Tag Archives: NBER

US teacher strikes were good, actually

Originally published in Vox on August 25, 2024.
—–

Few things have bedeviled education policy researchers in the US more than public school teacher strikes, driven by educators on the vanguard of resurging labor activism. While union membership nationwide continues to decline, nearly one in five union members in the US is a public school teacher — and their high-profile, disruptive strikes generate significant media attention and public debate.

But do these strikes work? Do they deliver gains for workers? Do they help or hurt students academically?

Answering these questions has been challenging, largely due to a lack of centralized data that scholars could use to analyze the strikes. The Bureau of Labor Statistics used to keep track of all strikes and work stoppages across the country, but since its budget was cut in the early 1980s, the agency has only tracked strikes involving more than 1,000 employees. Given that 97 percent of US school districts employ fewer than 1,000 teachers, the majority of teacher strikes are not federally documented.

Now, for the first time ever, researchers Melissa Arnold Lyon of the University at Albany, Matthew Kraft of Brown University, and Matthew Steinberg of the education group Accelerate have compiled a novel data set to answer these questions, providing the first credible estimates of the effect of US teacher strikes.

Their data set — which covers 772 teacher strikes across 610 school districts in 27 states between 2007-2023 — took four years to compile. The three co-authors, plus seven additional research assistants, reviewed over 90,000 news articles to plug the gaps in national data. Their NBER working paper, published on Monday, provides revealing information about the causes and consequences of teacher strikes in America, and suggests they remain a potent tool for educators to improve their working conditions.

Teacher strikes lead to significant wage increases on average, regardless of length

By and large, teacher strikes in the US are not common, nor are they lengthy work stoppages. The median number of strikes per year over the 16-year study was 12.5, with the typical strike lasting just one day. Sixty-five percent of strikes ended in five days or less. Their longest identified strike was 34 days in Strongsville, Ohio in 2013.

Almost 90 percent of the teacher strikes identified involved educators calling for higher salaries or increased benefits, and the researchers found that, on average, strikes were successful in delivering those gains. Specifically, the strikes caused average compensation to increase by 3 percent (or $2,000 per teacher) one year after the strike, reaching 8 percent, or $10,000 per teacher, five years out from the strike.

More than half of strikes also called for improved working conditions, such as lower class sizes or increased spending on school facilities and non-instructional staff like nurses. The researchers found that strikes were also effective in this regard, as pupil-teacher ratios fell by 3.2 percent and there was a 7 percent increase in spending dedicated to paying non-instructional staff by the third year after a strike.

Importantly, the new spending on compensation and working conditions did not come from shuffling existing funds, but from increasing overall education spending, primarily from the state level.

That these strikes were effective is notable, particularly since labor strikes overall have not been associated with increases in wages, hours, or benefits since the 1980s. The study authors suggest strikes among public school teachers may be a more “high-leverage negotiating tactic” than other unionized fields because teachers can be less easily replaced by non-unionized workers or tech automation.

Perhaps surprisingly, the researchers find no relationship between whether a strike is short or long in terms of the effect it has on teacher salary.

Lyon of the University at Albany thinks that part of why teachers may be so successful in achieving such significant increases is because teacher strikes can send public signals in ways other labor strikes often can’t.

“Because education is such a salient industry, even a one-day strike can have a big impact,” she told me. “News media will pick it up, people will pay attention, and parents are going to be inconvenienced. You have these built-in mechanisms for attracting attention that other types of protest do not.” Another study she co-authored with Kraft earlier this year found that teacher strikes more than double the probability of US congressional political ads mentioning education, underscoring their power in signaling the need for educational change.

Students were not academically harmed by the strikes

Previous research on teacher strikes in Argentina, Canada, and Belgium, where work stoppages lasted much longer, found large negative effects on student achievement from teacher strikes. (In the Argentina study, the average student lost 88 school days.)

In contrast, the researchers find no evidence that US teacher strikes, which are much shorter, affected reading or math achievement for students in the year of the strike, or in the five years after. While US strikes lasting two or more weeks negatively affected math achievement in both the year of the strike and the year after, scores rebounded for students after that.

In fact, Lyon said they could not rule out that the brief teacher strikes actually boosted student learning over time, given the increased school spending associated with them. A recent influential meta-analysis on school finance found that increasing operational spending by $1,000 per student for four years helped student learning.

