On the eve of the Seattle teachers strike in September 2015, the Seattle Times condemned the impending walkout, accusing the union of “stiff-arming more than 50,000 kids and their families.” Yet the teachers insisted that their strike was about children’s education, not just teacher pay, and commanded widespread support from parents and the community at large.

Seattle teachers and administrators reached an agreement in one week, but the question of how unions affect public education is far from settled. According to the recent Education Next poll (see “The 2015 EdNext Poll on School Reform,” features, Winter 2016), the public is divided as to whether teachers unions have a positive or negative impact on schools, and, until now, researchers have been unable to document the effects of collective bargaining on students’ long-term outcomes.

Today, more than 60 percent of teachers in the United States work under a union contract. The rights of teachers to unionize and bargain together have expanded dramatically since the late 1950s, when states began passing “duty-to-bargain” (DTB) laws that required school districts to negotiate with teachers unions in good faith. Recently, though, states such as Wisconsin, Indiana, Michigan, and Tennessee have sought to weaken the ability of teachers unions to negotiate contracts in K–12 education.
Advocates for these restrictions claim that unions have a negative effect on the quality of public education and, therefore, students’ life chances. Those in favor of teacher collective bargaining, on the other hand, argue that unions make the education system more effective by empowering teachers who are in the classroom and by giving them a role in shaping their working conditions. Due to data limitations, however, empirical research has not credibly addressed the critical question of how teacher collective bargaining influences student outcomes.

In this study, we present the first evidence on how laws that support teacher collective bargaining affect students’ employment and earnings in adulthood. We do so by first examining how the outcomes of students educated in a given state changed after the state enacted a duty-to-bargain law, and then comparing those changes to what happened over the same time period in states that did not change their collective-bargaining policies.

We find no clear effects of collective-bargaining laws on how much schooling students ultimately complete. But our results show that laws requiring school districts to engage in collective bargaining with teachers unions lead students to be less successful in the labor market in adulthood. Students who spent all 12 years of grade school in a state with a duty-to-bargain law earned an average of $795 less per year and worked half an hour less per week as adults than students who were not exposed to collective-bargaining laws. They are 0.9 percentage points less likely to be employed and 0.8 percentage points less likely to be in the labor force. And those with jobs tend to work in lower-skilled occupations.

Teacher Collective Bargaining in the United States

In the first half of the 20th century, teachers unions in the United States were predominantly professional organizations that had little say in contract negotiations between teachers and school districts. Starting with Wisconsin in 1959, however, states began passing union-friendly legislation that either gave teachers the right to collectively bargain or explicitly mandated that districts negotiate with unions in good faith. Duty-to-bargain laws in particular give unions considerable power in the collective-bargaining process, because they make it illegal for a district to refuse to bargain with a union, and because most of them require state arbitration if the two sides reach an impasse. The enactment of such laws led to a sharp rise in the number of teachers who joined unions and in the prevalence of collectively bargained contracts.

Between 1959 and 1987, 33 states passed duty-to-bargain laws (see Figure 1); just 1 (New Mexico) has done so since. Of the 16 states without such a law, 9 have legislation that permits teachers unions and districts to bargain if both sides agree to do so. In the remaining 7 states (Arizona, Georgia, Mississippi, North Carolina, South Carolina, Texas, and Virginia), collective bargaining is prohibited either by statute or by court ruling (see Figure 2).

How Might Collective Bargaining Affect Schools and Students?

Collective-bargaining laws strengthen teachers unions and give them greater influence over how school districts allocate their resources. A typical collective-bargaining agreement addresses a remarkably broad range of items: unions negotiate over salary schedules and benefits; hiring, evaluation, and firing policies; and rules detailing work and teaching hours, class assignments, class sizes, and nonteaching duties. By increasing union membership, collective-bargaining laws also heighten the influence of teachers unions in education politics at the state level.

Critics of teacher unionization argue that collective bargaining in public education has reduced school quality by shifting resources toward teachers and away from other educational inputs and by making it more difficult to fire low-performing teachers. Stronger unions may also have made it  harder for states to adopt policies aimed at improving school quality through enhanced accountability or expanded school choice.

Other arguments, however, suggest that stronger unions may benefit students. First, to the extent that teachers have expertise in creating effective learning environments, giving them more say over how resources are allocated could lead to better educational outcomes. Second, giving teachers a greater voice in the structure of their working environments could lead them to become more productive and could attract more effective teachers into the profession. Finally, teachers unions could use their political muscle to support additional investment in public education and other policies that might enhance school quality.

In sum, there’s little dispute that collective bargaining alters how school districts operate and shifts the balance of power in state education politics, but there is wide disagreement over whether these changes affect student outcomes negatively or positively. This disparity of opinion highlights the importance of turning to empirical evidence.

Our Study

The central challenge in studying the effects of collective-bargaining policies is that states with strong protections for collective bargaining tend to be very different from states with weaker protections. For example, the states without duty-to-bargain laws are located mainly in the South, where student achievement has historically been low for reasons unrelated to collective bargaining. States such as Massachusetts and Minnesota demonstrate that it is possible to have a relatively high-performing school system in the presence of strong unions, but they tell us very little about the effects of collective bargaining itself.

Our study overcomes this hurdle by examining how the outcomes of students educated in specific states changed over the years when most states enacted collective-bargaining laws. We focus on duty-to-bargain DTB laws specifically because these laws led to greater growth in unionization and collective bargaining than did other forms of state union laws. And we focus on entire states rather than on specific school districts, because the passage of a duty-to-bargain law might have consequences even for students in districts that did not unionize. Unions’ political activities influence education policies statewide, and nonunionized districts operating in a DTB state may tend to adopt policies supported by teachers in order to avoid unionization efforts.

We do not directly compare students educated in duty-to-bargain states with students in non-DTB states, because such comparisons would clearly yield outcome differences unrelated to collective bargaining (for instance, differences caused by higher or lower poverty rates). Also, we eschew simple “before and after” comparisons within a state, because again, any observed outcome differences could be the result of factors other than collective bargaining (for instance, social and political changes since the 1960s that affected K–12 education). Our strategy, therefore, is to compare the differences in outcomes for students educated in the same state (before and after the DTB law passed) to the differences in outcomes for students in non-DTB states over the same time period.

When making these comparisons, we adjust for the share of the student’s state birth cohort that is black, Hispanic, and white, and the share that is male. We also take into account two policy changes enacted by many states during this same time period that may have affected student outcomes: school finance reforms and changes in the generosity of state earned-income tax credits. If the rollout of those policies coincided with the passage of duty-to-bargain laws, unadjusted before-and-after comparisons could yield misleading results. Adjusting for these two variables turns out to make little difference in our results but strengthens our confidence that collective bargaining is responsible for the effects we document.

Our measure of the extent to which each student is exposed to collective bargaining varies from 0 to 1 and is defined as the proportion of the student’s school years in which a duty-to-bargain law was in effect in his or her state. A value of 1 means that a DTB law had been enacted by the time students in the birth cohort were six years old (in time for first grade); thus, they were exposed to the law throughout their entire K–12 education. The variable is 0 for students whose birth cohorts had no exposure, either because they were over 18 when a DTB law was passed or because they were born in a state that did not impose a duty to bargain.

Data

The data for our analysis come from two main sources. The first is the National Bureau of Economic Research collective-bargaining law dataset that contains, for each state and year since 1955, collective-bargaining laws for each type of public-sector worker. We combine the collective-bargaining information for teachers with 2005–2012 American Community Survey (ACS) data containing detailed information on the educational attainment and labor market success of representative samples of adults in each state.

We look specifically at ACS data for individuals between the ages of 35 and 49, because people in this age group typically have completed their education and are at a juncture when yearly earnings are indicative of lifetime earnings. We examine birth cohorts ranging from 1956 to 1977, which correspond to students who attended school from 1962 to 1995. As shown in Figure 1, these schooling years correspond with the dramatic rise in duty-to-bargain laws in the United States.

Results

These data enable us to examine the effects of teacher collective-bargaining policies on multiple indicators of students’ labor-market success. Taken as a whole, our results clearly indicate that laws supporting collective bargaining for teachers have adverse long-term consequences for students.

Earnings. We find strong evidence that teacher collective bargaining has a negative effect on students’ earnings as adults. Attending school in a state with a duty-to-bargain law for all 12 years of schooling reduces later earnings by $795 dollars per year (see Figure 3). This represents a decline in earnings of 1.9 percent relative to the average. Although the individual effect is modest, it translates into a large overall loss of earnings for the nation as a whole. In particular, our results suggest a total loss of $196 billion per year accruing to those who were educated in the 34 states with duty-to-bargain policies on the books.

Hours worked. Consistent with this reduction in earnings, we also find that exposure to a duty-to-bargain law throughout one’s school years is associated with a decline of 0.49 hours worked per week. This is a 1.4 percent decline relative to the average, and it suggests that a reduction in hours worked is a main driver of the lower earnings.

Wages. The reduced earnings caused by unionization could also reflect lower wages, and the evidence suggests a negative relationship between collective-bargaining exposure and wages. While this relationship is not statistically significant, it is consistent with our other results and suggests that teacher collective bargaining may also have a modest adverse effect on average wages.

Employment. The fact that teacher collective bargaining reduces working hours suggests that duty-to-bargain laws may also affect employment levels. In fact, when we use the share of individuals who are employed as the outcome variable, we find that duty-to-bargain laws reduce employment. Specifically, exposure to a duty-to-bargain law for all 12 years of schooling lowers the likelihood that a worker is employed by 0.9 percentage points. Duty-to-bargain laws have no impact on unemployment rates, however, suggesting that they reduce employment by leading some individuals to drop out of the labor force altogether.

Occupational skill level. Finally, we analyze the effects of collective bargaining on the skill level of a student’s selected occupation, as measured by the share of workers in that occupation who have any education beyond a high school diploma. The results suggest yet another negative effect: being exposed to a duty-to-bargain law for all 12 years of schooling decreases the proportion of such workers in an occupation by almost half of a percentage point (or 0.6 percent relative to the average). This effect is modest in size, but it implies that teacher collective bargaining leads students to work in occupations requiring lower levels of skill.

Educational attainment. The reduced earnings and labor force participation associated with teacher collective bargaining raise the possibility that affected students may have completed less education. Our analysis, however, finds little evidence of bargaining power having a significant effect on how much schooling students completed. This finding is surprising in light of the substantial labor-market effects we document, but it comports with prior research that has found no effect of duty-to-bargain law passage on high-school dropout rates.

Additionally, educational attainment is but one measure of the amount of human capital students accumulate. Even if students do not complete fewer years of education, they may be acquiring fewer skills while they are in school. We believe that our results concerning earnings and employment are driven by other aspects of school quality that are not reflected in educational attainment, and they reinforce the importance of studying labor-market outcomes directly in order to understand how major reforms such as the enactment of teacher collective-bargaining laws affect students’ life outcomes.

Policy Implications

This study provides the first comprehensive analysis of the effect of teacher collective bargaining on the long-term educational and labor market outcomes of affected children. We find that exposure to a duty-to-bargain law while in grade school lowers earnings and leads to fewer hours worked, reductions in employment, and decreases in labor force participation. Occupational skill level also declines. However, educational attainment is unchanged by exposure to these laws.

