One Researcher's Viewpoint on Policy Issues Relating to What have we - - PowerPoint PPT Presentation

University of Pennsylvania Behrman 1

One Researcher's Viewpoint on Policy Issues Relating to ‘What have we learned, and what’s next?’

• n the MDGs

Jere R. Behrman

University of Pennsylvania, Philadelphia, PA USA Policy Conference on Reaching the MDGs (12 June), 6th General Poverty & Economic Policy (PEP) Research Network Meeting Sheraton Lima Hotel, Paseo de la Republica 170, Lima, Peru 9-16 June 2007

University of Pennsylvania Behrman 2

Millennium Development Goals (MDGs)

• MDG 1: Eradicate extreme poverty and hunger
• MDG 2: Achieve universal primary education
• MDG 3: Promote gender equality and empower women
• MDG 4: Reduce child mortality
• MDG 5: Improve maternal health
• MDG 6: Combat HIV/AIDS, malaria and other diseases
• MDG 7: Ensure environmental sustainability
• MDG 8: Develop a global partnership for development

Established in 2000 with targets for 2015, monitoring; Lustig provides more details.

University of Pennsylvania Behrman 3

• Also considerable & increasing research on these &

related issues in developing world.

• Question arises, what have we learned from this

research that is germane to policies related to MDGs?

• This paper gives perspective of one development

economics researcher with experience in Africa, Asia and Latin America on this question, with emphasis on MDG1-7 (Lustig on MDG 8 as well).

• Some might wish that a research perspective would

result in set of magic bullets – “Do this. Do not do that” – but world too complicated & information too limited to provide such a simple list.

• But hopefully following six general points constitute a

perspective that helpful for policy considerations.

University of Pennsylvania Behrman 4

• 1. Essential to place research and policy

implications within framework for basic policy motives: (1) efficiency and (2) distribution

• Efficiency: social = private rates of return; otherwise

can improve welfare of all or of some at no cost to

• thers.
• Distribution: e.g. lessen poverty
• Tradeoffs (opportunity costs of resources) versus

win-win (e.g., educational spillovers on technology adoption; improving markets for capital, insurance, information with particular benefits for poorest)

University of Pennsylvania Behrman 5

• 2. Important to assess policy options in

terms of their relative economic costs

• Policy hierarchy (e.g., MDG 4 child mortality targets could

be obtained with many policies, but costs differ)

• Economic costs, public and private costs (distortion costs),

not governmental budgetary costs (includes transfers); Economic benefit-cost ratios.

• Copenhagen Consensus: 8 leading economists (4 Nobel

Laureates) & later UN ambassadors prioritized policies: 1 civil conflicts; 2 climate change (MDG7); 3 communicable diseases (MDG6); 4 education (MDG2); 5 financial stability; 6 governance; 7 hunger & malnutrition (MDG1); 8 migration; 9 trade reform; 10 water & sanitation (MDG6, 7).

• 4-5 projects in each, B-C ratios by experts. Challenging, but
• informative. E.G. CC7 hunger & malnutrition (MDG1)

University of Pennsylvania Behrman 6

Example: Present discounted value of shifting one LBW infant to non-LBW status in low-income country, 5% discount rate

$580 Sum of PDV 8 $45 Intergenerational benefits 4 $23 Reduced costs of chronic diseases 41 $239 Increased cognitive ability 17 $99 Increased physical productivity 14 $80 Reduced neonatal and infant/child illness costs 16 $93 Reduced infant mortality % of column PDV

University of Pennsylvania Behrman 7

Sensitivity of benefits to averting low birthweights to changes in discount rate

47% 100% 170% 351% % of 5% discount rate 273 580 1378 2037 Sum of PDV 7 45 219 422 Intergenerational benefits 1 23 132 239 Reductions in costs of chronic diseases 69 239 600 846 Gains from increased cognitive ability 28 99 249 351 Gains from increased physical productivity 78 80 81 81 Reduced illness costs 88 93 $95 $96 Reduced infant mortality 10% 5% 3% 1%

University of Pennsylvania Behrman 8

Benefit-Cost Ratios for these Opportunities

University of Pennsylvania Behrman 9

University of Pennsylvania Behrman 10

• 3. Policies usually have unintended
• r indirect effects:
• Increase resources for some & change incentives for

behaviors for individuals and families and other entities, including service providers (e.g., in health and education) and governmental bureaucrats (e.g., rents from policy-created restrictions, patronage).

