HISPANICS IN HIGHER EDUCATION AND THE TEXAS TOP TEN PERCENT LAW

Abstract

This paper examines the consequences of changes in Hispanic college enrollment after affirmative action was banned and replaced by an admission guarantee for students who graduate in the top 10% of their high school class. We use administrative data on applicants, admittees and enrollees from the two most selective public institutions and TEA data about high schools to evaluate whether and how application, admission and enrollment rates changed under the three admission regimes. Despite popular claims that the top 10% law has restored diversity to Texas's public flagships, our analyses that account for secular changes in the size of graduation cohorts show that Hispanics are more disadvantaged relative to whites under the top 10% admission regime at both UT and TAMU. Simulations of Hispanics’ gains and losses at each stage of the college pipeline reveal that affirmative action is the most efficient policy to diversify college campuses, even in highly segregated states like Texas.

Introduction

Texas higher education has been in the spotlight since the Fifth Circuit Court outlawed the use of race and national origin in college admissions¹. The following year several campuses, including the University of Texas at Austin (UT) and Texas A&M University (TAMU), registered sharp declines in the number of black and Hispanic first time freshmen (Barr, 2002).² In response to the judicial ban the 75^th Texas legislature passed H.B.588—the uniform admission law—which guarantees admission to any Texas public university to seniors who graduate in the top decile of their class. Popularly known as the top 10% law, H.B.588 also specified 18 factors that universities should consider in admitting students who do not qualify for automatic admission, including socioeconomic status, second language ability, and indications that the applicant overcame adversity (see Long and Tienda, 2008a). The uniform admission law was fully in force for the fall, 1998 admission cohort (for a more thorough review of H.B.588, see Long and Tienda, 2008).

Spearheaded by the late Congresswoman Irma Rangel, H.B.588 was intended to increase college access to a wide spectrum of the Texas population by attracting the very best students of every high school to the State's flagship universities (Holley and Spencer, 1999; Montejano, 2001). Initially the law was praised as a race-neutral alternative to affirmative action that both rewarded merit and broadened college access (Tienda and Sullivan, 2009). Supporters claimed that the percent plan helped restore diversity to the flagship campuses, albeit by capitalizing on segregation (Tienda and Niu, 2006), by removing the standardized test score barrier (Alon and Tienda, 2007). Over time, however, the top 10% law has become as controversial as the affirmative action regime it replaced. Detractors argue that percent-based admission regimes not only are a disguised form of affirmative action, but that they also are unfair to high achieving students ranked below the 90^th percentile who graduate from competitive high schools. Although the landmark 2003 Grutter³ decision reversed the judicial ban on race preferences, the top 10% law remains in force until repealed by the Texas legislature. In effect, between the early 1990s and the present, judicial and statutory decisions produced four different college admission regimes in Texas:

• Pre-Hopwood: affirmative action permitted (pre-1996);⁴

• No preferences: Judicial ban on affirmative action (1997);

• Top 10% law with continued judicial ban (1998-2003);

• Post-Grutter: affirmative action permitted, top 10% law remains in effect (2004 – present).⁵

Researchers have focused on this aspect of post-secondary decisions, especially on the admission advantage enjoyed by minority applicants (Bowen and Bok, 1998; Long and Tienda, 2008). College administrators and legislators measure success in achieving campus diversity based on freshman enrollment, and to a lesser extent, graduation rates. Despite their centrality in shaping the composition of entering classes, with few exceptions (e.g., Long, 2005; Card and Krueger, 2005; Brown and Hirschman, 2006; Long and Tienda, 2008a; Koffman and Tienda, 2008), researchers have paid much less attention to variation and change in application rates. Partly this reflects data constraints and partly the fact that litigation largely focuses on institutional admissions decisions, not individual decisions to apply or, conditional on acceptance, to enroll.

Several analysts have begun to fill this gap, including researchers interested in the Texas admission guarantee for top 10% graduates. For example, Long and Tienda (2008a) show that the elimination of affirmative action and implementation of the percent plan, which directly impacts the admissions systems only of the most selective institutions, also produces substantial indirect effects at other institutions. Using a variety of empirical methods, they find that average test scores of applicants to less selective institutions rose following the change in admission criteria, as students with high test scores who did not qualify for the admission guarantee applied to a broader set of institutions. Furthermore, as the share of top 10 % applicants at UT-Austin rose, the steady assent in the test scores of their applicants stagnated (see also Niu and Tienda, 2010).

