, which we assumed to address the issues of reforms in central banks, treasuries and ministries of finance.
In the second step we sought to narrow down the pool, so that it contained only the most relevant contributions in the literature. To this end, we imposed additional exclusion criteria on the initial sample. As this review focuses on macroeconomic bureaucracies, studies about reforms in peripheral or local public organisations and administrative reorganisations were excluded. Additionally, organisational or policy change processes taking place in international or supranational institutions, such as European Monetary Union, or organisations, such as European Central Bank, were excluded as well. Articles that fit the review criteria -empirical and/or theoretical research on institutional reform/change and/or policy entrepreneurship in key macroeconomic bureaucracies were then analysed with special reference to theories and findings directly related to institutional change in these bureaucracies, rather than monetary and fiscal policy reforms. Twenty-two articles were found to be relevant at the end of this elimination process.
Finally, a third step was essential to overcome possible weaknesses such as the omission of relevant research. Following this review, we decided to broaden our pool by two additional selection processes. First, we added four articles to our sample based on our reading of the 22 relevant articles. 1 Second, although our sample included articles in Thomson & Reuters Web of Knowledge, we reasoned that this database may have excluded some relevant researches. Consequently, the use of other search engines (i.e. JSTOR, Sage, & Wiley) yielded four additional results. 2 Thus, our final sample was comprised of 29 articles which were all reviewed thoroughly and classified according to their research designs, methods and theoretical approaches (see Appendix 1).
The authors of this article jointly analysed these papers in the sample. Following Webster and Watson (2002), our coding criteria aim to satisfy principles of relevance and of feasibility. Our codes are based on (1) author, (2) year of publication, (3) journal, (4) empirical focus (e.g. central banks), (5) theory focus (e.g. institutional theory, multiple streams approach) and (6) key findings and/or arguments. Coding was performed independently by both authors summarised in an Excel spread sheet. This involved an iterative process where multiple rounds of coding and discussion among the authors took place until they reached an agreement on all codes and the final classification (Miles, Huberman, & Saldana, 2014). 1 We checked why these four articles were not included in the initial pool that we acquired in the web-based search process.
three of these four articles - Mcnamara (2002) and Quaglia (2005aQuaglia ( , 2005b) -do not contain the keyword 'reform' in their topics, and one - Polillo and Guillén (2005) -was tagged as a 'review' but not as an 'article' in SSci.
2 our findings about the absence of these three articles in the Web of Knowledge SSci despite their availability in alternative online databases are as follows: Kawasaki (1993) did not appear among the results of our web-based search as the digital topic archive of thomson and Reuters' SSci do not cover the journal's issues -namely the Journal of Public Policy -for the time when the articles were published. Record of this journal is available from 2009 onwards. King (2005) was tagged as a 'review' but not as an 'article' in SSci records and Jung (2008) was published in a journal -Pacific Focus -which was indexed in 'area studies' but not in political science and public administration. as a result, these three articles were not included in the initial pool which we acquired after the web-based search despite their presence in the digital archive.

Number of publications and topical interests
Our research shows that despite the importance of the central bank, MoF, and treasury for national economic performance, various dimensions of institutional change and policy entrepreneurship in these key macroeconomic bureaucracies have not been analysed extensively enough in political science, public policy and public administration literatures. As presented in the previous section, the final sample of the most relevant contributions included just 29 articles published during the 36 years between 1980 and 2015.
The research on institutional reform in economic bureaucracies is dominated by studies on central banks. Twenty-one papers (72%) focused on central banks, whilst four (14%) and two papers (7%) focused on MoF and treasuries, respectively (see Table 2). The interest in central banks is largely due to world-wide statutory reforms granting central banks legal independence from political interference in the 1990s. However, given the interconnectedness of monetary and fiscal policy, it is surprising that there were only three studies (10%) that offered cross organisational comparisons of institutional change/reform and only one study that offered a cross-country comparison (3%).

Methodology, countries studied and theoretical approaches
In terms of research design, we observe that single-case design (22 articles, 73% of the total) is the most common research design (see Table 3). There is only one few-cases design. There is a balanced distribution of single-country case studies between developed and developing countries. Of those 29 studies, seven papers (24%) focus on developed economies including the UK (three), Israel (two), Australia, Italy and the USA, and four papers (14%) on central bank independence (CBI) reform in developing countries including Turkey (two), Chile, China and Mexico. 3 Only one article analyses CBI in a less developed country -namely Uganda. In regard to MoF, authors analyse Japan (one), Israel (two), Saudi Arabia, the Netherlands and South Korea. Two articles on treasury reforms offer case studies on China and New Zealand. A cross-country case study in our sample focuses on Italy, the UK and Germany.
