Transparency, Accountability, and Global Education Policy

Transparency and Accountability in Public Services

• Inherent versus Instrumental Value of Information:

  • Transparency and information access are inherently beneficial in promoting learning and opening institutional veils.
  • Greater transparency does not inherently translate into improved material outcomes, such as public healthcare service quality.

• The Uganda Healthcare Report Card Experiment (Björkman & Svensson):

  • Experimental Setup: Conducted in Uganda in the early 2000s; randomly assigned citizen report cards distributed at public meetings across $25$ health clinics, compared against $25$ control clinics that did not receive the intervention.
  • Reported Outcomes: Substantial across-the-board improvements were observed across nearly all indicators:
    • Wait times for patients to see medical staff decreased.
    • Healthcare worker attendance rates increased noticeably.
    • Physical facility cleanliness improved visibly.
    • Prescription drug leakages dropped (mitigating diversion into the extensive black market for pharmaceuticals operating in Sub-Saharan Africa).
    • Vaccination coverage rose significantly.
    • Child mortality for children under five declined by $\frac{1}{3}$, amounting to an estimated $515$ children saved within the clinic catchment areas.
  • Methodological Limitations: Sample size was small ($25$ treatment clinics vs. $25$ control clinics), representing the bare minimum required to identify very large statistical effects. The intervention was bundled (combining citizen report cards and community meetings), making it impossible to isolate which specific mechanism produced the result or whether reporting to administrative superiors drove outcomes.

• Replication and Factorial Evaluation (Raffler, Poehler, & Parkinson):

  • Design: Implemented a larger sample replication in Uganda approximately ten years after the initial evaluation, using an unbundled $2 \times 2$ factorial design:
    • $\frac{1}{4}$ received both citizen report cards and public meetings.
    • $\frac{1}{4}$ received public meetings only.
    • $\frac{1}{4}$ received report cards only.
    • $\frac{1}{4}$ served as pure control.
  • Findings: The combined intervention arm showed modest improvements in healthcare process quality, but these did not translate into distal health outcomes such as child mortality reductions or improved vaccination rates.
  • Explanations for Divergence: The initial effect sizes were exceptionally large relative to standard public health interventions; baseline conditions in public systems shifted over the intervening decade; human subjects learn, update behavioral strategies, and adapt to monitoring mechanisms over time.

Political Scorecards and Voter Behavior

• Uganda Parliamentary Scorecard Study (Humphreys & Weinstein):

  • Intervention Design: Comprehensive performance scorecards were compiled for Ugandan Members of Parliament (MPs) by a research team including Adam Harris, measuring plenary attendance, committee participation, and legislative bill sponsorship.
  • Coefficient Plot Interpretation: If confidence intervals cross zero ($0$), the difference between treatment and control is statistically indistinguishable from zero; intervals entirely excluding zero indicate statistically significant effects.
  • Electoral Intentions vs. Politician Effort: Scorecard dissemination shifted voter intentions as hypothesized: MPs with scores above the norm gained voter support among past supporters and non-supporters alike, whereas MPs with poor scores lost voter support across both groups.
  • Behavioral Adjustment of Politicians: Low-performing MPs exhibited no subsequent performance improvement. Instead, politicians leveraged charisma and public meetings to reframe performance metrics (e.g., claiming presence in the local constituency was superior to attending committee meetings in the capital). Because of politician capabilities and local patronage dynamics, the informational intervention failed to affect actual re-election outcomes.
  • Institutional Context: Uganda operates as an autocratic regime where parliamentary elections remain competitive and opposition MPs win seats, whereas presidential elections face significant irregularities.

• Vietnam National Assembly Scorecards (Malesky, Schuler, & Tran):

  • Context and Assignment: Implemented in Vietnam, an authoritarian context with less electoral competition than Uganda. Scorecards evaluating delegate performance were randomly assigned to approximately $\frac{1}{4}$ to $\frac{1}{3}$ of delegates in the National Assembly.
  • Perverse Outcomes: Monitored delegates significantly reduced parliamentary effort and experienced decreased re-election rates, driven largely by failure to secure renomination from party screening bodies. The transparency intervention caused net harm to delegates, including those with previously high performance.