Although highly informative, Long and Tienda (2008a) only consider the subset of students who actually applied to a post-secondary institution, independent of changes in the number of potential applicants (i.e., the size of high school graduation cohorts). To address this limitation, Koffman and Tienda (2008) analyze administrative records from UT and TAMU, making two important extensions. First, they compare changes in the applicant pools according to the economic status of their high school; and second, they evaluate application behavior relative to the size of high school graduates in specific years. They find that graduates from affluent schools are significantly more likely to seek admission at one of the public flagships compared with their peers who graduated from high schools that served students of low to moderate socioeconomic status. This generalization holds before and after the change in admission regime. More importantly, Koffman and Tienda show that the admission guarantee did little to raise flagship application rates from poor high schools.

Building on these insights, we investigate the consequences of changes in Texas college admission policies for Hispanics, the fastest growing segment of the State's college-age population. Using over a decade of administrative data for the two flagship campuses, we consider how Hispanic and white students fared across the three policy regimes in force between 1992 and 2003.⁶ To motivate the empirical analysis, we provide a brief overview of the changing demography of Texas higher education. Following a discussion of data and methods, we report changes in Hispanic and white application, admission and enrollment rates across the three policy regimes. The conclusion reconciles our findings with those of other studies using similar methods and discusses the implications of dismantling the top 10% law for Hispanics.

Our analysis is novel in two ways: First, we compute application rates by merging high school-specific data about graduates with college applicants from those schools. This is important given the rapid growth of the college-eligible population in Texas during the observation period (WICHE, 2008). Second, our simulation exercises go beyond conventional impact assessments by quantifying Hispanics’ gains and losses at each stage of the college pipeline under the three admission regimes. Despite popular claims that the top 10% law has restored diversity to UT and TAMU (Wilgoren, 1999), our results show that Hispanics are worse off relative to whites than they were under affirmative action.

Demography of Texas Higher Education

Owing both to high levels of immigration and high immigrant fertility, Texas is one of the nation's fastest growing and most rapidly diversifying states. Between 1994 and 2004, the number of diploma recipients rose 50 percent, from 163 to 244 thousand (Tienda and Sullivan, 2009). High school graduation rates improved by almost 11 percentage points between 1994 (pre-Hopwood) and 2003 (pre-Grutter)—rising from 56 to 67 percent (Swanson, 2006)—but large differences remain between whites and disadvantaged minorities.⁷ Many Hispanic students do not complete high school; nevertheless the number of Hispanic high school graduates rose 78 percent during this period, raising their share of Texas diploma recipients from 29 percent in 1994 to 35 percent by 2004 (Tienda and Sullivan, 2009). White students are more likely to graduate from high school than Hispanics, but their share of the high school population has been shrinking. In 1994 whites earned 56 percent of diplomas awarded in Texas, but by 2004 their share dropped to 48 percent. WICHE (2008:107) projections indicate that Hispanics will earn 38 percent of diplomas in 2008, compared with 43 percent for whites.

Not all high school graduates pursue post-secondary education of course, but the larger graduation cohorts imply intensified competition for access to the selective public institutions among the college-bound. Although Texas's post-secondary system expanded in response to higher demand, its growth failed to keep pace with demographic trends. Texas postsecondary enrollment rose 27 percent between 1994 and 2004, which is well below the 50 percent increase in the number of high school graduates during the same period (Tienda, 2006).Texas differs from the nation and many other states in another important respect that bears on the college squeeze—namely, the preponderance of two-year institutions within its post-secondary education system. At the national level, enrollment growth at two- and four-year institutions was relatively similar during this period - around 19-20 percent - but this was not the case in Texas, where two-year institutions registered a 37 percent enrollment increase during the period. It is noteworthy, moreover, that total enrollment at two-year institutions had surpassed that of four-year public institutions in 1995, at least one year before the Hopwood decision (THECB, 2005).

The change in Texas college admission regimes over a short period of time coupled with appreciable increases in the number of college-eligible Hispanics raises several questions about their representation in higher education: First how did Hispanic application, admission and enrollment rates to the University of Texas at Austin (UT) and Texas A&M University (TAMU) change over the three policy regimes? That applications surge does not, ipso facto, signal rising application rates because the increases may represent secular changes in the growth of high school graduates. Second, how have Hispanic-white gaps in application, admission and enrollment rates changed across policy regimes? Finally, what are the enrollment implications of changes in Hispanics’ application and admission behaviors? In addressing these questions, we illustrate the policy consequences of changes in Texas college admission policies for Hispanics—the fastest growing segment of its college-age population.

