Agent's Beliefs Research Articles

Learning Generates Long Memory

We consider a prototypical representative-agent forward-looking model, and study the low frequency variability of the data when the agent's beliefs about the model are updated through linear learning algorithms. We nd that learning in this context can generate strong persistence. The degree of persistence depends on the weights agents place on past observations when they update their beliefs, and on the magnitude of the feedback from expectations to the endogenous variable. When the learning algorithm is recursive least squares, long memory arises when the coefficient on expectations is sufficiently large. In algorithms with discounting, long memory provides a very good approximation to the low-frequency variability of the data. Hence long memory arises endogenously, due to the self-referential nature of the model, without any persistence in the exogenous shocks. This is distinctly dierent from the case of rational expectations, where the memory of the endogenous variable is determined exogenously. Finally, this property of learning is used to shed light on some well-known empirical puzzles.

Listing Behaviors and Housing Market Outcomes

We examine whether, and how, listing strategies impact sale prices in residential home sales. Literatures in housing economics, negotiations, and auctions offer diverse predictions around this question. On the one hand, housing studies typically treat home prices as an objective function of property and neighborhood characteristics. Yet, the large and robust literature on anchoring effects (Tversky and Kahneman, 1974) suggests a positive relationship between listing prices and sale prices. Finally, evidence from the auctions literature suggests the opposite pattern through herding behaviors. We analyzed more than 14,000 transactions, taking into account observable property heterogeneity, geographical location and timing of the sales. We find that higher starting prices are indeed associated with higher selling prices, consistent with anchoring. For the average home in our sample, over-pricing between 10 and 20% leads to an increase in the sale price of $117-$163. This effect is particularly strong in areas with higher rates of mortgage foreclosure or serious delinquency. Additional analyses show that our results are unlikely to be driven by seller motivations or unobserved home qualities. We contrast our findings with recommendations and private beliefs of real estate agents, who provide services and advice for about 90% of home sales in the US.

Inflation Risk Premia in the Term Structure of Interest Rates: Evidence from Euro Area Inflation Swaps

We estimate inflation risk premia in the Euro area using inflation swaps. By proposing a no-arbitrage model for econometric analysis, and estimating it using Markov Chain Monte Carlo, we find estimates of inflation risk premia, that on average show an upward sloping term structure, with 1 year risk premia of 13 bps and 10 year risk premia of 52 bps, however with fluctuation in risk premia over time. Our estimates suggest that surveys are important in identifying inflation expectations and thus inflation risk premia. We relate estimates of inflation risk premia to agents beliefs, and find that skews in short term inflation perceptions drive short term inflation risk premia, where beliefs on GDP growth drive longer term risk premia.

A game-theoretic model of interactions between agents with different beliefs in the form of linear cognitive maps

AbstractA game-theoretic model of conflicts of interests between agents in semistructured situations is considered. Linear cognitive maps are used as a model of agent beliefs about situations. Cognitive maps of different agents may vary.

Oral health: locus of control, health behavior, self-rated oral health and socio-demographic factors in Istanbul adults

Objectives. To determine oral health control beliefs of Istanbul adults using the Multidimensional Oral Health Locus of Control Scale (MOHLCS) after confirming its factorial validity and to examine the relationships between these beliefs, self-rated oral health, oral health behaviors and socio-demographic factors. Material and methods. The MOHLCS was administered to a sample of 1200 subjects aged ≥18 years in Istanbul chosen using a quota-sampling method (response 88%). The relationship between the MOHLCS and oral health behaviors, self-rated oral health and socio-demographic factors was assessed after confirming the factorial validity of the MOHLCS. Results. The MOHLCS demonstrated satisfactory internal reliability. Factor analysis results showed a new four-factor solution, namely Internal, Dentist, Chance, and Socialization agents. Multivariate analysis showed that female gender, younger age, higher socioeconomic status, more frequent daily toothbrushing, and regular dental check-ups were associated with higher Internal beliefs, while older age, lower educational level, lower socioeconomic status, low toothbrushing frequency, and symptom-orientated dental attendance were associated with higher Chance beliefs. Being unmarried and low toothbrushing frequency were associated with lower Dentist beliefs. Males and older subjects had lower Socialization agents beliefs. Internal, Dentist and Chance beliefs were significantly associated with self-rated oral health. Conclusions. Compared with the original factor structure, the new factor structure had better goodness of fit for this sample. Self-rated oral health, socio-demographic factors, and oral health behaviors were significantly associated with oral health control beliefs. These beliefs may be useful for planning oral health promotion programs and for formulating advice given by oral health professionals about their patients' oral health behaviors.

