Notesale: Turn your study into money

Document Preview

Extracts from the notes are below, to see the PDF you'll receive please use the links above

---

INTERTEMPORAL DECISION MAKING
Idea - Paper - Questions
Topic I: Time preferences
Q
...
What is discount function? in the concept of exponential discounting & quasi-hyperpolic discounting
- Discount function ​reflects impatience​: we discount for future consumption relative to now
- Discount function assumed to b​e DECREASING in t​: a unit increase in consumption today worth
more to you today than a unit increase in consumption tomorrow
- Discrete: D(s) < D(t) for s>t
- Continuous: D’(t) < 0 : decreasing function
Q
...
What is the implication of EDU ? What is the evidence that it doesn't work?
- Implications of EDU depends on the fact that in EDU, discount rate is constant

Stationarity:

−ρt

ρ
Discount rate is
constant

-

discount (difference between utility) must be equal
The preference between 2 consumption adjustments depends ONLY on the time distance separating
the two adjustment -> ​only the time difference matters, not the benchmark of time
Prove: The function of EDU

But: Empirical evidence
- Thaler (1981): one apple today > 2 apples tmr ; one apple 100 days < one apple 101 days

Time consistency:

-

the ranking of outcome-time pairs does not change whether the individual is asked at s or t
no incentive to deviate ex-post, ex-ante, no dynamic preference reversals
Prove that the optimal plan under EDU is time consistent?

But: Empirical evidence:

-

Leeuwen (1998); people made advance choices and immediate choices differently for type of snacks
(healthy vs not)
Loewensteen (1999: people made advance choices and immediate choice differently (lowbrow and
high brow)
Gym membership

Q
...
Behavioural - (Quasi) Hyperbolic discount function - model of multiple selves (across time)
Phelps Pollak Laibson ----> DISCRETE-TIME DISCOUNT FUNTION, QUASI-HYPERBOLIC DISCOUNT
FUNCTION

-

-

-

Capture the fact that discount rate also discounts over time: higher discounting rate between the
present and the future than between any future time periods -> ​TIME INCONSISTENCY - present
bias
Hyperbolic (Continuous):
- Discount rate is a decreasing function in time
- very very sharp between now and next period, slowly in the future
- faster to reach lowest point
Quasi-hyperbolic (Discrete) - Laibson 1997 :
- introduce β ε (0,1) to capture​ present bias​, all the future periods’ discount function is
multiplied by β => the discount function discounts over time, ​the discount rate is not
constant but also decreasing function in time -> compare MRS
- discount rate is sharp between now and the next period then declines as a hyperbolic

Q
...
Welfare of TC, sophisticated and naive
- Would these change in behaviour make themselves worse off or better off?
- use benchmark of a self 0 who doesn't need to take any decision in period 0 and weights all future
periods equally?
Q
...

- If workers are sophisticated they​ may demand the dominated work contract​ to make
their future selves work harder
...


Q
...
3 months between last entrance and contract determination
- Monthly contract more likely to stay enrolled (stick to the status quo - auto renew)

Ariely - Deadlines and procrastination (2002002) -> CDs and effectiveness (prob)
STUDY 1:
- 2 groups: evenly-spaced deadlines or set deadlines themselves
- People are willing to self-imposed deadlines at costs, they want to control themselves
- Deadlines are earlier for each paper than the last possible deadlines
- Only 32% are set for final week, 12% are set for last date
-

Effectiveness: self-control has lower grades -> people do not control deadlines optimally

STUDY 2:
- 3 groups: even, end, self-imposed and there is PENALTY for delay

-

Effectiveness: self-imposed better than imposed end, but worse than imposed evenly

Ashraf, Karlan and Yin, 2006, Savings held at banks -> Effective, ppl need it are hyperbolic
- 3 groups: Commitment treatment (SEED) , Marketing treatment (savings but not SEED), control
- Baseline survey: ​28% are hyperbolic:​ against stationary -> prefer the immediate reward now,
prefer the higher but more distant is distance time frame

-

-

Demand for commitment:
- When offer, 28% people do want SEED -> ppl aware and want help
Effectiveness of commitment device on saving
- Compared with control, save $235 more in 6 months
- Compared with control, 10
...
, Niederle, M
...
(2015), Working over
time: Dynamic inconsistency in real eort tasks, Quarterly Journal of Economics, 130(3), 1067-1115
...

Second, in choices over effort, we find substantial present bias
...

