Samenvatting Experimental and Behavioural Economics
Lecture 1a. Introduction
Experimental economics is an empirical method, a way to gather data.
Goals of empirical economics:
- draw causal inferences
- make predictions about ‘new’ policy measures
- examine effect of interventions (treatments)
Treatment effect: difference between the outcome with or without treatment
Counterfactual: expressing what has not happened but could, would, or might under
differing conditions
Randomization
Treatments are randomly assigned. This ensures:
- Control group: some subjects are not exposed to the treatment
- Ceteris paribus comparison: subjects in treatment and control are similar on
average
Randomization ensures a reliable counterfactual (if enough participants) and avoids
confounding effects(=alternative explanations for results that can’t be excluded)
Methodology: Economics as an experimental discipline
Lecture 1b Introduction to behavioural economics
Experiment: A controlled economic environment in which experimental subjects make
decisions that the experimenter records for the purpose of scientific analysis.
controlled economic environment : The individual economic agents together with an
institution through which the agents interact.
Economic experiments study simplified economic reality: data can neither mimic the ‘real
world’ nor perfectly mirror formal models.
- participants make decisions with house money
- models often assume risk-neutrality, probably not the case in the lab.
Traditionally, experiments have been done in the laboratory, while(other) empirical research
examined naturally-occuring setting (also called a field setting)
Free-rider problem: everyone would like others to contribute
Economics vs Psychology:
, ● Different interests: economists focus on behavior in an economic context including
restrictions and institutions (e.g. markets, games), while psychologists focus more
on individual, unrestrained, behavior.
● Stylistics differences: Economists are more theory-based and focus more on
outcomes, while psychologists focus on the process
● Design and procedures: Induces-value theory in economics, experimental
deception/manipulation (i.e. Lying to subjects about elements of the experiment)
Internal validity:
- Is the attribution of causality warranted?
- Is the variation in y caused by variation in x?
Experiments should score high on internal validity, it’s the degree to which conclusions
about results are correct.
External validity:
Does the causal relation from x to y generalize over subjects and environments?
More than one experiment is needed to ensure this.
Parallelism precept:
Laboratory finding should generalize to comparable field settings.
Criticism of laboratory experiments:
- Lab is artificial
- Students are ‘WEIRD”(Western, Educated, Industrialized, Rich, Democratic) group
of people
- Not much at stake
Possible reactions on this criticism:
- Lab experiments are one way of looking at the truth’,and other types of data
collection are complementary and maybe needed(like field data)
- Run more experiments,e.g.with other subject pools and with high stakes
You can add external validity with experiments in a (more) field setting
The lab vs. field classification is not absolute (it’s a gradual scale)
Six factors that can be used to determine the field context of an experiment:
The nature of the ….
● … subject pool
● … information that subject brings to the task
● … commodity
● … task or trading rules applied
● … stakes
● … environment that the subject operates in
, - Conventional laboratory experiment(Lab): employ standard subject pool of
students, abstract framing,imposed set of rules.
- Artifactual field experiment(AFE): same as conventional lab experiment but with
non-standard subject pool
- Framed Field experiment(FFE): same as artifactual field experiment but with field
context in either the commodity, task, or informations set that subjects can use
- Natural Field experiment(NFE): same as a framed field experiment but where the
environment is one where the subjects naturally undertake these tasks and where
subjects don't know that they are in an experiment
From lab to field:
Stoop, Noussair & Van Soest(2012, JPE): measure cooperation between fishermen
- Lab: students, abstract framing
- Groups of 4 subjects, allowed to catch at most 2 fish
- Receive € 1 for each fish caught
- For each fish you don't catch, the other group receive € 0.50 each
- AFE: Fishermen in the lab, abstract framing
- FFE: Fishermen on the side of the pond, instructed
Contribution on Experimental Economics
Vernon Smith: Developed methods for laboratory experiments in economics
Daniel Kahneman: Integrated economic analysis with insights from cognitive psychology,
in decisions under uncertainty
Elinor Ostrom: Demonstrated how local property can be successfully managed by local
commons without centralized
Richard Thaler: Built a bridge between economic and psychological analyses of individual
decision-making. He has been instrumental in creating the field of behavioural economics .
