The MSEE Sweet Spot:
Does graduate school pay off? Here's which degrees have the highest returns, study finds
Why Stopping Short of a PhD Often Pays — and How AI Is Rewriting the Calculus
A landmark 2025 NBER study confirms what many engineers have long suspected: the marginal return on an EE doctorate, once opportunity costs are factored in, is narrower than the credential implies — and a disrupting AI labor market is tightening the calculus further.
April 4, 2026 | Analysis & Opinion
BLUF — Bottom Line Up Front
A 2025 National Bureau of Economic Research working paper by Yale economist Joseph Altonji and Vassar's Zhengren Zhu, drawing on administrative records for 121 graduate degrees in Texas, finds that a master's degree in electrical engineering yields a cost-adjusted lifetime return of approximately 4 percent — and a PhD in EE, excluded from the study's scope because it falls outside professional master's and doctoral categories, carries an even more ambiguous payoff once four-to-six years of foregone industry salary are priced in. For the majority of practicing engineers targeting industry careers, the evidence increasingly suggests the MSEE represents the optimal terminal degree. That calculus is now further complicated by AI-driven restructuring of the EE job market, which is simultaneously creating premium demand for systems-level expertise and compressing entry-level roles — further eroding the traditional justification for doctoral study as a career accelerator.
The Study and Its Findings
Every few years, an economics paper lands with enough empirical rigor to reframe a question that practitioners have been answering by intuition for decades. The February 2025 working paper by Joseph G. Altonji of Yale University and Zhengren Zhu of Vassar College — published as NBER Working Paper No. 33530 — is that paper for graduate engineering education. Using administrative data from the Texas Education Research Center, Altonji and Zhu estimated the causal effects of earning any of 121 distinct graduate degrees on long-run earnings, isolating degree-specific returns from the confounding factor of student selection.
The headline finding is that graduate programs increase earnings by 17 percent on average — but the variance across fields is so wide (standard deviation of 0.176 log points) as to make the average nearly meaningless for individual decision-making. The professional degrees dominate: medicine (MD) delivers a cost-adjusted lifetime return of 173 percent; pharmacy (PharmD), 68 percent; and law (JD), 41 percent. Engineering master's degrees, by contrast, return 4 percent for electrical engineering, 4 percent for mechanical engineering, and just 2 percent for computer engineering — after accounting for tuition costs and foregone earnings during enrollment.
Critically, the study notes that high absolute post-graduation earnings do not imply high returns. Electrical engineering master's graduates earn a post-degree average of roughly $130,953 annually, according to the Texas data — but they also entered graduate school already earning $83,092, one of the highest pre-enrollment baselines of any graduate population. The incremental lift, net of cost, is modest precisely because these students were already high earners before they enrolled.
"High postgraduate earnings don't always mean higher returns on investment. Engineering offers high postgraduate average earnings but only moderate returns on investment." — Zhengren Zhu, Vassar College, quoted in Inside Higher Ed, March 2026
Where the PhD Fits — and Where It Doesn't
The Altonji-Zhu study explicitly excludes research PhDs from its scope — confining its analysis to master's degrees and professional doctorates (MD, JD, PharmD). That methodological choice actually illuminates the PhD's economic position by implication: the paper's framework for cost-adjusted returns, if applied to a research doctorate in EE, would confront an even starker arithmetic.
A fully funded EE PhD program at a major research university typically provides a stipend of roughly $32,000 to $52,000 annually — with institutions like the University of Michigan guaranteeing around $41,000 — plus tuition coverage. Stanford, MIT, and comparable programs provide similar packages. Against an MS-holding engineer who enters industry at $95,000 to $120,000, the forgone earnings over a four-to-six year program represent a gap of $250,000 to $450,000 in cumulative pre-tax income, before counting compounding investment value.
PayScale data from 2021 places the average EE/ECE PhD salary at $149,000 — a figure consistent with Salary.com data showing senior hardware engineering PhDs averaging $114,930 and ECE professors averaging $144,115. A rough industry estimate is that a PhD commands a salary premium of $5,000 to $25,000 annually over an MS-level peer with equivalent experience. At that differential, full break-even on the opportunity cost can require eight to ten years of post-graduation earnings — and, as Quora respondents with direct industry experience observe, "PhDs who are at the lower end of the scale might never make up the years of lost income."
