At low levels of schooling (the left side of the graph), the marginal benefit is high: you are learning skills—the alphabet, for example—that your brain can digest and that have a high payoff in society. At the same time, the marginal cost—the opportunity cost of foregone earnings—is low: even if it’s legal for them to work, six-year-old kids are not very productive in the labor market. The early years of primary school, by this calculation, have a much higher benefit than cost. As you grow up, however, your potential wage grows, and thus the marginal cost of schooling rises. All the while, the marginal benefit declines with more time in school, your brain being less adept at learning the now-more-esoteric subjects. But you should stay in school as long at the marginal benefit exceeds the marginal cost. However, there comes a point at which the marginal cost of staying in school is greater than the marginal benefits, and going to school past that point is a losing proposition. Optimally, you should stop just before this happens: when marginal benefits precisely equal marginal costs, denoted by the dashed line in the figure.

Now, let’s think about some policies...

Compelling, tricking, or bribing people to spend more years in school does so when the gains from schooling have been largely exhausted. (Numerous fashionable policy initiatives do this to some extent. I won’t name names.) You can see this in the graph: those last years of schooling (just to the left of the dashed line) come after the gap between MB and MC has mostly disappeared. In fact, policies that bump you to either side of your optimal choice will have small effects on lifetime income. The poor will still be poor, albeit with different amounts of education than they would have had otherwise. This, too, is seen in the graph. At the optimal choice of schooling, you are just indifferent between attending and dropping out, because marginal benefit equals marginal cost. It follows that if you attend a little more or a lit less instead of your optimal plan, well... marginal benefit is still pretty close to marginal cost. To a first approximation, you don’t lose much lifetime income by attending a little bit more or little bit les school. (Students of 1st-year grad economics will recognize this as the ‘envelope condition’: once you’ve optimized something, small deviations from your optimal plan have small only effects.)