Data and Methods

The empirical analyses use data from the Texas Education Agency (TEA) about graduates from Texas public high schools and administrative data on applicants, admittees and enrollees to UT-Austin and TAMU for the years 1993-2003 that were compiled by the Texas Higher Education Opportunity Project (THEOP).⁸ We use the publically available TEA data to compute Hispanic and white application rates to UT-Austin and TAMU for each policy regime. Special and alternative schools are excluded because their students may differ systematically from standard public high schools in their college going behavior.⁹ Analyses of application rates include 942 public high schools that were in operation throughout the observation period. Nearly 95 percent of Texas seniors graduate from public high schools and this share has not changed since the early 1990s (WICHE, 2008). We employ weights to account for school variation in the size and ethnic composition of graduating classes.¹⁰

We then draw on institutional administrative data that includes the admission and enrollment status of all applicants to UT and TAMU to evaluate their admission and enrollment outcomes across policy regimes. TEA data include measures of senior class size, which institutions use to compute class rank, as well as various indicators that portray heterogeneity among Texas high schools, such as public/private status (from NCES) and the percent of students ever economically disadvantaged (from TEA). We merge these school attributes to records of individual applicants. Finally, we simulate admission and enrollment changes in the post-affirmative action period under several scenarios about application and matriculation behavior for each group.

Application Rates

Table 1 reports the application rates to UT-Austin and TAMU for Hispanic and white Texas public school seniors across the three policy regimes. The average Hispanic application rate to UT-Austin fell from 3.6 to 2.8 percent during the policy period when neither race nor rank preferences were exercised. Although Hispanic application rates to UT improved under the top 10% regime compared with the no preference year, they remained below the level observed under affirmative action. The white application rate to UT also fell in 1997 from 7.2 to 6.6 percent, but rebounded to its affirmative action level under the top 10% law. Comparable data for TAMU show a steady drop in application rates for both Hispanic and white Texas high school graduates after affirmative action was banned, but the decline was appreciably larger for Hispanics. In part the steady drop reflects a provision in the law that allows rank-eligible students to select their campus, but the surge in the number of graduates across regimes is also responsible (Tienda and Sullivan, 2009). Nevertheless, Hispanics experienced larger declines in application rates than whites once affirmative action was disallowed. The greater decline for Hispanics at TAMU relative to UT is partly due to the general perception that the former is less hospitable to minorities (Koffman and Tienda, 2008).

Table 1.

Average Application Rates to UT-Austin and Texas A&M from Texas Public High School Seniors, 1993-2003

	University of Texas at Austin			Texas A&M University
Policy (Years)	Hispanic	Whites	Difference	Hispanic	Whites	Difference
Affirm. Act. (‘94-‘96)	3.61	7.19	-3.58	3.23	9.48	-6.25
No Policy (1997)	2.77	6.56	-3.79	2.74	9.30	-6.56
Top 10% (‘98-’03)	3.26	7.13	-3.87	2.22	9.06	-6.84

Note: Percentages are for the 942 public high schools in operation in Texas from 1993 through 2003. Data for Texas A&M extend through 2002.

Although the percentages in Table 1 seem small, they represent substantial changes in the number of Hispanic applicants because the population of graduates is large and growing rapidly. To illustrate, Table 2 simulates the number of additional Hispanic applicants under two hypothetical behavioral scenarios: 1) if the Hispanic application rate remained at its affirmative action level; and 2) if Hispanics applied at the same rate as whites for each policy period. Assuming that Hispanic application rates remained at their affirmative action levels implies that an additional 380 and 221 Hispanics would have applied to UT-Austin and TAMU, respectively, when neither race nor class rank preferences were used. Moreover, during the first four to five years of the top 10% regime, UT-Austin and TAMU would have gained an average of 243 and 691 Hispanic applicants, respectively, had their application rates remained at their pre-Hopwood level.

Table 2.

