Households’ Behaviors and Systemic Financial Instability

Experimental Insights and Agent-based Simulations

for Macroeconomic Policy Analyses

Paola D’Orazio

Department of Economic-Quantitative and Philosophical-Educational Sciences

University “G. D’Annunzio” Chieti-Pescara, viale Pindaro, 42, 65127 Pescara, Italy

1 OBJECTIVES AND STAGE OF

THE RESEARCH

In the last two years my research has focused on

macroeconomic ﬁnancial fragility. In particular, on

how to bring different methodologies together to

study endogenous generated crises

Economics profession is currently engaged in a de-

bate on which are the best methodological tools in or-

der to study the dynamics of the real economy and

adequately address important policy issues and so-

cial concerns. My research endorse the view accord-

ing to which traditional economic theories and the

models they give rise to (namely, the DSGE models)

are ill-equipped to manage serious crises which of-

ten emerge in real world economies. In my research

project I am trying to address these issues by develop-

ing an “experimentally” microfounded Agent-based

model (ABM), which will account also for the con-

sistency of stocks and ﬂows

The main research issues concern both the micro and

the macro level, i.e., agents and the environment in

which they act and interact. For the micro level, I have

designed an experiment in order to gain insights into

agents’ behaviors. For the macro level, I plan to build

an ABM where agents are estimated - rather than cal-

ibrated - by using experimental data.

I plan to start

working on the Agent-based macro structure once ﬁn-

ished the micro level analysis, i.e. (a) performing the

experiment and data analysis and (b) building the be-

Endogenous crises are those generated caeteris paribus

by non-price interactions, localized learning processes, in-

vestment practices. For a more comprehensive analysis of

ﬁnancial fragility and its effects on real economy see (Min-

sky, 1970, 1974, 1992).

This is a relevant aspect of the model that will be built.

Detailed explanation and motivations for this requirement

are given in 2.3. For an historical and methodological per-

spective on Stock-Flow consistency nad related models see

(Caverzasi and Godin, 2013).

This point is explained in section 4.

havioral rules for artiﬁcial agents. Once the ABM will

be set up, I will perform policy experiments in order

to evaluate the effectiveness and “proﬁtability” of dif-

ferent policies.

2 RESEARCH PROBLEM

The interest in this research originates from the in-

creasingly widespread opinion among scholars who

claim for the need of new tools to cope with the com-

plexity of socio-economic systems. The new tools

must be able to allow for the building of microfoun-

dations of Macroeconomics by considering also the

feedback effect the macro system has on individual

agents. Analytically solvable pure theoretical macro

models are of little help for policy guidance (Velupil-

lai, 2011) because they are based on assumptions that

are far from reality and do not consider any interac-

tion among heterogeneous agents (Kirman, 1992). In

order to consider the macro economy as a complex

endogenously organized system, researchers should

be able to study both the micro level, i.e., agents, and

the macro level, i.e., the environment. Concerning

the former, behaviors that guides the decision-making

process and the heterogeneity of agents’ population

must be incorporated into the model. Regarding the

latter, models have to deal with the complexity and

instability of the macro environment, which is indeed

a central feature of the choices that economic agents

face.

2.1 The Micro Level

As Simon argued in 1959, to explain agents’ behavior

in the face of complexity, the theory must incorporate

at least some description of the processes and mech-

anisms through which the adaptation takes place. Si-

mon claimed that the emergence of new areas of the-

ory and application, in which complexity and change

are central facts, led to the demand for a fuller picture

D’Orazio P..

Households’ Behaviors and Systemic Financial Instability - Experimental Insights and Agent-based Simulations for Macroeconomic Policy Analyses.

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

of economic man (Simon, 1959).

The “behavioral revolution” (Akerlof, 2002; Kahne-

man, 2003; Camerer et al., 2004), which has unfolded

in recent decades following in Simon’s footsteps, has

contributed to the increase of the explanatory power

of Economics by providing it with more realistic psy-

chological foundations and “practical” ways in which

behaviors can be incorporated into the models (Kao

and Velupillai, 2011).

