Summary experimental and behavioral economics
Week 1: introduction and methodology
- Lecture 1a: introduction to experimental economics
Experimental economics is an empirical method, a way to gather data.
The goal is to draw causal inferences, make predictions about new policy measures and to
examine the effect of interventions (treatments).
What do experiments tell us about theory?
Descriptive approach: testing whether actual behavior satisfies the
assumptions about human behavior on which theories are based (most
psychologists).
Prescriptive/normative approach (as if): testing whether the
conclusions/implications of theories offer good(-enough) approximations
of actual behavior, even when its assumptions are not precisely satisfied
(most economists).
Correlation vs causation
- We denote the treatment as a dummy variable, with value 1 if the subject is in the
treatment group, and 0 if the subject is in the control group.
- Y1i = outcome with treatment for person i
- Y0i = outcome without treatment for person i
- However, we cannot observe both outcomes as one person cannot be in the treatment
and control group simultaneously we lack a counterfactual (the value of Y if the
subject did not receive the treatment when he or she was in the treatment, or the other
way around).
- We can overcome this by testing more people and using a control group. These people
need to be somewhat similar to have a trustworthy counterfactual (do not construct
one group with only men and one with only women, etc.).
Economic experiments
- These allow control through randomization.
- This means that the treatment is randomly assigned. This ensures:
, o Control group: some subjects are not exposed to the treatment
o Ceteris paribus comparison: subjects in treatment and control are similar on
average.
- Randomization, therefore, ensures a reliable counterfactual if there are enough
participants.
- The only thing that is different for the treatment compared to the control group is, the
treatment.
- Confounding effects need to be avoided (these are alternative explanations for the
results, for example, by having more differences between the control and treatment
group).
Laboratory experiments
- We focus on laboratory experiments. These are organized, such that everyone knows
that they are participating in an experiment.
- To participate in an experiment, the following things must apply:
o You perform simple decision task
o All information needed is provided at the spot
o Decisions have real consequences: you may earn money
o Individual decisions or games against other participants
o Everything is anonymous
- Lecture 1a: introduction to behavioral economics
Behavior is often irrational according to the neo-classical model. Behavioral economics uses
social, cognitive and emotional factors in understanding the economic definitions of
individuals and institutions. This all has to do with the concept of bounded rationality/ will-
power / self-interest. Psychology is often used in behavioral economics as well.
Behavioral economics can be seen as the field, and experimental economics as the
tool/method.
- Lecture 1b: introduction to behavioral economics (continued)
This is precisely what we do in the lectures. We begin with a known theory (game theory,
individual decision making, social preferences and market experiments), then we test this in
an experiment. Later, we update our theory with the findings of our experiment (if this is
found on a large scale, we do not update a theory if just one experiment found something).
The first principles of neoclassical economics
, - The first principles are rationality and self-interest. However, lately there is lots of
evidence that shows that people are not always self-interested (altruism) or rational
(bounded rationality).
- There are 2 systems of cognition system 1 (intuition) and system 2 (reasoning).
- We often think with our first system, which is fast and uses emotions. Intuitive
thoughts come to mind spontaneously. A key principle is accessibility: the more we
can imagine something to happen, the quicker we will think about this.
- Behavioral economics understands that people make systematic mistakes due to
system 1 and tries to put them in perspective of theories.
- Lecture 1b: methodology
Economics as an experimental discipline
Reasons for an experiment
1. Theory falsification: Do people play a mixed strategy Nash equilibrium in practice?
2. Applicability of competing theories: when is one more successful than the other?
3. Robustness / theory stress tests: when does the theory (not) work?
4. Generate data for policy advice: how do people respond to different institutions?
5. Test-bed mechanism: new institutions that are too complex for theory
6. Discover empirical regularities: in the absence of accurate theory
7. Measure individual characteristics: risk aversion, impatience, etc.
8. Simulation of natural economic processes (rare): macroeconomic experiments
9. Pedagogical purposes: teach students about alternative theories
Economic experiments
An experiment is a controlled economic environmental in which experimental subjects make
decisions that the experimenter records for the purpose of scientific analysis.
A controlled economic environmental are the individual agents together with an institution
through which the agents interact.
