Decisions and Organizations

The origins of the application of the notion of perfect rationality in political science may be found at the roots of the so-called rational choice theory. According to a couple of well-known proponents of this approach, the role of rational choice theorists is “to build models that predict how individuals’ self-oriented actions combine to yield collective outcomes” (Amadae and Bueno de Mesquita 1999, p. 270).

Nevertheless, in a deeper sense, two different sources—both just after World War II—may be found that explain this general aim of rational choice theories. On the one hand, a first source was the realization that computers were capable of symbolic processing (Simon 1996b), and thus computer programs could be created that “would properly model those mental processes involved in intelligent human behavior whatever they may be: information search, problem solving, learning, invention or creativity” (Casanovas and Noriega 2007, p. 16).

On the other hand, a different (though related) source was game theory, developed after World War II by John von Neumann and Oskar Morgenstern.

In effect, the application of game-theoretical approaches to political science and, in particular, to international relations is one of the keys to interpret the popularity of this theory of rational decision among political scientists until our days. The other is the elegance of their theory of decision.

Most models in this stream of political analysis start with a set of prior assumptions (Bennett 1995):
Some early formulations of the rational-choice approach in political science felt strongly attracted to this view of human rationality, following the basic scope and methods of their leading counterparts in economics. But a particular elaboration of this set of assumptions for political science was not fully developed until the end of the 1950s, Anthony Downs and William H. Riker perhaps being the most prominent and cited representative authors of this shift.⁷ In different ways, both authors set up the theoretical framework (Downs 1957) and most of the analytic tools (Riker 1958, 1962) upon which mainstream rational choice theories were and still are built and developed.

A foundational approach in modern political science such as Downs’ An Economic Theory of Democracy (Downs 1957) combined a “neoclassical” approach to rationality with an account of how both individuals and organizations behaved in the political process. But while Downs’ contribution to the theory of rationality is very modest (the basic neoclassical assumptions remain untouched in his An Economic Theory of Democracy (1957)), it certainly implied a momentum in the analysis of democratic societies, thus embodying a new and fresh look at political behavior.

Downs’ notion of rationality draws directly on Kenneth Arrow’s view of economic rationality (Downs 1957, p. 6): “A rational man is one who behaves as follows:
As to institutions, Downs argues that “the government provides the framework of order upon which the rest of society is built” (Downs 1957), which is a close formulation to that of Herbert A. Simon (1983) who—as we saw in the Introduction—views organizations as providing “a stable environment […] that makes at least a modicum of rationality possible”.⁹ In particular, Downs sees political order as a guarantee to reduce uncertainty in choice, in the sense that a stable political order—i.e., a stable institutional or organizational environment—provides the individual with an acceptable degree of prediction regarding other people’s actions. Yet Downs does not go beyond this point as far as institutional analysis is concerned. Instead, such organizations as political parties or governments are presented in his model as rational, power-seeking agents, thus obviating any reference to organizational or institutional factors affecting decision-making in some way.

Downs’ model contains a number of relevant methodological assumptions. First, it assumes a radical methodological individualism according to which collective behavior must be viewed as aggregate individual actions. In order to account for these actions, both humans and organizations such as governments are viewed as goal-oriented, self-interested agents. Furthermore, their goals are represented by preferences, which agents try to maximize .¹⁰ Since information is costly and incomplete, Downs’ model must account for uncertainty, i.e., “any lack of sure knowledge about the course of past, present, future, or hypothetical events” (Downs 1957, p. 77), which the author concedes can be reduced by devoting certain amount of resources to information.

Therefore, Downs assumes a model which is partially based on the traditional neoclassical view of individual decision-making (utility maximization ) but introducing uncertainty (or information cost) as a correction factor to the model, which is especially relevant for his account of persuasion and, in fact, for the mere existence of political parties in the political process. Put in a simple way (Downs 1957, Chap. 6), since people lack knowledge and complete information, both government and the opposition party—assuming a two-party system—use persuasion in order to provide voters with the information they need to decide their vote, i.e., in order to gain the voters’ vote.¹¹ Although Downs’ book contains numerous assertions about the world (factual assertions), no evidence is provided.

Thus Downs’ is a deduced model,¹² one that presents politics as a market where voters’ preferences are sold and bought, where parties (and governments) are sellers and voters are buyers, and where persuasion and ideology are explained by the mere existence of uncertainty.

