Experiences with Vouchers for Basic Education: A PrincipalAgent Perspective

1
Experiences with Vouchers for Basic Education A
Principal-Agent Perspective

• Varun Gauri, DECRG
• October 17, 2002

2
The Archetypal Education System

• Centralized management and hiring
• General revenue finance
• Direct payment of operating costs
• Students assigned to schools
• Remuneration and promotion based on negotiated,
  experience-related criteria

3
Voucher Programs

• Parents choose schools public / private
• Intense incentives on enrollments
• Managerial autonomy to respond to demand

4
Three Simultaneous Reforms

• Autonomy without intense incentives (Chicago, El
  Salvador)
  
• Choice without autonomy (Minnesota, Detroit, de
  facto developing countries)
  
• Intense incentives without choice (high stakes
  accountability, merit pay)
  

5
Principal-Agent approach to vouchers

• The state is the principal
• The schools are the agents
• Enrollments are the sole measure of school
  effort
  
• School payments ? ?(e x ?y)

6
The Estimate of Effort

• Prima facie plausibility
• Potential systematic bias conflate student body
  with quality of education, and latter with effort
  
  
• Noise remoteness, birth rate, migration, returns
  to education, cultural expectations on who should
  be in school

7
Evidence on Bias and Noise

• Small impact in US studies (Myers 2001, Rouse
  1998), Chile (Hsieh and Urquiola 2001, McEwan and
  Carnoy 1999)
  
• Students flee minority schools in New Zealand
  (Fiske and Ladd 1999, 2000) and Sweden (Daun
  2002)
  
• School selection in Chile (Gauri 1998, Espinola
  1995) and New Zealand

8
Productivity of More Effort

• Limited if teacher training inadequate, textbooks
  unavailable, rules restrictive, students
  malnourished, teachers politicized, difficult to
  recruit teachers Cote dIvoire (Michaelowa 2001,
  World Bank 1999)
• Substantial if surplus capacity in private
  schools Colombia (Angrist et al 2001)

9
Relative compensation across activities

• Activities rewarded at lower rates will receive
  little or no effort
  
• Teacher training Cote dIvoire
• Curricular innovation Chile, New Zealand, UK
  (Walford 2002)

10
Value of Monitoring

• Rewards for exaggerating enrollment rates,
  loosening academic standards
  
• Anecdotal evidence in Chile, Cote dIvoire
• Evidence to the contrary in Colombia

11
Risk Averse Agents

• The greater the risk aversion, the higher the
  welfare cost from intense incentives
  
• Teachers lobby not to be fired, transaction costs
  of selling schools
  
• Few schools closed in Chile, New Zealand

12
Bottom Line

• Bias and noise in the estimate of effort
  include co-variates and controls
  
• Need monitoring, risk-averse agents, limited
  returns to effort vouchers are costly
  
• Useful over range where marginal returns to
  effort are high poor students, excess capacity
  in private schools
  
• Useful for disadvantaged students (?)