Repeated games are a fundamental concept in game theory, where players interact in a sequence of similar games rather than just once. This chapter introduces the key ideas, concepts, and historical background of repeated games.
Repeated games involve players engaging in a series of strategic interactions. Unlike one-shot games, where players make decisions without considering future encounters, repeated games allow for the accumulation of experience, the development of reputations, and the possibility of long-term cooperation. This dynamic is crucial in understanding various real-world scenarios, such as business negotiations, diplomatic relations, and evolutionary biology.
The importance of repeated games lies in their ability to model situations where players' actions have lasting consequences. This can lead to different outcomes compared to one-shot games, as players may choose strategies that maximize their long-term gains rather than immediate payoffs.
To delve into repeated games, it is essential to grasp several key concepts and terms:
These concepts provide the building blocks for analyzing and solving repeated games.
The study of repeated games has evolved significantly over the years, driven by advancements in game theory and its applications. Early contributions include the work of Robert J. Aumann and Sergiu Hart, who developed the concept of correlated equilibria and the folk theorem, respectively. These foundational works laid the groundwork for understanding the dynamics of repeated interactions.
Later developments, such as the contributions of John Harsanyi, Reinhard Selten, and Robert D. Lucas, further enriched the theory by introducing concepts like trigger strategies, evolutionary stability, and the role of information in repeated games. These developments have not only deepened our theoretical understanding but also expanded the applicability of repeated games to various fields, including economics, biology, and political science.
The evolution of repeated games reflects the dynamic nature of game theory itself, continually adapting to new challenges and insights.
This chapter delves into the fundamental concepts of static games and their transition to repeated interactions. Understanding the differences and similarities between these two types of games is crucial for grasping the nuances of strategic behavior in various economic and social contexts.
Static games are one-shot interactions where players make decisions simultaneously, without the influence of future interactions. These games are characterized by their simplicity and the lack of temporal dynamics. Key concepts include:
Examples of static games include the Prisoner's Dilemma, the Battle of the Sexes, and the Stag Hunt. These games serve as building blocks for understanding more complex interactions.
Repeated games differ from one-shot games in that they involve multiple interactions over time. This repetition allows for strategic considerations based on future payoffs and the potential for building reputations. Key differences include:
In contrast, one-shot games lack these temporal dynamics and the associated strategic complexities. The focus is solely on the immediate payoffs.
Repeated games introduce several strategic considerations that are not present in one-shot games. These include:
Understanding these strategic considerations is essential for analyzing and predicting behavior in repeated interactions.
In the next chapter, we will explore finite repeated games in more detail, focusing on their structure, properties, and solution concepts.
Finite repeated games are a fundamental concept in game theory, where players interact over a fixed number of stages. Unlike infinite repeated games, the number of interactions is predetermined, allowing for the use of backward induction to analyze strategies and outcomes. This chapter delves into the structure, properties, and solutions of finite repeated games.
The structure of a finite repeated game is characterized by a finite number of stages, denoted by \( T \). In each stage \( t \), players choose actions \( a_t \) from a set of available actions \( A \). The payoff in each stage is determined by a stage game payoff function \( u_t(a_t) \). The overall payoff for the game is often a sum or discounted sum of the stage payoffs.
Key properties of finite repeated games include:
Solving finite repeated games involves determining the optimal strategies for players given their knowledge of the game's structure and the number of stages remaining. The backward induction method is a common technique used to solve these games.
Backward induction works by solving the game in reverse order, starting from the last stage and working backwards to the first stage. At each stage, players consider the optimal action given the subsequent stages' outcomes. This process ensures that players make decisions that are consistent with their objectives over the entire duration of the game.
Backward induction is closely related to the concept of subgame perfection. A strategy profile is subgame perfect if it is optimal for all players in every subgame of the original game. In finite repeated games, subgame perfection ensures that players' strategies remain optimal as the game progresses.
To illustrate, consider a simple finite repeated game with two players, Alice and Bob, and two stages. In the second stage, the payoff matrix is:
Alice's payoffs: \( \begin{bmatrix} 3, 1 \\ 0, 2 \end{bmatrix} \)
Bob's payoffs: \( \begin{bmatrix} 1, 0 \\ 2, 3 \end{bmatrix} \)
Using backward induction, we can determine the subgame perfect equilibrium. In the second stage, both players would choose the action that maximizes their payoffs, given the first stage's outcome. This process is repeated for each stage, ensuring that the strategies are consistent and optimal throughout the game.
