Boundedly Rational Agents Research Articles

Boundedly Rational Agents in Sequential Posted Pricing

Boundedly Rational Agents in Sequential Posted Pricing

Bringing markets back into economics: On economy wide self - coordination by boundedly rational market agents

Markets, I argue, matter significantly for macroeconomic development. The reason for this strange statement is that policy practice in industrialized market economies is routinely governed by models that ignore the role of markets. Policy makers are therefore at risk of misunderstanding the economic situation and of mismanaging the economy.Some macroeconomic models with sufficient nonlinearities and a production lag, two pioneering articles by Day (1982,1983) demonstrated, exhibit phases of erratic fluctuations (or deterministic chaos). Shortly thereafter Eliasson (1983, 1984) observed the endogenous emergence of similar unpredictable irregularities in simulating a complex, agent based, market self-coordinated economy wide model implemented on Swedish data. The irregularities originated in a failure of markets to coordinate production, employment, and investment decisions as they were pushed excessively by policy to become more efficient. In this agent based model of an Experimentally Organized Economy, called MOSES, progress is governed by a dominant flow of successful business experiments conducted by firms, trailed by a flow of mistaken projects. The coordination of these flows by autonomous market agents was shown to be easily disturbed by external interferences, for instance unwitting policies. Markets, simulation experiments however also showed have great endogenous capacities to restore disturbed order. Large private investment mistakes at the agent level, experiments furthermore indicate, incur only minor social costs if production on the site is promptly terminated, and markets organized to efficiently reallocate resources. Bold private investments should therefore be encouraged, corrections of identified mistakes made without delay, and the efficiency of markets to reallocate resources paid attention to. I conclude that central policy in “Sweden like industrial economies” can only effectively be pursued by taking advantage of the endogenous capacity of functioning markets of effective self-coordination. Coordination failure being created by policy failure masquerading as market failure should be considered before market failure is blamed.

Learning to Play General-Sum Games against Multiple Boundedly Rational Agents

Learning to Play General-Sum Games against Multiple Boundedly Rational Agents

Estimation of heuristic switching in behavioral macroeconomic models

This paper addresses the issue of empirical validation of macroeconomic models with behavioral heuristics and a nonlinear switching mechanism. Heuristic switching is an important feature of modeling strategy since it uses simple decision rules of boundedly rational heterogeneous agents. The simulation study shows that the proposed simulated maximum likelihood method well identifies behavioral effects that remain hidden under standard econometric approaches. In the empirical application, we estimate the structural and behavioral parameters of the US economy. We are specifically able to reliably identify the intensity of choice that governs the models’ nonlinear dynamics. Our empirical results thus lay the foundation for studying monetary and fiscal policy in a behavioral macroeconomic framework.

The impacts of interest rates on banks’ loan portfolio risk-taking

Using an agent-based model, we investigate how interest rates affects banks’ risk-taking in terms of the profile of their lending to real sector firms. Our agent-based model considers five types of agents: banks, depositors, the Central Bank, firms, and the clearinghouse. Banks are bounded-rational agents with adaptive strategies. In different setups, depositors are either noisy agents or bounded-rational agents that withdraw their deposits when they have concerns over their banks’ solvency. The other players’ behaviors are used as a reference to understand how these main agents respond strategically to different incentives and situations. Some of our findings recover stylized facts available in the literature: (1) when interest rates decrease, there is an increase of real sector loans, particularly for riskier clients; (2) the interbank market plays a fundamental role in banks’ liquidity management; (3) banks avoid borrowing resources from the Central Bank; (4) when the interest rates increase, banks increase the level of capital buffers and the Capital Adequacy Ratio (CAR). We also present new insights regarding the relationship between interest rates and bank risk-taking, opening an avenue to investigate the banks’ learning process dynamics. Finally, working with different parameter sets, we find a rich set of banks’ behaviors in different interest rate regimes.