A Fixed Point Convergence Theorem in Euclidean Spaces and its Application to Non-Bayesian Learning in Social Networks

I characterize the fixed points of continuous and (neighbor-)constricting functions f:ℝ^{v}→ℝ^{v} and show that each recursively defined sequence x^{k 1}=f(x^{k}), k=0,1,2,... converges to a fixed point of f. The results are applied to generalize the existing results on convergence of beliefs of non-Bayesian agents in social networks of DeGroot (1974), DeMarzo, Vayanos and Zwiebel (2003), and Golub and Jackson (2010).

On the reciprocal interaction between believing and feeling: an adaptive agent modelling perspective

An agent’s beliefs usually depend on informational or cognitive factors such as observation or received communication or reasoning, but also affective factors may play a role. In this paper, by adopting neurological theories on the role of emotions and feelings, an agent model is introduced incorporating the interaction between cognitive and affective factors in believing. The model describes how the strength of a belief may not only depend on information obtained, but also on the emotional responses on the belief. For feeling emotions a recursive body loop between preparations for emotional responses and feelings is assumed. The model introduces a second feedback loop for the interaction between feeling and belief. The strength of a belief and of the feeling both result from the converging dynamic pattern modelled by the combination of the two loops. For some specific cases it is described, for example, how for certain personal characteristics an optimistic world view is generated in the agent’s beliefs, or, for other characteristics, a pessimistic world view. Moreover, the paper shows how such affective effects on beliefs can emerge and become stronger over time due to experiences obtained. It is shown how based on Hebbian learning a connection from feeling to belief can develop. As these connections affect the strenghts of future beliefs, in this way an effect of judgment ‘by experience built up in the past’ or ‘by gut feeling’ can be obtained. Some example simulation results and a mathematical analysis of the equilibria are presented.

The dynamics of efficient asset trading with heterogeneous beliefs

This paper analyzes the dynamic properties of portfolios that sustain dynamically complete markets equilibrium when agents have heterogeneous priors. We argue that the conventional wisdom that belief heterogeneity generates continuous trade and significant fluctuations in individual portfolios may be correct but it needs some qualifications. We consider an infinite horizon stochastic endowment economy populated by many Bayesian agents with heterogeneous priors over the stochastic process of the states of nature. Our approach hinges on studying the portfolios that decentralize Pareto optimal allocations. Since these allocations are typically history dependent, we propose a methodology to provide a complete recursive characterization when agents believe that the process of states of nature is i.i.d. but disagree about the probability of the states. We show that even though heterogeneous priors within that class can indeed generate genuine changes in the portfolios of any dynamically complete markets equilibrium, these changes vanish with probability one if the true process consists of i.i.d. draws from a common distribution and the support of some agent's prior belief contains the true distribution. Finally, we provide examples in which asset trading does not vanish because either (i) no agent learns the true conditional probability of the states or (ii) some agent does not know the true process generating the data is i.i.d.

Public communication in justification logic

Justification Logic is a framework for reasoning about evidence and justification in multi-agent systems. Most accounts of Justification Logic are essentially static, in that the (justified) beliefs of agents are immutable. In this article, we add public communication, a dynamic operation of belief change studied in the area of Dynamic Epistemic Logic, to the language of Justification Logic. Introducing notions of bisimulation for the languages of Justification Logic with and without public communication, we catalogue the expressive relationships that exist between almost all of the well-known static fragments of Justification Logic and then determine whether the addition of public communication affects the various expressive relationships existing between these fragments.

