You are on page 1of 12

UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS

EC345 BEHAVIOURAL ECONOMICS, 2022/23

EC345 Behavioural Economics: Theory & Applications

Seminar 1: Introduction and Prospect Theory (with sketch answers)

The answers below are for guidance only and should be treated as such.

1. ‘Behavioural economics should enhance, not replace, traditional economics’. Comment


on this statement.

A good answer will begin by explaining what is behavioural economics and give a brief history (how it
gained momentum around 1950s as it was easier to target standard economics theory. Why there was
a need for behavioural economics and finally give an overview of how it works.

• identify normative assumptions or models that are universally used by economists, such
as Bayesian updating, expected utility, and discounted utility
• identify anomalies—i.e., demonstrate clear violations of the assumption or model, and
painstakingly rule out alternative explanations, such as subjects’ confusion or transactions
costs
• use the anomalies as inspiration to create alternative theories that generalize existing
models
• construct economic models of behaviour using the behavioural assumptions from the
third step, derive fresh implications, and test them

2. Explain the Ellsberg paradox. What assumption is violated here? Is ambiguity aversion a
good explanation for this paradox?

30 balls 60 balls

Gamble balls
Red Black Yellow
A $100 $0 $0

B $0 $100 $0

C $100 $0 $100

D $0 $100 $100

In most cases people will choose A over B and D over C. This is in violation of independence axiom
(a good answer will explain this). Ambiguity aversion can explain the paradox since it is thought
that betting for or against the known information (red ball) is safer than betting for or against the
unknown (black ball). Here, A≻B because the proportion of red balls is known and D≻C because
the proportion of black and yellow balls is known.

1
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

3. Why is the value function in Prospect theory S-shaped?


The value function is S-shaped for three main principles.
i. reference dependence: The value function is defined on
deviations from a reference point
ii. Loss aversion: losses loom larger than gains. Value function is
normally concave for gains (implying risk aversion), commonly
convex for losses (risk seeking) and is generally steeper for losses
than for gains.
iii. Diminishing sensitivity: The principle of diminishing sensitivity
implies that the effect of the change diminishes with distance to the reference point. A
difference between a gain/loss of 100 to 200 appears larger than a difference between 1100
and 1200.

4. Explain the disposition effect using the paper by Odean (1998). Why the disposition effect
is usually not observed during the month of December?
The disposition effect is the tendency of investors to hold losing investments too long and sell winning
investments too soon. Odean (1998) analyses trading records for 10,000 accounts at a large discount
brokerage house and finds that the investors demonstrate a strong preference for realizing winners
rather than losers.
Odean (1998) illustrates the disposition effect with the help of the following example: Suppose, for
example, that an investor has five stocks in his portfolio, A, B, C, D, and E. A and B are worth more
than he paid for them; C, D, and E are worth less. Another investor has three stocks F, G, and H in her
portfolio. F and G are worth more than she paid for them; H is worth less. On a particular day the first
investor sells shares of A and of C. The next day the other investor sells shares of F. The sales of A and
F are counted as realized gains. The sale of C is a realized loss. Since B and G could have been sold for
a profit but weren’t, they are counted as paper gains. D, E, and H are paper losses. So for these two
investors over these two days, 2 realized gains, 1 realized loss, 2 paper gains, and 3 paper losses are
counted. Realized gains, paper gains, realized losses, and paper losses are summed for each account
and across accounts.
Then two ratios are calculated:
𝑟𝑒𝑎𝑙𝑖𝑧𝑒𝑑 𝑔𝑎𝑖𝑛𝑠 1
𝑃𝐺𝑅 = =
𝑟𝑒𝑎𝑙𝑖𝑧𝑒𝑑 𝑔𝑎𝑖𝑛𝑠 + 𝑝𝑎𝑝𝑒𝑟 𝑔𝑎𝑖𝑛𝑠 2
𝑟𝑒𝑎𝑙𝑖𝑧𝑒𝑑 𝑙𝑜𝑠𝑠𝑒𝑠 1
𝑃𝐿𝑅 = =
𝑟𝑒𝑎𝑙𝑖𝑧𝑒𝑑 𝑙𝑜𝑠𝑠𝑒𝑠 + 𝑝𝑎𝑝𝑒𝑟 𝑙𝑜𝑠𝑠𝑒𝑠 4

2
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

A large difference in the proportion of gains realized (PGR) and the proportion of losses realized (PLR)
indicates that investors are more willing to realize either gains or losses. Any test of the disposition
effect is a joint test of the hypothesis that people sell gains more readily than losses
The disposition effect is usually not observed during the month of December as it is the month in
which tax is calculated, so individuals have incentive to realize losses in order to declare less profit and
pay less tax.