Experimental economics is the study of economics with data generated under controlled
conditions, either in the field or in the lab. Experimental economics can also be seen as a
research method for behavioral economics.
It is useless to try to replicate certain things in a lab:
- Field environment
- Precise assumptions of a model (which omits details)
, Economic experiments study simplified economic reality: data can neither mimic the real
world nor perfectly mirror formal models.
There are different types of experiments:
- Tests of behavioral hypotheses/ theory falsification
- Theory stress tests/ sensitivity tests link laboratory to more naturally occurring
markets
- Searching for empirical regularities
Data sources
- Experimental vs. happenstance data: experimental data are deliberately created under
controlled conditions. Happenstance data are a by-product of ongoing uncontrolled
processes.
- Laboratory data vs. field data: laboratory data are gathered in an artificial
environment. Field data are gathered in a naturally occurring environment.
- Costs to gather happenstance and experimental data are high (unless it has been
collected by government in the case of happenstance data).
There are different types of decisions in an experiment:
- Individual
- Strategic, against another participant
Note: experimentalists claim that an economic experiment with proper
incentives (e.g. monetary incentives) is economic reality, because the
decisions have economic effects.
Economics vs. psychology
Both use experiments to gather data, but there are some differences:
- Different interests: economics focus on behavior in an economic context including
restrictions and institutions, while psychologists focus more on individual,
unrestrained, behavior.
- Stylistic differences: economists are more theory-based and focus more on outcomes,
while psychologists focus on the process.
- Design and procedures: induced value-theory in economics, experimental deception/
manipulation (lying to subjects about elements of the experiment) is not allowed in
economics. Rewards in philosophy are often not salient. Economics uses most of the
time a flat fee or no money at all.
Natural vs. laboratory setting
Natural or field setting Laboratory setting
Real and detailed (high external validity) Simplified economic reality
Given environment that happens to be what Controlled: opportunity to isolate effect and
it is: no repetition, no deliberate to learn causalities step by step
manipulation
Not controlled: allows only for a limited Internal validity high
analysis
Internal validity often limited External validity questionable
Induced-value theory: proper use of
Week 1: introduction and methodology
- Lecture 1a: introduction to experimental economics
Experimental economics is an empirical method, a way to gather data.
The goal is to draw causal inferences, make predictions about new policy measures and to
examine the effect of interventions (treatments).
What do experiments tell us about theory?
Descriptive approach: testing whether actual behavior satisfies the
assumptions about human behavior on which theories are based (most
psychologists).
Prescriptive/normative approach (as if): testing whether the
conclusions/implications of theories offer good(-enough) approximations
of actual behavior, even when its assumptions are not precisely satisfied
(most economists).
Correlation vs causation
- We denote the treatment as a dummy variable, with value 1 if the subject is in the
treatment group, and 0 if the subject is in the control group.
- Y1i = outcome with treatment for person i
- Y0i = outcome without treatment for person i
- However, we cannot observe both outcomes as one person cannot be in the treatment
and control group simultaneously we lack a counterfactual (the value of Y if the
subject did not receive the treatment when he or she was in the treatment, or the other
way around).
- We can overcome this by testing more people and using a control group. These people
need to be somewhat similar to have a trustworthy counterfactual (do not construct
one group with only men and one with only women, etc.).
Economic experiments
- These allow control through randomization.
- This means that the treatment is randomly assigned. This ensures:
, o Control group: some subjects are not exposed to the treatment
o Ceteris paribus comparison: subjects in treatment and control are similar on
average.
- Randomization, therefore, ensures a reliable counterfactual if there are enough
participants.
- The only thing that is different for the treatment compared to the control group is, the
treatment.
- Confounding effects need to be avoided (these are alternative explanations for the
results, for example, by having more differences between the control and treatment
group).
Laboratory experiments
- We focus on laboratory experiments. These are organized, such that everyone knows
that they are participating in an experiment.
- To participate in an experiment, the following things must apply:
o You perform simple decision task
o All information needed is provided at the spot
o Decisions have real consequences: you may earn money
o Individual decisions or games against other participants
o Everything is anonymous
- Lecture 1a: introduction to behavioral economics
Behavior is often irrational according to the neo-classical model. Behavioral economics uses
social, cognitive and emotional factors in understanding the economic definitions of
individuals and institutions. This all has to do with the concept of bounded rationality/ will-
power / self-interest. Psychology is often used in behavioral economics as well.
