The Learning Recession Started in 2013. Here Is What Recovering Districts Did Differently.
The conventional story about American student achievement goes like this: schools were on a steady upward path until COVID hit, school closures created sweeping learning loss, and districts have been digging out ever since. A report from the Education Scorecard challenges that story at its foundation.
According to the Scorecard, a research project led by economists and education scholars at Dartmouth, Harvard, and Stanford, US students entered a learning recession around 2013. That is roughly seven years before the first pandemic-era school closure.
The finding reframes what recovery actually requires. If the decline predates COVID, then catching students back up to a pre-pandemic baseline is not the goal. The goal is reversing a slide that the field has been living with for more than a decade. And the districts that have managed to reverse it share a pattern that has less to do with how much they spent on teacher training and more to do with what they did after the training ended.
What the NAEP Data Actually Shows
The Education Scorecard draws on National Assessment of Educational Progress (NAEP) data going back to 1990. The picture it paints has two distinct chapters.
From 1990 to 2015, 4th and 8th grade students in the US gained roughly two grade levels of math achievement. That represents sustained, measurable progress across 25 years. Then it started to unwind.
What is striking is not just the reversal, but its timing. According to Thomas Kane, the Harvard economist who helped create the Scorecard, reading scores fell roughly as much before the pandemic as they did during its peak. COVID accelerated a decline that was already underway.
Morgan Polikoff, an education professor at the University of Southern California, told The 74 that the Scorecard was notable for its ambition to “tell the whole story” of the downturn, even without dispositive proof of its causes. That framing matters for how districts respond. A short-term emergency calls for a short-term fix. A decade-long recession calls for a change in how the system supports teaching.
Two Hypotheses, No Causal Proof
The Scorecard researchers identify two leading explanations for why the decline started around 2013. They present these as hypotheses, not proven causes.
The first involves the rollback of federal accountability under No Child Left Behind (NCLB). In 2011, the Obama administration began granting states waivers from NCLB requirements. More than 40 states ultimately received them. By 2014, fewer than 10 percent of schools were flagged for missing learning benchmarks, compared to the high percentages the original law had surfaced. The report also tracks how quickly accountability faded from public attention: media references to federal accountability fell 97 percent after 2017. Whether accountability drove the earlier gains, and whether its rollback drove the decline, remains contested. The researchers flag the correlation. They do not claim causation.
The second hypothesis involves social media. Pew Research, cited in the report, found that 46 percent of US teens reported being online almost constantly by 2022. The rise in smartphone and social media use among school-aged children roughly tracks the timing of the decline. Here too, the researchers are careful: a Stanford study examining phone bans found improvements in student psychological well-being but mostly no academic improvement after two years. The relationship between screen time and learning outcomes is real but not simple.
The honest answer, based on what the Scorecard presents, is that researchers do not yet know which factor matters more, or whether other forces are also at work. The report is a diagnosis, not a prescription.
What the Recovering Districts Did
The more actionable part of the Scorecard focuses on recovery. Approximately 108 districts posted sizable gains in both math and reading since 2022. Another 450 or so showed gains in at least one subject. The list includes Atlanta, Birmingham, Alabama, and Compton, California.
The Scorecard authors produced a case study on the District of Columbia Public Schools (DCPS), which offers the most specific look at what recovery actually looks like on the ground.
DCPS reading achievement for grades 3 through 8 now exceeds 2018 levels by roughly half a grade level. That is not a return to the pre-pandemic baseline. It is improvement beyond it.
Two specific levers appear in the DCPS case study. The first is a proprietary K-5 English curriculum. The second is stipends paid to teachers for completing specialized literacy training. The second lever is the one worth sitting with. DCPS did not just adopt new curriculum and hand it to teachers. The district paid teachers for the time it takes to learn how to use it well, which is to say it paid for the part that comes after the initial rollout. Professional development was treated as a sustained investment, not a one-day event.
Mississippi’s statewide reading reforms are cited as a state-level example showing similar results at scale, again built on extended teacher support rather than a single training push.
Why the Pattern Points to Between-Session Coaching
The recovery data lines up with a finding that research on teacher development has produced consistently for decades. Curriculum and training are inputs. Whether they change classroom practice depends on what happens after the training ends.
