E STIMATING COHERENCY B ETWEEN SURVEY DATA A ND INCENTIVIZED

In Section 5, we discuss the identification and estimation of the structural parameters of the model of the ex-ante value of higher education estimated from the survey, which is used to generate the distribution of financial aid valuations in the survey. [...] In practice, this means that the intercept terms of the expression relating individual consumption while in school (c̃ei ) and the consumption value (cv e i ) to the ranks of the primitive factors cannot be distinguish from the location parameters of the monetary and non-monetary costs and the consumption value (cve0) and therefore renders the separa- tion of consumption while in school and post-s. [...] Second, with respect to post-education benefits, knowledge of the minimum level of consumption achievable (equal to that of the outside option) along with the normalization of mbe0 to 0, allows us to estimate the location parameter of the non-monetary benefits of education and therefore separate the per-period utility of attending education from the utility of post-education labor market experienc. [...] 5.2.3 Likelihood and Estimation Method As indicated earlier, we separate the estimation of the model from the estimation of the distribution of the primitive factors and benefits and costs of education. [...] 5.3.4 Implications for Financial Aid Returning to the main objective of the paper, which is to obtain two separate distributions of the welfare gains of financial aid offers in the experiment, we now use the structural parameters obtained from survey data to predict the welfare gain of the specific offers found in the experiment (found in Table 1).