At a glance
University compensation and talent acquisition teams are benchmarking academic pay against peers while budgets tighten and discipline-level market rates shift. A practical look at peer group design, total value comparison and what transparency rules now demand.
Compensation teams went into this year's cycle holding a number that looked defensible on paper. Average full-time faculty salaries in the United States rose 2.3 percent between fall 2024 and fall 2025, according to preliminary results from the American Association of University Professors' annual Faculty Compensation Survey. Consumer prices rose 2.7 percent over the same period. Real pay fell 0.4 percent, and average full-time salaries sit roughly 9.5 percent below where they were in fall 2019 once inflation is stripped out.
The College and University Professional Association for Human Resources (CUPA-HR) tells the same story from a different angle. In its 2025-26 workforce data, tenure-track faculty received a median increase of 1.8 percent — the smallest of any employee group, and the fourth year running they have finished last. In real terms, tenure-track salaries are 11.7 percent below 2019-20 levels. Across the ten years CUPA-HR has tracked, tenure-track increases have never once beaten inflation.
| Employee group | Median increase, 2025-26 | Real change vs 2019-20 |
|---|---|---|
| Non-exempt staff | 3.0% | minus 0.9% |
| Administrators | 2.9% | not reported in the summary |
| Professionals | 2.8% | not reported in the summary |
| Non-tenure-track faculty | 2.0% | minus 6.8% |
| Tenure-track faculty | 1.8% | minus 11.7% |
| Consumer price index | 2.7% | reference point |
None of that answers the question a hiring manager actually has on a Tuesday afternoon: what do we offer an associate professor of accounting at a mid-sized public university, in a city where living costs have climbed faster than the pay spine? A national median is a starting point and nothing more. The work of academic salary benchmarking is building the comparator that answers that question, then being able to explain in writing why the answer is fair.
Peer groups decide the answer before the data arrives
Ask a compensation office which peers it uses and you learn more about the institution than any salary table will tell you. The peer group sets the target, which is why most institutions settle on a fixed set and defend it for years.
Iowa State University works from three groups: ten peer institutions designated by the Iowa Board of Regents, a 29-institution set submitted to the federal Integrated Postsecondary Education Data System (IPEDS), and a group drawn from institutions in the fourth quartile for total expenses. Virginia Tech follows an ordered sequence when data are thin — the group set by the State Council of Higher Education for Virginia first, then public land-grant universities, then doctoral institutions with total expenses above $900 million. The University of Central Florida reviews its peers formally every two years, with the president, provost, chief analytics officer and chief human resources officer all in the room.
The details matter more than the labels. Wake Forest University applies local cost-of-living data alongside national academic peers, because national benchmarks alone would push its targets too high for its region. UCF does something similar from the opposite direction: sitting in the Orlando metro area, it often finds national academic peers too low to be competitive and pulls in local market data instead. Virginia Tech keeps an aspirational group on hand, private doctoral institutions nationwide, and reaches for it when a specific role has chronic turnover.
There is a governance dimension too. A 2026 study in the Journal of Information Systems Applied Research criticised the common practice of building separate peer groups for base salaries and for high-demand disciplines, arguing that subjective filters and post-hoc additions such as aspirational institutions invite equity challenges. The authors proposed a single peer group built from stakeholder-identified variables such as enrolment, finances and student outcomes, deliberately excluding salary data. The principle holds either way: if the peer group can be redrawn after the numbers arrive, nobody will trust the numbers.
Discipline is the pay band that actually moves
Institution-wide averages hide the only variation that drives offers. CUPA-HR's twenty-year analysis of faculty pay by discipline found business faculty were the highest-paid discipline for nine consecutive years and recorded the largest median salary increase of any field, 66.2 percent, between 2003-04 and 2023-24. Health professions more than doubled its faculty headcount over the same period. English language and literature ranked among the ten lowest-paid disciplines every year and has roughly 40 percent of its faculty off the tenure track.
The finding that should stop a compensation lead cold is this one: no discipline's median salary kept pace with inflation across those two decades. Not one. A benchmark that says your department is at market may still be describing a slow real-terms cut, and telling a candidate that is very different from telling them they are being paid competitively. The CUPA-HR report sets out the discipline-level numbers in full.
Then there is the internal damage external benchmarking leaves behind. Salary compression is when junior and senior colleagues end up too close together; inversion is when the new hire earns more than the people who will mentor them. Association to Advance Collegiate Schools of Business (AACSB) data cited in a study of the problem showed newly hired assistant professors in business arriving at 2 percent above incumbent assistants, and new doctorates hired at rates 10 percent above current assistants, 8 percent above current associates and only about 7 percent below current full professors. By 2008 the gap between full and associate professors had narrowed to 13 percent while the gap between associates and assistants was just 2 percent. The market sets the entry price. The internal ladder sets everything above it, usually years out of date.
Jacqueline Bichsel, CUPA-HR's associate vice president of research, put the practical consequence plainly: for many faculty the only chance at a significant rise is promotion, and even those events are now producing smaller increases than expected, as she told Inside Higher Ed. A benchmark that ignores the promotion step is not measuring a career. It is measuring a starting line.