These results contribute new information to the contentious debate occurring in many states over limiting the collective-bargaining rights of teachers. For example, in 2011 Wisconsin passed legislation that greatly reduced the ability of teachers to bargain with school districts (see “Limits on Collective Bargaining,” features, Fall 2013), and in 2014 Michigan passed a public employee right-to-work law that sought to limit union negotiating power. Not surprisingly, teachers unions and their allies responded to these laws with fierce opposition.

At the core of this debate lies the question of how teacher collective bargaining affects student outcomes. Our results suggest that lawmakers in Wisconsin and Michigan have evidence on their side. However, we urge caution when generalizing these findings to current students, because the cohorts we analyze in this study, most of whom attended school in the 1970s and 1980s, were educated in an environment very different from today’s. Some of the adverse effects of teacher collective bargaining we document could be driven by how teachers unions interacted with aspects of the school system that are no longer relevant. On the other hand, the economy’s growing demand for skilled workers may mean that policies affecting human capital accumulation matter more now than ever. Future research should investigate whether and how the effects of teacher collective bargaining have changed over time.

Moreover, our results say little about why the enactment of collective-bargaining laws has harmed student outcomes. Perhaps collective bargaining has made it more difficult for school districts to dismiss ineffective teachers or to allocate teachers among schools. Or perhaps the political influence of teachers unions at the state level has interfered with efforts to improve school quality. Identifying the factors at play could shed light on the most promising strategies for reform. In the meantime, however, our evidence points to the conclusion that collective bargaining in public education has been a bad deal for American students.

Michael F. Lovenheim is associate professor of policy analysis and management at Cornell University and a faculty research fellow at the National Bureau of Economic Research. Alexander Willén is a doctoral student in policy analysis and management at Cornell University.      

Ocean animals was the theme in pre-kindergarten classes at a California school in early May. Some pre-K teachers introduced “octopus” and “tentacle,” while others taught “pulpo” and “tentaculo.” In all the pre-K classes, children acted out vocabulary words with hand movements, sang songs, and played a guess-the-ocean creature game. Then they moved to tables, where some of them painted paper octopuses, while others gingerly smelled, touched, and then dangled little octopuses from a local fish market.

Down the hall, kindergartners wrote about their favorite desert animals, talked with a partner about where cacti grow, and chanted about biomes:

Arid deserts drying

Luscious forests growing

Polar caps freezing

Green prairies growing …

First graders discussed a story their teacher had read aloud in which a grandfather remembers courting his wife. In Common-Core style, they cited “clues” from the text of the grandfather’s feelings and learned words to describe emotions.

“How do you know he’s happy?” asked Heidi Conti, the teacher.

“He ‘winked’ at the boy,” answered a student.

“Good,” responded Conti. “You made an inference.”

Ninety-five percent of students at Redwood City’s Hoover School, in San Mateo County, come from low-income and working-class Latino families, and nearly all start school as English language learners (ELLs). The elementary and middle school piloted the Sobrato Early Academic Language (SEAL) program in 2009 in hopes of raising reading and math scores and moving more students to the college track.

Programs like SEAL offer a fresh approach to educating English language learners. The focus in schools is shifting “from the language of instruction to the quality of instruction,” says Kenji Hakuta, a Stanford professor who specializes in language learning. As a result, long-standing debates about whether English learners should be taught only in English or also in their native tongue feel increasingly obsolete.

Spanish—and Somali

Close to 5 million U.S. students—about 9 percent of public school enrollment—are ELLs. Three-quarters of them were born in the U.S. and are the children—or grandchildren—of immigrants, according to a Migration Policy Institute analysis of 2013 U.S. Census data. Nearly 80 percent of ELLs come from Spanish-speaking homes, but the rest may speak Chinese, Vietnamese, French/Haitian Creole, Arabic, or any one of hundreds of other languages. In Maine, the most common language spoken by ELLs is Somali (see Figure 1). South Carolina’s second most-common language, after Spanish, is Russian. Illinois schools enroll students from families that speak Arabic, Polish, Chinese, and Urdu.

ELLs usually learn “social” or “playground” English quickly, but many struggle to master the “academic” English vocabulary needed to read complex texts, write clearly, and understand concepts.

Pushed by No Child Left Behind’s (NCLB) accountability measures and the college-for-all movement, educators nationwide have raised expectations for children from immigrant families. Despite moves to ease limits on bilingual education in California and Massachusetts, more ELLs are now learning in English, taught by teachers who use an array of strategies to reach nonfluent students.

Students need strong reading and writing skills in English to have any chance of success in college. Long-term ELLs—those who haven’t reclassified after five years—often drop out of high school or graduate without the skills needed to train for a job or pass a community college class. Reaching English proficiency by middle school is critical for success in high school and beyond. Those who do are likely to take college-prep courses. Those who don’t are not. Most educators would like students to be bilingual and bicultural, but college readiness comes first.

“We have kids who start school in kindergarten as English learners and they’re still English learners 12 years later”—if they stay in school, says SEAL director Laurie Olsen. With funding from the Sobrato Family Foundation, Olsen designed SEAL to move students to English proficiency by 3rd grade. The pre-kindergarten to grade 3 program is aligned with the Common Core State Standards, Next Generation Science Standards, and social studies standards. Elementary schools without a pre-K work closely with feeder preschools.

Starting in pre-K, children talk, sing, chant, move, explore, experiment, and play in language-rich, text-rich, information-rich environments. They dictate stories to volunteers, write letters, keep journals, and see their writing “published” in bound books. SEAL teachers help students develop the sophisticated vocabulary they’ll need to read, discuss, and write as they move through school.

English language development is taught in the context of thematic science and social studies units, broadening children’s knowledge of the world. In all subjects, including math, teachers use “strong, powerful, complex” language.

Most new Hoover students start in bilingual classes, with parental waivers from California’s English immersion policy.

Before SEAL, Hoover’s ELLs showed little progress on the California English Language Development Test, says principal Amanda Rothengast. Many entered and left middle school as ELLs.

Whether they start in Spanish or English, “our SEAL kids are reclassifying as proficient in English by 4th grade,” says Rothengast.

SEAL is spreading quickly in California, as schools seek ways to teach Common Core standards to all students. Sixty-five preschools and elementary schools will use SEAL this school year, says Olsen.

The Rise of Bilingual Ed

The Civil Rights Act of 1964 banned discrimination on the basis of race, color, or national origin in programs receiving federal funds. The Bilingual Education Act of 1968, also known as Title VII, provided grants for programs to meet the “special educational needs” of young students with limited English skills. It encouraged but did not require bilingual instruction.

In the 1974 Lau decision, the U.S. Supreme Court held that failing to provide services for students learning English violated the Civil Rights Act. Chinese families in San Francisco had challenged placement of children in mainstream classrooms with no help in learning English.

That same year, Congress passed the Equal Educational Opportunities Act (EEOA), which required public schools to take “appropriate action” to help students overcome barriers to English fluency, ensure “access to the core curriculum,” and integrate them with native English speakers, as much as possible.

That was perceived as a mandate for teaching children in their parents’ language, writes Paul Peterson in Saving Schools. Furthermore, while bilingual education was sold as the best way to help students learn English, advocates also wanted to help children maintain their parents’ language and culture. In the Nixon administration, Secretary of Health, Education, and Welfare Elliot Richardson told Congress that districts must “recognize and value” minority children’s “cultural environment” to develop their self-esteem.

“In other words,” explains Peterson, “a Republican administration was endorsing bilingual education as an end in itself, not simply as a means for immigrants to become proficient in English.”

Federal policy never mandated native-language instruction, but some states required it, if there were enough children speaking the same language at the same grade level. “In the 1970s and 1980s, policies and practice favored bilingual education, in which children were taught partially or entirely in their native language, and then transitioned at some point during the elementary grades to English-only instruction,” wrote Johns Hopkins researcher Robert E. Slavin and colleagues in a study published in 2010.
Advocates of transitional bilingual education said children would develop their native-language skills and learn academic subjects, then make the switch to English. Students were expected to be bilingual and biliterate by the end of elementary school. But many were not proficient in either language. In some schools, bilingual classes separated children from the least-educated families from their peers and limited their access to native English speakers.

For many years, children from non-English-speaking families “were placed in separate classes with lower demands,” says Carrie Hahnel, deputy director of research and policy analysis at Education Trust-West.

Rosa Torres left Redwood City in 1990 when Hoover’s principal refused to transfer her daughter out of undemanding bilingual classes. In Cupertino, which had too much language diversity to offer native-language instruction, daughter Angelica worked hard to catch up, Torres told Glenn Garvin for a 1998 Reason magazine story, titled “Loco, Completamente Loco.” A college graduate now, Angelica speaks little Spanish and her own daughter speaks none. Her granddaughter earns honors on every report card, says Rosa Torres proudly. In her mind, that’s what counts.

“In the old model, expectations were very low,” confirms Veronica Aguilar, director of English learner support in the California Department of Education.
English grammar and vocabulary were taught out of context, says Hakuta. Little time was spent on science or social studies, so students didn’t build a strong, broad foundation of knowledge.

“When I was a bilingual teacher, there wasn’t enough rigor,” recalls Frances Teso, who taught in a San Jose elementary school before founding a charter school. “We called it the Pobrecito Syndrome. ‘Those poor kids, their parents aren’t educated. They have so many problems.’ It’s true, but what can we do here at school about it? We didn’t use data to tell if kids had learned the lesson or not.”

Since bilingual teachers were in short supply, schools often hired aides to teach in Spanish or imported teachers from Spanish-speaking countries who spoke little English. “We never had the teachers to pull off bilingual education,” says Michael Kirst, president of the California State Board of Education and a Stanford education professor emeritus.

No Más Pobrecito

The popular revolt against bilingual education started in California. Ron Unz, a software entrepreneur, read about a Los Angeles protest: immigrant parents were demanding their children be taught English. Unz wrote and financed an initiative requiring English immersion, unless parents sign a waiver requesting a bilingual alternative. Proposition 227, known as English for the Children, won a 61 percent majority in 1998.

Arizona voters passed a similar measure in 2000, and Massachusetts followed suit in 2002. About one-third of ELLs nationwide live in those three states.
According to a U.S. Education Department (ED) analysis, from 1993 to 2003 the proportion of English learners receiving “some” or “significant” native-language instruction decreased from 53 percent to 29 percent.

After the passage of Proposition 227, the proportion of California ELLs in bilingual education classes dropped from 30 percent to 8 percent, according to a five-year follow-up study by American Institutes for Research and WestEd for the state education department.

Many California ELLs never met their district’s criteria to exit the program, the report found. Less than 40 percent achieved “fluent English proficient” status after 10 years in California schools. After 227 passed, reclassification rates rose.

Achievement also rose for ELLs and other students over the period, the study reported. Other accountability-related reforms were implemented at the same time, however, and may have had an impact. The study found no clear advantage for English immersion or bilingual education. High-performing schools employed skilled teachers, who used data to assess teaching and learning and adjusted instruction based on their students’ performance, the study found. These schools had “a well-defined, rigorously structured plan of instruction” for English learners.

“What matters most in the education of English language learners is the quality of instruction,” Slavin and colleagues from Johns Hopkins concluded. To isolate the effect on student learning of the language of instruction, the researchers followed students who were all learning reading via the Success for All curriculum but were randomly assigned to structured English immersion or transitional bilingual education. The groups had similar English and Spanish reading skills by 4th grade, the study found.