- In the distance time frame, more 9% work for the first work date
- Initial allocation: Week 1 decides for week 2 & week 3
- Later allocation: Week 2 decides for week 2 & week 3
Third, we study commitment demand, documenting that at price 0 roughly 60% of subjects prefer
commitment to flexibility
...
​(Sophisticated demands commitment devices)

Experimental evidence:
The Ultimatum game: shows that people might incorporate fairness of themselves and of others (but
it could be because of strategic)
1 - take it or leave it offer and 2 accept (split) vs not (nothing for both)
Econ theory: 1: offer x =0 , 2 accept any x > = 0 (all offers)
Beha: many 1 offer more than 0 (50% offer equal split), 1/3 of 2 accept all offers

The Dictator game: shows fairness (not just strategic to gain some money)
- 1 - divide the pie for 2 MUST accept
Econ: 1 offer 0
Beha: offer less than Ultimatum game (dominate position - strategic in terms of UG) but still positive
(fairness)
Meta-analysis: average giving rate : 28% ​, equal split (fair) : 17% and nothing split: 36%
The Public Good game: shows that we take into account other utilities when calculate ours
- everyone put x in and then 2x gonna be divide between everyone
- if every1 put in -> best allocation but every1 wanna free ride
Econ: contribute = 0 because selfish, utility max by individual
- with punish
Econ: contribute = 0 because punish is costly
Findings: small and global scale

Fehr and Gächter (2000)

Decline contributions toward free-riding when
played many times

Herrmann, Thöni and Gächter (2008)

Varies in rate of contribution between DIFFERENT
COUNTRIES, but in general decline when played
many times

Fehr and Gächter (2000): introduce opportunity
to punish - costly for the punisher and reduce
payoff for the person being punished
Econ: no change, contribute = 0 because punish
is costly

Average contribution go toward full contribution
when played many times

Herrmann, Thöni and Gächter (2008):

Higher the level of punishment, lower average
contribution for all periods (replay)

-> Without punishment, people feel unfair if they contribute and others don't -> decline contribution towards
free riding
...

The Trust Game: People do incorporate trust
- A give B x; B get backs 3x; B gives A m
Econ: x = 0, m = 0
Findings:
Bellemare and Kröger (2007)

Johnson and Mislin (2011) - meta analysis

heterogeneity in both investment and returns ( but
not 0)

...
Theory of social preferences: All fairness models modify the utility function so that the utility an individual
derives does not only depend on her own payoff
- Outcome-based model (doesn't matter how you get there): utility depends on own payoff and others
payoff
- Intention-based model: utility depends on intentions as well
Q
...
5: he will offer M/2
- B = 0
...
5: he will offer 0
=> Consistent with empirical finding that some individuals give 50% of their endowment, while other
individuals give 0 or some value in-between (but no individuals give more than 50%)
Prediction: Ultimatum game
Please read the notes on a4

Q
...
, 2008: can choose to punish at my cost if
player A take away money from me

when I know that actions based on intentions, I
would likely to punish more
Intentions-treatment: A is a human decision
No intentions-treament : machine

Q
...
pecuniary/money concerns): people give moeny
value to late arrivals
- But sometimes, extrinsic incentives are needed to control and encourage motivations: increase min
working hour up to a certain threshold might not bring negative effect anymore
...
Incomplete contract: no price for being late

...
Small fine -> late not important -> still very late

...
Trust condition:


...



...
Diistrust by the investor, manifested by the
imposition of the fine, crowds out reciprocity
('distrust aversion') -> when u don't trust me &
impose a fine on me, i will not transfer back

Fehr and Rockenback (2006): impose control
(locked doors, minimum effort…)
Restrict/Control: locked door, min effort
Don't restrict/Trust: incomplete contract, open
doors

Key findings

...
TRUST IS NOT ALWAYS BETTER THAN CONTROL

...
on
- frequency of different types: ​selfish,
inequality averse, distrust averse
- strength of incentives/degree of control:
min effort = 1,5,10 (if x control = 20, control
actually better)
- Ultimate management goal: develop
contracts that motivate opportunistic
agents (lazy ones, free riders) without
de-motivating intrinsically motivated ones
(don't underestimate them)

There are different types of people (selfish, distrust
aversion, inequality aversion)

Q
...
What are the theories of SOCIAL preferences
1
...
Intentions based model: distrust taken into account
- Intentions: kind or unkind (Falk and Fischbacher, 2006)
-> utility of an individual does not only depends on his/her payoff

DECISION MAKING UNDER RISK AND UNCERTAINTY
-

Risk​: The probability distribution of outcomes is known objectively (1/6 roll dice value 6)
Uncertainty:​ The probability distribution of outcomes is not known ( might rain, might not rain)

DECISION UNDER RISK
1
...
linear in probabilities

...
)
2
...

