Week 1. Lecture 1- Introduction to experiments
Honesty in managerial reporting
The experiments reveal considerable honesty.
the reason that individuals reported less honestly under the Modified Trust Contract is that
they wanted to achieve a desired distribution of the total surplus between themselves and
the firm.
Their result suggests that individuals’ willingness to lie is endogenous, in other words, the
extent of honesty depends on the type of contract individuals receive.
Revelation Principle= firms can always get honest reporting if they pay enough for it.
Conventional agency model= individuals experience no disutility from lying.
Threshold model = because individuals experience a small fixed disutility from lying, they
report honestly for all payoffs less than their personal disutility threshold and report to
maximize wealth for all payoffs at or above the threshold.
Types model= agents are one of two types: ‘‘ethical’’ (fully honest) or ‘‘economic’’ (willing to
tell any lie necessary to maximize wealth).
Trade-off model= individuals will trade off ethical behavior with the monetary payoff for
unethical behavior. In our setting, this means individuals will lie more as the payoff to lying
increases.
Method:
- They did three experiments
o Setting 1→ the more the manager lies, the more he earns. The less he lies,
the less he earns.
o Setting 2→ identical to Experiment 1 except that the dollar payoffs for lying
were increased by a factor of 5.
o Scenario 3→ Experiment 3 introduces the Modified Trust Contract (MTC),
which is the Trust Contract from Experiment 1 with a production hurdle i.e., a
cost above which there is no production
Class notes on the paper
Research question: how much do people lie to benefit themselves?
Why do we care about this in accounting? → employees often report on their private
information (accounting performance, budgetary needs, etc.)
The study examines whether managers’ preferences include honesty as well as wealth, and
whether managers’ preferences for honesty increase the value of communication.
,Standard economic thinking: people will lie to the max when they can get away with it. →
these two studies: is this really true?
Honesty in managerial reporting (Evans et al.)
Do people have any preferences for being honest? If so, do these preferences matter
enough for us to care about (i.e., materiality)?
They measure honesty by looking at cost reports:
- Local managers know more about costs→ but managers can benefit by inflating cost
reports (i.e., being dishonest)→ if people have intrinsic preferences for honesty,
then they will not inflate to the max.
But can we just look at cost reports?
- How do we know whether a given cost report is honest of dishonest?
- How do we know whether honest reports are due to honest preferences?
How can we measure honesty (measurement validity)?
- By using an artificial setting designed precisely to isolate honest preferences
Experimental task:
- MBA student participants acting as managers. The managers report their production
costs to HQ.
- Actual cost is random (between 4 and 6 lira)→ managers learn actual cost before
reporting. HQ only knows that the cost is between 4 and 6
- Managers earn money by reporting high costs
How do we know whether a given cost report is honest or dishonest?
- Dishonesty= reported cost – actual costs→ the researchers can calculate this
because they know the actual costs.
Why would managers not request the maximum budget?
- To avoid punishment or appear honest? HQ doesn’t know actual cost; anonymous
interactions
- Risk buffer? Participants know the actual cost when submitting their budget
- To ensure HQ accepts request? HQ has to accept any budget request
- To project competence? Actual costs don’t depend on skill or effort (it’s random)
- Small stakes? Increasing payoffs doesn’t change much (experiment 2)
- Innate aversion to lying? … is this all that’s left?
, The Effect of Honesty and Superior Authority (Ranking. Schwartz, and Young)
Slack is dishonesty→ slack= budget request – actual quest→ closely replicates Evans et al.
Measurement validity
- Evans et al. → because people did not lie to the max, they must have preferences for
honesty
- RSY→ not lying to the max could reflect things other than honesty→ so RSY are
skeptical about the second question of measurement validity
RSY questioned external validity too→ claim: Evans et al. miss a key aspect of reality
- Giving the agent final authority→ ethical dilemma
- Giving the principal final authority→ strategic dilemma
Predictive validity framework
- Conceptual model→ the research question we
want to get an answer to/does x cause y
Internal validity→ to which extent does independent
variable cause dependent variable.
Measurement validity→ are the survey question capture
in a valid way the leadership style
External validity→ can we generalize what we have
learned from the research to the other settings.