• E.g., Nutrition programs targeted towards children

but families redistribute

• E.g., High administrative & logistic costs of in-kind

programs

• E.g., Poorly targeted
• Such concerns behind Conditional Cash Transfer

programs (e.g., Mexican PROGRESA)

University of Pennsylvania Behrman 11

• 4. Policies likely to be more effective the

more closely targeted to the real objective:

• Often policies targeted to intermediate, not ultimate
• bjectives.
• E.g., MDG 2 and 3 on school enrollments & gender equality.
• Schooling attainment or learning of real interest.
• Enrollment & attendance targets create incentives for

schools to over-report, but not to assure learning.

• School enrollment often lower for girls; therefore concern

about disadvantaging girls. But in some cases (e.g., Malawi, Mexico), boys fail & repeat or drop out & re-enter school more & have higher enrollment, but lower

• attainment. Therefore PROGRESA higher scholarship for

girls incentive to increase gender gap in school attainment.

University of Pennsylvania Behrman 12

• Positive example: teacher absence perceived problem

that limits student learning.

–Recent policy evaluation experiment in rural India: schools provided cameras with unalterable time/date mechanisms; teacher bonuses depending on teachers present -- increased teacher presence & student tests.

• Related but different example: MDG 4-6 emphasis on

traditional health problems of developing countries – communicable, maternal, perinatal and nutritional conditions (CMPNC). But non-communicable diseases (NCD) are larger & predicted growing share of health

• problems. So focus on CMPNC may divert attention

from more important health problems.

University of Pennsylvania Behrman 13

Chart 7A. % Composition of DALYs Projected for Three Major GBD/WHO Categories for All Developing Countries

41% 32% 50% 54% 13% 13% 14% 0% 10% 20% 30% 40% 50% 60% 2005 2015 2030 CMPNC NCD Injuries 37% 46%

University of Pennsylvania Behrman 14

Table 3. Ranking of Top Causes among Projected DALYs for All Developing Countries and for Low-Income Developing Countries

Ranking of Top Conditions Percentage Shares of Total DALYS for Leading Projected Causes Aggregate Tripartite Category

• f Causes

All Developing Countries Low- Income Developing Countries All Developing Countries Low-Income Developing Countries Causes Ranked for All Developing Countries in 2005 2005 2030 2005 2030 2005 2015 2030 2005 2015 2030

Neuropsychiatric conditions NCD

1 1

1 2 12.2% 13.0% 13.3% 11.4% 11.7% 11.8% Cardiovascular diseases NCD

2 3

3 4 9.9% 10.2% 10.9% 10.2% 10.2% 10.5% Unintentional injuries Injuries

3 4

2 1 9.3% 9.5% 9.5% 11.3% 11.5% 11.8% Perinatal conditions CMPNC

4 9

4 9 6.7% 5.6% 4.1% 7.0% 5.9% 4.3% HIV/AIDS CMPNC

5 2

6 3 6.0% 8.2% 11.0% 5.9% 8.6% 11.2% Respiratory infections CMPNC

6 10

5 10 6.3% 4.7% 3.1% 6.2% 4.8% 3.3% Sense organ diseases NCD

7 5

9 6 5.0% 6.0% 7.3% 4.5% 5.1% 6.0% Malignant neoplasms NCD

8 7

8 8 4.5% 5.1% 5.8% 4.6% 5.1% 5.8% Respiratory diseases NCD

9 6

11 5 3.8% 4.7% 6.1% 4.0% 5.0% 6.4% Diarrheal diseases CMPNC

10 13

10 12 4.3% 3.2% 2.1% 4.3% 3.2% 2.2% Intentional injuries Injuries

11 8

7 7 3.5% 3.8% 4.1% 4.9% 5.3% 6.0%

University of Pennsylvania Behrman 15

• 5. Policy effectiveness depends on

context so not necessarily transferable

• Effectiveness depends on market, policy, cultural

environment:

–Improved child nutrients depends on infectious disease environment –Increasing textbooks depends on teacher quality –increased mothers’ schooling impact on child schooling depends on labor markets for women (e.g., US vs. rural India)

• Therefore not so simple as blindly emulating specific

“best practice” policies.

University of Pennsylvania Behrman 16

• 6. Likely considerable gains to collecting

good information & undertaking good systematic analysis of policies

• Many determinants of outcomes of interest (e.g.,

maternal health - MDG 5, water quality - MDG 7); some are not easily observable (e.g., innate ability, health, motivation for MDG1-6; soil & water qualities for MDG7).

• Therefore associations between some policy & some
• utcome not likely to reveal policy impact; individuals (or
• ther entities) exposed to policy not likely to be same as

those who are not wrt unobserved characteristics.

• E.g., for MDG 2 & 3, those who attend school or attend

better schools likely to differ from those who do not with respect to ability, motivation and family background.

University of Pennsylvania Behrman 17

• Challenging counterfactual question: “What would be

impact on person exposed to policy change in comparison with same person who at same time is not exposed to the policy change?”

• To answer need good information (baseline,

longitudinal, indirect effects) & means of establishing control for counterfactual question (experiment, statistical methods such as matching, instrumental variable, fixed effects or structural models).

• Probably best-known large-scale experiment Mexican

PROGRESA anti-poverty/human resources program with initial random assignment of 506 rural communities for treatment & then matching: showed value and improved program & aided survival.

University of Pennsylvania Behrman 18

• Other interesting smaller-scale examples; i.e., for

MDG2 and MDG3:

–Flip charts in W. Kenya (association not causal) –Teacher absence, monitoring (cameras) & incentives in rural Rajasthan, India –Lottery for private school vouchers in urban Colombia

• Systematic data collection and analyses have costs,

but to attain MDGs takes considerable resources and savings in obtaining them more effectively would seem often to exceed information and analysis costs.

University of Pennsylvania Behrman 19

Conclusions

• Recent research has some important insights for

selecting policies likely to improve attainment of MDGs.

• Not magic bullets (i.e., policy innovations that will

have high benefit-to-cost ratios everywhere), but are examples in particular contexts that suggest consideration elsewhere. – E.g., Returns to human resource investments (health, nutrition education) high in many contexts, with synergies among different types of investments and with early life particularly important, often suggesting some “win-win” possibilities of increasing longer-run productivies and efficiencies in addition to attaining MDG-like goals.

University of Pennsylvania Behrman 20

– E.G. Conditional cash transfers & very specific incentives to improve specific problems ( e.g., teacher and health-care worker absence) seem very promising for helping to attain several of MDGs at least in contexts in which they have been explored.

• But

– Contexts differ & incentives for participants in policy chain – from policy makers to implementers to clients – also differ, with many unintended & indirect effects resulting from policies. – Information problems also severe.

University of Pennsylvania Behrman 21

– In particular contexts, likely to be multiplicity of policies that might help attain any particular MDG with different benefit-to-cost ratios, great variance in benefit-to-cost ratios across MDGs and between pursuing MDGs and other policies that might help enhance welfare of citizens.

• Therefore more serious efforts at collecting and systematically

analyzing information to understand and evaluate specific policies in particular contexts is likely to lead to enhanced knowledge about desirable policy choices to attain specific MDG goals and other goals in varying contexts in different countries in Africa, Asia and Latin America.