The most cited evidence on this point comes from Bruce Joyce and Beverly Showers, whose work has been replicated since the 1980s. They found that workshop training on its own transfers to actual classroom practice roughly five percent of the time. Add structured follow-up in the classroom, the kind of coaching that helps a teacher try a practice, reflect on what happened, and adjust, and that transfer rate climbs to roughly 90 percent. More recent work points the same direction. A 2018 meta-analysis by Matthew Kraft, David Blazar, and Dylan Hogan found that instructional coaching has significant positive effects on both teaching quality and student achievement.
This is the lever underneath the DCPS results. The district did not simply buy a curriculum. It paid for the sustained, supported practice that turns a curriculum into changed instruction. That is what the stipend was for.
For district and instructional leadership, the DCPS example raises a practical problem. The approach worked, but the district had both the funding for teacher stipends and the infrastructure to deliver specialized literacy training at scale. Most districts do not have those resources sitting idle. The replicable lesson is not DCPS’s budget. It is the principle underneath it: the investment that moves outcomes is the one that supports teachers after the initial training, not the training itself.
That principle also sits at the center of the current debate over federal funding. Title II-A, the main federal program supporting educator professional learning, represents roughly $2.19 billion annually and is currently being debated in Congress, with proposals on the table to consolidate or reduce it. However the funding question resolves, the recovery data suggests the more consequential variable is not how much is spent on teacher development but what kind of development that money buys.
Scaling the Lever Without a DC-Sized Budget
The hard part for most districts is not understanding that between-session support matters. It is delivering it. A teacher who completes a training program and then returns to a school with no follow-up, no structured reflection, and no mechanism for ongoing practice tends to revert. Closing that gap has traditionally required hiring more instructional coaches, and coaching capacity is exactly what most districts cannot afford to add.
This is where AI-supported coaching changes the math. AI Coach by Edthena gives every teacher in a building access to structured, on-demand coaching between formal cycles, without adding to existing workloads or requiring a school to hire additional coaches. Teachers reflect on their own practice, set a goal, and work through it with a virtual coach on their own schedule. It is a supplement to in-person coaching, not a replacement for it, and it extends the kind of sustained follow-up the DCPS results were built on to districts that cannot fund teacher stipends and proprietary curriculum at the same time.
The districts posting the gains the Scorecard describes are not running single-day workshops and hoping practice changes. They are building systems around teacher growth. Curriculum is one input. Sustained, structured support between sessions is the other, and it is the one that has historically been out of reach for districts without DCPS-level budgets. That is the part that is now affordable in a way it was not five years ago.
Frequently Asked Questions
When did the US learning recession begin?
According to the Education Scorecard, a research project led by scholars at Dartmouth, Harvard, and Stanford, US students began losing academic ground around 2013. Reading scores fell roughly as much before the COVID-19 pandemic as they did during its peak, meaning the decline predates the pandemic by about seven years.
What caused the learning recession?
Researchers have identified two leading hypotheses: the rollback of No Child Left Behind accountability starting in 2011, and the rise of social media and smartphone use among school-aged children. The Scorecard authors note these are not proven causes. The relationship between these factors and student outcomes is still being studied.
Which districts have recovered from learning loss?
Approximately 108 districts showed sizable gains in both math and reading since 2022, according to the Education Scorecard. The District of Columbia Public Schools (DCPS) is highlighted as a case study: reading achievement for grades 3 through 8 now exceeds 2018 levels by roughly half a grade level.
What did DC Public Schools (DCPS) do to improve reading scores?
The Scorecard case study identifies two levers: adoption of a proprietary K-5 English curriculum and stipends paid to teachers for completing specialized literacy training. Both were treated as sustained investments rather than one-time events, with the stipends specifically funding the ongoing practice that follows initial training.
Why doesn’t teacher training alone improve student outcomes?
Research on professional development is consistent on this point. Bruce Joyce and Beverly Showers found that workshop training on its own transfers to classroom practice roughly five percent of the time, while adding structured coaching follow-up raises that figure to roughly 90 percent. The variable that predicts whether training changes instruction is the sustained, supported practice that comes after it, not the training event itself.