A Belief Revision Framework for Revising Epistemic States with Partial Epistemic States

Belief revision performs belief change on an agent's beliefs when new evidence (either of the form of a propositional formula or of the form of a total pre-order on a set of interpretations) is received. Jeffrey's rule is commonly used for revising probabilistic epistemic states when new information is probabilistically uncertain. In this paper, we propose a general epistemic revision framework where new evidence is of the form of a partial epistemic state. Our framework extends Jeffrey's rule with uncertain inputs and covers well-known existing frameworks such as ordinal conditional function (OCF) or possibility theory. We then define a set of postulates that such revision operators shall satisfy and establish representation theorems to characterize those postulates. We show that these postulates reveal common characteristics of various existing revision strategies and are satisfied by OCF conditionalization, Jeffrey's rule of conditioning and possibility conditionalization. Furthermore, when reducing to the belief revision situation, our postulates can induce most of Darwiche and Pearl's postulates.

Shopper: A System for Executing and Simulating Expressive Plans

We present Shopper, a plan execution engine that facilitates experimental evaluation of plans and makes it easier for planning researchers to incorporate replanning. Shopper interprets the LTML plan language, which extends PDDL in two major ways: with more expressive control structures, and with support for semantic web services modeled on OWL-S. LTML's command structures include not only conventional ones such as branching, iteration, and procedure calls, but also features needed to handle HTN plans, such as precondition-filtered method choice. Unlike conventional programming languages, LTML supports interaction with the agent's belief store, so that its execution semantics line up with those assumed by planners. LTML actions extend PDDL actions in having outputs as well as effects, which means that they can support actions that sense the world; an important special case of this is semantic web services, which reveal information about a state hidden from the agent. To support experimentation as well as action in the real world, Shopper accommodates multiple, swappable implementations of its primitive action API. For example, one may interact with real web services through SOAP and WSDL, or with simulated web services through local procedure calls. We describe novel features of LTML, the interpretation strategy, swappable back-ends, and the implementation.

Theory of Argumentation in Financial Markets

This paper aims to explore the relevance of the Theory of Argumentation TA in the complex area of financial reporting. Specifically, we investigated the scope of the phenomenon of persuasion in advertising. It examines advertisements in publications notable economic movement in Colombia. The financial communication is important to distinguish how to run the models of behavior based on beliefs of agents. Consequently, investors' beliefs can also change systematically with changes in market prices. This paper is the first part and its purpose is to prepare from the Theory of Argumentation TA an application to the financial sector in Colombia.

Investigating the Neural and Cognitive Basis of Moral Luck: It's Not What You Do but What You Know.

Moral judgments, we expect, ought not to depend on luck. A person should be blamed only for actions and outcomes that were under the person’s control. Yet often, moral judgments appear to be influenced by luck. A father who leaves his child by the bath, after telling his child to stay put and believing that he will stay put, is judged to be morally blameworthy if the child drowns (an unlucky outcome), but not if his child stays put and doesn’t drown. Previous theories of moral luck suggest that this asymmetry reflects primarily the influence of unlucky outcomes on moral judgments. In the current study, we use behavioral methods and fMRI to test an alternative: these moral judgments largely reflect participants’ judgments of the agent’s beliefs. In “moral luck” scenarios, the unlucky agent also holds a false belief. Here, we show that moral luck depends more on false beliefs than bad outcomes. We also show that participants with false beliefs are judged as having less justified beliefs and are therefore judged as more morally blameworthy. The current study lends support to a rationalist account of moral luck: moral luck asymmetries are driven not by outcome bias primarily, but by mental state assessments we endorse as morally relevant, i.e. whether agents are justified in thinking that they won’t cause harm.