Institutional analysis is used in 23 studies (or 79%). 4 In our sample, seven articles accord with historical institutionalism due to their emphasis on both formal and informal rules, 3 in this classification, we considered these countries' economic conditions for the period covered in this study. 4 Rational choice, organisational, historical institutionalism, constructivist and discursive institutionalism are five variants of institutional analysis (campbell, 2004;campbell & Pedersen, 2001;Schmidt, 2008Schmidt, , 2010. they all share a 'common view that agents' choices and actions are influenced by the institutional environment in which they are embedded (i.e. 'institutions matter'), and they have their own assumptions about relationships between actors and their environment generating the agency vs. structure debate' (Bakir, 2013, p. 43). norms and conventions that inform reform in economic bureaucracies in a historical context (Boylan, 1998(Boylan, , 2001Cohen, 2015;Jung, 2008;Kawasaki, 1993;Keating, 2011;Maman & Rosenhek, 2007). They point out the importance of institutionalised behaviour patterns among actors -usually marked by conflict-driven dynamics -in reform processes which mostly follow external shocks, such as the outbreak of crises. They highlight how domestic institutions inform reform process. Four articles seem to fall under organisational institutionalism, which often focuses on cultural conventions and cognitive frameworks that inform agency behaviour (Bodea & Hicks, 2014;Marcussen, 2005;Polillo & Guillén, 2005;Rapaport, Levi-Faur, & Miodownik, 2009). Reforms occur when these conventions and frameworks within or among organisations diffuse. Only one article adopts rational choice institutionalism, assuming that the institutional change is a rational response by politicians who are shaped by their perceptions about other significant political actors' material preferences and powers (Bernhard, 1998). In addition to these studies, which mostly adopt one variant of institutional perspectives, 11 articles combine two or more varieties of institutionalisms and offer broader analytical frameworks (Bakir, 2009a;Dellepiane-Avellaneda, 2013;Epstein, 2006;Johnson, 2006;King, 2005;Maman & Rosenhek, 2009, 2012McNamara, 2002;Patel, 2009;Quaglia, 2005aQuaglia, , 2005b. These studies highlight the significant role of policy entrepreneurs and/or ideational entrepreneurs who utilise strong ideas in domestic policy-making processes to deliver reform outcomes. However, there are only three papers which explicitly use Kingdon's multiple streams analysis (Bakir, 2009a;Patel, 2009;Rapaport et al., 2009). The remaining six papers (21%) in our sample are purely descriptive and tend not to convey a strong theoretical base or analytical leverage (Ang, 2009;Caporale, 2003;Chen, 1995;Hawtrey, 1997;Şahin, 2012;Wallis, 2010). In the sample, there are only five papers that use quantitative methodology and benefit from institutional analysis (Bernhard, 1998;Bodea & Hicks, 2014;Marcussen, 2005;Polillo & Guillén, 2005;Rapaport et al., 2009).

Central questions and their answers
We identified a wide array of questions around which the current research on institutional change in monetary and fiscal bureaucracies is shaped. However, we observed that almost every author addresses at least two dimensions of the following broad question: 'When, how and why do administrative/legislative reforms in these bureaucracies take place?' To address this, scholars identify various macro-level (structural and institutional) and microlevel (organisational and/or individual agency) explanations. Not surprisingly, there are a plethora in the literature. These include explanations based on pressures arising from structures (e.g. economic crises that open windows of opportunity for change, pressure from economic globalisation processes, and political regime changes), institutions (e.g. ideas/ norms driving endogenous actors seeking legitimacy in their environment) and agents (e.g. interest and idea-guided politicians, policy entrepreneurs). Articles in our final sample often refer to more than one structural-, institutional-and agency-related factor that influences institutional change. Inevitably these three factors are not discrete.

When do reforms in macroeconomic bureaucracies take place?
The literature offers both exogenous and endogenous explanations for the timing of institutional change. The most frequently referred to exogenous factor in contributing to reforming environments is economic crisis. Maman and Rosenhek (2007, p. 253), for example, categorise the acute hyperinflation crisis in Israel in the 1980s as 'the most important condition that facilitated the institutional reform' in central banking (see also Bakir, 2009a). In addition to crisis, a 'policy window' for reform might also be opened with general elections (Patel, 2009). In another post-crisis reform study on the MoF and Economy in Korea, it is suggested that economic crisis opens 'a rare window of opportunity' for leaders 'to pursue reform measures that may have been impossible in the absence of crisis' (Jung, 2008, pp. 132-133). In addition to these studies focusing on domestic economic crises, we observe that international crises also attract the attention of scholars in analysing the timing of bureaucratic reform. For example, Kawasaki (1993) argues that the 1969-1971 international monetary crises triggered the reform process in the MoF in Japan. These studies emphasise the significance of regional and/ or worldwide shocks that challenged existing institutional arrangements influencing economic bureaucracies. Further, they highlight the role of the IMF as an international intergovernmental organisation and the EU as a supranational organisation, which coerce politicians and holders of international financial capital (guided by material interests), into pursuing reform. There are also endogenous explanations of institutional change in macroeconomic bureaucracies. The endogenous explanations mostly focus on the decisions and actions of individuals (e.g. idea-guided politicians, policy entrepreneurs, ideational entrepreneurs, central bankers), epistemic communities, and organisational actors (e.g. central banks). Indeed, ideas matter when they are effectively put into action by institutional entrepreneurs to bring about institutional change (see, for example, Bakir, 2009a;Kawasaki, 1993). These explanations relate to world-culture theory (see Meyer, Boli, Thomas, & Ramirez, 1997) and organisational institutionalism (see DiMaggio & Powell, 1983;Scott, 2001). In this perspective, exogenous globalisation pressures were endogenously interpreted by states that delivered CBI reform. For example, Polillo and Guillén (2005, p. 1794) refer to 'globalisation pressures of normative, coercive or mimetic' processes that generate environmental pressures at the state and/or central bank levels towards institutional isomorphism, and convergence in central banking practices (see also Marcussen, 2005). It is assumed that states supply independence to central banks as they compete with each other to gain or maintain their legitimacy in the international community.