Downs’ work has been also considered—both by its critics (Simon 1985) and supporters (Riker 1961)—as one of the first contemporary attempts to articulate a coherent model of politics bringing the mathematical and game-theoretical apparatus from economics to political science. Reading his Economic Theory of Democracy, one can certainly conclude that the author applies the maximizing utility principle to politics, but in actuality his application is rather verbal, not extremely formalized. This might mean that Downs offered a general model in partial concordance with the main objective of his famous work, namely “to provide […] a behavior rule for democratic government and to trace its implications” (Downs 1957, p. 3). He undoubtedly provides the model, but obviously its implications for political science were far from foreseeable at that moment.

On the other hand, a first systematic application of this general model—and of game theory and social choice theory , with the subsequent powerful analytic apparatus—to particular political phenomena was Riker’s early work on voting (Riker 1958). This was a first hint of one of the most influential authors in what shall be known as positive political theory (Riker and Ordeshook 1973) or, more generally, rational choice theory .¹³ Yet the actual benchmark of this approach will be Riker’s Theory of Political Coalitions (Riker 1962), an application of the theory of n-person games to political coalitions, and a canonical text in rational choice literature.¹⁴

As noted above (in agreement with Rothstein 1998), Riker’s radical methodological individualism and his turn in the mid-1950s to game-theoretical explanations of political phenomena represents an extreme view in our human-organization interaction model. Regarding his theory of rationality, its basic principles are—like Downs’—borrowed from neoclassical economics and game theory¹⁵ and they may be seen as a set of principles and assumptions to be shared and accepted as foundational by future rational choice theory practitioners.

As stated above, Riker’s work has usually been described as a radical individualistic approach to politics. Methodological individualism is apparent throughout his work after his first relevant contribution to election theory (Riker 1958) and it is maintained along later works. As a result, Riker and Ordeshook (1973, p. 1) contended in their Introduction to Positive Political Theory that “[people] are bundles of opinions about nature and of preferences about the alternatives nature offers them”.

As to rationality in particular, it is assumed that “people behave as if they sort out and logically arrange the preferences in their bundles”. Besides, the authors further assume that these people “are able to use the sorted bundles for decisions so that, when faced with a choice, they can choose as directed by their preferences”. In conclusion, “[s]uch behavior we describe as rational” (Riker and Ordeshook 1973, p. 8).

Riker’s first broad definition of rationality runs as follows (Riker 1962, p. 22):¹⁶

Given social situations within certain kinds of decision-making institutions (of which parlor games, the market, elections, and warfare are notable examples) and in which exist two alternative courses of action with differing outcomes in money or power or success, some participants will choose the alternative leading to the larger payoff. Such choice is rational behavior and it will be accepted as definitive while the behavior of participants who do not so choose will not necessarily be so accepted.

Austen-Smith and Banks (1998, pp. 263–264) offer a formal account of the very same assumptions or premisses:

The primitives of any rational choice-theoretic formal model of politics include a specification of a set of the relevant individuals, denoted N; a set of feasible alternatives or outcomes, X, and, for each individual in N, a description of her preferences over the set X. […]

To formalize the idea of individual preferences, each individual in N is assumed to possess a binary preference relation on X, denoted R _i , where, for any two alternatives x and y, “” is shorthand for the statement “according to individual i, x is at least as good as y”. From R _i one can define i’s strict preference relation, P _i (where “” reads “according to individual i, x is strictly better than y”) and i’s indifference relation, I _i (where “” reads “according to individual i, x and y are equally good”). […]

A number of restrictions are established for this model, though, most of them being aimed at assuring that “individual-level decision problems are well defined” (Austen-Smith and Banks, 1998, p. 264), namely that possible flaws of collective rationality are never to be ascribed to individual irrational behavior but to collective choice paradoxes and problems. These restrictions refer to individual preferences regarding their completeness, consistency, continuity, and convexity (Austen-Smith and Banks 1998).