“We used to think that what’s good for all students is good for English learners, but now we think what’s good for English learners is good for all students,” says Robert Linquanti, who directs English learner evaluation and accountability support at WestEd.

Beyond the Bilingual Wars

The emphasis on the quality of instruction has “cooled off” the bilingual wars, says Stanford’s Hakuta.

“When 227 passed, I thought it would be a disaster,” says Teso. “Now I think it was a good thing in some ways. It eliminated a lot of low-quality bilingual programs and opened the door to better-quality programs.” In 2007, Teso founded Voices College-Bound Language Academy, a high-performing K‒8 charter school in San Jose that uses a modified dual-immersion model to teach English and Spanish.

As old-style bilingual programs are phased out, such “dual-immersion” or “two-way” bilingual schools are gaining in popularity. This model often mixes the children of English-speaking parents with ELLs, offering qualified teachers and a coherent, rigorous curriculum.

Houston is expanding dual-immersion schools, including a brand-new school where students will learn in Arabic for half the day. The programs have drawn white, Latino, and black students from English-speaking families, say school officials.

The push for bilingualism often comes from English-speaking parents. “Parents who live in affluent suburban communities want their children to learn a second language,” says Hahnel at Education Trust-West.

Kate Menken, associate professor of linguistics at the City University of New York, says New York City parents “see dual immersion as an alternative to gifted and talented programs.”

Miami-Dade superintendent Alberto Carvalho wants to stop teaching 30 minutes of Spanish a day in elementary school—nobody learns much—and offer a rigorous, intensive dual-immersion program for students with grade-level skills.

But Miami schools can’t find enough teachers who are literate in Spanish, according to the Miami Herald. “When you go into the intermediate grades and you really have to teach grammar rules…you can’t do that just by having a Hispanic last name,” says Beatriz Zarraluqui, Miami-Dade’s director of bilingual education and world languages. It’s likely that dual immersion will remain a popular and high-quality option, but the shortage of bilingual teachers will limit its reach.

English immersion remains controversial—with educators, if not with parents.

Arizona’s ELL approach—intensive English instruction for newcomers—has survived a series of legal challenges. In June 2015, the U.S. Court of Appeals for the Ninth Circuit ruled that Arizona’s approach doesn’t segregate students illegally or deny access to academic content.

The Massachusetts legislature is considering Language Opportunities for Our Kids (LOOK), which would make it easier to offer bilingual programs.
Californians will vote in November 2016 on the Multilingual Education Act, which would repeal most of 227, allowing children to be placed in non-English instruction without parental waivers.

“I see no political or educational support,” for a return to the old days, says Kirst.

Unz notes that polls show that Latino parents want their children taught in English.

In Los Angeles, some Spanish-speaking parents lie on home-language surveys to keep their children from being classified as English learners, the Associated Press (AP) reported in 2014. They fear their children will be assigned to low-level classes or pulled out of class for tutoring.

Nieves Garcia, who came to Los Angeles from Mexico at the age of six, remained an ELL even after she mastered English. Garcia’s husband doesn’t speak English, but she lied on the survey to keep her daughter from being labeled. “I just said we spoke English, English, English, and English,” Garcia told the AP.

NCLB and Common Core

Rising state standards and accountability initiatives have spotlighted the weak academic progress of many ELLs. No Child Left Behind required states to test most ELLs and report their subgroup scores. Schools faced increased pressure to move students to English fluency and raise reading and math scores.  Around the time of the law’s passage in 2001, test scores of Hispanic students in reading and math did begin to rise but remain well below those of their non-Hispanic white peers (see Figure 2).

“Since NCLB became law, every state has upgraded its monitoring of the academic performance of English language learners,” said Don Soifer of the Lexington Institute, testifying before a congressional committee in 2006. “A common trend has been toward a single statewide method for identifying, assessing, and redesignating ELLs.”

“NCLB was the instrument that, after decades of incursions on the original Bilingual Education Act of 1968, finally removed all references within the Department of Education to Bilingual Education,” wrote Patricia Gándara and Gabriel Baca in the journal Language Policy in 2008. “The Office of Bilingual Education and Minority Languages Affairs (OBEMLA) became under NCLB, the Office of English Language Acquisition, Language Enhancement, and Academic Achievement for Limited English Proficient Students, popularly known as OELA.”

NCLB could stand for “No Child Left Bilingual,” say Menken and Cristian Solorza, an instructor at Bank Street College of Education. Despite New York’s pro-bilingual law, New York City schools are replacing bilingual programs with English-only programs to meet accountability requirements, according to their research. Principals said they switched to English to improve test scores, which are used to judge schools’ performance. If students are going to be tested in English, administrators want them taught in English.

Widespread adoption of Common Core standards is also accelerating the move away from bilingual education, Menken and Solorza believe. Schools are teaching in English to prepare students for Core-aligned tests. In addition, it’s hard for teachers to find high-quality Spanish-language curricula aligned to the Core.

The recent House and Senate revisions of No Child Left Behind retained both annual testing and the requirement that scores be reported separately for various subgroups of students within each school, including English language learners. As a result, it seems likely that ELLs’ progress toward English proficiency will continue to be factored into school accountability measures under a new Elementary and Secondary Education Act.

“We see increased national attention to English learners,” says Maria Millard, a policy analyst at Education Commission of the States (ECS). States are monitoring former ELLs’ performance for longer. Washington State is tracking former ELLs through high school.

With better state data comes greater understanding of student differences. The 12-year-old refugee who’s been out of school for several years has very different—and more expensive—needs than the five-year-old preschool graduate who speaks a mix of Spanish and English.

Much of the action is happening at the state level, where “states are scrambling to rethink funding formulas,” says Micah Ann Wixom, an ECS policy analyst. Some are shifting to formulas that link dollars to students’ needs.

At the federal level, funding hasn’t kept pace with the growing number of students from non-English-speaking families. ED put money into competitive grants rather than categorical programs such as Title III, the main vehicle for federal ELL funding. Title III funding peaked at $750 million in 2010‒11 and is estimated to be $737 million in 2015‒16.

The department is offering NCLB waivers to states that agree to teach all students, including ELLs, to high-level “college and career-ready” standards, such as the Common Core and Next Generation Science standards. In the department’s new framework, the top goal is to “ensure all English Learners are college and career ready for a global society by building on students’ linguistic and cultural assets.” Civil rights guidelines released this year also stress providing access to high-level curricula.

Getting Real

English learners have struggled to make it through high school. Now they’re supposed to meet much more rigorous standards. How will they do in the new era?

“Common Core raises cognitive demands and expectations for all students,” says Hahnel of Education Trust-West.

Teachers are worried, but there’s also “real enthusiasm” for the standards, says WestEd’s Linquanti. “These are brave and exciting times.”

Back at Hoover in Redwood City, kindergartners look at Van Gogh’s Sunflowers and draw their own still-life depictions of plants while learning vocabulary such as “germinate.”

In 1st and 2nd grade, children can put on a lab coat at the “inquiry center” to be a geologist studying rocks or a paleontologist studying fossils. They write their findings in journals.

When a class studies insects, parents are asked to take them for a walk to see how many insects they can identify, then explain how they know that’s an insect. On the classroom walls are visual aids such as charts, graphics, timelines, Venn diagrams, and photos to build comprehension.

Because of Common Core, “there’s more focus on kids using language to explain their reasoning, construct an argument, and point out evidence in the text,” says Hakuta.

Common Core math requires students to explain their answers, for example. “Scores will go down for those who’ve traditionally done well in math, especially for immigrants.”

It’s worth it, Hakuta believes. “We’re getting real about what it means to be prepared for college.”

Joanne Jacobs, the author of Our School, is a freelance writer and education blogger at joannejacobs.com.

Education analysts often compare U.S. schools to those in Finland, Korea, Poland, even Shanghai. Surprisingly, the nation of Germany rarely appears in this discourse, even though it has much in common with the United States. Each of the two nations is the largest democracy, with the biggest economy, on its continent. And each has a diverse population, strong unions, a federal system of government, demand for a skilled workforce, and a school system that in 2000 was badly in need of reform.

After examining schools and public opinion in both countries, our team at Education Next, together with scholars at the University of Munich, were left with an intriguing question: why have German schools made significant progress since the turn of this century, while U.S. schools have not?

You can read our report at hks.harvard.edu/pepg, but here are some key facts:

Back in 2000, U.S. and German students at age 15 were performing at roughly the same level on international tests in reading, math, and science, and shortly thereafter, a spirited school-reform movement was launched in both countries. And yet, by 2012, German 15-year-olds were outscoring their U.S. peers by 32 points in math, a difference representing more than a year’s worth of learning. In science, scores were 27 points higher in Germany, and in reading, 10 points (Figure 1).

Notably, the German gains did not come at the price of equity. In both countries, the difference between the scores of students in the upper and lower quarters of the test-score distribution narrowed considerably between 2000 and 2012. Gap closing was steeper in Germany (Figure 2).

Larger gains in effectiveness and equity may be related to higher German teacher salaries. In 2013, at the higher-secondary grade level, entry-level German teachers received 51 percent more in wages, while those with 15 years of experience received 41 percent more (roughly $70,000 in Germany compared to $50,000 in the United States). Beginning elementary-school teachers earned about $47,500 in Germany, about $10,000 more than the entering U.S. teacher.

Somehow, the Germans find a way of paying these generous salaries without imposing higher costs on taxpayers. Per-pupil expenditures for secondary education run about $10,300 in Germany and $12,700 in the United States. At the elementary-school level, the numbers are $7,600 in Germany, $11,000 in the United States.

Despite the fact that Germans are getting more for less, they are not as satisfied as Americans are with their local schools and teachers, according to our polls. On the traditional A-to-F scale, 47 percent of people in the United States give their local schools an A or a B, as compared to just 42 percent of Germans. Similarly, the typical U.S. adult gives an A or a B to 51 percent of the teachers at the local school, while Germans rate just 41 percent of their teachers that highly.

Why did German schools improve after 2000 when U.S. schools did not? Was it because school reform in Germany was pushed forward by a consensus among state-level political leaders, educators, teachers unions, and the public at large, while in the United States, union and partisan opposition quickly emerged?

Why are costs in Germany so low when teachers are paid so well? Is it because German schools are run by the states, with little federal direction and no local school boards at all? Are operations more efficient when schools are run mainly by one tier of government instead of by an often fractious federal-state-local partnership?

Why do Germans give teachers lower grades, when students there are learning more? Is it because German students, if they want to get ahead, must pass exams at about age 10 and again upon finishing secondary school? When students must pass high-stakes exams, do both the parents and the public at large expect more from their teachers?

Of course, no firm conclusions can be drawn from any comparison between just two countries. But it’s still worth pondering these questions when two similar countries see such dramatic differences in educational outcomes.

–Paul E. Peterson

At a time when the national conversation is focused on lagging upward mobility, it is no surprise that many educators point to poverty as the explanation for mediocre test scores among U.S. students compared to those of students in other countries. If American teachers in struggling U.S. schools taught in Finland, says Finnish educator Pasi Sahlberg, they would flourish, in part, because of “support from homes unchallenged by poverty.” Michael Rebell and Jessica Wolff at Columbia University’s Teachers College argue that middling test scores reflect a “poverty crisis” in the United States, not an “education crisis.” Adding union muscle to the argument, American Federation of Teachers president Randi Weingarten calls poverty “the elephant in the room” that accounts for poor student performance.