Esther Duflo: Introduced a new approach to obtain reliable answers about best ways to
fight poverty
Designing and conducting experiments
experimental …
- … data
- … variation and control
- … techniques
- … subjects
- … procedure
Experimental Data
What is an observation?
Casual vs. Statistical observation:
Casual: any relevant information recorded
Statistical: observations that can be used for statistical tests (hard facts)
quote of the day:
, ‘’Scientific knowledge must be based on hard facts, but soft facts may inspire new
research’’
Terminology:
- Round(run, periodortrial): logically or organizationally indivisible set of decisions
- Cohort(independent subject group):all subjects that interact during a session
- Session:all rounds that are conducted with same group of subjects at one occasion
- Treatment(block):one or more sessions with an identical economic
environment,i.e. with the same configuration of treatment variables
- Experiment:the collection sessions of one or more related treatments
Experimental Design: specification of sessions in one/more treatments
Data descriptors: single vs. compound (aggregated) observations
independence requirement?
A stochastic variable X is independent of a stochastic variable Y, if the conditional
distribution of X does not vary with the value of Y.
Procedural independence: Observations are assumed to be independent,because they
are generated by subjects that could not interact, due to the experimental procedure.
Hypothesized independence: Observations are assumed to be independent, even though
subjects could have interacted given the experimental procedure. This is only justified,if it is
statistically tested.
Partner matching: group composition is the same (fixed)throughout the session
Stranger matching: group composition changes each round
Uniformity requirement: Uniformity may be assumed, if all observations are generated by
the same experimental procedure and the same process of recruitment.
Experimental variation and control
different types of variables:
Treatment Variables: parameters of the economic environment that are set at specific
values by the experimenter in order to test treatment differences in behavior and outcome.
Focus variables: The outcome variables in which we are interested
Nuisance variables: the variables that we do not want to study, but must record in case
they affect the results
Confounding effects (alternative explanations for the treatment effect that can’t be
excluded) should be avoided using controlled variation ( ceteris paribus comparison, only
1 parameter is changed)
Lecture 1a. Introduction
Experimental economics is an empirical method, a way to gather data.
Goals of empirical economics:
- draw causal inferences
- make predictions about ‘new’ policy measures
- examine effect of interventions (treatments)
Treatment effect: difference between the outcome with or without treatment
Counterfactual: expressing what has not happened but could, would, or might under
differing conditions
Randomization
Treatments are randomly assigned. This ensures:
- Control group: some subjects are not exposed to the treatment
- Ceteris paribus comparison: subjects in treatment and control are similar on
average
Randomization ensures a reliable counterfactual (if enough participants) and avoids
confounding effects(=alternative explanations for results that can’t be excluded)
Methodology: Economics as an experimental discipline
Lecture 1b Introduction to behavioural economics
Experiment: A controlled economic environment in which experimental subjects make
decisions that the experimenter records for the purpose of scientific analysis.
controlled economic environment : The individual economic agents together with an
institution through which the agents interact.
Economic experiments study simplified economic reality: data can neither mimic the ‘real
world’ nor perfectly mirror formal models.
- participants make decisions with house money
- models often assume risk-neutrality, probably not the case in the lab.
Traditionally, experiments have been done in the laboratory, while(other) empirical research
examined naturally-occuring setting (also called a field setting)
Free-rider problem: everyone would like others to contribute
Economics vs Psychology:
, ● Different interests: economists focus on behavior in an economic context including
restrictions and institutions (e.g. markets, games), while psychologists focus more
on individual, unrestrained, behavior.
● Stylistics differences: Economists are more theory-based and focus more on
outcomes, while psychologists focus on the process
● Design and procedures: Induces-value theory in economics, experimental
deception/manipulation (i.e. Lying to subjects about elements of the experiment)
Internal validity:
- Is the attribution of causality warranted?
- Is the variation in y caused by variation in x?
Experiments should score high on internal validity, it’s the degree to which conclusions
about results are correct.
External validity:
Does the causal relation from x to y generalize over subjects and environments?
More than one experiment is needed to ensure this.
