Getting A’s on the three F’s: The Educational and Fiscal Opportunities in Decentralization

1. Getting A’s on the three F’s: The Educational and Fiscal Opportunities in Decentralization Lant Pritchett Varad Pande

2. Outline (but with several big “asides”) Decentralization and Primary Education Reform in India: The Two Questions An Analytical Approach to Decentralization – Using First Principles Fiscal issues A ‘Modest Proposal’ for Reform

3. Decentralization And Primary Education Reform in India India is Trying to Decentralize Indian Needs Primary Education Reform • 73rd and 74th Amendments to the Constitution devolve political power to local governments • States modify Acts and hold elections to local governments • Functions in XI and XII Schedules devolved to local governments • Continued low enrolment, completion and achievement despite central and state schemes (and Sarva Shiksha Abhiyaan) • Much lower educational indicators for poor, who suffer disproportionately Question 1: How to Do Education Delivery in this Decentralizing Context Question 2: Is Well-Designed Decentralization the Right Model for Primary Education Reform?

4. 10 Fiscal Administrative Political 5 0 UP Rajasthan MP AP Kerala Poland Chile Colombia Decentralization in India Has to date been Unbalanced and Incomplete Unbalanced Decentralization in India in an International Context ‘Unbalanced’ Decentralization • Political Decentralization has happened (election of PRIs)but… • Administrative and Fiscal Decentralization remains weak Decentralization Index (0-10) Source: World Bank (1999)

5. Decentralization and Education • Decentralization is no panacea—there are enormous risks to decentralization that could worsen service delivery • Badly designed decentralization also has fiscal risks (and missed opportunities) • Getting the “three F’s” properly aligned is difficult

6. Decentralization is Neither a Necessary nor Sufficient Condition for High Quality Education Correlation between Decentralization Indicators and Primary Education Outcomes Source: Analysis based on data from Governance and Service Delivery Indicators Dataset, Pippa Norris, Kennedy School of Government, (2004) No evidence of high correlation between decentralization and education outcomes in cross-country analysis

7. Assign “vertical shares” (revenues) Central State Local Assign functions (across tiers) Education Health Roads Water The problem that a State Finance Commission Faces Functions Finance ?

8. A plausible sounding procedure: • Assign functions (across tiers of PRI) by sector and activity (including mandates for minimum standards—e.g. universal primary education) • Cost those functions at the standards demanded/expected • Calculate the vertical share to transfer adequate resources to each tier to carry out its functions—make revenue assignments

9. Our Analytical Approach To Effective Decentralization First Principles of Public Financeand Accountability Functional Allocation and Recommendations Status quo Analysis Unbundling What are the key functions and activities in Primary Education? Who provides these functions and activities today in India? How to decide who should provide which functions and activities? What changes are required to the current system of service delivery? • Disaggregate into functions and activities: • Logical, mutually exclusive and exhaustive (MECE) functional classification • Consistent with Indian public administration delivery system • Analyze de jure functional allocation of each tier by studying existing legislation • Analyze de factofunctional allocation (on-the-ground picture) by conducting surveys and interviews • Based on this systematic analysis, propose changes to current system • Develop a detailed proposal for reform of primary education that draws on the findings of our analytical approach • Use First Principles of Public Financefind optimal allocation from public finance perspective • Use First Principles of Accountability to address the accountability issues

10. Step 1: Unbundling Question: What are the Key Functions and Activities in Primary Education? Functions and Activities in Government Primary Schooling in Rural India Output: A Mutually Exclusive and Exhaustive Classification of Primary Education into Functions and Activities

11. Step 2: Status Quo Analysis Question: Who provides these functions and activities today? Functional Allocation in Primary Education – DE JURE Lessons • Current system is highly centralized with little role for PRIs / User Groups • Schools / teachers have little or no autonomy… or accountability Functional Allocation in Primary Education – DE FACTO

