Do Institutions Produce Institutional Change? The New Historical Institutionalism and Analytic Innovations in the Theory of Change*

This essay discusses the problem of endogenous institutional change in the context of the new historical institutionalism. It reviews the critique of traditional theorizing on institutional change and offers a comparative analysis of innovative approaches to institutional change on the contemporary agenda of historical institutionalism in comparative political science. The analysis focuses on the logic underlying the conceptual and analytical transformations in the methodological debate of how to expand the explanatory capacity of traditional models by introducing institutional variables in them. We review quasi-parametric models, incremental change models and those models seeking to introduce the dynamic interaction between ideas and institutions. The key argument is that, over this decade, institutional change models, driven by the introduction of new concepts, new modes of theorizing, and new mechanisms of change, have undergone a remarkable analytical transformation. These innovations have allowed political scientists to deal with change endogenously.


Introduction
T he study of endogenous institutional change represents a central theme in the new research agenda produced by new institutionalism in comparative political science.The key analytical challenge for endogeneity lies in the question of how possible it is to introduce institutional factors to explain change that was traditionally considered as being produced by exogenous, i.e., non-institutional factors.Analysts working in this tradition usually underscore the fact that, in interpreting and explaining change, institutional variables are to be understood as the building blocks on which to evolve new theories and models.
Considerable analytical effort in the neoinstitutional tradition has led to a new research frontier that has been seeking to more seriously address endogeneity issues.
Institutional analysis has been faced with a remarkable revolution in the ideas and concepts underpinning theories of endogenous institutional change.The new institutionalism has offered a new generation of theories on how and why institutional elements drive institutional change.Traditional models and theories are being reviewed, while there is a rising wave in conceptual, theoretical and methodological innovations on the question of change.
This article aims to review innovative approaches to the new historical institutionalism generated since the 2000s. 1 The choice for the new historical institutionalism as a case for comparative theoretical analysis stems from the fact that it is precisely in this strand of institutional research that lie the key concerns regarding innovation and the refinement of models designed to approach institutional change-related phenomena in political science.
What is striking is the rising concern with expanding the analytical capacity of the traditional models and refining the status of endogenous institutional variables.
The models and theories analyzed herein represent plausible alternatives to go beyond the limits of the exogenous explanations traditionally considered in institutionalist research work.Emphasis on analytical innovations allows us to see which new frontiers are being expanded as a result of the increasing introduction of new concepts, new theorizing and, foremost, of a range of mechanisms that render it possible to approach the problem of change by relying mostly on institutional factors.This article seeks to more closely understand how such transformations have taken place and which conceptual and theoretical changes were produced within the neoinstitutionalist debate on the actual possibility of conferring centrality on endogenous, i.e., institutional, factors in analyzing change.
The understanding of such innovations focuses primarily on the theoretical logic, concepts and mechanisms introduced to explain change.To be sure, as this is a field of

The Traditional Explanation for Change in the New Historical Institutionalism
The analysis of institutional change is one of the core problems of the so-called new institutionalisms in political science. 2  Rothstein (1996) considers the analysis of change one of the most problematic issues on the institutional analysis agenda.Traditional approaches to the analysis of change tend to rely on functionalist, historical or cultural explanations to account for the demand for certain institutional arrangements, all of which have proven insufficient to explain why and how the latter are generated.
Theorizing on change actually depends on how to formulate models capable of generating plausible and testable hypotheses of why and how certain patterns of interplay between agents, resources and institutions are relevant in some conditions while not in others.Campbell (2006) argues there are at least three "critical frontiers" of key relevance to the new institutionalisms: (1) theorization about change; (2) the capacity to specify the mechanisms driving institutional change and (3) the challenge of embedding ideational analytical categories.The production of analytical innovations in the study of institutional change emerges as one of the essential fields for the survival and development of the new institutionalism.
To Goodin (1996) institutional change must be analyzed as arising from the isolated (or combined) action of variables associated with institutional design, selective competition and the action of exogenous shocks.Analysis is to focus on determining in which conditions certain combinations of institutional and non-institutional causes operate in order to produce change (or non-change), as well as identifying their key causal mechanisms.Traditional approaches to change in the new historical institutionalism 3 encompass two conceptual frameworks: incremental models and punctuated equilibrium models. 4While incremental approaches conceive of change as the outcome of a gradual accumulation of "small transformations and marginal adjustments to causal factors external to institutions over time", punctuated equilibrium models conceive change as "radical alterations" in incremental patterns.These models conceive of the occurrence, in specific contingencies, of great transformations, usually brought about by non-institutional, exogenous factors. 5 Traditional theorizing therefore focuses on analyses that underscore how exogenous causes bring about institutional change over time.Change is understood as disruption, a deviation, an alteration in relation to those models hinged on status quo stability, order and maintenance.Traditional explanations, whether based on incremental or on punctuated equilibrium models, draw on the isolated or combined action of two key causal mechanisms: diffusion processes (e.g. of ideas, policies, technologies etc.) and the action of legacies and path-dependence. 6 As argued by Peters (2000), Peters, Pierre andKing (2005), Pierson (2000;2004) and Mahoney (2000), legacies and path dependence, as important mechanisms for structuring institutional maintenance and stability, are directly associated with the institutional arrangements' self-reinforcing and positive feedback dynamics.Institutions are usually stable and tend to endure over time by relying on powerful institutional reproduction mechanisms.
