Health, Skills, and Parentingtmwcenter.uchicago.edu/wp-content/uploads/2018/12/James...2018/11/28...

Build Skills to Benefit From Later Life StaticComplementarity

Heckman Health, Skills, and Parenting

Returns Per Unit Invested

Job Training

Schooling

Preschool programs

Programs targeted toward the earliest years

Prenatal programs

Post-SchoollSchool4–50–3Prenatal0

Rate

of

Retu

rn t

o I

nvest

ment

in H

um

an C

apital

Heckman Health, Skills, and Parenting

Powerful Evidence For Effectiveness of Early Interventions

Heckman Health, Skills, and Parenting

Demonstrates the Power of Family Influence

Heckman Health, Skills, and Parenting

Successful Interventions Percolate Throughout the LifetimeThrough Multiple Channels

Heckman Health, Skills, and Parenting

Some Long-Run Evidence on the Effects of Quality EarlyInterventions

• Many successful early childhood interventions followed over thelife cycle operate primarily through boosting non-cognitiveskills of participants and parenting skills. IQ is often barelybudged for interventions past age 3.

• Long term evaluations of interventions often provide a differentassessment of the effectiveness of interventions than do shortrun evaluations.

Heckman Health, Skills, and Parenting

Long-Term Evaluations are Essential

• Literature filled with large—even miraculous—short-termevaluation results that fade out when evaluated long-term.

• Filled with “curricula” rather than a deep understanding of thefamily and how successful programs replicate beneficial familyenvironments.

Heckman Health, Skills, and Parenting

Perry Preschool Project

Heckman Health, Skills, and Parenting

Starts at Age 32 hrs a Day – Two Years

Heckman Health, Skills, and Parenting

Curriculum: Plan, Do, Review

Heckman Health, Skills, and Parenting

10% Rate of Return Per Dollar Invested(through age 40 and ignoring health benefits)

Heckman Health, Skills, and Parenting

Ongoing Work:Follow-up Through Mid-Fifties

Heckman Health, Skills, and Parenting

Figure 7: Male Cognitive Dynamics

79.2 94.9 95.4 91.5 91.1 88.3 88.4 83.7

77.8 83.1 84.8 85.8 87.7 89.1 89.0 86.0

75

80

85

90

95

100

105

Treatment

Control

IQ

4 5 6 7 8 9 10Entry

Age

Treatment

Control

Heckman Health, Skills, and Parenting

• Worked primarily through noncognitive channels.

• Early interventions promoting skills lower the probability ofengaging in unhealthy behaviors in adulthood.

Heckman Health, Skills, and Parenting

Figure 8: Perry Preschool Program: Histograms of Indices of PersonalitySkills and CAT Scores

Panel A. Externalizing behavior, Panel B. Externalizing behavior,

control treatment

(Bigger is better) (Bigger is better)

0

0.2

0.4

0.6

0.8

1

Density

1 2 3 4 5

Density

1 2 3 4 50

0.2

0.4

0.6

0.8

1

Source: Heckman et al. (2013).

Heckman Health, Skills, and Parenting

Figure 8: Perry Preschool Program: Histograms of Indices of PersonalitySkills and CAT Scores, Cont’d

Panel C. Academic motivation, Panel D. Academic motivation,

control treatment

1 2 3 4 5

Density

0.6

0.5

0.4

0.3

0.2

0.1

01 2 3 4 5

Density

0.6

0.5

0.4

0.3

0.2

0.1

0

Source: Heckman et al. (2013).

Heckman Health, Skills, and Parenting

Child Preferences, Beliefs, and Skills Induced By Intervention

Heckman Health, Skills, and Parenting

Figure 9: Felt as belonging to school at 19 (sign of factor); feels able tochange things in life at 40; has little control over things at 40

0

.1

.2

.3

.4

.5D

ensity

−3 −2 −1 0 1 2Factor Score

Control Treatment

Average Treatment Effect, Pooled: .48 (p−value: .01)Average Treatment Effect, Females: .5600000000000001 (p−value: .03)Average Treatment Effect, Males: .4 (p−value: .06)

Heckman Health, Skills, and Parenting

Figure 10: Invites friends home frequently at 15 (sign of factor); helpothers at 27

0

.1

.2

.3

.4

.5D

ensity

−3 −2 −1 0 1Factor Score

Control Treatment

Average Treatment Effect, Pooled: .39 (p−value: .03)Average Treatment Effect, Females: .44 (p−value: .12)Average Treatment Effect, Males: .35 (p−value: .07)

Heckman Health, Skills, and Parenting

Decomposition of Treatment Effects, Males

90 100%0 10 3020 40 50 60 70 80

Personal Behavior

Cognitive Factors

Other Factors

Socio-Emotional State

CAT total*, age 14(+)

Employed, age 19 (+)

Monthly Income, age 27 (+)

No tobacco use, age 27 (+)

# of adult arrests, age 27 (-)

Jobless for more than 2 years, age 40 (-)

Ever on welfare (-)

Total charges of viol.crimes with victim costs, age 40, (-)

Total charges of all crimes, age 40 (-)

