SOME REFLECTIONS ON HOW INSTITUTIONS INFLUENCE STYLES OF INNOVATION

by J. Rogers Hollingsworth

University of Wisconsin (Madison)

 Paper presented to the Swedish Collegium for Advanced Study in the Social Sciences (SCASSS) 26 September 2002

 Comments will be appreciated

Abstract

The study of institutions and innovativeness is presently high on the agenda of the social sciences. There is increasing concern with how the institutional makeup of societies leads to variation in their styles of innovation. However, before there can be significant advance in the study of this problem, it is important that we have a better understanding of how to do institutional analysis. Every social science discipline – with the exception of psychology – has at least one distinctive strategy for doing institutional analysis. And it is because of the lack of consensus as to the appropriate boundaries and content of institutional analysis that we have limited ability to make theoretical advances in understanding how the institutional makeup of a society impacts on its innovativeness. Recognizing that this is a serious problem for the social sciences, this paper attempts to structure the field of institutional analysis and takes the first steps in relating it to the study of a society’s style of innovativeness.

Keywords

Incremental and radical innovations; institutional arrangements; institutional complementarity; institutions; organizations; path-dependency; social system of production, varieties of capitalism

INTRODUCTION

This paper addresses two issues which are currently high on the agenda of the social sciences: (l) why do societies vary in their style of innovativeness? and (2) how should we go about doing institutional analysis? While these are often treated as separate issues, this paper makes some effort to relate the two subject areas to one another.

For some years, a great deal of social science literature has argued that a country’s innovative capacity is linked to its international competitiveness (Landes, 1969, 1998; Nelson, 1993). Yet we are greatly lacking a theoretical understanding of why countries vary in their innovative styles. Why, for example, do some advanced industrial countries, time and time again, make radical breakthroughs in basic and applied science and develop radically new products – and even new industries? And why do other countries rarely make any radical breakthroughs in basic and applied science or in product development, but continuously make incremental advances in knowledge, improve upon newly developed products, and even become the dominant producers in these market segments?

Why the innovative styles of countries vary is a complex problem. But much of the variation in innovative styles across societies is due to their institutional configuration. Institutions may either constrain or facilitate innovativeness (Hage and Hollingsworth, 2000; Edquist, 1997; Langlois and Robertson, 1995), but at present we do not have a good understanding of how the institutional makeup of a society is associated with its style of innovativeness. This is due to the fact that the comprehension of the institutional structures of societies is in a state of confusion. Hence a major argument of this paper is that before we can understand how the institutional configuration of a society influences its style of innovativeness, we must first identify the various components of the institutional makeup of a society and understand how these components are related to each other. The latter part of the paper explains how the institutional makeup of a society is related to its style of innovativeness.

At present, it is difficult to relate institutional analysis to innovativeness because our theories of both areas are poorly developed. Even so, we are now at a strategic moment, because we can build on a rich body of literature on institutionalism and innovativeness in order to advance our theory of each and to integrate the two fields.

For some years, much of the literature on technological innovation has emerged from a focus on the firm (Dosi, 1988; Fransman, 1994; Langlois and Robertson, 1995; Whitley, 2000). For example, Alfred Chandler’s great corpus of work (1962, 1977, 1990) has tended to emphasize how the success of a firm’s technological innovativeness – both across countries and over time – has been influenced primarily by whether it has the right strategy and structure. For Chandler, firms which have had the right strategy and structure have ended up having the organizational capabilities which permitted them to have the economies of scale and scope to develop cost advantages over their competitors (Teece, 1993). Chandler’s work has had a profound influence on the management literature of the past couple of decades. Hence, in the Chandlerian tradition, much of that literature has suggested that the key to understanding the competitive advantage of firms is to identify their strategies and organizational structures.

Over time, another literature has emerged which emphasizes the importance of the institutional environment of organizations for understanding why firms in some countries excel in some industries but not in others, and why the firms in a specific country may excel in a particular industry at one time but may eventually lose that advantage (Landes, 1969, 1998; North, 1981, 1990; Hollingsworth, Schmitter, and Streeck, 1994). More recently, Richard Nelson and other scholars have been advancing this literature by attempting to integrate the literature on institutions, organizational strategy, and technological innovation (Nelson, 1994, 1995a, 1995b, 1996; Mowery and Nelson, 1999; Murmann, 1998; Arora et al., 1998; Hollingsworth and Hage, 2000).

