Historical Institutional Analysis
Avner Greif
8 April 2001
Preface
To be provided
-IIContents
Part I: Preliminaries
1. Motivation and Issues
2. On Analyses of Economic Institutions in the Social Sciences
Part II: Self-enforcing Institutions: Institutional Statics as an Equilibrium
3. Institutions as Rules and Beliefs
4. Organizations and Self-enforcing Institutions
5. The Institutions Within Us: Endogenous Preferences, Norms, and Internalized Beliefs
Part III: Self-Reinforcing Institutions: Institutional Dynamics as a Historical Process
1. Why Institutional Analysis is Inherently Historical?
6. The Influence of Past Institutions on the Rate of Change: Institutional Inertia and
Endogenous Institutional Change
2. Why Institutions are the ‘Carriers of History’? The Influence of Past Institutions on
the Direction of Institutional Change
Part IV: Institutional Complexes and Empirics
7. Why Institutions are the Producers of History?
8. On Empirics: Combining Induction and Deduction in Studying the Observable
Implication of Beliefs
Part V: Conclusion
9. Concluding Remarks
4/8/01 - 1
Chapter 1: Motivation and Issues
In economics, political science, and sociology it has become widely accepted that institutions
matter.1 There is less agreement, however, about what economic institutions are and how to study
them. Economic institutions are defined, for example, as rules, beliefs, norms of behavior,
regularities of behavior, or organizations (such as firms, courts, parliaments, social networks, and
communities). Furthermore, the main approaches to institutional analysis also differ in their basic
methodological postulates regarding the origins, dynamics, and implications of institutions. While
some advance a deductive, agency perspective others advance an inductive, historical, structural
perspective. Because distinct definitions of institutions have been treated in the literature as
mutually exclusive and because of the above methodological divide, we still have a limited ability to
address three inter-related questions that are at the heart of social science and history: What are
institutions? Why do societies evolve along distinct institutional trajectories? Why do societies
often fail to adopt the institutional structures of other more successful societies?
The Challenge of Institutional Analysis
Among the obstacles for advancing a comparative analysis of institutions over time is the
elusive nature of the concept of institutions. There are many, seemingly alternative, definitions of
the term institutions and varied approaches to studying them within the disciplines of economics,
political science and sociology. The following are some of definitions commonly used in these
fields. Institutions are:
1. “The rules of the game” in a society (North 1990: 4) and “the sets of working rules” that
“contain prescriptions that forbid, permit, or require some action or outcome,” and that are
“actually used, monitored, and enforced” (Ostrom 1990: 6).
2. Organizations such as firms, parliaments, tribes, families, communities, and universities.
(E.g., Williamson 1985; Granovetter 1985; North and Weingast 1989; Nelson ??.)
3. Beliefs. Cultural or shared beliefs about others’ behavior or about the world around us and the
relationships between actions and outcomes in it. (E.g., Weber 1958 [1904]; Denzau and
North 1994; Greif 1994a; Calvert 1995; Lal 1998; Aoki forthcoming.)
1 Conviction in the importance of institutional analysis is well reflected in its recently gained
momentum the social sciences. For surveys of these developments, see, for example, Eggertsson
(1990); Furubotn and Richter (1997); Hodgson (1998); Milgrom and Roberts (1995); Williamson
(1996); Hart (1995); Bardhan (1991); and Barzel (1989); Greif (1996b, 1997a, 1997c, 1997d,
1997e, 1998) in economics. Weingast (1996); Bates et al. (1998), Thelen (1999) in political
science and, Coleman (1990); Gravnovatter (1985); Powell and DiMaggio (1991, introduction);
Smesler and Swedberg (1994); Scott (1995); and Brinton and Nee (1998) in sociology. See also the
papers in the May 1998 volume of The American Economic Review regarding the new institutional
economics.
4/8/01 - 2
4. Norms of behavior that were internalized by members of the society and hence influenced
their behavior. (E.g., Ullmann-Margalit 1977; Elster 1989; Platteau 1994).
5. Regularities of behavior or “social practices that are regularly and continuously repeated”
(Berger 1977; Schotter 1981; The Penguin Dictionary of Sociology: 216; Young 1998:
preface.)
The main approaches to institutional analysis in the social sciences differ in more ways than
how they define their object of study. They also differ in their basic methodological postulates
regarding the origins, dynamics, and implications of institutions. Some advance a deductive, agency
perspective while others advance an inductive, historical, structural perspective.
Approaches that adopted the deductive, agency perspective place the individual decisionmaker
at the center of their analysis and study institutions deductively as equilibrium outcomes
reflecting the inter-relationships among individuals’ objectives, possibilities, and the environment
within which they interact. Institutions are therefore considered to reflect human actions and social
process and are postulated not to endure beyond the conditions that led to their emergence.
Politicians, for example, aspire to create the rules that serve their political and economic objectives
best. If these objectives or the political process of rule formation changes, so will the resulting rules.
Similarly, conventional rules of behavior emerge spontaneously through the interactions of
individuals in a given environment and will change following an environmental change. The point of
departure for such institutional analysis is therefore at the micro-level of the individuals whose
interactions in a particular environment give rise to an institution.