The Effort of Reasoning: Modelling the Inference Steps of Boundedly Rational Agents

In this paper we design a new logical system to explicitly model the different deductive reasoning steps of a boundedly rational agent. We present an adequate system in line with experimental findings about an agent’s reasoning limitations and the cognitive effort that is involved. Inspired by Dynamic Epistemic Logic, we work with dynamic operators denoting explicit applications of inference rules in our logical language. Our models are supplemented by (a) impossible worlds (not closed under logical consequence), suitably structured according to the effect of inference rules, and (b) quantitative components capturing the agent’s cognitive capacity and the cognitive costs of rules with respect to certain resources (e.g. memory, time). These ingredients allow us to avoid problematic logical closure principles, while at the same time deductive reasoning is reflected in our dynamic truth clauses. We further show that our models can be reduced to awareness-like plausibility structures that validate the same formulas and we give a sound and complete axiomatization with respect to them. This approach to the agent’s internal deductive reasoning is finally combined with actions of external information.

Macroeconomic dynamics under bounded rationality: on the impact of consumers’ forecast heuristics

This paper considers various types of forecast heuristics to examine the effects of boundedly rational agents on macroeconomic dynamics. Given the baseline New Keynesian model, we seek to find the expectation formation process that is most suitable in describing economic adjustments over the business cycle. In particular, impulse response analysis is used to compare the performances of the macroeconomic model under bounded rationality and under rational expectations. The results show that the fluctuations in consumer confidence mainly explain the degree of persistence in consumption. We conclude that a model under bounded rationality with a heuristic-induced switching process can qualitatively provide a good fit to the data that is equivalent to a model under rational expectations.

Naive Learning Through Probability Overmatching

Naïve Learning in a Binary Action, Social Network Environment In “Naïve Learning Through Probability Overmatching,” I. Arieli, Y. Babichenko, and M. Mueller-Frank consider an environment where privately informed agents select a binary action repeatedly observing the past actions of their neighbors in a social network. Rational inference has been shown to be exceedingly complex in this environment. Instead, this paper focuses on boundedly rational agents that form beliefs according to discretized DeGroot updating and apply a decision rule that assigns a (mixed) action to each belief. It is shown that naïve learning, where the long run actions of all agents are optimal given their pooled private information, can be achieved in any strongly connected network if beliefs satisfy a high level of inertia and the decision rule coincides with probability overmatching. The main difference to existing naïve learning results is that here it is shown to hold (1) for binary rather than uncountable action spaces and (2) even for network and information structures where Bayesian agents fail to learn.

Monetary Policy with a State-Dependent Inflation Target in a Behavioral Two-Country Monetary Union Model

In this paper we study the implementation of a state-dependent inflation target in a two-country monetary union model characterized by boundedly rational agents. In particular, we use the spread between the actual policy rate (which is constrained by the zero-lower-bound) and the Taylor rate (which can become negative) as a measure for the degree of ineffectiveness of conventional monetary policy as a stabilizing mechanism. The perception of macroeconomic risk by the agents is assumed to vary according to this measure by means of the Brock-Hommes switching mechanism. Our numerical simulations indicate a) that a state-dependent inflation target may lead to a better macroeconomic and inflation stabilization, and b) may even lead to an enlarged fiscal space, i.e. a lower debt-to-GDP ratio if the risk premium’s reaction to a higher inflation target is not too large.

The Impacts of Monetary Policy on Banks’ Loan Portfolio Risk-Taking

Using an agent-based model, we investigate how monetary policy affects banks' risk-taking in terms of the profile of their lending to real sector firms.Our agent-based model considers five types of agents: banks, depositors, the Central Bank, firms, and the clearinghouse. While banks and depositors are bounded-rational agents with adaptive strategies, the other players' behaviors are used as a reference to understand how these main agents respond strategically to different incentives and situations. Some of our findings recover stylized facts available in the literature: (1) when the monetary policy eases, there is an increase of real sector loans, particularly for riskier clients; (2) the interbank market plays a fundamental role in banks' liquidity management; (3) banks avoid borrowing resources from the Central Bank; (4) when the monetary policy is restrictive, banks increase the level of capital buffers and the Capital Adequacy Ratio (CAR). We also present new insights regarding the relationship between monetary policy stances and bank risk-taking, opening an avenue to investigate the banks' learning process dynamics. Finally, we show that banks tend to grow when the monetary policy stance eases.

Electric demand response and bounded rationality: mean-field control for large populations of heterogeneous bounded-rational agents.

