Career change at 50: what a certification can and cannot do about age

What the callback numbers actually show

David Neumark, Ian Burn and Patrick Button sent over 40,000 job applications to real vacancies and varied only the applicant's age. For administrative jobs, which made up about 61 percent of the eligible ads, the callback rate was 14.4 percent for applicants aged 29 to 31, 10.3 percent for applicants aged 49 to 51, and 7.6 percent for applicants aged 64 to 66. Combining all four occupations they studied, callback rates were about 18 percent lower for the middle-aged group and about 35 percent lower for the oldest group. The evidence was strongest and most robust for women. For men, the authors found considerably less evidence once they corrected for biases in earlier study designs, so a man reading this should not assume the same size of effect applies to him. Two things follow. First, the penalty appears at the callback stage, before any conversation about what you can do. Second, the resumes in this study were otherwise matched. The older applicants were not less qualified. They were older. That is why the honest starting point for anyone planning a career change at 50 is that the problem being solved by a certification and the problem showing up in your inbox may be two different problems.

The six-second resume number, and where it came from

Somebody will tell you that recruiters spend six seconds on a resume, and that this is why you are being screened out at 52. Follow the number back. It comes from a press release issued on 21 March 2012 by TheLadders, a paid job-search company that also sold resume-writing services, announcing an in-house eye-tracking study of what the release calls 'dozens of professional recruiters' over a ten-week period. The write-up was a white paper posted on the company's own blog. It was not published in a journal, it was not peer reviewed, no sample size is stated in the announcement, and the company selling the finding also sold the fix for it. This page therefore puts no number on how long anyone looks at your resume, because no source used here measures it. The reason to mention it at all is that the six-second claim quietly pushes you toward paying for cosmetic resume work, when the experiment that did measure age in hiring found the gap survived changes to what the resume said.

The study already tested whether skills fix it

This is the part that matters for anyone about to spend money. The same researchers built skill variation into the resumes, precisely because the most common explanation for age discrimination is that employers assume older applicants have weaker computer skills. They could then measure what happens when a resume lists computer skills. For administrative jobs, listing computer skills did help older applicants more than younger ones, which is consistent with employers assuming older candidates lack them when nothing is stated. But the size of that effect was 0.034 against an age gap of 0.09 in the same model. The authors wrote that this 'by no means accounts for the age difference'. Read plainly: adding a skills signal recovered roughly a third of the gap in the one occupation where it worked at all, and did nothing clear in sales. They also found no significant difference in callbacks between resumes showing low experience and resumes showing experience commensurate with age. Rewriting the resume to look younger did not fix it either. So a certification is a partial instrument against a partial cause. If a vendor tells you a certificate will make your age irrelevant, that claim has been tested and it failed.

The law protects employees more than it protects applicants

The Age Discrimination in Employment Act covers people aged 40 and over, and applies to employers with twenty or more employees for each working day in each of twenty or more calendar weeks in the current or preceding year. Small employers are outside it entirely. There is a second limit almost nobody explains. Discrimination claims come in two forms: disparate treatment, meaning someone acted on your age deliberately, and disparate impact, meaning a neutral rule fell harder on older people. In January 2019 the Seventh Circuit, sitting en banc in Kleber v. CareFusion Corporation, held that the plain language of section 4(a)(2) of the ADEA protects employees from disparate impact age discrimination but does not extend that protection to outside job applicants. Dale Kleber was 58. The job asked for 3 to 7 years and no more than 7 years of legal experience. He had more. A 29-year-old was hired. His disparate impact claim was dismissed and the dismissal was affirmed. Practically, this means an experience cap that screens out everyone over about 40 is much harder to challenge as an outside applicant than as a current employee. Deliberate treatment claims under section 4(a)(1) still cover applicants, but they require showing intent, which is exactly what a screening algorithm and a polite rejection email are designed not to reveal.