Taking this theoretical framework into account, the essence of politics is then “the selection of the preference of some person (or the potential preference of some person) to be the choice of society” (Riker and Ordeshook 1973, p. 2).¹⁷ It refers to three different processes or problems that in my opinion synthesize Riker’s research program:
This third problem—outcome production from choices—appears in Riker’s work more general than the others in the sense that it has not deserved an as thorough treatment as the other two. Yet at the same time it is recognizable throughout all his work, though in my opinion is best represented by his notion of heresthetics, or the art of winning in politics either referred to as political manipulation (Riker 1986), or simply rhetoric (Riker 1996).

Nonetheless, it is precisely this factor what enables us to introduce Riker’s notion of the role of institutions in the game of politics. Briefly stated, Riker’s theory of political behavior requires a self-interested, rational agent who acts in a strategic, game-shaped environment. Riker’s argument would be that so long as the output of politics refers both to the results that can be arrived at and those that ought to be obtained in social choice (Riker and Ordeshook 1973), institutions may have a role in setting the agenda of what is possible and desirable as game outputs. Like Downs, the basic assumption is that institutions are the rules of the game of politics, and that as rules they can be changed favoring particular actors.

This idea is apparent in the first of Riker’s relevant papers pertaining to the rational choice approach (Riker 1958), in which the author explored the risk of irrational collective behavior (voting paradoxes) in the US Senate. Riker concludes that “[w]ith the consequences exposed, it may be that the Congress will wish to change its Rules in order to avoid the risk of behaving irrationally” (Riker 1958, p. 362). But the author immediately warns that these rules may also enable perverse behavior and results, in the sense that “the present Rules allow a skilled minority to defeat an unsophisticated majority; or, to put it differently, a committee chairman to control House action” (Riker 1958, p. 362).

On one hand, Riker focuses the function of agenda control on particular actors that “guide the operation of the body” within which collective decision-making takes place (Riker 1982).¹⁹ This refers to different strategies these actors may carry out in order to seize the power of the decision-making body by taking control of some aspects of the voting procedure, i.e., establishing the order in which alternatives are presented, or generating themselves other alternatives to be voted upon.²⁰

On the other hand, Riker’s acknowledgement of this “institutional” fact does not mean that the author takes on an institutionalist analysis of the political process at all in the sense of carrying out a thorough analysis of the institutional mechanisms that take place in agenda setting or manipulation. Instead, this phenomenon is coherently analyzed from a strategical, game-theoretical perspective in which a number of self-interested, strategical-behaving agents seek to alter voting procedures in such a way that they can take profit of such phenomena as majority cycling.²¹

Although Riker’s work on this particular issue does not seem to reach the systematic treatment as the other two dimensions of politics as he saw it, most of his later works are devoted to this phenomenon (Riker 1986, 1996), and his view of the potential role of institutions in political output was to be highly influential in rational choice-based, new institutionalist approaches.

In conclusion, this brief review of the first attempts to present rational choice models of political behavior shows that these models tended to overweigh individual decision making as the result of the application of preference maximizing strategies in different political contexts. In these models, rationality is defined drawing upon Arrow’s development of neoclassical economic concept of rational action.

These principles imply that rational decision making has at least the following features:
The influence of Riker’s contributions is apparent in numerous examples of this kind of analysis that may be found throughout the early work of a group of his students, who were to become the flagship of the rational choice theory in political science.²² These authors, and others, elaborated on the model to focus on party behavior (Ordeshook 1970), Pareto optimality in elections (Ordeshook 1971), game-theoretical rationality in elections (Riker and Zavoina 1970; Riker 1976; McKelvey 1976), spatial models of party competition (Hinich and Ordeshook 1970; McKelvey 1975; McKelvey and Ordeshook 1976; Aldrich and McKelvey, 1977), various aspects of collective choice in legislatures or similar bodies (Riker 1958; Schwartz 1977; McKelvey 1979; Weingast 1979; Ordeshook and Winer 1980), explorations on the (im)possibility of social choice functions under particular requirements (Ferejohn et al. 1982; Ferejohn and McKelvey 1983), and the relation between information and the aggregation of preferences (McKelvey and Ordeshook 1985a, b; McKelvey and Page 1986; McKelvey 1986; McKelvey and Ordeshook 1986; McKelvey 1987; McKelvey and Page 1990; Nielsen et al. 1990; McKelvey and Patty 2006).

As to organizations or institutions, these models do not explicitly deny that decisions are actually made within institutional environments, but the game-theoretical framework used to analyze political phenomena implies that “the variation among institutional, psychological, and ideological components of behavior are minimized” (Riker 1967, p. 642). Therefore, no institutional arrangements are explicitly taken into account that shape these individuals’ preferences or their very ability to perform decisions.