Estimated Additional Hispanic Applicants under Two Behavioral Scenarios

	If Hispanics’ Application Rates Remained at Affirmative Action Level a		If Hispanics had Whites’ Application Rates for Each Policy Regime b
Policy (Years)	UT-Austin	TAMU	UT-Austin	TAMU
Affirm. Act. Annual Avg.	---	---	1,525	2,668
No Policy (1997)	380	221	1,702	2,948
Top 10% Annual Avg.	243	691	2,604	4,683

Note: Data for TAMU extend only to 2002. .

^aThe following formula was employed for each cell: Additional Applications = (HispGrads * HAppRtAA – HispApps), where HispGrads are the total number of Hispanic high school graduates during the given policy period, HAppRtAA is the Hispanic application rate during Affirmative Action, and HispApps are the total number of Hispanic applicants from Texas high schools during the same policy period. Numbers were calculated from Texas Education Agency data.

^bThe following formula was employed for each cell: Additional Applications = (HispGrads * WhtAppR – HispApps), where WhtAppR is the white application rate during the given policy period.

The second counterfactual – that of equal Hispanic and white application rates across policy regimes – implies substantially more Hispanic applicants to both UT-Austin and TAMU under affirmative action, averaging respectively, 1,525 and 2,668, annually. Even more striking is the simulated implication that if Hispanic and white application rates were equivalent during the top 10% period, UT and TAMU each would have received approximately 2,604 and 4,683 additional Hispanic applicants annually. That the simulated estimates increase under the no preference and the top 10% regimes reflects both growing disparities in application rates between Hispanics and whites (Table 1) and larger pools of high school graduates. One conclusion is that the policy emphasis on admissions may be partly misplaced; a focus on boosting application rates, particularly from underrepresented high schools, appears to be warranted.

These estimates are likely to be conservative because the TEA data used as school-specific denominators for application rates only include public high school seniors, which accounted for 67 and 72 percent of all applications received by UT-Austin and TAMU in 1997, the year neither race nor class rank preferences were in force, and approximately 70 percent of the applicants for both universities under the top 10% regime. Approximately 30 percent of applicants to UT-Austin and TAMU were private school attendees, out-of-state students, international students, or non-traditional students who applied at least one year post-high school. It is conceivable that our simulated changes in Hispanic applications would be even larger if we were able to perform similar calculations for all Texas high school graduates who deferred their college careers either for financial or personal reasons.

Admissions and Enrollment

Campus diversity depends not only on the size and composition of the applicant pool, but also admission and enrollment rates. Admissions are constrained both by policy and institutional carrying capacity, namely the size of the freshman class that can be accommodated given physical and human capital resources. Table 3 shows that, unconditional on applicant characteristics, Hispanics enjoyed an admission advantage relative to whites under affirmative action (3.2 and 12.3 percent at UT and TAMU, respectively), but faced lower admission prospects compared with whites under both alternative admission regimes. Because test scores were considered for admission under affirmative action, the conditional advantage would likely be larger given Hispanics’ lower test scores. Hispanic admission rates at both flagships were lowest under the top 10% law. At UT the decline in Hispanic admission rates occurred after the no policy period, while the drop at TAMU began with the repeal of affirmative action. Compared with affirmative action, Hispanic applicants to TAMU witnessed declines in admission rates of 10 percentage points in 1997 and 15 points under the top 10% regime. By contrast, whites’ admission rates rose during the no policy period (by 11.5 percentage points at UT and 6 points at TAMU), but their admission rates returned to affirmative action levels under the top 10% law mainly because of institutional constraints on the size of the freshman class.

Table 3.

Admission and Enrollment Rates, and Share Admitted and Enrolled at UT-Austin and Texas A&M across Three Admission Regimes

	Admitted						Yield
	UT-Austin			Texas A&M University			UT-Austin			Texas A&M University
	(Admitted / Applicants) × 100						(Enrolled / Admitted) × 100
Group	Affirm. Action (‘90-‘96)	No Policy (1997)	Top Ten Percent (’98-’03)	Affirm. Action (‘92-‘96)	No Policy (1997)	Top Ten Percent (‘98-‘02)	Affirm. Action (‘90-‘96)	No Policy (1997)	Top Ten Percent (‘98-‘03)	Affirm. Action (‘92-‘96)	No Policy (1997)	Top Ten Percent (‘98-‘02)
Whites	71.7	83.2	71.4	73.9	79.9	74.0	59.1	64.0	61.7	61.8	57.4	64.6
Hispanics	74.9	75.4	67.7	86.2	76.2	70.8	54.5	64.3	60.2	50.3	46.5	51.7
	Composition of Admittees (%)						Composition of Enrollees (%)
Whites	.65.1	65.6	61.3	73.0	74.9	76.6	66.4	66.8	62.3	78.2	80.5	82.2
Hispanics	15.9	12.4	14.5	14.4	11.2	10.9	15.0	12.7	14.4	12.5	9.8	9.4
Totals	81.0	78.0	75.8	87.4	86.1	87.5	81.4	79.5	76.7	90.7	90.3	91.6

