Returns to higher education can reflect either an increase in human capital or job market signalling. In this paper, I isolate the human capital resulting from higher education, capitalizing on a distinctive feature of Norway’s implementation of the Bologna reform. Students who enrolled in 1999 inadvertently acquired more human capital compared to those who enrolled in 2000, despite earning identical degrees. This anomaly, two sets of graduates with identical degrees, of which one has more education, allows me to answer a key question concerning investment in higher education. How much does an increase in human capital increase an individuals’ earnings, in the absence of self-selection and signalling? I use rich longitudinal register data to estimate returns to university-acquired human capital. 15-20 years after first enrolling, receiving reform-induced human capital leads to gaining 1.5 percentile ranks in the earnings distribution. While I can rule out negative effects, the precision of the estimates does not allow me to distinguish between moderate and large effects. Using a novel 3-step cross-fit test-and-select estimator exploiting first-stage heterogeneity reduces standard errors by 16%. Human capital increases long-run earnings potential even when employers cannot observe it directly, but signalling matters when first entering the labor market.

We investigate how inequality acceptance depends on power abuse. Running a large- scale pre-registered two-stage experiment, we measure inequality acceptance through spectators redistribution choices. We randomize whether a worker can decide their initial earnings, potentially abusing their power for their own gain at the expense of their co-worker by allocating high earnings to themselves. Then, impartial spectators redistribute. We find that spectators give significantly lower earnings to workers de- ciding their own initial earnings, compared to workers with identical but externally decided initial earnings. Spectators redistribute substantially when confronted with power abuse, while they accept meritocratic initial earnings regardless of who pro- poses. While most spectators redistribute to achieve meritocratic shares, a minority of spectators switches to active punishment in the face of power abuse. Thus, allowing for power abuse dismantles a consensus to implement meritocratic (and unequal) pay.