In summary, finite repeated games offer a structured framework for analyzing strategic interactions over a fixed number of stages. By employing backward induction and ensuring subgame perfection, players can determine optimal strategies that maximize their payoffs over the entire duration of the game.
Infinite repeated games are a fundamental concept in game theory, extending the analysis of static games to scenarios where interactions occur indefinitely. This chapter delves into the structure, properties, and solutions of infinite repeated games, distinguishing between discounted and undiscounted games.
Infinite repeated games are characterized by an infinite sequence of stages, where players make decisions at each stage based on the history of play. The key properties include:
The structure of an infinite repeated game can be represented as a sequence of stages, each involving a static game. The payoffs at each stage are typically discounted to reflect the time value of money or the diminishing importance of future payoffs.
Solving infinite repeated games involves finding strategies that are optimal for all players in the long run. The key concepts include:
Grinding strategies are essential for ensuring that a player can sustain a minimal level of payoff, while trigger strategies provide a mechanism for enforcing cooperation. Folk theorems, which will be discussed in detail in Chapter 6, provide a framework for understanding the set of achievable payoffs in infinite repeated games.
Infinite repeated games can be categorized into discounted and undiscounted games based on the treatment of future payoffs:
Discounted games are often used to model situations where the time value of money is significant, such as in financial markets. Undiscounted games, on the other hand, are more appropriate for scenarios where the focus is on long-term average performance, like in labor contracts or repeated interactions in social settings.
In both discounted and undiscounted games, the key challenge is to find strategies that balance immediate gains with long-term sustainability. This requires a deep understanding of the game's structure and the players' preferences.
Trigger strategies are a fundamental concept in the study of repeated games, particularly in the context of infinite repeated games. They provide a mechanism for players to condition their future actions on the history of play, allowing for more complex and realistic strategic interactions.
A trigger strategy is defined by a pair of strategies: a trigger condition and a response. The trigger condition specifies a deviation from a prescribed path of play, while the response outlines the actions to be taken once the trigger condition is met. The simplest form of a trigger strategy is a punishment threat, where a player threatens to deviate from a cooperative path if the other player does not cooperate.
For example, consider a simple prisoner's dilemma game repeated infinitely. A player might use a trigger strategy where they cooperate as long as the other player cooperates. If at any point the other player defects, the first player will defect in all subsequent rounds as a form of punishment.
In finite repeated games, trigger strategies can be analyzed using backward induction. The player who moves last in the game has a strategy that specifies their action based on the history of play. This strategy can be seen as a trigger strategy where the trigger condition is the deviation from the prescribed path, and the response is the action taken in the last round.
In infinite repeated games, trigger strategies become more powerful. Players can use trigger strategies to enforce cooperation over an infinite horizon. For instance, in the repeated prisoner's dilemma, a player might use a grim trigger strategy, where they cooperate until the other player defects and then defect forever. This strategy ensures that cooperation is sustained as long as the other player continues to cooperate.
Trigger strategies have wide applications in various fields, including economics, political science, and biology. They are used to model situations where cooperation can be sustained through the threat of punishment. For example, in international relations, trigger strategies can be used to model the behavior of countries that threaten military action if another country violates a treaty.
However, trigger strategies also have limitations. They often rely on the assumption of perfect monitoring and enforcement, which may not hold in real-world situations. Additionally, the effectiveness of trigger strategies can be sensitive to the specific parameters of the game, such as the discount factor in infinite games.
In summary, trigger strategies are a powerful tool in the analysis of repeated games. They allow for the modeling of complex strategic interactions and have broad applications in various fields. However, their effectiveness is subject to certain assumptions and limitations.
Folk theorems are a fundamental concept in the study of repeated games, providing insights into the strategic behavior of players in situations where the game is played multiple times. These theorems are so-called because they are widely known and accepted among game theorists, despite not being formally proven in the traditional sense. They offer powerful predictions about the outcomes of repeated games under certain conditions.
Folk theorems typically state that in a repeated game with a finite number of players and actions, if the one-shot game has a Nash equilibrium, then there exists a subgame-perfect Nash equilibrium in the repeated game where players cooperate and follow a strategy that is optimal in the long run. The key idea is that players can commit to a cooperative strategy even if they cannot enforce it in a single stage of the game.
To illustrate, consider a simple example: the Prisoner's Dilemma. In a one-shot game, the Nash equilibrium is for both players to defect. However, in a repeated game, folk theorems suggest that players can reach a subgame-perfect Nash equilibrium where they both cooperate in every period, provided that the discount factor is sufficiently high. This is because players can threaten to defect in future rounds, making cooperation in the present rational.