Approaches that adopted the inductive, historical, structural perspective emphasize that
institutions are structures that do not reflect agents’ needs and possibilities but shape these needs and
determine these possibilities. Institutions structure human interactions and mold individuals and their
social and cultural worlds. They therefore transcend the situation that led to their emergence and
constitute part of a society’s historical heritage. Beliefs, norms, and social structures that were
crystalized in the past, for example, are part of the structure within which individuals interact. One
has to study institutions inductively, utilizing dynamic, historically-contingent models in which past
institutions can have a lasting influence. The point of departure for such institutional analysis is
therefore at the macro-level of the historically-determined structure within which individuals
interact.
These two perspectives - the deductive, agency perspective and the inductive, structural,
historical perspective - separate the main approaches to institutional analysis in the social sciences.
Within sociology, works in the tradition of Weber (e.g., 1949) maintain that institutions reflect the
interactions among individuals, while works in the tradition of Durkheim (e.g., 1950) consider
institutions to be societal features that “impose themselves upon” individuals (p. 2). Within
economics, transaction cost economics assumes that institutions are instrumental transaction costs
optimizing responses to environmental conditions (e.g., Williamson 1985), but in economic history
and evolutionary economics it is common to identify institutions with history-dependent, and not
necessarily functional, behavior (e.g., North 1990; Hodgson 1998). Rational choice analysis in
political science examines institutions as instrumental outcomes using equilibrium analysis, while
historical institutionalism emphasizes that they reflect a historical process (Thelen 1999).
These definitions of institutions and approaches to their analysis have been treated in the
literature as mutually exclusive, thereby curtailing the ability to advance institutional analysis and
benefit from integrating the insights of various approaches. This book is based on the postulate that
the key to further advancing institutional analysis is understanding the common aspects of various
4/8/01 - 3
definitions of institutions, developing a unifying concept of the object of study, and exploring the
complementary relationships between the above two perspectives.2
This postulate is inspired by various studies of historical institutions and their dynamics that
indicate the deficiency of each of the above definitions of, and perspectives on, institutions. Each is
appropriate for addressing particular issues, but fails to provide a comprehensive understanding of the
social underpinnings of regularities of behavior and their dynamics. At the same time, works on
historical institutions have revealed that the various definitions share much more than catches the
eye and hence advancing an encompassing framework is feasible. Furthermore, the various
approaches complement each other. Hence, furthering institutional analysis beyond the confines of
each approach and exploring, for example, institutional statics and dynamics within the same
framework, requires revealing the commonality, distinctions, and inter-relationships among all the
approaches to institutional analysis.
The need for, and the potential benefit of, integrating various lines of institutional analysis
have been noted by many students of institutions, such as Douglass North (1990) and James Coleman
(1990). Nevertheless, despite the recent surge in inter-disciplinary institutional analysis, little
interaction and cross-fertilization exists among scholars embracing distinct definitions of, and
perspectives on, economic institutions. This situation is well reflected in the fact that recent
important works on economic institutions either refrain from defining institutions or advance a
particular definition at the expense of alternative ones. (E.g., North 1990; Eggertsson 1990;
Furubotn and Richter 1997; Weingast 1996; Young 1998; Aoki forthcoming.) At the same time,
during the 1990s, incremental analytical contributions and empirical studies indicated the ability to,
and provided the framework for, integrating various lines of institutional analysis. This book brings
together and builds upon these developments.
Common Aspects in Different Approaches to Institutional Analysis
2 For tentative expositions of the various elements of the proposed approach, see Greif (1994,
1997a, 1997b, 1998a, 1998b, 1998c, 2000a, 2000b).
Despite transparent differences, approaches adopting various definitions of institutions have
a lot in common. First, all of them are concerned with the behavioral implications of factors (such
as rules, beliefs, norms, or organizations) that are social in the sense that they are neither determined
by technology nor by our genetics. Furthermore, institutional analysis is not concerned with social
factors that are temporary in nature or relate to particular individuals. Institutional analysis is
concerned with social factors that have a widespread and lasting influence and that govern the
relationships among individuals who occupy particular social positions (such as parents and children,
lenders and borrowers, or employers and employees). In other words, institutional analysis is
concerned with social factors that generate regularities of behavior among individuals who occupy
particular social positions.
Second, irrespective of the particular social factor a certain approach to institutional analysis
emphasizes, common to all of them is the postulate that this social factor is exogenous to each of
the individuals whose behavior it influences although it is endogenous to the society. These social
factors are exogenous to each of the individuals whose behavior they influence in the sense that they
are beyond their control, but they are nevertheless endogenous to the society and reflect the
interactions among the individuals whose behavior they influence or other individuals. Legal rules are
a good example. They reflect the legal and political processes of rule making in which the individuals
whose behavior they govern, or other individuals, take part. Similarly, the norms that a society’s
members were socialized into are exogenous to each of them, but reflect endogenous social processes.
