Is School-Based Financial Education Effective?

The Ministry of Education (MINEDU) partnered with the Superintendency of Banks and Insurance (SBS) and the Center of Studies (CEFI) of the Peruvian Association of Banks to develop a pilot that provided financial education to high school students. Together, they developed student workbooks for each of the last three grades in high school (equivalent to ninth, tenth, and eleventh grades in the United States) as well as a teacher's guide. The team also designed and implemented a 20-hour teacher training plan divided in 5 sessions, which included a training component on the financial literacy contents (4 sessions) as well as a pedagogical one (1 session). MINEDU encouraged teachers to attend the training sessions conducted before the school year started. School principals were requested to facilitate teacher participation in the training and participants received both a transport subsidy (mostly in kind) and a full meal during the workshop. The content of the students' workbooks varies by grade and the content was delivered during the regular classes of the course "History, Geography, and Economics'' (HGE).

2016-08-01

2016-12-20

financial literacy, socioemotional skills, consumption and saving habits, credit behavior, academic outcomes

Financial literacy is measured with entry and exit exams designed taking into account the content of the workbooks. Socioemotional skills is measured through self-rated scales of conscientiousness, self-control, impulsiveness, time-preferences, and risk aversion. Conscientiousness, which is closely related to deliberative thinking, was measured using the Big Five Scale for this attribute [Pervin and John, 1999]. Self-control is measured by Tangney et al. [2004]’s scale, while impulsiveness is measured by the Barratt Impulsiveness Scale (BIS-11) [Orozco-Cabal et al., 2010]. Time inconsistency is defined as in Ashraf et al. [2006]. Although preferences and personality traits are
self-reported, they are measured relying on extensively tested scales that are specifically designed to be self-rated.

Consumption and saving habits are measured also through self-reported survey data.

Credit behavior up to three years after the intervention was provided by EQUIFAX, a private credit bureau that concentrates loans data from almost all lenders in the Peruvian credit market. EQUIFAX's data captures an individual's credit standing at the time in which she is searched, both in terms of their positive and negative records. Loan records report the current balance by default status and source of the funds, distinguishing between loans from banks and financial institutions supervised by the SBS from those that come from formal but non-regulated institutions such as microfinance NGOs and cooperatives. In addition to loan balances, the credit bureau's data also captures late or skipped payments of service bills and credit cards.

grades and grade progression

MINEDU's academic records provide data for all high school students enrolled in any of the 300 schools of the sample in 2016. These data contain individual-level information on cumulative grades by course and grade progression at the end of three consecutive academic years, one before and two after the intervention was launched. Access to these records offers the possibility to estimate treatment effects on academic outcomes among students in the sample with survey data as well as among the universe of students in the experimental sample of schools (~60,000 at baseline).

The universe of interest was restricted to full-day public high schools in urban areas in six regions of the country: Lima and Callao, Arequipa, Piura, Junin, Puno, and San Martin. Due to logistic reasons, the universe was further restricted to Local Education Management Units which were close to cities and with a high number of schools under its supervision. After imposing some additional restrictions (directly managed by the MINEDU, single-grade schools, and number of students by grade above the 5th percentile and below the 95th percentile), the final universe included 308 schools. The restricted universe was stratified by region. Schools are paired by their similarity within each of the six strata. This procedure returns 150 matched pairs, yielding an experimental sample of 300 schools.

Randomization done in office by a computer

Schools

Yes

300

20,000

10,000 (150 schools in each treatment arm)

To establish the number of schools required for the evaluation, power calculations were performed with the following parameters: significance level of 0.05, statistical power of 0.8, minimum detectable effect of 0.1SD, R2 of the outcome equation of 0.1, intra-cluster correlation of 0.1, and a sample size of 40 students per grade. Under these assumptions, about 300 schools were required, 150 in each treatment arm.