This account of institutions might be interpreted as to be some kind of “rules of the game” (Downs) in the logical sense that decisions are not made in an empty world. These rules might also be interpreted as outlining the available alternatives in a decision situation (Riker), thus foregoing future rational choice-based institutional analyses that shall be commented on in the next two subsections.

The interest in institutions by rational choice theorists originated with the intention of bridging the gulf between abstract models of collective choice (in an empty and ahistorical world) and “more specialized formalizations which incorporate institutional specifics” (Shepsle 1985, p. 7).²³ In particular, the main claim made to traditional social choice models was that “[n]early thirty years of research on social choice has produced a large body of theoretical results. […] While attention has been devoted to the mechanisms by which individual revealed preferences are aggregated into a social choice, rarely are other aspects of institutional arrangements treated endogenously” (Shepsle 1979, p. 27).

Therefore, it might be said that current rational choice theory has changed some of its foundational premises regarding the role of institutions in decision-making. Or otherwise stated, this approach to political phenomena has enriched its analysis by introducing institutions and their operation into its analysis. For, while it is still founded upon the idea of an individual or set of individuals with a well-defined set of preferences, institutions appear—following Riker’s statements about it—either as the main source of available alternatives, their order, and the shape in which the information available is presented (Shepsle 2006a), or as equilibrium institutions (Shepsle 2006b).

Therefore, in the next subsection we will deal with those rational-choice theories that assume a negative view of institutions (E, in Table 2.3), highlighting the discussion that led to the introduction of institutions. Then, Sect. 2.3.3 will outline rational-choice theories that assume a positive view (B).

The general idea of this branch of rational choice theory regarding institutions—coherently with their foundational premises—is that they are the “rules of the game” (North and Weingast 2000) or “game forms” (Shepsle 2006b) that act as a constraint to agents’ action. In other words, these rules establish “the set of players, what actions players may take (must take, may not take, etc.), informational conditions under which they make their choices, the timing of moves, the role and timing of exogenous events, and the outcome that results from a distribution of choices and the realization of all random factors” (Shepsle 2006b, pp. 1031–1032). These rules may be formal (e.g., contracts) or informal (e.g., social taboos) (North 1990).

Shepsle’s paper on institutional arrangements and equilibrium in social choice models (Shepsle 1979) was one of the first attempts to make explicit the role of institutions in rational choice models. Therein the author focuses on structure-induced equilibria, i.e., “the play of the game” itself (Shepsle 2006b), in the sense that a number of institutional mechanisms such as rules of jurisdiction or amendment control have “an important independent impact on the existence of equilibrium and, together with the distribution of preferences, […] [they co-determinate] the characteristics of the equilibrium state(s) of collective choice processes” (Shepsle 1979, p. 29). Thus institutional arrangements affect outcomes.²⁴

A year later Riker conceded that “we cannot leave out the force of institutions” (Riker 1980a, p. 432) because they can “affect social outcomes”. In this work institutions are understood as the embodiment of a set of conventions, namely “rules of behavior, especially about making decision” in the sense that “decisions are customarily made within the framework of known rules, which are what we commonly call institutions” (Riker 1980a, p. 443). These institutions—Riker argues—affect somehow the content of decisions,²⁵ and therefore produce biases in decisions, social and individual.

Then the author concludes that since decisions are affected both by preferences and by institutional arrangements, institutions are worth studying, appreciating Shepsle’s effort to include these factors—institutions as structure-induced equilibria—in rational choice models. Yet Riker objects that since institutions are mere sets of rules and that actors in social choice may want to change them in order to induce particular equilibria or outcomes in the future, institutions may be seen simply as “more alternatives in the policy space and the status quo of one set of rules can be supplanted with another set of rules” (Riker 1980a, p. 445).²⁶ Thence—Riker goes on—institutions are to be interpreted as “congealed” preferences, and if preferences lack equilibrium, “then also do institutions”. Therefore, institutions are not to be considered “fixed exogenous constraints [but] changeable [and endogenous] features of the political landscape” (Shepsle 2006b).

