Spencer Pantoja

I am a game theorist working on questions in the union and intersection of political economy and organizational theory. My primary research agenda is on political organizations function internally; connecting and contrasting the theory of the firm to political parties, bureaucracies, dictatorships and other non-firm organizations. 

I can be contacted at spantoja at stanford dot edu. [CV]

Working Papers

An Formal Theory of Parties in Legislatures

I model parties in legislatures as information sharing mechanisms. A party forms when a group of legislators share their private information on their own policy preferences with a designated member—the party “whip”. The whip uses this information to craft policies that benefit the party as a whole and that, over time, make all members better off. In my theory, the party is essential—the same outcomes are not attainable if legislators share their information directly or if all information is made public. Information sharing has its downsides, however, as the whip may exploit the information shared with her. The ability to depose a party whip constrains her incentives, enabling legislators to trust the whip, and for the party to work. This party logic leads to new insights. Counterintuitively, the party is weakest, and often fails, when members are ideologically aligned as the threat of dissolution is then not sufficient to keep everyone well-behaved.

Presented at: 2023 Asian School in Economic Theory; 9th Workshop on Relational Contracts; 2024 Society for the Advancement of Economic Theory (SAET) Conference; 2024 Midwest Political Science Association Conference (Session Organizer); 2024 Society for Institutional & Organizational Economics; GAMES 2024, the 7th World Congress of the Game Theory Society (Session Organizer); WUSTL Conference for Early Career Scholars in Theoretical Political Economy; University of Chicago Harris Political Economy Lunch; UC Berkeley Research Workshop in American Politics. 

Works in Progress

Relational Contracts with Contestable Control Rights: Theory and an Application to Autocracies  with Scott Gehlbach and Zhaotian Luo
Presented at: University of Chicago Harris Political Economy Lunch; Utah Organizational & Political Economics Winter Conference; Aspen Political Economics Conference; 2025 Western Political Science Association Conference; 2025 Society for Institutional & Organizational Economics;  2025 American Political Science Association Conference; Stanford Political Economy Seminar.

We examine a dynamic agency model with matching in which agents can attempt to seize power from the principal. In this environment, a stable regime is a relational contract in which the ruler compensates the agent, the agent exerts effort on behalf of the ruler, and the agent does not attempt to seize power. We identify three principles governing such regimes. Under the first principle, exemplified by Uganda under Idi Amin, high resource rents encourage the dictator to rule essentially alone, sharing power with no close individual. The second principle, which we associate with Stalin’s henchman Vyacheslav Molotov, states that rulers who do recruit agents select those who are neither too incompetent nor too competent. Finally, the third principle, which we suggest characterizes Ataturk’s relationship with Turkish statesman Ismet Inonu, says that more competent autocrats are willing to recruit more competent agents—a form of assortative matching.

A Model of Competitive Relationship Building---Conflictual Enforcement in Relational Contracts with Álvaro Delgado-Vega

We study the problem of two great powers in an IR setting who are in zero-sum conflict with each other, both trying to develop relational contracts amongst a shared set of smaller countries. Our methodological innovation is allowing both great powers to influence equilibrium selection—with one principal, we can conceive of the principal’s best equilibrium as if they had the ability to choose and announce which equilibrium would be played. In our setup, each great power announces their own strategy sequentially, essentially dictating the terms of their relational contracts with the smaller countries, but unable to set the terms of the relational contracts of the other great power. Thus, the first-moving great power has to set the terms of their relationships knowing the second power can come in and either try to wreck their strongest relational contracts (i.e. alliances) or target any smaller country the first power ignored.

 We find that when the power of the great powers is imbalanced, there are a continuum of equilibria that the great powers are indifferent to, but are Pareto-rankable for the smaller powers. This happens because of the competition in relationship building—if only one player was a relationship builder in this particular setup, we would never reach a Pareto inefficient outcome. We have upfront transferrable utility and the great powers effectively have commitment power, so the great power would be better off offering the more efficient equilibrium for a cut of the surplus gains (if they were alone). But with two powers, each one sabotages each other’s attempt to extract these surplus gains, thus surplus is left on the table. This indifference to Pareto-rankable equilibria suggests small tie-breaking whims amongst the great powers could lead to large welfare differences amongst the smaller countries.