Therefore, if risk aversion over small stakes ->
super high risk aversion over large stakes

3
...
Endowment effect/Framing effect: Support Prospect theory
Endowment effect: mug - reference level (loss
- Mug experiment: Kahneman, Knetsch
and Thaler, 1999: ​ref level (own vs not
and gain)
own) determines the outcome (WTA = 2x
- people assign more value to things merely
WTP)
because they own them
=> Individuals seem to treat the endowed mugs as
- can be explained with loss aversion:​ losing
part of their reference levels whereas individuals
something you own weighs heavier than without mugs consider not having a mug as their
gaining something you do not own
reference point
- Loss aversion can give rise to a status quo
bias - giving something for free and then
threatening to withdraw it is more effectiv
than promising a reward

Framing effect: Asian disease (loss aversion)
- systematic reversal in preference
depending on whether the problem is
presented in terms of gains or in terms of
losses
Framing effect: Gamble option
- choices between sure option (lose 20 or
keep 30) and gamble (lose all keep all)
- when presented in terms of loss, ppl are
loss aversion and would gamble

-

Asian disease (Kahneman et al)
Lives saved: risk averse -> sure option
Death: risk loving -> gamble
Dù gì cũng mất rùi thì mất nốt :<

-

gamble choose more in LOSS (loss aversion
-> don’t like losing)
Heterogeneity in framing effects among
people (% increase of gambling)
framing and brain activity: ACC opponency of rational and emotional

-

5
...
5x
- Treatment H mỗi lần invest mỗi lần biết; Treatment L: invest 3 lần ko biết, cuối 3 periods biết 1 lần

Results
- Intuition: inform less -> anticipate less loss when they invest
- Treatment L then hence invest more (because treatment H has higher chance of seeing losses and ppl
are loss aversion)
Others: medal, cab driver in NY ( evaluate earnings according to income target, Labout supply elasticity is
negative after a certain level)
Medal - silver and bronze
-> predict: silver happier

Silver (loss domain compared to gold) vs bronze (gain domain compar
to losers)
=> Reference point: expectations, social ref

Labor supply: increase in
wage -> increase in working
hours (reduce the chance of
loss aversion)

Bundling at one income -- negative elasticity of supply
- Income target hypothesis
- Increase earnings -> Work less because reach
=> Reference point: target income

Having a predetermined fixed Bunching at the f (fixed rate)
payment gives ppl incentive => Reference point: f - expectation
to do the piece rate earnings
close to that

DECISION UNDER UNCERTAINTY: unknown probability (rain/norain)
1
...

-

individuals form beliefs about the number of red balls: r and blue balls 1-r
make choice consistent with that subjective believe
Critique: Ellsberg Paradox (Ambiguity aversion)
Two urn (50;50 and unknown-1-unknown)
If you believe urn 2 have more red than blue, there must be a case between 2 options you choose urn
2

-

Ambiguity aversion: ​an individual rather chooses an alternative where the probabilities are known
than an alternative where the probabilities are not known
...
Limited attention and Salience
-

Model with visible and opaque component: ​V = v + o
introduce​ θ : how attentive a person is , only applies to the opague
-> ​ ​V^ = v + θ o
S: the more salient (noticeable) the opaque, the higher the θ

-

People do pay more attention to the salient part
Price is more salient than sales tax => elasticity of demand is higher for price than tax
Excise tax more salient than sales tax => elasticity of demand of is higher for excise tax than sales tax

Important study: ​Pre-tax price vs
...