4/8/01 - 4
The common focus on social factors exogenous to each individual but endogenous to the
society reflects a third aspect common to all the approaches to institutional analysis (with the
exception of those that consider institutions as regularities of behavior). All approaches consider
institutions as social factors that generate regularities of behavior by enabling, coordinating, and
motivating individuals to take particular actions. (I use the terms motivating and constraining
interchangeably to denote the social factors that induce individuals to behave based on either rewards
or punishments.) Hence, institutions have to be social factors exogenous to each individual whose
behavior they influence, because whatever is under the direct control of an individual (whatever one’s
choice variable is) is not a social factor that enables, coordinates, and motivates his behavior. One’s
choice is made within the contours spanned by the relevant institutions and it is part of the
regularities of behavior that these institutions imply. For example, the social factors that influence
the selection and enable the use of various contractual forms by a firm constitute the relevant
institution, while the contract a firm actually offers its employees is the behavior this institution
implies.
Defining Institutions: One Definition and Multiple Contents
Hence, although seemingly distinct, various approaches to institutional analysis share a basic
concern: they are all directly or indirectly concerned with social factors that generate a regularity of
behavior by enabling, coordinating, and motivating (or constraining) behavior while being
exogenous to each of the individuals whose behavior they influence but endogenous to the society.
Different approaches to institutional analysis concentrate on distinct social factors. Each gives a
different meaning to the same definition. Indeed, it seems appropriate to give a distinct meaning to
this definition. Distinct social factors can prevail in various historical episodes with respect to a
particular transaction and generate the same or different regularities of behavior. For example, legal
rules, norms, and belief in God’s punishment following a particular action can direct behavior in
economic exchange.
The possible multiplicity of social factors influencing behavior implies there is a need to
differentiate between social factors that can and do influence behavior. Accordingly, this work
refers to the social factors that actually contribute to generating regularities of behavior in a
particular historical episode as (relevant) institutional elements. An institutional element is a
social factor that contributes to generating regularities of behavior among individuals with a
particular social position and in a particular transaction while being exogenous to each of the
individuals whose behavior it influences but endogenous to the society. (For simplicity of exposition
I will, henceforth, omit these qualifications and sometimes refer to social factors as potential or
feasible institutional elements.)
But while each of the various approaches to institutional analysis focuses on a particular
institutional element, the approach advanced here emphasizes that understanding the social
underpinnings of behavior - namely, achieving the objective of all approaches to institutional
analysis - requires going beyond the analysis of a particular institutional element. As various recent
theoretical considerations and analyses of historical and contemporary institutions indicate,
understanding the social underpinnings of behavior requires examining the inter-relationships among
various institutional elements.
For example, formal or informal rules of behavior do not generate regularities of behavior
unless they are enforced. Hence, analysis of the rules and the process through which they are
generated has to be complemented by an analysis of how they are enforced. Considering
enforcement, however, often requires examining such social factors as beliefs, norms, and
organizational features (such as a community or a court). Similarly, identifying organizations as
institutions often provides an incomplete understanding of the social underpinnings of behavioral
regularities because, as game theory forcefully illustrates, organizations provide a framework within
which distinct behavior can be generated and prevail as an equilibrium outcome. For example,
members of a community can punish a member who cheated other members. But other patterns of
4/8/01 - 5
intra-community behavior can theoretically prevail as an equilibrium outcome and have been found
to prevail.
Accordingly, this work examines various institutional elements as inter-related parts of a
larger whole - an institution - with which a particular regularity of behavior is associated. The
discussion is first aimed at advancing the study of institutional statics. It will highlight how
institutions generate and reflect endogenous behavioral regularities in a society, thereby creating what
sociologists often refer to as a (perceived or actual) “social order” that members of the “society take
for granted” (e.g., Jepperson 1991: 145). But the discussion is also aimed at advancing the study of
institutional dynamics, emphasizing why it is a historical process in which the rate and direction of
institutional change reflect past institutions.
In advancing this position this work differs from approaches that identified institutions with
a particular institutional element, such as rules, beliefs, or organizations. Such approaches ignored, by
and large, the challenge and potential importance of examining the inter-relationships among various
institutional elements. Some consider all institutional elements to be a part of a larger “institutional
matrix” (North 1990) whose analysis is beyond the scope of institutional analysis per se. Others
assert that the distinction between various institutional elements is superficial. Rules and
organizations, for example, are often considered particular types of beliefs (e.g., Calvert 1995).
Each of these positions is appropriate for various analytical purposes and in specific empirical
studies, but this work argues that such positions either run the risk of defining institutions in a way
that is too abstract or vague to guide a positive, empirical analysis, or lose insights that can be
derived by recognizing the distinctions and inter-relationships among various institutional elements.
An Integrated Approach to the Study of Institutions
An objective of this book is, therefore, to present a conceptual and analytical framework
enabling us to capture important aspects of the inter-relationships among various institutional
elements. This framework is aimed at highlighting and fostering the analysis of how various
institutional elements constitute inter-related parts of a larger whole - an institution - with which a
particular regularity of behavior is associated. This work presents this framework and elaborates on
its usefulness in studying institutions and their dynamics.
This book particularly concentrates on the interplay between the institutional elements that
various approaches to institutional analysis define as institutions (namely, rules, beliefs, norms, and
organizations). It does so while examining endogenous institutions, namely, those in which all
regularities of behavior are generated and maintained by institutional elements that are subject to the
analysis. The ability to conduct a general examination reflects the recent development of various
analytical frameworks, particularly game theory, information economics, and evolutionary
economics. These frameworks enable us to analyze interactions among individuals with little prior
knowledge and information in situations in which all behavior has to be endogenously motivated.