2.2 The Macro Level

The second problem of the present research is related

to the deﬁnition of the economy as a Complex Adap-

tive System (Tesfatsion, 2003).

ABMs allow to study the “micro-to-macro map-

ping” (Epstein, 2007) as well as social complexity.

They are indeed ﬂexible, precise and consistent mod-

eling tool, providing the ability to explore learning

and adaptation phenomena and increasing our under-

standing of economic dynamics (Holland and Miller,

1991). In their “early steps” (LeBaron, 2000), ABMs

have been mainly used as a tool to scale up to the

macro level and to perform counterintuitive hypothe-

ses regarding individual behavior or to gain insights

into existing human experiments.

However, as Epstein (2007) points out, “even per-

fect knowledge of individual decision rules does not

always allow us to predict macroscopic structure”.

The microspeciﬁcation of the ABM is thus crucial:

individual behaviors “observed” in the experimental

laboratory become the solid foundationsof the model.

Moreover, information collected from experimental

micro-systems differs from that obtained from empir-

ical data (Smith, 1982) in that the former offers richer

insights into the individual and collective dynamics

of a model. Experimental data could thus represent a

solution to the problem of availability of micro data

which threatens the research on the microfoundations

of macroeconomic models (Carroll, 2012b).

2.3 The Stock-Flow Consistency

Requirement

Adding a stock-ﬂow consistency (SFC) requirement

to the model helps to overcome the fallacy of compo-

sition critique

. Moreover, it helps in detecting sec-

toral instability - the household sector in the present

research project - by considering agents’ balance

sheets.

It concerns the presumption that what is true of each

single part of a whole is necessarily true of the whole as

well. Basically, it questions the aggregation process used

by “standard” macroeconomic models.

SFC models are speciﬁc kinds of macro models

that try to coherently integrate all stocks and ﬂows of

an economy. The two main components are: (1) the

accounting framework, and (2) the behavioral equa-

tions. The ﬁrst of these two components usually relies

on a set of matrices reproducing the balance sheets,

the transactions, and the capital gains of each of the

institutional sectors into which the economy is orga-

nized. The second component is a set of behavioral

equations modeling all the transactions not directly

determined by the accounting structure of the econ-

omy.

Their roots lie in the study on “money ﬂows” of

Morris A. Copeland (Copeland, 1949, 1962)

. Re-

cently, SFC models are drawing new attention by re-

searchers in that they are developing new ways to use

the SFC framework as a development of the exist-

ing general aggregative models (Kinsella et al., 2011;

Seppecher, 2012; Raberto et al., 2012).

Until now SFC models are designed as systems

of stock-ﬂow consistent equations describing the laws

of motion of the economy at the aggregate level.

The combination of SFC and Agent-based modeling

tools will yield more complete macro models. On

one hand, the ABM can display consistency between

stocks and ﬂows; the result is a framework that en-

sures the compatibility of real and ﬁnancial variables.

On the other hand, the ABM provide explicit micro-

foundations to macroeconomic relations.

3 STATE OF THE ART

Agent-based models have been used to study ﬁndings

from human subject experiments. In his review, Duffy

(2006) calls the attention on the interaction between

ABM and Experimental Economics by claiming that

they are “natural allies”. He points out three main ar-

eas in which ABM have been used in this way. The

ﬁrst one is the so-called “zero-intelligent” agent ap-

proach based on the Gode and Sunder (1991) seminal

paper. It consists of models with very low rational-

ity constraints which made evident that self-interested

and rational behavior is observed mainly in highly

structured and constrained markets. A second line of

research focuses on reinforcerment and belief-based

models of agents’ behaviors. Finally, there are Evolu-

tionary Algorithms where individual learning is more

The intuition of Copeland was to enlarge the social ac-

counting perspective - which had been used mainly in the

study of national income - to the study of money ﬂows. He

laid the foundations for an economic approach able to in-

tegrate real and ﬁnancial ﬂows of the economy (Copeland

(1949), p. 254).