Since we do not presently have an adequate theory on how institutions, organizations and technologies co-evolve, we are not at a stage to test a set of formal hypotheses which flow from some well-defined model. Hence a number of us have independently and collectively been developing descriptive studies of how institutions, organizational capabilities and technologies co-evolve so that particular societies and organizations at specific moments in time excel in particular kinds of innovations. The goal of this kind of work has been to develop, by working inductively, a better understanding of the processes of how technologies, organizations and institutions co-evolve across a number of industries and countries.

A variety of endowments in the institutional environment provide economic actors/organizations with initial advantages or disadvantages for particular types of technological activity (Murmann, 1988: ch. 7). But over time everything is dynamic, and the larger global and institutional environments, the capabilities of organizations and their performance all co-evolve and feed back onto one another. However, institutional environments differ widely from one society to another, and the successful firms and organizations are those which can best adapt their activities to the institutional environment within which they are embedded. Once a number of organizations in a particular industry are successful, however, they may be able to engage in collective action to modify their institutional environment in order to enhance their innovativeness and their technological competitiveness.

Figure 1   Institutional environments, organizations and innovativeness

As these comments suggest, the literature cited above provides some potential for understanding why societies are more competitive in some sectors than in others. However, to extend and complement this work, we need a greater understanding of and consensus on the meaning of institutions and the institutional environment of organizations. Hence, this paper is written in the spirit of complementing the agenda described above. One of its goals is to develop a map of what can be called the terrain of institutional analysis. It assumes that until we have a map of this terrain, scholars working in the broad field of institutional analysis will not comprehend how their work interrelates.

Institutional analysis is now rather high on the research agenda of the social sciences. However, we need to be aware of the obstacles confronting us as we attempt to advance an agenda of institutional analysis. There is no consensus as to what is meant by institutions or by institutional analysis. These terms are very widely used, but they are used with different conceptualizations, and the scholars who use them share little common ground. Until scholars have some consensus about the meaning of the concepts they use, their potential to bring about effective advancement of knowledge is somewhat limited. Thus, the widespread interest in several academic disciplines in the concepts ‘institutions’ and ‘institutional analysis’ may well promise more than it can deliver, given the organizational and disciplinary fragmentation of contemporary universities.

There are many different approaches to the study of institutions (Nelson and Sampat, 1998): there are the new and the old institutionalisms (Stinchcombe, 1997; Hodgson, 1998; Langlois, 1986, 1989); there is historical institutionalism (Steinmo et al., 1992; Zysman, 1994; Immergut, 1998; Katznelson, 1998); and several of the social sciences have their own distinctive approaches to the study of institutions (Hall and Taylor, 1996; Hodgson, 1988; Eggertsson, 1990; Finnemore, 1996; Scott, 1994; Calvert, 1995; Hechter and Kanazawa, 1997).

The following comments reflect some of the confusion in utilizing the concept ‘institution’. Nobel laureate Douglass North in his book Institutions, Institutional Change, and Economic Performance (1990: 3) defines institutions as ‘rules of the game in a society’. To North, institutions are constraints which shape human interaction and the way that societies evolve through time. On the other hand, Andrew Schotter argues that institutions ‘are not rules of the game’. Rather, institutions are the behavior that follows from rules. Briefly, he is concerned with what actors do with rules, but not with what the rules are (Schotter, 1981: 155). Many other examples might be given to illustrate the heterogeneity of approach to institutions and institutional analysis. Even if scholars were to agree with North that rules and norms are institutions, they would not necessarily agree on what a rule is. Shimanoff, for example, has identified more than 100 synonyms for the concept ‘rule’ (Shimanoff, 1980: 57; Ostrom, 1986: 5).

Another critical issue in the institutional literature is the relationship between institutions and organizations. North (1990), following from his definition of institutions, argues that institutions and organizations are distinct entities. Which organizations come into being and how they evolve through time is influenced by a society’s rules and norms, that is by its ‘institutions’. On the other hand, a number of recent organizational sociologists see very little difference between institutions and organizations. For them, rules and norms are institutions which unfold in tandem with organizational structures and processes, and changes in organizational forms internalize and reflect changes in the society’s rules and norms. Using this perspective, a whole subdiscipline within sociology called the ‘new institutionalism’ in organizational analysis has emerged. The ‘institutionalist’ perspective on organizations assumes that the kinds of organizations which actors create are dictated by the cultural norms and rules in which they are embedded (Powell and DiMaggio, 1991; Zucker, 1988, 1991).

The importance of this disagreement about institutions is obvious. If institutions are so critical for understanding our societies, it is important that we come to some systematic consensus as to what institutions are, and how they influence social actors and the organizations that they create. If we cannot do so, we risk talking past one another, and losing the opportunities for cumulative knowledge based on articulation of widely shared theoretical understanding.