Clearly, for some positive institutional analyses there is no need to examine fully endogenous
institutions. It may be appropriate in a particular study to assume, for example, that the
enforcement provided by a non-strategic third party such as the state is exogenous, or that the
formal rules governing voting in a parliament are binding, or that one’s religious beliefs or communal
affiliation can be taken as given. Taking these or other particular institutional elements as
exogenous in the analysis constitute considering a special case of the more general one of studying
endogenous institutions.
This book begins by providing a brief introduction to the study of institutions in economics
and their relationship to institutional analysis in political science and sociology. This survey
presents the broad development of institutional analysis in economics up to circa 1990. Later
developments are incorporated in the body of the book that also explores the inter-relationships
among various institutional elements and how they can constitute inter-related parts of a larger
4/8/01 - 6
whole - an institution - with which a particular regularity of behavior is associated.3 In exploring the
role of various institutional elements in such an institution, the analysis considers transactions or
interactions as the basic units of analysis. (The importance of the transaction as the basic unit of
analysis has been stressed by Commons, Coase, Williamson, and North, and is in contrast to the use
of goods as the basic unit of analysis in neo-classical economics.)
3 These ideas build on the works of many scholars to which I refer below.
When enforcement is endogenous (namely, when we cannot assume that people will follow
rules) the role of rules is to guide and coordinate behavior in the transaction under consideration.
Such rules can and do take many forms: they can be formal or informal, implicit or explicit, tacit or
well articulated, reflect normative values or the dictates of the powerful. Rules can emerge
spontaneously or deliberately and they can be formulated quickly or reflect a long period of
experimentation and learning. But in all cases they are behavioral instructions that are common
knowledge among members of the society. Not every rule, however, is an institutional element -
namely, a social factor that actually influences behavior. Unless people follow a rule, it remains
hollow with no influence on behavior.
The institutional elements that motivate individuals to either follow or ignore rules are
beliefs and internalized constraints such as norms. For a system of beliefs to be an institutional
element it has to be shared by members of the society. Such shared, cultural beliefs can be of two
kinds - behavioral beliefs and internalized beliefs. The first are beliefs that individuals hold regarding
the behavior that others will assume in various contingencies that may or may not actually transpire.
The second are internalized beliefs regarding the structure of our (and potentially other) world(s) and
the implied relationship between actions and outcomes. Such beliefs reflect, for example, the human
tendency to try to understand the world around them. Similarly, internalized constraints such as
norms are the socially constructed behavioral standards that have been incorporated into one’s
superego and hence influence one’s behavior by becoming a part of his or her preferences. Together,
behavioral beliefs, internalized beliefs, and internalized constraints motivate one to act in a particular
way. At the aggregate level, therefore, they determine which rules of behavior will indeed be
followed.
Organizations - either formal ones such as Parliaments and firms or informal ones such as
communities and business networks - have three inter-related roles. First, organizations produce and
harbor rules of behavior; second, they contribute to the perpetuation of internalized constraints; and
third, they influence the set of beliefs that can prevail with respect to a particular transaction. This
latter role of organizations has received little attention: organizations influence the set of beliefs that
can prevail in a society mainly by linking transactions. For example, a court must exist in order for
people to believe that legal punishment will deter cheating. Hence, organizations alter (or reflect the
alteration) of the rules of the game relevant to the decision-makers in the original transaction under
consideration by, for example, introducing a new player (the organization itself), by changing the
information available to players, or by changing payoffs associated with certain actions.
Roughly speaking then, this book explores and develops a simple idea: When enforcement is
endogenous, the role of rules is to coordinate on particular beliefs and mold and reflect behavioral
norms. Organizations provide the arena within which rules are generated and the mechanism that, by
linking transactions, contributes to the emergence and perpetuation of particular beliefs and
internalized constraints (henceforth, norms) that motivate individuals to follow rules. Cultural,
shared beliefs are therefore central to the current conceptualization of institutions (as stressed also in
Greif 1994a, Calvert 1995, Weingast 1995a, Aoki 2001 forthcoming).
Bridging the Methodological Divide: Agency and Structure
4/8/01 - 7
It is misleading, however, to examine an institution by considering each institutional element
in isolation from the others. For a system of institutional elements to constitute an institution they
have to render each other effective in generating particular regularities of behavior. Rules should
coordinate on beliefs that actually influence behavior, behavior should lead to the emergence of these
rules and confirm the underlying beliefs, while organizations should facilitate the specification of
these rules, enable these beliefs, and, whenever appropriate, be the endogenous result of beliefs and
behavior. To capture the inter-relationships among various institutional elements that render them
effective in generating regularities of behavior, the approach advanced here requires that for a system
of rules, beliefs, norms, and organizations to prevail - to constitute an endogenous institution
generating a particular behavior - it has to be self-enforcing.
An institution is self-enforcing when the actual and expected behavior of each of the
interacting individuals generates the institutional elements that enable, coordinate, and motivate
other individuals to follow the associated regularity of behavior. The behavior of each individual
leads others to act, or to be expected to act, in a way that motivated each person to behave in this
way to begin with. In other words, the institutional elements that influence the behavior of each
individual lead to behavior that maintains these institutional elements. The structure that each
individual faces induces him to take the actions that, on the aggregate level, create this structure.
