As the share of all income going to the top 1 percent has risen over the past four decades, so has the share of top incomes coming from labor income relative to capital income. The rise in labor income is mainly due to the explosion in executive compensation over the same period—mostly because of the increase in executives being paid with stocks, options, and bonuses. The principal-agent model explains the reason for such compensation instead of a flat salary. Yet hundreds of papers in economics, finance, accounting, and management have reached no consensus on whether executive compensation is efficient or whether empirically it conforms to the prediction of the principal-agent theory. In this article, we argue that this lack of consensus is due to two issues: The first is a measurement issue, and the second is that the exact prediction of the principal-agent model depends on many objects unobservable to the econometrician. We illustrate how using theory-based estimation together with a model-motivated measure of total compensation can help overcome these issues. Finally, using a model-consistent measure of compensation and theory-based estimation, we conclude that executive compensation broadly conforms to the principal-agent theory; however, each situation and the variables used have to be carefully modeled, identified, and estimated.