But does the room actually contain the elephant?

To prove that poverty is the major factor driving America’s meager academic achievement, at least two of the following three claims need to be established:

1. Poverty is related to lower levels of student learning.

2. America’s poor students perform worse than other countries’ poor students.

3. The poverty rate in the United States is substantially higher than the rates in countries with which it is compared.

Let’s examine each in turn.

Is Poverty Related to Lackluster Learning?

To this first question, the answer is obviously in the affirmative. That’s not to say “poor children can’t learn.” It is to say, rather, that there’s long been a clear connection between families’ socioeconomic status and students’ academic achievement. As can be seen in Figure 1a, states with higher percentages of students from low-income families report lower average scale scores in 8th-grade math on the National Assessment of Educational Progress (NAEP). The same connection between poverty and academic performance can be observed at the school level (see Figure 1b).

Why do kids from low-income families tend to score so much lower on average than their more-affluent peers? Is it something about poverty itself, that is, a lack of financial resources in the family? This is likely the case, as financial stress can create “toxic” conditions in the home and also make it difficult (if not impossible) for parents to afford the tutoring, educational games, summer camps, afterschool activities, and other educational experiences that middle-class and upper-middle-class students take for granted (and that almost surely boost their achievement).

But it’s not just about money. Poverty is associated with a host of other social ills that have a negative impact on learning. For instance, children in poverty are much more likely to be living in single-parent families headed by young, poorly educated mothers. Poverty is also associated with higher rates of alcoholism and other substance abuse in the home; greater incidence of child abuse and neglect; and heightened family involvement in the criminal justice system. All of these are well-known “risk factors” that are associated with lower test scores as well as with a greater likelihood of dropping out of high school.

So, yes, in general, poverty and factors correlated with low family income are strongly related to low test scores.

Do U.S. Students from Low-Income Families Underperform Their Peers Overseas?

The next question is whether U.S. students from low-income families are lower-scoring than those in other countries. To explore this question, we’re obliged to wrestle with measurement issues. The problem is complicated because no international data set contains both good measures of family income and good measures of student test-score performance.

The best available information is to be found in the data collected by the Program for International Student Assessment (PISA), which is sponsored by the Organization for Economic Co-Operation and Development (OECD). PISA, for its own analyses, uses an index of economic, social, and cultural status (ESCS) that looks at parent occupation and education, family wealth, home educational resources, and family possessions related to “classical” culture. PISA analysts use the index to stratify each country’s student population into quartiles.

Not everyone will agree with the way the ESCS index is constructed, but the data presented in Figure 2 are nonetheless quite instructive. The test scores of students in the bottom quartile of the ESCS index are plotted against those of students in the top quartile. If students in these two quartiles did equally well in each country (as compared to similarly situated students in other countries), then the dotted regression line displayed in green would have a steeper slope, and every dot would fall exactly on that line. As you can see, the actual pattern is not that perfect, as some countries, such as Belgium and France, are relatively better at teaching the higher-status students, while other countries, such as Canada and Finland, do relatively well at instructing students from lower-status families. But notice that the United States falls almost exactly on the regression line. It does equally well (or equally poorly, if you prefer) at teaching its least well-off as those coming from families in the top quartile of the ESCS index.

If we look at a different marker of socioeconomic status, parental education levels, we find a similar pattern. In the U.S., for instance, parents without a high school diploma are much more likely to be in poverty than their better-educated peers, and their children are much more likely than their peers to be low-performing and to drop out of school themselves.

In a study that examined whether some countries are particularly effective at teaching students from disadvantaged backgrounds, Eric A. Hanushek, Paul E. Peterson, and Ludger Woessmann find little difference in the rank order of countries by the performance of students from families where a parent had a college education and the rank order of countries by the performance of students whose parents had no more than a high school diploma. They find that if a country is comparatively effective at teaching the first group, it tends to be no less effective (as compared to others) at teaching the second. The United States performs as expected, proving not to be especially effective at teaching students from the best-educated or the least-educated families. The authors write,

Overall, the U.S. proficiency rate in math places the country at the 27th rank among the 34 OECD countries that participated in the Program for International Student Assessment (PISA). That ranking is somewhat lower for students from advantaged backgrounds (28th) than for those from disadvantaged ones (20th).

There is no evidence that disadvantaged students in the United States are underperforming other countries’ disadvantaged students. If anything, it is the “advantaged” U.S. students (those whose parents have a high level of education) who are falling short in international comparisons.

Is America’s Child-Poverty High Compared 
to Rates Elsewhere?

So far we’ve acknowledged that poverty is, in fact, strongly (and negatively) related to achievement. But we’ve also demonstrated that disadvantaged students in the United States are performing as expected, given the performance of better-situated U.S. students.

But if more students are poor in the U.S. than in other countries, it is still possible that students from low-income families are dragging down U.S. national averages. If that is true, poverty could still be the elephant in the classroom.

But does the U.S. have a greater proportion of low-income students than other countries?

For those educators quoted at the beginning of this essay, the answer is yes. They assert that the U.S. has a sky-high child-poverty rate compared to other developed countries.

To support their claim, they use a measure that assumes all families with less than half the median income in the country are by definition “poor.” Figure 3 shows relative child-poverty rates for selected countries.

In the U.S., median family income is about $52,000 per year, so any family earning less than $26,000 a year is said to be poor. The measure excludes any income from governmental transfers.

Relying on measures of relative poverty is appealing for its simplicity, but it is a highly misleading approach because it’s more a measure of income inequality than of poverty.

To see how relative poverty rates can mislead, let’s look at how they compare to absolute poverty rates for the general population in the American states. In Figure 4, we report the proportion of people living in households that earn less than half of their own state’s median income (basing state median incomes on the 2013 Census Current Population Survey). We also show each state’s absolute poverty rate as it is traditionally defined: the percentage of all people in the 
state living in households below the federal poverty line, which is currently set at $24,250 for a family of four.

For some states, whether one looks at relative poverty or at absolute poverty makes little difference. Arizona, Mississippi, and Louisiana have a lot of poor people however you slice the data.

But notice where wealthier states like Massachusetts and Connecticut appear on the graph. Their absolute poverty rates are among the lowest in the country. But their relative poverty rates are above average—higher than Texas, Tennessee, and Oklahoma. Massachusetts has a higher relative poverty rate than Georgia, Kentucky, and Alabama.

Of course, Massachusetts doesn’t really have more poverty than Alabama—but it does have more income inequality.

The same dynamic plays out when we use relative poverty rates to compare countries. Many of the U.S. households that are counted as poor on a relative measure would be considered middle class on an absolute measure.

Using 2010 data, Timothy Smeeding, founder of the Cross-National Data Center in Luxembourg, reports absolute poverty using a methodology that takes into account all forms of income, including social welfare benefits (see Figure 3). By this measure, the U.S. absolute poverty rate is lower than the United Kingdom’s, virtually the same as Germany’s, and just barely higher than Finland’s. To be sure, the U.S. still has too much poverty. But once social welfare benefits are included, and we look at absolute instead of relative poverty, the U.S. is hardly an outlier.

It’s important to note that the absolute poverty rates shown in Figure 3 are for the general population, not for children. It’s possible that absolute child-poverty rates would look quite different. But we have no way of knowing, because the data to calculate those rates across a large number of countries do not currently exist.
What we can say definitively is that relative poverty rates can be highly misleading. We ran a regression analysis to estimate the relationship between states’ absolute and relative poverty levels and student achievement, and the result was clear: absolute poverty is a powerful predictor of achievement, while the relationship between relative poverty and test scores in the U.S. is weak and not statistically significant (see Figure 5).

Relative poverty is also a weak predictor of student achievement internationally. In another analysis, we compared relative child-poverty rates  to PISA mean math scores in 2009—and once again found only a weak and statistically insignificant relationship.

In short, relative poverty rates, which are only weakly related to student achievement both in the U.S. and abroad, are erroneously used to explain America’s academic struggles. They seem to indicate that the U.S. has an unusually large population of low-income individuals, but in fact they simply demonstrate an unusually high degree of income inequality. Using absolute poverty rates—which are related to student achievement within the U.S.—we see that the proportion of Americans who are poor is quite typical by international standards.

Conclusion

Critics of education reform are certainly correct when they say that poverty is a major factor in lackluster academic performance.

Still, poverty is an issue for virtually every nation on the planet. Where reform critics get it wrong is when they claim that America’s average scores are dragged down by the particularly poor performance of low-income students, or that the advantaged kids are doing just fine.That is objectively untrue. And its scores are not dragged down by an unusually high proportion of poor students, as measures of absolute poverty find the U.S. not to be an outlier at all.

America’s mediocre performance is remarkably consistent. Yes, affluent students outperform poor students. But they don’t outperform their peers overseas.

This doesn’t imply that reform, as currently formulated, is on the right track. Why U.S. student performance is mediocre is a topic worthy of study and debate, as is how to help students at all points on the economic spectrum perform better.

What it does show is that poverty can’t explain away America’s lackluster academic performance. That excuse, however soothing it may be to educators, politicians, and social critics, turns out to be a crutch that’s unfounded in evidence. We need to stop using it and start getting serious about improving the achievement of all the nation’s students.

Michael J. Petrilli is president of the Thomas B. Fordham Institute, where Brandon L. Wright is managing editor and policy associate.

---

Update, November 18, 2015:

The method used to calculate absolute poverty rates in 2010, as reported in Figure 3 of “America’s Mediocre Test Scores,” required estimations from data made available by Timothy Smeeding. The note to Figure 3 does not include the estimating procedures used to calculate absolute poverty rates for 2010, which are available here. Some commentators have questioned those procedures, but similar results are obtained for 1999-2000 by Smeeding when poverty rates are calculated as 125% of the U.S. Department of Agriculture’s poverty line in his 2006 paper, “Poor People in Rich Nations: The United States in Comparative Perspective.” In both cases U.S. poverty rates are only modestly above that in several comparable countries. Differences are unable to explain the much larger differences in student achievement between the United States and other countries. For further discussion, please see “Is America’s Poverty Rate Exceptional? It depends on how you define poverty.”

	Percentages as given in Figure 3, “America’s Mediocre Test Scores,” Petrilli and Wright, 2015	Percentages as given in “Poor People in Rich Nations: The United States in Comparative Perspective,” Timothy Smeeding, 2006 (Using data from 1999-2000 and calculating poverty as 125% of U.S. poverty line)
United Kingdom	18.8	32.8
United States	15.3	19.5
Canada	9.9	17.5
Finland	14.6	17.5
Germany	15.2	17.3
Netherlands	7.4	17.3
Austria	*	15.3
Sweden	*	13.8
Belgium	*	12.2

* Not mentioned in Figure 3, “America’s Mediocre Test Scores,” Petrilli and Wright, 2015

David J. Deming sits down with EdNext’s Marty West to discuss his new study on the effects of a test-based accountability system in Texas on the Education Next Podcast.

---

When Congress passed the No Child Left Behind Act of 2001 (NCLB), standardized testing in public schools became the law of the land. The ambitious legislation identified test-based accountability as the key to improving schools and, by extension, the long-term prospects of American schoolchildren. Thirteen years later, the debate over the federal mandate still simmers. According to the 2015 EdNext poll, about two-thirds of K–12 parents support annual testing requirements, yet a vocal minority want the ability to have their children “opt out” of such tests (see “The 2015 EdNext Poll on School Reform,” features, Winter 2016). Teachers themselves are divided on the issue of high-stakes testing.