To say that for an institution to prevail it must be self-enforcing emphasizes that
comprehensive understanding of social outcomes requires considering all behavior and its
underpinning institutional elements as endogenous. Even kings and dictators are not exogenous to a
society and their ability to rule and influence others’ behavior reflects interactions between them and
other members of their societies. Their ability to rule is a self-enforcing outcome. Similarly, the
ability of such rulers to influence behavior depends on the credibility of their threats or promises to
punish or reward those who behave in a certain way. (E.g., Levi 1988).
By considering institutions as self-enforcing systems, the approach advanced here transcends
the dichotomy between the agency perspective and the structural perspective by recognizing the
distinctions and inter-dependencies among individuals, behavior, and institutional elements.
Concurring with the former perspective, this work emphasizes that for an endogenous institution to
prevail it has to be self-enforcing and for it to perpetuate over time, it has to be reinforced by human
actions and social processes. It also agrees with the latter perspective, however, and emphasizes that
institutional elements are exogenous to each individual and are enduring social features that can
transcend the situation that led to their emergence. Individuals create institutions and institutions
mold individuals and shape their interactions over time and situations. In a sense, this work advances
an analysis that begins in the spirit of the methodological approach advanced by Weber (e.g., 1949),
that places individuals at the center of the analysis. It then proceeds to study institutions in the
spirit of the methodological approach advanced by Durkheim (e.g., 1893, 1895), that views
institutions as molding individuals and their social and cultural worlds, thereby constituting an
independent force in directing actions.
Clearly, however, for various analytical purposes it is not necessary to consider all
institutional elements and associated behavior as endogenous and self-enforcing in order to examine
the issue at hand. In considering how a particular rule, say a tax law, is formed in the US Congress,
one may take the rules governing interactions among members of Congress or the ability to enforce
the law as exogenous to the analysis. Indeed, in conducting positive, empirical institutional analyses
it is generally the case that taking some institutional elements as given is appropriate for studying
the issue at hand. The language spoken by the members of the US Congress, for example, is
endogenous to their society, but examining its self-enforceability while considering the
determination of tax policy is clearly redundant.
Hence, an empirical institutional analysis is likely to entail considering particular rules as
being exogenously enforced on the interacting parties and some behavior and institutional elements
as exogenous to the analysis. Indeed, the distinction between endogenous, self-enforcing
institutions and institutions in which some institutional elements and behavior are taken as
4/8/01 - 8
exogenous highlights the fact that a society’s institutions are usually hierarchical. The institutions
that one member of a society faces are exogenous to him or her although they can be a choice
variable for other individuals or organizations. An analysis in which some institutional elements are
considered as exogenous is partial to the more general case of self-enforcing, endogenous institutions.
Hence, the discussion in this work concentrates mainly on the latter institutions.
Bridging the Methodological Divide: Integrating Institutional Statics and Dynamics
Considering self-enforcing institutions does not seem to be a promising starting point for
examining institutional dynamics. Clearly, if an environmental, exogenous change causes an
institution to no longer be self-enforcing, it will be intentionally changed or altered through
unintentional experimentation and mistakes. But how do self-enforcing institutions endogenously
change? Why do institutions tend to persist in a changing environment?
The challenge of combining the study of endogenous institutional change with the study of
self-enforcing institutions is well known: if an institution is self-enforcing, it is in equilibrium and all
changes must have an exogenous origin. Hence, approaches to institutional analysis that adopted the
agency perspective and, for example, game-theoretic analysis, have focused on institutional statics
rather than dynamics. The importance of examining endogenous institutional change, however, is
transparent and the deficiency of a framework that excludes endogenous institutional change by
assumption is substantial.4 The approach advanced here, enables us to consider how past institutions
influence their rates and direction of institutional change.
My position is that institutions persist in changing environments and cause endogenous
institutional change by reinforcing or undermining themselves. An institution is reinforced when
the social processes implied by it or other institutions (weakly) increase the range of situations in
which the behavior associated with this institution is self-enforcing. Examining such institutional
reinforcement requires considering the relationship between reinforcing processes and quasiparameters
that are the social and technological factors linking institutional statics and endogenous
dynamics.
4 Furthermore, for the study of institutional dynamics a rational choice framework has the
promise of bringing various lines of institutional analysis closer. This is particularly the case
regarding the deductive and historical approaches to institutional analysis in economics and rational
choice institutional analysis and historical institutionalism in political science. On the need for, and
benefit from, such integration, see, for example, Thelen, 1999.
4/8/01 - 9
Quasi-parameters have in common two properties: they can be gradually altered by the
behavior and processes an institution implies but their marginal change will not cause the behavior
associated with that institution to instantaneously change. These two properties imply that they are
parametric - exogenous and fixed - for the study of the institution at a given point in time, but we
have to consider them as exogenous and variable when studying institutional perpetuation in the long
run. Demography, wealth distribution, technological and other knowledge, preferences, various
beliefs, political power, and social networks are examples of social features that can be quasiparameters.
