Is salary decided at the end of the interview process?
People often treat salary negotiation as the decisive moment in a job switch: what number to ask for, how much of an increase to request, and when to discuss the offer.
A large Japanese placement dataset suggests that part of the outcome may be shaped much earlier.
Indeed Recruit Partners Research Center analyzed 313,485 successful Recruit Agent job switches from April 2022 through March 2026 using Oaxaca-Blinder decomposition.
¥604,000 of a ¥764,000 gap was not simply a same-job pay difference
The average post-switch gender salary gap was ¥764,000, with women's pay at 84.6 when men's pay is indexed to 100.
The study attributed ¥604,000, or 79.2%, to composition differences: differences in the jobs and observed characteristics associated with where men and women moved.
The remaining ¥159,000, or 20.8%, was an evaluation/treatment component left after aligning observed factors such as age, occupation, industry and the salary advertised for the job.
The largest single component was the salary posted for the job itself
Within the composition component, the biggest factor was the advertised salary of the job where the switch was completed, calculated from the midpoint of the stated salary range.
That variable alone accounted for ¥376,000 of the overall gap. The jobs secured by men carried advertised salaries roughly ¥770,000 higher on average than the jobs secured by women.
A large part of the gap therefore appears before final offer negotiation: people are landing in different salary bands of the job market.
Access to higher-priced jobs is itself shaped by prior career history
Other composition factors widening the gap included age at ¥124,000, current management status at ¥40,000, occupational experience at ¥39,000 and current employment status at ¥29,000.
In the dataset, men were about four times as likely as women to be managers in their current jobs, while women were more than twice as likely to be in non-regular employment.
This does not mean a person can simply apply to a higher-paying posting and raise salary. Management responsibility, occupational experience and employment history help determine which postings are realistically accessible.
Career Experience → Accessible Jobs → Posted Salary can shape the next offer
For an individual, the question may need to move upstream from 'How much should I negotiate?' to 'What salary band of jobs can my current experience actually get me into next?'
Responsibility, management exposure and specialist experience can change the set of jobs a person can credibly enter, and the posted salary range of those jobs then anchors the final offer.
Banseog reads this as Career Experience → Accessible Jobs → Posted Salary Range → Final Compensation. Negotiation matters, but the price of the accessible market may already have been shaped before negotiation begins.
At the higher end of the salary market, the location of the gap changes again
The researchers repeated the decomposition across four pre-switch salary groups, Q1 through Q4.
In the lowest group, Q1, the post-switch gap was ¥125,000 and 67.1% was composition difference. In the highest group, Q4, the gap widened to ¥458,000 and 61.5% was the evaluation/treatment component.
The aggregate market is dominated by differences in where people land, but at higher salary levels the residual component becomes more prominent. The bottleneck can shift as a career moves up-market.
The ¥159,000 residual should not be read automatically as discrimination
The study explicitly warns against equating the ¥159,000 evaluation/treatment component with discrimination.
The dataset cannot observe every relevant factor, including job grade, working hours and details of salary negotiation at hiring. Oaxaca-Blinder decomposition separates the observed gap into explained and residual components; it does not identify a single cause for the residual.
The sample is also not a random sample of all Japanese workers. It consists of successful job switches through Recruit Agent, so the percentages should not be generalized mechanically to the entire labor market.
The career question is not only today's salary, but the price band of tomorrow's jobs
A role can have economic value beyond the salary it pays today if it adds management responsibility, scarce occupational evidence or experience that opens a different job market later.
That does not make every lower-paid developmental role worthwhile. It means salary decisions can be evaluated on two horizons: what the job pays now and what range of jobs the experience makes accessible next.
The dataset makes that upstream part of compensation formation unusually visible.
BANSEOG VIEW | The salary label can start forming before the offer stage
The striking result in 313,485 placements is not only the ¥764,000 average gender gap, but where much of that gap appears.
Nearly four-fifths was associated with composition differences, and the largest single factor was the advertised salary of the job that was actually secured.
For a career decision, the practical question is therefore not only how well to negotiate the next offer. It is whether today's experience is expanding the salary band of jobs you can realistically enter next.
BANSEOG VIEW
Banseog View — Career Experience → Accessible Jobs → Posted Salary → Final Compensation
Post-switch salary can diverge before final negotiation through differences in the jobs people can access.
Composition accounted for 79.2% of the overall gap, but in the highest pre-switch salary quartile the evaluation/treatment component accounted for 61.5%.
Career decisions can be evaluated not only by current pay, but also by whether the experience expands the price band of jobs accessible next.
SOURCES
Primary sources and references
- Indeed Recruit Partners Research Center — Where does the gender gap in post-switch wages come from?
October 1, 2026. Confirms the 313,485 successful Recruit Agent placements, ¥764k gap, 79.2%/¥604k composition component, 20.8%/¥159k evaluation-treatment component, ¥376k posted-salary contribution, Q1/Q4 decomposition and interpretation limits.
The sample consists of 313,485 successful Recruit Agent job switches from April 2022 through March 2026 and is not a random sample of all workers in Japan. The evaluation/treatment residual can contain unobserved factors such as grade, working hours and hiring-stage salary negotiation and should not be equated automatically with discrimination. Career Experience → Accessible Jobs → Posted Salary Range → Final Compensation is Banseog's decision-oriented interpretation, not a causal model asserted by the study.