NCLB required that states test students in math and reading each year, that average student performance be publicized for every school, and that schools with persistently low test scores face an escalating series of sanctions. We now have ample evidence that these requirements have caused test scores to rise across the country. What we don’t know is: Do these improvements on high-stakes tests represent real learning gains? And do they make students better off in the long run? In fact, we know very little about the impact of test-based accountability on students’ later success. If academic gains do not translate into a better future, why keep testing?

In this study, we present the first evidence of how accountability pressure on schools influences students’ long-term outcomes. We do so by examining how the test-based accountability system introduced in Texas in 1993 affected students’ college enrollment and completion rates and their earnings as adults. Though the Texas system predates NCLB, it was implemented under then governor George W. Bush and it served as a blueprint for the federal legislation he signed as president nearly a decade later. More important, it was implemented long enough ago to allow us to investigate its impact on adult outcomes, since individuals who were in high school in the mid- to late 1990s have now reached adulthood.

Our analysis reveals that pressure on schools to avoid a low performance rating led low-scoring students to score significantly higher on a high-stakes math exam in 10th grade. These students were also more likely to accumulate significantly more math credits and to graduate from high school on time. Later in life, they were more likely to attend and graduate from a four-year college, and they had higher earnings at age 25.

Those positive outcomes are not observed, however, among students in schools facing a different kind of accountability pressure. Higher-performing schools facing pressure to achieve favorable recognition appear to have responded primarily by finding ways to exempt their low-scoring students from counting toward the school’s results. Years later, these students were less likely to have completed college and they earned less.

In short, our results indicate that school accountability in Texas led to long-term gains for students who attended schools that were at risk of falling below a minimum performance standard. Efforts to use high-stakes tests to regulate school quality at a higher level, however, did not benefit students and may have led schools to adopt strategies that caused long-term harm.

The Accountability Movement

A handful of states, such as Texas and North Carolina, began implementing “consequential” school accountability policies in the early 1990s. Under these policies, performance on standardized tests was not only made public but was also tied to rewards and sanctions. The number of states with consequential school-accountability policies rose from 5 in 1994 to 36 in 2000.

Under the accountability system implemented by Texas in 1993, every public school was given one of four ratings: Low-Performing, Acceptable, Recognized, or Exemplary. Schools were rated based on the overall share of students who passed the Texas Assessment of Academic Skills tests in reading, writing, and mathematics; attendance and high-school dropout rates were also considered. Pass rates were calculated separately for four subgroups—white, African American, Hispanic, and economically disadvantaged—if such subgroup made up at least 10 percent of the school’s population. Schools were assigned an overall rating based on the pass rate of the lowest-scoring subgroup-test combination (e.g., math for whites), giving some schools strong incentives to focus on particular students and subjects. (Because the state’s math test was more difficult than its reading test, low math scores were almost always the main obstacle to improving a school’s rating.) School ratings were often published in full-page spreads in local newspapers, and schools that were rated as Low-Performing underwent an evaluation that could lead to serious consequences, including layoffs, reconstitution, and school closure.

The accountability system adopted by Texas bore many similarities to the accountability requirements of NCLB, enacted nine years later. NCLB mandated reading and math testing in grades 3 through 8 and at least once in high school, and it required states to rate schools on the basis of test performance overall and for key subgroups. It also called for sanctions on schools that failed to meet statewide targets for student proficiency rates. Finally, the system required states to report subgroup test results and to increase their proficiency rate targets over time.

Too Good to Be True?

Scores on high-stakes tests rose rapidly in states that were early adopters of school accountability, and Texas was no exception. Pass rates on the state’s 10th-grade exam, which was also a high-stakes exit exam for students, rose from 57 percent to 78 percent between 1994 and 2000, with smaller yet still sizable gains in reading (see Figure 1).

The interpretation of this so-called Texas miracle, however, is complicated by studies of schools’ strategic responses to high-stakes testing. Research on how high-stakes accountability affects test performance has found that scores on high-stakes tests tend to improve with accountability, often dramatically, whereas performance on low-stakes tests with a different format but similar content improves only slightly or not at all. Furthermore, studies in Texas and elsewhere have found that some schools raised their published test scores by retaining low-performing students in 9th grade, by classifying them as eligible for special education (or otherwise exempting them from the exam), and even by encouraging them to drop out.

Clearly, accountability systems that rely on short-term, quantifiable measures to drive improved performance can lead to unintended consequences. Performance incentives may cause schools and teachers to redirect their efforts toward the least costly ways of raising test scores, at the expense of actions that do not boost scores but may be important for students’ long-term welfare.

Our study overcomes the limits of short-term analysis by asking: when schools face accountability pressure, do their efforts to raise test scores generate improvements in higher education attainment, earnings, and other long-term outcomes?

Our Study

An ideal experiment to address this question would randomly assign schools to test-based accountability and then observe changes in both test scores and long-term outcomes, comparing the results to those of a control group of schools. Such an experiment is not possible in this case because of the rapid rollout of high-stakes testing in Texas and (later) nationwide. And unfortunately, data limitations preclude us from looking at prior cohorts of students who were not part of the high-stakes testing regime.

Instead, our research design compares successive grade cohorts within the same school—cohorts that faced different degrees of accountability pressure owing to changes in how the state defined school performance categories over time. Beginning in 1995, each Texas school received its overall rating based on its lowest subgroup-test pass rate. That year, at least 25 percent of all tested students in a high school were required to pass the 10th-grade exit exam in each subject in order for the school to receive an Acceptable rating. This standard rose by 5 percentage points every year, up to 50 percent in 2000. The standard for a Recognized rating also rose, from a 70 percent pass rate in 1995 and 1996 to 75 percent in 1997 and 80 percent from 1998 onward. In contrast, the dropout and attendance-rate standards remained constant over the period we study. We use these changes in performance standards to estimate the “risk” that each school will receive a particular rating, and we compare cohorts who attended a school when it was on the brink of receiving a Low-Performing or Recognized rating to cohorts in the same school in years that it was all but certain to be rated Acceptable—and therefore plausibly “safe” from accountability pressure.

Most research on school accountability has studied how schools respond to receiving a poor rating, but our approach focuses instead on the much larger group of schools that face pressure to avoid a Low-Performing rating in the first place. Because the ratings thresholds rose over time, the set of schools experiencing the most pressure also changed. Consider, for example, students in a school that was plausibly safe from accountability pressure in 1995 but was at risk of a Low-Performing rating in 1996. Students in the 1996 cohort are likely quite similar to students in the class before them, except for the fact that they were subject to greater accountability pressure. (Our analysis does include controls for various ways in which those cohorts may have differed initially, such as by incoming test scores and demographic makeup.) By comparing grade cohorts who faced different degrees of accountability pressure, we can ascertain how much their level of risk affects not only 10th-grade exam scores but also how much schooling they completed and their earnings later in life.

Findings

We find that students, on average, experience better outcomes when they are in a grade cohort that puts its school at risk of receiving a Low-Performing rating. They score higher on the 10th-grade math exam, are more likely to graduate from high school on time, and accumulate more math credits, including in subjects beyond a 10th-grade level.

Later in life, these students are 0.6 percentage points more likely to attend a four-year college and 0.37 percentage points more likely to graduate. They also earn about 1 percent more at age 25 than those who were in cohorts whose schools were not facing as much accountability pressure. The earnings increase is comparable to the impact of having a teacher at the 87th percentile, in terms of her “value added” to student achievement, versus a teacher at the value-added median (see “Great Teaching,” research, Summer 2012).

Since the Texas state test was a test of basic skills, and the accountability metric is based on pass rates, schools had strong incentives to focus on helping lower-scoring students. While schools surely varied in how they identified struggling students, one reliable predictor that students might fail the 10th-grade exam was whether they failed an 8th-grade exam.

In fact, when we take into account 8th-grade failure rates, we find that all of the aforementioned gains are concentrated among students who previously failed an exam. These students are about 4.7 percentage points more likely to pass the 10th-grade math exam, and they score about 0.2 standard deviations higher on the exam overall (see Figure 2). More importantly, they are significantly more likely to attend a four-year college (1.9 percentage points) and earn a bachelor’s degree (1.3 percentage points). These impacts, while small in absolute terms, represent about 19 and 30 percent of the mean for students who previously failed an 8th-grade exam. We also find that they earn about $300 more annually at age 25.

In contrast, we find negative long-term impacts for low-scoring students in grade cohorts attending a school in a year when it faced pressure to achieve a Recognized rating. Students from these cohorts who previously failed an exam are about 1.8 percentage points less likely to attend a four-year college and 0.7 percentage points less likely to earn a bachelor’s degree, and they earn an average of $748 less at age 25. This negative impact on earnings is larger, in absolute terms, than the positive earnings impact in schools at risk of being rated Low-Performing. However, there are fewer low-scoring students in high-scoring schools, so the overall effects on low-scoring students roughly cancel one another other out. Again we find no impact of accountability pressure on higher-achieving students.

What worked well. Higher test scores in high school do not necessarily translate into greater postsecondary attainment and increased earnings in adulthood, yet our study demonstrates that, for many students, accountability pressure does seem to positively influence these long-range outcomes. Additional knowledge of mathematics is one plausible explanation for these favorable impacts on postsecondary attainment and earnings. Accountability pressure could have caused students to learn more math through: 1) additional class time and resources devoted to math instruction and 2) changes in students’ later course-taking patterns, sparked by improved on-time passage of the exit exam.

Indeed, we find an average increase of about 0.06 math course credits per student in schools that face pressure to avoid a Low-Performing rating. We also find that the impacts on both math credits and long-range outcomes grow with cohort size and with the number of students who previously failed an 8th-grade exam, suggesting that students particularly benefited from accountability pressure when it prompted schoolwide reform efforts.

Prior research has demonstrated that additional mathematics coursework in high school is associated with higher earnings later in life, and that even one additional year of math coursework increases annual earnings by between 4 and 8 percentage points. In our study, controlling for the amount of math coursework reduces the effects of accountability pressure on bachelor’s degree receipt and earnings at age 25 to nearly zero, and lowers the impact on four-year college attendance by about 50 percent. This suggests that additional math coursework may be a key mechanism for the long-term impacts of accountability pressure under the Texas policy.

Additionally, we find some evidence that schools respond to the risk of being rated Low-Performing by adding staff, particularly in remedial classrooms. This response is consistent with studies of accountability pressure in Texas and elsewhere that find increases in instructional time and resources devoted to low-scoring students, and provides another possible explanation for the positive effects of accountability pressure for certain students.

Dangers of a poorly designed system. Despite finding evidence of significant improvements in long-range outcomes for some students, those same improvements were not enjoyed by others. Why might an accountability system generate seemingly contradictory results?

As mentioned earlier, high-stakes testing poses the risk that it may cause teachers and schools to adjust their effort toward the least costly (in terms of dollars or effort) way of boosting test scores, possibly at the expense of other constructive actions. Thus, one can try to understand the difference in impacts between the two kinds of accountability by asking: in each situation, what was the least costly method of achieving a higher rating?