Behavioral economics can be seen as the field, and experimental economics as the
tool/method.
- Lecture 1b: introduction to behavioral economics (continued)
This is precisely what we do in the lectures. We begin with a known theory (game theory,
individual decision making, social preferences and market experiments), then we test this in
an experiment. Later, we update our theory with the findings of our experiment (if this is
found on a large scale, we do not update a theory if just one experiment found something).
The first principles of neoclassical economics
, - The first principles are rationality and self-interest. However, lately there is lots of
evidence that shows that people are not always self-interested (altruism) or rational
(bounded rationality).
- There are 2 systems of cognition system 1 (intuition) and system 2 (reasoning).
- We often think with our first system, which is fast and uses emotions. Intuitive
thoughts come to mind spontaneously. A key principle is accessibility: the more we
can imagine something to happen, the quicker we will think about this.
- Behavioral economics understands that people make systematic mistakes due to
system 1 and tries to put them in perspective of theories.
- Lecture 1b: methodology
Economics as an experimental discipline
Reasons for an experiment
1. Theory falsification: Do people play a mixed strategy Nash equilibrium in practice?
2. Applicability of competing theories: when is one more successful than the other?
3. Robustness / theory stress tests: when does the theory (not) work?
4. Generate data for policy advice: how do people respond to different institutions?
5. Test-bed mechanism: new institutions that are too complex for theory
6. Discover empirical regularities: in the absence of accurate theory
7. Measure individual characteristics: risk aversion, impatience, etc.
8. Simulation of natural economic processes (rare): macroeconomic experiments
9. Pedagogical purposes: teach students about alternative theories
Economic experiments
An experiment is a controlled economic environmental in which experimental subjects make
decisions that the experimenter records for the purpose of scientific analysis.
A controlled economic environmental are the individual agents together with an institution
through which the agents interact.
Experimental economics is the study of economics with data generated under controlled
conditions, either in the field or in the lab. Experimental economics can also be seen as a
research method for behavioral economics.
It is useless to try to replicate certain things in a lab:
- Field environment
- Precise assumptions of a model (which omits details)
, Economic experiments study simplified economic reality: data can neither mimic the real
world nor perfectly mirror formal models.
There are different types of experiments:
- Tests of behavioral hypotheses/ theory falsification
- Theory stress tests/ sensitivity tests link laboratory to more naturally occurring
markets
- Searching for empirical regularities
Data sources
- Experimental vs. happenstance data: experimental data are deliberately created under
controlled conditions. Happenstance data are a by-product of ongoing uncontrolled
processes.
- Laboratory data vs. field data: laboratory data are gathered in an artificial
environment. Field data are gathered in a naturally occurring environment.
- Costs to gather happenstance and experimental data are high (unless it has been
collected by government in the case of happenstance data).
There are different types of decisions in an experiment:
- Individual
- Strategic, against another participant
Note: experimentalists claim that an economic experiment with proper
incentives (e.g. monetary incentives) is economic reality, because the
decisions have economic effects.
Economics vs. psychology
Both use experiments to gather data, but there are some differences:
- Different interests: economics focus on behavior in an economic context including
restrictions and institutions, while psychologists focus more on individual,
unrestrained, behavior.
- Stylistic differences: economists are more theory-based and focus more on outcomes,
while psychologists focus on the process.
- Design and procedures: induced value-theory in economics, experimental deception/
manipulation (lying to subjects about elements of the experiment) is not allowed in
economics. Rewards in philosophy are often not salient. Economics uses most of the
time a flat fee or no money at all.
Natural vs. laboratory setting
Natural or field setting Laboratory setting
Real and detailed (high external validity) Simplified economic reality
Given environment that happens to be what Controlled: opportunity to isolate effect and
it is: no repetition, no deliberate to learn causalities step by step
manipulation
Not controlled: allows only for a limited Internal validity high
analysis
Internal validity often limited External validity questionable
Induced-value theory: proper use of
