When to Job-Hop and When to Stay: The 3-Year Data on Both Paths

Somewhere along the way, “job-hopping is bad” flipped into “staying is for suckers,” and both eras were wrong for the same reason: they turned a market condition into a moral rule. The honest answer to “should I hop or stay?” is a number that changes with the labor market plus a handful of facts about your specific seat. Let’s get the real numbers on the table, run the three-year math both ways, and build a decision rule you can reuse for the rest of your career — because you’ll face this question every two to four years whether you like it or not.

What the data actually says in 2026

The cleanest ongoing measurement of the hop-versus-stay question is the Federal Reserve Bank of Atlanta’s Wage Growth Tracker, which follows the same individuals’ wages through time using Census survey microdata. As of July 2026 it shows median wage growth of 4.4% for job switchers versus 3.6% for job stayers — a 0.8-point switching premium. That’s a real edge, but notice two things folklore misses. First, it’s the median switcher; roughly half do worse than that, and some switch into pay cuts. Second, the premium breathes with the market: it blew out to multi-point spreads during the 2021–22 frenzy, then compressed so hard in stretches of 2025 that stayers briefly matched or beat switchers. A number that can invert is a tide you check, not a law you obey.

Meanwhile, staying is what most people actually do most of the time. The BLS’s latest employee tenure data puts median tenure at 3.9 years overall — 3.6 for women, and just 2.7 for workers 25–34. Read that age split honestly: changing jobs every ~3 years through your late twenties and early thirties isn’t deviant behavior that requires explaining in interviews; it’s the statistical center of your cohort.

The three-year table, both paths

Folklore compares one raise to one offer. The real comparison is three years of compounding, because the hop’s premium arrives once while raise trajectories keep running. Take $90,000 today, and use honest 2026-ish numbers: staying means roughly 3.6% annual growth if you’re treated normally; hopping means a realistic 10% jump (well above the median switcher — this is the good offer case), then normal growth after.

Three readings of that table, in increasing order of importance:

1. The good hop wins big. About $29,000 more over three years than passively staying — and every future raise percentages off the higher base, so the gap keeps widening after the table ends.
2. The promoted stay closes most of the gap. A real internal promotion — not a title pat, a documented case landing an 8–10% bump — recovers roughly two-thirds of the hop’s advantage without the switching costs.
3. The passive stay is the only clearly losing row. Which reframes the entire question. It was never really “hop vs. stay.” It’s “hop vs. stay-and-push” — and the row you should fear isn’t either of those; it’s drifting in the first column while telling yourself you’re in the third.

What the wage data can’t see

The tracker measures wages. Careers run on more variables, and the big ones argue in both directions:

Costs of hopping the data hides: unvested equity and bonuses left behind (check your vesting dates before interviewing, not after); the last-in position if layoffs come — formal seniority systems are rare, but “last hired” is a real informal exposure; the productivity trough of relearning an organization; and the compounding value of deep context — the person who’s seen three planning cycles catches problems the brilliant newcomer can’t. Also, bluntly: benefits reset. Waiting periods, leave-eligibility clocks, PTO accrual — a hop can zero clocks you didn’t know you were running.

Costs of staying the data also hides: salary compression, where your 3%-a-year drifts below what new hires are offered for your own role (this is precisely the thing a market-rate check every six months catches); skill narrowing, where you become excellent at this company rather than your field; and network stagnation. The resume-pattern worry cuts both ways too — a string of sub-18-month exits genuinely does raise hiring-manager eyebrows, but so does a decade in one seat with one title.

Reading your own situation against the data

Aggregate numbers earn their keep only when you localize them, so here’s how to translate each national figure into a personal one:

Your switcher premium isn’t 4.4% — it’s whatever offers say it is. The tracker reports the median across every industry and seniority; your personal premium might be triple that (underpaid, hot specialty) or negative (top of band, cooling field). The only instrument that measures it is the market itself: two or three exploratory interviews a year, run ethically and without burning your cover, tell you your actual number. Treat recruiter screens as free data collection — you’re allowed to conclude “staying is correct” from them, and telling a recruiter “this confirmed I’m well-positioned where I am” is a perfectly respectable end to a process.

Your tenure baseline is your industry’s, not the nation’s. The 3.9-year median averages federal clerks and line cooks. Tech, media, and agency worlds run materially shorter cycles; government, healthcare systems, and academia run longer. Hiring managers calibrate “job-hopper” against their field’s rhythm — three two-year stints read differently at a startup than at a utility. Look at the actual tenures of people two levels above you in the org you want to join; that’s the norm you’re being measured against.

Watch the sequencing more than the frequency. The pattern that damages resumes isn’t short stays — it’s short stays without visible progression. Three three-year chapters, each with a title or scope jump, tell a story of demand. Three three-year chapters at the same level tell a story of restlessness. If you’re going to hop, hop upward or wider, and make the progression legible in the first line of each role’s description.

And price the transition costs before, not after. Sum your unvested value, your bonus timing, and the benefits clocks that reset. If leaving in March costs $9,000 more than leaving in July, that’s not a reason to stay — it’s a reason to time the going. Companies time their generosity; you’re allowed to time your exits.

A decision rule that doesn’t require forecasting

You can’t predict the market, so anchor the decision to things you can verify. Once a year — this slots naturally into the annual review you run on yourself — answer four questions with evidence, not feelings:

1. Am I at market? Verified range, three sources, dated. More than ~10% below with no correction in motion → the market is paying a bonus for your resume; go collect it.
2. Is my growth curve here real? A promotion or scope jump with a date attached in the last 18 months, or credibly scheduled in the next 12? Written down, or wishful?
3. Is my skill set compounding or narrowing? Would a stranger pay for what I learned this year?
4. What’s the tracker saying? When the switcher premium is fat, external moves are on sale. When it’s thin — like 2026’s 0.8 points — internal pushes and tight-year raise plays compete surprisingly well, and a hop needs to clear a higher bar of role quality, not just salary.

Score it plainly: at market + real growth + compounding skills = stay and push, whatever the tracker says. Under market + no dated growth + narrowing = start interviewing, whatever the tracker says. The tracker breaks ties and sets your asking price.

My own record, for honesty’s sake: I’ve hopped twice and stayed-and-pushed twice, and my worst decision was a stay — year six at one employer, where loyalty had quietly become inertia and the raise that “was coming” arrived two years late and three points short. The data would have flagged it in year four: I was under market, nothing had a date on it, and I didn’t check. The table above isn’t investment advice or a crystal ball — it’s a mirror. Run your own numbers in it once a year, and neither folklore — the old loyalty kind or the new hustle kind — gets to make this decision for you.