The emergence over time of self-reinforcing outcomes that shape similar future outcomes is an important reasoning of traditional institutional change theory.Institutional inertia stems from the fact that certain sequences of events over time tend to reinforce self-enforceable institutional models that are decoupled from efficiency, legitimacy, or other performance criteria.The key to understanding institutional change is to assume that conditions exist whereby agents create and find ways to exploit institutional legacies.
Hence, the traditional analysis of change in the new historical institutionalism builds on the assumption that change is related to the occurrence of critical junctures external to institutions.These external events tend to produce conditions for a rupture with a given institutional order stemming from a transformation in agency-structure interaction 7 .Katznelson (2003) suggests that critical juncture models approach this problem by addressing the relation between agency and structure in a new way, one that is gaining prominence in comparative political science and that is underpinned on a periodization and preferences approach.Processes of change in relation to the existing institutional order are seen as outcomes that, for a number of reasons, shape distinct relationship patterns between agents (preferences) and structures (periodization). 8 The occurrence of critical junctures is a key assumption in traditional institutional bpsr (2011)  Flávio da Cunha Rezende analysis in the political science as well in sociology.Stinchcombe (1968) considers there are two basic types of explanation for institutional continuity and stability: constant causes and historical causes.The latter would be fundamental to understanding the role and meaning of critical junctures in the interpretation of change.Based on the critical junctures approach, Collier and Collier (1991) make an important contribution to the field of comparative policy studies with their analyses of labor-policy reforms for the case of six Latin American countries.Capoccia and Kelemen (2007) provide substantial analysis of the concept of critical junctures and of their relevance to the more recent analysis of change in new historical institutionalism frameworks in comparative political science.Katznelson (2003) argues that historical-comparative analyses of change in political science should focus primarily on the role of agents and their preferences in these critical junctures.Attention to these processes would shed light on the mechanisms that disrupt stability, thus making it possible to understand how change is made possible.
Critical junctures represent therefore temporal configurations that enable the realigning of agents' beliefs, preferences and strategic choices in relation to new institutional models.Katznelson (2003) considers that institution-centered approaches 9 represent fruitful ways to more adequately address the tension between periodization and preferences for the analysis of change.The author suggests that the analysis of change should be built on models that do not confer ontological superiority either on agency, as do those models centered on rational choice, or on social structures.
In assuming the existence of critical junctures, institutionalist analysis makes it possible to understand how such "atypical" events enable agents and their preferences to generate change by breaking with the status quo.Pierson (2004) pushes further the boundaries of traditional theorizing by positing the existence of four conceptions in the traditional analysis of change: (1) the critical junctures theory, where explanations are fundamentally associated with the action of external shocks; (2) theories that assume institutional malleability, where change is accounted for by the action of losing coalitions in the context of a political process; 10 (3) theories that emphasize multidimensional aspects and interactional effects, where institutional change is driven by the effects of the interaction between multiple realities or institutions 11 and, lastly, (4) theories that focus on the action of reformers and view changes as being set in motion by political agents endowed with capabilities that enable them to undertake reforms in face of typical collective action problems.
Despite some variability in key concepts and institutional outcomes, the main analytical emphasis of traditional models in the new historical institutionalism rests on a causal connection between critical junctures and the emergence of new institutions.Critical junctures are therefore conceived of as configurations endowed with causal powers capable of sharply altering structural constraints operating on social agents in a given context.In conferring causal powers on exogenous, non-institutional factors, traditional analysis would be better equipped to understand order and stability rather than change.
Institutional models are admittedly limited in demonstrating how "institutions endogenously change", i.e., building on cases that transcend agency, on the one hand, or structure, on the other.The analysis of change often tends to consider legacy-driven agency reconfigurations as decisive and, usually, to considerably underestimate the ways whereby gradual changes tend to yield sharp changes in institutions.
More recently, theories of change in the new historical institutionalism tradition are facing strong criticism.Are these theories indeed capable of embedding institutional factors in the explanation of processes of change by breaking with the strong degree of exogeneity of traditional conceptions?The answer is positive, and the ongoing debate on the neoinstitutional research agenda has given rise to mounting critique founded on, as we shall see in the next section, two pillars: the question of exogeneity and the status quo bias of institutional stability.The new generations of institutionalist studies represent, to a large extent, a response to these problems.In the next section we review and discuss the most important critiques on the traditional conceptions of change.

Critique and Limits of the Traditional Theories of Change
In this section we review the critique emerging from the new historical institutionalism research agenda on the problem of change.As mentioned earlier, despite a diversity of stances on the problem of change, the key arguments underpinning the critique rest on two themes: endogeneity and the neoinstitutional stability bias premise 12 .
The problem of endogeneity lies in the consideration that institutionalist models and theories rely on analytical strategies that confer significant causal powers on factors deemed exogenous to institutions.Changes would be caused by alterations in parameters lying outside the institutions, i.e., in social structures or in the agent's strategic choices.
The problem of resorting to non-institutional variables lies in incorporating a high level of functionalism in the theoretical framework, which then starts to treat changes in terms of functional "needs" to adapt to patterns that are efficient, legitimate, rational, or better adapted to the contexts in which they are embedded.Institutions change to adapt themselves to context-driven outcomes, the decision-making environment or changes in parameters exogenous to the institutions.If such conceptions suffice to explain change, then there would be no room for incorporating institutional variables in the model.
The second emerging critique refers to the conservative bias contained (implicitly or explicitly) in institutionalist theories.Neoinstitutionalists usually build on the premise that "to understand the effect of institutions on action and behavior, it is necessary to bpsr (2011) 5 (1) 135 129 -152 Flávio da Cunha Rezende build models on the assumption that institutions should be understood as being stable over time".This adherence to stability-based assumptions is typical of the so-called firstgeneration institutional models.Though the "stability" assumption in theoretical models proves useful in understanding order-related phenomena, it is rather poor in accounting for complex processes of change.