Total # of lifetime arrests, age 40 (-)

Total # of adult arrests, age 40 (-)

Total # of misdemeanor arrests, age 40 (-)

Total charges of all crimes with victim costs, age 40 (-)

Any charges of a crime with victim cost, age 40 (-)

Heckman Health, Skills, and Parenting

Effects of Perry at Late MidlifeJoint Work with Ganesh Karapakula

Heckman Health, Skills, and Parenting

Crime

Heckman Health, Skills, and Parenting

Figure 11: Cumulative conviction counts for violent misdemeanors forPerry men

µ

µ

µ τ τ τ τ µ µ

Heckman Health, Skills, and Parenting

Figure 12: Cumulative conviction counts for property, violent, ordrug-related misdemeanors in the pooled sample

µ

µ

µ τ τ τ τ µ µ

Heckman Health, Skills, and Parenting

Figure 13: Probability of ever having been convicted more than once forviolent misdemeanors in the male subsample

µ

µ

µ τ τ τ τ µ µ

Heckman Health, Skills, and Parenting

Employment

Heckman Health, Skills, and Parenting

Figure 14: Monthly earnings (in 2017 USD) for male participants

µ

µ

µ τ τ τ τ µ µ

Heckman Health, Skills, and Parenting

Figure 15: Employment rate for male participants

µ

µ

µ τ τ τ τ µ µ

Heckman Health, Skills, and Parenting

Figure 16: Employment rate for female participants

µ

µ

µ τ τ τ τ µ µ

Heckman Health, Skills, and Parenting

Health

Heckman Health, Skills, and Parenting

Table 1: Post-midlife health effects on Perry men

Control Treatment Worst-caseMean Mean p-value

High total cholesterol 0.944 0.708 0.018High C-reactive protein 0.542 0.346 0.030Weekly homecooking rate 4.333 7.724 0.041Monthly bedridden rate 0.032 0.015 0.065

Heckman Health, Skills, and Parenting

Table 2: Post-midlife health effects on Perry women

Control Treatment Worst-caseMean Mean p-value

Hair cortisol 89.29 39.01 0.024Regular exercise 0.250 0.435 0.079Diabetes 1.000 0.826 0.037Treated for substance usage 0.150 0.000 0.027Prolonged uninsured status 0.200 0.043 0.059

Heckman Health, Skills, and Parenting

Cognitive and Noncognitive Skills

Heckman Health, Skills, and Parenting

Table 3: Significant effects on cognitive and socioemotional outcomes inthe pooled sample

Control Treatment Worst-caseMean Mean p-value

General intelligence (Empiri-cal Bayes (EB) score)

−0.194 0.187 0.054

Positive personality (EB scorebased on self-rating)

−0.158 0.155 0.033

Positive personality (EB scorebased on self and external rat-ings)

−0.211 0.216 0.040

Heckman Health, Skills, and Parenting

Table 4: Significant effects on Perry men’s cognitive and socioemotionaloutcomes

Control Treatment Worst-caseMean Mean p-value

General intelligence (Empiri-cal Bayes (EB) score)

−0.239 0.245 0.051

Positive personality (EB scorebased on self-rating)

−0.135 0.144 0.030

Positive personality (EB scorebased on self and external rat-ings)

−0.149 0.169 0.058

Heckman Health, Skills, and Parenting

Parenting Mechanisms Generating Treatment Effects

Heckman Health, Skills, and Parenting

Parental response to Perry Preschool Program after 1 year experience oftreatment:

10

20

30

40

50

60

Pro

port

ion

−.015 −.01 −.005 0 .005 .01 .015Belief in Importance of Parenting

Control Treatment

Heckman Health, Skills, and Parenting

Parental Warmth, Perry Preschool

Heckman Health, Skills, and Parenting

Parental Authoritarianism, Perry Preschool

Heckman Health, Skills, and Parenting

Intergenerational Externalities: The Effects of the PerryPreschool Program on the Children of Perry Participants

Joint work with Ganesh Karapakula

Heckman Health, Skills, and Parenting

Selected Outcomes for All Children of the Perry Participants

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Completedhigh school

In goodhealth

Employedfull-time

Neversuspended

Neverarrested

P .0849 P .0624 P .0548 P .0347 P .0792

Part

icip

ant-

level avera

ge o

f child

ren's

outc

om

es

Control group's mean Treatment effect (difference-in-means)

P: Worst-case randomization test-based exact p-value

Heckman Health, Skills, and Parenting

Figure 17: Summaries of selected intergenerational outcomes

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 17: Summaries of selected intergenerational outcomes

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 18: Selected outcomes of the male children of the pooledparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 19: Selected outcomes of the female children of the pooledparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 20: Selected outcomes of the pooled children of the maleparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 21: Selected outcomes of the male children of the maleparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 22: Selected outcomes of the female children of the maleparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 23: Selected outcomes of the pooled children of the femaleparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting

Figure 24: Selected outcomes of the male children of the femaleparticipants

≥

≥

≥

≥

≥

≥

≥

≥

≥

µ

µ

µ τ τ τ τ µ µ

≥

Heckman Health, Skills, and Parenting