Why do reforms take place?
There are several articles that mainly rely on rational choice explanations to explain why central banking reform takes place. Some of the scholars focus on the role of political leaders who initiate reform processes as they seek to preserve both their own interests and those of economic elites by institutionalising policy preferences; in the event that political power is lost, these preferences would then endure in the political economy of their country (Boylan, 1998(Boylan, , 2001. Here, it is argued that institutional change takes place because politicians adopted the CBI reform due to a pressing functional political or economic problem. For example, Boylan (1998Boylan ( , 2001, in two case studies on Chilean and Mexican CBI reforms, reported that institutional change is most likely when the political regime changes from an authoritarian to a democratic one. In this view, CBI takes place in these developing countries because authoritarian elites fear the populism that may be endemic to new democracies and know that a regime change is imminent, [thus] they can be expected to create autonomous central banks to lock in a commitment to price stability over the long haul. (Boylan, 2001, p. 5; see also Boylan, 1998, p. 444) Other functionalist explanations note that politicians adopt a central bank reform because they respond to the lobbying of an interest group (e.g. banking sector) 'that is more committed to price stability than is the median voter to reducing the central bank's risk in pursuing anti-inflationary policies' (Posen, 1995, p. 256). In this perspective, politicians are influenced by economically powerful groups such as financial constituencies and donors (Keating, 2011).
In contrast to these functionalist perspectives, McNamara (2002, p. 48) argues that 'governments choose to delegate [CBI] not because of narrow functional benefits but rather because delegation has important legitimizing and symbolic properties which render it attractive in times of uncertainty or economic distress' (emphasis added, see also Simmons & Elkins, 2004). It is widely held that institutional change in central banking takes place when norm-driven endogenous actors operate in 'the culture of neoliberalism … ,[with] the spread of [CBI being] a fundamentally social and political phenomenon, rooted in the logic of organizational mimicry and global norms of neoliberal governance' (McNamara, 2002, pp. 48-49). Thus, CBI reform has been 'determined by a social process of crossnational institutional diffusion' (McNamara, 2002). For organisational (or sociological) institutionalists, 'the greater a country's exposure to foreign trade, investment and multilateral lending' (Polillo & Guillén, 2005, p. 1776, and/or greater its need to attract foreign direct investment and sovereign borrowing (Bodea & Hicks, 2014), the more independent is its central bank. Adopting a similar line of argument, treasury reform in China was initiated by the key reformers in the communist party to increase fiscal control and reduce corruption in a world of economic globalisation (Ang, 2009). Development of the booming Chinese market economy lacked a centralised treasury system which has been a norm in developed countries: 'the traditional treasury system was unsuited to the demands of a complex market economy' (Ang, 2009, p. 266; for the analysis of central banking reform in China, see Chen, 1995).
The role of interests and ideas are also noted in the reform process. For example, Johnson (2006) highlights that the very same reform agenda for central banking reform in Hungary and the Czech Republic diffused from the EU level to the domestic sphere through norm internalisation and external incentives (i.e. EU membership criteria).
In contrast to constructivist institutional perspectives, historical institutionalist accounts focus more on path-dependent patterns of conflicts among proponents and opponents of reform. These conflicts are aimed at advancing the ideas and material interests of each side through domestic coalitions and bargaining, which affects the reform outcome in the domestic political economy (for example, Bakir, 2009a;Maman & Rosenhek, 2007, 2012. Agency matters in understanding reforms in macroeconomic bureaucracies. In contrast to most of the constructivist accounts of diffusion and punctuated equilibrium, several articles have shown that crises or globalisation pressures are neither necessary nor provide sufficient conditions for institutional change. King (2005), for example, in a case study on central bank reform in England, shows that institutional change does not necessarily occur because of economic/political crises, uncertainty, pressures of international financial community, inter-state competition or external coercion by international intergovernmental organisations. Instead, the reform took place in 1997 when ideational entrepreneurs, 'an epistemic community of monetary experts'/central bankers, convinced Labour's Chancellor of the Exchequer, Gordon Brown, to 'champion' this reform (King, 2005, p. 113). In this view, institutional change took place because 'the ideas promoted by an epistemic community of monetary experts influenced [key politicians]' (King, 2005, p. 96); moreover, these ideas were momentous because their adoption 'satisf [ied] politicians' perceptions of their electoral self-interest' (King, 2005, p. 117). Politicians who pursue a reform agenda are often convinced by monetary experts and cultural processes (King, 2005;Maman & Rosenhek, 2007;Marcussen, 2005;McNamara, 2002;Patel, 2009).