We need not only conceptual clarification as to what institutions are, but also greater consensus as to how to study institutions. No scientific field can advance very far if the practitioners do not share a common understanding of the key concepts used in their analysis (Ostrom, 1986: 4). But with our universities and academic associations so fragmented into different disciplines and into various subspecialties within disciplines, it is difficult to advance the theoretical agenda of institutional analysis within the academy. Indeed, the disciplinary fragmentation of the modern university is a major barrier to the theoretical advancement of the study of institutions and innovations as well as most other hybrid fields of research (Hollingsworth, 1984; Hollingsworth and Hollingsworth, 2000). And it will be only as a result of effective communication across diverse fields of knowledge that our study of institutions and innovativeness will be effectively advanced.

MULTIPLE LEVELS OF INSTITUTIONAL ANALYSIS AND INSTITUTIONAL CHANGE

There are innumerable signs that we are living in a time of great institutional change: the demise of the Soviet empire; the processes of European political and economic integration; rapid transformation in parts of the global economy; the disintegration of the family structure; the weakening of voluntary associations and the decline in political participation in a number of advanced capitalist societies; the weakening not only of welfare states but of the autonomy of nation states as well. European law is superseding national law and is even changing complete national legal systems. The list could go on and on.

Even though scholars discuss institutional change at length, their ability to measure the rate of institutional change is very limited. And more crucial than the limited ability of scholars to measure institutional change is their very limited understanding of how to build new institutions. One of the reasons for these shortcomings is that the social sciences are deficient in a theory of institutions. The building of new institutions and redressing the decline of some of the most important institutions of our societies are among the most important problems of our time. If we are to advance in the development of a theory of institutions, we need to work collaboratively across the social sciences, and we need to define the parameters of institutional analysis.

First level of analysis

This paper attempts to make some modest contribution to outlining the parameters of institutional analysis. At the outset, we need to recognize that when we engage in institutional analysis, we must be sensitive to multiple levels of reality. As suggested above, most scholars who engage in institutional analysis do not participate in any coordinated activity with each other, and their activity is fragmented into a variety of disciplines and subdisciplines. To establish some coherence to the field of institutional analysis, we need a map of the field so that those working in one area can see where their research fits in relation to other areas and other practitioners on the map.

Table 1 presents such a map, with multiple levels at which institutional analysis occurs. Theoretically, each of these areas on the map is interrelated with each other level. However, the various areas/components on the map are arranged in descending order of stability or permanence. Those components at higher levels of reality are more permanent and durable, while those at lower levels change more rapidly.

Were scholars doing institutional analysis able to reach some consensus about where their own work fits in relation to all other practitioners in the field, there would be increased potential for all practitioners to communicate with each other. By analogy, once geneticists, crystallographers, biochemists, etc. had a good understanding of how their investigations were related to each area of molecular biology, the field quickly was able to make theoretical advances (Judson, 1979).

At the first level, there are the basic norms, rules, conventions, habits and values of a society. These are the most fundamental properties of institutions and are the most enduring and resistant to change. Rules, norms, conventions, etc. are institutions, but are only one component of what constitutes institutional analysis. As Burns and Carson (2002) point out, most human activity is organized and regulated by norms and rules and systems of rules. These concepts are extremely important for institutional analysis, as they exert the greatest influence on the nature of the components of institutional analysis at the next four levels which are depicted in Table 1. In most forms of social analysis, it is extremely important that we understand the social and cognitive conditions that lead to compliance or non-compliance with rules, and the conditions which lead to changes in rules.

Table 1   Components of institutional analysis

1   Institutions = norms, rules, conventions, habits and values (see North, 1990; Burns and Flam, 1987).

2   Institutional arrangements = markets, states, corporate hierarchies, networks, associations, communities (Hollingsworth and Lindberg, 1985; Campbell et al., 1991; Hollingsworth et al., 1994; Hollingsworth and Boyer, 1997).

3   Institutional sectors = financial system, system of education, business system, system of research (Hollingsworth, 1997; Streeck, 1992).

4   Organizations (Powell and DiMaggio, 1991).

5   Outputs and performance = statutes; administrative decisions, the nature, quantity and the quality of industrial products (Hollingsworth, 1991, 1997); sectoral and societal performance (Hollingsworth and Streeck, 1994; Hollingsworth et al., 1990; Hollingsworth and Hanneman, 1982).

Note: The five components in this table are arranged in descending order of permanence and stability. That is, norms, conventions, etc. are more enduring and persistent than each of the other components of institutional analysis. Each component is interrelated with every other component, and changes in one are highly likely to have some effect in bringing about change in each of the other components.