Exogeneity Pierson (2000) holds that a most critical point in traditional analysis of change in the new historical institutionalism lies precisely in the core assumption that changes are due to the action of exogenous causes, often understood as external shocks.Greif and Laitin (2004) consider it to be analytically inadequate to understand institutional change processes only as factors that are exogenous to institutions.To them the key problem of institutional change theories lies in the explicit difficulty of demonstrating how processes of stability and change coordinate.Lieberman (2002) considers that the various types of institutionalisms exhibit blind spots that can, on closer analysis, be understood as reductionism, prevalence of structure over agency, and exogeneity in addressing institutional change.Institutionalist models place great emphasis on the analysis of order and stability, of patterns of regularity, yet they pay little attention to complex institution-driven processes of change.
Institutional models are often limited when it comes to explaining changes brought about by the institutions themselves.When it comes to forecasting new institutional designs, institutionalist frameworks tend to focus more intensely on strategic action and on the conditions in which political agents can create and transform political institutions.
However, they are ill-equipped to show how institutions change with institutions as the starting points, i.e., driven by causes that transcend both agency and structure.
Numerous authors consider that the neoinstitutionalist traditional legitimation explanations suffer from the "infinite regress" problem of exogeneity.To explain changes in behaviors, phenomena or interest-driven processes, neoinstitutional theories generally assume a highly-correlated causal connection between institutional changes over time t and antecedent institutional variables.In pursuing this type of explanation, analysts create a new problem: which antecedent events brought about a subsequent change, and so on and so forth, endlessly.At some point of the analysis, neoinstitutional theories should consider that the sources of change might lie "outside" institutions, thus significantly affecting the endogeneity of the models.
Another major shortcoming of the traditional models lies in their inattentiveness or absence in relation to the fact that "political institutions" are multidimensional and, interests between social agents.Institutions should not be understood as "exogenous rules", as mere incentives for agents.Especially in the political realm, where institutions are laden with conflicts and ambiguity yielded by the "interaction" or "friction" between the various institutional realities.Mahoney and Thelen (2010) argue that what the new institutionalisms are lacking is the production of more consistent theories on institutional change that may account for the interplay of the endogenous and exogenous factors that engender the mechanisms that ensure the transformation or maintenance of institutions over time.It is not about reducing the analysis of change to its endogenous dimension but rather, about providing more satisfactory ways to understand change and, in particular, gradual reform processes. 13

Premises for institutional stability
The critique regarding the conservative bias of first-generation institutional theories 14 lies in the consideration of the problem generated by models and theories that assume that institutions are stable over time.This important assumption is made necessary to understand the effects of path dependence on behavior and action in shaping agential preferences, the construction of identities, strategic action, and the decision-making process (Harty 2005, 52).There seems to be a consensus among some authors working on the recent historicalinstitutionalist research agenda that emphasis on legacies and path dependence sharply constrains a more adequate analytical treatment of change.Stability premises institutional change analysis on well-known mechanisms, such as increasing returns, lock-in and positive feedbacks.However, these prove insufficient to account for complex processes of change, especially in political institutions.March and Olsen (2006) argue that in theorizing on endogenously-driven institutional changes analysts should build on the following core assumptions: (1) change represents an essential characteristic of institutional arrangements.This precludes assuming ex ante the conservative bias of institutional stability on which traditional theories are premised and, (2) that change be explained on the basis of institution-based analytical categories.Boas (2007) argues that the second-generation new historical institutionalism would be steadily parting ways with models hinged on the concept of path dependence.The analytical focus of the new studies is shifting towards approaches capable of conferring causal powers upon institutional variables, for one, and of working with models that allow conflating endogenous and exogenous factors, for another.These new approaches have been fruitful in driving attention to multiplicity and variability in change-related phenomena by building on a number of mechanisms such as layering, conversion and drift, as suggested by authors like Thelen (2000;2005).Flávio da Cunha Rezende Mahoney (2000) and Thelen (1999) consider that traditional models should be expanded for the purpose of breaking away with the notion of increasing returns and its connection with institutional change explanation.They suggest that it is heuristically positive for the new generations of institutionalists to pay substantial attention to other mechanisms such as incentives structures, conventions, social norms, and for analytical categories produced by the interaction between agents and institutions.They criticize the important fact that traditional neoinstitutionalist research had made little progress in investigating how legacies affect agents and their choices at multiple levels of analysis, especially when dealing with gradual change issues.Pierson (2004) holds that historical-institutional models of change suffer from three basic problems: (1) identifying and differentiating processes of change, in that it has thus far proved hard to accurately specify which typical change patterns should occur in certain conditions, as well as their differentiation; 16 (2) intensive reliance on case-study research designs.This problem lies in the greater likelihood that these research designs will introduce a high degree of bias and indeterminacy in the selection and (3) a primary focus on the immediate causes of institutional change.Derived from the option for case studies, studies of change are seriously limited when it comes to accounting for the important role of factors considered structural that are shaped throughout the slow processes of erosion of antecedent configurations. 17 The traditional case study literature tends to confer substantial causal power on the strategic role of actors committed to institutional change and reform, while overlooking deeper structural factors and causes acting in a diversity of conditions and contexts.There is limited understanding of the types of action that lead to a reform's success or failure over time.Case-based models are more clearly lacking when analysts seek to evolve concepts and to generate new typical mechanisms and plausible hypotheses.Pierson (2004) advances the important point that understanding the conditions necessary for the emergence of various types of change is not possible merely by building on the analysis of demands for reforms.Hence the importance of a more in-depth analysis of failures, resilience and ambiguity in responding to such demands.A major shortcoming of traditional institutional models lies in approaching change and stability as being separate.