There are studies that blend constructivist and rational choice institutionalisms to offer agency-level explanations for reform in economic bureaucracies. Dellepiane-Avellaneda (2013) also highlights that change in the constitution of the Bank of England in 1997 was guided by Brown's personal ideational preferences and political strategy as a political entrepreneur. Similarly, in regard to the reform in MoF in Japan, the leadership of the Vice-Minister of Finance had a significant impact in overcoming the policy-making deadlock in the post-crisis period, as he used 'the norm of consensus-building to his advantage' (Kawasaki, 1993, p. 125).

How do reforms take place?
There are several explanations for how reforms in economic bureaucracies take place. Following McNamara (1998), constructivist institutionalists offer three mechanisms to illustrate how ideas are translated to policies and institutional change: First, a period of policy failure leads to the collapse of the old paradigm and the search for new solutions. While this policy failure may be accompanied by a crisis, the main feature is a period of greater uncertainty. Second, a new paradigm emerges offering a clear policy solution that is advocated by the epistemic community and implemented by politicians in a few states. Politicians in other states monitor the results of these test cases to judge whether the policy is effective or not. Third, politicians in other states proceed to emulate this policy, embedding the new paradigm in their own institutional framework. (King, 2005, p. 99) In this constructivist perspective, [o]rganisational models are diffused across borders through the perceptions and actions of people seeking to replicate others' success and legitimise their own efforts at reform by borrowing rules from other settings, even if these rules are materially inappropriate to their local needs. (McNamara, 2002, p. 48; see also Bodea & Hicks, 2014;Marcussen, 2005;Polillo & Guillén, 2005;Rapaport et al., 2009) Similarly, several studies highlighted the role of the spread of neoliberal policy prescriptions as well as the pressure and support of international intergovernmental economic organisations, such as the IMF, World Bank, Bank for International Settlements, Organization for Economic Cooperation and Development (OECD), for such reforms (Epstein, 2006;Johnson, 2006;Jung, 2008).
However, these studies do not show how international norms are filtered in domestic political economies. Indeed, institutional change in macroeconomic bureaucracies is the product of domestic political conflicts and bargaining in formulating and implementing public policies where multiple individual and organisational agents play decisive roles in policy-making processes.
In contrast to constructivist institutional perspectives, several studies using historical institutionalism have emphasised that institutional change in macroeconomic bureaucracies is determined by the domestic institutional environment that filtered the conflicts, struggles and contestations in political economies. These studies highlight interactions among external and internal dynamics and institutions. For Maman and Rosenhek (2007, p. 257), there were two major mechanisms of institutional change operated and interacted within [CBI reform in Israel]: a mechanism of inter-state dominance exercised by an external actor -the American administration -and a mechanism of expert power exercised by a cross-national network of Israeli and American academic economists.
In a later article about inflation targeting as an institutional arrangement, they focused on inter-bureaucratic political conflicts between the Israeli Central Bank and MoF as a major mechanism through which worldwide diffusion of institutional practices takes place … We contend that, like other processes of institutional change, its [inflation targeting] adoption is not the outcome of smooth processes of diffusion, learning and acceptance of more rational and efficient practices, but rather the result of political conflicts between [state] actors seeking to advance their institutional interests. (Maman & Rosenhek, 2009, p. 217, 219) There are also several historical institutionalist studies that focus on why reform initiatives fail in domestic political economic struggles. Keating (2011), in his study on democratisation in Uganda, shows that CBI reforms were halted due to a severe conflict between the pro-reform executive branch and counter-reform legislature. Prior to the adoption of multipartism, members of parliament -formed in a no-party setting, hence not bounded by any party program or party discipline -were free to oppose CBI reform initiatives championed by the executive. This led to deadlock within the government, thereby impeding CBI reform. According to Keating, the crisis between the executive branch and legislature over the CBI reform accelerated the democratisation process. In addition to the conception of multipartism as a necessary institutional condition of democracy, the executive branch perceived that the adoption of multipartism would lead to a more tractable legislative power in contrast to the no-party parliament which the country had during the CBI reform deadlock.
The pressure of external actors such as the IMF and EU can also influence who gets what in domestic political conflicts. For example, Epstein (2006Epstein ( , p. 1020 argues that international organisations could depoliticise reform and facilitate CBI reform in Poland 'through persuasion, argumentation, and coalition [that] cultivated a social consensus' . If they failed to do so, the reform processes would be marked by polarisation at the domestic level leading to a politically turbulent and unfruitful change process, as has been the case for the EU-led agricultural reform in the same country. The focus on 'agency' enables the researchers to better reveal the causal mechanism of institutional change (i.e. how institutional change actually takes place). As such, a number of studies focus on endogenous sources of institutional change in macroeconomic bureaucracies. They argue that institutional change arises from the decisions and actions of agents such as political leaders (Boylan, 1998), ideational entrepreneurs (Marcussen, 2005;Patel, 2009), policy and institutional entrepreneurs (Bakir, 2009a), epistemic community of central bankers (King, 2005), treasurers (Patel, 2009), and states (McNamara, 2002;Polillo & Guillén, 2005;Rapaport et al., 2009). For example, institutional change takes place when 'policy entrepreneurs, with joint membership in domestic and transnational policy communities, mediate various ideas and discourse within and among these communities in a punctuated institutional equilibrium [i.e. economic crisis]' (Bakir, 2009a, p. 572).