12. Step 3: Use First Principles: Public Finance What are the First Principles of Public Finance? Principle Explanation Implication • Activities with significant economies of scale should be done by a higher level of government Unit Cost of production declines as scale of production increases Economies of Scale • Activities with significant externalities should be done at a higher level of government, so that the ‘external’ effect can be ‘internalized’ in the system The actions of one agent affects other agents Externalities • Equity may imply financial support to education at a large enough geographic scope to allow for redistributive transfers to equalize across smaller units Need for equitable spread in inputs, process or outcomes Equity • The more heterogenous the demand for the activity is likely to be, the more locally it should be done Variation in local needs and preferences between regions Heterogeneity of Demand

13. Step 3: Use First Principles – Public Finance Economies of Scale What is the size of ‘optimally sized catchments area’ of a service provider to jurisdiction? State Population: 30-80 million Zilla Parishad Population: 800,000-2 million Block Panchayat Population: 60,000-600,000 Gram Panchayat Population: 2,000-20,000

14. Step 3: Use First Principles – Public Finance Economies of Scale Optimal Catchments Areas for LEVELS of Education Universities State Population: 30-80 million • Whereas the optimal catchments area for a primary school is at village level (GP jurisdiction)… • …the optimal catchments area for a University is district / state level (State jurisdiction) Zilla Parishad Population: 800,000-2 million Block Panchayat Population: 60,000-600,000 Gram Panchayat Population: 2,000-20,000 Secondary Schools Primary schools

15. Step 3: Use First Principles – Public Finance Economies of Scale Optimal Catchments Area for FUNCTIONS of Primary Education Setting learning achievement standards Monitoring and Evaluation (process and outcome) State Population: 30-80 million • Whereas the optimal catchments area for Operations is at village level (GP jurisdiction)… • …the optimal catchments area for a Setting Learning Standards is state level (State jurisdiction) Zilla Parishad Population: 800,000-2 million Planning Block Panchayat Population: 60,000-600,000 Asset Creation Gram Panchayat Population: 2,000-20,000 Primary School Operation

16. Step 3: Use First Principles – Public Finance Economies of Scale Functional Allocation according to Economies of Scale Criterion`

17. Step 3: Use First Principles – Public Finance Equity Equity Concerns are Important Gross Enrolment Rate - UP (2003-04) Gross Enrolment Rate - Kerala (2003-04) Source: DPEP MIS Database http://www.dpepmis.org/webpages/edatabase.htm • Major inequity / variations in primary education levels exists between districts within states • This creates equity rationale for centralization of education delivery • This does not imply all functions in primary education should be done by higher levels of government…. • As long as Standard Setting and M&E are done by higher level of governments, AND redistributive equalizing fiscal transfers are done, goal of equity can be achieved Equity Doesn’t Require Centralization

18. Step 3: Use First Principles – Public Finance Equity Functional Allocation according to Equity Criterion

19. Step 3: Use First Principles – Public Finance Functional Allocation according to First Principles of Public Finance Standard Setting and Monitoring by State Government, and Asset Creation and Operation by low-level PRIs is effective

20. Step 3: Use First Principles - Accountability What are the First Principles of Accountability? Discretionary Transaction-Intensity Inferring Performance Activities that require decisions to be made using individual judgment in local context Activities that require several repeated transactions at the local level Where can performance be better inferred – locally or technically Examples Does the service require decisions made in local context? (Discretionary) Does the service require many local transactions? (Transaction Intensive) Activities that are Discretionary, Transaction-Intensive and Locally Observablerequire local input for accountability Implication