In our data, the student populations of schools at risk of a Low-Performing rating were, on average, 23 percent African American and 32 percent Hispanic, and 44 percent of students were poor. The mean cohort size was 212, and the mean pass rate on the 8th-grade math exam was 56 percent. Since the overall cohort and each tested subgroup were on average quite large, these schools could only escape a Low-Performing rating through broad improvement in test performance. In contrast, school populations closer to the high end of the performance spectrum were only about 5 percent African American, 10 percent Hispanic, and 16 percent poor, with a mean cohort size of only 114 and a mean pass rate of 84 percent on the 8th-grade math exam. Thus, many of the schools that were aspiring to a Recognized rating could achieve it by affecting the scores of only a small number of students.

One example of how a small school might “game the system” is by strategically classifying students in order to influence who “counts” toward the school’s rating. Indeed, we find strong evidence that some schools trying to attain a Recognized rating did so by exempting students from the high-stakes test. These schools classified low-performing students as eligible for special education services to keep them from lowering the school’s rating (special education students could take the 10th-grade state test, but their scores did not count toward the rating).

In schools that had a chance to achieve a Recognized rating, low-scoring students who were not designated as eligible for special education in 8th grade were 2.4 percentage points more likely to be newly designated as such in 10th grade, an increase of more than 100 percent relative to the 2 percent designation rate in other schools. The designation of low-scoring students as eligible for special education was more common in schools where a small number of students had failed the 8th-grade exam, making it easier for educators to target specific students. We also find a small but still noteworthy decrease of 0.5 percentage points in special education classification for high-scoring students in these schools.

As a result of this strategic classification, marginal students in certain schools were placed in less-demanding courses and acquired fewer skills, accounting for the negative impact of accountability pressure on long-term outcomes for those students. In essence, those students did not receive the attention they needed in order to improve their learning.

Summing Up

Why do some students benefit from accountability pressure while others suffer? Our results suggest that Texas schools responded to accountability pressure by choosing the path of least resistance, which produced divergent outcomes. The typical school at risk of receiving a Low-Performing rating was large and had a majority nonwhite population, with many students who had previously failed an 8th-grade exam. These schools had limited opportunity to strategically classify students as eligible for special education services. Instead, they had to focus their efforts on truly helping a large number of students improve. As a result, students in these schools were more likely to pass the 10th-grade math exam on time, acquire more math credits in high school, and graduate from high school on time. In the long run, they had higher rates of postsecondary attainment and earnings. These gains were concentrated among students at the greatest risk of failure.

In other schools, the accountability system produced strong incentives to exempt students from exams and other requirements. In these schools, accountability pressure more than doubled the chances that a low-scoring student would be newly deemed eligible for special education. This designation exempted students from the normal high-school graduation requirements, which then led them to accumulate fewer math credits. In the long run, low-scoring students in these schools had significantly lower postsecondary attainment and earnings.

In some respects, though not all, the accountability policy in Texas served as the template for No Child Left Behind, and thus our findings may have applicability to the accountability regimes that were rolled out later in other states. In Texas, and under NCLB nationwide, holding schools accountable for the performance of every student subgroup has proven to be a mixed blessing. On the one hand, this approach shines light on inequality within schools in an attempt to ensure that “no child is left behind.” On the other hand, when schools can achieve substantial “improvements” by focusing on a relatively small group of students, they face a strong incentive to game the system. In Texas, this situation led some schools to strategically classify students as eligible for special education, which may have done them long-run harm.

What policy lessons can we draw from this study as Congress works out a new iteration of theElementary and Secondary Education Act to replace NCLB? First, policy complexity can carry a heavy cost. As many other studies have shown, high-stakes testing creates strong incentives to game the system, and the potential for strategic responses grows as the rules become more complicated. The second lesson is that, at least in Texas, school accountability measures only worked for schools that were at risk of receiving a failing grade. Therefore, the federal government might consider approaching school accountability the way the Food and Drug Administration regulates consumer products. Instead of rating and ranking schools, the feds could develop a system that ensures a minimum standard of quality.

David J. Deming is associate professor of education and economics at the Harvard Graduate School of Education. Sarah Cohodes is assistant professor of education and public policy at Teachers College, Columbia University. Jennifer Jennings is assistant professor of sociology at New York University. Christopher Jencks is the Malcolm Wiener Professor of Social Policy at the Harvard Kennedy School.

President Obama’s proposal for tuition-free community college, issued earlier this year, seems to have laid down a marker for the Democratic Party. Vermont senator Bernie Sanders is touting his plan for free four-year public college on the primary trail; Massachusetts senator Elizabeth Warren called for “debt-free college” in a high-profile speech; and former senator and U.S. secretary of state Hillary Clinton has proposed her own plans for tuition-free community college and “no-loan” tuition at four-year public colleges.

In this forum, Sara Goldrick-Rab, professor of educational policy studies and sociology at the University of Wisconsin–Madison and co-author of a paper that helped shape the president’s plan, calls for an even more expansive effort—one that includes funding for students’ living and other expenses while they pursue an associate degree at any public institution. Andrew Kelly, director of the Center on Higher Education Reform at the American Enterprise Institute, argues that the Obama plan will not address low rates of college readiness and student success but will strain public budgets and crowd out innovation.

• “The Economy Needs More Workers with Associate Degrees,” by Sara Goldrick-Rab

• “Tuition Is Not the Main Obstacle to Student Success,” by Andrew P. Kelly

At a June 2015 education forum hosted by the National Journal, U.S. senator Lamar Alexander threw cold water on recent proposals to make the first two years of college free—because he thinks it already is. The senator explained, “Two years of college for a low-income student is already free, or nearly free.”

It would be wonderful if that were true. But while the senator’s statement is a reasonable description of how college used to be (at least for some), today’s reality is quite different. Attending college is far too expensive for many Americans. The 2012 National Postsecondary Student Aid Study, conducted by the National Center for Education Statistics, found that after taking all grants and scholarships into account, attending one year of community college runs dependent students from low-income families more than $8,000 in tuition, fees, and living costs (see the green “Net price of attendance” bars in Figure 2). In other words, while it is true that more students are qualifying for and receiving a federal Pell grant, the price after the discount from the Pell is climbing higher and higher (see Figure 1). This is largely driven by living costs, which must be covered if students are to focus their time and energy on school rather than work. Federal loan limits are too low to fully cover these costs, but they are the true costs of degree completion. Tuition and fees are the price of access—living costs are the price of success.

The whole concept of higher education is under debate in America today—public versus private versus for-profit, preprofessional versus liberal arts, in-person versus online. One thing, however, should be clear to all: when large numbers of people can’t afford college at all, the system is broken.

The American people know this. A 2014 Gallup-Lumina Foundation study found that, of those surveyed, a whopping 96 percent said that postsecondary education was important, and 79 percent believed that college wasn’t affordable for everyone in this country who needed it. This is not a partisan issue. Americans want and need college to be much more affordable.
Responding in part to the needs of employers, communities around the country are calling for plans to make at least the first few years of college very inexpensive or even free. President Obama’s proposal, America’s College Promise, would eliminate tuition at community colleges for eligible students, reducing the average annual cost by about $3,800.

Senator Alexander’s reluctance to embrace Obama’s plan is ironic, since Alexander’s home state has been a leader on the issue. In 2014, Tennessee governor Bill Haslam, a Republican, created the Tennessee Promise, which uses lottery funds to cover tuition and fees not covered by the Pell grant or other state assistance to make two years of community college more affordable for all Tennessean high-school graduates. Obama’s plan builds on Haslam’s.

An Associate Degree 
Makes a Difference

Decades ago, a high school education was enough for most folks to earn a middle-class living, build a family, and live out the American dream. Strong U.S. manufacturing and three decades of high economic growth (from the 1940s to the 1970s) sustained millions of relatively high-paying jobs for high school grads.

But the American economy has changed. Completing 12 years of school doesn’t cut it in the current job market. U.S. employers complain of a skills gap. They need many more employees with technical skills than they can find. Workplaces require more technological ability, better social skills, and a broader grounding in multiple disciplines. This is exactly what a postsecondary education offers.

President Obama’s focus on the first two years of college makes sense. Two-year associate degrees pay off. The unemployment rate for people with associate degrees is 25 percent lower than for those with just a high school diploma. Forty percent of people who earn associate degrees go on to earn higher degrees. If the associate degree is completed by the time a person is 20 years old, then 60 percent of the time it is a milestone toward a higher degree. And people with even a couple years of college make significantly more money, on average, than those with none.

What Obama Left Out

The proposals from President Obama and Governor Haslam are a good start, but we shouldn’t restrict the choices of students too much. Students from low-income families, like their more-affluent peers, should be able to attend the public institutions that best fit their academic talents and personal and professional goals. Four-year state universities should not be reserved for the financially privileged.

Obama’s plan has another missing piece. It is critical that we ensure that students not only begin degrees but also complete them. Some may get an associate degree in two years, while others may work more outside of school and take longer to finish. Focusing on completing the associate degree and not on a two-year time period can give students the flexibility to achieve their goals in ways that work best for their individual situations.

To make this work and to fix a long-standing problem in higher education, public four-year colleges and universities must acknowledge that the first degree is an associate degree, not a bachelor’s degree. Two years of credits at a community college result in an associate degree, but students at four-year institutions who lack the money or support systems to persist beyond two years leave with no degree at all. The associate credential should be awarded at both two-year and four-year colleges, even if the institutions differ in the specific programs of study they offer. Doing so will increase competition while enhancing outcomes for students.

While state-by-state and campus-by-campus proposals are already beginning to help students, a national approach is needed to ensure every American has a viable path to achieving a postsecondary degree and a better future. The federal government is still the biggest provider of financial aid. If the federal government doesn’t work with states and communities to achieve their goals, then it is effectively working against them.

A federal-state partnership to make an associate degree free can also bring much-needed accountability to America’s public colleges and universities. Inadequate and inequitable spending hinders college completion rates. A new partnership could help states ensure that spending is sufficient to cover the resource costs of a high-quality sub-baccalaureate education; that spending on instruction takes precedence over amenities; that enrollment capacity expands rather than contracts; and that exclusionary admission practices are reduced. Campuses should be discouraged from chasing luxury dorms and glittery amenities for out-of-state recruitment and instead focus on quality, affordable, near-campus housing. Accountability for these changes should be a prerequisite for the new federal support. These would be part of the terms of a new system, a rejiggered Financial Aid 2.0.

Claiming limited resources, many state and federal proposals have so far focused on tuition. Other costs, like fees, books, and living expenses, go uncovered. This is unfortunate and myopic, especially since these other costs constitute more than half—and often as much as three-fourths—of the total cost of attending college. Covering tuition facilitates access to college, but the other costs need to be addressed to ensure completion. Thus, to be truly effective, “free” (to the student) must truly mean that all costs are considered.

Reduce Waste and Profiteering

With or without the changes I suggest above, many might wonder if there is any realistic path to funding President Obama’s proposal in this era of severe fiscal restraint. This is a critical issue.

We can start with better use of current dollars. Today, federal taxpayer dollars spent on financial aid are spread broadly across all years of education and all types of schools. Many of these schools are private and for-profit institutions that the government cannot hold accountable. We should confine federal spending to public, high-quality, affordable college opportunities. This is not meant as a slight to private or for-profit schools—many of which are wonderful places to attend college. But like many businesses, the private sector of higher education can flourish without government subsidies and may perform even better without government intervention.

