[Excerpt] It is well-known that test scores are correlated with students’ socio-economic backgrounds. Hence to the extent that colleges are successful in “buying” higher test score students, one should expect that their enrollment of students from families in the lower tails of the family income distribution should decline. However, somewhat surprisingly, there have been no efforts to test if this is occurring. Our paper presents such a test. While institutional level data on the dollar amounts of merit scholarships offered by colleges and universities are not available, data are available on the number of National Merit Scholarship (henceforth NMS) winners attending an institution on scholarships that have been funded by the institution itself, rather than the National Merit Scholarship Corporation (henceforth NMSC). These institutional scholarships are awarded to high test score students only if they attend the institution. Our research strategy is to estimate if an increase in the number of recipients of these scholarships at an institution is associated with a decline in the number of students from lower and lower middle income families attending the institution, other factors held constant. We measure the number of these students by the number of Pell Grant recipients attending the institution.