ICAART2014-DoctoralConsortium

complicated.

Brian Arthur was among the ﬁrsts in exploring the

idea of calibrating an algorithm to reproduce human

behavior (Arthur, 1991, 1993). He called the atten-

tion on the need to go beyond the assumption of ra-

tionality, suggesting some ways to model economic

choices. He did not want to design a learning algo-

rithm or automaton that maximizes some criterion.

Rather, he aimed at designing an algorithm that can

be “tuned to choose actions in an iterated choice sit-

uation the way humans would ”(Arthur, 1991, p.354).

To calibrate the algorithm in a way that could be de-

ﬁned as a “good indication” of human behavior, he

used the results of an experiment performed in 1952-

53 by Robillard at Harvard University. His results

- and tests of ﬁtness - showed that the automaton

was able to replicate those behaviors also in differ-

ent choice problems, than those for which it was cali-

brated.

From Arthur on, Genetic Algorithm (GA) has

been extensively used in Economics; simulations’

results show GAs successfully replicate experimen-

tal behaviors in different environments (see Arifovic,

1996, 2000; Dawid, 1996, for a comprehensive sur-

vey). However, in the ABM literature artiﬁcial agents

have usually been considered as “equally smart”

(Chen and Yu, 2011) and they are built by relying on

available theories on individual decision-making (see

Raberto et al., 2012; Dosi et al., 2010, 2013, among

others). The exploration and induction of agents’ be-

haviors by means of the experimental method is thus

a more recent strand of research.

4 METHODOLOGY

Households’ ﬁnancial behaviors (in particular, their

“debt love”/debt aversion) will be put under the “mag-

nifying glass” by means of the experimental method

in order to detect ﬂaws in traditional theories of in-

tertemporal consumption/saving decisions. The ex-

perimental method will allow also to explore actual

decision-making processes and heterogeneity across

agents. Data collected from experiments will be used

to analyze the aggregate implications of households’

behaviors and to eventually estimate the behaviors of

artiﬁcial agents that will interact in the artiﬁcial envi-

ronment. As showed in Giulioni et al. (2013), the use

of experimental data allows to go beyond the standard

parameters’ calibration procedure.

The household sector will be included in a broader

project, i.e. a macroeconomic ABM, in which the

ﬁrm sector and the banking sector will be also consid-

ered. The “integrated” experimental-ABM methodol-

ogy will allow studying both how the macro dynam-

ics evolve and the eventual endogenously generated

crises.

4.1 Experimental Insights on

Intertemporal Decision-making

Experimental Economics complements computa-

tional, theoretical and empirical works (Davis and

Holt, 1993; Binmore, 1999; Samuelson, 2005) in that

it helps identifying behavioral rules agents use in eco-

nomic decision-making (Kahneman, 2003). Indeed,

the experimentation process is not a “simple” addi-

tional element of the modeling process; rather it inter-

acts at a deeper level with the limited cognitive abili-

ties of economic agents.

Several studies have been highlighting that indi-

vidual preferences - and macro feedbacks - are ob-

servable and controllable in the laboratory such that

economic modeling becomes richer and more realis-

tic (Smith, 1982, 2002). In this sense, the experimen-

tal approach follows both the warnings of the well-

known Lucas’ critique (Lucas, 1986) and the Lucas’

invitation

. In his “invitation”, Lucas stresses the rel-

evance of experiments in shedding light on humans

decision making processes, and therefore on the eco-

nomic analysis as a whole.

In my research, experiments are used to closely in-

vestigate human behaviors in an attempt to contribute

to the research stream concerned with the building

of new microfoundations for macroeconomic models.

This method also allows for “detecting” agents het-

erogeneity and how they interact with each other.