Tax credits to offset tuition bills are another way national policy chases its own tail. Cutting the credits could free up money to fund more effective college access strategies, like Obama’s proposal.

Making an associate degree entirely free can be the centerpiece of reform that would dramatically reduce waste in higher education by dealing with the complex and often inadequate web of financial aid that often results in students leaving institutions without degrees. Clearly, with students paying more than $8,000 a year for community college after all financial aid, fixing financial aid is not nearly enough. We won’t succeed in tinkering our way toward accountability by trying to attach strings to existing programs in an effort to promote risk sharing. But the kind of transformative effort I’m calling for would initiate an entirely new approach to boost both students’ and universities’ bottom lines and yield dividends for economies in communities nationwide.

Worth the Money
In my work as an education researcher, I’ve listened to the stories of hundreds of students from all sorts of backgrounds and circumstances. They all understand how important it is to go to college. But too often their hopes are dashed by a system that delivers few opportunities for low- and middle-income students. We can do better.

The benefits of offering the first degree for free are evident for our economy, our country, and our children. Business leaders will get access to the larger pool of skilled, certified workers they are clamoring for. Connecting American workers with good jobs that require an associate degree or higher will help us start to rebuild our waning middle class.

Now is a good time to bring this plan to the forefront of higher education policy. Congress is currently reauthorizing the federal Higher Education Act. At the same time, statehouses are considering a patchwork of reforms—some good, some bad. The iron is hot. Rather than stringing along an antiquated and broken system, it’s time to get to work in order to make the first degree free, stimulate the economy, and provide real security for the middle class.

I was heartened to see President Obama introduce America’s College Promise. While parts of the president’s plan need improvement, the proposal is a step in the right direction.

This is part of a forum on community college. For an alternate take, please see “Tuition Is Not the Main Obstacle to Student Success” by Andrew P. Kelly.

The Obama plan for tuition-free community college, America’s College Promise, will go nowhere in the current Congress. But it’s still important to understand what the idea would actually mean for American higher education. For while the media fixated on the “free tuition” part, and skeptics immediately bemoaned the likely cost, the reform is much more radical. Specifically, it would move American higher education from a voucher-funded market to a system with a free public option much like traditional K‒12 public schools.

The idea is rooted in a reasonable conclusion: the current path of federal student aid policy is unsustainable. While grants and loans have boosted college access, flooding the market with easy money that has few strings attached leaves federal policymakers at the mercy of colleges and state policymakers who fund public institutions. Increases in federal aid provide little incentive for these other players to worry about rising tuition, and new evidence suggests that the availability of student loans encourages schools to charge more. When tuition goes up, policymakers increase federal aid to bring out-of-pocket costs down again, only to see those resources gobbled up by future tuition increases.

Advocates of free public college look at this status quo as proof that the market-based model has failed and cannot possibly work. As Sara Goldrick-Rab and Nancy Kendall argue in “Redefining College Affordability,” placing the power in the hands of consumers, with all the constraints that hamper their ability to make a rational investment, fails to hold colleges accountable. The voucher-based market encourages colleges to “increase tuition and fees without regard to how these affect diverse students and families, as long as ‘the market’ supports [it]” and to “pay greater attention to the demands and desires of students and families that can pay more for their services.”

The answer, advocates argue, is to revisit the decision to fund students with a voucher in the first place. Instead, we should move away from the market and toward a model where the feds fund public colleges directly. In return, those public institutions pledge not to charge any tuition or fees. And because the feds directly control the purse strings, the argument goes, they’ll have greater leverage to set quality standards, hold colleges accountable, and require them to implement chosen reforms. Indeed, President Obama’s proposal reads like a federal reform plan for community colleges, complete with a laundry list of new requirements for states and schools.

The Wrong Prescription

Obama’s plan is provocative, but it’s deeply flawed. First, thanks to federal aid, most low- and middle-income students already pay no net tuition to attend community college, yet student outcomes at two-year colleges are poor. Second, free community college could lead students to “undermatch,” which would lower their odds of completing a degree. Third, Washington cannot regulate community colleges to success. Fourth, the combination of tuition caps and direct public funding could actually lead to rationing. Finally, a free public option would stifle innovation and competition.

Let’s start with the most basic objection: community colleges are already tuition-free for most low-income students (see Figure 2). According to the College Board, students with annual family incomes under $65,000 received enough grant aid to cover the entire cost of tuition at community colleges in 2011‒12 (the most recent federal data available). In fact, dependent students in the lowest income quartile received enough grant and scholarship aid to cover the cost of tuition and leave $3,100 to cover other expenses (the average independent student received enough grant aid to cover tuition with $1,800 left over). Middle- and upper-income students pay a modest amount in net tuition, but that is because government aid is progressive. Free community college would provide a windfall to these families who would pay to send their kid to college in the absence of a free public option.

Though many students already attend community college tuition-free, student success rates are discouragingly low at these institutions. The National Student Clearinghouse estimates that less than 40 percent of students who start at a public two-year college complete a degree or certificate within six years. According to federal data, just one-third of students from the lowest income quartile who started at a public two-year college in 2003‒04 finished a credential in six years. Among independent low-income students at two-year colleges, the completion rate was 22 percent. Even the most affluent dependent students struggled, graduating at a rate of just 42 percent.

It’s not clear why President Obama’s plan to waive tuition at community colleges would automatically improve these outcomes. Now, as Goldrick-Rab points out, students have to pay expenses beyond tuition—books and supplies, room and board, and transportation—and she and Kendall propose a state-funded stipend equal to 15 hours a week at the region’s “living wage.” Obama’s plan would not be as generous, although low-income students would no longer have to spend Pell grant funds on tuition and could use them to defray other expenses (it is a “first-dollar” program).

Living expenses are part of the price of attendance for many community college students—especially adults who are living on their own—and all students have to pay for books. To the extent these expenses exceed grant and loan limits, that’s an argument for experimenting with beefier vouchers for needy students and increased loan limits for others, not creating an entitlement to free tuition for all.

Furthermore, simply throwing money for living expenses at students is unlikely to remove other clear obstacles to success and may well exacerbate them. For instance, how would free college improve student readiness? Federal data show that 68 percent of public two-year college students have to take at least one remedial course; the average student who starts at a two-year college takes 2.9 remedial courses. Very few of these students complete a degree or certificate. Free college tuition won’t fix American high schools, and conditioning cash for living expenses on college attendance would likely draw in even more students who are unprepared for college-level work.

That brings us to the second problem: free community college could actually lower rates of student success. Ironically, the president’s proposal would likely run counter to one of the administration’s other priorities—decreasing the rate at which low-income students “undermatch” at less-selective colleges. The growing literature on undermatch suggests that enrolling in a college that is less selective than they are academically qualified to attend reduces students’ chances of graduating. For instance, a study by Bridget Terry Long and Michal Kurlaender found that Ohio students who started at community colleges were 14.5 percent less likely to finish bachelor’s degrees within nine years than similar students who started at four-year colleges.

Of particular importance here is how policy might induce students to enroll in lower-quality schools. In a study of the Adams Scholarship in Massachusetts, which provides high-achieving students with a merit-based scholarship to attend an in-state public college, Josh Goodman and Sarah Cohodes found the scholarship led recipients to choose less-selective colleges than they would have, and that they graduated at lower rates than peers who attended better schools. In another study, Goodman and colleagues found that lower-achieving students who scored just above the SAT score necessary for admission to a four-year college in Georgia were substantially more likely to complete a bachelor’s degree than those just below the cutoff, most of whom wound up enrolling in a community college.

Of course, it could be the case that free community college would benefit those who would not have attended at all, counterbalancing the negative effects of undermatch. Economist Jeff Denning has shown that a $1,000 drop in the price of Texas community colleges increased enrollment rates among students who would not have enrolled otherwise, but the effects on degree completion were far from definitive. It’s plausible that these students benefit from accessing some college. The question is, at what cost?

Third, proponents recognize that the poor performance of community colleges is a significant weakness in free college plans, but they have outsize expectations about the ability of federal rules to turn these troubled institutions around. Advocates like Goldrick-Rab argue that direct funding will give federal policymakers more power to control the quality of public colleges through rules and accountability requirements. The White House “fact sheet” on America’s College Promise lists what states and colleges would have to do: participating colleges would have to “adopt promising and evidence-based institutional reforms to improve student outcomes,” while states would have to coordinate high schools, community colleges, and four-year schools to reduce remediation rates and, to create incentives to improve, “allocate a significant portion of funding based on performance, not enrollment alone.”

These best-laid plans sound good, but they dramatically overestimate the federal government’s ability to fix colleges. As Rick Hess reminds us, while federal rules can tell people to do something, they cannot force them to do it well.

You’d expect the recent experience with K‒12 School Improvement Grants (SIG) to raise red flags about the feds’ ability to fix troubled schools with more money and mandates. The SIG program has provided more than $5 billion in federal grants to 2,000 of the country’s lowest-performing schools while requiring them to choose one of four “turnaround” models. The results have not been encouraging. While about two-thirds of SIG schools did register modest gains in reading and math, scores at one-third actually declined over the period. The University of Washington’s Robin Lake summed up the evidence: “Given the amount of money that was put in here, the return on investment looks negligible at this point.”

Community college reform is equally challenging. The privately funded Achieving the Dream initiative, which started in 2004 as an effort to apply evidence-based reforms to more than 200 community colleges in 35 states, has disappointed. A 2011 study of the initial cohort of 26 community colleges found that, with a few exceptions, student outcomes remained largely unchanged. It will not be any easier to dictate community college improvement from Washington.

Fourth, it’s not even a slam-dunk that free community college would increase access. Capping tuition at free is effectively a price control that limits college spending to whatever the public is willing to invest. But it would not change the cost structure of colleges themselves, and additional funding may even relax incentives to become more efficient. Proponents argue that existing federal aid resources could cover current costs at public institutions. What happens once the cost of delivery and enrollments increase?

Look at it this way: some economists argue that traditional college will inevitably become more expensive because of the “cost disease.” Like any other service industry that relies on highly educated labor to deliver a product that is heavy on interaction, it is difficult to improve the productivity of traditional college teaching. But to keep pace with wages in the rest of the economy, colleges must pay professors more, which raises costs year over year. Add in the ever-expanding corps of administrators and support staff, and you’ve got a recipe for rising costs. Meanwhile, federal projections predict that college enrollment will grow over time, putting further strain on public budgets. If the public’s generosity doesn’t keep pace with these increases, schools that are prohibited from charging tuition will have to turn students away. Insisting on free tuition could lead to rationing, not open access.

California’s experience during the recession is instructive here: caught between state-mandated caps on tuition and state funding that failed to keep up with an enrollment boom, the community colleges turned away some 600,000 students and reduced course offerings by 21 percent. In response to reduced per-pupil funding from the state, the California State University system reduced enrollment targets for 2015‒16, and trustees have discussed the idea of no longer accepting freshmen at these campuses. Government-imposed tuition caps can wind up keeping students out instead of letting more in.

Finally, a free public option would crowd out innovation and limit competition. There is plenty of innovation going on outside of the public sector, from competency-based programs to career boot camps to new forms of credentialing. Rather than asking how reforms can encourage an array of options (public, private, or for-profit) to emerge that fit the needs of today’s students, the free-public-college crowd wants to simply cram more people through the same old expensive, mediocre model of education.