Apart from politicians and policy entrepreneurs, for Quaglia (2005a), civil servants and their ideas played a major role in institutional change in the national central bank and the Ministry of the Treasury and Economic Planning in Italy. Institutional change in these economic bureaucracies was able to take place because strong neoliberal ideas '[found] an audience' among 'influential civil servants dealing with macroeconomic policies' (Quaglia, 2005a, p. 559). In a similar vein, transnational learning and communication among central bankers also contributed to the agency of central bankers (Marcussen, 2005;McNamara, 2002;Polillo & Guillén, 2005).
Central banks themselves are also regarded as institutional entrepreneurs that introduced and promoted institutional change (King, 2005;Maman & Rosenhek, 2009, 2012. For example, the Central Bank of Israel employed 'persuasive and credible discursive strategies [e.g., framing CBI as a requirement of globalisation]' to perform a major role in institutional change processes (Maman & Rosenhek, 2012, p. 336).
What are the enabling conditions for the agency of actors? The enabling conditions at actor level relates to both social positions, and individual characteristics and skills of agency. In regard to human agency, a policy entrepreneur becomes an institutional entrepreneur when he purposely initiates institutional change and steers the change process in all stages of public policy (Bakir, 2009a). Such individual institutional entrepreneurs may represent a powerful agency with multiple identities that enable him to operate in different ideational realms as decision-maker (e.g., politician and bureaucrat), academician (e.g., theoretician and intellectual), framer (e.g., spin doctor publicizing favourable interpretation of some of the neoliberal ideas), and mediator [between domestic and transnational policy communities]. (Bakir, 2009a, p. 587) In particular, policy entrepreneurs build broad coalitions supporting their entrepreneurship activity in the domestic political economy. This is because they carry, connect and decontextualise 'different kinds of ideas such as programs and paradigms and [utilise effectively] communicative and coordinative discourses to affect policy and institutional changes' (Bakir, 2009a, p. 588).
There are also ideational entrepreneurs such as academics, bureaucrats and experts promoting policy change. Their essential resource is access to key politicians (King, 2005;Maman & Rosenhek, 2009, 2012Patel, 2009). Furthermore, civil servants in macroeconomic bureaucracies as ideational entrepreneurs (e.g. central bank and treasury senior bureaucrats) have intangible resources that are significant for 'engineer[ing] major policy changes' via 'technocratic capture' (Quaglia, 2005a, p. 546). Here intangible assets refer to 'information, including access to empirical data; technical knowledge, which can take the form of specific policy paradigms; and organizational culture, especially a strong civil service ethos that also affects the perception of a given bureaucracy as an apolitical body' (Quaglia, 2005a, p. 547).
There are also tangible resources: 'first and foremost, tangible assets, such as existing degrees of independence, and intangible assets, such as expertise, authority, and the strategies of the bank' (Quaglia, 2005b, p. 554). For example, intangible resources make 'possible the build-up of a particular kind of "credibility, " which in turn enables certain civil servants to extricate certain parts of macroeconomic policy-making from everyday domestic politics' (Quaglia, 2005b, p. 558). Likewise, their institutional positioning as the nexus between the academic and the policy-making fields provides them with an invaluable political resource to act as institutional entrepreneurs, allowing them to deploy concepts, causal models and empirical claims provided by academic economics in their struggles with opponents. (Maman & Rosenhek, 2009, p. 223; see also King, 2005;Maman & Rosenhek, 2012)

Discussion
Several issues related to methodology, theory, gaps and directions for future research in the literature emerged from our review.

Methodology
Most papers we reviewed in this article use qualitative research designs relying primarily on case studies rather than quantitative analysis. This is probably due to the suitability of qualitative methods in this research field where authors are interested in 'why' and 'how' questions rather than 'what' questions. They examine detailed relationships and reform processes in real-life contexts, where they do not have control over dependent and independent variables, and benefit from multiple sources of evidence (see, Yin, 1994). These studies typically rely on multiple data sources including interviews, official records and secondary written sources. There is only one study based solely on archival data sources (Patel, 2009). Responding to the question of 'when, why and how institutional change takes place' requires an in-depth analysis of various actors' decisions and actions in domestic political economic processes. Thus, single-case designs offer analytical leverage to better diagnose who the actors and what the processes of change are in particular cases.
Although most of the articles are based on single-case design, some of these articles set case designs in a comparative fashion. Authors who apply a within-case comparison in single-case settings succeed in showing the organisational level differences by holding context-specific conditions constant (Epstein, 2006;Quaglia, 2005a). Additionally, scholars who make a cross-temporal comparative analysis in a single-case design capture the common patterns and mechanisms which remain the same in different time frames (Bakir, 2009a;Patel, 2009;Wallis, 2010). Finally, there is a single-case design related to CBI reform which is examined in a comparative perspective via two shadow cases (Quaglia, 2005b). These examples actually show that single-case design does not solely mean a one-dimensional analysis of a particular case; it may well accommodate a comparative analytical perspective. In regard to inductive theory building, there is only one article that offers an integrated framework which shows that international, national and micro-institutional levels of interactions affected the CBI outcome (Quaglia, 2005b).