21. Aside on Accountability: Quality of Government Primary Schooling in India is very low…especially for the Poor Framework for Judging Quality Enrolment Completion Achievement Percent of 6-14 yr olds currently in school (1998-99) Percent of children with Inadequate Primary Learning Achievement for Math of Age for Class 5 or in Grade 5 Percent of 15-19 yr olds completed at least Grade 8 (1998-99) Average78.8% Not reaching Grade 5 81.7% Scoring < 50% in Grade 5 (Govt School) Average49.6% 67.6% 69.7% 53.8% Note: Framework adapted from Warwick and Reimers (1995); Enrolment and Completion data from ‘Educational Attainment and Enrollment Profiles’,Filmer (2004); test scores analysis for government and government aided schools only based on data from Singh, Jain, Gautam and Kumar (2005) …+ Anecdotal Evidence of low quality (Pratichi Trust, PROBE Report, etc.)…+ Evidence of student migration from government to private schools

22. AOA: Learning achievement from Baseline study in AP Mechanical learning to a large extent. Single digit addition : 18% of students (for example 3+3) Single Digit Subtraction: 12% of students (for example 9-2) Counting Kites : 54% of Students Highest percentage in any class is 77% of Students (which is low) Best performances are in number concepts in maths and in following instructions and recognising letters/sounds associated in language

23. AOA: Learning by age * Data based on the testing of 17,608 children in 4 random blocks in the District of Jaunpur, UP. Child tests are weighted by total number of children in village divided by number of children tested in village.

24. AOA: The low quality is caused by ‘systemic’ and not ‘managerial’ failures; and thus requires an institutional solution Evidence Teacher Absence Rates by Country % “Best practice” level of absence Teacher Absenteeism Country Source: Kremer et al (2004), pg 9 Level of Parental Dissatisfaction (West Bengal) Grades given by Parents to Child’s school (USA) Level of Parental Satisfaction Source: The Pratichi Education Report (2002); Berliner and Biddle (1994) pg 113 Other Evidence • ‘Non’ Teaching: “Teaching practices have been reduced to a minimum…it has become a way of life” (PROBE) • Growth of Private Schooling…which meets the differentiated(quality) demand…even among the poor (Tilak et al.)

25. Result: People voting with their feet (and pocketbooks) out of the sectors:

26. AOA: What is ‘Accountability’? – Demystifying the Elements of the Accountability Relations There are Five Features to Any Accountability Relationship Feature What Example 1: Buying a Sandwich Example 2: Going to a Doctor Delegation You give a task to the accountable ‘agent’ • You ask for a sandwich • You go to the doctor to be treated Financing You give the ‘agent’ the money to do the task • You pay for the sandwich • You pay the doctor for the treatment Performing The ‘agent’ does the assigned task • The sandwich is made for you • The doctor treats you to try cure your ailment Informing You find out how well the ‘agent’ has done the work • You eat the sandwich which informs you of its quality • You see if you are feeling better – you assess the performance of the doctor Enforcing You reward good performance and punish bad performance • You choose whether to buy a sandwich from the seller the next time, affecting his profits • You go to him next time (if he was good) or choose to go somewhere else if not

27. Systemic failures are rooted in Inadequate Accountability Relationship 1: VOICE – The Long Route, first leg Policy Makers • Citizens can elect Policy Makers based on quality of public services they receive (VOICE)… • But, Citizens… • may vote along ethnic / ideological lines (DELEGATION) • may not be able to attribute service to appropriate Policy-Maker (INFORMING) • may think Policy-Maker tenure is too short (ENFORCING) VOICE Citizens

28. Systemic failures are rooted in Inadequate Accountability Relationship 2: COMPACT, long route second leg Policy Makers • Well-meaning Policy Makers can influence Service Providers to improve service delivery performance (COMPACT)… • But, Policy Makers… • may not be able to specify exact contracts (DELEGATION) • may not be able to monitor Service Providers (INFORMING) • may not be able to alter incentives of Service Providers (ENFORCING) COMPACT Service Providers

29. Systemic failures are rooted in Inadequate Accountability Relationship 3: CLIENT POWER – The Short Route • Clients can play major role in tailoring service-mix to their local needs and monitoring Service Providers (CLIENT POWER) • But, Clients… • often do not have choice (lack of competition) • often cannot infer provider effort (INFORMING) • often do not have powers to discipline bad Service Providers (ENFORCING) Citizens Service Providers CLIENT POWER

30. Pratichi Trust (A. Sen’s NGO) report on West Bengal: “According to you what are the main problems of primary schooling?” The Sabhapati of a Panchayat Samiti in Puruliya needed no time to reply: “Teachers do not teach.”

