econ 980o health, education and...

32
ECON 980o Health, Education and Development Lecture 2 September 25, 2008

Upload: others

Post on 16-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

ECON 980oHealth, Education and Development

Lecture 2September 25, 2008

Page 2: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Motivation: Preston curve

Page 3: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

• 3 possible interpretations:

‐ Income leads to health

‐ Health leads to income

‐ Third factor behind both (geography, institutions, politics, etc)

→ Difficult to isolate causal effect of health on income

Income‐health gradient

Page 4: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Multiple potential health-to-wealth pathways:

Direct productivity outcomes:– Health/nutrition increases labor market productivity

(Thomas et al, 2004)– Child nutrition increases productivity of schooling

(investment effect)– Ill health has large externalities

Particularly true for infectious disease ….

Lower life expectancy :– Reduces investment in education– Reduces savings rates – Increases population growth

To test theory, need to isolate particular channel

This class, focus on second and third channels

Page 5: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Why is effect of health on schooling potentially even more important for growth than labor market effects?

• Child interventions thought to have investment effects, so benefits of health intervention multiply over lifetime

What do we mean by this?

- Human capital investments- Biological investments (cognitive development)

• Another motivation: What is education production function? Why don’t people invest more in education in developing countries?Is it purely credit constraints?

Page 6: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Intestinal helminth (worm) infectionsIntestinal helminths among most widespread diseases in less

developed countries• 1.3 billion people infected with roundworm• 1.3 billion with hookworm• 900 million with whipworm• 200 million with schistosomiasis

Disease features:• Particularly concentrated among school-age children & in

sub-Saharan Africa• Worm load (# of worms) matters for health

Health impact:• Light infections often asymptomatic• Severe infections frequently cause anemia, malnutrition,

stunting, wasting, listlessness, and abdominal pain

Page 7: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Non-health impact:Education

• What channels important?

• Sickness lowers school attendance

• Hypothesized that children with intense infections less attentive in school and thus show reduced educational achievement

Spillovers

• Transmitted through contact with infected fecal matter

• What do we mean by spillovers in this context?

→ For parasitic and infectious diseases, reducing number of worms in one person reduces rate of infection in others

Page 8: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Prevention and treatment• Transmitted through poor sanitation & hygiene, bathing in

infected water

Treatments: 1. Behavioral change (shoes, latrines, bath in clean water)2. De-worming drugs

Advantages:– Low-cost – Easy to administer (single-dose oral therapies)– Virtually no side effectsDisadvantages:– Reinfection (need to take every 6 months to stay clean) – Girls over 12 typically excluded because of embryotoxicity

School-based de-worming particularly cost-effective. WHY?

Page 9: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Kremer/Miguel hypotheses:

• Even small health effect could have important effect on educational attainment and returns to schooling

• Externalities are also likely to be large because of way they spread

(i.e. de-worming one kid equivalent to de-worming ?? kids)

Page 10: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Why Kremer/Miguel paper important:

Previously much debate over whether to spend money on de-worming drugs

• Clear estimate of prevalence, but health impact of worms thought to be low based on medical studies

→ Not viewed as high priority policy area given preventable diseases with higher morbidity and mortality

In influential Cochrane review published in the British Medical Journal, Dickson et al. (2000) claim that:

“The evidence of benefit for mass [deworming] treatment of children … is not convincing. In light of these data, we would be unwilling to recommend that countries invest in programs that routinely treat children with anthelmintic drugs.”

Page 11: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Randomized experiment: Kenyan primary school de-worming project• NGO-initiated program to administer de-worming drugs

(albendazole, praziquantel) to students in primary school

• 75 primary schools and 30,000 children in single region of Kenya (Busia)

• Health education component (preventive health behavior)

• Take-up rate approximately 70 percent

Page 12: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Evaluation: Track outcomes of participating students and a comparison

group after the experiment

Want to know effect of drugs on:

• Health• School Attendance• School Performance

Key feature of intervention: For political/ethical reasons, all schools in districts had to be treated

So how do they get a control group?

Page 13: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Experimental design:

De-worming phased in to schools in three stages, each school randomly assigned to one of three start dates

Group A (25 schools) in 1998-2001Group B (25 schools) in 1999-2001Group C (25 schools) in 2001

Phase-in most common form of randomized policy evaluation

Treatment Group 1998 1999 2000 2001 Group 1 (25 schools) T T T T Group 2 (25 schools) C T T T Group 3 (25 schools) C C C T

Page 14: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

How that translates into a regression:(Note: data from 1998, 1999)

ijtiijtititijt euXTTY ++′+⋅+⋅+= δββα 2211

How do you interpret ?

How do you interpret ?

Average effects of de-worming treatment in years 1 and 2 of program.

What do you expect coefficient on T2 to look like?

1̂β

2β̂

Page 15: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Experimental complications:

1. Low take-up (what is this?)Many students don’t take drug, mainly because they’re absent

So who do we put in the treatment group?

Possible approaches : • Intent-to-treat: Follow all students in treatment schools regardless of take-up

• Treatment on treated: Estimate effect only among those who take medicine

Which approach is better? Why are they useful?What assumptions necessary?

Page 16: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Other experimental complications

1. Contamination (#1)

Untreated students can get outside treatment for worms

Why is this a problem?

Why not a big problem in Busia:

• Low willingness-to-pay for de-worming medication• Main remedies herbal treatments not thought to have

large effects• Gives conservative treatment effect. Why?• Effect of school de-worming over and above potential

influence of home remedies

Page 17: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Other experimental complications

2. Contamination (#2)

Students assigned to control schools may benefit from treatment of others

Why?