Understanding the mechanisms that produce reforms requires, analytically, understanding the mechanisms that inhibit them.Reforms should be perceived as "transformation" of or "disruption" with the mechanisms that produce stability; hence, theories appropriate for the analysis of change should seriously endeavor in the task of unveiling the logic of institutional inertia.Mahoney and Thelen (2010) and the multiple orders model developed by Lieberman (2002).

Analytical Pathways to Institutional Change
In this section we analyze three of the main emerging theories on the neoinstitutional agenda for the treatment of change.The purpose is to understand how these models suggest new formulations for the problem of endogeneity.Three important innovation routes in contemporary institutional research are discussed: the quasi-parametric model as suggested by Avner Greif (2006), the gradual change model proposed by Mahoney and Thelen (2010), and the multiple orders model suggested by Lieberman (2002).These models have become important references for new studies of the problem of institutional change.

Gradual policy change
The second innovation case analyzed herein is the new historical institutionalism model proposed by Mahoney and Thelen (2010).Based on extensive comparative research on change, they develop the analytical elements necessary to underpin a theory of gradual endogenous change.They start from the assumption that the main analytical shortcomings of the new institutionalisms lie in the latter's excessive emphasis on exogenous elements, and that more satisfactory models for the analysis of change should balance the causal powers attributed to endogenous and exogenous factors.The model's starting point is the assumption that gradual change exists and that it can be endogenously produced.
In accounting for gradual change the authors suggest the need to consider institutions on the basis of their political elements, i.e., more directly linked to the distributional conflict arising from the allocation of scarce resources.A distributional concept of institutions opens up possibilities for suggesting that institutional arrangements are fraught with dynamic tensions and continuous pressure for change rather than stability.
When seen from a distributional perspective, institutional arrangements, political institutions in particular, are conceived of as resource mobilization and allocation arenas A main implication of this shift from traditional concepts is that change and stability are essentially intertwined.This conception shows that the new models must seriously consider institutional self-enforcement over time as a function of the agents' continuous mobilization for political resources and for overcoming conflicts and ambiguity.Institutional stability would not take place naturally, automatically, spontaneously but rather, it would arise from the actions of agents engaged in the distributional conflict for resources, directly related therefore to power fluctuations arising from the political conflict.
Gradual change analysis should not be driven solely by (endogenously or exogenously produced) variables linked to changes in the distribution of power but, fundamentally, also by questions related to rule enforcement.Agency compliance gains causal centrality in this model.
The problem of change is triggered by the degree of ambiguity yielded by the implementation of rules and decisions in the context of the distributional conflict.Farrell and Héritier (2005) is a good example of this approach to understand the role of ambiguity in policy implementation and the mechanisms of change for the case of regional integration institution.
The key category in explaining change is the ambiguity arising between institutions and change implementation.The authors contend that typical institutional ambiguities provide space for agents to interpret, debate, and contest rules, thus providing for endogenous change.Ambiguity and the problem of rule implementation would not be linked to the level of institutional formalization.The authors consider it important to take into account that, even when it comes to complying with highly formal rules, agents find space for rule interpretation, opening up considerable conflict-generating gaps over meaning, application and specific forms of institutional resource allocation.Driven by the rule-making policy and the interaction between agents and institutions, these gaps allow the emergence of new institutional models.
The authors are quite clear in specifying the "space" wherein analysts are to find the main analytical categories that matter to more accurately explain endogenous institutional change.These categories, the authors contend, are bound to occur as the result of causes situated between the rules and their implementation, i.e., variables embedded in the "politics of institutional change policies".The context as well as the political conflicts over the allocation of scarce resources plays a key role in this analysis.

Ideas and institutions
The third model of endogenous change analyzed is the one presented by Lieberman (2002), where it is critical to embed ideational and institutional analytical categories.The author's basic assumption is that of seriously considering the role of ideas to account for dynamic processes of change while further adhering to empirical realities.Institutionalist models and theories should seriously consider agency-related analytical categories by taking into account the dynamic interaction between agents, their mental models and the existing institutions.
The incorporation of agency is to be considered on the basis of categories focusing on the processes that lead to the shaping of beliefs and the structuring of preferences, and on knowledge-related elements, understandings and expectations arising from the continuous interaction with institutions, thus requiring analytical treatment.Ideas play a decisive role in the creation of new institutional arrangements and can more accurately shed light on the causal mechanisms involved in the production of change.Lieberman (2002) suggests that institutional theories, when reconfigured in accordance with an ontology of politics "as situated in multiple and not necessarily equilibrated order", can more adequately make sense of complex processes of change 25 .He considers it critical to introduce variables associated with ideas and values, thus going beyond reductionist institutional theorizing in political science, which tends to construe conflict and political cooperation as decisions by rational agents situated in a one-dimensional space on the basis of structured preferences.Ideation increases the likelihood that models will embody the multidimensional nature typical of political phenomena, as well as enabling analysts to move beyond the rational-choice premise of considering the interests and beliefs of the Flávio da Cunha Rezende agents as fixed.The key to interpreting change lies in understanding how tensions between institutions and cognitive models can, in specific conditions, lead to a reformulation of the political agents' incentives and strategic opportunities.