Later, Shepsle (2006b) revisited Riker’s objection to the static treatment of institutions. Shepsle’s counterargument to Riker’s is that institutions may well still be considered as structurally inducing equilibria since, as it is usually the case in democratic legislatures, the same kind of majority is not sufficient for both policy decisions and institutional change decisions. In the latter case, numerous constitutions require special majorities in order to make collective decisions upon particular issues (such as the very structure of the state, or to remove a monarch). Then, from this standpoint institutions can be viewed as something more than Riker’s “changeable features of the political landscape”.²⁷

Shepsle’s response to Riker’s objection reveals a central tension between the study of institutions and rational choice theories, one that focuses on the importance of considering what is the object of research at any given time and, most importantly, what are the units of analysis. On the one hand, Riker’s endogenous consideration of institutions is critical to the study of the origins of institutions, institutional change, and evolution. If institutions are endogenous to the political process understood from the strategic model of politics, then they cannot strictly matter (Przeworski 2007). Yet, on the other hand, even if institutions are seen as endogenous to the political process, comparative, micro-level analyses on how agents “exploit their institutional environment” (Shepsle 2006b) may still be in need to account for variation in institutional settings as game forms. This partly explains the interest on how critical institutional arrangements such as electoral systems (Benoit 2004; Boix 1999; Helmke and Meguid 2008), democratic regimes (Acemoglu and Robinson 2005; Boix 2003) and the law (Hadfield and Weingast 2011) were adopted, as “primal settings” (Shepsle 2006b) or rules of the game that respond to their creators’ preferences (Greif and Kingston 2011).

The micro-level analysis of institutional arrangements as constrains has led numerous rational choice theorists to elaborate on the traditional view of institutions as settings for agenda manipulation. Indeed, the renewed consciousness on the role of institutions led to an important change in rational choice theories “moving from models of voting—especially models of pure majority rule with their attendant emphasis on voting cycles and system instability—to an understanding of how institutional rules can shape collective choice” (Moe 1987, p. 476), focusing on both the rules themselves and the mechanisms set up to enforce them as separate instances.

Coherently, Weingast (1998, p. 169) has called this agenda-centered approach the “setter model”. According to it, institutions set the agenda of alternatives among which a decision is to be made. In simple setter models, a single individual, group, or organization holds the exclusive power to set the defined set of alternatives according to which voters, other organization members or any particular group of decision-makers will construct their utility function. This general framework, though, permits a variety of approaches and applications including the study of constitutions (Riker 1984; Acemoglu 2005; Weingast 2005; Colomer 2006; Shane 2006), executives (Rhodes 2006a), jury decisions (Guarnaschelli et al. 2000), judicial behavior (Epstein and Knight 2000), and international relations (Niou and Ordeshook 1987, 1990, 1991, 1994a, b, c; Schultz and Weingast 2003).

Yet the basic setter model has privileged legislative bodies as the main object for the application of rational choice models, since they constitute interesting and typical collective choice scenarios.²⁸

In the 1980s, during the first institutionalist wave of rational choice theory, a typical institution—or set of institutions—for collective choice such as the US Congress deserved most attention by rational theorists.²⁹ These legislative studies show a varied number of objects, ranging from internal legislative rules to the relations between legislative bodies and other actors.³⁰

For instance, Shepsle and Weingast (1981) offer an explanation of the way institutional arrangements—among other features—can produce biases to expected rational behavior in legislative settings. In effect, the authors explain the “empirical regularity of universalism in the realm of distributive policies”, which means that in numerous legislative bodies rational legislators tend to support “coalitions well in excess of minimal wining size”.³¹

Moreover, Shepsle and Weingast (1982, 1984b) generalize this model exploring the institutional arrangements that underpin “the operation of majority rule legislatures”.³² Following this vein, Denzau et al. (1985) offer a model of legislators’ sophisticated voting, and Krehbiel (1985 1986a, c) has showed other intricacies of legislative voting and has explored methods to better outline legislative decision-making from a rational choice perspective (Krehbiel 1986b).³³

Shepsle and Weingast (1987), Krehbiel et al. (1987), and Krehbiel (1990) focus on the legislative process (proposing, amending, and vetoing) in order to model and explain legislative committee power—Krehbiel (1988) present a case study on Senate voting—, while Ordeshook and Schwartz (1987) and Ordeshook and Palfrey (1988) produced a couple of influential studies on the precise process of agenda formation and control. Furthermore, Ferejohn and Krehbiel (1987) advanced the study of budget decisions by members of Congress. Austen-Smith and Riker (1987, 1990), and Ordeshook and Palfrey (1988) pay attention on information distribution in a legislative, agenda-setting scenario, and Gilligan and Krehbiel (1988, 1989, 1990, 1994) have devoted much attention to the relation between information and legislative rules.