Dynamic Public Opinion with Steven Callander

We investigate the decision of a politician deciding not only what platform to announce, but when to announce it in the face of evolving public opinion. In a dynamic setting, a politician decides each period if they will announce a stance on a particular issue, and if so, what there stance is. With some probability, each period, the politician privately observes which stance is the “correct stance” (where “correct” can be objective, like whether vaccines are effective, or subjective, representing the opinion society will eventually converge on, like gay marriage). Each period a public signal imperfectly informative on this true stance is observed. The game ends in a finite time horizon, and the politician wants to maximize the final perception that they are informed. If they announce the final popular position before it becomes popular (like being against the Iraq war in 2003), that shows strong evidence that they are informed, but it comes with the biggest risk of being wrong. 

Our main result is that when it becomes hard to determine if the politician was right (the signal is inherently noisey), politicians pool on announcing in the first period. Intuitively, if the politician cannot be proven right or wrong based on the public signal, there are strong incentives to signal that they are informed by announcing early. Waiting reveals the politician has yet to become informed, which is the downside to waiting. This can be overcome by the possibility of becoming informed, but the benefits of becoming truely informed are too muted if the signal is too weak, and thus the downsides of waiting outweigh the upsides. This is the worst outcome for both the politician and the voters—by announcing immediately, the politician has the least amount of opportunity to actually become informed before committing to a platform. 

How is a Dictatorship (not) like a Firm? with Scott Gehlbach and Zhaotian Luo 

Presented at: 2025 Society for Institutional & Organizational Economics 

In this perspectives style article, we investigate the extent to which formal models in organizational economics (ostensibly with firms in mind) could be applied to dictatorships, what alterations would need to be made in this new context of dictatorship, and how the alterations might also teach us something about firms. We focus on three features much more salient in dictatorships than in firms: dependence of incentives on survival of principal, availability of any formal contracts, and the ability of agent to replace principal. 

Goal Complexity and Organizational Conflict with Yunus Aybas

“In actual organizational practice, no one attempts to find an optimal solution for the whole problem. Instead, various groups of decisions are made by specialized units of the organization. . .We do not have to postulate conflict in personal motivations in order to explain conflicts [across units in an organization]. They could, and would, equally well arise if each of the organizational decision-making roles were being enacted by digital computers.” —Herbert Simon (1964) 

The above Herbert Simon quote inspired this project. Every firm, whether they make ball bearings, bottle caps, or corrugated cardboard, is filled with organizational conflict, often across departmental lines. Clearly, the employees did not have these strong opinions about cardboard before joining the firm—thus, something about occupying their specialized organizational roles leads to the conflict in perspectives. How should the design of specialized subunits/departments take into account the endogenously generated conflict this creates?

 Economics has models of organizational design in which we allocate knowledge and specializations (Marschak and Radner, 1972; Van Zandt, 1999; Garicano, 2000). However, in all of these models (collectively called the team theory literature), all players have the same preferences— that is, there is no conflict. Economics also has models which capture the problem of resolving conflict between departments (Alonso, Dessein and Matouschek, 2008; Rantakari, 2008), where a department has both specialized knowledge and biased preferences towards that knowledge. But these papers never look back on where these departments came from—shouldn’t the same person designing solutions to resolve this conflict anticipated the conflict they were creating by making many specialized departments? 

In this paper, we model the tradeoff inherent in organizational design of adding another perspective into the organization—more aggregate knowledge within the organization, but more disagreement on how any given decision should be made that somehow must be resolved. An organization faces a stream of “problems”, which are subsets of a space of “problem attributes”. For each problem, they must make a decision, whose performance depends on the set of attributes pertaining to the problem. A department is a player who specializes in a subset of problem attributes—the smaller the subset, the more accurately the player understands how to address that subset of attributes, but we assume they only care about how solutions perform with respect to their department’s subset of the attribute space. Thus, generalists are more aligned with the organizational objective (although less knowledgable about any given detail).

 We characterize the optimal allocation of expertise and management of conflict. We find conditions under which we should find an interior number of departments, in spite of the fact that we have no labor costs in adding a department and more departments always means more aggregate knowledge. In doing this exercise, we explain why organizations of different types have different internal structures and conflict management styles.