 The approach to the study of institutions employed here argues that norms, rules, habits, conventions and values both reflect and shape the preferences of actors. Norms, rules, habits, conventions and values influence who and what are included in different types of decision making, how information is processed and structured, what action is taken (Shepsle, 1986, 1989). It is through norms and rules that behavior is judged to be democratic, fair or egalitarian.

Burns and Flam (1987) point out that in any society there are multiple rule systems. Within a family there are rules for decision making, often quite different from rules and norms for decision making for a professor in a classroom, or for the customer in a bank. Despite the heterogeneity of rule systems, there are meta rules and norms which encompass lesser rule systems. Otherwise there would be such contradictory rule systems that a society would be paralyzed. The existence of meta rule systems permits different rule systems to intersect with each other so that ambiguities can be resolved. The greater the pluralism and complexity of a society, the more ambiguity there is about meta rules and norms in a society, and of course, all ambiguities never completely disappear. Overall, the degree to which separate rule systems are interlinked is an empirical problem. There are different sectors, groups and interests pursuing their own action logic, but through higher-order meta principles and rules there can be order, consensus and coherence in a society. It is through a set of meta rules that class and ethnic conflict in societies are contained. (For elaboration, see Burns and Flam, 1987 and Archer, 1996.)

In many respects, our understanding about norms, rules, habits, conventions and values influences our perspective on how societies are constructed and how they change. Many new institutionalists’ (Posner, 1992; Schotter, 1981; Williamson, 1975, 1985) tend to assume that at one time there was a state of nature and that there was a movement from individuals to institutions – an approach often called methodological individualism (Popper, 1961; Hodgson, 1998, 1999). And of course there are innumerable instances which methodological individualists cite to demonstrate that individuals create new rules of behavior. For example, it is possible for actors to change the rules of driving, so that instead of driving on the left side of the road, drivers adopt a new rule and drive on the right.

This article, however, tends to equate social habits and institutions. As Hodgson and others remind us (Hodgson, 1988, 1989, 1997, 1999; Grafstein, 1992; Camic, 1986; Johnson, 1992; Nelson and Winter, 1982; Veblen, 1899), social habits are the results of earlier choices and are a means of avoiding endless deliberation. Because cognitive frameworks are learned through habit, individuals rely on the acquisition of such cognitive habits before reason, communication, choice or action are possible.

Whereas Schotter (1981) and other game theorists take the individual as an agent unencumbered by previous habits, Field (1984) and others have stressed that there can be no games without prior norms and rules, and thus a set of norms and rules must be presumed at the start. Those who attempt to explain institutions from individual behavior alone are using a bad strategy (Hodgson, 1998).

The position here – heavily influenced by Hodgson (1998, 1997, 1988) – is that individuals are embedded in a complex institutional environment and that institutions not only constrain but also shape individuals (also see Hollingsworth and Boyer, 1997; Hollingsworth, Müller, and Hollingsworth, 2002). It would be a mistake, however, to get involved in an infinite regress in order to determine which came first – individuals or institutions. Of course, institutions are formed and changed by individuals, just as individuals are shaped and constrained by institutions. But at a macro level, it is institutions that provide a cognitive framework whereby individuals can cope with their reality. In this sense, the micro and macro worlds are intertwined. At the macro level, there is considerable stability, but at the micro level, individuals have a significant level of autonomy, and there can be widespread diversity. As Hodgson (1988) reminds us, most institutions in a temporal sense exist prior to the individuals in any given society.

It would be a serious mistake to downplay the importance of individuals and micro level analysis as we study institutions. In the final analysis, it is at the level of individuals that norms, rules, habits, conventions and values exist. An individual is born into and socialized into groups and a society, and this is how one early in life acquires a sense of appropriate forms of behavior. Because of the way that individuals are socialized into a world of rules, norms, habits, conventions and values, it is unnecessary for individuals to restructure the world anew every day (Douglas, 1987; Elster, 1989; Archer, 1996). Every action does not have to be seriously reflected upon. For this reason, institutions provide cognitive frameworks for individuals, make their environments predictable, provide the information for coping with complex problems and environments. In the words of Johnson (1992: 26), ‘Institutions reduce uncertainty, coordinate the use of knowledge, mediate conflicts, and provide incentive systems. By serving these functions institutions provide the stability necessary for the reproduction of society’. However, each society has different forms of habits, rules and norms and hence different incentives and disincentive systems for learning and forgetting, for processing information. But because individuals have varying degrees of autonomy, individuals and groups can deviate from the prescribed forms of behavior in a society. And of course, these changes at the level of individuals become important in understanding processes of societal change.