In this vein, Weingast (1989) explores the impact of the open rule in the US Congress and uses it to predict legislative behavior, and Cox (2000) reflects generally on the way legislative rules affect the outcome of legislation. Also, McKelvey and Riezman (1992) treat seniority as an endogenous rule that legislative institutions enforce in order to procure the legislators’ reelection, and Krehbiel (1993, 1999a, b, 2000) explores political parties’ strategic behavior to control legislative outcomes.³⁴ Finally, other studies (Lupia and McCubbins 1994) incorporate learning processes to enrich legislative behavior models.³⁵

Variations on the same theme assume that the setter that determines the content of the agenda is not always clear, and that influence may be rather obscure. This fact has recently favored rational choice approaches to bureaucracy. For example, Moran and Weingast (1982) offer an explanation of the way the U.S. Congress influences decisions made by (supposedly) independent regulatory agencies. In the same line, Weingast and Moran (1983) expand the model to explore bureaucracy and its influence on the agenda, while McCubbins and Schwartz (1984) and Calvert et al. (1989) build theory and apply it to particular cases in order to outline bureaucratic discretion, and to explain political motivations in congressional political behavior regarding regulatory decisions (see also McCubbins 1985). Indeed, further attempts to include the role of bureaucracy in decision making models have been interpreted as to “hold the promise of moving toward a well-integrated perspective on the larger institutional system”, implying “a pioneering move toward a coherent general theory of political institutions” (Moe 1987, p. 476).³⁶

In conclusion, so far we have seen that while the first records of rational choice models tended to be unidimensional by neglecting the influence of institutional arrangements in collective decision making, this assertion does not hold since the mid 1970s and 1980s. At that moment a new institutionalism pervaded political science including the most influential rational choice theories. Secondly, we have showed that a dominant branch of this approach adopts Riker’s basic agenda-setting model.

Despite the variety of approaches to institutional operation turns to be evident, a number of features seem to remain constant in almost all institutionalist rational-choice approaches reviewed so far. This model holds on a number of assumptions, namely:
Consequently, since authors have focused mostly on how decision outcomes may be manipulated, given both fixed assumptions on human rationality and fixed (though changeable) institutional environments, actual decision processes are hardly approached or studied in detail. Therefore, scant attention has been paid to show evidence of the actual processes that enable decision makers make their decisions, namely, the actual interaction (if any) between humans and the organizations within which they act.

However, this view on institutions as a constraint, being dominant in rational choice theory, is not unique. Next section is devoted to briefly outline an alternate view of institutions within rational choice theory, a perspective that departs from the traditional view of institutions to embrace an account of their influence in decision processes that we may consider as rather positive.

The first question one must ask when attempting a review of rational-choice approaches with a positive view of organizational factors to decision making is whether it makes any actual sense to defend such a position around organizational influence on decisions if at the same time one assumes humans are perfectly rational.

At a philosophical level, for instance, it could fit a Hobbesian view of the political process, in the sense that institutions at least enable human decision-making under certain reasonable conditions (security, peace, etc.). In the case of Hobbes himself, the part “Of Man” in his Leviathan (Hobbes 1651) contains an important auxiliary assumption about the human condition according to which humans, in order to achieve their goals, can be at least as naturally aggressive as rational.³⁷ In such a case, then, it is hard to think that, though aggression is rational, it is possible to make other (perfectly) rational decisions in an environment of permanent mutual aggression. Therefore, organizations, if not enhance human decision making, at least enable it.

A positive account of institutions (in our sense) for collective choice from a rational choice perspective may only be found in what has been generally understood as one of the two standard ways rational choice theorists have interpreted institutions, i.e., institutions as equilibria (Shepsle 2006b) or endogenous institutions (Calvert 1995; Shepsle 2006a), in opposition to institutions as exogenous constraints, as were the ones presented in the previous section. This does not mean that all endogenous interpretations of institutions are to be interpreted as positive accounts of institutions in our sense, but we shall see that it is reasonable to think that all such positive accounts in rational choice theory are developed under this perspective.