4.1.1 The Experimental Design: Some

Theoretical and Methodological

Considerations

The designed experiment will address two issues in

intertemporal consumption models. On one side, the

experimental approach will allow to investigate the

ability of subjects to solve the task of intertempo-

ral optimization and the extent to which subjects be-

have according to standard optimization rules (Car-

roll, 1996, 2012b). In this way, the experiment will

take the theoretical predictions seriously and test if

consumers are able to carry out a dynamic intertem-

poral (utility) optimization problem. On the other

side, the experimental method will be used to study

the pervasiveness of debt in the liabilities side of

households’ portfolios. The model will thus account

As described in the Experimental Economics entry in

the New Palgrave Dictionary of Economics (Duffy, 2008).

Households'BehaviorsandSystemicFinancialInstability-ExperimentalInsightsandAgent-basedSimulationsfor

MacroeconomicPolicyAnalyses

also for the role ﬁnancial innovations have been gain-

ing in capitalistic economies in the last decades.

The experimental design is grounded on methods

and results of the early experiments performed on in-

tertemporal consumption. The ﬁrst attempt of testing

how closely the predictions of the optimality theory

ﬁt the actual behavior of subjects in an experimental

setting is the paper by Hey and Dardanoni (1987). My

experiment is closely related to the more recent exper-

iment performed by Ballinger et al. (2003) (although

I do not focus on intergenerational learning) and it

is designed in the footprints of the “learning to opti-

mize” experiments (LtOEs) where subjects are asked

to directly make economic decisions (to consume, in-

vest, trade, produce, etc.)

4.1.2 The Experimental Design: The Model

The starting point is the building of a benchmark

intertemporal consumption model which is used to

compute the optimal (theoretical) solution that will be

compared to our experimental data in order to assess

if there is a deviation from the optimal behavior. The

major innovation introduced concerns the relaxation

of the standard budget constraint by allowing borrow-

ing, hence debt.

The solution for the intertemporal consumption

problem is found by backward induction, following

the method for microeconomic dynamic stochastic

optimization problems. Among others, we consid-

ered the methods developed by Carroll (2012a) and

Stachurski (2009). Given that an explicit solution

to the problem does not exist (it is implicitly char-

acterised by an Euler equation), numerical methods

are needed in order to ﬁnd an optimal policy func-

tion (for saving and borrowing tasks) to be compared

to subjects’ decisions in the lab. This methodologi-

cal choice is of particular importance because in the

experiment we should be able to control for all possi-

ble confounding factors and focus on few variables of

interest.

For the standard intertemporal consumption

model set up, I compute the policy function for a con-

sumption/saving task which in turn will be used as a

benchmark for experimental data.

In order to have a simple and tractable model I

do not consider the discount factor; a separate exper-

iment is necessary in order to elicit - thus estimate -

the discount parameter of subjects in the lab. I indeed

LtOEs differ from “learning-to-forecast” experiments

(LtFEs) in that LtFEs aim atelicitating subjects’ forecasts of

the relevant endogenous variables such as the market price,

interest rate or wages. See, Arifovic (1996) and Smith et al.

(1988).

decided to leave this investigation for another exper-

iment (Cubitt and Read, 2007; Coller and Williams,

1999; Harrison et al., 1995).

The intertemporal settings that will be ana-

lyzed will be basically two. The perfect certainty

model (deterministic) which would be the ”candi-

date” framework to test the predictions of the model

built on the rational expectations assumptions, and a

stochastic version in which there is uncertainty about

the future income stream. The latter will be useful to

assess which kind of expectations (adaptive, etc) arise

among experimental subjects.

Th experimental subject pool will be composed by

students and workers

in an attempt to address the

usual criticism about the external validity of experi-

mental data and results.