These views are not meant to imply that the type of institutional analysis proposed herein approximates a general theory of society. This is clearly not the case. However, it is intended to represent the first steps in a mapping exercise of the boundaries of institutional analysis and to suggest a few methodological insights for studying institutions.

One should try to see norms and rules as continuous and not as dichotomous entities, to recognize that they come in varying strengths. Legro (1997) has suggested that we assess the robustness of norms and rules with three criteria: their simplicity, their durability and their concordance. Simplicity refers to how well actors understand norms and rules, how well they can be applied within a specific situation. Some norms and rules can be so complex that actors can have considerable difficulty in applying them in specific cases. Durability addresses the issue of how long norms and rules have been in effect – in short, in order to assess their level of legitimacy. While the position of this article is that norms, rules and values are quite durable, they do vary in this respect. Concordance refers to how widely applied a norm or rule is. This addresses the degree to which a rule is a meta rule, the degree to which it incorporates the heterogeneity of other norms and rules. In sum, the clearer the norms and rules of a society, the longer they have been in existence, and the more widely applicable they are, the greater their impact on a society. Hence, the more robust the norms and rules, the greater their impact on a society, and the less their robustness, the greater their flexibility and the less their effect on shaping a society’s outcomes and performances.

Because norms, rules and values are quite durable, they play an important role in shaping the history of societies, thus contributing to a great deal of path-dependency. Actors attempt to adjust to their contemporary environment, but since they are products of the past, the historical legacy of norms, rules and values influences the decisions they make. Although actors have some capacity to alter the course of their history, they are constrained by their past, and the degree to which they can move beyond their past is limited. As Lanzara (1998), Johnson (1992) and others have argued, societal inertia is a basic feature of institutions. They provide the basic stability necessary for change. Degrees of history are continuously reproduced by the way in which the inhabitants of societies are socialized. History matters, but at critical points in history, there is punctuated equilibrium (Somit and Peterson, 1992). During most periods of history, there is considerable stability in the norms, rules and values of a society, but at critical moments, norms, rules and values can quickly and dramatically be redefined. At all times, norms, rules, habits, conventions and values are influencing each of the other components in the institutional framework discussed below, but these other institutional components feed back and can modify rules, norms, conventions, etc. (Murmann, 1998).

Second level of analysis

The norms, rules, habits, conventions and values of a society lead to the next level of analysis – the institutional arrangements which are involved in the coordination of various economic actors: producers and suppliers of raw materials, knowledge, etc.; processors of raw materials, information; workers; customers of raw materials, finished products, information, etc.; financiers; governmental and other types of regulators. These actors regularly engage in contests to resolve various economic problems in virtually all sectors of society: how are prices to be set? What quantity of various products is to be produced? How are standards of various products, processes, etc. to be set? How is the quality of products and processes to be determined? How are various societal processes to be financed? In order to confront these problems and to address the conflicting positions of economic actors as they address these problems, societies develop various institutional (governance) arrangements for coordinating different actors. These consist of markets, various types of hierarchies and networks, associations, the state, communities and clans (see Hollingsworth and Boyer, 1997; Campbell et al., 1991: ch. 1).

Each of these particular kinds of coordinating mechanism has many different types. For example, there are many types of states (e.g. the regulatory state, developmental state, authoritarian state, welfare state), on which there is an extensive literature (Kim, 1997). Similarly, there are different types of markets, networks, different kinds of associations, etc. (Boyer, 1997; Hage and Alter, 1997; Schneiberg and Hollingsworth, 1990). When we do institutional analysis, we must engage in configurative analysis, recognizing that actors are not coordinated or governed by a single type of institutional arrangement. Some of the literature discusses industrial sectors as though they were coordinated or governed by a single institutional arrangement, whereas in fact each sector of an economy is coordinated by a configuration of institutional arrangements. Some configurations coordinate actors in certain problem areas, while other configurations of institutional arrangements coordinate actors in addressing other problems. The types of configurations which are dominant in a society are somewhat stable and tend to persist over time within a society (Hodgson, 1999).

When one mode of coordination is dominant in a society, it will influence the role which other coordinating modes will play. Hence, the strong role of the state in the Soviet Union influenced the role of markets, associations, etc. in the governance of the Soviet economy. Similarly, the prominence of particular modes of coordination in a society influences its style of innovation.