Why is this a problem?

Page 18: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Recall necessary conditions for unbiased estimator:E[Net Effect]=E[YA

T=1-YBT=0 ] if and only if:

1. E[YBT=1]=E[YA

T=1] (both groups have the same response to treatment),

and

2. E[YAT=0]=E[YB

T=0] (both groups have same outcome if not treated)

Which condition violated by presence of externalities?

If untreated group affected by treatment, condition #2 violated

Page 19: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Why do externalities create complications with treatment estimation in randomized experiments?Untreated children benefit from treatment →Control group is contaminated

So instead of condition #2, we get: E[YAT=0]<E[YB

T=0]

How does this change the difference-in-difference estimate?

Does standard DID underestimate or overestimate the real benefit of the drugs?

Page 20: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Effect of de-worming treatment on worm infection:

0102030405060

Treated Students Comparablestudents incomparison

schools

Untreatedstudents intreatmentschools

Moderate-to-Heavy Helminth Infection Rates:

Page 21: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

How do we correct for this?(1) Randomization happened at school level, so compare

treatment and control schools that are far apart

Does this give you an unbiased estimate of the benefit of the program?

Does it allow you to measure the size of the externality?

No! – you get the total effect only

(2) Make use of predictions about the expected size of externality

This solves two problems: (what?)• Can get unbiased estimate of effect on control group• Can separate treatment effect from externality

Page 22: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

How do we predict size of externalities?What does size of externality depend on?

• Type of worm (life cycle, #eggs, how travels)• Mobility of children• Proximity, connectedness of water• Population density

How does this work?

If no externalities, rate of improvement among treatment and control groups should be independent of above factors

In practice?Interact above variable with treatment indicator and add to

OLS regression

Page 23: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Empirical SpecificationIdentification relies on the randomized design:

Yij is?health outcome of individual i in village j

Ti is ?indicator var for individual assignment to program

Xij is?a vector of village and individual characteristics

Nit is?total number of individuals at distance d from i

is?number of individuals distance d from i assigned to program

TitN

ijtiitTitijtititijt euNNXTTY ++⋅+⋅+′+⋅+⋅+= φγδββα 2211

Page 24: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Empirical Specification

Which parameter measures the externality?

What does this parameter measure in words?

Change in outcome with each additional treated person living d distance away

ijtiitTitijtititijt euNNXTTY ++⋅+⋅+′+⋅+⋅+= φγδββα 2211

Page 25: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Empirical Specification

How do we measure the average effect of de-worming on students in the treatment group in years 1 and 2?

ijtiitTitijtititijt euNNXTTY ++⋅+⋅+′+⋅+⋅+= φγδββα 2211

Page 26: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Empirical Specification

Average effect of de-worming on treatment group in year 1:

Average effect of de-worming on treatment group in year 2:

Are externalities ever likely to work in opposite direction?

ijtiitTitijtititijt euNNXTTY ++⋅+⋅+′+⋅+⋅+= φγδββα 2211

TitN⋅+ γβ1

TitN⋅+ γβ2

Page 27: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

What additional sources of variation could be used to further isolate externalities?

Page 28: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Example:

NP=number of students with indoor plumbing within 3 km radius

Using both sources of variation:

Can you think of an experimental method of measuring externalities?

(what can be randomized to give a prediction about size of externality?)

2211 ijtiPit

TPitdit

Titijtititijt euNNNNXTTY ++⋅+⋅+⋅+⋅+′+⋅+⋅+= ϕλφγδββα

Page 29: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

RESULTS

Health impact after 1st round of treatment:

• Prevalence of hookworm, roundworm, and schistosomiasissignificantly lower in Group 1

• Fewer Group 1 report having been sick

• Group 1 students have better height-for-age

• Worm prevention techniques education failed

Also bit impacts on health through externalities:

What are two types of externalities measured?

• Within-school and between school

Page 30: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Impacts on health through externalities:

• Cross-school externalities

– Infection rates 26 percentage points lower per 1000 pupils in Group 1 schools within 3 km

– Infection rates 14 percentage points lower per 1000 pupils between 3-6 km away

• Localized, within school externalities

– Rates of moderate-to-heavy infections substantially lower in Group 1 who did not receive medical treatment than among comparison students in Group 2 schools

Page 31: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Impacts on education:(Participation measured during quarterly unannounced visits)

Treatment associated with dramatic gains in school participation, large externalities:

• Baseline absence rate (30%) reduced by ¼

• Total effect: Increase in school participation ~0.14 yrs/ child

• What about impact on what is learned in classroom?– No significant impact on test scores (no evidence of

cognitive impact)– What do you make of this?

• Was it cost-effective?Very! $3.50 per additional year of schooling

Page 32: ECON 980o Health, Education and Developmentsites.fas.harvard.edu/~ec1386/Lecture_Notes_Fall_2008/Lecture2_Fa… · – Child nutrition increases productivity of schooling (investment

Conclusions(1) Externalities important part of health effect

Over 2/3 of returns to de-worming from externalities

(2) Easy to conclude intervention has small effect when externalities:

Treatment very cost effective, but only meets strictest standard when externalities taken into account

• Cost: $20/DALY saved per person vs $5 per total DALYs• Externality benefits 75% of total DALY reduction

This means that must incorporate externalities into estimates whenever may be present – double bias in experimental estimates

Can externalities ever work in opposite direction?