31. AOA: Accountability at the local level…Pratichi Trust report: We, the villagers, forced him to come to school at 10:30 and made him stay in the school until 3:30…” Says a villager in Puruliya

32. AOA: local client power? “We, the villagers, forced him to come to school at 10:30 and made him stay in the school until 3:30…” Says a villager in Puruliya Result of this attempt at local accountbility? “He slept, got his body massaged by the children, and read the newspaper. How long can you watch him leaving aside your own work?”

33. Systemic failures are rooted in Inadequate Accountability The Three Relationships of Accountability in Service Delivery Policy Makers VOICE COMPACT Citizens Service Providers CLIENT POWER Source: World Development Report, 2004

34. AOA: The key missing accountability relations in Primary Educationbetween parents and teachers are Informing and Enforcing Feature of Accountability Is this the Key Problem? Evidence Delegation Is it important for your child to be educated? Somewhat: Diffused objectives but Parents want to delegate (i.e., send children to school) Source:PROBE Report, pg 14 Yes 89% Yes 98% Average Salaries as multiple of Income per capita by Region Average = 6.1 Note: Uttar Pradesh figure is for 1995-96; all others are quoted from Carnoy and Welmond (1996); Source: Mehrotra and Buckland (2001) Financing 8-13.5 No, teachers are paid very well relative to other countries Informing % Households Aware of Existence and activities of Community Structures Yes, no usable info on class performance; parents not actively involved in school / teaching matters Source: CAG Report (2001) Enforcing Yes, parents have little or no powers to discipline or reward teachers • Kremer et al (2004) find only one case of teacher dismissal ever in 3000 government schools surveyed • No monetary incentives possible for high performing teachers under current government teacher terms of employment Teacher Absence Rates by Indian States All India Average 25% Performing % Yes, many teachers are not performing adequately

35. AOA: Alternative Explanations for Low Quality are Unlikely to be Sufficient Enrollment rate among 5-11 year olds 1 % Budget expansion approaches don’t seem to have worked in India – e.g.: District Primary Education Program (DPEP) Lack of Adequate Public Expenditure DPEP Districts Non-DPEP Districts (with same criteria) Source: Jalan and Glinskaya (2003) 2 Presence and Demand for Child Labor • Child labor is not big enough – NSS (1993) shows only 8 percent • Most child laborers work as family labor, so their time is fungible • Even if child labor is ‘high’, direction of causality is not clear 3 Parents’ Attitude to Education in India Lack of Parent Interest Is it important for a boy to be educated? Is it important for a girl to be educated? Parents, even poor parents, seem keen to educate their children No Response 1% No 1% NoResponse 1% No 10% Yes 98% Yes 89% Source:PROBE Report, pg 14

36. End of long aside on accountability? • Publicly provided primary education is not performing well • Accountability is a key problem • Decentralization alone won’t improve accountability • But possibly decentralization informed by accountability analysis can improve services

37. Step 3: (back from long aside on accountability) Use First Principles - Accountability What are the First Principles of Accountability? Discretionary Transaction-Intensity Inferring Performance Activities that require decisions to be made using individual judgment in local context Activities that require several repeated transactions at the local level Where can performance be better inferred – locally or technically Examples Does the service require decisions made in local context? (Discretionary) Does the service require many local transactions? (Transaction Intensive) Activities that are Discretionary, Transaction-Intensive and Locally Observablerequire local input for accountability Implication

38. Step 3: Use First Principles –Accountability Functional Allocation according to First Principles of Accountability Discretionary, Transaction-intensive and Locally Evaluatable activities like Asset Creation and Operations should be decentralized to PRIs