4.2 The Economy as a Complex Adaptive

System

Agent-based modeling is of particual interest for

my research since it allows for taking into account

the possibility of endogenous co-evolution of micro-

behaviors and institutions (namely, the macro struc-

ture), the heterogeneity and interactions among eco-

nomic agents that can lead the system to breakdown

without external interference, i.e., shocks. In this,

ABMs are a versatile tool: observed dynamics are

open-ended (not closed form) and they allow for an

ergodic state of the system, i.e., an equilibrium, which

is an emergent and optional outcome (Delli Gatti

et al., 2008). While DSGE models are based on the

centralized information processing structure, ABM

takes a bottom-up approach that starts modelling real-

istic microfoundations and ends up analyzing the re-

sulting aggregate behaviour. The dynamics of aggre-

gate variables are the result of complex, continuously

and endogenously changing micro-structure. This

yields substantial advantages in modelling macroeco-

nomic policies (LeBaron and Tesfatsion, 2008).

However, there are some methodological prob-

lems related to ABM. Caeteris paribus, the empir-

ical model validation

and robustness checks of re-

sults. Indeed, the large ﬂexibility of the setup (start-

ing values) of agent-based models and the number of

selected parameters give many “degrees of freedom”

to the researcher. This in turn poses serious chal-

lenges to the use of ACE models for the evaluation

and design of economic policy measures. This prob-

lem could be (partially) overcome by using the ex-

They have been recruited throught the ORSEE software

(Greiner, 2004).

For a broader discussion on this issue see (Fagiolo

et al., 2007a,b).

ICAART2014-DoctoralConsortium

perimental method to build (i.e., estimate) agents that

populate the artiﬁcial environment.

4.3 Genetic Algorithm “Revisited”

The application and development of the methodology

presented and discussed in this paper beneﬁts mainly

from theoretical and “practical” insights of Brock and

Hommes (1997) seminal paper. It is also inspired

from the improvements of “heuristics switching mod-

els”, in which agents have a set of imple forecasting

heuristics (adaptive, trend extrapolating and so on)

and choose those that had a better past performance.

However, in order to account for a “complete”

agents’ heteroegeneity and give the model better mi-

crofoundations, the researcher needs tools that allow

to consider the whole set of behavioral rules ﬁnded in

experimental data. In the last decades, many schoars

have been engaged in this line of research so that het-

erogenous, interacting agents in ABM are designed in

many different ways (for a comprehensive survey see

Chen (2012)).

At the present stage of my research I am consid-

ering retaking Brian Arthur’s main arguments and re-

sults (as discussed in section 3) and go beyond the cal-

ibration process by estimating the behavioral rules for

artiﬁcial agents. Indeed, the main point of the present

research project is that by means of the experimental

method it is possible to gain insights into human be-

havior and collect information on heterogeneity of the

population. Experimental data can thus be used to es-

timate, rather than “merely” calibrate, the behavioral

rules that guide artiﬁcial agents’ actions

In order to ﬁt the ABM using experimental data, I

am considering the implementation of a Genetic Al-

gorithm, Classiﬁer Systems or a combination of both.

GA is a ﬂexible optimization procedure which has

been extensively used in Economics (Arifovic, 2000,

1996). Classiﬁer systems is a machine learning sys-

tem that learns syntactically simple string rules, called

classiﬁers (Booker et al., 1989; Birchenhall, 1995).

The decision will come after a comparison and careful

analysis of the pros and cons - ad the related effective-

ness - of the application of the two techniques to the

problem under scrutiny.

In the evaluation of which evolutionary method

and technique will best ﬁt for my project, I am

also concerned about the economic soundness of this

tools. I am indeed considering both the concerns

and the warning of Waltman et al. (2011); Dawid and

Hansen and Heckman (1996) offer an interesting per-

spective on model calibration and estimation methodolo-

gies. It also discusses the related empirical “hidden dan-

gers”.

Dermietzel (2006) about the economic interpretation

of evolutionary algorithms based on binary encoding

and other techniques adopted from computer science

without any modiﬁcation.

5 TIMING AND EXPECTED

OUTCOME

The research presented in this paper represents a doc-

toral research project which is still in progress. It

beneﬁts from the integration of different methodolo-

gies and modeling techniques. The last 2 years have

been devoted to the study of both the experimental

and the agent-based methods and to ﬁgure how they

can beneﬁt from each other in order to build a macro

model. The experimental phase is ready to be imple-

mented; once performed I will use experimental data

to build the behavioral rules for artiﬁcial agents and

thus populate the artiﬁcial economy, i.e., the ABM.