It’s possible higher wages could reduce teacher burnout, or the need to work second jobs, leading to improved performance in the classroom. Still, Lyon explained, it’s also possible that increased spending on teachers would not lead to higher student test scores, if wage gains went primarily to more experienced teachers, or to pensions, or if teachers were already maximizing their effort before the strike.

Strikes were more common in conservative, labor-unfriendly areas

Overall, the researchers found that teacher union density has fallen more sharply than previously recognized. According to federal data, 85 percent of public school teachers reported being in a union in 1990, falling to 79 percent in 1999, and then to 68 percent by 2020.

“As someone who studies unions, that statistic alone is still pretty surprising to me,” Lyon said. “And it came from the federal Schools and Staffing Survey, which is one of our best data sources.” Tracking teacher union membership can be complicated because of mergers, and because the two national unions — the American Federation of Teachers and the National Education Association — include non-teachers and retired teachers in their ranks. Still, even with the drop, the 68 percent dwarfs that of the private sector, where just 10 percent of workers are in unions.

Roughly 35 states have laws that either explicitly ban or effectively prohibit teacher strikes, but those laws haven’t stopped educators from organizing labor stoppages. (Nearly every state in the #RedforEd teacher strikes from 2018 and 2019 — including Arizona, Kentucky, West Virginia, and Oklahoma — had banned teacher strikes.)

In compiling their data set, Lyon, Kraft, and Steinberg included both legal strikes and illegal work stoppages, including mass walk-outs, “sick-outs” (when teachers call in sick en masse), or so-called “wildcat strikes” (when educators strike without the support of union leadership).

Perhaps counterintuitively, they found strikes were more common in more conservative, labor-hostile states, something they attributed mostly to large-scale coordinated strikes across districts happening more often in those places. Individual district strikes were more likely to occur in liberal areas, where such actions are legal.

The teacher uprisings over the last decade have helped boost support from parents and the broader public, who report in surveys backing for educator organizing and increased teacher pay. The percentage of the public who see teacher unions as a positive influence on schools rose from 32 percent in 2013 to 43 percent in 2019, according to Education Next polling. A majority of the US public supports teachers having the right to strike, which suggests educators may be comfortable using this tactic going forward.

School Suspensions, Test Scores, and Lead Poisoning

Originally published in The American Prospect on June 9, 2017.
—–

Over the past several years, education advocates and civil rights groups have been sounding the alarm on the harms of exclusionary school discipline policies. Critics say these punishments—suspensions, expulsions, and school-based arrests—are increasingly doled out for minor infractions, and disproportionately given to students of color.

A National Bureau of Economic Research (NBER) working paper published in May adds a new wrinkle to the debate on disparities in school discipline: Economists found causal evidence linking young children with higher exposures to lead in their bloodstream with an increased probability of getting suspended from school and placed in juvenile detention.

Anna Aizer, a Brown University economist, and Janet Currie, a Princeton University economist were the first researchers to look at the relationship between lead exposure and school discipline. While others have explored links between lead exposure, criminal activity, and cognitive development, this study breaks new ground by tracking individual children over time.

One reason why researchers hadn’t studied lead and school discipline before is due to insufficient data. Most children exposed to lead tend to come from disadvantaged backgrounds, but even if a researcher had identified a relationship between lead levels and future disciplinary infractions—they could not determine if lead was driving the relationship, or the if underlying poverty was the cause.

Getting precise measurements of lead has also been a challenge for researchers. Blood tests are relatively cheap and easy to administer, but depending on when the blood is drawn, the results may or may not accurately capture a child’s exposure because lead remains in an individual’s bloodstream for a little over a month before their organs absorb the metal. In other words, researchers worry that they potentially underestimate children’s exposure to lead when they analyze blood tests.

Only 11 states and the District of Columbia even mandate blood-lead tests for children, which exacerbates the problem of measuring lead. Researchers estimate that maybe a quarter of all U.S. children nationally ever get screened. A Reuters investigation published last year documented how millions of kids fall through the cracks, even among those who are required to undergo state-mandated screening.