5 Self-reinforcing processes can thus reflect, for example, individuals’ responses to the
incentives institutions entail or the feedback from behavior to preference formation, knowledge,
information, and wealth distribution.
To consider concrete examples, ignore the possible inter-relationships among institutions. In
this case, an institution is reinforced if it is self-reinforcing, namely, exhibiting a positive feedback.
The behavior and processes this institution entails increase the range of situations in which the
behavior associated with it will prevail. For example, if norms evolve as many scholars have claimed
(e.g., Sugden 1989), through peoples’ tendencies to attribute normative value to behavior that was
followed in the past, this process self-reinforces the institution that originally generated this
behavior.
By considering institutional enforcement and reinforcement, the approach advanced here
helps integrate the analysis of institutional statics and endogenous institutional dynamics. For an
endogenous institution to prevail at a given point in time, it has to be self-enforcing. For an
endogenous institution to perpetuate over time, however, it has to be weakly self-reinforcing or
reinforced by the behavior and processes implied by other institutions. (Henceforth, I will refer to
both self-reinforcement and enforcement as (self-)reinforcement.) When (self-)reinforcement
transpires, institutional inertia occurs: past behavior will prevail in a larger parameter set.
Marginal environmental changes will not undermine past patterns of behavior. Indeed, behavior can
prevail even in situations in which it would not have emerged to begin with.
But an institution will change when negative (self-)reinforcement transpires. In such cases,
institutions are self-destructing: they foster processes that change quasi-parameters in a way that,
over the long run, undermine the self-enforceability of the associated behavior leading to
institutional change. Consider the following example. Suppose that beliefs in collective punishment
within a community lead to a particular regularity of behavior. To study this institution, we have to
examine this community, the beliefs, and behavioral rules as a self-enforcing system of institutional
elements generating that behavior. We have to examine why each member of the community is
endogenously motivated to retain his membership in the community, hold these beliefs, follow the
behavioral rules, and participate in a collective punishment. But even if such self-enforceability
prevails at a particular point in time, this does not imply that this institution will perpetuate over
time. For this to be the case, the institution has to be (weakly self-)reinforced. But this may not be
the case. For example, the economic success of the community implied by the collective
punishment may lead it to grow over time. Growth can undermine the self-enforceability of beliefs in
collective punishment because information transmission within a larger group may be too slow to
deter deviation. Similarly, each member of the community can become, over time, sufficiently
wealthy so that the threat of communal punishment will no longer be strong enough to make past
patterns of behavior self-enforcing. Such self-destructing processes are not merely a theoretical
possibility. Greif (2000a), for example, documents how self-enforcing institutions that supported
5 Institutional elements and their attributes, such as community size and distribution of beliefs,
can be quasi-parameters.
4/8/01 - 10
impersonal exchange in pre-modern Europe gradually eroded their economic efficiency and political
viability.
Differentiating between institutional elements, institutions, and the associated behavior also
enables us also to examine how past institutions influenced the direction of institutional change.
Past institutional elements, such as beliefs, communities, political organizations, and norms, are
enduring cultural, political, and social features that transcend the situations that led to their
emergence. Other social factors that were shaped by past institutions such as knowledge, wealth
distribution, and sunk organizational cost, are also enduring features that transcend the situation from
which they emerged. Because of their enduring and transcending natures, past institutional elements
and other social factors that past institutions implied influence the direction of institutional change.
Hence, when a new situation arises in a society because of exogenous or endogenous change,
institutions do not emerge anew reflecting only environmental conditions. They evolve in a spirallike
manner in which past institutional elements provide the foundation for, and direct the
emergence of, new institutions. For example, communities and political organizations that were
formed in the past constitute part of the (“endogenous”) rules of the game in new situations while
individuals carry beliefs that were crystalized in the past with them to new situations. But past
institutional elements only direct and do not determine new institutions. If they do not become part
of a new self-enforcing institution, they will decay over time and vanish. Institutions are outcomes
emerging from within and interacting with the legacy of past institutions (as the structural, inductive
perspective for institutional analysis holds). But for institutional elements inherited from past
institutions to persist, they have to become a part of the new institutions. Environmental factors
are important in directing institutional change (as the agency, deductive perspective for institutional
analysis holds).
An Analytical Framework for the Study of Self-enforcing and Self-reinforcing
Institutions
Advancing the study of institutions as an integrated, self-enforcing and (self-)reinforcing
system requires an explicit analytical framework. Such a framework has to contain a deductive
theory restricting the set of admissible arguments based on a priori, general principals. The
importance of such a restriction is well-reflected in the observation that promising institutional
research - from Veblen and Commons in economics to Parsons and Selznick in sociology - “fell into
disfavor, not because they asked the wrong questions, but because they provided answers that were
either largely descriptive and historically specific or so abstract as to lack explanatory punch”
(Powell and DiMaggio 1991: 2).
The analytical approach advanced here makes use of game theory to enable us to deductively
limit admissible self-enforcing institutions and empirically verify conjectures regarding the
prevalence of particular beliefs. There are many benefits to applying game theory to institutional
analysis. For example, a problem associated with the study of the behavioral implications of beliefs
is that beliefs are non-observable and even statements about them may convey little information
about the actual beliefs that people hold. Game theory facilitates overcoming this difficulty.