The exceptions matter, however. Research scientist roles at institutions like NVIDIA, Qualcomm, Apple, Google, and AMD increasingly require or strongly prefer a doctorate for work involving novel algorithm development, chip architecture research, and foundational AI hardware design. National laboratory positions at Los Alamos, Lawrence Livermore, or Sandia — and aerospace defense research at organizations like DARPA — similarly treat the PhD as a near-prerequisite for principal investigator status. And for any aspiring tenure-track faculty member, the doctorate is simply non-negotiable.
The Selection Bias Problem
One of the most technically significant contributions of the Altonji-Zhu paper is its treatment of selection bias — the persistent methodological flaw in earlier studies comparing graduate and non-graduate earnings. Simply observing that PhD engineers earn more than MS engineers conflates degree value with the baseline characteristics of people who choose to pursue doctorates. Those individuals typically exhibit higher academic ability, stronger risk tolerance, and more intense research motivation — traits that would have generated above-average earnings regardless of credential.
By exploiting the longitudinal administrative data structure of the Texas ERC records, Altonji and Zhu estimate earnings before and after degree completion for the same individuals, controlling for undergraduate major and pre-enrollment earnings trajectory. Their approach isolates the causal wage effect of the degree itself, rather than the characteristics of degree-seekers. The 4 percent cost-adjusted return for EE master's degrees reflects this rigor — it is the increment attributable to the MSEE specifically, not to the type of person who earns one.
The implication for PhD decision-making is sobering: much of the apparent wage premium for doctoral engineers may reflect selection, not causation. The engineers who complete EE PhDs and land high-compensation research positions would likely have performed well even without the doctorate — especially given the already-high earnings baseline of EE professionals entering graduate school.
The AI Wildcard
The 2025–2026 research landscape surrounding graduate engineering ROI has acquired a disruptive new variable: artificial intelligence. The question is no longer merely whether a PhD pays off relative to an MS — it is whether the traditional rationale for doctoral training, namely access to research-intensive career tracks, remains stable as AI reshapes what research-intensive work looks like.
A February 2026 paper by Dallas Fed economist J. Scott Davis found that AI is simultaneously reducing entry-level hiring and raising wages for experienced workers in AI-exposed occupations — a bifurcation that has direct implications for the traditional PhD pipeline, which depends on a supply of entry-level research and teaching assistants as training inputs. Harvard economists Lawrence Katz and Claudia Goldin, in September 2025, found that the college wage premium has barely moved since 2000, with the San Francisco Fed attributing the stagnation primarily to reduced demand for cognitive routine work. And the World Economic Forum reported in early 2026 that AI skills now command a 23 percent wage premium — versus only 8 percent for an isolated bachelor's degree.
Within EE specifically, AI is restructuring the work itself at every level. As Matthew Graham, senior group director at Cadence, told Semiconductor Engineering in January 2026: "There absolutely is a set of skills that will no longer be required to do chip design." AI-assisted EDA tools are compressing design cycles from years to months; neural receivers are outperforming classical signal processing algorithms; and predictive maintenance systems are displacing manual fault-diagnosis workflows. The practical effect is that many tasks historically used to justify PhD-level deep specialization — exhaustive literature synthesis, simulation-based optimization, systematic algorithm benchmarking — are now partially or substantially augmented by AI tools accessible to MSEE-level engineers.
Roles that did not exist five years ago — AI Systems Engineer, Edge AI Architect, Robotics Hardware Developer — are commanding premium compensation without requiring the doctoral credential, because they demand systems integration agility over depth of specialization in a single research domain. As NVIDIA CEO Jensen Huang has repeatedly put it, engineers using AI will replace engineers not using it — a framing that privileges adaptive application over academic specialization.
"We don't need that many junior engineers to go deep into the abstraction. They just need to be at the right level, and still they can work on the same things and gain experience." — Industry engineer quoted in Semiconductor Engineering, January 2026
The Practical Verdict for Industry-Bound Engineers
The data portrait that emerges from the combined weight of the Altonji-Zhu findings, BLS earnings data, AI labor market research, and practitioner consensus is coherent and surprisingly durable across sources: for the large majority of EE professionals targeting industry careers in defense, aerospace, semiconductors, telecommunications, or power systems, the MSEE is the economically optimal terminal degree.
The MSEE provides the credential floor required by most senior engineering roles; enables immediate entry at a salary of $95,000 to $120,000; avoids four to six years of compounding opportunity cost; and, in the emerging AI-integrated workplace, offers sufficient theoretical depth to productively direct, validate, and integrate AI-generated technical work. The Physics Forums engineering community — representing practicing engineers across multiple decades of experience — broadly corroborates this: "The vast vast majority of engineers that I have come in contact with over the past several decades have either BS or MS degrees, mostly BS, although MS is becoming more important these days."