Note: Number of observations is 169,547 and 140,472 for UT-Austin and TAMU, respectively. Composition of admittees is calculated as group specific admits / total admits. Similarly, composition of enrollees is calculated as group specific enrollees / total enrollees.

The second panel in Table 3 reports how the admission pools changed under the three policy regimes.¹¹ The first two columns show that the ban on affirmative action resulted in smaller shares of Hispanics at both public flagships. The white share of admittees was unchanged at UT, but rose about two percentage points at TAMU. Under the top 10% regime, Hispanics partly regained their share of UT's admittee pool, but their representation did not fully rebound to the affirmative action level. At TAMU, the share of white applicants in the admittee pool increased after affirmative action, while Hispanics’ representation continued a downward spiral. These generalizations about changes in the ethno-racial composition of admittee pools hold even if tabulations are restricted to admissions from Texas public high schools.¹² This is not surprising considering that approximately 80 percent of all students admitted to both universities are from Texas public high schools.

The right panel of Table 3 provides information about the changing composition of enrollment. The top rows, which portray enrollment rates, known as “yield,” reflect individual decisions that take into account alternative offers of admission as well as financial and other unobserved personal circumstances. Although the changes in admission policy since the repeal of affirmative action narrowed Hispanic-white enrollment differentials among admitted students, enrollment rates are based on a much smaller pool of Hispanics compared with whites. During the year that neither race nor class rank preferences were in force, both Hispanic and white admittee yields rose at UT but fell at TAMU. Under the top 10% regime, yield rates rebounded at TAMU, and slightly exceeded the affirmative action rates for both groups; however, at UT, yield rates fell slightly for both groups, albeit at levels above pre-Hopwood years.

The second row of the right panel reports the composition of enrollee pools at both flagships and represents the measure of campus diversity used by administrations and reported by the media. At both flagships, Hispanic's relative share of enrollment was highest under affirmative action; however, at TAMU, Hispanic representation did not rebound under the top 10% regime, as occurred at UT. Again these generalizations hold even when comparisons are restricted to graduates from in-state public high schools. Collectively, these tabulations indicate that the top 10% policy was not as salutary for Hispanics as generally assumed even at UT. The share of Hispanics in the admittee and enrollee pools was highest during the affirmative action period at both flagships, and their admission prospects were lower than those of whites under both post-affirmative action admission regimes.

Admission Policy and the College Pipeline: A Simulation

To illustrate the implications of the Hispanic-white admission and enrollment gaps during a period of rising demand for college statewide, we estimate the “cost” in slots following the repeal of affirmative action. Specifically, we simulate the number of admitted and enrolled students each group would have gained or lost if affirmative action had not been rescinded and the shifts in admission regimes were inconsequential. Importantly, the simulations account for the carrying capacity of both UT and TAMU during the three policy regimes, which is vitally important during a period of rapid growth in the college-eligible population.

The top panel of Table 4 shows that if the share of students admitted to UT for each group remained at affirmative action levels, an additional 393 Hispanics would have gained admission to UT during the no policy period. This can be interpreted as the policy effect in student-units, or the cost in admissions that resulted from changes in group shares after the repeal of affirmative action. Therefore, the negative number for whites indicates a modest gain of 58 slots; clearly the largest beneficiaries of the judicial ban were other groups, but Asians in particular (Long and Tienda, 2008). Under the top 10% regime, whites incurred substantial admission costs – approximately 550 slots per year had their share of the admittee pool not dropped, but Hispanics incurred an additional 204 losses per year under the uniform admission law. By contrast, at TAMU whites gained approximately 219 additional admits during the no policy period and about 500 annual additional admits under the top 10% regime. Yet if Hispanic admission shares remained at their affirmative action level, they would have gained approximately 366 and 493 additional admits during the no policy period and the top 10% regime, respectively.

