Nonprofit workers earn less on average than for-profit workers. Existing empirical work leaves open the question of whether this is driven by a willingness to work for less (the "labor donation hypothesis"). Wage gaps have consistently been found to be present in some industries and absent in others. In this paper, I consider when we should expect labor donations to nonprofits to generate wage gaps and, in doing so, offer an explanation for the previous inconsistent results. I highlight the importance of nonprofits' labor demand. Specifically, it is only in nonprofit employers' interest to maintain low wages if their labor demand does not exceed the number of workers who are willing to work for less. Otherwise, nonprofits must raise wages to attract other workers. This yields the prediction that wage gaps should be largest when the nonprofit share of labor within an industry is low. As nonprofit share increases, wages should equalize. Using economy-wide Census microdata, I provide evidence consistent with this prediction. I use more detailed data from the nursing home industry to better control for market conditions and rule out alternative explanations. I find that the quality of work in nonprofit nursing homes is highest in localities with low nonprofit share, where wage gaps are largest. This provides evidence that the relationship between wage gaps and nonprofit share is driven by "motivated types" sorting into nonprofit jobs.