The band is the smallest part of the offer
Where the pay spine is negotiated collectively, the headline percentage is only one lever. In the United Kingdom, the Universities and Colleges Employers Association (UCEA) made a final offer of 1.4 percent from 1 August 2025 for the institutions it represents, alongside removal of the bottom two spine points, and noted that around half of staff would also be eligible for an increment worth up to 3 percent. That increment, not the headline, is what a new starter experiences in year one. The full 2025-26 round is documented by UCEA, including the union claim of RPI plus 3.5 percent or a flat £2,500, and the joint work on spine reform that followed.
In Australia the same logic plays out through enterprise agreements, which lock in classification rates, loadings and allowances for the life of the agreement. RMIT University's 2024 agreement runs to a nominal expiry of 31 December 2026 and defines full rate of pay to include all loadings and allowances. Benchmarking there is really bargaining preparation: the comparison you build this year becomes the evidence you table two years from now.
Geography is the adjustment most often done badly. Tufts University's faculty market analysis, run with Willis Towers Watson, drew on CUPA-HR's national faculty survey using custom peer groups and then applied a 14 percent geographic cost-of-labor differential for Boston, sourced from ERI survey data as of 1 April 2025. Tufts also said plainly that most institutions phase the resulting salary changes over three to five years. The methodology is published in full, which is rather the point.
And the part of the offer that never makes it into the pay band: pensions, leave, contract security, teaching load. Among 555 institutions reporting starting rates for part-time faculty in the AAUP's survey, the median was $3,121 per course section. Only 32.7 percent contributed to retirement for some or all of those faculty, and 30.6 percent contributed to medical cover. Candidates read those numbers before they read the salary.
Transparency is turning the benchmark into a published document
For institutions operating in the European Union, benchmarking has stopped being an internal exercise. Directive (EU) 2023/970 on pay transparency carried a transposition deadline of 7 June 2026. It requires employers to give applicants the starting salary or a salary range before the interview, and forbids asking about pay history. It also requires pay structures built on objective, gender-neutral criteria assessed against four factors: skills, responsibility, effort and working conditions.
The European Commission and the European Institute for Gender Equality published detailed guidance in March 2026, including a default points framework that weights skills at 40 percent, responsibility at 35 percent, effort at 15 percent and working conditions at 10 percent, for a maximum of 1,200 points. Each 10 percent band of the total score is meant to define a comparable category of workers. Lewis Silkin's analysis sets out how the framework is expected to operate in practice.
Reporting starts at 7 June 2027 for employers with 250 or more staff, on an annual cycle, with smaller employers phased in to 2031. Where a reported gap of 5 percent or more in any category cannot be justified on objective grounds and is not fixed within six months, the employer must run a joint pay assessment with worker representatives. The burden of proof sits with the employer.
Read that alongside your peer benchmarking and the risk is obvious. If the answer to why one role pays less than another is that the peers pay it that way, the peer data had better be current, documented and applied consistently. Benchmarking data is now evidence.
Selling the role when salary is not the ceiling
Most institutions cannot win a bidding war for a clinical psychologist or a machine-learning lecturer, and pretending otherwise wastes everyone's time. What they can do is stop hiding the parts of the offer candidates genuinely weigh: whether the contract is continuing, how progression works, what the teaching load is, whether sessional staff have a conversion route, and what the pension is worth.
Two changes do most of the work. Publish the band, with a stated midpoint and a clear account of what moves someone through it, which turns a negotiation into a conversation. Then name the ceiling honestly and say what sits beside it. A candidate comparing a 12-month contract at a lower salary against a continuing post with a documented workload model is not comparing salaries at all.
The pattern on the candidate side is consistent: ads that withhold the range get fewer serious applicants, and the ones who do apply arrive with a number in mind that nobody has corrected. There is more on this in how employer branding works in higher education recruitment, and it applies whether you are hiring one lecturer or forty.
A benchmarking checklist that survives a scrutiny panel
None of this is complicated. It is just frequently skipped.
- Fix the peer group in writing before you pull any data, use the same set for every role in the exercise, and publish it.
- Record the fallback sequence for roles where data are thin — narrow group first, then broader, then local market. Virginia Tech's written methodology is the model.
- Match on discipline code, rank and tenure status rather than on department name. The University of Louisville's study used Classification of Instructional Programs (CIP) codes, and fell back to professional association surveys for dentistry, medicine, nursing and librarians where the national faculty survey lacked the detail.
- Age every data set to your decision date using a documented cost-of-labor adjustment, not cost of living.
- Benchmark total cash, not base only: market supplements, clinical loadings, stipends and sessional rates all change the comparison.
- Apply a geographic differential that is sourced and dated, and say which city it refers to.
- Run a compression and inversion check before you make the offer, not after the incumbent complains.
- Report the share of your population within 10 percent of market and within 5 percent. A consultancy rule of thumb applied in the University of Central Arkansas study treats within 10 percent as competitive, within 5 percent as highly competitive, below 90 percent as possible misalignment and below 85 percent as significant misalignment.
- Test the result for a 5 percent gap by category, with worker representatives in the room, before a regulator asks.
- Write the method down in plain language and put it where faculty can read it.
What to do on Monday
Pick the peer group, write down why it is the right one, and give the same set to every hiring manager in the institution. Then take the roles you have been struggling to fill and check what your own data says against the market for the discipline, at that rank, in that city. If the gap is wider than 10 percent, that is not a recruitment problem. It is a budget decision somebody has been avoiding, and no job ad will fix it.
Photo by Sasun Bughdaryan on Unsplash
- higher-education-recruitment
- faculty-pay
- cupa-hr
- pay-transparency
- pay-bands