In principle, the approach that views institutions as equilibria does not imply a radical departure from the traditional—we might say Downsonian—view of institutions as “rules of the game”. But these rules or institutions are not observed as external objects or factors to be accounted for in the analysis of decisions made in the political arena. An example of institutional analysis of the external type would be seeing market behavior as a cooperative game in which some exogenous elements exist (property rights, legal principles, etc.) in order to enforce agreements between agents (e.g., contracts). Yet as we have seen in the previous section, political processes (such as legislature choice) are not usually analyzed as cooperative but as non-cooperative games, because doing otherwise would pose the problem of lacking “any exogenous enforcement mechanisms” (Shepsle 1985, p. 8).³⁸ Thence the internalization of institutions as inner, not explicit rules for modeling collective choice in the political field. In effect, these rules are assumed to be provided by the actors themselves in their interactions and, even more, “[they] reflect the willingness of (nearly) everyone to engage with one another according to particular patterns and procedures (nearly all the time)” (Shepsle 2006a, p. 25). This view, then, “focuses on how interactions among purposeful agents create the structure that gives each of them the motivation to act in a manner perpetuating this structure” (Greif and Kingston 2011).

This focus on willingness, perpetuity, and ultimately on equilibrium are the main reasons why we count this perspective on the “positive” side of rational choice institutionalism. For, institutional arrangements here are not viewed as constraints to action or decision but rather as mechanisms that enable collective choice where action is based upon the expectations agents have regarding the behavior of others.

This approach is still recent and to our knowledge only a few examples or applications of this notion of institutions from a rational choice perspective may be found in the main literature. Shepsle (2006a, p. 25) calls Calvert (1995) “one of the intellectual architects of this perspective”, though Weingast (1998) includes this approach in a broader branch of rational choice studies on institutional stability and, in particular, links it to the concept of self-enforcement, thus focusing on Douglas North’s contributions to institutional evolution and stability.³⁹ As Engerman and Sokoloff (2008, p. 129) point out, these institutional arrangements “might be written or unwritten, formally spelled out or informally known, but, in any case, they come to be accepted by the relevant members of society”.

This path leads to a major change in the focus of research. The previous view focused on how the institutional framework shaped behavioral outcomes—variation may be observed both in outcomes and in institutional constrains, but players abide by the rules. In institutions as equilibria, two different scenarios may be envisaged. First, when an institution is in equilibrium with its environment and it is resistant to exogenous shocks, it can be viewed as a constraint, since “the behaviors it fosters appear to be equilibrium resopnses to the opportunities provided by this (equilibium) arrangement” (Shepsle 2006b). If, on the other hand, variations in the environment shake the foundations of the institution so that departures from the rules may be observed among agents that create an entirely new strategic scenario, the institution-as-constrain model is no longer adequate to account for institutional change. The view of institutions as equilibria, then, ultimately leads to the study of how agents “are motivated to follow institutionalized patterns of behavior” (Greif and Kingston 2011) and how these motivations change—how the parameters that characterize the “primal strategic settings” can change in response to exogenous shocks (Shepsle 2006b).⁴⁰

Be that as it may, the main literature on the endogenous approach to institutions from rational choice theory assumes that it has two main differences with respect to the exogenous perspective. The first one regards the level of analysis. The second, the nature of institutions.

As to the first question, in comparison to the view of institutions as exogenous constraints to action, the endogenous perspective is considered to constitute a different level of analysis, namely, it “takes the study of institutions to a deeper level” (Weingast 2002, p. 661). This means that while the perspective on institutions presented in the previous section tended to treat institutions as a given set of rules (of the game), or as a given or fixed environment, the endogenous perspective sets forth the need for institutions, i.e., why do humans need such environments to perform certain decisions. This is the main reason why most reviews of this perspective turn to Douglas North’s (and colleagues) literature on institutional history, change, and stability, thus providing “microfoundations for macro-political phenomena” (Weingast 2002, p. 661).