The expected outcome of the present research is in-

deed the development of an “integrated” methodol-

ogy of research which is based on the interaction of

Agent-based modeling tools and Experimental Eco-

nomics method, with the “additional” requirement of

SFC.

The combined use of Experimental Economics

method and Agent-based Computational Economics,

enriched by the “informed intuition” of stock-ﬂow

consistency, will allow to develop more real-like

macro models and more practical tools to guide pol-

icymakers. The ABM will be then used to perform

some policy experiments, i.e., analyses on the effects

and effectiveness of different ﬁscal and/or monetary

policies. The problem of availability of micro data

(Carroll, 2012b) for a “proper” estimation of agents’

behavioral rules will be overcome by relying on data

collected from experiments. Finally, the present re-

search is also aiming at contributing to the improve-

ment of the ABMs’ calibration/estimation techniques.

REFERENCES

Akerlof, G. A. (2002). Behavioral macroeconomics and

macroeconomic behavior. American Economic Re-

view, 92(3):411–433.

Arifovic, J. (1996). The behavior of the exchange rate in the

genetic algorithm and experimental economies. Jour-

nal of Political Economy, 104(3):510–41.

Arifovic, J. (2000). Evolutionary algorithms in macroe-

conomic models. Macroeconomic Dynamics, 4:373–

414.

Arthur, W. B. (1991). Designing Economic Agents that

Act like Human Agents: A Behavioral Approach to

Households'BehaviorsandSystemicFinancialInstability-ExperimentalInsightsandAgent-basedSimulationsfor

MacroeconomicPolicyAnalyses

Bounded Rationality. American Economic Review Pa-

pers and Proceedings, 81(2):353–359.

Arthur, W. B. (1993). On designing economic agents that

behave like human agents. Journal of Ecolutionary

Economics, 3:1–22.

Ballinger, T. P., Palumbo, M. G., and Wilcox, N. T.

(2003). Precautionary saving and social learning

across generations: an experiment. Economic Jour-

nal, 113(490):920–947.

Binmore, K. (1999). Why Experiment in Economics? Eco-

nomic Journal, 109(453):16–24.

Birchenhall, C. R. (1995). Genetic algorithms, classiﬁer

systems and genetic programming and their use in the

models of adaptive behaviour and learning. The Eco-

nomic Journal, 105(430):788–795.

Booker, L. B., Goldberg, D. E., and Holland, J. H. (1989).

Classiﬁer systems and genetic algorithms. Artiﬁcial

Intelligence, 40(1-3):235–282.

Brock, W. A. and Hommes, C. H. (1997). A rational route

to randomness. Econometrica, 65(5):1059–1096.

Camerer, C. F., Loewenstein, G., and Rabin, M. (2004). Ad-

vances in Behavioral Economics. Princeton Univer-

sity Press, Princeton.

Carroll, C. (2012a). Lecture notes on solution methods for

microeconomicdynamic stochastic optimization prob-

lems. Lecture Notes - John Hopkins University.

Carroll, C. D. (1996). Buffer-stock saving and the life cy-

cle/permanent income hypothesis. NBER Working

Papers 5788, National Bureau of Economic Research,

Inc.

Carroll, C. D. (2012b). Representing consumption and sav-

ing without a representative consumer. In Measuring

Economic Sustainability and Progress, NBER Chap-

ters. National Bureau of Economic Research, Inc.

Caverzasi, E. and Godin, A. (2013). Stock-ﬂow consistent

modeling through the ages. Economics working paper

archive, Levy Economics Institute, The.

Chen, S.-H. (2012). Varieties of agents in agent-based com-

putational economics: A historical and an interdisci-

plinary perspective. Journal of Economic Dynamics

and Control, 36(1):1–25.