However, most of the qualitative studies reviewed have several methodological weaknesses (Ang, 2009;Caporale, 2003;Chen, 1995;Hawtrey, 1997;Wallis, 2010). They do not justify case selection (e.g. type of the case), sampling (e.g. selection of interviewees), data collection (e.g. triangulation), and data analysis (e.g. distinction between data offered by interviewees and data induced by researcher, and/or incorporation of emerging themes into aggregate theoretical constructs). Thus, findings in the past research are subject to validity tests in future studies.
In contrast to qualitative studies, large-N designs with inferential statistical applications offer robust tests for hypotheses which are mostly formulated within the confines of rational choice and organisational institutionalisms (Bernhard, 1998;Bodea & Hicks, 2014;Marcussen, 2005;Polillo & Guillén, 2005;Rapaport et al., 2009). However, such studies in our sample (only five out of 29 studies) are marked by two key common methodological weaknesses. First, they do not offer insight into the causal mechanisms and processes which operate in domestic political economies producing institutional change. Second, they are designed for theory testing and do not seem to contribute to theory building.

Theoretical approaches
Many articles offered variants of institutional theory and public policy theory. A close look at the theories suggests that variants of institutional theory or their combinations (27 papers, 93%) are widely used. In addition to institutional theory, multiple-policy stream analysis (Kingdon, 1995) in particular is a dominant public policy approach. For example, it is used 'to study policy diffusion in a global context' where nation states as organisational actors adopt CBI (Rapaport et al., 2009, p. 697), agenda setting (Patel, 2009) and policy diffusion, and institutional change where policy entrepreneurs steer institutional change processes (Bakir, 2009a).
The vast majority of these articles aim at theory testing rather than theory building in empirical papers. There are five narrative studies (17%) which are not very clear about their theoretical approach (Ang, 2009;Caporale, 2003;Chen, 1995;Hawtrey, 1997;Wallis, 2010). These studies are classified as atheoretical narratives (for a good example of an analytic narrative in the sample, see Dellepiane-Avellaneda, 2013). Further, it is important to note that all theorising endeavours do not necessarily command analytical leverage (see for example, Şahin, 2012).
Only three articles (8%) are concerned with adding to or further developing existing theory (Bakir, 2009a;Quaglia, 2005a;Rapaport et al., 2009). Rapaport et.al., links the multiple streams analysis to diffusion of CBI across nations. It is argued that A state's (i.e., agent's) decision to implement a policy idea is based on the interplay of three conditions: (i) whether the problem a particular policy addresses is very salient, i.e., ranked high on the public agenda; (ii) whether solutions that have been put forward to some problems can be used to fit others; and (iii) the acceptance of the policy solution of the political establishment. (Rapaport et al., 2009, p. 701) In doing so, authors make a welcome contribution to organisational institutionalism by offering an agent-based model to evaluate these three under-explored aspects of the diffusion process.
Kingdon's framework is also adapted to institutional analysis to illustrate how policy entrepreneurs became institutional entrepreneurs, embedded in institutional and structural contexts to steer various actors in public policy processes towards change. In so doing, institutional theory and policy theory achieve more explanatory power. For example, it is argued that institutional entrepreneurship is most likely when policy entrepreneurs with multiple identities operating in the intersects of various policy subsystems mobilise various ideas and discourse for policy and institutional changes, resolve conflicts within and among domestic and international policy communities, and steer the translation of these ideas into policy outcomes in the various stages of policy-making processes (Bakir, 2009a, pp. 572, 575). They operate across problem, solution and streams. This perspective proved highly useful to bridge variants of institutional approaches and public policy theory.
By utilising historical and organisational institutional approaches, Quaglia (2005a) offers one of the rare and important examples of a comparative case study of the Bank of Italy's path to independence versus the experiences of the Bank of England and the Bundesbank. She suggests an integrative, three-tier analytical framework articulated at the international (i.e. diffusion of neoliberal ideas and pressures of international markets and organisations), national (i.e. bargaining between suppliers and those who demand CBI) and organisational (i.e. a central bank's utilisation of its 'tangible assets, such as existing degrees of independence, and intangible assets, such as expertise, authority, and the strategies of the bank') levels of analysis (Quaglia, 2005a, p. 554).

Gaps in the literature, directions for future research and limitations
Our review reveals important methodological weaknesses and theoretical gaps in the literature. The methodological weaknesses in the existing research are twofold. First, the existing research using qualitative-or quantitative-based methods test theories or hypotheses with data. There is a scarcity of rigorous and transparent inductive qualitative research design where the main emphasis is on theory building. Transparency is one of the key features of scientific research and 'renaissance in case research' (Ketokivi & Choi, 2014:1, cited in Bakir, 2017). Yet in most of the papers we reviewed, there is no methodology and/ or research design section. Thus, transparent and rigorous inductive research is need to generate theory from data.