Example used: ​sales tax in the US ​which is added at the cashier
Questions:
- Tax salience (tax inclusive prices) have on consumer demand behaviour?
- Do consumers underreact to taxes that are not included in the posted price? -> the evidence of θ
Framework:
Interpret the last equation
- In the case of full attention -> RHS nominator = 0 -> θ = 1 -> full attention case !!
- In the case of limited attention but when then don't pay attention at all
- RHS = 1 -> θ = 0 -> !!
-

The RHS:
- Nominator: posting tax inclusive price
- Denominator: change in demand given the price increase corresponding to the tax
- Explain whether θ = 0 or θ = 1

Study 1: sales tax of toiletries in department store
Design
- DDD: triple difference estimator
- compared the “within-treatment-store" DiD estimator DD(TS) to a “within-control-store" DiD
estimator DD(C)
- Triple difference estimator: DDD = DD(TS) - DD(C)

-

Elimination of within-store (some stores are affected by shocks) and within-product time
trends (some products are affected by shocks)
- Other assumptions that this depends on to make it consistent:
- Rule the specific shock to the specific categories at the specific store at the specific
period - buy shampoo at that exact Boots period

Results
- DDD = -2
...
sign at 1% level
- Making the tax salient​ reduced demand by 7
...
35​: ​an increase in tax of 10% reduces demand by the same amount as an increase in
price of 3
...
88%
a 1% rise in the gross-of-sales-tax price is estimated to reduce beer consumption by 0
...
Cognitive Limitations: ability, knowledge, background
- human cognitive systems might have limited capacity (`limited bandwidth')
- Manitest 1
- individuals might lack knowledge about economics/finance to make important economic/financial
decisions
- Lusardi and Mitchell, 2007: Financial Literacy - questions and savings
- Bertrand and Morse, 2011: Financial Literacy - payday loans
- important in understanding the sources and consequences of socioeconomic inequality (background)
Important study: Limited Bandwidth - Mani et al
...
Stylized Facts (U
...
)
2
...
Do people know they have self-control problem? Do they think their savings are optimal? - ie: are the
sophisticated or are they naive or are they time consistent
Study by Choi et al (2009)
- People believe they should save more: 68%
- People plan to increase their savings : 24%
- People actually do so: 3%
-> EVIDENCE: SELF-CONTROL PROBLEM!!

3
...

- “Defined contribution”​: employee can​ decide an amount of pre-tax earnings to an
account​, can decide what to invest using that money => EMPLOYEE are responsible for the
decision step: to open the savings account, the % , the investment decision
- “​Defined benefit”:​ employer promises a pension for employee in the future based on
performance and earnings history
Standard theory: “nudging” & “default rate" have no effects on savings
INTRODUCE “Automatic enrolment” employees are enrolled, % are set, and investment to a money
market fund -> US, 401k account
Standard model : ​no change in savings because they always act optimal and ppl are free to choose choices and
no procrastination etc
...
, 2001:​ preset enrolment, contribution rate
and investment choices have good effects
...
For a time-consistent agent, the ranking of
outcome-time pairs does not depend on whether the individual is asked about it at time t or at some
other time s
- This implies that individuals do not have an incentive to deviate from the exante plans they have
made
...
Put
differently, there is no disagreement about the optimal savings rate at a given point in time between
current selves and future selves
...
In a standard model, we typically assume that agents are fully rational (i
...
they
have full information and the cognitive capability to process that information), the additional
information should not affect people's desired savings rates
...
A large proportion (78%) of those offered the plan joined
ii
...
Average savings rates of participants rose from ​3
...
6%​ over the course of 40 months
-

A standard economic model would have predicted no one joining the plan and no change in the
savings rates because under the model, individuals would be time consistent & choosing the optimal
savings rate already
...
Using behavioural tools to change people's behaviour can be
thought of as paternalistic and should therefore be avoided
...

- `Libertarian/soft paternalism' is a philosophy that advocates designing institutions that help people
make better decisions but do not impinge on their freedom to choose (Thaler and Sunstein, 2003)
...

- Notice that there is always some default option if no action is taken
...

- alternative: no default but ​mandated active choice (regulate)
f) Examples of nudging: “Putting the fruit at eye level counts as a nudge
...
, 2011: Implementation intentions and vaccination rates : with date and time, the
received vaccination rate increased

-

Rozin et al, 2001: more accessible healthier eating at school
The Nudge Unit:
- making public services more cost-effective and easier for citizens to use
- improving outcomes by introducing a more realistic model of human behaviour to policy
- enabling people to make 'better choices for themselves'

Criticism:
- “Iam an adult
...