The Metaketa Initiative and Multi-Country Information Interventions

• Architecture of the Metaketa Coordinated Trials:

  • Organized by a consortium including Macartan Humphreys, coordinating simultaneous field experiments across six sites in five countries (Brazil, Mexico, Uganda with two independent teams, Burkina Faso, and Benin) between 2014 and 2017.
  • Studies utilized pre-harmonized interventions, common outcome metrics, and pooled statistical analysis to evaluate whether supplying objective politician performance data influences vote choice in elections.

• The Uganda Mobile Text Intervention Arm:

  • Recruited an experimental sample of approximately $18,000$ subjects via mobile telephone numbers across rural villages, achieving a sample size roughly ten times larger than comparable project arms.
  • Monitored and reported specific local service metrics prior to local council elections (honeymoon elections held shortly after presidential contests):
    • Local government budget irregularities and audit discrepancies.
    • Road smoothness and roughness measured via tablet motion sensors.
    • Water access parameters, including walking distance, pump waiting times, and user fees.
    • Health clinic operational indicators matching the Björkman and Svensson metrics.
    • Primary school literacy and numeracy metrics collected in partnership with the civil society organization Uwezo, which uses volunteers to administer household-level literacy and math diagnostics to children.

• The India Project Termination:

  • The proposed experimental arm in India intended to disseminate candidate criminal histories to voters (where approximately half of all sitting MPs possess criminal records).
  • Politicians utilized state power and police intervention to shut down the project within days of launch, illustrating the capacity of entrenched elites to suppress accountability information.

• Empirical Results of the Metaketa Consortium:

  • The pooled meta-analysis across all six studies yielded a precisely estimated null effect ($0$), demonstrating that objective performance information delivered by researchers does not systematically alter aggregate voting behavior.
  • In the large Uganda sample, high statistical power permitted detection of very small, symmetric effects at the district council level: voters rewarded incumbents with fewer budget irregularities and penalized those with higher irregularities.
  • Zero effect was observed for district chairpersons (governors), about whom voters already possessed substantial information and established partisan attachments.

Theoretical and Intrinsic Value of Education

• Critique of Transactional Education Models:

  • Viewing education purely as a financial transaction (tuition traded for grades, which are then traded for employment) mischaracterizes human capital acquisition.
  • In public systems, state subsidies offset costs (e.g., the State of Texas subsidizes public university tuition by approximately $\frac{2}{3}$, meaning students pay roughly $\frac{1}{3}$ of true educational costs).
  • Grades correlate imperfectly with actual learning; students may acquire perfect marks via rote execution without conceptual understanding, or receive lower grades in rigorous environments while mastering underlying principles.

• Independent Economic and Social Returns:

  • Labor Diversification: Education facilitates employment outside prevailing low-skill village occupations (e.g., crafts, healthcare, primary education).
  • Productive Efficiency: Educated agricultural workers introduce innovations in cultivation and harvesting techniques, raising total factor productivity.
  • Intergenerational Spillovers: Maternal educational attainment strongly predicts reductions in infant mortality and improvements in child nutrition and health habits.
  • Civic Agency: Literacy and education enable voters to interpret governmental decision-making and hold representatives accountable.

• The Intrinsic Worth of Knowledge:

  • Learning possesses non-instrumental, hedonic, and cognitive value rooted in human curiosity (a trait in which humans rank in the $99^{\text{th}}$ percentile relative to other species).
  • Acquiring knowledge alters internal preferences, core values, and epistemic beliefs independently of external economic payoffs.