Regarding the second question, from this perspective institutions can be viewed either as structured bodies of rules and arrangements, or rather as unstructured patterns of behavior (Shepsle 2006a). Therefore, rules may be formal or informal, though the focus on self-enforcement may promote studies on the latter. In our opinion this is one of the reasons that may explain the less success of this approach in rational choice theories, in comparison to its (negative) alternative.⁴¹ In effect, endogenous institutions are seen as “equilibrium ways of doing things” (Shepsle 2006a, p. 26), or “as an equilibrium of behavior in an underlying game” (Calvert 1995, p. 58). Calvert further elaborates that “there are no institutional ‘constraints’ or ‘preferences’ aside from those arising out of the mutual expectations of individuals and their intentions to react in specific ways to the actions of others, all in an attempt to maximize their utility in a setting of interdependency. Institution is just a name we give to certain parts of certain kinds of equilibria” (Calvert 1995, p. 74).

This approach from rational choice institutionalism has different instances in recent political science production. First of all, and following North’s contribution to the origins and evolution of economic institutions, the endogenous perspective has given rise to new historical accounts of political institutions. Secondly, it has promoted the study of unstructured patterns of behavior as endogenous institutions that lead to different kinds of equilibria.

As to the first group of studies, North and Weingast (1989) explore how certain institutional arrangements in seventeenth-century England allowed for a government commitment to property rights and thus to economic growth. Yet while traditional approaches to institutional change had usually focused on the establishment of a set of rules that enforced the king’s obligation and action, this new perspective tries to set up a model to show that the institutional changes implemented after the Glorious revolution “altered the incentives of governmental actors in a manner desired by the winners of the Revolution” (North and Weingast 1989, p. 804).

Summing up, “[t]hese changes reflected an explicit attempt to make credible the government’s ability to honor its commitments”, securing private rights in the mid-long run and thus promoting economic growth. Therefore, the authors show an alternate model of historical institutional analysis to that of rules as a constraint or control to power; instead, they promote a positive view of institutional change in the sense that the new arrangements enabled the king’s commitment to be credible and in consequence that private rights such as property rights were guaranteed.

A second example is given by Weingast (1998) who applies this model to explain American political stability before the Civil War, in the sense that a set of rules such as “the balance rule were necessary to implement and maintain American federalism during the antebellum era” (Weingast 1998, p. 178).

More recently, Bueno de Mesquita (2000) delivers an alternate endogenous interpretation of the origins of sovereignty. In his model, the origins of the modern territorial state model are to be found in the Concordat of Worms rather that in the Treaty of Westphalia. In a few words, the “Concordat of Worms created a property right over territory (specifically, bishoprics) […] [that] created competition between kings and the pope over political control. The contest for control manifested itself, in part, through the erection of new political institutions” (Bueno de Mesquita 2000). Moreover, while the church promoted institutions that tended to demote economic growth (in order to maintain its influence and dominance over monarchies), kings tended to design institutions that promoted economic growth as a way to take the church’s weight off their shoulders. This change had further institutional consequences, namely, “[t]he competition that was institutionalized at Worms fomented the construction of new executive, judicial, taxing, and legislative apparatus that eventually emerged as the state as we know it” (Bueno de Mesquita 2000). However, not all endogenous approaches have a historical perspective. As noted above, a set of authors have rather focused on the informal way rules are institutionalized, delivering studies on unstructured institutions. An example of equilibrium pattern of behavior from this perspective is senatorial courtesy. It refers to an American practice by which a presidential appointment might be revoked if either the senior senator of the president’s party, or the senators from the state to which the appointment applies raise objections to it.⁴²

A further example explores the self-enforcement of some endogenous institutions that in the long run lead to social cooperation. In this sense, the studies by Axelrod (1980a, b), Axelrod and Hamilton (1981), Axelrod and Dion (1988), and Axelrod (2006) on the application of the prisoner’s dilemma (in an artificially created computer tournament) show that an institution such as tit-for-tat⁴³ can lead to stable cooperation on successive repetitions of the same game.⁴⁴ Weingast (2002) observes that it shows that “in some contexts, the incentives of repeated play are sufficient to enforce long-term social cooperation”, to conclude that “a fundamental aspect of institutions is that they provide the means for the enforcement of cooperation”.

In conclusion, though a precise description of institutions from this perspective is sometimes perceived as difficult, it is assumed that they “structure human actions by providing incentives that shape economic and political behavior” (Engerman and Sokoloff 2008