Classical game theory enables us to deductively restrict beliefs, even those regarding behavior in
situations that will never transpire, and consider their behavioral implications in a strategic situation.
Furthermore, it enables us to consider the strategic process of rule formation and the implications of
incomplete and asymmetric information. Similarly, evolutionary game theory enables us to consider
the spontaneous process through which rules and behavior evolve and the implications of
interactions among individuals with a limited understanding of their surroundings and abilities to
deduct and learn.6
6 The use of game theory for institutional analysis follows such studies as Schotter 1981; Greif
1989, 1994; Young 1993; Calvert 1995; and Aoki forthcoming. Game theory is still being refined
and enhanced and the approach presented here is not committed to game theory in its current form
4/8/01 - 11
Classical and evolutionary game theory can also be used to explore various aspects of
(self-)reinforcement (e.g., Greif 2000a). But the game-theoretic approach (which has been
extensively used by scholars who adopted the agency perspective) is inadequate for studying various
aspects of reinforcement. Game theory, for example, has a limited ability to explore how
institutions mold individuals, their norms, capacities, knowledge, and their cultural and social worlds.
Game theory has nothing to say, for example, about the appropriate specification of norms or
preferences in a particular historical episode, or the process through which they are shaped and
formed.
But these various specifications and processes can be integrated with a game-theoretic
framework. To achieve such an integration, we have to complement game theory with other
theories and insights developed elsewhere in the social sciences and used mainly in approaches
adopting the structural, inductive perspective. This work thus explicitly states the need for, and
explores the ability to use, game theory as a springboard for a broader, multi-disciplinary framework
relying on theories and insights developed in various social science disciplines, such as sociology,
social psychology, political science, and cognitive science.7 (In pursuing this integration I will follow
and expand on such works as Geanakoplos, Pearce, and Stacchetti 1989; Sugden 1989; Rabin 1993;
Greif 1994a; and Kuran 1995.)
Bridging the Methodological Divide: Deduction and Induction in Institutional Analysis
and, indeed, specifies its various limitations. More generally, throughout the discussion I attempt to
separate the conceptual and analytical frameworks. The former is the issue that we are attempting
to study while the latter is the analytics of doing so. Hence, the approach as a whole can
accommodate advances in the analytical framework. I concentrate on a single analytical framework
to simplify the presentation and have chosen game theory for reasons elaborated upon in the
following chapters.
7 Needless to say, I am ill equipped to pursue such an integration in much depth. But the
framework developed below makes exact the issues that require such integration and hence, I hope, it
will contribute to a dialog.
4/8/01 - 12
Studying an institution as a self-enforcing and (self-)reinforcing system of institutional
elements that generate regularities of behavior does not presuppose that either the agency, deductive
perspective, or the structural, inductive perspective is appropriate. Furthermore, the above
conceptualization of institutions does not imply any particular process through which institutions
emerge: they can be intentional outcomes or an unintended consequence; they can be designed by
individuals fully informed and knowledgeable about the situation or emerge unintentionally among
individuals with limited knowledge and information. Studying institutional elements accommodates
the case in which an institution is endogenous to the interacting individuals (expressing themselves,
for example, in a convention) as well as the case in which institutions governing the behavior of
some are the choice variables of others (expressing themselves, for example, in legal rules).
Similarly, such a study does not assert that institutions fulfill particular functions, such as efficiency
or distribution.
More importantly, the reliance on game theory and the view of institutions as composed of
inter-related institutional elements contributes to bridging the gap between the deductive and the
inductive perspectives in institutional analysis. Although game theory enables deductively restricting
the set of admissible, self-enforcing institutions, it indicates the importance of inductive analysis. On
the one hand, game theory reveals that the details of a situation profoundly influence the set of
possible self-enforcing institutions while multiple self-enforcing patterns of behavior and beliefs can
emerge in a given situation. On the other hand, game theory also reveals the difficulty of
deductively selecting among alternative outcomes and provides a limited guide to selection among
them.
Furthermore, because past institutions influence the rate and direction of institutional change
there is a need to integrate deduction and induction in institutional analysis. Institutions are selfenforcing
systems that can respond to changes in their environment but past institutional elements
may transcend the situation in which they emerged and influence the trajectory of their change.
Institutional dynamics is a historical process in which past institutions influence the rate and
direction of institutional change.
Hence, the above approach supports the view that institutional analysis is inherently
historical. One has to study institutions inductively within the context of the broader historical -
economic, political, social, and cultural - processes of which they are an integral part. Historical
analysis is required to identify the relevant transactions, expose institutional elements that can be
considered exogenous to the issue at hand, expose the details required for an appropriate contextspecific
model, form a hypothesis regarding the relevance of a particular institution, and reveal the
historical process of institutional selection. Positive, empirically-oriented, institutional analysis
necessitates a continual dialogue between history and theory. It requires an “interactive process of
theoretical and historical examination” (Greif 1997a: 87) that combines the social scientists’
emphasis on deduction and the historians’ on induction.8
Combining theory and history in institutional analysis is also motivated by recognizing that
there is more to institutions than selection from the set of all those technologically and physically
possible. Institutions reflect and embody a society’s collective knowledge so it is misleading to
initiate an institutional analysis by considering all the technologically possible institutions in a given
situation. As game theory indicates, even in relatively simple situations such an analysis is very
complicated and therefore likely to be a misleading representation of the set of institutions relevant
to the process of institutional selection. It risks imposing our conscious understanding of the
situation on the historical actors. Instead of learning how the knowledge prevailing in a particular
historical society is reflected and embodied in its institutions, we could superimpose our cognition on
the analysis and examine historically irrelevant alternatives. Combining induction and deduction is
8 The exact meanings of the terms “deductive” and “inductive” in this work are provided in
chapter 6.