Table 4.

Estimated Changes in Admits and Enrollees under the Null Hypothesis of “No Change” since Affirmative Action ^a

	Change in admits if shares of students admitted remained at affirmative action levels
	University of Texas at Austin					Texas A & M University
		No Policy Regime		Top 10% Regime			No Policy Regime		Top 10% Regime
Group	Af. Act. Shares	% Δ in Shares	Cost in Admits	% Δ in Shares	Cost in Admits	Af. Act. Shares	% Δ in hares	Cost in Admits	% Δ in Shares	Cost in Admits
Whites	.651	0.5	-58	-3.8	550	.730	1.9	-219	3.6	-506
Hispanics	.159	-3.5	393	-1.4	204	.144	-3.1	366	-3.5	493
Other	.190	3.0	-335	5.2	-754	.126	1.2	-147	-0.1	13
Totals	1.000	0.0	0	0.0	0	1.000	0.0	0	0.0	0

	Change in enrollees if shares of admitted students who enrolled remained at affirmative action levels
Group	Af. Act. Shares	% Δ in Shares	Cost in Enrollees	% Δ in Shares	Cost in Enrollees	Af. Act. Shares	% Δ in Shares	Cost in Enrollees	% Δ in Shares	Cost in Enrollees
Whites	.664	0.5	-32	-4.1	354	.782	2.3	-142	4.0	-340
Hispanics	.150	-2.3	160	-0.6	51	.125	-2.7	171	-3.1	271
Other	.186	1.8	-128	4.7	-405	.093	0.4	-29	-0.9	69
Totals	1.000	0.0	0	0.0	0	1.000	0.0	0	0.0	0

Note: Number of observations is 169,547 and 140,472 for UT-Austin and TAMU, respectively. Data for TAMU Exclude 2003

^aCost in admits = (total regime admits × group specific affirmative action share) – groups’ actual admits during regime. Cost of admits is divided by 5 for UT and 4 for TAMU to obtain the yearly average during the top ten percent regime. The previous formula is repeated using the analogous information for enrollment.

The bottom panel of Table 4 reports the change in enrollees if the group shares in enrollment remained at affirmative action levels. The patterns are similar to those for admissions. At UT, the slight increase in whites’ enrollment share during the no policy period suggests that they benefitted from the ban on affirmative action, albeit slightly; however, the four percent decline in their share of admittees during the top 10% regime cost them about 354 enrollees annually. For Hispanics, their decline in share of admittees who enrolled at UT after the repeal of affirmative action cost them 160 enrollees during the no policy period and about 51 slots annually under the top 10% regime. At TAMU, whites benefitted greatly under the top 10% regime increasing their annual numbers by 340 as the number of Hispanics on campus fell 171 to 271 annually post-Hopwood.

Explaining Group Differences in Admission and Enrollment Rates

Although informative, the findings discussed above do not account for group differences in characteristics associated with college admission prospects. In particular, the observed minority-white gaps likely reflect group differences in academic outcomes and high school quality, which is related to application behavior and college readiness (Long and Tienda, 2010; Koffman and Tienda, 2008). Thus, in the next set of analyses we examine how admission and enrollment outcomes—shown in Model 1 of Table 5—change after accounting for variation in applicants’ achievement in Model 2 (i.e., SAT/ACT test scores and class rank) and high school characteristics (i.e., class size, public-private status, and percent of students ever economically disadvantaged). The set of high school characteristics are only included in Model 3; rather than report the point estimates, inclusion of these covariates is indicated in the last row of Table 5. Given that the data are for the population of students who applied to both UT and TAMU from 1993 to 2003, we do not provide p-values.

Table 5.