3
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/2023

EC345 Behavioural Economics: Theory & Applications

Seminar 2: Intertemporal Choices and Neuroeconomics (with sketch answers)

The answers below are for guidance only and should be treated as such.

1. Compare between the Habit formation models and Utility from Anticipation model. Which
model can better explain addictive behaviour?

The main idea behind Habit Formation model is that individuals form some consumption habit
based on their past consumption and this has an effect on current consumption. Although habit
formation is often said to induce a preference for an increasing consumption profile, it can, under
some circumstances, lead a person to prefer a decreasing consumption profile. The direction of
the effect depends on things such as how much one has already consumed and, perhaps most
importantly, whether current consumption increases or decreases future utility. There are
different representations of the habit formation models; a very basic idea is that period 
instantaneous utility comprises of current consumption C and a state variable Z. This Z is
considered as a state variable which enters into the utility function. Also, Z can be taken to mean
all previous period consumption or some periods.

A model of Utility from Anticipation on the other hand says that our consumption utility in time
period t is affected by consumption in a future period t+k. The basic idea is that you get utility
from imagining an experience you will have in the future. It is possible to get either negative
(disutility) or positive utility from anticipation, depending on the nature of the experience. For
example, you can get positive utility now from anticipation of a holiday in the future. On the other
hand, you may get disutility (negative utility) from anticipation of an exam. Utility derived from
anticipation must be small relative to the actual utility, but it can still have some strong effect.

Habit formation models are more suited and have been applied extensively to explain addiction
or utility from addictive goods.

1
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/2023

2. “People are present biased regardless of their wealth”. Comment on this statement in reference
to the findings from Tanaka, T., Camerer, C. F., & Nguyen, Q. (2010) and your understanding of
present bias.

Present bias is when an individual prefers a smaller present reward to a larger later reward but
shows preference reversal when both rewards are equally delayed. We can demonstrate present
bias with a simple example. When asked would you prefer to get: (a)£110 in 31 days or (b)£100 in
30 days? And would you prefer to get: (c)£100 today or (d)£110 tomorrow? Often people choose
(a) and (c). This violates the key assumption that time is discounted at constant rate (delaying or
accelerating two dated outcomes by a common amount should not change preferences between
the outcomes). This behaviour can be explained by the presence of a present bias which implies
that we are very impatient in the short run.

In the paper by Tanaka, T., Camerer, C. F., & Nguyen, Q. (2010), the authors aim to empirically test
a general model of intertemporal choices to test exponential, hyperbolic, quasi-hyperbolic
discounting, and a more general form of discounting. Their hypothesis is motivated by the finding
in literature that wealthier people are more patient and have a lower discount rate as measured
by the exponential discounting model. In the paper they estimate the three factors: conventional
time discounting (r), present bias (β), and hyperbolicity (θ) of the discount function. The
parameters are estimated based on the choices made by subjects in a series of intertemporal
binary choices. Specifically, subjects make 75 choices between smaller rewards delivered today
and larger rewards delivered at specified times in the future as follows: Option A: receive x dong
today; or Option B: receive y dong in t days. The reward x varies between 30,000 to 300,000 and
the time delay t varies between three days and three months. One of the main empirical results
from the paper states that People are present biased regardless of their wealth. This statement is
based on the finding in the paper where the authors estimate the coefficient of β to be 0.644. This
indicates that subjects are present biased. This value of β (degree of present bias) is comparable
to estimates from a variety of other studies. In a regression the authors find that none of the
income variables (Income, Relative income or Mean village income) explains individual difference
in present bias (β).

2
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/2023

3. In the paper titled ‘Oxytocin increases trust in humans’, Kosfeld et al. (2005) investigate the
effect of Oxytocin on prosocial behaviour. Describe their study and results in details. Why did
the authors conduct the ‘risk experiment’ along with the trust experiment? (5)

Oxytocin is a hormone, which induces labour in human and nonhuman mammals, during lactation
of young animals and during mating. Oxytocin is conjectured to play a key role in different social
behaviours.

