Research

Gender-specific patterns in educational attainment create dual sources of talent misallocation. Using Portuguese administrative data (1995-2021), we document that women systematically underenroll in STEM fields (2.2 % vs 4.4 % for men) but overenroll in non-STEM college education (14 % vs 6.8 % for men), while men disproportionately avoid higher education altogether. These patterns, together with the underperformance of female entrepreneurs across all educational levels, generate aggregate costs to the economy. We develop a general equilibrium model where agents make endogenous educational and occupational choices, incorporating both gender-specific education costs and entrepreneurial productivity differentials. Our quantitative experiments reveal the relative importance of each distortion. Equalizing women's education costs to male levels reduces GDP by 0.9 %, as lower college participation offsets gains from increased female STEM enrollment, while just increasing men's participation in non-STEM college fields increases GDP by 2.4 %. The STEM channel does not seem quantitatively relevant, since equalizing women's participation to men only increases GDP by 0.02 %. However, equalizing entrepreneurial productivity parameters across genders increases GDP by 5.3 %, highlighting the relevance of the distortions that make female entrepreneurs underperform relative to men.

The gains from a more skilled workforce depend on firms’ ability to finance complementary investment. Using Portuguese employer–employee records linked to firm accounts, I measure own-account intangible investment from graduates’ oc- cupations and earnings. Firms with more intangible capital use skilled labor more intensively in production, and early intangible investment predicts subsequent em- ployment growth. I develop a heterogeneous-firm model in which skilled workers produce output or build intangible capital that raises labor productivity. Limited internal resources delay intangible investment and the expansion of tangible opera- tions in productive young firms, restricting their ability to absorb additional skilled labor. Calibrating the model to Portugal, I compare stationary equilibria following an increase in the skilled share of labor from 9.6 to 20.5 percent. The output gain is 9 percent larger without financial frictions, and the output cost of frictions rises from 2.6 to 3.3 percent as skill supply expands. Financial frictions also shift addi- tional skilled employment toward established firms: firms younger than five absorb 37 percent without frictions, compared with 17 percent with frictions.