Second, despite a large and growing literature on administrative reform, the comparative study on monetary and fiscal bureaucracies has been generally neglected to date. In the absence of comparative research design, it is difficult to generalise the causal explanations for other organisational contexts and to expand the provided theoretical explanations. There is a need for within and cross-country comparisons as well as inter-and intra-bureaucracy comparisons. Empirical research done thus far has been primarily through single, in-depth, longitudinal case studies. Generalisability of the causal explanations provided by most of the articles that we covered in our literature review is subject to further verification. Further, methodological relevance of the many cases for theory building or theory testing is not discussed in most of the articles. As such, cases seem to be 'given' rather than 'selected' . Given the lack of a research design discussion, most articles seem to adopt a single-case design to overcome the difficulties related to data availability and collection necessary to perform a multiple-case comparison. Hence, to test the external validity of these insights and to welcome new ones, we need more rigorous research designs with multi-case and cross-country comparative research focusing on which combinations of interdependent variables lead to institutional change (or persistence).
Unsurprisingly, in the absence of a rigorous inductive research method, most of the previous research fails to explain (why) and explore (how) causal mechanisms operating at the contextual (structural and institutional) and actor (organisational and individual) levels ultimately result in the collective outcome (e.g. institutional change).
To address these weaknesses future research might benefit from recent methodological advances informed by the logic of grounded theory in management and organisation studies towards a rigorous inductive qualitative research design, the Gioia Method (Gioia, Corley, & Hamilton, 2013;Gioia & Thomas, 1996;Gioia, Thomas, Clark, & Chittipeddi, 1994). Indeed, the most recent interdisciplinary research (Bakir, 2017) shows the utility of this method for the political science, political economy and public policy disciplines in addressing these weaknesses which are endemic to qualitative research.
There are also several theoretical gaps in the literature which should be addressed in future research. Public policy scholars who study administrative reform have paid relatively little attention to how context conditions agency behaviour, whilst institutionalists have paid little attention to the mobilisation of various domestic and international actors towards institutional change, the formation of coalitions and conflicts among these actors, and the resolution of these conflicts in domestic political economic struggles in the internationalised policy domains (see Bakir and Jarvis, 2017).
There are several theoretical weaknesses of the past research. First, public policy scholars under-explore how and why structures (i.e. broader material and cultural contexts within which actors and institutions are embedded), institutions (i.e. formal and informal rules that guide behaviour of actors through the logic of appropriateness and logic of instrumentality), and agency-level enabling conditions (e.g. multiple identities) inform agency behaviour in institutional change processes. How institutionally embedded organisational and human agents are enabled to engage in bureaucratic reform and what strategies they adopt is not subject to theoretical analysis in the literature (see for example, Maman & Rosenhek, 2009, 2012Marcussen, 2005;McNamara, 2002;Polillo & Guillén, 2005). Despite references to what organisational institutionalists (Battilana & D' Aunno, 2009) call field-level (i.e. structural and institutional), organisational-level and individual-level enabling conditions for institutional work, there are only two papers explicitly discussing the theoretical underpinnings and empirical implications of these enabling conditions for agency behaviour and institutional change (Bakir, 2009a;Quaglia, 2005a). In this regard, a recent research shows that the examination of policy capacity in the context of central banking proves highly useful (Bakir and Coban, Forthcoming).
There is also a need to consider a policy entrepreneur's agency in institutional change processes that calls for fresh perspectives, complementary to existing ones (see Bakir, 2013, pp. 42-59). This may include, for example, an exploration from the organisational institutionalism perspective of the link between the concepts of 'institutional work' -purposive actions of human agency 'aimed at creating, maintaining and disrupting institutions' (Lawrence, Suddaby, & Leca, 2009, p. 215) and 'the enabling conditions for individual actors' to 'engage in institutional work' (Battilana & D' Aunno, 2009, p. 41) or the use of discursive institutionalism (Schmidt, 2008(Schmidt, , 2010 in order to account for the discursive strategies of institutional entrepreneurship. As Bakir (2009a, p. 573) argues, this helps to endogenise the individual agency (e.g. policy entrepreneur) that 'carries ideas and utilizes discourse in various stages of public policy-making toward institutional change' , and has been largely ignored in the literature. Incorporating 'normative and cognitive ideas (i.e. what was said) and discourses (i.e. who said what to whom, where, when, how, and why) to institutional and public policy analyses will help us better understand how, when, where, and why policy entrepreneurs' decisions and actions affect institutional change. In a similar vein, it may invite deep insights into debates on public policy to understand the role of structures and institutions in conditioning agency behaviour. There is certainly a need for future research on structural, institutional and agency-level (organisational and individual) sources of enabling conditions that inform the agency of agents.
