Children's early skills are strong predictors of later learning outcomes. Research aiming to disentangle the causal effects of early skills from unmeasured, stable characteristics related to learning throughout development demonstrates that unmeasured confounders explain a large portion of the effects of early skills previously identified. To date, such research has been conducted exclusively in high-income Western countries, where education systems are better funded and overall learning outcomes are much higher than in low- and middle-income countries. The present study examined these issues in Ghana, a lower middle-income country in West Africa, using longitudinal data over 6 years (N = 2,012; 49.7% girls). We first used multilevel regression to estimate the associations of preschoolers' social and cognitive skills on their fourth and fifth grade outcomes. Next, we employed the random intercept cross-lagged-panel model to test whether the estimated effects of preschool on later skills are sensitive to a model that attempts to adjust for stable confounding factors during this developmental period. Our findings indicate interindividual stability in math and literacy, as well as codevelopment of math, literacy, and executive function during early and middle childhood. We contribute evidence on children's skill-building dynamics in a global context, with implications for how to optimize intervention programs and policies aiming to support children's academic achievement and learning.