Estimates of Policy Effects on Admission and Enrollment at UT and TAMU

	Admitted						Enrolled a
	UT-Austin			TAMU			UT-Austin			TAMU
Ind. Variables	(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)
Group (Whites Ref.)
Hispanics	.032	.133	.121	.123	.179	.169	-.046	-.092	-.095	-.115	-.154	-.148
Policy by Group
No Policy(NP)	.115	.123	.120	.059	.056	.044	.061	.062	.039	-.044	-.048	-.065
NP × Hisp.	-.110	-.125	-.138	-.160	-.161	-.176	.019	.030	.018	.006	.013	.017
Top 10%(TT)	-.003	-.004	-.030	.001	.003	-.004	.058	.067	-.014	.028	.025	.008
TT × Hisp.	-.069	-.081	-.088	-.154	-.156	-.169	.026	.026	.043	-.015	-.011	-.007
Academics
SAT/ACT b	---	.010	.011	---	.008	.009	---	-.004	-.003	---	-.004	-.004
TT Class rank	---	.282	.261	---	.309	.274	---	-.062	-.060	---	-.014	-.013
Constant	.717	-.575	-.679	.739	-.265	-.415	.536	1.057	.976	.618	1.107	1.120
R2	.008	.305	.326	.009	.245	.259	.005	.027	.106	.024	.041	.045
Controls Included	No	No	Yes	No	No	Yes	No	No	Yes	No	No	Yes
N =	210,037			156,848			151,900			117,060

Note: Control variables include sex, class size, percent of high school receiving free or reduced lunch, and public/private status of high school. Indicator variables for students with missing values on each covariate are also included in the regressions.

^aData excludes students who enrolled but were not granted formal admission (e.g., waitlisted, deferred enrollment), which corresponds to 3 percent of the sample at UT.

^bSAT/ACT are composite scores divided by 10; ACT scores were converted to the SAT scale. Therefore, the estimates represent the average change in the outcome associated with every 10-point increase in test scores along the SAT scale.

Results of the first model, which are reported in the left-hand panel of Table 5 for both UT and TAMU, predict the proportion of students admitted to each institution during each policy regime relative to the white admissions rate under affirmative action. As such, the coefficients in the first model replicate the findings reported in the top panel of Table 3 as proportions rather than percentages. For example, Model 1 for UT shows that relative to the proportion of whites admitted to UT during affirmative action (given by the constant of .717), a greater proportion of Hispanics (.032 more) gained admission to UT during the same time period. These proportions correspond to the 71.7 percent (.717 × 100) and 74.9 percent ([.717 + .032] × 100) admission rates observed in Table 3 for whites and Hispanics, respectively. The first coefficient for each policy regime shows that the proportion of whites admitted to UT increased by .115 during the year when neither the top 10% or affirmative action were in force (No Policy or NP), and decreased by -.003 during Top 10% admission regime relative to their admission rate under affirmative action. These coefficients correspond to the 83.2 percent ([.717 + .115] × 100) and 71.4 percent ([.717 - .003] × 100) displayed in the top panel in Table 3 for white students at UT. The second coefficient associated with each policy regime represents the Hispanic-white difference in admission rates during the no policy year (NP × Hisp.) and the Top10% (TT × Hisp.) regimes relative to the admission rates for both groups under affirmative action. For example, the -.110 for NP × Hisp. corresponds to the 75.4 percent admission rate for Hispanics in Table 3 ([.717 + .032 + 115 - .110] × 100).

The second model for each institution shows the same information as Model 1 after accounting for variation in applicants’ high school academic achievement. Thus, the coefficients represent the admission rates for each policy regime (in proportions) for Hispanic and white students with similar achievement levels. The second model shows that the admission rate for UT during affirmative action was 13.3 percent greater for Hispanics with achievement levels comparable to their white counterparts. Hispanics experience greater declines in admission rates than whites during each policy regime after affirmative action, even once achievement levels and high school characteristics are controlled.

The findings for TAMU show some differences from those observed for UT. First, the Hispanic advantage in admission rates over whites during affirmative action was 12.3 percent rather than 3.2 percent. Second, the changes in policy regimes resulted in much lower admission rates for Hispanics relative to whites compared with UT (as evident in the magnitude of the interaction terms represented by NP × Hisp. and the TT × Hisp.). Accounting for achievement and accentuates the magnitude of declines in Hispanics’ admission rates compared to whites, but especially under the top 10% regime. As expected, the coefficients for achievement show that admission rates increase for students with higher test scores and those ranked highly in their class. As revealed by the final model, high school characteristics are unrelated to gaps in admission rates (compare estimates with Model 2).