-

Education affects earnings
Intergenerational earnings​ persistence is high (cha giàu con giàu) while social mobility (nghèo lên
giàu) is low
Social-economic status affects
- University enrolment
- SES gap in enrolment even conditional on ability (perceptions)
...
34)
UK, More deprived -> lower Higher Education participation
- Among the top scoring in those groups, difference between the richest and the poorest is
around 22
...
5% point increase in attending HE)

2
...
Standard theory
- Payoff maximising: discounted streams of future payoffs w uni + cost > discounted streams of future
payoffsw/o uni
b
...

c
...
Test: raise enrolment!!
Bettinger ​et al (2012):
application to financial aid
(FAFSA form​) for poor (info, and
assistance)

Control: info booklet
Info only treatment : info +
service estimation of eligibility
and costs
FAFSA treatment :
info+service+assistance to
complete FAFSA


...
Provide info only doesnt help ->
should provide​ assistance

Hoxby and Turner ​(2013):
info/application fee waivers for
good poor to apply to selective
college (when they/their parents
don't understand about lower
costs, higher benefits)

6 different groups
Control group, Application
guidance, Net costs, Fee waiver,
ECO(all 3), Parent AN


...
Intent to treat(ECO): lower

...
Other 3: guidance and fee waive
kinda effective, net costs (just
info) ambiguous
Jensen (2010): info about
earnings for poor
1 Perceived earnings and real
education level (Present bias)
2 Info and education decision
3 Info on returns and better
decision (bounded rationality)

Bettinger​ et al, 2012:
3 groups:
- Control: info booklet
- Info only treatment : info + service estimation of eligibility and costs
- FAFSA treatment : info+service+​assistance to complete FAFSA
2 smaller sub group
- Dependent Participants: Parents received infor
- Independent Participants: Self-received
Results:
- Help students in FAFSA treatment, whether dependent or independent
- Info-only doesn't help, even negative but insig
...
37 yr

2 Students do perceive wrongly
- overestimate earnings with primary school by 11% ​an​d underestimate earnings with
secondary school by 14%​ -> bounded rationality, kinda different weights (present bias)

3
...
Info and real results
- Treatment:​ 0
...


Summary:
- perceived returns to education predict/correlate educational investment decisions
​informing students​ about actual earnings of people with different educational attainment acts
- (i) students' perceptions about the returns to education as well as: less overestimate for
primary, less underestimate for secondary
- (ii) students' actual educational attainment decisions: 0
...
When
compared to other interventions aimed at increasing school attainment, Jensen admits that​ CCTs
interventions such as the PROGRESA/Oportunidades program in Mexico can have more impact,
particularly for those at the lower ends of the income distribution, and can be less costly,
however
...
Fact
- Women under-represented in STEM
- STEM has higher salary
Reasons:
- difference in skills
- difference in tastes
- difference in preferences for environment, endurance
- or difference in​ beliefs​: perceptions form actions
2
...
(2014): competitiveness measured by
the above experiment -> track in secondary school
(ban ABCD)


...
differences in preferences: competition

...
​ gender differences explains 20% gender diff in
track choices

Buser and Yuan (2016): women give up more (once
loss randomly, will they re-enter)

Labo

...
women
...
differences in characters: discouraged by negative
feedback
Field:

...
(2012): nature vs nurture
competitiveness in gender

only in patriarchal (phụ hệ): after puberty, jump in
male who compete
-> socialization process (puberty) can shape
competitiveness, not gender
...
and Vesterlund, L
...
, Becker, A
...
, Enke, B
...
, Sunde, U
...
Preference Module - Falk et al 2015a
What is it? ​combined survey and experiments
- First comprehensive ​experimentally-validated preference survey ​module
- Idea: select those survey items (quantitative and qualitative) that​ (jointly) predict behavior​ in
incentivized choice experiments best
- do the experiments with the small group of ppl in order to select the best survey items that
can capture behaviour on the​ incentivized choice experiments
...
-> use staircase to see switch
point as well
- Trying to capture the incentivized experiments by doing survey

1
...
Is the heterogeneity in preferences larger between-country or within-country?
- Variation can be mostly explained (higher than 80% across all 5) by WITHIN-COUNTRY VARIATION
rather than BETWEEN COUNTRIES

3
...
g
...
e
...