The proof of folk theorems involves constructing a strategy profile where players commit to a cooperative path. This typically requires a credible threat of punishment for deviations. For instance, in the repeated Prisoner's Dilemma, players can agree to cooperate in all periods and threaten to defect forever if either player deviates. The threat of future defection makes current cooperation rational, even though defection is the dominant strategy in the one-shot game.
Folk theorems have significant implications for game design and the analysis of real-world situations. They suggest that the structure of repeated interactions can lead to cooperation even when individual rationality suggests otherwise. This has applications in various fields, including economics, political science, and biology.
For example, in economics, folk theorems help explain why firms might engage in long-term contracts or why countries might sign treaties. In political science, they provide insights into the stability of cooperation in international relations. In biology, they can be used to model the evolution of cooperative behavior among organisms.
However, folk theorems also highlight the importance of the discount factor. If the discount factor is too low, players may not find it credible to cooperate in the future, leading to defection in the present. This underscores the role of time preferences and the structure of repeated interactions in determining outcomes.
Several variations and extensions of folk theorems have been explored in the literature. One important extension is the folk theorem for infinite repeated games. In this setting, the theorem states that if the one-shot game has a Nash equilibrium, then there exists a subgame-perfect Nash equilibrium in the infinite repeated game where players cooperate in all periods, provided that the discount factor is sufficiently close to 1.
Another variation considers repeated games with incomplete information. In these games, players have private information that affects their payoffs. Folk theorems in this context often rely on the concept of credible commitment, where players can commit to a cooperative strategy despite their private information.
Folk theorems have also been extended to more complex game structures, such as games with multiple equilibria or games with imperfect monitoring. These extensions help broaden the applicability of folk theorems to a wider range of real-world situations.
Repeated games with incomplete information extend the framework of repeated games by introducing uncertainty about the players' types or preferences. This chapter explores how such incomplete information affects strategic interactions and outcomes in repeated games.
Bayesian games provide a framework for analyzing situations where players have different types, and their actions depend on the type they believe they are playing against. In the context of repeated games, signaling becomes crucial as players may use their actions to convey information about their types.
For example, consider a repeated game where one player is a "high type" and the other is a "low type." The high type may signal its type through consistent cooperative behavior, while the low type may behave more selfishly. The other player, observing these actions, updates their beliefs about the opponent's type and adjusts their strategy accordingly.
Private information refers to the situation where each player has private knowledge that the other players do not possess. In repeated games, private information can lead to complex dynamics, as players must decide whether to reveal their private information and how to do so strategically.
One key concept in this context is the "secrecy" problem, where players may prefer to keep their private information private to avoid exploitation. However, revealing information can also build trust and facilitate cooperation. The balance between secrecy and revelation is a critical aspect of repeated games with private information.
Credible commitment is essential in repeated games with incomplete information, as players must commit to certain strategies that are self-enforcing. A player's reputation can significantly influence its behavior in future interactions, as other players may adjust their expectations based on past actions.
For instance, a player with a good reputation may be able to commit to a cooperative strategy even if it is not in their immediate self-interest. Conversely, a player with a poor reputation may find it difficult to commit to cooperation. The dynamics of reputation and commitment are complex and depend on the specific structure of the repeated game and the players' beliefs.
In summary, repeated games with incomplete information introduce rich and dynamic strategic interactions. The concepts of Bayesian games, private information, credible commitment, and reputation all play crucial roles in understanding how such games unfold and what outcomes can be expected.
Evolutionary games provide a framework for understanding how strategies evolve over time in populations. When combined with the repeated interactions found in repeated games, evolutionary games offer insights into the dynamics of strategy adoption and the emergence of cooperative behavior.
Evolutionary games are a branch of game theory that studies how strategies evolve in a population over time. Unlike traditional game theory, which often assumes rational decision-making, evolutionary games consider the adaptive processes that drive the evolution of strategies. Key concepts in evolutionary games include replicator dynamics, evolutionarily stable strategies (ESS), and the concept of fitness.
In an evolutionary game, each individual in a population chooses a strategy based on its fitness, which is determined by the payoffs received from interacting with other individuals. Over time, strategies that perform better (i.e., yield higher payoffs) become more prevalent in the population.
When evolutionary games are combined with the repeated interactions found in repeated games, the dynamics of strategy evolution become even more complex. In a repeated game setting, individuals interact multiple times, and their strategies can be influenced by the outcomes of previous interactions.