The PhD retains clear justification in a narrower set of scenarios: genuine passion for and aptitude in basic research; intention to pursue academic tenure; a credible path to principal investigator or chief scientist roles at top-tier research organizations; or a plan to leverage doctoral research into commercializable intellectual property. Semiconductor startup founders and venture-backed deep-tech entrepreneurs sometimes find the PhD network, the publications record, and the patent portfolio critical competitive assets.
But as a general proposition for the student sitting at a master's qualifying exam and weighing whether to stay — the accumulated evidence in 2026 suggests the correct answer is increasingly: take the MSEE, enter the workforce, and let compounding income and industry experience do what four more years of stipend-funded laboratory work cannot.
A Critical Caveat: Geographic and Institutional Generalizability
The Altonji-Zhu findings must be interpreted with one significant structural limitation clearly in view: the Texas Education Research Center data captures graduate students enrolled at Texas institutions — predominantly UT Austin, Texas A&M, Texas Tech, and the University of Houston — who subsequently enter the Texas labor market. No Ivy League institutions, no UC system schools, no MIT, no Stanford, no Caltech are represented. For engineering in particular, that exclusion is not minor. It is potentially decisive.
U.S. Department of Education College Scorecard data, updated March 2026, makes the gap concrete. Five years after receiving a bachelor's degree in electrical engineering, median salaries by institution tell a strikingly different story from the Texas average of $130,953. UC Berkeley EE graduates report a median of $240,462 — MIT graduates $190,731 — Stanford $187,527 — and Carnegie Mellon $183,814. Cornell places at $131,283 and UCLA at $127,528. Even UT Austin, the flagship Texas institution most likely to be well-represented in the Altonji-Zhu sample, comes in at $123,101, with Texas A&M at $107,294.
At UC Berkeley, MIT, Stanford, and Carnegie Mellon, the five-year median EE salary is not a modest increment above the Texas average — it is a different number in a different labor market altogether. — College Transitions Dataverse, derived from U.S. Dept. of Education College Scorecard (March 2026)
| Institution | Median EE Salary (5 yr) | vs. Texas Study Avg. |
|---|---|---|
| UC Berkeley | $240,462 | +84% |
| MIT | $190,731 | +46% |
| Stanford | $187,527 | +43% |
| Carnegie Mellon | $183,814 | +40% |
| Cornell | $131,283 | +0.3% |
| UCLA | $127,528 | −3% |
| UC San Diego | $120,436 | −8% |
| UT Austin | $123,101 | −6% |
| Georgia Tech | $109,110 | −17% |
| Texas A&M | $107,294 | −18% |
| Texas Study Average (MSEE) | $130,953 | — |
Source: U.S. Dept. of Education College Scorecard, via College Transitions Dataverse (March 2026). Five-year median salary for EE bachelor's graduates entering the workforce. Comparison column references the Altonji-Zhu (2025) Texas post-grad MSEE average earnings of $130,953.
Several structural factors explain the elite-program premium. First, these institutions feed directly and preferentially into the highest-compensation labor markets in the United States: Silicon Valley semiconductor and AI firms (NVIDIA, Apple, Qualcomm, Google, AMD), Bay Area hyperscalers, and Seattle tech campuses where total compensation packages — base salary plus restricted stock units plus signing bonus — routinely place early-career engineers in the $180,000 to $300,000 range within five years of graduation. Second, the five-year salary window in College Scorecard data captures initial equity vesting cycles at these firms, which are a substantial fraction of total compensation. Third, the alumni networks and recruiting pipelines at elite programs are institutionally embedded in ways that fundamentally alter the job-offer distribution available to graduates.
It is also worth noting a methodological nuance that slightly complicates direct comparison: the College Scorecard figures reflect primarily bachelor's-level graduates entering the workforce, while the Altonji-Zhu figures measure the returns to graduate enrollment specifically. At elite schools, a substantial fraction of EE graduates proceed directly to MS or PhD programs rather than industry — which may actually concentrate the bachelor's-to-employment pipeline toward the most aggressively recruited candidates, pushing the five-year median upward.