Political orders are laden with conflict of values, ideas, and these are decisive in rendering it possible to account for a two-tiered interaction: across institutions, and at a deeper level, between ideas and institutions.Political orders would be charged with uncertainty and ambiguity, thus dramatically increasing the potential to produce change.
This conception interprets processes of change as arising from tensions (frictions) between institutional models and ideas.
However, the great innovation in the adequate approach proposed by Lieberman (2002) lies in its dynamic conception of a more "balanced" approach for the interaction between ideas and institutions.The key problem is how to incorporate agency and its connection with institutional orders.He recommends that multiple-order models be designed that would strike a balance between institutional and ideational causal powers.It is precisely the dynamic interaction between ideas and institutions that would provide the basic clues for a more satisfactory understanding of endogenous institutional change.
The model's key hypothesis is that the likelihood of abrupt change in political institutions (as opposed to incremental change) will be the more accentuated in conditions wherein the level of tension across political orders is more prevalent.The model's core analytical category is the radical decomposition of the notion of a single political order, typical of traditional models, into its constituent parts (formal and informal), with the aim of considering to what extent they overlap, conflate, or conflict with one another, and how these interacting configurations produce change.A comparative advantage of the model is that it considers both institutional and ideational interactions as the explication's core constituent elements.In analytical terms, this situational and relational understanding of change makes it possible to consider elements associated with the specific way variables (or causes) coordinate in specific historical conditions, broadening the scope of the traditional models that deterministically emphasize the causal power of legacies.

Final Remarks
This article reviews the main critiques and analyzes substantial innovations generated by the neoinstitutionalists in the construction of theories of endogenous institutional change in comparative political science.
The construction of new models reflects, to a large extent, the production of responses to critiques of traditional new historical institutionalism first-generation approaches, which placed excessive emphasis on exogenous causes and usually contained a strong conservative Furthermore, it has also become clear that assuming that institutions change constantly does not necessarily entail that order and stability are not necessary categories to understand change.On the contrary, order and change are deeply interlinked, and it should not be assumed there exists a clear analytical line separating the two situations.
More robust theories of change must be guided by conceptions enabling a more integrated approach to these two sides of the problem.
Analytical treatment of the problem of endogenous institutional change is, in large measure, coupled with questions regarding the incorporation of agency.Far from assuming that changes are primarily structured in critical junctures, the more recent models seek to incorporate agency by drawing on conceptions that make it possible to understand how decisive certain fundamental dimensions, such as those associated with subjective and informal elements -values, mental models, beliefs, cognitivity -, are to account for how institutional reforms are produced by agents and their interactions with the institutions.
Another important question revealed through analysis is that the new theories of change have steadily reduced the level of exogeneity of traditional theories of change.The new models seek a number of alternatives toward reducing the excessive causal powers conferred on parameters external to the institutions, i.e., on social structures or on the agents' strategic choices.Eventually the explications come to combine, to some extent, endogenous and exogenous explanations.However, the main argument of the neoinstitutional theories is that the various models should consider institutional variables more clearly than traditional conceptions do.This argument is the key to considering the important fact of explaining change with institutions as the starting points.While representing just the first steps in such direction, the cases reviewed provide insights as to the future prospects surrounding the question.Future generations of work may come to contribute more decisively in the analysis of change.

Flávio da Cunha Rezende
The roads taken by these analytical innovations may be considered the first steps toward securing the advance and survival of the neoinstitutionalist tradition in political science (and, more broadly, in social sciences).It is known that the latter depends on the relative success of theory in accounting for the chronic problem of institutional transformation, which must be more creatively addressed by comparative studies that may lead to deeper comparative analysis of institutional change processes.Despite the breakthroughs, the analysis of institutional changes is still wanting as a field of reflection in the neoinstitutional tradition.
A number of political scientists working in this research tradition continue to persuasively state that traditional theories have failed in interpreting the complex problem of change.Why they fail and how to overcome the problem in constructing new theories was the basic problem addressed in this short comparative work.
Comparative reflection over the critiques and analytical innovations for the case of historical neoinstitutionalism has become relevant, in that the study of change will continue to pose one of the greatest challenges for the research agenda of contemporary political science in a world in constant change on several levels of analysis.The analyses developed in this brief study reveal that, despite significant progress in the theorizing of institutional change, there is an urge for the construction of new institutional models.
As mentioned earlier, the aim of this work on comparative theory is not to launch

Notes
1 Neoinstitutionalism constitutes a main paradigm in comparative political science.Hall and Taylor (1996) argue that the new institutionalism in political science seeks to understand three fundamental problems: institutional genesis, development and change.March and Olsen (2006) argue that neoinstitutional models are premised on the assumption that existing institutional arrangements have considerable causal power to explicate how and why institutions emerge and transform themselves.Since the beginning of the last decade comparative political science has relied on the intensive use of an institutionalist conceptual framework to analyze change, thereby prompting a great deal of theoretical and methodological debates in the field, among which is the problem of endogeneity, herein focused on more in depth.
2 While institutionalist analysis in comparative political science includes a number of variables and sub-variables, it is also widely known that this field of knowledge is characterized by a great deal of conceptual stretching and diversity, and by a plurality of terms, with regard to its core categories.Herein, however, I deliberately restrict the use of the expression "new institutionalisms" to a form equivalent to that used by Hall and Taylor (1996), who subdivide new institutionalism into three broad analytical clusters: historical institutionalism, sociological institutionalism and rational-choice institutionalism.