The increased penetration of renewable energy sources into existing power systems induces challenges in supply-demand balancing. Demand-side flexibility is seen as an option to accommodate variability and limited predictability from renewable energy generation. Heat pumps at residential level, if well coordinated, can be one of those flexibility sources. The complexity involved is high though, since their coordinated operation combines control, population effects and the fact agents may actually not behave as rational decision-makers. We describe here a coordinated control framework that accounts for those aspects altogether. Decentralized model predictive control for large populations of heterogeneous agents is employed. As the cost to be minimized is affected by the population behaviour as a whole through the electricity price, the decentralized control is re-thought as a mean-field game. Existence and uniqueness of a Nash equilibrium are discussed while the Picard-Banach algorithm is used as a solution approach. It is extended to the case of bounded-rational agents. The impact on system dynamics of modelling agents as bounded rational is illustrated through numerical simulations. This article is part of the theme issue 'The mathematics of energy systems'.

Markov-game modeling of cyclist-pedestrian interactions in shared spaces: A multi-agent adversarial inverse reinforcement learning approach

Understanding and modeling road user dynamics and their microscopic interaction behaviour at shared space facilities are curial for several applications including safety and performance evaluations. Despite the multi-agent nature of road user interactions, the majority of previous studies modeled their interactions as a single-agent modeling framework, i.e., considering the other interaction agents as part of the passive environment. However, this assumption is unrealistic and could limit the model's accuracy and transferability in non-stationary road user environments. This study proposes a novel Multi-Agent Adversarial Inverse Reinforcement Learning approach (MA-AIRL) to model and simulate road user interactions at shared space facilities. Unlike the traditional game-theoretic framework that models multi-agent systems as a single time-step payoff, the proposed approach is based on Markov Games (MG) which models road users' sequential decisions concurrently. Moreover, the proposed model can handle bounded rationality agents, e.g., limited information access, through the implementation of the Logistic Stochastic Best Response Equilibrium (LSBRE) solution concept. The proposed algorithm recovers road users' multi-agent reward functions using adversarial deep neural network discriminators and estimates their optimal policies using Multi-agent Actor-Critic with Kronecker factors (MACK) deep reinforcement learning. Data from three shared space locations in Vancouver, BC and New York City, New York are used in this study. The model's performance is compared to a baseline single-agent Gaussian Process Inverse Reinforcement Learning (GPIRL). The results show that the multi-agent modeling framework led to a significantly more accurate prediction of road users’ behaviour and their evasive action mechanisms. Moreover, the recovered reward functions based on the single-agent modeling approach failed to capture the equilibrium solution concept similar to the multi-agent approach.

Learning when to say no

We consider boundedly-rational agents in McCall's model of intertemporal job search. Agents update over time their perception of the value of waiting for an additional job offer using value-function learning. A first-principles argument applied to a stationary environment demonstrates asymptotic convergence to fully optimal decision-making. In environments with actual or possible structural change our agents are assumed to discount past data. Using simulations, we consider a change in unemployment benefits, and study the effect of the associated learning dynamics on unemployment and its duration. Separately, in a calibrated exercise we show the potential of our model of bounded rationality to resolve a frictional wage dispersion puzzle.

Evolutionary economics under nonconvexity and externalities

This article examines the evolutionary behavior of boundedly rational agents under nonconvexity and externalities. While allowing for technical and allocative inefficiency, we evaluate how producer and consumer behavior can use evolutionary rules that eventually lead to Pareto efficient allocations. We study how markets can support the convergence to efficiency, stressing that externalities and nonconvexity can require nonlinear pricing. In an evolutionary economy, we show how pricing and decentralized decisions can support moves toward Pareto efficiency under bounded rationality, nonconvexity and externalities.

