Digital Libraries for Families
Introduction
Kalil, Liu, Mayer, Rury, and Shah (2026) evaluate a digital library built to provide children from disadvantaged households access to a wider selection of books. Families received an electronic tablet preloaded with Children and Parents Engaged in Reading (CAPER), an application carrying 207 vetted picture books. Every other function of the tablet was disabled: families could not reach the internet, use other applications, or download anything. For comparison, the average child owns about 52 books at kindergarten entry at the 10th income percentile and 131 at the 90th, so the library raised low-income children’s book access to the level of their more advantaged peers.
The paper studies the program through an 11-month randomized controlled trial run in Chicago from December 2019 to October 2020. It recruited 607 children aged three to four from 13 subsidized preschools and randomized them, stratified by school, into four groups: a control group that received an activity book with crayons and stickers; a tablet group that received the CAPER tablet; a reminders group that received the tablet plus four text messages a week reminding the parent to read; and a goal-setting group that received the tablet plus four weekly messages setting and tracking a reading-time goal. Children’s language skills were measured at baseline and at the end of the intervention. The analytical sample is 289 families.
Access to the digital library alone raised children’s language skills by 0.29 standard deviations relative to the control group, significant at the 1 percent level, moving the average child from roughly the 40th percentile of the national distribution to about the median. The gains are concentrated among the children who began furthest behind: those with baseline CELF scores below the median improved by about 0.6 standard deviations, closing roughly 15 percent of the baseline language gap. The paper finds no statistically significant evidence that the text messages improved outcomes beyond the tablet and cannot reject equality of treatment effects across the three treatment arms.
The paper uses these results to estimate the MVPF of providing a family with a CAPER tablet.
Pays for Itself
Net Cost
The paper estimates the net cost to government of providing one CAPER tablet as the direct cost of the program less the additional tax revenue generated by the children’s higher earnings in adulthood.
The paper reports the direct cost as “$300 per CAPER tablet, plus some additional costs for digital books.”
The fiscal externality is the government’s share of the projected earnings gain described in the willingness-to-pay section below. Applying an average tax-and-transfer rate of 20 percent to the $15,987 present value of the additional lifetime earnings, the government recovers $3,197 per child.
The net cost of the components the paper quantifies is therefore $300 – $3,197 = -$2,897 per child.
Willingness to Pay
The paper measures willingness to pay as the present value of the lifetime earnings gains that the intervention generates for the children.
The input is the tablet group’s effect of 0.29 standard deviations on the CELF assessment, which the paper describes as the most conservative of its estimates. Using conversion factors from Anger et al. (2025) and Hanushek and Woessmann (2009), under which a one standard deviation increase in cognitive skills correlates with a 12 percent rise in earnings, a 0.29 standard deviation gain implies an expected 3.5 percent increase in earnings.
The paper applies that percentage to an annual salary of $29,741 (the median income for U.S. adults without a high school diploma, according to the 2020 U.S. Census), which gives an annual earnings gain of $1,040. Assuming a 30-year working period, a 5 percent discount rate, and no income growth, the present value of the lifetime earnings gain is $15,987.
The paper measures willingness to pay as the pre-tax present value of the earnings gain. Netting out the 20 percent tax-and-transfer share that net cost already credits to the government would give a post-tax willingness to pay of $12,790. The MVPF is infinite on either measure, as net cost is still negative and willingness to pay is still positive.
The families’ own valuation of the tablet and the digital library as an in-kind good is not counted, suggesting this may be a lower bound on the program’s total benefits to the beneficiaries.
MVPF Summary
With a positive willingness to pay and a negative net cost, the MVPF of providing a CAPER tablet is infinite.
The projection carries the language gain measured at the end of the 11-month intervention forward into adulthood without any adjustment for fade-out. The paper has no long-term follow-up and acknowledges that it may overestimate long-term benefits if the initial gains decline. Reading through the CAPER app itself fell over the course of the intervention: Figure A1 shows a steady decrease across all treatment arms, which the paper attributes to a novelty effect, and of the 296 minutes of app reading logged over the 45 weeks, 250 were logged in the first 23 weeks. The infinite result does not turn on the language gain persisting in full: because the $300 tablet is small relative to the $3,197 of projected tax revenue, net cost stays negative even if only a small fraction of the measured gain survives into adulthood. That margin narrows as the unquantified cost of the digital books rises.
The paper also cautions that the result should not be generalized to unrestricted tablets. Every function of the device other than the reading application was disabled, and an open-access tablet could displace reading with other uses.
References
Anger, Silke, Bernhard Christoph, Agata Galkiewicz, Shushanik Margaryan, Frauke Peter, Malte Sandner, and Thomas Siedler (2025). “A Library in the Palm of Your Hand? A Randomized Field Experiment with Low-Income Children.” IZA Discussion Paper No. 17322. IZA Institute of Labor Economics. https://docs.iza.org/dp17322.pdf
Hanushek, Eric A., and Ludger Woessmann (2009). “Schooling, Cognitive Skills, and the Latin American Growth Puzzle.” NBER Working Paper No. 15066. http://www.nber.org/papers/w15066
Kalil, Ariel, Haoxuan Liu, Susan E. Mayer, Derek Rury, and Rohen Shah (2026). “A Digital Library for Parent-Child Shared Reading Improves Language Skills for Young Disadvantaged Children.” European Economic Review, 186: 105323. DOI: https://doi.org/10.1016/j.euroecorev.2026.105323
Policy Features
- Category
- Education
- Sub-Category
- Child Education
- Beneficiary Type(s)
- Children under five, Parents
- Average Age
- 3.8
- Average Income
- 24,629
- Country of Implementation
- United States
- Year of Implementation
- 2020
- Empirical Method
- RCT
- Research Type
- Primary
- Peer Reviewed
- Yes
- MVPF Publication Link
- doi.org/10.1016/j.euroecorev.2026.105323