We analyze large nationally representative surveys of the labor force from three developing Asian economies (India, the Philippines and Thailand) at two points in time separated by a decade or more. Secondary and tertiary education attainment rose in the interim while the Mincerian education-wage profile became more convex. We document these shifts, allowing for inter-cohort dynamics. Returns to secondary education fell. Returns to college rose for older workers everywhere and for young workers in India, but fell for young Thais and Filipinos. We develop a new decomposition that permits us to attribute the shifting returns to education to the evolving structure of employment and inter- and intra-industry wage patterns. Secondary returns fell sharply in every sector as secondary-educated workers rapidly became available, while employment structures shifted slowly to absorb them. Conversely, rising returns within modern services were instrumental in lifting the returns to tertiary education. More manufacturing jobs will enable the Philippines to leverage higher growth from its human capital stock. Returns to secondary education in India have come to depend less on the manufacturing sector as manufacturing employment growth has been concentrated in low-skill sub-sectors. The intercohort divergence in returns to college arises in the Philippines and Thailand because excess young college-educated workers are pushed into low-wage or low-return jobs, while older college graduates are more likely to work in modern services. As modern service employment grows slowly, the largest and growing share of services employment has been in low-wage traditional services. From an employment perspective, “services-led development” therefore appears to be a red herring.

We study why the relationship between education and log-wages has become more convex in India, the Philippines and Thailand. To do so, we develop decompositions connecting returns to education, and shifts in those returns, to the evolving structure of employment. Returns to college depend mostly upon high-skill service jobs. While relative demand for college graduates in that sector generally rose, its employment share grew slowly, pushing workers with secondary education into less skill intensive services. Services employment, which grew fast, therefore became more menial on average. These polarizing trends in services account for the growing convexity of the Mincerian wage profile. The effects of industrialization on the returns to secondary education depend upon the composition of manufacturing employment. Slow structural transformation when educational attainment increases rapidly causes education inflation, and drives down the returns to secondary education. This constrains governments' seeking to use educational expansion to alter the wage distribution.

In the absence of data measuring the number of years of schooling required to perform particular jobs, we propose a new approach to testing for overeducation. Overeducation is confirmed by rising education levels in menial jobs that offer very low returns to education. Overeducation is deemed a systemic problem if these jobs absorb a growing share of the educated labor force. Normatively, overeducation should sometimes be seen as a shortage of skilled jobs, not as a surplus of educated workers. We use a decomposition of the returns to education to clarify the distinction between equity and efficiency rationales for expanding access to education when returns to education are low. We find substantial evidence of overeducation in the Philippines, mild evidence of it in India, and none in Thailand.