Systemic Obstacles, Absenteeism, and Policy Interventions

• Teacher Labor Markets and Institutional Design:

  • Teacher efficacy relies heavily on instructional awareness ("withitness") and attention management, qualities that standard colleges of education frequently fail to screen or develop.
  • Compensation structures dictate applicant pools. In Finland, national policy aligns teacher salaries with those of physicians, lawyers, and engineers, attracting top academic talent and securing high educational rankings worldwide. In contrast, low compensation and high demands in systems like the United States produce severe retention challenges.

• Physical and Structural Deficits in Education Systems:

  • Infrastructural deprivation—such as absent desks, inadequate study spaces, lack of books, or rodent infestations—undermines instructional efficacy.
  • Centralized, rigid curricular frameworks (e.g., long daily schedules from 8:00 AM to 5:00 PM or 6:00 PM) create uneven learning outcomes across public and private tiers.
  • Historical legacies of education systems, originally designed by elites for elite reproduction, perpetuate institutional exclusion.

• Parental Perceptions and the Educational "Lottery Ticket" Model:

  • As highlighted by Abhijit Banerjee and Esther Duflo, impoverished parents often conceptualize schooling as an all-or-nothing lottery. When asked to construct collages representing the purpose of schooling, parents predominantly select luxury items (automobiles, diamonds, yachts).
  • Parents frequently believe that secondary education yields exponential labor market returns relative to primary schooling, an assumption contradicted by empirical evidence showing steady, roughly linear returns per additional year of schooling.
  • This misconception causes families with multiple children to invest scarce resources into a single selected child perceived as the "smartest," while withdrawing siblings from school to supply immediate household labor.

• Cost-Effectiveness of Schooling Interventions (Abdul Latif Jameel Poverty Action Lab / J-PAL):

  • Informing Parents on Returns to Education: Informing parents about the measured economic returns of schooling generates the largest gain in school participation per $\$100$ spent due to near-zero implementation costs.
  • Child Deworming: Inexpensive pharmacological treatment yields massive gains in school attendance per dollar spent.
  • School Feeding Programs: Providing meals increases attendance years but incurs high baseline costs, as subsidies fund meals for students who would attend regardless.
  • Conditional Cash Transfers (CCTs): Pioneered in Mexico with randomized evaluations, CCTs disburse state cash transfers to mothers via debit cards conditional on child school attendance and clinic vaccinations. While highly effective at poverty relief and attendance, CCTs exhibit low marginal cost-effectiveness in boosting schooling alone because substantial funds are transferred to inframarginal families whose children would already attend.
  • Monitoring Absenteeism: Installing tamper-proof cameras to verify teacher presence at the start and end of the instructional day reduced teacher absenteeism from $42\%$ to $21\%$ (a $21$ percentage point reduction) in informal single-teacher schools, resulting in measurable test score gains.

The Learning Crisis: Schooling Versus Actual Learning

• Pritchett's Field Observations in Uttar Pradesh:

  • Lant Pritchett documented a diagnostic intervention in the Indian state of Uttar Pradesh (population between $100\,\text{million}$ and $200\,\text{million}$) where an NGO evaluated basic numeracy and literacy among rural children before convening a public village assembly.
  • Children aged $11$ to $12$ in the third and fourth grades could neither recognize the letters of their native alphabet nor identify script directionality after years in physical attendance.
  • At the village meeting, an agricultural laborer confronted the local headmaster, stating that he had sacrificed his life working like a donkey in the fields on the promise that education would alter his child's trajectory, only to discover his fifth-grade son could not read and faced the exact same fate.
  • The school headmaster publicly dismissed the parents, stating that "donkeys come from donkeys" and that sending untalented children to school naturally yielded untalented outcomes. This response illustrated complete power insouciance, as the headmaster's pay, civil service tenure, and career were insulated from parental oversight.

