The Economics of Scaling Early Childhood Programs: Lessons from the Chicago School

Beginning with innovative ideas from Theodore Schultz in 1961 about the value of education, to Gary Becker’s pathbreaking development of human capital theory in 1964, to James Heckman’s trailblazing studies beginning in the early 2000s about the importance of early childhood development, UChicago economics has expanded our understanding of how investments in education, training, and health increase an individual’s knowledge, skills, and productivity, which leads to future earnings for themselves and to economic growth for society. 

However, despite the decades of research by these and other researchers at UChicago and around the world, and despite many well-run and often successful policy implementations at a local level, we still face difficult challenges bringing those programs to scale. We have developed influential new theories, in other words, and we have gathered important empirical evidence from focused programs, but we have yet to master the science of scalability. That is the challenge—how to bring policy to scale—that motivates this new paper from UChicago Prof. John A. List, and he uses early childhood development as his exemplar.

For those interested in an overview of UChicago’s contributions to human capital theory and early childhood development, a reading of List’s introduction would prove rewarding. However, for the purposes of this research brief, we will focus on List’s theoretical approach to the scalability challenge. To begin, we must first review List’s recent introduction of a new framework meant to reorient research design, “Option C thinking.” This new framework reorients the research design itself, rather than adding a treatment arm onto a standard A/B test A/B test: in economics, this is shorthand for a randomized controlled experiment that compares two versions of a variable—a control (A) and a treatment (B)—to determine which produces a better economic outcome. A/B testing involves randomly assigning subjects to different conditions to minimize bias , shifting focus from establishing efficacy under controlled conditions to anticipating how such interventions will scale. That is, the time to worry about scalability is not after results are obtained from a small-scale A/B test, but before beginning that test. Success is not defined by an effective A/B research design, but rather by its effective scalability. Absent a fair amount of luck, research framework lacking Option C thinking is doomed to policy irrelevance.

This is a theoretical paper, which means that it does not explicitly offer empirical evidence, though List does corroborate this work’s theoretical findings with recent empirical research. The contribution here is to formalize Option C thinking and to offer a framework for ongoing research. To do so, List revisits Heckman’s microfounded model microfounded model: a framework where aggregate economic behavior is derived from the optimizing decisions of individual agents of skill production, extending the framework to explicitly incorporate the mechanisms that undermine scaled interventions. The model includes heterogeneous children with dynamic skill formation, where early investments exhibit complementarity with initial endowments and later-life outcomes. Please see the working paper for details, but List’s model reveals the following insights with application to scalability:

• Targeted interventions that address disadvantaged children can offer greater returns than universal programs, a finding that formalizes Heckman’s intuition about the alignment between equity and efficiency. This result holds for a plausible range of parameter values in the model, and List notes it would hold even more strongly under a policy that places extra weight on outcomes for disadvantaged children.
• Traditional A/B testing overestimates benefits by ignoring “ voltage drops voltage drops: term coined by List (along with voltage effect) and explicated in a 2022 book, voltage drops occur when a promising idea, policy, or intervention loses its effectiveness, impact, or profitability when it is applied to a larger scale. Voltage drops stem from five causes: false positives (bad data), incorrect population assumptions (general population does not mirror small-scale population, non-scalable key ingredients (“secret sauce” at micro level cannot be replicated at scale), high costs of scaling (benefits outweighed by costs), and negative spillover effects (costs to other areas hurts the overall outcome). :” declining effects from unrepresentative samples and situations, rising marginal costs, supply-side quality degradation, and general equilibrium spillovers. 
• Optimal program scale balances marginal benefits marginal benefits: the additional satisfaction, utility, or revenue a consumer or producer gains from consuming or producing one extra unit of a good or service. This value is incremental and generally decreases as more units are consumed, a concept known as diminishing marginal utility. against these voltage drops; policymakers must mitigate such effects through design adjustments that anticipate the magnitude of such declines at scale. 
• The model’s empirical predictions align with several recent empirical studies, indicating its applicability across diverse domains, including the microdynamics of skill formation to family investment decisions, peer effects, targeted interventions, and explicit analyses of scaling challenges.

This work answers List’s own challenge to incorporate scalability into research design by offering an applicable theoretical framework. But it is only a first step. Further theoretical research will offer more effective guidance, and, over time, real-world examples of successful A/B/C interventions will set the standard for policy-focused research.

Written by David Fettig • Designed by Maia Rabenold