Much of the literature on institutional arrangements (e.g. forms of economic coordination) remains fragmented and unintegrated. It is helpful to array these various forms of coordination in a two-dimensional taxonomy, as in Figure 2. On the vertical dimension, the economist’s view of a self-interested agent is contrasted with a more sociological perspective, according to which social rules, obligation and compliance shape human actions. On the horizontal dimension, there is another continuum (i.e. the distribution of power). At one extreme of the dimension, one finds many and relatively equal agents interacting (e.g. as in a well-organized spot market). At the other extreme, inequality in power results in a hierarchical form of coordination which structures the interaction between principals and agents or between leaders and followers. Where a society falls along the distribution of power is influenced by the rules, norms, values, etc. which are dominant in a society at a particular moment in time.

 Institutional arrangements can be visualized on two dimensions: the nature of action motive on the one hand, and the distribution of power on the other. Markets (cell 1) combine self-interest with horizontal coordination transactors, and they reflect sensitivity to concerns about supply and demand, thus providing ex post an unintended equilibrium. Paradoxically, the more pure and perfect the market competition, the greater the need for codified rules of the games for coordinating economic transactions. Thus, collective associations (cell 6) and/or various forms of state intervention (cell 4) are required to develop and enforce rules for transacting partners (Schneiberg and Hollingsworth, 1990; Streeck and Schmitter, 1985a). This is an example of how the norms and rules and the institutional arrangements of a society are intertwined as reflected in a configuration of institutional or governance arrangements.

Along the horizontal axis, actors can be joined in an organization or a firm: a hierarchy is the generic terms for this institutional arrangement (cell 2). Along the horizontal line, one recognizes the difference between transactions in a market and transactions within a firm. The well-known works of Coase (1960, 1981) and Williamson (1975, 1985) utilize the concept of transaction costs in explaining the emergence of corporate hierarchies.

There are also various types of networks (cell 5) which exhibit mixes of self-interest and social obligation, with some actors being formally independent and equal. Yet in some networks (the large firms and their subcontractors), there is unequal power and initiative. Networks may comprise all kinds of actors; some consist only of firms but others include associations and the state (Hage and Alter, 1997).

The vertical axis deals with action motives. Toward the upper part of Figure 2, actors are engaged in individualistically oriented behavior, whereas toward the lower part, actors are more engaged in collective behavior and strive to cope with problems of common interest. Cell 3 – communities and clans – consists of institutional arrangements based on trust, reciprocity or obligation, and thus are not derived from the pure selfish computation of pleasures and pains. This is an unconventional form of coordination for most neoclassical economists (however, see Arrow, 1974), but not for many anthropologists, political scientists and sociologists (Streeck and Schmitter, 1985a; Polanyi, 1944; Gambetta, 1988; Fukuyama, 1995; Sabel, 1992; Putnam, 1993, 2000).

In the neoclassical paradigm, theorists argue that actors engage in forms of exchanges that best promote their individual interests. If some structural conditions are fulfilled (absence of increasing returns to scale, the reversibility of transactions, absence of uncertainty, and complete contingent markets, with no collusion between actors), then the invisible hand theorem applies, and market-type activity functions quite well and also provides the optimum for society, thereby combining efficiency, harmony and order. However, an excess of market activity may well lead to ruinous competition and excessive conflict (Hirsch, 1976; Hirschman, 1986; Polanyi, 1944). There is variation in the extent to which ruinous competition occurs, depending on the social context in which transactions take place. Thus, it is important that we be sensitive to the institutional context in which transactions are embedded and that we understand the degree to which social bonds exist at both the micro and macro levels of analysis. Micro bonds facilitate exchanges in a society, but at the societal level social bonds exist at the level of the collective – in the community or region, and among members of racial, religious and ethnic groups. All other things being equal, the more powerful the social bonds among transacting partners, the more economic competition is likely to be restrained. Thus, most transactions occur not simply in an impersonal, calculative system of autonomous actors unrestrained by social ties – as implied by the neoclassical paradigm – but in the context of social ties, variation in the strength of which leads to variation in levels of trust and transaction costs (Etzioni, 1988: 211; Granovetter, 1985; Streeck and Schmitter, 1985a; Hollingsworth and Boyer, 1997; Hodgson, 1988, 1999; Schneiberg and Hollingsworth, 1990).

Another form of multilateral institutional arrangement is various types of associations (cell 6). Unlike networks, clans and communities, associations are more formal organizations. Whereas markets, corporate hierarchies and networks tend to coordinate economic activity among different types of actors (e.g. producers with suppliers, capital with labor), associations typically coordinate actors engaged in the same or similar kinds of activities. Business associations and labor unions are some of the most common forms of associations for coordinating economic activity in capitalist economies (Schneiberg and Hollingsworth, 1990; Streeck and Schmitter, 1985a).