• Empirical Deficits in Developing Country Education Systems:

  • Schooling Ain't Learning: Pritchett's findings demonstrate that increasing years of school enrollment fails to produce proportional literacy or numeracy gains.
  • Arithmetic Deficits: Mastery of basic double-digit vertical addition tops out around $70\%$ by grade five.
  • Conceptual Failures: On ruler measurement assessments requiring students to determine the length of a pencil displaced by $1\,\text{cm}$ (running from $1\,\text{cm}$ to $6\,\text{cm}$, true length $5\,\text{cm}$), the vast majority of students incorrectly guess $6\,\text{cm}$. This error persists largely unchanged between grade four and grade eight, reflecting reliance on rote memorization inherited from colonial-era schooling systems rather than conceptual problem-solving.
  • Cross-National Learning Distributions: In low- and middle-income countries, the proportion of students reaching the top $10\%$ of global learning distributions is near zero, whereas in South Korea it reaches nearly $20\%$ and in Denmark it sits around the expected $10\%$.
  • Systemic Versus Household Determinants: Disaggregating student performance by wealth reveals that children from the top $20\%$ and bottom $20\%$ of household income within lower-income nations track each other closely; cross-national differences in learning are driven by total education system performance rather than individual household wealth.

• Pathways for Educational System Reform:

  • Banerjee and Duflo emphasize that teaching basic functional skills requires relatively short teacher training when matched with appropriate instructional mindsets.
  • Curricula must depart from rigid, factory-style colonial architectures toward self-paced instructional models (such as Montessori methodologies) and remedial instruction targeted to student learning levels.
  • Mobile smartphones and cellular connectivity present scalable platforms for personalized software instruction, though field experiments reveal that without deliberate structuring, children utilize digital devices primarily for entertainment rather than pedagogical growth.

Health Interventions: The Deworming Controversy

• The Kremer and Miguel Primary School Deworming Evaluation:

  • Conducted by Michael Kremer and Edward Miguel in Western Kenya, evaluating mass pharmacological treatment of intestinal helminths (worms) administered through primary schools.
  • Initial Findings: Parasite treatment generated substantial, highly cost-effective reductions in school absenteeism.
  • Long-Term Labor Market Returns: Longitudinal follow-up surveys tracked subjects into adulthood, reporting that an additional year of deworming exposure in childhood raised adult earnings by $20\%$ to $30\%$.

• Methodological Critiques and External Validity:

  • Epidemiological Pushback: Medical researchers and epidemiologists contested the magnitude of the findings, arguing that the nutritional mechanisms could not credibly yield employment and wage gains of that magnitude.
  • Randomization Procedures: The experimental design utilized alphabetical sorting of school names to assign treatment and control status rather than computerized pseudo-random generation, a departure from contemporary experimental standards.
  • Contextual Specificity: The intervention occurred in an environment characterized by exceptionally high baseline parasite burdens in Western Kenya; the estimated treatment effects fail to generalize to settings where helminth prevalence is modest or low.

Questions & Discussion

• Question / Objection on the Value of Education for Rural Laborers:

  • Query: What rational justification exists for sending a child to school when local labor opportunities are entirely confined to manual agricultural labor, such as cane cutting in the Dominican Republic, where parents and grandparents have worked identical jobs?
  • Response: Schooling carries returns beyond manual wage rates in a single industry. First, village economies include non-agricultural roles (e.g., haircutting, craft production, teaching, healthcare) that require literacy and numeracy. Second, educated agricultural workers apply operational and agronomic innovations that improve output. Third, maternal education directly improves infant survival and family health management. Fourth, civic competence enables citizens to evaluate public policy and leadership.

• Student Contributions on Student and Teacher Motivation:

  • Prompt: What mechanisms successfully resolve pupil and instructor absenteeism in under-resourced schools?
  • Student Response (Victoria): School attendance requires actionable accountability. Providing school nutrition (free meals) provides a direct physical incentive to attend. For instructors, raising base salaries while conditioning payment on verifiable daily attendance—docking wages or terminating employment for unexcused absences—mirrors standard professional labor contracts.
  • Student Response (Lou): Structural factors within centralized systems (such as in France) create severe disparities. Daily requirements spanning 8:00 AM to 5:00 PM or 6:00 PM, dilapidated facilities, lack of seating, and hygiene issues undermine teacher and student motivation regardless of individual instructor competence.