Summary: In spite of a large literature on the causal effects of financial aid, the impact of funding on college outcomes remains an area of ongoing research with important unanswered questions. Studying the impact of financial aid presents the ongoing challenge of how to identify causal effects, as aid receipt is rarely random and at least partly driven by individual student characteristics. We use a unique administrative dataset to evaluate the causal impact of the 21 Century Scholars Program (21C) in Indiana, eligibility of which is based on family income and modest merit requirements. We use as-good-as random variation in income eligibility during a student's eighth grade year to identify the impact of the funding on college enrollment, college quality, degree completion, and post-college earnings. Although scholarship recipients outperform state averages in all these measures, an intent-to-treat framework shows that virtually all of these differences are due to selection. We propose that increasing the 21C take-up rate could improve the realized impact of the program by enrolling students who are otherwise unlikely to succeed in college. Background: Indiana's 21 Century Scholars Program (21C) provides up to four years of in-state college tuition funding for eligible students. Middle school students with family income below 185% of the Federal income poverty level are eligible to enroll in the program, which requires them to maintain a C+ grade point average, complete certain steps towards college readiness, and abide by program conduct rules. Students must graduate from an Indiana public high school and attend an Indiana institution to receive the funding. 20,000 students are receiving 21C funding from the state and the average annual award size has been around $8,000. 21C funds can be only be used to pay for tuition, but can be used before other financial aid awards such as Pell grants, meaning that the award significantly adds to the total amount of grant aid that low-income students can receive. Among other financial aid programs, 21C is similar to the Susan Thompson Buffett Foundation grant (Autor, Pallais, and Angrist, 2022) in that 21C is a place-based scholarship with modest merit requirements, family income caps, and comparatively generous award levels. Nearly half of Indiana high school students are eligible for the program, but only about 40% of eligible students participate in the program and fewer than 8% of students go on to actually receive 21C funding. While 21C recipients have college completion rates that are higher than the state average (48 percent versus 20 percent), these group differences cannot be interpreted as causal effects. 21C participants are positively selected due to the requirements of the program enrollment process and are not representative of their non-participating, low-income peers (St. John et al., 2002, 2005; Toutkoushian et al., 2015). Research Question: Our objective is to identify the causal impact of 21C on college success and post college earnings, and to gauge the cost effectiveness of the program for the state of Indiana. Data: We use data from Indiana's Management Performance Hub, a centralized repository that aggregates and merges administrative records on K-12 education, college outcomes, employment, and earnings. The data include selected grade school outcomes (test scores, grade progression), high school academic outcomes (attendance, grades, graduation), college enrollment, degree completion, and employment and earnings. The data are accessible through a secure MPH portal. All analyses are carried out remotely. Only aggregates (including summary statistics, regression coefficients, and figures) can be extracted from the MPH environment. Importantly, we are able to observe individual Free or Reduced Price Lunch (FRPL) status by year, and 21C funding receipt for the universe of 2012-2016 high school graduation cohorts, a total of 400,000 students. Descriptive statistics confirm that this sample is representative of Indiana's high school student population in subsequent years, and well-measured sample attributes allow for transparent comparisons with other states or regions. Because the income requirements for 21C are the same as the FRPL eligibility limits, we use FRPL status during 7 and 8 grade as a proxy for 21C eligibility. Students must sign up for 21C during 7 or 8 grade, with a hard deadline of June 30 at the end of their 8 grade year. We measure the effect of quasi-random variation in FRPL during grades 7 and 8 on our outcome(s) of interest. Specification: As a first stage, we regress the marginal impact of FRPL eligibility during middle school on 21C funding receipt, holding constant the number of years of FRPL eligibility between grades 5-11, cohort year, geography, race, and gender. This first stage yields an estimated effect of FRPL eligibility on 21C funding receipt of 4.6 percentage points on a baseline 21C grant receipt rate of 7.8%. The F-statistic for the first stage is 274, eliminating concerns of a weak instrument. We then use an intent to treat framework and regress our second stage outcomes of interest on FRPL eligibility in middle school; these outcomes include high school graduation, college enrollment and graduation, and earnings 4-6 years after expected high school graduation. To calculate the program impact on compliers (local average treatment effect), we use two-stage least squares. Findings/Conclusion: This study is the first to use a quasi-random research design to evaluate the 21C program. We find precisely estimated null effects of program eligibility on education and earnings outcomes. Two-stage least squares shows null effects for participants as well. The preliminary results indicate virtually all of the positive differences observed between 21C recipients and no-recipients is due to selection, and that the students who currently make use of the program are the ones most likely to attend college with or without the support of 21C funding. The question of the 21C efficacy is especially relevant to policy makers as the current state legislature is debating a bill to enroll all eligible Indiana students into the program by default. We propose that expanded program participation or improved program retention could help the program reach marginal students whose college outcomes would be affected by additional aid.