4/8/01 - 13
thus essential for institutional analysis: history is necessary to identify the relevant institutions and
theory is required to deductively restrict hypotheses, reveal lines of causation, and generate
predictions that can be compared with the evidence.
To put it another way, the physical and social worlds within which individuals interact are
very complex. At the same time, individuals are endowed with limited rationality, knowledge,
cognitive ability, and information. (E.g., Denzau and North 1994.) We do not have a theory of how
people build mental constructs that enable them to make sense of their world and direct their
behavior. Accordingly, this work explores how to combine induction and deduction to reveal various
aspects of these constructs. In particular, it uses game theory to better formalize and understand the
rationale beyond particular patterns of behavior. Game theory enables us to evaluate the
implications of the particularities of individuals’ understanding of the world around them and exposes
the internal logic of this understanding.9
The above framework thus fosters our ability to address the questions motivating this
research. Societies evolve along distinct institutional trajectories because past institutions influence
the evolution of quasi-parameters, and past institutional elements provide the initial conditions in
the process of institutional adjustment following an endogenous or an exogenous change. The
process through which institutions evolve over time and situations is one in which new institutional
elements emerge to complement existing ones.
Hence, societies often fail to adopt the institutional structure of those societies that are more
successful because such an adoption is feasible only with respect to particular institutional elements,
such as rules and organizations, but it is not possible with respect to other institutional elements, such
as beliefs and norms. This implies that transferred institutional elements will fail to generate the
same behavior. The policy implications are that adopting some institutional elements may be
ineffective. To alter behavior, one has to first understand the institutions that prevail in the
adopting society and use this knowledge to devise complementary institutional elements that will
generate the desired behavior.
Empirics
A positive approach must lend itself to empirical evaluation. The approach presented here
can be tested since it makes its concept of institutions explicit and utilizes deductive theory to
restrict admissible arguments, expose lines of causation, and generate predictions that can be
evaluated empirically. Hence, although the study of institutions often does not lend itself to
quantitative analysis, the interactive, empirical methodology developed here facilitates evaluating
hypotheses in a way that is compatible with, but does not depend on, using such an analysis.
9 Inductive game theory explores strategic situations whose structures are not known to the
players. See Kaneko and Matsui 1999; Aoki forthcoming, chapter 9.
4/8/01 - 14
Such an empirical analysis does not begin by deductively examining the set of all possible
institutions in a given situation. It first combines historical and theoretical analyses to form and
evaluate a hypothesis regarding the relevant institutions based on evaluating and synthesizing microlevel,
historical and comparative evidence with predictions and insights from context-specific
models.10 Furthermore, it explicitly recognizes that some institutional features have to be taken as
exogenous. Once we have identified the relevant exogenous and endogenous institution and
understood the forces leading to its perpetuation, we can then examine how such factors as strategic
interactions, evolutionary forces, and awareness of alternatives made the other possible institutions
irrelevant and how past institutions have influenced the rate and direction of institutional change.
Historical Institutional Analysis
The approach developed in this work can be referred to as Historical Institutional Analysis
(HIA). It is motivated by, and attempts to gain insights through, comparative studies over time and
space. It recognizes the importance of combining history and theory in a positive, empiricallyoriented,
multi-disciplinary analysis of institutions as outcomes and societal features. Although some
institutional features have to be taken as exogenous in any empirical study, HIA is constructed to
help us take as endogenous more aspects of institutions than is usually the case. At the same time, its
explicit lines of causation restrict the set of admissible historical accounts so this approach is more
permissive (in considering how past institutions cause and direct institutional change) than standard
game theory. Nevertheless, in considering the historical development of institutions and their
implications, the proposed methodology is far less permissive (in allowable interpretations) than the
standard historical analysis.
Clearly, HIA is still evolving and many of its aspects will benefit from further elaboration and
evaluation, but it is sufficiently mature to merit exposition. The approach is “scientific” in the sense
that its constituting elements are explicit and it is based on refutable hypotheses and objective
information and measures even though in such scientific inquiries in general one cannot avoid a
degree of subjectivity in historical research. The researcher must, for example, select the issue,
identify the significant historical actors, and evaluate the extent to which predictions are confirmed.
HCIA is thus not an ultimate path to truth, but a methodology that incorporates a much broader
range of phenomena than an ordinary economic analysis. It thereby fosters the study of historical
issues that require an examination of this broader class of phenomena.
10 Surprisingly little attention has been devoted to delineating the use of game theory for nonstatistical
empirical analysis. Among the notable exceptions are Sutton (1991), Greif (1996; 1997a;
1997c), and Bates et al. (1998).