3 Pierson and Skocpol (2002) argue that the new historical institutionalism research in political science is focused on generating explanations for relevant phenomena and issues based on three interrelated characteristics: assuming time as a key analytical category, specifying sequences, and focusing on institutional contexts and configurations.Analysts working in this tradition are usually concerned with understanding variations relative to the occurrence of certain context-and history-bound phenomena.
4 With the seminal works of Baumgartner and Jones (1993) the analysis of change based on punctuated equilibrium-intensive models becomes very frequent in traditional public policies analysis, with a number of applications especially in the key areas of fiscal behavior and social policies in multiple research designs.Givel (2010) and Jensen (2009) offer excellent analyses of the evolution and use of these models in the field of public policy analysis.
7 Gorges (2001) claims that new institutional approaches to change fail to provide adequate explanations for institutional change.This failure is mainly due to their reliance on variables such as critical junctures, path dependency leaving institutions behind.He suggests that better theories should specify with more accuracy the conditions these variables matter as well be more precise about the causal relevance of institutions.New institutional theories should be more explicit about the causes of change and the key links between these factors and the mechanisms of institutional change.Doing so, however, he argues, the traditional approaches could mean abandoning their emphasis on the primacy of institutions in developing explanations for political phenomena.
8 The analysis of the problem of agency, i.e., of preferences, offered by Katznelson (2003), for institutional change analysis clearly departs from the agency-structure conceptualizations as conceived by Sewell Jr. (1992) and Swidler (1986).These models work with an artificial divide between agency (preferences) and structure (periodization), wherein periods are considered either as "settled" or "unsettled".Katznelson (2003) builds on the rather important assumption of the inappropriateness of conventional models in comparative social research to address change in that they reduce the role of agents.In opting for a structural-holistic ontology, this research tradition emphasizes lock-in processes to explain institutional persistence, continuity and inertia over time.He argues that more adequate analyses of change should, whenever possible, draw closer to critical realism conceptualizations (Archer, 1995), where agency and structure are analytically treated as mutually constitutive, and to the morphogenetic approach to change.9 Institutionalist models that are able to address agency and structure in a more interactionist fashion may prove fruitful in accounting for processes of change in political science.Embedding analytical categories that may render it possible to understand the manner whereby choices and preferences are shaped, in addition to the institutions' distributional character, the action of social networks and cognitive models, are potential pathways to go beyond the often assumed notion of external shocks advanced by the first generation of institutional-historicist theories.
The existing institutional models tend to create the likelihood (and not the determination) of certain equilibria to occur wherein the role of agents embedded in these arrangements, as well as their preferences, is critical.
10 This argument is originally attributed to William Riker, who considers it important to think that institutional change in politics constitutes a possibility set in motion by the organized action of groups and coalitions who suffer losses in the institutional system in relation to winning groups.Democracies represent a clear example of malleable political systems insofar as they make it possible for groups defeated in a given election to win in a future election.
11 This formulation is included in the original argument developed by Orren and Skowronek (1994), who hypothesize change in political institutions as driven by the "abrasive effects" of the various political institutions in interaction.The thesis is that, as institutions interact, they generate tensions and opportunities that yield the conditions for change.
12 In an influential article in the modern debates in the new institutional agenda of change, Przeworski (2004a) express his analytical ceticism about the potential and primacy of new institutional explanations of change in political science.Przeworski (2004b) discusses this ceticism for the case of institutions and economic development making the claim for endogeneity and the key problem of the causal "primacy of institutions" that supports the basic claim of institutional theory: institutions matter.
13 They consider that, when "institutions are treated as causes, scholars are too apt to assume that big and abrupt shifts in institutional forms are more important or consequential than slow and incrementally occurring changes".Recent works on institutional change, as those produced by Pierson (2004), Streeck and Thelen (2005) and Caporaso (2007) would be better equipped to explain gradual change by building on models that distance themselves from the exogenous-driven, punctuated equilibrium approaches.
14 The debate agenda of the first generation of institutional studies was centered on the effects generated by institutions or on the explanation of institutional order and stability.Analysis of endogenous institutional change (i.e., with institutions as the starting points) is held to be critical for the research agenda of the so-called second institutionalist generation in comparative political science.
15 Thelen (2003, 37) emphasizes that one of the decisive problems for the advancement of theorization on institutional change lies in the analysts' ability to more accurately identify factors and conditions driving (or inhibiting) the emergence of specific types of change in institutions.The importance of accuracy in necessary conditions in comparative research is suggested by Goertz (2005).
16 This fact is particularly problematic given the crucial importance of positive feedback in institutionalization processes.If agents adapt to and align with existing institutions, the equilibria derived will self-reinforce over time.In face of this trend, change is likely to occur if the models consider the occurrence of slow erosions or the reconfiguration of antecedent conditions for institutional reproduction.Reform-driven analyses may commit the critical error of overlooking these slow erosions as causal processes and placing greater emphasis on the final stages of change.This fact is usually accounted for in political science by the known argument that the creators of new political institutions, faced with the problems of uncertainty and intergenerational choice inconsistency, usually conceive institutions (and public policies) to stand up to the "uncertainties" of politics, just as suggested by Moe (1990).New institutions are usually constituted of rules devised to face and endure temptations for control by and reversal to antecedent models.In this sense political institutions are typically known for their change resistance.