Heterogeneous trading strategies with adaptive fuzzy Actor–Critic reinforcement learning: A behavioral approach

The present study addresses the learning mechanism of boundedly rational agents in the dynamic and noisy environment of financial markets. The main objective is the development of a system that “decodes” the knowledge-acquisition strategy and the decision-making process of technical analysts called “chartists”. It advances the literature on heterogeneous learning in speculative markets by introducing a trading system wherein market environment and agent beliefs are represented by fuzzy inference rules. The resulting functionality leads to the derivation of the parameters of the fuzzy rules by means of adaptive training. In technical terms, it expands the literature that has utilized Actor–Critic reinforcement learning and fuzzy systems in agent-based applications, by presenting an adaptive fuzzy reinforcement learning approach that provides with accurate and prompt identification of market turning points and thus higher predictability. The purpose of this paper is to illustrate this concretely through a comparative investigation against other well-established models. The results indicate that with the inclusion of transaction costs, the profitability of the novel system in case of NASDAQ Composite, FTSE100 and NIKKEI255 indices is consistently superior to that of a Recurrent Neural Network, a Markov-switching model and a Buy and Hold strategy. Overall, the proposed system via the reinforcement learning mechanism, the fuzzy rule-based state space modeling and the adaptive action selection policy, leads to superior predictions upon the direction-of-change of the market.

Non-Bayesian Learning

A series of experiments suggest that, compared to the Bayesian benchmark, people may either underreact or overreact to new information. We consider a setting where agents repeatedly process new data. Our main result shows a basic distinction between the long-run beliefs of agents who underreact to information and agents who overreact to information. Like Bayesian learners, non-Bayesian updaters who underreact to observations eventually forecast accurately. Hence, underreaction may be a transient phenomenon. Non-Bayesian updaters who overreact to observations eventually forecast accurately with positive probability but may also, with positive probability, converge to incorrect forecasts. Hence, overreaction may have long-run consequences.

Customary International Law and the Place of Normative Considerations

A large part of the theory of customary international law subscribes to the idea that we can determine the content of international custom by attending to certain social facts about the behaviour and intentions or beliefs of international agents. I argue that this idea is mistaken insofar as it excludes from the determinants of customary legal content the only considerations that could explain which social facts are significant, by how much, and which propositions of law they support: normative considerations.

Believers' estimates of God's beliefs are more egocentric than estimates of other people's beliefs

People often reason egocentrically about others' beliefs, using their own beliefs as an inductive guide. Correlational, experimental, and neuroimaging evidence suggests that people may be even more egocentric when reasoning about a religious agent's beliefs (e.g., God). In both nationally representative and more local samples, people's own beliefs on important social and ethical issues were consistently correlated more strongly with estimates of God's beliefs than with estimates of other people's beliefs (Studies 1-4). Manipulating people's beliefs similarly influenced estimates of God's beliefs but did not as consistently influence estimates of other people's beliefs (Studies 5 and 6). A final neuroimaging study demonstrated a clear convergence in neural activity when reasoning about one's own beliefs and God's beliefs, but clear divergences when reasoning about another person's beliefs (Study 7). In particular, reasoning about God's beliefs activated areas associated with self-referential thinking more so than did reasoning about another person's beliefs. Believers commonly use inferences about God's beliefs as a moral compass, but that compass appears especially dependent on one's own existing beliefs.

Departures from Rational Expectations and Asset Pricing Anomalies

We investigate the potential of the consumption CAPM with pessimism, doubt, and the availability heuristic in the agent's beliefs to resolve the equity premium and risk-free rate puzzles. Using the nonlinear GMM estimation techniques, we find that doubt and the availability heuristic play an important role in explaining the cross-section of asset returns. However, when taken alone, these deviations from rational expectations cannot resolve the equity premium and risk-free rate puzzles. This result is robust to the assumption that the expected value of an uncertain prospect is nonlinear in the subjective outcome probabilities.

