Why Externally Tested Credentials Hold Their Value

← Back · ACT II · Chapter 7: Value Is Not Promised; It Is Proven

Most organizations build capability by adding to it. They buy a platform, fund a training program, sponsor certifications, stand up a center of excellence, and count each addition as progress. The additions are real enough. Whether any of them produced an ability the organization can use on demand is usually assumed rather than checked.

A recent study puts a price on that assumption, in the one market that has to pay for it: the market for workers.

In a Brookings Institution working paper, Eduardo Levy Yeyati, Ian Seyal and Sophia Henn examined 37.7 million US worker resumes and matched 54.3 million credentials against the occupations of the people holding them. The study set out to answer a labor question, which credentials raise wages. Its results speak just as directly to how organizations build capability.

What the Market Actually Buys

A first credential relevant to a worker’s job is associated with a 3.8 percent wage premium, against 1.8 percent for a first credential unrelated to the job. Few people would be surprised by that gap.

What happens with the second and third credential is more interesting. Each additional relevant credential adds about 1.0 percent. Piling up irrelevant ones shows no return at all, and sometimes a negative association, so accumulation without relevance can leave a worker worse off.

Verification Is the Variable

The mechanism shows up when the credentials are split by type.

Certifications typically require a proctored exam, regular renewal and validation by a third party. A first relevant certification is associated with a 4.1 percent premium, and each additional one with another 2.2 percent. For workers without a college degree the figures are 7.1 percent for the first and 3.5 percent for each one after, so the value keeps building.

Badges, which Levy Yeyati describes as often low-stakes participation badges, behave differently. A first badge carries a 5.3 percent premium even when it has nothing to do with the job, which the authors read as a signal that separates the holder from someone with no credential at all. Accumulating relevant badges goes the other way, with a penalty of 3.0 percent.

Both sit on the same line of a resume. What separates them is that, for a certification, someone independent tested the claim and the holder has to renew it.

The Same Test Runs at the Enterprise

Look at what a certification actually asks for: an exam, a renewal date, and a check by someone outside. That is the same test an organization’s capabilities should have to pass. The labor market has put a price on it, one person at a time.

Picture an enterprise that has bought forty tools, run eleven training programs and published a capability roadmap. In most cases nobody outside the team that built those abilities has tested them, none of them come up for renewal, and if an ability fades, nothing in the plan records it. What the organization holds looks a lot more like the badge portfolio than the certification record.

The labor market does not pay workers for that kind of accumulation, while many capability plans still count every item on the list as progress.

Subtraction Is the Discipline

Here is a plainer way to define capability. Take away everything the outcome does not need. What is left counts as capability only if it still works when someone who did not build it tests it.

That changes the question an executive asks of the plan. Instead of what to add next quarter, the question becomes which of the abilities already on the list would pass an independent test today. In many organizations the honest answer is that nobody outside the delivery team has ever looked.

There is a reason capability plans grow faster than capability. Adding is easy to approve and report, and it produces something to show within the quarter. A test produces only a result, and the result is often unwelcome. So organizations keep buying and skip the test, sometimes for years, until an ability that has sat on the list for a decade turns out not to work when it is needed.

What Happens When the Standard Changes and Nothing Tests It

There is a large piece of evidence for what a declared standard does when no independent check sits behind it, and it comes from hiring rather than from a capability plan.

Harvard Business School and the Burning Glass Institute studied 11,300 roles at large US firms that had removed a degree requirement from their job postings. Inside those roles the share of hires without a bachelor’s degree rose about 3.5 percentage points, which is a real change. But only about 3.6 percent of US roles dropped the requirement in the years studied, so across the market the net movement came to 0.14 percentage points, an estimated 97,000 hires out of roughly 77 million in a year.

Look at what happened across the firms that made the same change. About 37 percent followed through and hired more people without degrees. Around 45 percent changed nothing at all, and roughly a fifth ended up hiring a smaller share than before. One change to the posting, three different results, and nothing in any of those organizations was obliged to go back and ask which one it had produced.

It is easy to read that study as a hiring problem and reach for better assessments and structured interviews. Those are sensible. What they leave alone is the original question, because the firms in the study were not bad at hiring. They changed something they said about themselves and added nothing that could tell them later whether it was true.

A capability plan behaves the same way. An item turns green when a program is funded or a seat is filled, and nothing in the organization is then obliged to check whether the outcome was ever produced.

The Table Many Organizations Cannot Complete

Take three capabilities from the organization’s own plan. For each, write down the evidence that it produced its outcome more than once, who outside the delivery team verified it, when that was last re-checked, and what it costs per unit of outcome.

Many organizations cannot complete a single row. The exercise takes an afternoon, and it can take more off the capability plan than the next program would add.

A supplier whose training or platform does build a working ability gains from that table, because it is the one check that separates its product from the tools that were bought and never tested.

Levy Yeyati notes, citing Credential Engine’s count, that more than 1.5 million non-degree credentials now exist in the United States, and warns that without better quality infrastructure their spread risks creating “more noise than signal.” Enterprises face the same problem from the buying side, with more money at stake. The check that works for them is the one a certification already gets: someone outside the team runs it again on a schedule, against a definition held by the buyer.

References: Eduardo Levy Yeyati, Ian Seyal and Sophia Henn, “The market value of non-degree credentials,” Brookings Institution Working Paper 196; Eduardo Levy Yeyati, “The market value of non-degree credentials: New evidence from 37 million US workers,” CEPR VoxEU, January 31, 2026. Harvard Business School and the Burning Glass Institute, “Skills-Based Hiring: The Long Road from Pronouncements to Practice,” February 2024.

Keep reading: ACT II · Chapter 7: Value Is Not Promised; It Is Proven