Human Capital at Home: Evidence from a Randomized Evaluation in the Philippines

Introduction

We conduct a randomized controlled trial in the Philippines with over 1,000 households to promote active parental engagement in their child’s education. We collect detailed learning outcome data in three waves over the course of a year and a half. We evaluate an intervention called Family Academy which provided numeracy and phonics educational instruction for children ages 3 to 5 for 10 weeks. Households enrolled in the program earned just about $2 to $3 per day, substantially below the poverty line in the Philippines.

The program involved an in-person coach who visited households for two 45-minute sessions each week to conduct learning sessions with the child as well as promote positive parenting in general. One of the sessions focused on math, and the other on phonics. Example activities included children reciting the alphabet through song or recognizing numbers by counting fingers. The activities were targeted at developing mathematical skills such as recognizing colors, shapes, and numbers, as well as phonics skills such as identifying letters, sounds, and reciting the alphabet. The sessions involved parents as active participants, with a ‘tell-show do’ approach: the coach first introduced the learning activity, then demonstrated it, and finally invited the parent to lead the activity themselves.

We find that Family Academy improves learning outcomes substantially, by 0.44 to 0.46 standard deviation gains in basic numeracy and phonics skills. These learning outcome gains in math have some persistence over time with 0.18 standard deviation impacts over a year after the intervention with high statistical significance (p-value = 0.005). Phonics skills also persist with 0.18 standard deviation gains, although statistical significance is lower (p-value = 0.136). These effects are considered large in a literature where over half of education interventions have no positive effect (Angrist et al. 2020), and the median effect size is 0.1 standard deviations (Evans and Yuan 2022).

We find parents are highly engaged in the educational instruction of their child, revealing that parents even in low literacy settings can effectively engage in education. However, alternative parental behaviors and investments in education are not affected, such as positive parenting, suggesting that most of the learning gains are realized through the channel of direct parental engagement in instruction.

This RCT was funded by the Global Innovation Fund.