Finally, there is the state (cell 4), which is an institutional arrangement quite unlike any of the others. It is the state that sanctions and regulates the various non-state coordinating mechanisms, that is the ultimate enforcer of rules of the various mechanisms, that defines and enforces property rights, and that manipulates fiscal and monetary policy. Overall, it is the state that influences the total incentive system of a society (Lindberg and Campbell, 1991; Johnson, 1992: 40). At the same time, the state may also be an economic actor by engaging directly in production and exchange relations.

Institutional arrangements in Figure 2 are constrained by the social context in which they are embedded. Depending on the nature of that embeddedness, there is variation in the collective forms of governance, some of which are specified in the lower part of the figure. Some modes of coordination in the bottom and upper levels of the typology are often mixed together, though one coordinating mode is likely to be more prominent than another. But actors are often engaged in complex configurations involving several kinds of institutional arrangements.

Each of these various institutional arrangements has its own logic – its own rules, its own procedures for enforcing compliance, its own norms and ideologies which help to reduce the costs of enforcement. These are summarized in Table 2, which provides further elaboration about the various coordinating mechanisms one finds in almost every capitalist society. Table 2 lists the organizational arrangements and their structures, rules of exchange, and means of enforcing compliance associated with each type of coordinating mechanism. While each type of institutional arrangement has various positive features, each institutional arrangement also has failures, and these are featured in Table 3. It is the contest between those who support and those who oppose these various institutional arrangements that tends to lead to transformations in institutional arrangements over time (Campbell et al., 1991; Campbell, 1997).

    Again, it is difficult to conceive of pure institutional arrangements – either exclusively markets or exclusively hierarchies – since it has been well known since Adam Smith’s Wealth of Nations that the division of labor within the firm cannot be disentangled from the existence and extent of the market. Institutional arrangements have their own distinctive form of efficiency (good static efficiency for the market, dynamic efficiency for firms) and inefficiency, often leading to considerable inequality. Neither networks nor communities are panaceas for economic coordination, without being configured with other types of institutional arrangements. Networks and communities may solve certain issues, but they raise other, no less severe problems. It is important to recognize the imperfection of any single institutional arrangement in order to comprehend the origin and transformation of any other institutional arrangement (Hodgson, 1999: ch. 3).

   As suggested above, each institutional arrangement is configured with other institutional arrangements. Usually, one particular institutional arrangement is more dominant in a particular configuration, but because each type of institutional arrangement has its own strengths and weaknesses, there is no simple structural logic in the governance or coordination of a society. Each institutional arrangement constrains the other, but the inherent tension among the various institutional arrangements within a configuration contributes to changes in the governance or coordination of a society. The routines and logics of each of these configurations of institutional arrangements provide constraints and incentives for actors. The way that actors perceive the incentives and constraints of these governance configurations leads to a particular market logic, and it is the specific market logic of a society which influences its specific capacities and weaknesses. The inherent strain among the different logics in a configuration of coordination helps to provide the flexibility for a society to adapt to new circumstances (Hollingsworth, 1991).

   If there were a society with pure markets or a society coordinated only by the state, there would be too much rigidity and too little diversity to cope with the vast uncertainty in the global environment. A society with very little diversity in its coordination mechanisms would have little capacity to adapt to new circumstances. Configurations with considerable diversity of institutional arrangements provide for a certain amount of incoherence in governance, but they also provide for the capacity to adapt to new circumstances. While the Soviet Union was heavily dominated by the state as a coordinating mechanism, there were always functioning very weak markets in the system. Moreover, the feudal society of the twelfth and thirteenth centuries consisted not only of hierarchical relationships between serfs and masters but also of urban guilds, clerical hierarchies and markets (Johnson, 1992: 38). In sum, the robustness of institutions often depends on multiple and diverse principles and logics of action, on the inconsistency of principles and procedures, on patterned forms of disorder (Lanzara, 1998; Orren and Skowronek, 1991: 320, 329).

* Note: For allocative efficiency and X-efficiency, see Leibenstein (1966, 1976)

Third level of analysis

The next level consists of the institutional sectors of a society. The rules, norms and values of a society influence the array of institutional arrangements, and levels one and two influence the nature of and the relationships among various institutional sectors, and all three levels are intricately linked together to form a social system of production. Together, all of these components influence the performance of economic sectors within a society as well as the performance of the total society. A social system of production is the way that a society’s institutions (see first level of analysis above), its institutional arrangements (see second level of analysis above), and its institutional sectors are integrated into a social configuration. Recently, a group of scholars have developed the concept “institutional complementarity” which is somewhat similar to the ideas developed here. Institutional complementarity is the idea that particular institutional sectors and coordinating mechanisms function all the better because of the way they interact with other sectors and coordinating mechanisms (Amable, 2000; Hall and Soskice, 2001). The notion of institutional complementarity links together all the give levels of institutional analysis presented in Table 1 into an overall configuration.