Education reformers have seen this movie in K‒12, and they know how it ends. Those who have fought for school choice should recognize free community college for what it really is, and instead push for reforms that expand options and choice—not limit them.

This is part of a forum on community college. For an alternate take, please see “The Economy Needs More Workers with Associate Degrees” by Sara Goldrick-Rab.

The digital revolution occurring in schools has focused predominantly on online education in its various forms—including fully online courses, learning management systems, games, and mobile applications—to personalize learning and boost the performance of all students.

To optimize the learning experience for each student, new school models may benefit from leveraging other types of digital tools—from wearable devices that track student metrics to video cameras that capture and digitize key learning moments to screens that read the expressions on students’ faces to help determine how emotionally engaged they are in their learning.

Research at North Carolina State University, for example, shows that software that tracks facial expressions “can accurately assess the emotions of students engaged in interactive online learning and predict the effectiveness of online tutoring sessions.” Assessing engagement is important because research in neuroscience is finding that emotional responses play an important role in learning.

Another study, in the peer-reviewed journal IEEE Transactions on Affective Computing, found that software was able to make judgments about students’ levels of engagement that were as reliable as those of human observers, and that these video-based engagement scores predicted post-test scores better than pre-test scores could.

These studies appeared in 2013 and 2014, respectively, and companies like Pearson have been experimenting with technologies like these for several years. Yet these technologies have had little impact in actual K–12 schools. Using all of them productively still seems far off in the future, as in combination they can create more work for teachers without providing what teachers and students actually need.

If these technologies are to enable K–12 schools to boost each student’s learning, how might that unfold?

The theory of interdependence and modularity that Harvard Business School professor Clayton Christensen developed many years ago sheds some light.

In the early days of most new products and services, leading providers tend to offer products with proprietary, interdependent architectures. The reason? The technology is still immature, and the ways the parts within the new system interact are not yet well understood and are therefore unpredictably interdependent. The organization must therefore integrate to control every critical component of the system and develop them in concert to make any part of the system function at a high enough level to satisfy users. In other words, in order to do anything, the organization must do nearly everything.

Gustavus Franklin Swift’s approach in the 19th century to butchering, marketing, and selling beef illustrates the point. At that time, because there was no technology for transporting meat long distances, the beef industry lacked significant economies of scale and beef was sold on an exclusively local basis. So Swift integrated. He centralized butchering in Kansas City, which meant he could process beef at a very low cost. Then Swift designed the world’s first ice-cooled railcars. He even made ice cabinets and sold them to retail shops throughout the Midwest and Northeast so that once the beef arrived, it would stay fresh. One key to Swift’s ability to market beef in far-flung regions was the assurance he could give customers that the beef was still safe to consume after it had traveled from the stockyards of Chicago to the market. Because a clear understanding of refrigeration and meatpacking processes did not exist at the time, Swift had to control the entire process to ensure that the temperature and storage practices were consistent. In other words, to revolutionize the beef industry Swift had to expand beyond his so-called core competencies and introduce new, interdependent lines of business.

But as an industry matures and products and services improve, there is a shift. The unpredictable interdependencies within a service become better understood and predictable, and suppliers of less integrated, more modular products can become industry leaders. This shift happens as a service’s raw performance becomes good enough to get the job done, so customers start to prioritize the flexibility that modularity offers over the increased performance that integration makes possible.

Because modular parts fit and work together in well-understood, crisply codified ways and can be developed in independent work groups or by different organizations working at arm’s length, standards arise that dictate how different components must interact. For example, a light bulb and lamp have a modular interface. Engineers have lots of freedom to improve the design inside the light bulb as long as they build the stem so that it can fit the established socket specifications.

Importantly, pure interdependence and modularity are the two ends of a spectrum. Most architectures fall somewhere between. There isn’t a “right” place to be. Instead, organizations are more likely to succeed when they match the type of architecture to their particular circumstances.

Although some of the tools have been around for several years, it’s still relatively early in the pursuit of personalized learning, and the various technologies are still underperforming. As a result, pursuing an integrated and proprietary approach to developing the technologies and controlling how they interact with the school’s teachers, physical architecture, and philosophy may be critical. In other words, we may need to see more schools take what Andreessen Horowitz, a leading venture-capital firm, calls the “full-stack start-up” approach—the idea that a start-up builds a “complete, end-to-end product or service,” and controls even the nontechnology components of a solution if those will perform better when integrated with the technology.

Although most district schools aren’t equipped to take this approach—and having them try wouldn’t be advisable—there are schooling networks emerging to tackle this work.

AltSchool, a private micro-school network (see “The Rise of Micro-schools,” what next, Summer 2015), has attracted significant media attention because of the whopping $130 million in capital it has raised. The network is using a significant portion of this money to hire engineers to develop a full set of digital tools, including an online learning platform that supports its personalized learning playlists for each student; video cameras placed in every classroom, which allow teachers to record, document, and learn what works in different moments; and software that supports the administration and operations for its network of schools. If any school network would seem to be well positioned for experimenting with wearables and facial recognition, it would be AltSchool.

Summit Public Schools, a charter management organization with schools in California and the state of Washington, is taking a similar approach. It has brought in engineers to create a personalized learning playlist platform for its students and teachers.

Both networks ultimately want other schools to use the technologies that they are developing so they can have a wider impact. But it is an open question whether technologies that have been developed for specific schooling models with distinct philosophies about learning can be modularized for use by a school that doesn’t have a similar instructional model, philosophy, and internal capacity. AltSchool’s student-to-teacher ratio ranges from 8- to 12-to-1, and its founder, Max Ventilla says it aims to be a “Montessori 2.0 school.” Summit is using a complex competency-based model of blended learning that gives students significant ownership over their learning.

We can predict that before the wider world of schooling can benefit from these technologies, the performance of AltSchool and Summit will have to become reliable, and the interactions between the technologies and the different aspects of the school must be well understood. AltSchool and Summit might then begin to unbundle their offerings and develop clear standards that detail how the component parts must interact with each other.

Technology developed for a school with, for example, a 12-to-1 student-to-teacher ratio still may not export easily to a more traditional public schooling context. That may not be the only market for the technology, however. AltSchool could fuel the growth of independent schools in rural areas where it doesn’t plan to compete directly. Such schools might be better candidates for adopting AltSchool’s offerings.

Whether the platforms being developed at AltSchool and Summit succeed or whether wearables, video cameras that read expressions, and the like have an impact on education remains to be seen. But with schools now designing and building technology, the odds are better that we’ll see some technology breakthroughs that will help educators everywhere to rethink school.

Michael B. Horn is co-founder of and executive director of the education program at the Clayton Christensen Institute and executive editor at Education Next.

Presidents, Congress, and 
the Public Schools: The Politics of Education Reform
By Jack Jennings
Harvard Education Press, 2015, 
$35; 264 pages.

As reviewed by Jay P. Greene

Education reformers tend to have little interest in history. They are so convinced that the old system is broken and so focused on fixing it for the future that they often fail to consider what lessons might be learned from past efforts. Jack Jennings’s new book, Presidents, Congress, and the Public Schools, is a useful antidote to the ahistorical approach.

Jennings’s role as a staffer for the U.S. House Committee on Education and the Workforce placed him at the center of nearly a half century of federal education policymaking. Want to know about how the Elementary and Secondary Education Act (ESEA) began? Jennings was present at its creation and can speak about it authoritatively. Want to know about the Individuals with Disabilities Education Act or the Bilingual Education Act? Jennings was involved in their enactments as well. Jennings describes the motivations of those who authored these federal reforms, the political hurdles they faced, and the ultimate success or failure of those initiatives.

To his credit, Jennings does not act as a cheerleader for past reforms, including those in which he played an important role. For example, in describing the results of Title I, Jennings concludes, “In a nutshell, the billions of dollars spent on Title I had at best a modest effect on the academic achievement of the disadvantaged students who participated in the program…” On No Child Left Behind (NCLB), he writes, “So it truly was a mixed bag. The spotlight was directed on groups of students whose low performance could have been concealed in the past, and districts were held accountable for every school. The weakness, though, was that tests do not a good education make.”

In general, Jennings is less informative in assessing the effectiveness of federal reform efforts than in describing their origins and political struggles. His assessments are based more on a keen sense of what is politically sensible than on rigorous research. Of course, Jennings is not a researcher, and no one should read this book hoping to learn about the latest and best research findings. The appeal of the book is its firsthand history of major federal education reforms and its conventional wisdom account of their effectiveness.

The book’s weak understanding of research is most clearly seen in his analysis of the effectiveness of NCLB. Jennings examines gains on the National Assessment of Educational Progress (NAEP) as his primary method for determining whether NCLB has been
beneficial. He counts how often NAEP gains were greater in the decade before the act’s adoption than in the decade after for different grades, subjects, and subgroups. He describes NAEP as “the ‘gold standard’ of assessment,” seemingly unaware that the quality of the assessment does not compensate for the weakness of his simple pre‒post comparison research design in trying to determine the effectiveness of a program. Jennings also appears unaware that there are rigorous studies on the effects of NCLB and other high-stakes accountability systems, such as those by Thomas Dee and Brian Jacob (see “Evaluating NCLB,” research, Summer 2010) and those published by Stanford University researchers Eric Hanushek and Margaret Raymond in 2005, 
and Martin Carnoy and Susanna Loeb in 2002.

Jennings nonetheless captures what many elites in Washington, D.C., currently think about past reforms. That may be more important than knowing what rigorous research has to say for understanding future politics.

Unfortunately, Jennings’s prescriptions for the future are not very compelling. While acknowledging the limited effectiveness of past federal programs, he never seriously considers that federal solutions are simply unworkable. He has a somewhat charming but naive optimism that if we just change the design and increase the funding, things will be different next time. On this matter, Jennings may have lost touch with the thinking of one cadre of D.C. elites, whose disillusionment with federally based education reform has become palpable.

Oblivious to the growing opposition to this approach in Congress, Jennings uses the final section of the book to propose a new federal program, the United for Students Act, which is essentially Race to the Top on steroids. It would be bigger and better funded, but it would similarly offer extra money to states if they pursued certain types of policies, including preschool expansion, teacher quality reforms, extra funding for schools with extra challenges, and curriculum changes. Jennings thinks he is being respectful of federalism when he concedes that “a state should be able to choose to apply for the United for Students grant or not,” but he doesn’t seem to grasp that this is the equivalent of saying that states could choose to pay taxes for large programs that other states would get and they would not.

Jennings titles the section containing this new proposal “Fresh Thinking about the Federal Role in Education,” but there is little that is “fresh” in his thinking. Other than proposing that the new effort be better funded and focused on what he deems to be the critical issues, it is unclear how this new proposal should be expected to produce something dramatically different from the disappointing results of past efforts. Didn’t past efforts also represent significant increases in funding for their time? Didn’t the designers of past efforts also believe they were focused on the critical issues? Why will federal policymakers get it right this time if they haven’t managed to do so previously?

Perhaps Jennings’s era as the architect of major federal policy has passed. Jennings’s book is an interesting and informative window into the past, but we shouldn’t look to him for cutting-edge research or compelling proposals for future federal efforts. We can, however, hope that new generations of education reformers make use of Jennings’s accounts of past federal efforts in designing future initiatives that might be more effective.

Jay P. Greene is professor of education reform at the University of Arkansas.