The implications for the study's 4 percent cost-adjusted return figure for MSEE degrees are significant but nuanced. The Altonji-Zhu return estimate may reasonably represent the typical MSEE experience for the broad population of graduate students at public, non-elite programs — which constitutes the large majority of MSEE enrollment nationally. But for a student choosing between an MSEE at MIT or Stanford and forgoing the degree entirely, the relevant counterfactual is not the Texas average. The incremental return on an elite MSEE, measured against a Texas-school BSEE baseline, is likely substantially larger than 4 percent — and measured against a Bay Area tech labor market, may be substantially larger still.
The rank ordering of degree types in the study — professional degrees over engineering, engineering over social sciences — is almost certainly robust across geographies. But the absolute return percentages, particularly the low single-digit figures for EE, should be understood as lower-bound estimates for elite-program graduates entering coastal high-wage labor markets, not as universal benchmarks.
Policy Implications
The findings have implications beyond individual decision-making. The Education Department launched a graduate earnings disclosure tool in 2025 to help prospective students evaluate program-level ROI — a direct response to the kind of research the Altonji-Zhu paper represents. Preston Cooper, a senior fellow at the American Enterprise Institute, summarized the policy stakes bluntly: "Outside some highly compensated professional fields, including law and medicine, graduate school is a risky proposition."
For engineering specifically, graduate program accountability should incorporate cost-adjusted returns — not raw post-graduation earnings — as the central benchmark. A program producing graduates earning $130,000 per year may still be delivering a 4 percent return if its students' counterfactual trajectory would have landed them at $125,000. Transparency on that delta, degree-specific and institution-specific, is what prospective students actually need to make rational decisions.
Key Data Summary
| Degree | Pre-Grad Earnings | Post-Grad Earnings | Cost-Adjusted Return | Notes |
|---|---|---|---|---|
| MD | $48,650 | $181,691 | 173% | Highest ROI of any degree studied |
| PharmD | $49,892 | $132,460 | 68% | |
| JD | $55,521 | $132,520 | 41% | School rank matters significantly |
| MPA | $52,519 | $79,246 | 26% | |
| Nursing (MS) | $70,215 | $108,370 | 12% | |
| MBA | $72,248 | $112,458 | 13% | Higher-ranked programs yield more |
| Civil Engineering (MS) | $61,483 | $111,784 | 19% | Best engineering ROI in study |
| Computer Science (MS) | $69,419 | $110,106 | 6% | |
| Electrical Engineering (MS) | $83,092 | $130,953 | 4% | High absolute earnings, low marginal ROI |
| Mechanical Engineering (MS) | $73,429 | $119,551 | 4% | |
| Computer Engineering (MS) | $82,261 | $113,554 | 2% | Lowest among engineering disciplines |
| Psychology (MS) | $43,537 | $61,445 | -8% | Negative lifetime return |
| Clinical Psychology (MS) | $45,577 | $57,894 | -5% | |
| Social Work (MS) | $41,166 | $55,590 | -2% |
Source: Altonji & Zhu (2025), NBER Working Paper No. 33530, using Texas Education Research Center administrative data. Includes master's degrees and professional doctoral degrees; excludes research PhDs. Cost-adjusted returns reflect estimated change in lifetime income after accounting for tuition and foregone earnings during enrollment.