Parallelism precept:
Laboratory finding should generalize to comparable field settings.
Criticism of laboratory experiments:
- Lab is artificial
- Students are ‘WEIRD”(Western, Educated, Industrialized, Rich, Democratic) group
of people
- Not much at stake
Possible reactions on this criticism:
- Lab experiments are one way of looking at the truth’,and other types of data
collection are complementary and maybe needed(like field data)
- Run more experiments,e.g.with other subject pools and with high stakes
You can add external validity with experiments in a (more) field setting
The lab vs. field classification is not absolute (it’s a gradual scale)
Six factors that can be used to determine the field context of an experiment:
The nature of the ….
● … subject pool
● … information that subject brings to the task
● … commodity
● … task or trading rules applied
● … stakes
● … environment that the subject operates in
, - Conventional laboratory experiment(Lab): employ standard subject pool of
students, abstract framing,imposed set of rules.
- Artifactual field experiment(AFE): same as conventional lab experiment but with
non-standard subject pool
- Framed Field experiment(FFE): same as artifactual field experiment but with field
context in either the commodity, task, or informations set that subjects can use
- Natural Field experiment(NFE): same as a framed field experiment but where the
environment is one where the subjects naturally undertake these tasks and where
subjects don't know that they are in an experiment
From lab to field:
Stoop, Noussair & Van Soest(2012, JPE): measure cooperation between fishermen
- Lab: students, abstract framing
- Groups of 4 subjects, allowed to catch at most 2 fish
- Receive € 1 for each fish caught
- For each fish you don't catch, the other group receive € 0.50 each
- AFE: Fishermen in the lab, abstract framing
- FFE: Fishermen on the side of the pond, instructed
Contribution on Experimental Economics
Vernon Smith: Developed methods for laboratory experiments in economics
Daniel Kahneman: Integrated economic analysis with insights from cognitive psychology,
in decisions under uncertainty
Elinor Ostrom: Demonstrated how local property can be successfully managed by local
commons without centralized
Richard Thaler: Built a bridge between economic and psychological analyses of individual
decision-making. He has been instrumental in creating the field of behavioural economics .
Esther Duflo: Introduced a new approach to obtain reliable answers about best ways to
fight poverty
Designing and conducting experiments
experimental …
- … data
- … variation and control
- … techniques
- … subjects
- … procedure
Experimental Data
What is an observation?
Casual vs. Statistical observation:
Casual: any relevant information recorded
Statistical: observations that can be used for statistical tests (hard facts)
quote of the day:
, ‘’Scientific knowledge must be based on hard facts, but soft facts may inspire new
research’’
Terminology:
- Round(run, periodortrial): logically or organizationally indivisible set of decisions
- Cohort(independent subject group):all subjects that interact during a session
- Session:all rounds that are conducted with same group of subjects at one occasion
- Treatment(block):one or more sessions with an identical economic
environment,i.e. with the same configuration of treatment variables
- Experiment:the collection sessions of one or more related treatments
Experimental Design: specification of sessions in one/more treatments
Data descriptors: single vs. compound (aggregated) observations
independence requirement?
A stochastic variable X is independent of a stochastic variable Y, if the conditional
distribution of X does not vary with the value of Y.
Procedural independence: Observations are assumed to be independent,because they
are generated by subjects that could not interact, due to the experimental procedure.
Hypothesized independence: Observations are assumed to be independent, even though
subjects could have interacted given the experimental procedure. This is only justified,if it is
statistically tested.
Partner matching: group composition is the same (fixed)throughout the session
Stranger matching: group composition changes each round
Uniformity requirement: Uniformity may be assumed, if all observations are generated by
the same experimental procedure and the same process of recruitment.
Experimental variation and control
different types of variables:
Treatment Variables: parameters of the economic environment that are set at specific
values by the experimenter in order to test treatment differences in behavior and outcome.
Focus variables: The outcome variables in which we are interested
Nuisance variables: the variables that we do not want to study, but must record in case
they affect the results
Confounding effects (alternative explanations for the treatment effect that can’t be
excluded) should be avoided using controlled variation ( ceteris paribus comparison, only
1 parameter is changed)
