Michael Kremer (Harvard University) Evidence for …siteresources.worldbank.org/INTDEVIMPEVAINI/Resources/...Development Impact Evaluation Initiative Innovating in Design: Evidence

Development Impact Evaluation Initiative

Innovating in Design: Evidence for Impact in Health

Cape Town, South Africa December 10-14, 2012

Evidence for Impact in Health Scaling up innovative interventions in development/small changes, big results: design features for high impact

Michael Kremer (Harvard University)

1

0

1000

2000

3000

4000

5000

6000

United States 1900

Vietnam 2009

SSA 2009

Income per Capita (2005 dollars)

0

10

20

30

40

50

60

70

80

United States 1900

Vietnam 2009

SSA 2009

Life Expectancy

2

Overview

1. Impact of Health Investments

2. Consumer Behavior

A. Prevention

B. Treatment

3. Health Care Provision

3

Development impact of health investment: NTD example

Helminths widespread, treatment inexpensive and safe, diagnosis expensive

WHO recommends mass school-based treatment Prospective trial of Kenyan NGO program School absence falls by ¼ Epidemiological effects Neighboring schools, younger children Long run effects on work hours, earnings 60% rate of return Increased tax revenue > program cost

Response

Strong interest from PS of Ministry of Education, World Bank Inter-ministry committee Deworm the World Initiative at WEF Generates high-level political support Funds technical assistance

Inclusion in MoE plan, rollled out to 3.5 million children in 2009

Scandal in education ministry, donor support put on hold Identification of new donor for 5 year program reaching

more than 5 million children per year Programs in Bihar, Delhi, Rajasthan…

Public Goods: Spring Protection and Road Safety

$1K capital cost, negligible marginal cost

Randomized trial suggests 25% reduction in diarrhea, but substantial recontamination

Estimate willingness to walk, effects of different pricing and propery righs systems, prioritization among springs

Stickers in matatus encouraging consumers to speak up

(Kremer et al 2007) 6 1. Consumer Behavior 2. Health care provision

Determinants of Health

Consumer Behavior Prices Financial incentives (negative prices) Information

Quality of Provision Incentives Community monitoring and accountability

1. Consumer Behavior 2. Health care provision

Example: Deworming in Kenya

8

Price response (Kremer and Miguel 2007)

o 75% take-up when free, 18% with small user fee o No difference in take-up with family size despite per family fee structure o Sicker students not more likely to pay for drugs

1. Consumer Behavior 2. Health care provision

Impact of (Low) Prices on Preventative Health Take-Up

9

Does Paying Makes People More Likely to Use Health Goods?

Widely believed by practitioners

Could arise due to sunk cost fallacy

No evidence Bednets in Kenya & Uganda (Cohen &

Dupas 2010; Hoffman et al 2009)

Water purification in Zambia (Ashraf et al

2010)

10 1. Consumer Behavior 2. Health care provision

Financial Incentives (Negative Prices)

Large negative prices in conditional cash transfer

programs increase uptake of preventative health Health & education: Mexico (Gertler 2004; Schultz

2004) Health: Tanzania (Evans et al. 2012)

Small incentives also have remarkable impact HIV test results in Malawi (Thornton 2008) Vaccination in India Cash transfer in Malawi: Transfer amount does not

affect outcomes (Baird et al 2011)

11 1. Consumer Behavior 2. Health care provision

Implications

Implications for understanding consumer behavior Present-bias can explain several puzzles ▪ Prevention vs. Treatment ▪ Sensitivity to prices around zero

Implications for policy Partial cost sharing is typical compromise, but may be

worst of all worlds Zero prices, small subsidies Prioritization: cost effective prevention approaches for

communicable disease

Information

Often, little impact of health education Intensive school health education intervention had no impact on

worm prevention behaviors in Kenya (Kremer & Miguel 2007)

But in some cases, dramatic impacts Information on HIV prevalence by age in Kenya (Dupas 2011)

