Frantz, Christopher

The dynamics and evolution of self-governance systems are difficult to understand. Human choice plays a fundamental role. However, this relationship—especially collective behavior and institutional design—is poorly articulated in theory and research. Government involves Faustian bargaining—exchanging individual liberty for collective efficacy, security, and public good delivery. This exchange is central to concepts of the “State” and State-reinforced self-governance (SRSG)—how governments enable/constrain societal self-governance for public good provision and commons management. A behavioral theory of institutional design requires understanding of human motivation, reasoning, and decision-making, as well as learning and memory. This understanding must build on individual and group processes to articulate how boundedly rational agents conceptualize social-ecological dilemmas and governance systems. It must also account for dilemma stakeholders’ strategic positions, goals, and beliefs.

We address this challenge by building on prior attempts by Elinor and Vincent Ostrom to account for Bayesian reasoning and Faustian bargaining in constitutional choice. We integrate these perspectives with principles of social cognition and learning, developing both a conceptual framework and agent-based model (ABM) of the individual and collective learning and decision-making processes involved in creation of self-governing systems. The framework outlines core premises of boundedly rational constitutional decision-makers. Constitutional choice is conceptualized as a bargaining process, whereby stakeholders discuss alternative institutional designs in terms of (a) configurations of design features (e.g., collective choice and regulatory arrangements) that (b) bear on actors’ fundamental needs and liberties (e.g., self-determination, procedural justice, security) and (c) strategic goals. The ABM attempts to empirically test these assumptions with data from a lab experiment, investigating evolution of regulatory systems in a commons dilemma. We describe how constitutional agents form and update mental representations (mental models) via communication, and form preferences for particular institutional designs based on current mental models, goals, and needs.

We further employ the Institutional Grammar and communication coding techniques as the basis for institutional analysis and integration with SRSG and the ABM. This research informs behavioral theory underlying governance of commons, State influence, and collective action via constitutional decision-making.

The Institutional Grammar (IG) provides a data model for reasoning about how an organization “moves” through time via the underlying behavioral strategies, norms and rules – its institutions. If the position of an organization is described by its team members, activities, resources, and the relationships between them, then a data set of institutional statements can serve as an estimate of that organization’s velocity. It’s possible to extrapolate an informed estimate of an organization’s future position from knowledge about its current position and its velocity (i.e. its rules, norms and strategies for allocating resources), along with beliefs or projections about its future (external) circumstances. The choices that affect an organization’s velocity thus constitute the form of “steering” commonly called “governance”: the application of forces that directly or indirectly induce changes in organizational space.

This presentation explores recent developments in the formalization of the Institutional Grammar for computational methods, and shares learnings from the design and deployment of an ongoing research initiative focused on applying these developments in a real-world context through the experimental implementation of organizational state estimation at an engineering firm. This initiative involves wiring up a feedback system that uses the firm’s internal data infrastructure to continuously sense data about the organization itself, then parses that data via the IG perspective, and makes it visible to members of the firm – who will then be asked to reflect on its accuracy and usefulness. If it is found to be useful, research will shift to questions of dynamics: What is a sensible time scale for such self-observation and modeling? Shifting from a comparative-static view to a view that captures processes of change, and potentially exposes subtle trends not overtly observable? Can a continuous view of an organization’s “velocity” encourage or enable better “steering” and “navigation” – governance of a real world organization?

Understanding the processes that lead to the emergence and continued adaptation of institutions remains the raison d’être for the field of institutional analysis. While extant theory provides explanatory accounts for such processes, their postulates build on sampled empirical cases, commonly with emphasis on comparative-static information as well as context-specific arrangements and outcomes (e.g., Ostrom) or primarily theoretical engagement (e.g., Aoki).
One alternative approach to develop and test theories of collective action is to rely on computational modeling, which is carried by promises of full control of scenario specification as well as comprehensive data collection of antecedents, processes and outcomes. Previous work approached this challenge by variably focusing on norm emergence and inference (e.g., Axelrod (1986), Frantz et al. (2015)), the experimental parameterization of rules in commons scenarios (e.g., Smajgl et al. (2008), Ghorbani and Bravo (2016)), or explored it in support of real-world experimentation (e.g., Janssen and Ostrom, 2006).
Addressing the gap of capturing the institutionalization process comprehensively, we develop a cognitively plausible, yet transparent, architectural model of institutional agents that are able to capture the entire strata of institutions, as well as exhibit the associated capabilities to drive associated formation and transition processes. This includes the exploration, inference and adaption of behavioral strategies, the inference and socialization of behavioral norms as well as endogenous introduction and adaptation to novel rules.
To this end, the institutional agent model recognizes principles of experiential and observational learning, memorization, differentiated institutional reasoning, implicit social cognition, as well as autonomous decision-making.
Drawing on a moderately complex scenario consisting of different actor types and action spaces, we illustrate the principal functioning of this architecture in a social setting. Exploring the effects of varying social choice mechanisms (e.g., voting variants) and configurations, we a) identify conditions that drive variation in process and outcome, and b) compare the varying outcomes caused by different social choice mechanisms (and compositions thereof).
Captured under the umbrella of “Generative Institutional Analysis”, we highlight opportunities for the purpose of developing and testing theory using computational institutional models, as well as outlining opportunities for the use in empirical settings.

Constitutions play a key role in defining the structure, roles, rights, and responsibilities of nation states. Extant research has analyzed how features of constitutional design contribute to the materialization of democratic principles, including the separation of powers (V. Ostrom, 2008; Brinks, Levitsky & Murillo, 2019; Levitsky & Ziblatt, 2019). Recent research applying the Institutional Grammar (IG) highlights the utility of the IG in deciphering how legal concepts are embedded within constitutions (DeMatee, 2023). Nevertheless, systematic use of the IG to characterize how specific characteristics of democracy materialize in constitutional architecture is yet to be developed.
This paper intends to make a twofold contribution to the literature. First, we use the IG 2.0 to understand how features of democracy materialize in Constitutional texts, using the theory of separation of powers as a basis for our analysis. Second, we aim to test, refine, and expand the current characterizations of the constitutive functions proposed by Frantz and Siddiki (2022). As such, we aim to answer a) what types of entities are being constituted in the constitutions of nation states, b) what types of constitutive functions are assigned to said entities, in relation to the principle of separation of powers, and c) whether, and if so how, do these entities and corresponding functions capture the theoretical features of the separation of powers.
To do this, we use a comparative case study approach. We purposefully selected constitutions from six countries, two in North America, two in South America, and two in Europe, to consider different legal traditions (common law and civil law) and regimes to conduct the study. Once identified, we focus on a particular subject of the Constitution, the provisions related to the separation of powers, and code each institutional statement using the IG. We later look for patterns and differences in the way each Constitution captures the idea of separation of powers.