The career change tax: what a switch costs, measured

The folk price of a career change is a pay cut. Advice columns quote 10 to 20 percent, framed as tuition for the fresh start, and the market data seems to back the mood: by Q4 2025, only 56 percent of US new hires landed higher pay than their last role, down from 70 percent two years earlier (ZipRecruiter, via The Hill). The number everyone repeats is real. It is also about job changes in general, not career changes in particular, and nobody quoting it can tell you what switching from your occupation to a specific other one costs.

That second question is computable, so we computed it. Our pipeline holds live US postings with stated salaries for 163 occupations, at least 15 salary-stated listings each, and a skill graph of which occupations are measurably adjacent to which. Cross the two and you get 966 transitions with a salary median on both ends: the advertised price of the job you have next to the advertised price of the job you could move into. The gap between them is the career change tax, one transition at a time.

The tax is nearly zero. The moves are not.

The median tax across all 966 measured transitions is 2.3 percent, a long way from the folk 10 to 20, which on a $90,000 salary is about what a bad parking habit costs a year. And the distribution splits almost evenly. 52.5 percentof transitions end in an occupation whose advertised median is higher than the origin’s. A negative tax turns out to be the slight majority of the graph.

What the averages hide is the amplitude. The median switch moves advertised pay by 22.8 percent in one direction or the other. Career changes are not cheap; they are volatile, and the folk advice mistakes volatility for cost. The people who took the famous 20 percent cut and the people who quietly gained 20 percent are both inside that number. The difference between them is not luck. It is which wire they followed, and the wires are visible before you jump.

First, the fake raises

Run this analysis naively and the top of the raise table is a parade of mirages. Clinical research coordinator to physician shows a 223 percent raise. Sales representative to lawyer, 158. Event planner to pharmacist, 139. Every one of these is real arithmetic and fake advice, because the destination sits behind a license and, in the physician’s case, most of a decade of school. Our graph marks license-gated occupations, so 132 of the 966 transitions are excluded from every table below. What remains is the open market: moves where the gate is a hiring manager, not a bar exam.

The overlap paradox

Here is the finding we did not expect. Split the open transitions by skill overlap, the share of the destination’s demanded skills your occupation already covers, and the comfortable half underperforms. Above-median overlap: a median tax of 8.5 percent. Below-median overlap: a median raise of 8.5 percent.

The mechanism is unglamorous once you see it. Skills cluster inside pay bands, so the occupations most similar to yours tend to be priced like yours or slightly below; support roles neighbor support roles, and coordination jobs neighbor coordination jobs. The raises live one ring further out, where the overlap is real but partial, and the missing skills are the reason the destination pays more in the first place. Familiarity is comfort. It is not upside.

Where the raises hide

The useful quadrant is transitions with meaningful overlap and a positive gap: enough carryover that the interview is winnable, enough distance that the pay band changes. A sample from the top of that quadrant, all license-free, all with live listings on the board:

A gap is not an offer. Customer support to operations manager is a real wire in the graph and still a competitive interview against people already carrying the title. What the number says is narrower and more useful: the market prices these two skill sets 108 percent apart, and part of the bridge is already built. Whether the remaining part takes six months of visible work or two years is written in the skill gap, not the salary gap.

The same wire, billed in one direction

Because adjacency is mutual and pay is not, the graph holds 38 pairsof occupations measured in both directions where the identical wire pays going one way and charges coming back. Content strategist to developer advocate runs +106 percent; the return trip runs a 52 percent cut. Database administrator to software engineer, +82 out, 45 back. None of this is mysterious, which is what makes it useful. The tax belongs to the direction, not to the act of switching, and direction is the one variable entirely in the switcher’s control.

Where this leaves you

Three numbers decide a move, and all three are printed before you apply. The overlap says whether the interview is winnable. The salary gap says what the move pays, in either direction. The skill gap names what to build to cross it. If you hold a title and want the priced routes out of it, the instrument computes your own row of this table; if you want to see who is hiring on the other end today, the board carries the live listings, salary shown where the employer states it. And if the move you are weighing is a deliberate downshift, physician to scribe runs an 84 percent cut and people take it on purpose, the same numbers price the exit honestly too.