Design: Conduct a placebo-controlled hormone study that isolates the specific impact. This
provides causal information about the impact of the hormone. In the study, subjects play either
the role of an investor or a trustee.
• Both receive an initial endowment of 12 monetary units (MU)
• investor can send 0, 4, 8 or 12MU to the trustee
• experimenter triples each MU the investor transfers
• the trustee has the option of sending any amount between zero and his total amount available
back to the investor
• E.g., if the investor has sent 12MU, the trustee possesses 48MU (12MU own endowment +
36MU tripled transfer)
• hypothesis is that oxytocin increases the trusting behaviour of investors- investors in the
oxytocin group will show higher money transfers than those in the placebo group
Results:
• 45% in the oxytocin group showed the maximal trust level, (only 21% in the placebo group
showed maximal trust )
• 21% of the subjects in the oxytocin group had a trust level below 8 monetary units (MU), but
45% of the subjects in the control group showed such low levels of trust
• The investors’ average transfer is 17% higher in the oxytocin group

The ‘risk experiment’ is conducted to isolate the risk element coming from the social interaction
in the study. Risk experiment constitutes a powerful control for the effects of oxytocin on trusting
behaviour. Investors’ risk in the risk experiment is not generated through a social interaction, all
the indirect effects of oxytocin on the state of a subject, such as possible effects on mood or
calmness, would be present in both the trust and the risk experiment. The study concludes that,
the effect of oxytocin on investor behaviour is driven by the hormone and not risk aversion.

3
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/2023

4. What contribution has Neuroeconomics made to our understanding of present bias? Give
example of empirical evidence that supports your claim (for example, you can base your answer
on the paper by McClure et. al. (2004)).

Neuroeconomics contribution: Neuroeconomics explains that the rational/analytical part of the


brain is responsible for the usual exponential discounting, the affective part of the brain is
responsible for the present bias combining the two together results in a hyperbolic (or quasi
hyperbolic) discount rate.

Methodology + Results from McClure paper

In this paper the authors hypothesize that the discrepancy between short-run and long-run
preferences reflects the differential activation of distinguishable neural systems. Their key
hypothesis is that the pattern of behaviour can be summarised by the two parameters: β (which
reflects the special weight placed on outcomes that are immediate) and δ (which reflects a more
consistent weighting of time periods). They find that short-run impatience is driven by the limbic
system, which responds preferentially to immediate rewards and is less sensitive to the value of
future rewards-(this is termed as β). And long-run patience is mediated by the lateral prefrontal
cortex and associated structures, which are able to evaluate trade-offs between abstract rewards,
including rewards in the more distant future- (this is termed as δ).

Explain β – δ terms

It is possible to link the findings from McClure et. al. (2004) directly with the theory of hyperbolic
discounting (or quasi-hyperbolic discounting model) of time preference. The quasi-hyperbolic
time-discounting function—sometimes referred to as β - δ model, posits that the present
discounted value of a reward of value u received at delay t is equal to u for t=0 and to βδu for t>0;
where 0<β≤1 and δ≤1. The β parameter represents the special value placed on immediate rewards
relative to rewards received at any other point in time. When β<1, all future rewards are uniformly
down-weighted relative to immediate rewards. The δ parameter is simply the discount rate in the
standard exponential formula, which treats a given delay equivalently regardless of when it
occurs.

4
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

EC345 Behavioural Economics: Theory & Applications

Sketch Solutions for Seminar topics: Subjective Well-being and Utility, and Subjective

Well-being and the Economy

1) How might we measure well-being? What problems are there with reported measures of
subjective well-being, and what gives us some reassurance that the answers are not
merely random?

Solution:

Firstly, there are objective and subjective measures of well-being. Objective measures are
given on slide 8 of the SWB and U lecture notes, and are worth mentioning at the beginning
of any discussion, but our focus is on the subjective measures used in behavioural economics.

Self-reports of subjective well-being can be via survey sources, or via momentary evaluation
(e.g. experience sampling), or via the Day Reconstruction Method. Questions can focus on
immediate experience, or evaluations over some time period (up to and including global
evaluations of life as a whole), and questions can be framed in different ways - particularly a
relatively affective/hedonic manner (what Kahneman and Deaton call emotional well-being)
or evaluative/cognitive manner (measures of well-being as life evaluation). See slides 10 and
11. The measures are censored reports of what is assumed to be some aspect of underlying,
latent experience quality, via a reporting function (slide 13).