Rational learning model vs. limited attention model Fade out of education campaign effect Greater responsiveness to coarse information in Bangladesh (Bennear

et al 2011)

People tested negative for malaria still buy malaria meds in Kenya (Dupas et al 2012)

13 1. Consumer Behavior 2. Health care provision

Acute Treatment

Substantial expenditure on health treatment of doubtful utility (Das and Hammer, 2005; Banerjee et al., 2004)

Consumers bear substantial health risk (Gertler & Gruber 2002)

Low take-up of voluntary health insurance in Nicaragua (Thornton et al 2010)

Limited and mixed evidence on effect of price of clinic visits on take-up, health Exogenous increase in fees at clinics in Indonesia (Dow et al 2003)

Subsidized anti-malaria in Kenya (Cohen and Dupas 2011)

Mexican Seguro Popular health insurance program had no effect on healthcare utilization in short term (King et al 2009)

14 1. Consumer Behavior 2. Health care provision

Applying the Lessons: Program Design and Testing

Water infrastructure often expensive, recontamination frequent

Chlorination safe, effective, low-cost, but single-digit take-up under social marketing

Design for free distribution, convenience, habit formation, norm formation

15 1. Consumer Behavior 2. Health care provision

Chlorine Dispenser System

16

• Majority of households

test positive for chlorine, use sustained over time (Kremer et al 2011)

• Very low cost

• >400,000 people now have access in Kenya’s Western Province

• Successful pilots in Uganda, Haiti

1. Impact of health 2. Impact Mechanisms: Health Investment 3. Impact Mechanisms: Health care provision

Overview

1. Consumer Behavior

2. Health care provision

17

Background

18

Chaudhury et al. (2006)

1. Consumer Behavior 2. Health care provision

Poor Quality

Frequent use of unqualified providers in Tanzania and other countries (Leonard & Masatu 2007; Das & Hammer 2005)

19

6.3

4

1.5

0 1 2 3 4 5 6 7

Minutes spent per patient

Questions asked per patient

Physical exams per patient

Das et al 2008

Physicians in Tanzanian hospitals…

1. Consumer Behavior 2. Health care provision

Approaches to Addressing Government Failure in Service Delivery

Management reforms (e.g. supply chain management

Provider Incentives

Community mobilization for accountability

Paying for results

Contracting for health

20 1. Consumer Behavior 2. Health care provision

Provider Incentives

Pay for performance program in Rwanda: Significant increase in quality prenatal care but no impact on number of visits, immunization rates (Basinga et al 2010)

Non-financial rewards motivate public

health workers in Zambia (Ashraf et al 2012)

Monitoring program linked pay to nurse attendance in India (Banerjee et al 2008)

Initial attendance boost; disappears after 6 months (program undermined)

21 1. Consumer Behavior 2. Health care provision

Community Mobilization and Accountability

Community monitoring of health services in Uganda (Björkman & Svensson 2009) Provider absence 14% lower in treatment

facilities Vaccination rates increased 46% 46% more children received vitamin A

supplements Health improvements

Contrasting results elsewhere and weakness of pre-existing Ugandan efforts suggests implementation critical 22

Paying for Results?

Block grants to communities tied to results (Olken et al 2011)

Regular weight checks for children and iron tablets for pregnant women increased Malnutrition fell by 2.2

percentage points

23 1. Consumer Behavior 2. Health care provision

Contracting for Health in Cambodia

Management of healthcare contracted out to NGOs in some districts (Bhushan et al 2006)

Targeted health outcomes increased one standard deviation 42% increase in vitamin A and 36% increase in

prenatal care coverage Decreased absence rates among providers

• No change in non-targeted health outcomes • Increased government spending in contract districts offset by

decrease in private health spending

24 1. Consumer Behavior 2. Health care provision

Factors in Project Design

Consumer Prices and subsidies Design for convenience, habit formation

Provider Management Incentives Governance