The idea that institutions matter, nowadays, is commonplace. This should not be taken to imply that widespread agreement exists among the social sciences or even within disciplines about the nature and the role of institutions. Following the process of differentiation between the social sciences in the nineteenth century, each discipline has seen its own theoretical development, which has colored each perspective on institutions. Now that the notion of institution is widely recognized as a foundational concept in social theory, and an important subject of research in sociology, economics, history, politics, organizational theory, etc., it proves a difficult task to fit the various bits and pieces together. The fact that despite renewed attention to the institution concept, Hodgson (2006) recently published an article with the title "What are Institutions?" may be a case in point.
One important reason, we believe, for this disarray is the variety of perspectives on the "polarity between the individual and the social" (Burman 1979, 374-5), or between agency and structure. This issue concerns conflicting views on how social phenomena are to be studied and explained. The agency point of view takes the explanation of social facts to be rooted at the level of the individual, i.e. an explanation is to be built from the (given) preferences, expectations and motives or behavior of rationally acting individuals. It is argued that social phenomena are to be understood as the result of individual actions oriented toward the (expected) actions of others. This point of view is handsomely captured by Elster, when he claimed that "there are no societies, only individuals who interact with each other" (1989, 248). Objecting that social phenomena cannot be reduced to the properties of the individual parts but ought to be studied within the social system in which they occur, adherents of the structural approach emphasize that society is a reality sui generis. Individual behavior, interdependent and interwoven with behavior of others, unintentionally gives rise to structured regularities in processes, relatively autonomous with regard to the intentions and preferences of individuals. As Durkheim expressed this point of view: "The first and fundamental rule is: Consider social facts as things" (Durkheim  1947, 14).
This paper proceeds as follows. The setting for the agency-structure debate is briefly laid out in the first section without any intention to discuss exhaustively the vast literature on this central issue in the social sciences. The next section elaborates on how the concept of the institution is implicated in this discussion, introducing the notion of tension, perceived to exist by actors between institutions and their legitimation in the socio-cultural values subscribed to by a society or community. Given that such tensions need to be activated to induce institutional change, a following section argues the importance of the three constituent elements of order and institutions--interests, power and social-cultural values--in this respect. Taking social-cultural values and the legitimation these supply for institutional settings as a point of entry in our analysis of institutional change, the next two sections explore how perceptions of tensions that arise between institutions, concrete practices and behavior may be seen to produce such change. The sixth section discusses three types of tensions, drawing out the conditions of institutional change within our theoretical perspective, while section seven offers two exemplary cases. A concluding section ensues.
Agency and Structure
Given that in the social sciences opinions range from the view that structures or institutions "determine" individual behavior on the one end to the idea that social structures or institutions are the unplanned outcome of the interplay of individual behavior at the other end of the spectrum, divergent explanations of institutions and institutional change abound. Prompted by a growing dissatisfaction with the unrealistic nature of the assumption that economic processes take place within a cognitive, motivational and institutional vacuum, from the seventies onward the notion that institutions should be incorporated into economic analysis gained support. With the rise of new institutional economics, built on the same foundations as orthodox economic theory, it is generally assumed that institutions are to be explained as the outcomes of purposeful actions by instrumentally-oriented individuals.
Even though sociologists emphasize that social action has a logic of its own, it would be a misrepresentation to suggest that the differences between economists and sociologists may be cast in terms of agency versus structure. The agency--structure issue already figured as a bone of contention between sociologists Durkheim and Weber as well. While Durkheim ( 1947;  1964) developed a structural (collectivistic) approach and referred to sociology as the theory of institutions, Weber ( 1964) adopted an individualistic perspective, arguing that sociology is about meaningful action, and hence individual action, as only individuals can give meaning to action.
The same issue draws dividing-lines between old and new institutionalism in economics. New institutional economics sets the stage for analysis by assuming that institutions emerge out of the interactions of cost minimizing agents within a set context of institutions designed such that societally optimal outcomes result. Structure and agency thus having been predefined, the emergence of different types of economic institutions is explained in terms of different characteristics of transactions. Notwithstanding the behavioral and situational amendments to include institutional arrangements in economic analysis, the agency-based approach of neo-classical explanation has remained intact. Presented as solutions to problems of organization, institutions are seen to enhance efficiency, create predictability and reduce uncertainty, imposing barriers or constraints on behavior that affect the range of options open to the individual (Ingram and Clay 2000).
Adherents to evolutionary/institutional economics, in which the research program(s) of institutionalists like Veblen, Commons, Mitchell and Ayres have been revived, emphasize the role in economic decision-making of pre-existing habits and rules, values and cognitive frames and typically argue that the individual does not make decisions in isolation, but decisions are predefined by the social context in which the agent finds herself. Institutions have emergent properties that cannot be explained in reference to individual attributes or motives.
The same difference in explanatory perspective surfaces if views on institutional change are compared. An example of a structural approach is Veblen's account of institutional change as the result of the dynamics of technological change. An opposite view has been developed by North (1990), identifying changes in relative prices and changes in preferences as major sources of institutional change that drive political or economic entrepreneurs to actions that alter the institutional framework.
In view of this, Hay and Wincott have argued that "if institutionalism is to develop to its full potential, it must consider the relation between structure and agency" (1998, 951). As long as the structure-agency struggle continues, mainstream economics seems content to await future developments, in the meantime practicing an agency-based perspective. Williamson, for instance, has shown in a number of papers (1993; 1998; 2000) his growing awareness of the importance and influence of sociological factors such as values, norms and legal rules on the selection of (governance) structures, but assigned the task of analyzing and explaining this environment to other social sciences, boldly asserting that the insights from other social sciences may be fitted into his framework.
While mainstream economics purportedly contends itself with an agency-based perspective (cf. Davis 2003), the institutional approach has taken up the challenge, tending to a central position between the extremes of either agent or structure, rather than emphasizing their interactive relationship. As Lawson (1987, 969) put it: "individual agency and social structure and context are equally relevant for analysis each presupposes each other. Thus any reductionist account stressing analytical primacy for either individual agents or for social 'wholes' must be inadequate." The same point is made by Hodgson (1988).
This, however, is easier said than done and often either contributions to the structure-agency debate prohibit possibilities to give substance to the idea of institutional change, or the issue is likely to resemble the well-known Russian dolls in the sense that contributions to the theory of institutional change tend to reproduce the agency-structure debate at a different level of aggregation.
Criticizing theories that either take the individual/agent or society/structure as their point of departure, Giddens developed his theory of structuration to overcome the shortcomings of both these polar views. Central to structuration theory is the concept of "duality of structure" or the notion that agency and structure are to be considered as inextricably interwoven. Agents in interaction use the rules and resources that make up structures, and in doing so, they produce and reproduce such structures. "The constitution of agents and structures are not two independently given sets of phenomena, a dualism, but represent a duality ... the structural properties of social systems are both medium and outcome of the practices they recursively organize" (Giddens 1984, 25-6). Structure, rules, resources and roles only exists in the day to day routines of people (Jackson 2005), while interaction between agents is framed by the rules and...