Ideas about institutional complementarity have special relevance for the study of comparative capitalism. Thus, Aoki (2001) has argued that long-term employment is more likely to occur when the financial system provides capital on terms that are not sensitive to current rates of profitability. Hall and Soskice (2001: 18) argue that “where dense networks of business associations support collaborative systems of vocational training, those same networks may be used to operate the collective setting of standards.” Because of institutional complementarity, institutional practices of various types are not distributed randomly across societies. And while each type of capitalism has its own partisans, the argument here is not that one form of capitalism is superior to another. Each has its own logic and its own performance structures.

An institutional sector includes all organizations in a society which supply a given service or product, along with their associated focal organizations (e.g. major suppliers, funding sources, regulators, and so forth (Scott et al., 1994: 108, 117; Campbell et al., 1991; Hollingsworth et al., 1994). Institutional sectors include but are not limited to the society’s system of education, system of research, business system, financial markets, legal system and the state. The structure of the financial markets, system of training and education, industrial relations system, system of research, and state and legal systems are distinct and idiosyncratic in each society. In short, they are system specific.

All of these institutional sectors tend to cohere with each other, although they vary in the degree to which they are tightly coupled into a full-fledged system. While each of the institutional sectors has some autonomy and may have some goals that are contradictory to the goals of other institutional sectors with which it is linked, an institutional logic in each society leads institutions to coalesce into a complex social configuration (Hollingsworth, 1991). This occurs because the institutional sectors are embedded in a culture in which their logics are symbolically grounded, organizationally structured, technically and materially constrained, and politically defended. The configuration of institutional sectors usually exhibits some degree of adaptability to new challenges, but continues to evolve within an existing style. But under new circumstances or unprecedented disturbances, these institutional configurations are exposed to sharp historical limits as to what they may or may not do (Schumpeter, 1983; David, 1988; Arthur, 1988a, 1988b; Håkanssen and Lundgren, 1997).

Why do all of these different institutions coalesce into a complex social configuration, which is labeled here as a social system of production? The literature suggests two contrasting interpretations. Part of the answer – indeed a controversial one – is that these institutional sectors are functionally determined by the requirements of the practice of capitalism in each time and place (Habermas, 1975). Another explanation emphasizes the genesis of the actual configuration, via a trial and error process, according to which the survival of firms, regions or countries is the outcome of complex evolutionary mechanisms (Maynard-Smith, 1982; Nelson and Winter, 1982). However, the problem is even more complex. Markets and other mechanisms for coordinating relationships among economic actors place constraints on the means and ends of economic activity to be achieved in any society. The other coordinating mechanisms include different kinds of hierarchies, various types of networks and associations – trade unions, employers and business associations (Hollingsworth and Lindberg, 1985; Campbell et al., 1991). The logic of these various institutional arrangements provides actors with vocabularies and logics for pursuing their goals, for defining what is valued, and for shaping the norms and rules by which they abide (see Tables 2 and 3). In short, in contrast to the logic of the neoclassical paradigm, the argument here is that the dominant type of institutional arrangements places severe constraints on the definition of needs, preferences and choices of economic actors. Whereas the neoclassical paradigm assumes that individuals and firms are sovereign, this paper is based on the assumption that organizations are influenced by the hold that the institutional configurations making up a social system of production have on individual decision making (Campbell et al., 1991; Etzioni, 1988; Streeck and Schmitter, 1985a; Hollingsworth et al., 1994; Hollingsworth and Boyer, 1997; Magnusson and Ottosson, 1997; North, 1990; Hodgson, 1999).

Standard neoclassical economic theory has tended to downplay the role of production and consequently of firms. Even transaction cost theorists who are concerned with analyzing the firm as a coordinating mechanism have been relatively unconcerned with the various components of a social system of production. Indeed, as long as there was widespread optimism about the efficacy of Keynesian economics, there was relatively little concern among neoclassical economists with the supply side of the economy. Even in the opinion of most Keynesians, a group of experts should ideally be able to shape the size of aggregate demand while the supply side of the economy would be left to the two minimalist institutions of neoclassical economics – markets and managerial hierarchies. For several decades, however, it has become increasingly obvious that some of the most competitive and successful patterns of industrial output and industrial production in capitalist economies do not derive from the neoclassical prescription of unregulated markets and corporate hierarchies complemented by a neoliberal democratic state. Indeed, it is now well understood that certain highly