Table 1.

Baseline development accounting results and comparison to the literature

Human capital contribution	Our estimates				Estimates from the literature
	w = 0	w = 0.15	w = 0.20	w = 0.25	Ref. 5	Ref. 12	Ref. 7
h90/h10	2.24	3.35	3.82	4.34	2.00	2.10	4.70
(h90/h10)/(y90/y10)	0.11	0.16	0.19	0.21	0.09	0.22	0.21
var(log[h])/var(log[y])	0.07	0.14	0.18	0.21	0.06	0.07	0.26
	w = 0	w = 0.15	w = 0.20	w = 0.25	Ref.10	Ref.44	Ref.13	Ref.11
(ln[h90] – ln[h10])/(ln[y90] – ln[y10])	0.27	0.40	0.44	0.48	Nearly all	0.51	0.62	Potentially none

The variable y is real output per worker on average from 2000 to 2010; h is a measure of human capital constructed on the basis of both schooling and learning data on average from 2000 onwards. We include various decompositions of the role of human capital in explaining cross-country income differences based on the literature^5–13: h₉₀/h₁₀, (h₉₀/h₁₀)/(y₉₀/y₁₀), var(log[h])/var(log[y]), (ln[h₉₀] – ln[h₁₀])/(ln[y₉₀] – ln[y₁₀]). Subscripts refer to percentiles; for example, the decomposition (h₉₀/h₁₀)/(y₉₀/y₁₀) captures the ratio of human capital in the 90th relative to 10th percentile over the real output per worker in the 90th relative to 10th percentile. Variable constructions and decompositions are described in detail in the Methods. We assume rates of return to the learning component of human capital—denoted as w—on the basis of the microeconomic literature^42,43. We conduct sensitivity analyses with values w = 0.15, w = 0.20, and w = 0.25. When w = 0, our accounting de facto only includes schooling; for any value w > 0, we include learning as well as schooling. We include 131 countries in this development accounting exercise. Schooling data are from a previously published study³². GDP data on real output per worker are from the Penn world tables v.9.0¹⁴. Learning estimates are from our database. Literature estimates are derived from previously published papers^5–13,44.