Honesty in managerial reporting
The experiments reveal considerable honesty.
the reason that individuals reported less honestly under the Modified Trust Contract is that
they wanted to achieve a desired distribution of the total surplus between themselves and
the firm.
Their result suggests that individuals’ willingness to lie is endogenous, in other words, the
extent of honesty depends on the type of contract individuals receive.
Revelation Principle= firms can always get honest reporting if they pay enough for it.
Conventional agency model= individuals experience no disutility from lying.
Threshold model = because individuals experience a small fixed disutility from lying, they
report honestly for all payoffs less than their personal disutility threshold and report to
maximize wealth for all payoffs at or above the threshold.
Types model= agents are one of two types: ‘‘ethical’’ (fully honest) or ‘‘economic’’ (willing to
tell any lie necessary to maximize wealth).
Trade-off model= individuals will trade off ethical behavior with the monetary payoff for
unethical behavior. In our setting, this means individuals will lie more as the payoff to lying
increases.
Method:
- They did three experiments
o Setting 1→ the more the manager lies, the more he earns. The less he lies,
the less he earns.
o Setting 2→ identical to Experiment 1 except that the dollar payoffs for lying
were increased by a factor of 5.
o Scenario 3→ Experiment 3 introduces the Modified Trust Contract (MTC),
which is the Trust Contract from Experiment 1 with a production hurdle i.e., a
cost above which there is no production
Class notes on the paper
Research question: how much do people lie to benefit themselves?
Why do we care about this in accounting? → employees often report on their private
information (accounting performance, budgetary needs, etc.)
The study examines whether managers’ preferences include honesty as well as wealth, and
whether managers’ preferences for honesty increase the value of communication.
,Standard economic thinking: people will lie to the max when they can get away with it. →
these two studies: is this really true?
Honesty in managerial reporting (Evans et al.)
Do people have any preferences for being honest? If so, do these preferences matter
enough for us to care about (i.e., materiality)?
They measure honesty by looking at cost reports:
- Local managers know more about costs→ but managers can benefit by inflating cost
reports (i.e., being dishonest)→ if people have intrinsic preferences for honesty,
then they will not inflate to the max.
But can we just look at cost reports?
- How do we know whether a given cost report is honest of dishonest?
- How do we know whether honest reports are due to honest preferences?
How can we measure honesty (measurement validity)?
- By using an artificial setting designed precisely to isolate honest preferences
Experimental task:
- MBA student participants acting as managers. The managers report their production
costs to HQ.
- Actual cost is random (between 4 and 6 lira)→ managers learn actual cost before
reporting. HQ only knows that the cost is between 4 and 6
- Managers earn money by reporting high costs
How do we know whether a given cost report is honest or dishonest?
- Dishonesty= reported cost – actual costs→ the researchers can calculate this
because they know the actual costs.
Why would managers not request the maximum budget?
- To avoid punishment or appear honest? HQ doesn’t know actual cost; anonymous
interactions
- Risk buffer? Participants know the actual cost when submitting their budget
- To ensure HQ accepts request? HQ has to accept any budget request
- To project competence? Actual costs don’t depend on skill or effort (it’s random)
- Small stakes? Increasing payoffs doesn’t change much (experiment 2)
- Innate aversion to lying? … is this all that’s left?
, The Effect of Honesty and Superior Authority (Ranking. Schwartz, and Young)
Slack is dishonesty→ slack= budget request – actual quest→ closely replicates Evans et al.
Measurement validity
- Evans et al. → because people did not lie to the max, they must have preferences for
honesty
- RSY→ not lying to the max could reflect things other than honesty→ so RSY are
skeptical about the second question of measurement validity
RSY questioned external validity too→ claim: Evans et al. miss a key aspect of reality
- Giving the agent final authority→ ethical dilemma
- Giving the principal final authority→ strategic dilemma
Predictive validity framework
- Conceptual model→ the research question we
want to get an answer to/does x cause y
Internal validity→ to which extent does independent
variable cause dependent variable.
Measurement validity→ are the survey question capture
in a valid way the leadership style
External validity→ can we generalize what we have
learned from the research to the other settings.
