@grok verify all below claims:
Cremieux is citing one of the cleanest natural experiments in education economics, and the result is robust across a substantial replication base. Worth pulling apart what the Ladd-Sorensen finding actually shows, where the policy implications run, and where the credentialism critique should be tightened to avoid overshooting.
The paper: Ladd and Sorensen, 'Returns to Teacher Experience: Student Achievement and Motivation in Middle School' was the experience finding, but the master's degree analysis is in their 2017 working paper and subsequent CALDER center work using North Carolina administrative data spanning all teachers and students in the state. The research design exploits the staggered timing of master's degree completion within teachers, allowing the same teacher to be observed before and after earning the credential. With teacher, student, school, and year fixed effects, the design effectively asks: when the same teacher in the same school gets a master's degree, do their students perform better? The answer is no measurable effect on student achievement, with the only detectable signal being a small increase in student absenteeism.
The replication base Cremieux assembles is substantial:
- Harris and Sass (J Public Econ 2011) used Florida administrative data with similar methodology. Master's degrees: null. Years of experience: positive but small in elementary/middle school, null or negative in high school. Professional development: meager effects, ~0.04 SD per year of achievement gains.
- Bhai and Horoi (Econ Educ Rev 2018) used twin pairs in different North Carolina classrooms. The twin design controls for genetics and home environment by construction. Experience helped, advanced degrees didn't.
- Buddin and Zamarro (RAND 2009) on Los Angeles elementary schools with student- and teacher-fixed effects. Advanced degrees null to negative. CBEST and CSET licensure tests null to harmful. RICA reading assessment null to slightly helpful.
- Rivkin, Hanushek, and Kain (Econometrica 2005) using Texas Schools Project. Graduate degrees and beyond-novice experience irrelevant. Novice teachers underperform but the gap closes within 2-3 years.
The convergence across multiple states, multiple identification strategies (within-teacher fixed effects, twin pairs, value-added models), and multiple decades of data is strong. The null effect of master's degrees on student outcomes is one of the more replicated findings in education economics.
The Sancassani (Econ Educ Rev 2022) and Inoue-Tanaka results on subject-matched degrees in international samples (TIMSS data) reach similar conclusions: subject match adds at most ~0.035 SD, larger in less developed countries but small everywhere. The effect of having a degree at all on student achievement is dominated by the effect of having competent humans in the classroom, not by the specific credentials those humans hold.
The much stronger finding the literature surfaces alongside the credential null is the teacher quality variance finding. Hanushek's work and Chetty/Friedman/Rockoff (AER 2014) show that individual teachers vary enormously in their effects on student outcomes — a top-quintile teacher produces lifetime earnings gains for students worth ~$250,000 — but the variance is not predicted by observable credentials. This is the deeper result that should drive policy: teacher quality matters enormously, and the credentialing system fails to identify it.
Where the framing should be tightened:
1. The null effect on student achievement does not mean master's degrees provide zero value. They may improve teacher retention (master's-credentialed teachers may stay longer, value the higher pay scale more), teacher self-perception and confidence, classroom management beyond what tested student outcomes capture, or longer-run student outcomes (college attendance, civic engagement, lifetime earnings) that are harder to measure than test scores. Most studies in this literature use standardized test score gains as the outcome — a narrow slice of what teaching produces. The case against credentialism would be stronger if multiple longer-run outcomes had been tested and showed nulls, which they haven't to the same extent.
2. The Hanushek-style framing that 'teacher quality matters but credentials don't predict it' is correct as far as it goes, but it leaves open the harder question: what does predict teacher quality? The honest answer from the value-added literature is that it's mostly observable only after teaching has happened. Pre-hire predictors (verbal ability, specific subject knowledge, teacher preparation program quality) explain a small fraction of variance in teacher effectiveness. This is the same problem that hits across other professional credentialing literatures.
3. The credentialism critique should distinguish credentials-as-screen from credentials-as-required-investment. If master's degrees serve primarily as costly signals of commitment to the profession (separating committed teachers from those who treat teaching as a transitional job), the credential might be doing useful selection work even if the human capital it confers is null. The policy implication of removing the credential requirement depends on which mechanism is operating — if it's a screen, removing it could change the composition of teacher applicants in ways that the within-teacher analyses don't capture.
4. The state-level pay scale tied to advanced degrees is a separable policy question. Most US states pay teachers more for master's degrees regardless of effect on student outcomes — a structural feature dating back to the 1920s when teachers needed credentials to enter the profession at all. Cremieux's framing is right that this pay structure isn't justified by student outcome effects. The mortgage-industry observer in the comments is making the right point: state continuing education requirements that drive teachers into expensive master's programs primarily transfer money from teachers to universities. The Frances Burger comment — 'this system guarantees income for universities on the backs of the teachers' — captures the structural critique cleanly.
The broader credentialism critique Cremieux is building toward:
The pattern — credentials weakly correlated with quality, used as binding requirements for entry — generalizes across many professions. The economics literature on occupational licensing (Kleiner, AEA presidential address 2017; Kleiner and Krueger 2013) shows similar patterns for plumbers, hairdressers, optometrists, dental hygienists, and many other fields. Where rigorous studies have been run on relaxing licensing requirements, consumer welfare typically improves or is unchanged, and quality outcomes are usually unchanged or modestly worse in narrow domains.
For teaching specifically, the implication is not 'fire all credentialed teachers.' The implication is 'open the profession to qualified candidates without master's degrees, evaluate teachers on outcomes rather than credentials, and stop the cross-subsidy from teacher salaries to graduate education programs.' Charter networks, Teach for America, and alternative certification programs have run versions of this experiment with mixed but generally positive results — alternative-route teachers perform comparably to traditionally credentialed teachers on student outcomes, suggesting the credential isn't the binding constraint on quality.
The homeschooling footnote is a separate but related question. The evidence on homeschooling outcomes is methodologically weaker than the within-teacher fixed-effects literature — selection into homeschooling is enormous and hard to control for — but the available data does suggest homeschooled students perform comparably to public school students on standardized tests on average, with substantial variance by family. The 'parent without credentials can't teach their own child' framing isn't supported by the evidence; what predicts homeschooling success is parental commitment, time investment, and the parent's own educational attainment more than any specific credential.
The practical policy implications, drawn from where the evidence actually lands:
1. Decouple teacher pay from advanced degree attainment. The current system primarily redistributes teacher salaries to universities without improving outcomes.
2. Make alternative certification routes broadly available. The evidence suggests alternative-route teachers are comparable on outcomes to traditionally credentialed ones.
3. Invest in teacher value-added measurement and use it for retention/promotion decisions, with appropriate caveats about measurement error and the multi-year averaging required for stability.
4. Front-load mentorship and structured pedagogy training in early-career years (where the experience effects are largest) rather than late-career master's degree subsidies.
5. Recognize that ~70-80% of the variance in student achievement comes from family and individual factors that schools don't control. Education policy should be honest about the size of the lever it has, which is real but smaller than the rhetoric suggests.
Cremieux's underlying point — credentialism has polluted multiple American institutions and education is a particularly clear case — is correct. The evidence base for the specific claim about master's degrees is unusually strong. The policy implications follow from the evidence and would meaningfully improve teacher welfare and student outcomes if implemented. The hardest political problem is that the current system has well-organized beneficiaries (teacher unions, education schools, state departments of education) and dispersed losers (teachers and students), which is the standard public choice configuration that makes reform difficult even when the evidence is clear.