Behavioral and Experimental Macroeconomics and Policy Analysis: A Complex Systems Approach

This survey discusses behavioral and experimental macroeconomics, emphasizing a complex systems perspective. The economy consists of boundedly rational heterogeneous agents who do not fully understand their complex environment and use simple decision heuristics. Central to our survey is the question of under which conditions a complex macro-system of interacting agents may or may not coordinate on the rational equilibrium outcome. A general finding is that under positive expectations feedback (strategic complementarity)—where optimistic (pessimistic) expectations can cause a boom (bust)—coordination failures are quite common. The economy is then rather unstable, and persistent aggregate fluctuations arise strongly amplified by coordination on trend-following behavior leading to (almost-)self-fulfilling equilibria. Heterogeneous expectations and heuristics switching models match this observed micro and macro behavior surprisingly well. We also discuss policy implications of this coordination failure on the perfectly rational aggregate outcome and how policy can help to manage the self-organization process of a complex economic system. (JEL C63, C90, D91, E12, E71, G12)

Estimation of Heuristic Switching in Behavioral Macroeconomic Models

This paper offers a simulation-based method for the estimation of heuristic switching in nonlinear macroeconomic models. Heuristic switching is an important feature of modeling strategy since it uses simple decision rules of boundedly rational heterogeneous agents. The simulation study shows that the proposed simulated maximum likelihood method identifies the behavioral effects that stay hidden for standard econometric approaches. In the empirical application, we estimate the structural and behavioral parameters of the US economy. We are especially able to reliably identify the intensity of choice that governs the models’ nonlinear dynamics.

Bounded rationality and heterogeneous expectations: Euler versus anticipated-utility approach

By using Bayesian techniques, our paper investigates behavioral New-Keynesian DSGE models derived under two parsimonious alternatives to introduce heterogeneous expectations: the Euler equation and the anticipated-utility approach. First, we explore the relation between the expectation formation processes and the model determinacy for a broad range of parameterizations by using global sensitivity analysis and Monte Carlo filtering. Second, we perform model comparison to assess how much the two alternatives are consistent with macro and expectation survey data. Our main results are twofold: (1) model determinacy is strongly undermined by the presence of boundedly rational agents; (2) a behavioral model based on Euler equation approach fits the data decisively better than one based on anticipated utility.

Irreversibility of change in a model of boundedly rational agents exchanging goods and capital

This paper investigates the direction of change in a model of agents exchanging goods and capital. Agents have heterogeneous utility functions and are not perfectly rational but can refuse exchanges that decrease their own utility. In this model, an exchange is shown to always increase a quantity defined by the amount of production and the time preference of the agents.

Strategy-Proof Mechanism Design with Boundedly Rational Agents

This paper considers an implementation problem with bounded rationality of the agents. Bounded rationality presented here means that the agent might choose the agent's best response which is different from the agent's dominant strategy. To describe such a behavior, this paper introduces a new notion of equilibrium, called (n-k)-dominant strategy Nash equilibrium, in which at most k∈{0,1,⋯,n} boundedly rational agents might choose their best responses which are different from their dominant strategies and at least (n-k) rational agents choose their dominant strategies. In addition, to show what a socially optimal outcome collectively chosen under the existence of boundedly rational agents, this paper introduces a new notion of implementation, called k-secure implementation, which is double implementation in dominant strategy equilibria and (n-k)-dominant strategy Nash equilibria. In specific environments with k≤(n+1)/2, this paper shows that majority rule and the median voter rule satisfy k-secure implementability, but not secure implementability (Saijo, T., T. Sjostrom, and T. Yamato (2007) “Secure Implementation,” Theoretical Economics 2, pp.203-229) which is equivalent to n-secure implementability respectively.

Coalition formation with dynamically changing externalities

We consider multiple self-interested bounded-rational agents each of which has a goal it needs to achieve. Goals are achievable by executing a set of interdependent tasks. Some tasks exhibit time dependencies and may require sequential execution. For each agent, there may be several alternative sets of tasks that can achieve the goal. Execution of alternatives, may be more beneficial when done by a group of agents and not by a single agent. To jointly achieve goals, agents may form interdependent coalitions. Such coalition formation is computationally intractable. We nevertheless seek a practical solution that is not necessarily optimal yet acceptable by the agents. A solution where agents examine only coalitions in which they are members is inapplicable, as externalities are a major factor given task interdependencies. In this paper we study this coalition formation problem. We describe the problem and introduce a novel Multi-lateral Negotiation Protocol (MNP) that solves it by forming interdependent coalitions. We allow agents to heuristically make gradual concessions, revise their proposals and converge on specific alternatives, and nevertheless increase their expected gains.