39. Step 4: Optimal Allocation based on First Principles Analysis First Principles of Accountability First Principles of Public Finance Key Messages Functional Allocation in Primary Education –Based on First Principles Analysis • States do Standards Settingand Monitoring • PRIs assume responsibility for actual Operation • As much as possible as low as possible • Higher PRI tiers back-up on professionalism, technical

40. As discussed, Many Models of Education Decentralization have not worked well 1 Decentralize without unbundling • Indonesia Experience • Pushed all functions to district level (below provinces) • Bad monitoring systems means that little information about regional performance is available to stimulate competition or disseminate good or bad practices 2 Decentralize Non-Teacher Operations (But not Teacher Operations) • Latin America Experience • Pushed responsibility to local governments… • …But without enhancing operational control at local level (e.g.: Argentina) 3 Decentralize Functions, concurrently, without devolving F or F • India Experience (1995-2005) • Devolved functions (including primary education) to PRIs… • …But PRIs do not have adequate finance or control over functionaries to implement their mandate

41. But our Analysis Avoids these Pitfalls by Suggesting Countervailing Forces to build accountability via decentralization The Two Big Messages from Our Analysis 1 Strengthen Centre and State for Standard Setting and Monitoring • Consistent Standard Setting is Critical To Provide the Guiding Framework for Local Governments to Operate and Manage • Uniform Monitoring is Essential for Quality Control, Designing Rewards & Recognition Systems and Generating Credibility 2 Greater Operational Responsibility to PRIs, especially forTeachers

42. What we are not saying “just turn schools over to GPs/SMCs”—without clear “delegation” and without strengthening the “information” and “enforcement” this just pushes the problem lower. “Districts play no role” Districts play key role at the technical level in planning and pedagogical support

43. Here is where it gets interesting for “finance” • Suppose for a minute that a state government has decided to follow these recommendations and is going to devolve primary education to PRIs (for real). • What does the “finance” F corresponding to this “function” F look like? • It all depends on the “functionaries” F

44. Four alternatives: • Full-on voucher scheme (a la Chile)—equal transfer per (weighted) student to all schools, public and private (not discussed). • “Functionaries” remain a “state” problem so financing is dual (cash and in kind) • “Functionaries” are transferred to schools with a cash budget but wages/employment conditions fixed • “Functionaries” are transferred and “block grant” financing.

45. What is the key fiscal issue? • How does one compute the per child “standard unit cost” of a year of school (ignoring capital costs)? [That is, the formula for a block grant could have all kinds of weights—for girls, for SC/ST, for “marginality”—off of a standard unit cost] SUC=Teacher cost + non-teaching cost Teacher cost/child= (wage per teacher)*(number of teachers)/(number of children)

46. Teacher wages: the key parameter Desiderata: “Level and structure of compensation appropriate to attract, retain, and motivate teachers of desired quality.” • Structure is awful (for all goals) • Level of compensation?

47. Gov’t vs. organized pvt sector is high. Teachers vs. private sector is high. Existing teachers vs. “contract” or “alternative” teachers is high. International comparisons is high. Level of teacher compensation?

48. Pluses No political fight with teachers GPs/SMCs do control “operating” budgets and possibly some gains there. Minuses Accountability won’t work—no gains from decentralization No cost savings Pluses and Minuses of Alternatives: GPs/SMCs don’t H/F/A teachers, teachers are “assigned” and come “in kind”

49. Pluses Accountability might work at the local level Minuses: No fiscal savings No reallocation to more productive uses. With complete control and high wages corruption is inevitable. GPs/SMCs control teacher H/F/A, unit cost at current wages, wages fixed

50. Pluses Accountability might work at the local level Reallocation to productive uses (within education, between education and other uses?) Minuses: No fiscal savings to state GPs/SMCs control teacher H/F/A, unit cost at current wages, wages market determined