For measurement issues, see slide 15 – interpersonal comparability (interpretation of


question, scale), the bounded nature of the scale, sensitivity to context, distorted reporting,
and for global evaluations, errors in recollection and aggregation. There may be other
problems around the reporting function (particularly, in inferring curvature in the underlying
experience quality construct; or if the reporting function shifts over time). There are also
pervasive endogeneity issues in inferences about causality (slide 33).

If we were only looking at 1 or 2 responses, we would have good reason to be extremely


cautious. Fortunately, as we look over large datasets, we can have some degree of
reassurance to the extent that errors will average out (unless systematic, of course). More
importantly, subjective well-being responses are strongly correlated with a large number of
other measures of well-being (i.e. exhibit validity), such as physiological and mental health
measures, MRI brain activity and the evaluations of others. SWB measures also exhibit
reasonable reliability over time (slide 16). State-level life satisfaction also correlates strongly
with objective data on quality of life for US states (Oswald and Wu (2010), slides 17 & 18),
and SWB regressions display common patterns across a wide variety of datasets, exhibiting a
U shape relationship over age for example (see in particular slides 26, 28, 30 and 31 - Note:
the relationships with demographic variables mentioned on slide 32 are far less consistent
across datasets i.e. more setting/context-dependent).

Many of the measurement issues are discussed in Blanchflower and Oswald (2004) Sections 2
& 4 (or – not on the reading list – Krueger and Schkade (2008)).
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

2) Discuss whether the results of Benjamin et al (2012) are consistent with the idea that

people try to maximise subjective wellbeing in their choices.

Solution:

Benjamin et al (2012) explore whether people choose what they think would maximise
their SWB via 13 hypothetical decision scenarios, each with two alternatives, exploring
trade-offs between factors considered important in well-being (income, housing, time with
friends etc). Respondents are asked which alternative they would choose, and under which
they anticipate greater SWB, and the rankings are compared. The rankings coincide in
about 83% of cases, still leaving a substantial discrepancy – although it might be noted
that (i) this is on the basis of the SWB measures chosen – other, broader evaluative
measures (e.g. such as Cantril’s ladder) may reduce this, and (ii) here they are very
specifically looking at cases where there are potential trade-offs – in choice in general the
discrepancy is again likely to be smaller - although in their conclusion the authors point
out that it is in precisely in cases where there are such potential trade-offs between SWB
and other considerations, which are often important life decisions, that we might be
particularly cautious of using SWB as a proxy for utility/preferences (in the sense of
idealised choice utility - assuming this is revealed in these hypothetical choices).

Choice and happiness regressions are also undertaken to examine the extent to which
other factors predict choice once anticipated SWB is taken into account, finding significant
effects for things like family happiness, control over life, status, comfort and purpose (i.e.
systematic discrepancies between choice and anticipated SWB). However, it is notable that
anticipated SWB is by a very considerable margin the strongest predictor of choice, and
that the incremental R^2 (the increase in the variation in choice explained by additional
factors, i.e. the difference in R^2 between the full multivariate regression and a univariate
regression of choice on just anticipated SWB) is just 3% over all scenarios.

Overall the results suggest that people do not seek to maximise SWB exclusively (at least as
measured here), but that SWB is a uniquely important argument in the utility function.
Benjamin et al (2012) conclude that well-informed agents may still be willing to trade off
SWB for other things they care about, making SWB and preferences (or idealised choice
utility, under the assumption that these are revealed in hypothetical choices) distinct.
Students may offer alternative interpretations if they are reasonable, and recognise the
arguments made in the paper.
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

3) What is the empirical relationship between income and subjective well-being in the
United States? Include reference to the results of Deaton and Kahneman (2010) and
Luttmer (2005).

Solution:

There are of course a large number of papers looking at the relationship between income and
subjective well-being, but to narrow the discussion the question is deliberately focused on the
United States, and specifically the main results of Deaton and Kahneman 2010, and Luttmer
2005, both of which are included in the lecture notes (slides 6-8 and 14-16 respectively, in
Econ. of Happiness 2: SWB and the Economy).

The relationship between income and subjective well-being more generally is covered in
slides 5 – 16, and there is also the overarching issue of the Easterlin paradox of course (the
last section in the first set of Economics of Happiness slides).

There is a fairly modest but statistically significant positive relationship between SWB and
income in the cross section, with a gradient that is shallower in developed countries like the
US than in developing countries. Like many factors in subjective well-being, changes in
income are subject to adaptation, as well as social comparison effects. Although it appears
to be a reasonably important factor in subjective well-being, there are other factors
(particularly health and social relationships) that often exhibit stronger correlations, which is
important context for the relationship. Unemployment reduces subjective well-being by
considerably more than simply the effect of an equivalent loss of income.

