Gender parity in education cannot be measured by the gender parity in enrollment alone.
 
One of the targets of the Millennium Development Goals—gender parity in all levels of education—is widely considered to have been attained. However, measuring gender parity only through school enrollment is misleading, as girls may lag behind boys in other educational measures. We investigate this with four rounds of surveys from Bangladesh by decomposing households’ education decisions into enrollment, education expenditure, and share of the education expenditure allocated for the quality of education like private tutoring. We find a strong pro-female bias in school enrollment but pro-male bias in the other two decisions. This contradirectional gender bias is unique to Bangladesh and partly explained by the presence of conditional cash transfer programs. Although these programs promoted girls’ enrollment in secondary schools, they were largely ineffective in narrowing the gender gaps in academic performance and intrahousehold allocation of education resources. Gender parity in education cannot be truly achieved without addressing these gaps.