Managing Expectations and Fiscal Policy

Working Paper 2009-29 October 2009 Abstract: This paper studies an optimal fiscal policy problem of Lucas and Stokey (1983) but in a situation in which the representative agent's distrust of the probability model for government expenditures puts model uncertainty premia into history-contingent prices. This situation gives rise to a motive for expectation management that is absent within rational expectations and a novel incentive for the planner to smooth the shadow value of the agent's subjective beliefs to manipulate the equilibrium price of government debt. Unlike the Lucas and Stokey (1983) model, the optimal allocation, tax rate, and debt become history dependent despite complete markets and Markov government expenditures. JEL classification: D80, E62, H21, H63 Key words: Ramsey plan, misspecification, robustness, taxes, debt, martingale, expansion 1 Introduction Optimal policy design problems heavily exploit the rational expectations assumption that attributes a unique and fully trusted probability model to all agents. That useful assumption precludes carrying out a coherent analysis that attributes fears of model misspecification to some or all agents. It seems natural to ask the question: How should we approach policy design problems in macroeconomics when at least some agents distrust the model? This question is not just of academic interest but of particular practical relevance. Lack of confidence in models seems to have become pronounced in the recent financial crisis and has entered policy discussions. Caballero and Krishnamurthy (2008), for example, impute sets of probability models and a max-min criterion to private agents as a way to model Knightian uncertainty when a lender of last resort copes with flights to quality, whereas Uhlig (2009) appeals to uncertainty aversion to justify pessimism during bank runs. Our approach can be viewed as putting a particular structure on a decision maker's set of models and thereby on his pessimism. This additional structure opens up channels of influence for policy makers not present in the analyses of Caballero and Krishnamurthy (2008) and Uhlig (2009). (1) This paper features a notion of expectation management that is absent from the standard rational expectations paradigm. We formulate an optimal policy problem in which private agents' fears of model misspecification cause them to adjust their expectations in ways that a Ramsey planner recognizes and exploits, bringing the household's endogenous beliefs into the forefront of an optimal policy design problem because they affect equilibrium prices. We study a Ramsey fiscal policy problem in which a planner knows that a representative household distrusts a probability model for exogenous sequences of government expenditures, while the planner still trusts it. We start with the complete-markets economy without capital analyzed by Lucas and Stokey (1983), but modify the representative household's preferences to express his concerns about misspecification of the stochastic process for government expenditures. The planner can use a distortionary tax on labor income and issue state-contingent debt in order to finance the exogenous government expenditures. Our household expresses distrust of his model by ranking consumption plans according to the multiplier preferences of Hansen and Sargent (2001); when a multiplier parameter assumes a special value, the expected utility preferences of Lucas and Stokey (1983) emerge as a special case in which the decision maker completely trusts his probability model. (2) The Lucas and Stokey (1983) environment isolates essential dimensions of an optimal macroeconomic policy design problem in which a representative household's ambiguity about its statistical model creates an avenue that motivates the planner to manipulate the household's beliefs, because they affect equilibrium Arrow-Debreu prices. …

The incorporation of Moorean type information by introspective agents

The main task is to discuss the issue in belief dynamics in which philosophical beliefs and rational introspective agents incorporate Moorean type new information. First, a brief survey is conducted on Moore’s Paradox, and one of its solutions is introduced with the help of Update Semantics. Then, we present a Dynamic Doxastic Logic (DDL) which revises the belief of introspective agents put forward by Lindstrom & Rabinowicz. Next, we attempt to incorporate Moorean type new information within the DEL (DDL) framework, as advised by van Benthem, Segerberg et al. Though we maintain the principle of “the primacy of new information” from the literature on traditional belief revision theory, several unsuccessful ways are also presented. We then conclude that some special kind of success (weak success) can still be found in those revision processes although absolute success does not hold. At last, the relevant problem of “learnability” is re-considered through weak success.