Using 287 country-year samples [...] we document that all three axiomatically justified PAM measures of Chiappori et al. (2025) [...] decline with GDP per capita. We show, however, that country-level measures are biased because marriages are formed in local markets: they conflate within-region sorting with a segregation bias [...] and a PAM-composition bias [...]. The two biases affect the three measures differently [...]. A further problem is that the standard absolute college-completion threshold becomes uninformative in poor economies where attainment is near zero, rendering measures undefined or highly sensitive to small cell counts. To address both issues we propose a two-market search-and-matching framework [...] and gender-specific relative education percentiles as a threshold comparable across development levels. We derive closed-form expressions for all three measures and characterise the decomposition of each aggregate measure into a within-region sorting and the two biases.