Future research may consider such macro-and micro-level interactions to explain and explore the impact of causal processes and mechanisms on actor behaviour that generate institutional change. For example, structural complementarities such as the election of a new government in Australia, Australian policy and political pragmatism; institutional complementarities such as a new financial regulatory idea (i.e. Twin Peaks model), as well as a steering bureaucracy (i.e. Treasury Department); and the formation of an expert committee for financial system inquiry (i.e. Wallis Committee) and strong state capacity at the organisational level together reinforced the Australian Treasurer's policy and institutional entrepreneurship delivering radical institutional change in financial regulation (Bakir, 2013, pp. 116-134; see also Bakir, 2003Bakir, , 2009b. Second, both policy scholars and institutionalists are silent on the type of institutional change. To date, although there have been references to 'dramatic' , 'major' , 'radical' , 'substantial' and 'incremental' institutional changes by utilising institutional theory and public policy theory, definitions, measurements, patterns, and different types of institutional and policy change have not been sufficiently discussed (for example, Chen, 1995;Epstein, 2006;King, 2005;Marcussen, 2005). These intellectual silos have resulted in important gaps in the past literature. We need to learn more about how to identify different types of institutional and policy change when they occur and the mechanisms that cause them (Campbell, 2004, Chapters 2 and 3). For example, there are four modes of institutional change described by (historical) institutional theorists (Mahoney & Thelen, 2010, pp. 15-16): layering ('the introduction of new rules on the top of or alongside existing ones'), conversion ('the changed enactment of existing rules due to their strategic redeployment'), displacement ('the removal of existing rules and the introduction of new ones'), and drift ('the changed impact of existing rules due to shifts in the environment'). These modes can help public policy scholars to explain type of institutional change.
In future research, public policy scholars can bring 'state of the art' institutional theories back into the public policy field to identify different types of institutional and policy changes (Gunn, 2017). In this respect, public policy scholars may also benefit from a new taxonomy of policy change that moves beyond Hall's (1993) three orders of policy change. Indeed, Howlett and Cashore (2009, p. 42) are right in calling on students of policy change to 'distinguish between different levels of policy [ends and means]' , 'distinguish policy developments that move in slightly different directions over time but never deviate much from the status quo (policies in equilibrium), from those that move in the same (new) direction over time (cumulative change)' , and appreciate 'exogenous or endogenous sources of policy development. ' Institutionalists can also benefit from public policy theory in explaining how institutional change takes place. For example, institutionalists refer to a 'window of opportunity' as a critical juncture, but they rarely relate their analysis to the mobilisation of various actors, their conflicts and the resolution of these conflicts in domestic political economies (for exceptions, see Bakir, 2009a;Boylan, 1998Boylan, , 2001Keating, 2011;Maman & Rosenhek, 2007Quaglia, 2005a). In this context, for example, integrating the policy entrepreneurship concepts used in multiple policy stream analysis with institutional theory is likely to offer deep insights into how institutional changes take place in domestic political economies. However, although scholars offer an account of actors in institutional change, the notion of entrepreneurship in the public sector has been under-theorised. The role of the decisions and actions of individual agents in the introduction, diffusion and implementation of reform at various stages of public policy-making require further research.
In sum, past research is limited in appreciating the building of bridges between institutional theory and public policy theory to open the black box of institutional reform in macroeconomic bureaucracies. It is striking what little attention was paid in the past by institutional theorists and public policy and administration scholars to the search for opportunities for complementarity between institutional theory and public policy theory and to the study of institutional change in economic bureaucracies. Indeed, approaches bridging institutional and policy analysis may prove highly useful in understanding causes of institutional change and policy success (see Bakir, 2009a;Boin & Kuipers, 2008, pp. 42-43, 47;Peters, Pierre, & King, 2005, p. 1277Pierre, Peters, & Stoker, 2008, p. 233).
In line with our observations about the research design and theoretical approaches used in the literature, we reemphasise that the literature has limited contribution to theoretically and conceptually connected rigorous case studies. This requires further effort to turn bureaucratic reform in the context of policy entrepreneurship and institutional change into a worthwhile and viable research subject. One way of doing so could be to increase the intellectual contact and cross-disciplinary awareness among scholars of different disciplinary fields, who adopt or invent diverse and incommensurable conceptual frameworks to study different dimensions of the same phenomenon of change. Such dialogues are necessary to have better access to a growing body of literature in which theoretical explanations are discovered, or revisited and (re)tested among peers in an integrated research community.
Our study has also recognised its limitations due to its methodology. First, we retrieved our sample by reviewing articles archived in four databases of record SSCI, JSTOR, Sage and Wiley. Although these are the most inclusive, they may have omitted some relevant research. Thus, the use of additional databases might have yielded additional results. Second, we excluded monographs and edited book chapters, which might have offered theoretically informed empirical research on this topic. Third, the use of different keyword combinations generate additional results. Fourth, the filtering process is not immune from omissions. Consequently, we do not rule out that some relevant articles might have been omitted. Fifth, despite every care taken, we recognise our subjectivity regarding the classification of papers. Finally, we covered articles published in English only, excluding articles published in other languages. This is a recognised limitation. However, we believe that we adopted a rigorous systematic review protocol, complemented by our additive intervention, has reduced the likely-hood that we missed a finding that would critically alter our conclusions. Our pool covers the state of knowledge in the literature. organisational institutionalism countries reform their central banks in order to attract international capital, but that reform is driven not just by competition among states but also by imitation of social peers from networks of intergovernmental organisations note: *citations from the respective text is used where appropriate to illustrate key findings/arguments.