What the reemployment data says about how long this takes

The Bureau of Labor Statistics surveys displaced workers every two years. Its January 2026 survey covered 3.3 million people who lost jobs they had held three or more years between January 2023 and December 2025. By January 2026, 72.9 percent of those aged 25 to 54 were reemployed. For those aged 55 to 64 the figure was 57.3 percent. For those aged 65 and over it was 38.6 percent, with 48.9 percent out of the labour force entirely. Separately, in July 2026 the average duration of unemployment for people aged 55 to 64 was 33.1 weeks, against 23.7 weeks for all unemployed people aged 16 and over. Median duration was much closer across ages, which tells you the average is being pulled up by a minority of very long spells rather than by everyone taking a bit longer. That distinction matters for planning. Most older job seekers are not stuck. A significant minority get stuck badly. If you are already six or seven months in, you are in the tail, and the response to being in the tail is different from the response to being one month in. It usually means widening the target rather than deepening the credential.

The pay cut is now the bigger risk

In January 2024, about 62 percent of reemployed full-time displaced workers who reported earnings on the lost job were earning as much or more than before. In January 2026 that share was about 49 percent. Fewer than half. This is the number to hold in mind when you weigh a certification against simply taking the first reasonable offer, because it reframes the question. If the likely landing spot pays less than the old job regardless, then paying for a credential that raises the offer by nothing is a bad trade, and spending the same to enter a field with a different pay ceiling may be a good one. Work out the second number before you pay. Look at actual posted salaries in your area for the role the certification qualifies you for, not national averages and not the certifying body's own salary survey. If the posted range is below what you can earn by taking an adjacent job you are already qualified for, the certification is a cost with no return attached.

When the credential is the job rather than a signal

There is a category of work where a credential is not a persuasion device. It is a legal gate, and passing it makes you eligible in a way an employer cannot second-guess. Real estate salesperson licences, insurance producer licences, pharmacy technician certification and enrolled agent status all work this way. The IRS grants enrolled agent status to anyone who passes a three-part test covering individual and business returns, with no degree requirement anywhere in the path, and enrolled agents have the same unlimited practice rights before the IRS as attorneys and CPAs. Maintaining it takes 72 hours of continuing education every three years. The Pharmacy Technician Certification Board charges 129 dollars to apply and sit the exam, and offers a second eligibility pathway of 500 hours of pharmacy technician work experience for people who did not attend a recognised training programme. In both cases the decision maker is a scoring rule, not a hiring manager forming an impression of you from a document. That is the structural reason these credentials are worth more to a 52-year-old than a generic course certificate is. The trade is real though: licensed entry roles often start low, and commission-based licensed work can pay nothing for months.

Two cheaper moves to try before you buy anything

The first is to change who reads the application. The callback penalty is measured on cold applications to posted vacancies. It is not a measurement of what happens when a former colleague forwards your name to a hiring manager. Nothing in the evidence says a referral carries the same penalty, and referrals cost nothing but a list and a fortnight of messages. Work through everyone you have managed, everyone who has managed you, and every vendor and client contact from the last decade. The second is to remove the lawful, unfalsifiable objection. 'Skills may be out of date' is a rejection reason that never has to be written down and cannot be argued with. A current certification in the specific tool the job names converts that into a checkable fact. This is the narrow benefit that is genuinely real, and it is worth paying for only when a named requirement in the posting matches the credential exactly. If you are applying to fifty jobs and eight of them name the same certification in the requirements section, that is a signal worth acting on. If none of them name any certification, buying one is guessing.

How to decide this week

Take twenty postings you would genuinely accept. Copy the requirements section of each into one document. Count how many name a specific credential as a requirement rather than a preference. If the answer is zero or one, do not buy a certification. Your bottleneck is reach, not qualification, and the money is better spent on three months of runway while you work referrals. If the answer is five or more and they name the same credential, price it including the exam fee, retake fee, any required training hours and the renewal cycle, then check whether the roles pay more than what you can get without it. If the answer is that the credential is legally required to do the work at all, the calculation changes entirely: you are not buying a signal, you are buying eligibility, and the only question left is whether the entry-level pay in that field covers your bills while you build up. Write the answer down before you pay anyone. The most expensive outcome here is a half-finished credential in a field you had not checked the pay for.

What this does not fix

This post does not claim that a certification raises callback rates for older applicants. No study reviewed here measures that, and the one experiment that tested a skills signal found it recovered only part of the age gap in a single occupation. Nothing here is a prediction that you will be hired faster after paying an exam fee.