One of the key concepts in this context is the evolution of cooperation. In standard game theory, cooperation can be stable in repeated games through mechanisms such as trigger strategies and folk theorems. However, evolutionary games offer a different perspective, focusing on how cooperation can emerge and persist through the adaptive processes of natural selection.
For example, consider the Prisoner's Dilemma game, which is often used to study cooperation. In an evolutionary setting, cooperation can evolve and persist if it provides a fitness advantage. This can happen through mechanisms such as kin selection, direct reciprocity, or indirect reciprocity (reputation).
Evolutionary games and repeated interactions have wide-ranging applications in economics and biology. In economics, they are used to study the evolution of norms, standards, and conventions. For instance, evolutionary game theory can help explain how different standards of behavior (e.g., driving on the right or left side of the road) emerge and persist in a population.
In biology, evolutionary games are used to model the evolution of behaviors and strategies in animal populations. For example, they can be used to study the evolution of cooperative behaviors in social insects, such as ants and bees, or the evolution of competitive behaviors in predator-prey interactions.
One notable application is the study of the evolution of human cooperation. Evolutionary game theory has been used to explain the emergence of altruistic behaviors in humans, such as helping strangers or cooperating in large-scale social projects. This research highlights the importance of repeated interactions and the adaptive processes of natural selection in shaping human behavior.
Repeated games in experimental economics provide a unique platform to study strategic interactions under controlled conditions. This chapter explores the design, conduct, and implications of experiments involving repeated games.
Designing experiments that capture the essence of repeated games requires careful consideration of several factors. Researchers must choose the structure of the game, including the number of players, the payoff matrix, and the information available to participants. The design should also specify the number of repetitions and any potential termination conditions.
One common approach is to use laboratory experiments where participants interact in a controlled environment. This allows researchers to observe behavior over multiple rounds and analyze the evolution of strategies. Additionally, field experiments can be conducted to study repeated interactions in more natural settings, such as markets or organizations.
In both laboratory and field experiments, it is crucial to ensure that participants understand the rules and incentives of the game. Clear instructions and practice rounds can help mitigate any confusion and ensure that participants are motivated to perform well.
Experimental economics has yielded numerous insights into the behavior of individuals in repeated games. One of the key findings is the emergence of cooperation and trust. Participants often exhibit pro-social behavior, even when it is not in their immediate self-interest. This aligns with theoretical predictions from folk theorems, which suggest that cooperation can be sustained in repeated games.
Another important finding is the role of reputation and commitment. Participants who can establish a good reputation or make credible commitments tend to fare better in repeated interactions. This highlights the importance of long-term considerations in strategic decision-making.
Experiments have also revealed the impact of information on behavior. In games with incomplete information, participants may use signaling and other strategic communication to convey their types or intentions. This aligns with the theory of Bayesian games and signaling.
Despite their value, repeated games in experimental economics face several challenges. One of the main limitations is the potential for experimental bias. Participants may deviate from their rational self-interest due to cognitive limitations or social desirability bias.
Another challenge is the generalizability of experimental findings. Laboratory experiments often involve simple, stylized games that may not capture the complexity of real-world interactions. Field experiments can address this issue but may face logistical and ethical challenges.
Additionally, the design of experiments can influence the outcomes. Different incentives, instructions, and game structures can lead to varying results. Researchers must be cautious in interpreting findings and consider the robustness of their conclusions.
Future research in this area should focus on addressing these challenges and expanding the scope of experimental studies. This includes designing more realistic games, using larger sample sizes, and employing advanced statistical techniques to analyze data.
In concluding this exploration of repeated games, it is clear that the study of strategic interactions over time has deepened our understanding of game theory and its applications. This chapter will summarize the key concepts introduced throughout the book and discuss open problems and future research avenues in the field.
Repeated games, whether finite or infinite, have revealed that the repetition of interactions can lead to outcomes that differ significantly from those predicted by one-shot games. Key concepts include:
Despite the significant advancements, several open problems and research avenues remain:
The study of repeated games has broader implications for game theory and related fields. It highlights the importance of time and history in strategic interactions, challenging the static assumptions of classical game theory. Additionally, repeated games provide a framework for understanding cooperation, commitment, and evolution in various contexts, from economics and biology to political science and sociology.
As we look to the future, the continued exploration of repeated games will undoubtedly lead to new discoveries and applications, further enriching our understanding of strategic behavior and its consequences.
Log in to use the chat feature.