-

Do people's preferences correlate with individual behaviours?
Preferences do correlate with behaviours
Investigate​ predictive power of preferences​ for economic and social ​behaviors​, around the world
Time: Patience and accumulation decisions , hence GDP
- Dynamic theories in both micro- and macroeconomics highlight the crucial role of
time preference for accumulation processes and hence ultimately income
- Patience -> accumulation in physical cap, human cap and knowledge
- Consistency between theoretical work and empirical facts (Falk et al 2015)
- Raw correlation:​ Patience - Contemp
...
63 (R squared 40%)
- Risk: Risk taking and risky choices
- Social: Social preferences and social interactions

-

This is the nice way to test the meaningfulness of the survey: how can it predict behaviours in real
world

5
...
Income 0
...
> can we use these proxies for patience?
- Relationship b/w patience and proximate determinants extends to many other proxies for
human and physical capital as well as factor productivity
- Holds for both stocks (years of schooling, capital stocks,
...
)
- Correlations very strong: hold conditional on full set of covariates and sometimes
even conditional on GDP (not a third factor causing both patience and accumulation to
increase)
5a
...
What are the origins of the between-country variation in preferences
Reason 1: ​differences in preferences across countries have deep historical roots that go back thousands of
years, through variety of channel -> because of separation early/late but through WHICH CHANNELS?
- genetic
- experiences
- shocks
Test 1: According to Falk et al: historical events caused differences in preferences
(Becker/Enke/Falk, 2016)
Hypo: migration pattern “ Out of Africa"
- The idea:
- Expected difference in preference endowment between 2 populations at a specific point in
time would increase in the number of period that 2 pop were separated
- E (xiT- xjT) increase with no of separations

Separations -> source of channels:

Test 2: 3 class of proxies for​ temporal distance
- Genetic distance
- Linguistic distance
- Predicted migratory distance measures: WALKING TIME along the optimal route
between 2 locations
Framework:

-

dyadic regression
difference between preferences regressed on temporal distance between 2 populations
Control for country fixed effects to take out country specific unobservables , include disturbance term

Results:
- the larger the temporal distance between two countries, the larger their (absolute) difference
in risk aversion, altruism, trust, positive recip, (patience), (negative rep)
- R-squared is very high: lots of variation in pref between countries can be explained by
temporal distance
-

Holds for different proxies for temporal distance
Conditional on country FE, large vector of geographic, economic,
institutional controls


...
systematic preference between preferences and background char: age, gender, ability

...
patient -> higher accumulation -> higher GDP


...
Within
school/cohort
variation: control
for school fixed
effect, grade fixed
effect -> DID


...
Generosity and
prosocial

...
Classroom
behavior

Variation sources:

...
Within cohorts:
Only new entrants
are treated (year
2&3 in treatment,
grade 2 in delayed
treatment)


...
Within classroom:
Some small groups
have poor kids

+increased the
amount shared for
both poor (12%)
and rich (6%) ->
less discriminated,
fairer

...
Play-date: having
poor classmates or
poor study partner
increase
willingness to go
on playdate (lower
price)

...
Discipline slightly
worse in the case
of cursing in
treated classroom

Kose et al (2015):

Germany - Bonn

3 groups:


...
High SES and low

social preferences:
parental, home
environment:
Question:

...
Mentoring
programme
(guardian)

and Cologne

...
TL - CH: role of
parental and SES
differences

...
Social treatment
can close the SES
gap in prosociality
-> positive and
persistent


...


...

Rao paper - 2013 : social preferences
Motivation: diversity in schools can positively affect social beha and attitudes
Background: Delhi, India: where poor student are forced to be admitted to school
...

- Social interaction (teammates decision, “willingness to play" with other kids)
- Sports event
- Discrimination against the poor decreases as the stake increase
- But treated students discriminate less ( at 50ruppes, share of
discriminating only about 1/2 that of untreated)
- Willingness to play: having poor classmates (and study partner) increase WTP
(lowering the min payment to accept to play)
- Poor classmate -7 rupp
- Poor study partner: -8 rup
- Academic (Test scores, teacher comments)
- Test scores: no significant changes, negative effect on English
- Adverse effects on swearing, but no violent
...


-

-

Post treatment (Wave 2): When you give poor kid a mentor, their prosocial status
could be the same/as high as Control High SES (no significant different at 1% level) ->
close the SES gap​
...


Ie:
-

-

-

-

Correlation: prosocial by mother and mother's interaction can help close the SES gap that explains
prosocial interactions -> ​High SES, more time spent with parent and a more prosocial parent are
associated with higher levels of child prosocial attitudes
Causation:
- Treatment has strong effects to close the gap in prosocial attitudes
- Persistent effect (After wave 3)
Heterogenous:
- Prosocial of mother x Treat and Social interaction x Treat is negative -> Mentoring benefits
those who mothers don't care or mother are not prosocial!
Prosociality is malleable and early childhood interventions can change personality and
decreasing socioeconomic differences

---