Where an exam fits

• EA (Enrolled Agent), if you want practice rights granted by passing a test rather than a signal a manager can ignore, and can wait out the study time
• PTCB (Pharmacy Technician Certification), if you need a healthcare entry point, can live on a technician wage, and already have 500 hours of pharmacy work or access to a recognised programme
• Real Estate License Exam, if the licence is legally required where you live and you have savings to cover several months with no commission coming in
• CAPM (Certified Associate in Project Management), only if you already ran projects and five or more of the postings you would accept name a project management credential as a requirement
• CompTIA A+, only if the help desk postings in your area name it in the requirements section rather than under preferred

Common questions

Sources

1. 01Callback rates for identical resumes sent to administrative job vacancies, by applicant age 14.4 percent at ages 29-31, 10.3 percent at ages 49-51, 7.6 percent at ages 64-66
   Neumark, Burn and Button, NBER Working Paper 21669 (published in the Journal of Political Economy, 2019) checked 2026-08-28
2. 02Share of eligible job ads in the study that were administrative jobs about 61 percent of the total
   Neumark, Burn and Button, NBER Working Paper 21669, section 5 checked 2026-08-28
3. 03Number of job applications submitted in the correspondence study over 40,000
   Neumark, Burn and Button, NBER Working Paper 21669 checked 2026-08-28
4. 04Callback rates across all four occupations combined, relative to applicants aged 29 to 31 about 18 percent lower for middle-aged applicants and about 35 percent lower for older applicants
   Neumark, Burn and Button, NBER Working Paper 21669 checked 2026-08-28
5. 05Estimated effect of listing computer skills for older administrative applicants, compared with the age gap in the same model 0.034 against an age gap of 0.09, which the authors say by no means accounts for the age difference
   Neumark, Burn and Button, NBER Working Paper 21669, footnote 58 checked 2026-08-28
6. 06Reemployment rate in January 2026 for long-tenured displaced workers, by age 72.9 percent at ages 25-54 and 57.3 percent at ages 55-64
   U.S. Bureau of Labor Statistics, Worker Displacement: 2023-2025 (released 27 August 2026) checked 2026-08-28
7. 07Status in January 2026 of long-tenured displaced workers aged 65 and over 38.6 percent reemployed and 48.9 percent not in the labor force
   U.S. Bureau of Labor Statistics, Displaced Workers Table 1 checked 2026-08-28
8. 08Share of reemployed full-time displaced workers earning as much or more than on the job they lost about 49 percent in January 2026, down from about 62 percent in January 2024
   U.S. Bureau of Labor Statistics, Worker Displacement: 2023-2025 checked 2026-08-28
9. 09Average duration of unemployment in July 2026, not seasonally adjusted 33.1 weeks for people aged 55 to 64, against 23.7 weeks for all people aged 16 and over
   U.S. Bureau of Labor Statistics, Current Population Survey Table A-36 checked 2026-08-28
10. 10Who the Age Discrimination in Employment Act covers people aged 40 and over, at employers with twenty or more employees for each working day in each of twenty or more calendar weeks in the current or preceding calendar year
    U.S. Equal Employment Opportunity Commission, text of the ADEA checked 2026-08-28
11. 11Seventh Circuit en banc holding on ADEA disparate impact claims by outside job applicants affirmed dismissal on 23 January 2019; the job description required '3 to 7 years (no more than 7 years) of relevant legal experience', Kleber was 58 and had more, CareFusion hired a 29-year-old; the court held that section 4(a)(2) protects employees from disparate impact age discrimination but does not extend that protection to outside job applicants
    U.S. Court of Appeals for the Seventh Circuit, Kleber v. CareFusion Corporation, No. 17-1206, opinion via govinfo checked 2026-08-28
12. 12Cost and alternative eligibility route for pharmacy technician certification 129 dollars to apply and sit the exam; a second pathway of 500 hours of pharmacy technician work experience in place of a recognised training programme
    Pharmacy Technician Certification Board checked 2026-08-28
13. 13How the IRS grants enrolled agent status by passing a three-part test covering individual and business tax returns, with 72 hours of continuing education every three years to maintain it
    Internal Revenue Service checked 2026-08-28
14. 14Origin of the widely repeated claim that recruiters spend six seconds on a resume an in-house eye-tracking study announced on 21 March 2012 in a press release by TheLadders, a commercial job-search and resume-services company, covering 'dozens of professional recruiters' over a ten-week period, written up as a white paper on the company's own blog rather than in a peer-reviewed journal
    TheLadders press release via PR Newswire checked 2026-08-28

Related