18 Tsebelis (2002), in his well-known theory of veto players in comparative political science, advances the key argument that differentiation of the various political and institutional systems should be based on the configuration of their veto players, and that change is fundamentally derived from the collective action of these specific agents.He expands the concept that institutions are rules governing the action of agents and suggests the notion that institutions structure the decision-making sequences of agents endowed with veto power.Within institutions there exist agents with the power to shape the decision-making agenda, the agenda-setters.He hypothesizes that agenda-setting powers are inversely related to institutional stability.If the theoretical models specify veto players' preferences, the stance of the status quo, and the identity of the agenda-setter, they will render it possible to predict change in an accurate way.
19 Miller (2000) classifies these important veto players when analyzing processes of "institutional reform policies" as being of the "self-referencing" type.The emergence of these institutional veto players is contingent upon the type of reform policy under analysis.Moreover, not all veto points are "self-referencing", yet they are awarded such status as a function of the various types of institutional change.In some cases reformers can be created by referendum, weakening the capacity of political elites to block changes that are potentially damaging to their interests.
20 Deep equilibrium analysis is focused on understanding the underlying logic of why and how the interplay between certain factors tends to yield (in certain contexts) institutions situated at a point where the likelihood of reversal over time is highly unlikely.The understanding of these factors may provide a clue to understanding change.
21 Institutional change analysis would be premised on the key assumption that change is a function of interactional effects across multiple institutional realms.Pierson draws our attention to two important interactional effects when analyzing change: a) the occurrence of inter-institutional, self-reinforcing interactional patterns and b) inter-institutional loose coupling.In the first case, it is assumed that changes in certain institutional arrangements generate a high likelihood of occurrences in complementary institutions and, in the second case, ambiguities derived from power distribution across institutions are assumed which are fundamental in understanding change.
22 Conceptually, parameters represent factors deemed as exogenous to games (strategic interactions), that is, they cannot endogenously change in the short run.Quasi-parameters, in turn, are variable elements, endogenously produced by the outcomes of games over time.
Institutions yield outcomes that alter the dynamically fixed elements that govern and structure interactions between agents.It is also worth underscoring that discerning between parameters and quasi-parameters is sensitive to empirical observation and the objectives of each analysis being considered.
23 Institutional reinforcement processes occur over time when changes in quasi-parameters yield self-enforcing associated behaviors for a large set of situations.Self-enforcing institutions are undermining when changes in quasi-parameters decrease the set of situations wherein behaviors associated with the new conditions are sharply reduced.24 This type of mechanism is directly associated with cases that usually characterize the socalled punctuated equilibrium, typical of historical institutionalism's incremental approaches.Punctuated equilibrium models suggest that episodes of change be considered as radically deviating in relation to marginally-adjusted, organized, stable processes.
25 Based on neo-weberian claims, Roland (2004) proposes an interationist approach for change.His model understands change based on variables produced by the continuous interaction between slow-moving and fast-moving institutions.His claims for understand interactions between different institutions can provide reasonable theories about punctuated and gradual change.His model suggests a key difference between slow-moving and fast-moving institutions.
The former generally change slowly, incrementally, and continuously, whereas the latter are more given to rapid, discontinuous change in large steps.His neo-weberian claims confers causal powers for analytical categories related with context and culture.
of tensions, imbalances, clashes of values, preferences, and conflicting Pierson (2004) suggests four clusters on which analysts should focus their attention to build analytical categories that would make it possible to account for the mechanisms bpsr (2011) 5 (1) 138 129 -152 Do Institutions Produce Institutional Change?The New Historical Institutionalism and Analytic Innovations in the Theory of Change inhibiting change: problems related to coordination, action by veto players, asset specificity and the outcomes of positive feedback.Acting singly or in combination, these factors render reforms somewhat elusive. 17For the study of the analysis of change should consider it essential to understand the structure and configuration of the preferences of agents with veto power in a given institutional model. 18Agenda-setters are usually decisive in that they hold considerable clout over the capacity to protect institutions from proposals for change at the political level. 19Pierson (2004) suggests that theoretical breakthroughs in change analysis would be a function of addressing five problems on the research agenda: a) the rationale underlying the production of deep equilibria; 20 b) pathways of change; c) interactional effects among and between institutional realms; 21 d) gradual change and e) policy making.These critiques of traditional theoretical models are the building blocks on which a new set of institutionalist models are evolving that are better equipped to analyze change.The next section focuses on the comparative analysis of some typical models of this second generation of scholarly who introduced substantive analytical innovations for the treatment of change in the new historical institutionalism.Three specific cases are analyzed: Greif's (2006) endogenous change quasi-parameters; the gradual change theory proposed by

Greif ( 2006 )
makes use of a fruitful combination of repetitive game theory and new historical institutionalism to account for endogenous change analysis.He suggests the introduction of two interrelated concepts, quasi-parameters22 and institutional reinforcement, to overcome the acknowledged constraints of those approaches hinging on the notion of institutions as equilibria external to agents.consists of holding that parametric elements have a dual analytical role in institutional change analysis.In doing so the author goes beyond the traditional notion that assumes that rules of the game as "equilibria" over time suffer from a conservative bias in relation to order.Parameters can be considered differently depending on the particular time horizon.Analysts should note that institutions can be considered as parametric (fixed, exogenous) if we analyze their self-enforcing property in short temporal horizons; but as quasi-parametric, i.e., varying in a long-term perspective.Thus, quasiparameters represent institutional elements that can be endogenously altered and are decisive in understanding how institutions change without resorting to exogenous elements.The model presupposes that change is directly associated with the transformation of quasi-parameters and processes of reinforcement (or undermining) of the associated beliefs and behaviors of agents in relation to the existing institutional arrangements.Changes