Verified Sources and Formal Citations
- Altonji, J. G., & Zhu, Z. (2025). Returns to Specific Graduate Degrees: Estimates Using Texas Administrative Records. NBER Working Paper No. 33530. National Bureau of Economic Research. https://doi.org/10.3386/w33530 | https://www.nber.org/papers/w33530
- Altonji, J. G., & Zhu, Z. (2025). Do Graduate Degrees Pay Off? PEER Center at American University. https://www.peer-center.org/research/do-graduate-degrees-pay-off
- Cooper, P. (2026, March). Graduate School Is Riskier Than You Think. American Enterprise Institute. https://www.aei.org/education/graduate-school-is-riskier-than-you-think/
- Fain, P. (2026, March 31). Graduate School Pays Off for Pharmacists, Not Psychologists. Inside Higher Ed. https://www.insidehighered.com
- U.S. Bureau of Labor Statistics. (2024). Occupational Outlook Handbook: Electrical and Electronics Engineers. U.S. Department of Labor. https://www.bls.gov/ooh/architecture-and-engineering/electrical-and-electronics-engineers.htm
- U.S. Bureau of Labor Statistics. (2024). Earnings and Unemployment Rates by Educational Attainment, 2024. U.S. Department of Labor. https://www.bls.gov/emp/chart-unemployment-earnings-education.htm
- Research.com Editorial Team. (2026). Is an Electrical Engineering Master's Degree Worth It? ROI, Salary & Career Impact. Research.com. https://research.com/advice/is-an-electrical-engineering-masters-degree-worth-it-roi-salary-career-impact
- Research.com Editorial Team. (2026). Different Types of Electrical Engineering Degrees and Their Salaries. Research.com. https://research.com/advice/different-types-of-electrical-engineering-degrees-and-their-salaries
- SMU Graduate Articles. (2024). Career Guide: Electrical and Computer Engineer Jobs with a PhD. Southern Methodist University. https://gradarticles.smu.edu/advancing-the-field/career-guide-electrical-and-computer-engineer-jobs-with-a-phd
- ProFellow Editorial Team. (2026). Fully Funded PhD Programs in Electrical Engineering. ProFellow. https://www.profellow.com/fellowships/fully-funded-phd-programs-in-electrical-engineering/
- MIT EECS. (2025). EECS Graduate Funding. Massachusetts Institute of Technology. https://www.eecs.mit.edu/academics/graduate-programs/funding/
- Semiconductor Engineering Staff. (2026, January 29). AI's Impact On Engineering Jobs May Be Different Than Expected. Semiconductor Engineering. https://semiengineering.com/ais-impact-on-engineering-jobs-may-be-different-than-initial-projections/
- MRINetwork. (2025, July 12). AI Meets Electrical Engineering: What It Means for the 2025 Job Market. MRINetwork. https://mrinetwork.com/hiring-talent-strategy/ai-meets-electrical-engineering-what-it-means-for-the-2025-job-market/
- Centricity Search Group. (2025, July 31). Engineering the Future: How AI Is Reshaping Electrical Careers in 2025. Centricity Search Group. https://centricitysearch.com/engineering-the-future-how-ai-is-reshaping-electrical-careers-in-2025/
- Fortune / DNYUZ. (2026, April 4). College Grads in 'AI-Proof' Careers Like Psychology and Education Are Seeing Negative Returns on Their Degrees. https://dnyuz.com/2026/04/04/college-grads-in-ai-proof-careers-like-psychology-and-education-are-seeing-negative-returns-on-their-degrees/
- Amtec Inc. (2025). The Impact of AI and Automation on the Engineering Workforce. Amtec. https://www.amtec.us.com/blog/the-impact-of-ai-and-automation-on-the-engineering-workforce
- Liu, H., Wang, J., & Wang, Y. (2025). Cross-domain fluency and ethics in engineering education: A systematic review. IEEE Access. Retrieved from IEEE.org.
- edX Editorial Team. (2025). Is a Master's in Electrical Engineering Worth It for Your Career and Earnings? edX. https://www.edx.org/resources/is-a-masters-in-electrical-engineering-worth-it
- Grand Canyon University Blog. (2025, December). AI in Electrical Engineering: Evolving Together. GCU. https://www.gcu.edu/blog/engineering-technology/ai-electrical-engineering
- Physics Forums Community. (2019). Value of an engineering PhD versus a Masters — Industry. Physics Forums. https://www.physicsforums.com/threads/value-of-an-engineering-phd-versus-a-masters-industry.973717/
- College Transitions Dataverse. (2026, March). Median Starting Salaries (by Institution): Electrical Engineering. Derived from U.S. Department of Education College Scorecard. https://www.collegetransitions.com/dataverse/median-starting-salaries-electrical-engineering/
- MIT Admissions Office. (2025). Career Development FAQ: Average Starting Salary. Massachusetts Institute of Technology. https://mitadmissions.org/help/faq/career-development/
- Bloomberg News / Georgetown Center on Education and the Workforce. (2024). Best Colleges 2024: After Ivy League, Public Schools Offer Higher ROI. Bloomberg. https://www.bloomberg.com/graphics/2024-college-return-on-investment/
- Chetty, R., et al. / Opportunity Insights. (2023). Diversifying Society's Leaders? The Determinants and Causal Effects of Admission to Highly Selective Private Colleges. Harvard University. Reported in CNBC, August 2023. https://www.cnbc.com/2023/08/02/is-an-ivy-league-degree-worth-it-report-finds-advantages-beyond-pay.html
- Delta Wye Electric. (2025, September 26). Top Electrical Engineering Schools Ranked for 2025. Delta Wye. https://deltawye.com/top-electrical-engineering-schools/
Comments
Post a Comment