Yet Rhode Island, where Aizer and Currie did their study, presented some unique research opportunities. Unlike most states, the number of children in Rhode Island who get screened for lead is high, close to 80 percent. Not only that, but Rhode Island children are also screened three times on average during their first six years of life. This means that the chances of getting an accurate measure of lead exposure are significantly higher than usual since kids are tested multiple times.

Aizer and Currie studied Rhode Island children born between 1990 and 2004. They accessed state health department data on each child’s preschool blood lead levels from 1994 to 2010, and then linked that information to school suspension data for the 2007-2008 and 2013-2014 school years. The researchers also compared this information to data from the state’s juvenile detention facility and all Rhode Island correctional institutions.

“What we find is that there’s a pretty robust relationship between early childhood lead levels as measured by the blood tests and future disciplinary infractions,” says Aizer. Not only were children with elevated blood-lead levels more likely to be suspended, but Aizer and Currie found that suspended children were also ten times more likely to end up in juvenile detention.

The two most common sources of lead poisoning are old paint and soil. A robust movement to reduce lead in the environment didn’t really take off until 1971, when the U.S. surgeon general issued a policy statement on childhood lead poisoning, and the EPA’s first administrator declared that “an extensive body of information exists which indicates that the addition of alkyl lead to gasoline … results in lead particles that pose a threat to public health.”

The federal government finally banned lead-based paint in 1978, but older housing units may still contain the metal. Washington also phased out leaded gasoline between 1979 and 1986. While the amount of lead in soil has significantly decreased over time—because it’s been washed out, blown, or tracked away—the soil near roadways, especially in urban areas, may still contain lead that had seeped into the ground from automobile exhaust.

One strategy Aizer and Currie used to track the relationship between blood-lead levels and future misbehavior was to look at where kids with high lead exposure live. “We tried to address the question of confounding variables by looking at the same kids who live near the same roads at different periods of time [with the] same racial composition, the same income,” says Aizer. “The difference is they are living or born near roads fifteen years later, as the amount of lead in the soil near roads declined because of repaving or rain runoff or new turf.” The economists found that the dramatic declines in lead levels were linked to significant declines in disciplinary infractions.

Aizer and Currie’s findings related to race and gender were complex. They found that white girls with high lead levels were not likely to be suspended, but that was not true for African-American girls, or, generally, for boys. They also found that the relationship between lead and suspensions was much stronger for kids who received free lunches. While Aizer and Currie have not settled on definitive explanations for why lead had different effects on different children, they have developed some possible theories.

“There is some evidence that cognitive stimulation reduces or mitigates the negative impact of lead,” says Aizer, who explains that if a child who receives free lunches—a proxy for low-income households—gets less stimulation, they might react differently to the lead exposure compared to a wealthier student exposed to lead. Another explanation could be that schools respond differently to the misbehavior of different groups of students. Currie and Aizer say that there is evidence for both interpretations.

Their research comes on the heels of another working paper Aizer, Currie, and two other colleagues co-published last year. Using a slightly different research strategy but still involving Rhode Island children, they found that reducing children’s lead levels had significant positive effects on third grade reading test scores, especially for black and Hispanic students. “A one-unit decrease in average blood-lead levels reduces the probability of being substantially below proficient in reading by 3.1 percentage points,” the economists concluded.

While average childhood lead levels have fallen by more than 90 percent over the last 40 years, progress has been uneven, especially in poor urban areas with old housing stock.

Some states, like Rhode Island, have taken effective steps to reduce exposure to lead. In 1997, for example, Rhode Island officials implemented policies that required landlords to ensure that their rental properties were lead-free. One policy required all Rhode Island landlords to obtain “lead-safe certificates” if they intend to rent their properties. Another required landlords who own buildings where a child had elevated lead levels to mitigate the lead hazard or face legal action by the state attorney general.

In 2010, the U.S. spent more than $80 billion on corrections expenditures at the federal, state, and local levels, according to the Brookings Institution. They found that total correctional expenditures had more than quadrupled over the past 20 years. And in 2014, after surveying 46 states on their costs of juvenile detention, the Justice Policy Institute reported that average per-person costs for the most expensive juvenile confinement options reached $407 a day, or nearly $150,000 per year.

“Governments need to think about this,” says Aizer. “Crime is just an incredibly expensive outcome for a state, and lead mitigation is so much cheaper relative to that.”