in quasi-parameters can, depending on the conditions, generate mechanisms that either undermine or reinforce institutions over time.23 Yet, it is critical to consider the manner whereby the agents' beliefs and associated behaviors change over time.The institutional change analysis proposed is premised on the assumption that contextbound elements affect the self-enforceability of the agents' beliefs in the institutional models.Understanding how the process whereby beliefs and behaviors associated with institutions are self-enforcing (or self-undermining) over time gains relevance.Institutional undermining mechanisms are largely associated with the mode whereby incentives (sanctions/rewards) are created by the institutions themselves and interpreted by the agents.Institutional reputation mechanisms are typically undermining over time when the agents perceive a decrease in the future value of incentives associated with certain institutional behaviors.The model holds the idea that endogenous institutional change takes place when the connections between the expected beliefs and behaviors of agents over time become fragile.Institutional undermining processes may reveal how self-enforcing behaviors come to be transformed by quasi-parametric change.An arguably necessary condition for endogenous change is that new institutions should cause a reduction in the self-enforceability and reinforcement of agents´ behaviors associated with existing institutional arrangements.The specific mechanisms for endogenous change proposed by the theory are discernible by the way the connection between agents and their perception of quasi-parametric changes is organized.The model specifies three possibilities based on the visibility of the changes.In the first set there is scarce and inconsistent evidence of quasi-parametric change, which is barely recognized by agents.Institutional change mechanisms are bound to be associated with the behavior of agents in face of change-related risks; or, conversely, with the emergence of agents with a better knowledge of the context of change bpsr case, institutional undermining processes stem from a strategy based on the radical rejection of strategic behaviors caused by the agents' recognition ex post that the new behaviors are more adequate to the new context. 24Lastly, changes in behavior associated with institutions can occur when the agents perceive and understand the parametric changes.Agents start considering that behaviors that had previously been strategic become progressively less self-enforcing, while intentional and gradual institutional change is observed.Some typical institutional reform mechanisms are the generation of new behaviors, the specification of new normative models and the creation of new organizations.Endogenous change can be analytically approached by assuming it is driven by marginal changes in quasi-parametric values.Institutional change does indeed combine exogenous and endogenous elements.Generally, institutions do not reflect changes occurring exogenously but, rather, the manner whereby reinforcement and undermining processes are driven by agents and their interactions with belief models on institution over time.Accordingly, the author suggests not dissociating the study of institutional change from stability analysis.The study of stability represents a case in which quasi-parameters are considered as fixed and exogenous.In order to study change, analysts should more closely examine how such processes enable, or not, change to be introduced, by considering longterm variation in the quasi-parameters.
tensions and conflicts gravitate.Adherence to this conception allows the authors to distance themselves from conceptions of change that build on analyses favoring the notions of stability, self-enforceability and automatism.Change policies would be more adequately analyzed if premised on analytical categories based on the dynamic elements represented by power and conflict, and on agential tensions arising from resource allocation.Institutions are either constantly changing or vulnerable to change.
therefore a significant innovation in relation to traditional approaches insofar as it emphasizes the need to more deeply understand how agents organize themselves and the dynamic mechanisms whereby distributional coalitions are formed in rule-enforcement processes in a given context.By placing emphasis on the analysis of an institution's policy connection, the authors create analytical possibilities for more serious thought on the distributional effects associated with institutional models by drawing on the study of agency (values, beliefs, preferences, identity), typical coalition patterns, and, most importantly, of the political conflicts that drive and structure the processes of production of institutional change in multiple contexts and conditions.Hence the need to understand change not as abstraction but as emanating from a most fecund dialogue with empirical cases and comparisons across institutional reform processes the most variegated in context and conditions.
studies have managed to steadily move forward in reconstructing concepts, mechanisms and arguments that can address change in a manner both more sophisticated and more befitting to its complexity.As reviewed, there has been significant progress in the analysis of change in the new historical institutionalism.The development of plausible alternatives to account for change is directly linked to the development of new forms of apprehending change and to the embedding of new categories that are more sensitive to the problem of institutional dynamics.A review of these innovative models reveals that institutions are more dynamic than assumed by traditional approaches and exhibit a wide range of mechanisms that render the connection between theory and empirical reality much more complex.The institutional change models that have emerged are more sensitive to this diversity and admit of institutional incremental changes.
any new theory on endogenous change.Its contribution is merely that of showing the main critiques that have affected the design of new models and theories for the "case" of the new historical institutionalism.Analysis renders it possible to understand a remarkable analytical transformation enabled by the gradual introduction of concepts, analytical categories and a new set of causal mechanisms that have made it possible to approach processes of change on the basis of institutional factors.The acquisition of these new elements allows those researchers interested in analyzing institutional change to become more attentive to recent developments in terms of new theoretical formulations and designs.Designing more satisfactory theories of change requires, fundamentally, the analysts' skill in satisfactorily "dealing" with the critique embedded in the contemporary debate and, more broadly, the creation of new models that go beyond those herein reviewed.Lastly, it is worth underscoring that the development of more refined theories must be heedful of the questions of how to conflate traditional elements with the question of endogeneity.A theory of institutional change should not, for example, completely depart from social structures or exceed in embracing questions of agency.The success of a theory ultimately depends on the models' capacity to foster fruitful "integrations" between traditional models and, fundamentally, to determine in which cases and conditions these bpsr combinations can be undertaken.Greater attention to institutional elements themselves calls for a steady reduction in elements exogenous to institutions, as the former cannot account for change in its entirety.
political institutions are conceived of as very hard to undergo change or reform.