Parallel analyses for enrollment presented in the right hand panel show that Hispanics are less likely than whites to enroll at either flagship, conditional on admission, irrespective of the regime in place. Hispanics experienced greater increases in yield than whites at UT during the no policy period and top 10% regime relative to the affirmative action regime, despite persisting gaps. These findings persist even after accounting for achievement and high school characteristics. The enrollment findings for TAMU reveal that group differences in academic achievement and high school characteristics are not responsible for the patterns observed in Table 3. The shift to the no policy period was associated with similar declines in yield for both groups, but the rebound in yield during the top 10% regime was lower for Hispanics than whites. Finally, among the students granted admission, those with slightly lower levels of achievement decided to enroll. This might reflect that students with higher test scores are likely to gain admission into a greater number of schools and thus have more alternative options than admitted students toward the lower end of the achievement distribution (see Niu and Tienda, 2010).

In sum, the shift from affirmative action to the no policy period resulted in declines in the admission rates to both flagships for Hispanics relative to whites, even when comparisons are made among students with similar levels of achievement and from high schools with similar characteristics. Multivariate analyses that account for applicants prior academic achievement and high school characteristics modify admission and yield rates reported in Table 3 only negligibly.

Summary and Discussion

Our analyses show that changes in Texas college admission policies have been highly consequential for Hispanics, the largest and fastest growing segment of the State's population. Using data from the Texas Education Agency and from the administrative records of both UT-Austin and TAMU, we evaluate how Hispanic and white students fared across three policy regimes: affirmative action, no policy period, and the top 10% regime. Although it is commonplace to focus on admission and enrollment outcomes, our empirical analysis suggests that these outcomes are conditioned by the decision to apply (Long and Tienda, 2010; Koffman and Tienda, 2008). This conclusion echoes that reached by Brown and Hirschman (2006) based on the experience of the State of Washington, where use of race preferences in college admissions was outlawed by voters.

The empirical analyses produce three major findings. First, Hispanics’ application rates to the Texas flagship universities fell after affirmative action was banned; moreover, owing to rapid growth in the number of high school graduates, their disadvantage in percent of applicants relative to whites grew over time. Although the declines in application rates to both UT-Austin and TAMU averaged one percent or less, this implies an annual loss in Hispanic applicants ranging from nearly 204 to 390 at UT and from over 350 to nearly 500 at TAMU. Second, Hispanics’ admission rate to both UT-Austin and TAMU fell after the ban on affirmative action and reached its lowest point under the top 10% regime. This finding implies that the number of Hispanics eligible for enrollment to Texas flagship universities is reduced even further—a compounding of application and admission disadvantages that translates to fewer potential enrollees. Third, even with the declines in application rates for Hispanics since the repeal of affirmative action, our results suggest that Hispanics would gain substantial representation in Texas flagship universities if they maintained the admission shares they held during affirmative action. This result has profound policy implications that transcend admission regimes because they redirect attention away from the seemingly irresolvable differences about race or class rank preferences to encouraging greater numbers of qualified applicants to apply for admission.

That the expansion of the post-secondary education system has failed to keep up with the growth of the college-eligible population represents a formidable policy challenge because competition for access to the State's public flagships will continue to intensify in Texas, at least through 2015 (Tienda, 2006; WICHE, 2008); because legal and statutory challenges to race preferences and the top 10% plan show no sign of abating (Haurwitz, 2008; Schmidt, 2008); and because Texas invests less of its GDP on public education than several other states that have excellent public universities.¹³ Over the long term the post-secondary system will expand to accommodate slower growth of high school graduates, but the State faces enormous opportunity costs for continued underinvestment in the education of its fastest growing population. Educational underinvestment is seldom invoked as the culprit for the rising number of applicants denied admission to a four-year institution in the state, yet it is the ultimate cause of the college squeeze and a source of economic vulnerability for the state in the future. Comptroller Carole Keeton Strayhorn (2005) estimated a 500 percent return on every dollar invested in the state's higher education system.

In the short term, however, cultivating college-going cultures at under-resourced high schools is a potential high-impact, relatively low cost strategy to raise Hispanic college application rates. The Longhorn and Century Scholars programs developed by UT and TAMU, respectively, enabled economically disadvantaged top 10% graduates to attend their institutions. Domina (2007) provides some evidence that these programs, which were designed to benefit post-secondary institutions, also can improve high school climate by lowering absenteeism and fomenting college orientation via institutional partnerships. Finally, it warrants emphasizing that an admission guarantee cannot guarantee enrollment, particularly for students from limited economic means. That Hispanic students are disproportionately concentrated in low resourced high schools requires strong financial aid programs to ensure that successful applicants actually enroll and graduate from college.