Chen, S.-H. and Yu, T. (2011). Agents learned, but

do we? knowledge discovery using the agent-based

double auction markets. In Frontiers of Electrical

and Electronic Engineering in China, pages 159–170.

Springer.

Coller, M. and Williams, M. (1999). Eliciting individual

discount rates. Experimental Economics, 2(2):107–

127.

Copeland, M. A. (1949). Social accounting for moneyﬂows.

The Accounting Review, page 254264.

Copeland, M. A. (1962). Some illustrative analytical uses of

ﬂow-of-funds data. In The Flow-of Funds Approach to

Social Accounting, pages 195–238. National Bureau

of Economic Research, Inc.

Cubitt, R. and Read, D. (2007). Can intertemporal choice

experiments elicit time preferences for consumption?

Experimental Economics, 10:369–89.

Davis, D. D. and Holt, C. A. (1993). Experimental eco-

nomics: Methods, problems and promise. Estudios

Economicos, 8(2):179–212.

Dawid, H. (1996). Adaptive Learning by Genetic Algo-

rithms: Analytical Results and Applications to Eco-

nomic Models. Springer Verlag New York, Inc.

Dawid, H. and Dermietzel, J. (2006). How robust is the

equal split norm? responsive strategies, selection

mechanisms and the need for economic interpretation

of simulation parameters. Computational Economics,

28(4).

Delli Gatti, D., Gaffeo, E., Gallegati, M., Giulioni, G.,

and Palestrini, A. (2008). Emergent Macroeconomics.

Springer Verlag, Berlin.

Dosi, G., Fagiolo, G., Napoletano, M., and Roventini, A.

(2013). Income distribution, credit and ﬁscal policies

in an agent-based keynesian model. Journal of Eco-

nomic Dynamics and Control, 37(8):1598–1625.

Dosi, G., Fagiolo, G., and Roventini, A. (2010). Schum-

peter meeting keynes: A policy-friendly model of en-

dogenous growth and business cycles. Journal of Eco-

nomic Dynamics and Control, 34(9):1748–1767.

Duffy, J. (2006). Agent-based models and human subject

experiments. In Tesfatsion, L. and Judd, K. L., editors,

Handbook of Computational Economics, volume 2 of

Handbook of Computational Economics, chapter 19,

pages 949–1011. Elsevier.

Duffy, J. (2008). experimental macroeconomics. In

Durlauf, S. N. and Blume, L. E., editors, The New Pal-

grave Dictionary of Economics, volume 3. Palgrave

Macmillan, Basingstoke, second edition.

Epstein, J. M. (2007). Agent-based computational mod-

els and generative social science. In Generative So-

cial Science Studies in Agent-Based Computational

Modeling, Introductory Chapters. Princeton Univer-

sity Press.

Fagiolo, G., Birchenhall, C., and Windrum, P. (2007a). Em-

pirical Validation in Agent-based Models: Introduc-

tion to the Special Issue. Computational Economics,

30(3):189–194.

Fagiolo, G., Moneta, A., and Windrum, P. (2007b). A

critical guide to empirical validation of agent-based

models in economics: Methodologies, procedures,

and open problems. Computational Economics,

30(3):195–226.

Giulioni, G., Bucciarelli, E., Silvestri, M., and D’Orazio,

P. (2013). Agent-based computational economics and

experimental economics: a bridge to progress macroe-

conomics. manuscript submitted.

Gode, D. and Sunder, S. (1991). Allocative efﬁciency of

markets with zero intelligence (z1) traders: Market

as a partial substitute for individual rationality. Gsia

working papers, Carnegie Mellon University, Tepper

School of Business.

Greiner, B. (2004). An online recruitment system

for economic experiments. Forschung und wis-

senschaftliches rechnen 2003, GWDG Bericht.

Hansen, L. P. and Heckman, J. J. (1996). The empirical

foundations of calibration. Journal of Economic Per-

spectives, 10(1):87–104.

Harrison, G. W., Harstad, R. M., and Rutstrm, E. E. (1995).

