Pay rises by leaving

Why does changing employer often pay better than doing excellent work where you already are?

Someone does excellent work for four years and receives modest annual increases. They move to a similar job elsewhere and their pay jumps by considerably more than four years of raises delivered.

The employer has, in effect, paid a stranger more than they would pay someone whose performance they had directly observed for four years. That looks like a straightforward mistake, and it is common enough that it cannot be a mistake — so it must be the outcome of a system doing something other than what it appears to be doing.

Start with the two decisions and notice that they are made in different places, on different information, against different constraints.

The internal raise. It is made inside a structure. There is a budget, usually a percentage of payroll allocated before anyone knows what individuals deserve. There is a band for the role. There is an expectation that increases across a team will not vary wildly, since they will be compared. And critically, the manager awarding it usually cannot move someone substantially out of band without approvals that cost time and political capital.

The external offer. It is made in a market. The hiring firm is not comparing this person to their existing team's raises; they are comparing the cost of hiring against the cost of the role staying empty. That cost includes the work not being done, the recruitment already spent, and the risk of the search failing. They are bidding against other employers, not against a payroll budget.

So the same person is priced by two mechanisms: one asking what increase is defensible within a structure, the other what it will take to get them.

Now the second element, and it is the one that makes the gap persist rather than closing. Internal pay is sticky downward and visible sideways. An employer who matches the market for one person creates a comparison for everyone in the same band. If the raise is given, colleagues learn what the band really permits, and the cost is not one salary but a re-levelling across a group. If it is refused, the cost is one departure. Facing those two costs, the employer frequently prefers the departure — which is not stupidity but arithmetic.

That is why counter-offers appear only when someone is leaving, and why they are often generous when they do. The threat of departure converts the question from "what does the structure permit" to "what does replacement cost", which is the market question again. The employee has not become more valuable; they have forced the employer to price them the other way.

Third, information. An employer who has watched someone for four years knows far more than a hiring firm does — and that knowledge does not push pay up. The incumbent's outside options are unknown unless demonstrated, and pay tracks bargaining position, which is about alternatives rather than quality. Excellent work improves the case for a raise; an offer changes what the employer stands to lose.

And the last term compounds the rest: raises are usually percentages of a current salary. Someone who starts low stays low, because every subsequent increase is scaled to a base that was set at hiring. A move resets the base rather than incrementing it — which is why the effect is largest for people who were underpaid at entry, and why it is entirely possible for pay to become inverted, with new hires above longer-serving colleagues doing the same job.

This is a real pattern but not a universal law, and the honest statement is conditional. The gap is largest in occupations where skills transfer easily between employers, where hiring is competitive, and where internal structures are rigid. It is much smaller where firm-specific knowledge is valuable, where internal promotion ladders are genuinely used, where pay is set by collective agreement or a published scale, and in slack labour markets where employers are not bidding against each other. Anyone advising universal job-hopping is overstating it.

The costs of moving are real and usually left out. A move forfeits standing, relationships and knowledge of how the place works; it may reset probation, notice and accrual; and it carries a real chance the new job is worse than advertised. "Raise versus offer" compares two salary numbers, and the decision is not two numbers.

Accepting a counter-offer is a distinct decision with its own problems. It solves the pay gap and tells the employer this person will look elsewhere. Evidence on how often such acceptances end badly is mostly anecdotal and much of it comes from recruiters, who are not disinterested. I would treat confident figures with suspicion, while noting the structural point stands: the counter-offer proves the money existed.

The employer's behaviour is rational, and calling it short-sighted misses the mechanism. Matching one person's market value can genuinely cost more than replacing them, once the comparisons it creates are counted. That is a constraint rather than an excuse, and it explains why the pattern survives in well-run organisations. It also points at the fix, which is structural rather than exhortative: periodic market benchmarking of the existing payroll, published bands, and internal equity maintained deliberately rather than discovered when someone resigns.

The falsification test. If the mechanism is a structural gap between internally-set and market-set pay, then the effect should be largest where internal pay is most constrained and skills most transferable — and organisations that benchmark existing staff to market annually should show a smaller move premium than those that do not. If movers gained the same premium regardless of how the employer set internal pay, the account would be wrong and the premium would have to be explained by selection: better people being the ones who move.