Working Paper: NBER ID: w0141
Authors: Edward Lazear
Abstract: In this paper, another aspect of optimizing behavior is considered. Specifically, it asks whether variations in levels of attained schooling across groups can be explained by a model that assumes that capital markets are perfect and that individuals maximize wealth. The logic of the analysis proceeds as follows: First, a model is constructed that allows estimation of costs and returns to education for each individual, based on the assumption that all individuals face the same borrowing rates. Given costs and returns, one can obtain an optimal wealth-maximizing level of education for each individual. Differences between actually acquired and wealth-maximizing levels of education can then be calculated, and one can determine whether or not the residuals are systematically related to background variables. If, for example, low-income individuals have a consistently larger estimated wealth-maximizing level of education than actual level, one could conclude either that returns to schooling differed between groups or that capital market differences exist. The model allows these two explanations to be distinguished. Since differential returns are caused by wage differences across groups, the wealth-maximizing level can take these labor market variations into account. Any residual variation will be due to factors other than differential wage rates, presumably capital cost differences .
Keywords: Schooling Decisions; Family Background; Capital Markets; Education Economics
JEL Codes: I20; D91
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Variations in attained schooling levels across income groups (I24) | Capital market imperfections (G19) |
| Variations in attained schooling levels across income groups (I24) | Differential returns to education (I26) |
| Higher predicted levels of education for low-income individuals (I24) | Differing returns to education (I26) |
| Higher predicted levels of education for low-income individuals (I24) | Capital market imperfections (G19) |
| Systematic nature of residuals (C29) | Capital costs affecting schooling decisions (I21) |
| Wealthy families (G51) | Lower borrowing costs (G21) |
| Individuals from poorer backgrounds (I24) | Higher predicted levels of education than actual (I21) |
| Assumption of equal borrowing costs (G19) | Rejected at conventional levels of statistical significance (C12) |