Experimental methods and elicitation of values. Dis-

cussion paper serie b, University of Bonn, Germany.

ICAART2014-DoctoralConsortium

Hey, J. D. and Dardanoni, V. (1987). Optimal consump-

tion under uncertainty: An experimental investigation.

Economic Journal, 98(390):105–16.

Holland, J. H. and Miller, J. H. (1991). Artiﬁcial adaptive

agents in economic theory. The American Economic

Review, 81:365–370.

Kahneman, D. (2003). Maps of bounded rationality: Psy-

chology for behavioral economics. American Eco-

nomic Review, 93(5):1449–1475.

Kao, S. Y. F. and Velupillai, K. V. (2011). Behavioural eco-

nomics: Classical and modern. Technical report.

Kinsella, S., Greiff, M., and Nell, E. J. (2011). Income

distribution in a stock-ﬂow consistent model with ed-

ucation and technological change. Eastern Economic

Journal, 37(1):134–149.

Kirman, A. P. (1992). Whom or What Does The Repre-

sentative Individual Represent. Journal of Economic

Perspective, 6:117–36.

LeBaron, B. (2000). Agent-based computational ﬁnance:

Suggested readings and early research. Journal of

Economic Dynamics and Control, 24(5-7):679–702.

LeBaron, B. and Tesfatsion, L. (2008). Modeling macroe-

conomies as open-ended dynamic systems of interact-

ing agents. American Economic Review, 98(2):246–

50.

Lucas, R. E. (1986). Adaptive behavior and economic the-

ory. Journal of Business, 59:S401–S426.

Minsky, H. P. (1970). Financial instability revisited: the

economics of disaster. Board of Governors of the Fed-

eral Reserve System, Washington.

Minsky, H. P. (1974). The modeling of ﬁnancial instabil-

ity: an introduction. In Modelling and Simulation.

Proceedings of the 5th Annual Pittsburg Conference,

pages 267–272.

Minsky, H. P. (1992). The Financial Instability Hypothe-

sis. Working Paper 74, The Jerome Levy Economics

Institute of Bard College.

Raberto, M., Teglio, A., and Cincotti, S. (2012). Debt,

deleveraging and business cycles: An agent-based

perspective. Economics - The Open-Access, Open-

Assessment E-Journal, 6(27):1–49.

Samuelson, L. (2005). Economic theory and experi-

mental economics. Journal of Economic Literature,

43(1):65–107.

Seppecher, P. (2012). Flexibility of wages and macroe-

conomic instability in an agent-based computational

model with endogenous money. Macroeconomic Dy-

namics, 16(S2):284–297.

Simon, H. A. (1959). Theories of Decision-Making in Eco-

nomics and Behavioral Science. The American Eco-

nomic Review, 49(3):253–283.

Smith, V. L. (1982). Microeconomic Systems as an Ex-

perimental Science. American Economic Review,

72(5):923–55.

Smith, V. L. (2002). Method in Experiment: Rhetoric and

Reality. Experimental Economics, 5:91–110.

Smith, V. L., Suchanek, G. L., and Williams, A. W.

(1988). Bubbles, crashes, and endogenous expecta-

tions in experimental spot asset markets. Economet-

rica, 56(5):1119–51.

Stachurski, J. (2009). Economic Dynamics. Theory and

Computation. MIT Press.

Tesfatsion, L. (2003). Agent-based computational eco-

nomics: Modeling economies as complex adaptive

systems. Information Sciences, 149:263–269.

Velupillai, K. V. (2011). Towards an algorithmic revolu-

tion in economic theory. Journal of Economic Sur-

veys, 25(3):401–430.

Waltman, L., Eck, N., Dekker, R., and Kaymak, U. (2011).

Economic modeling using evolutionary algorithms:

the effect of a binary encoding of strategies. Journal

of Evolutionary Economics, 21(5):737–756.

Households'BehaviorsandSystemicFinancialInstability-ExperimentalInsightsandAgent-basedSimulationsfor

MacroeconomicPolicyAnalyses