Regarding the Deaton and Kahneman results (slides 6-8) for a sample of 450,000 people in
the US from 2008/09, poverty appears to exacerbate the effect of adverse circumstances,
such as loneliness or health problems. Both affective and evaluative measures of well-being
are (modestly) increasing in income, but in affective measures only up to a level of around
$75k (for this sample i.e. in the context of the US, 2008/09). This may be because of a non-
trivial impact on other well-being factors (such as social factors), because high incomes may
reduce our ability to enjoy the small things in life, and/or because of adaptation and social
comparison.

When Deaton and Kahneman look at the relative size of the measured relationships by
normalising coefficients with respect to the coefficient on ‘high income’ (those earning over
>$4000 per month, 58% of the sample), they again find a notable difference between
affective and evaluative measures. On the evaluative measure the high income relationship is
individually stronger than any other considered (although when taken together, other factors
dominate), but on affective measures even individually other measures (such as important
health and social factors) often exhibit stronger relationships.

The Relative Income Effect literature asserts that most of the impact of income, above a
certain lower threshold, may be driven by relative rather than absolute income, so that if
everyone increases their income the overall subjective well-being effect is greatly muted (as
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

in the Easterlin paradox). Luttmer (2005) examines this by using 2 waves from the National
Survey of Families and Households together with estimated earnings by locality (Public Use
Microdata Areas), finding a negative relationship between PUMA earnings and subjective
well-being measures that is highly statistically significant and robust to changes in
specification.

The variety of satisfaction measures used is also suggestive regarding the possible
mechanism. PUMA earnings are positively related to satisfaction with one’s place of
residence, suggesting the relationship is not simply a reflection of higher local costs. PUMA
earnings are however also negatively related to leisure time, satisfaction with leisure and
satisfaction with friends. Luttmer suggests people may be working more and giving up
leisure, potentially to try and mimic higher local income and consumption norms. He also
suggests that this would be consistent with systematic mispredictions of utility from goods
yielding extrinsic satisfaction, as adaptation and social comparison effects are neglected,
potentially due to projection bias (but this is of course speculative).

To contain the parameters of the discussion it is advisable to focus on these main results of
Kahneman and Deaton, and Luttmer. The lecture slides (6-8, and 14-16) are entirely sufficient
for these main results, but for further reading, the Kahneman and Deaton paper is quite
short, and the Luttmer results are discussed in section 4 of that paper. You could also discuss
the results in Blanchflower and Oswald (2004).

4) How does the curvature of the comparison term in a simple additive comparisons model
such as Clark and Oswald (1998) affect the optimal response to a change in reference?
Illustrate with the case of a positive exogenous shock to the income (and consumption)
of the social comparison reference.

Solution:
Clark and Oswald (1998) present a simple additive comparisons model of rational behaviour where
individuals also care about, and respond to, the reference actions of others, generating possible
reasons to imitate (or deviate from) others without simply assuming a need to conform (/deviate).
Whether the optimal response to a change in reference is to follow the change or to deviate depends
in the model on the curvature of the comparison term, with concavity generating optimal following
and convexity deviance. This is covered on slides 24-26 of the lecture notes (slide 27 relates to a more
slightly more complicated simulation from the paper that is beyond this particular question).

As an illustrative example, if the reference income (and associated consumption) increases due to a
positive exogenous shock, the individual’s utility decreases and they slide backwards down the curve
representing the comparison component of the model. If this comparison term is concave, the
marginal utility of income from this comparison term therefore increases, and so the optimal
response to this increase in reference action is to also increase self-action i.e. increase income (and
UNIVERSITY OF WARWICK, DEPARTMENT OF ECONOMICS
EC345 BEHAVIOURAL ECONOMICS, 2022/23

consumption, and decrease leisure), thus exhibiting following behaviour. Reference to a graph is
helpful:

If the comparison term is convex, the opposite will occur: the marginal utility of income
decreases, and the optimal response is to reduce income and associated consumption, to
deviate from the positive change in the comparison reference.

You might also mention more details of the model:

Since this models a negative externality, the emergent levels of income and consumption in
equilibrium (in the absence of cooperative mechanisms/institutions) are in general socially
suboptimal (Pareto inefficient).

You might also like