Why Some Stocks Have No Dividend Safety Score

Cardinal Health has paid a dividend for 42 consecutive years. It has raised the payment in 30 of them. Its earnings are positive, its history is complete, and its dividend record has been reconciled against its own filings.

It has no Dividend Health Score at all.

That combination looks like a bug and is not one. Here is what actually produces a blank, measured across every page on the site.

How often is the score missing?

78 of the 798 instruments on DividendAtlas carry no score, which is just under one in ten. They split into two groups that have nothing to do with each other.

All figures as of 29 August 2026.

The first group is a decision. The second is a limitation. Reading them as the same thing is what makes the blank confusing.

Funds are not scored, and that is deliberate

An ETF's distribution is not a dividend in the sense the score measures.

When a company pays a dividend, a board decides the amount, and that decision is supported or undermined by the company's earnings, cash flow and balance sheet. Those are the things a safety score reads. When an income ETF pays a distribution, it is passing through whatever its holdings paid it, minus costs. No board set it. No balance sheet stands behind it.

Scoring that on payout coverage would produce a number with no meaning. The honest treatment is to leave it blank and let the fund be judged on what actually governs it: what it holds, what it costs, and how it is constructed.

This is why the 41 are not a backlog. They are not waiting for data.

For a company, one input is missing far more often than the rest

The 37 unrated companies are a different matter, and they are strikingly consistent.

The payout ratio is the discriminator. 95% of unrated companies have none, against 5% of the ones that carry a score. This follows from what the score is: a judgement about whether a dividend is affordable. Without a measure of what the payment costs relative to what the business produces, there is nothing to form that judgement from, however long and clean the payment record looks.

The second row is starker in its way. Every single rated company has a current unbroken payment run. Not one company with a gap in its recent record carries a score.

That row turns out to be less mysterious than it looks. All 22 of those companies are missing a dividend for both of the last two fiscal years, and 21 of them for the last three. They are not recovering payers waiting to be re-scored. They have stopped paying, and a dividend safety score for a company with no current dividend would be measuring nothing. Thirteen of the 22 also have a gap at 2020 or 2021, which is the shape you would expect from the businesses involved.

Which companies end up here

The 37 are not spread evenly, and the concentration is informative.

By sector: 9 industrials, 7 consumer cyclicals, 5 technology, 4 healthcare, 4 energy, 3 financial services, and single figures across the rest. Airlines alone account for three of them.

That is close to a list of the industries that suspended dividends during the 2020 disruption and have not restored them. It is not a random sample of the market, and it is not a sample of weak businesses either. Several are large, profitable companies that simply chose not to resume a payment. The score has nothing to say about them because there is no payment to assess, which is a different statement from a low score.

The two the explanation does not cover

Being straight about the residual matters more than a tidy account.

Of the 37, two have both a current payment run and a published payout ratio and still carry no score: Ares Capital and Fidus Investment. Both are business development companies, a lending structure that sits closer to a fund than to an ordinary corporate, and both yield above 9%.

Our score carries separate profiles for the structures whose accounting genuinely differs, which is why REITs are judged on funds from operations rather than earnings. A lending vehicle of this kind has no such profile, and companies without one sit outside the scored set. We are describing where they land rather than claiming to have diagnosed each case, because two instruments is not a pattern and the honest answer is that this pair needs looking at individually.

The 42-year company with no score

Price

$235.13

Dividend yield

0.88%

Annual dividend

$2.06

Key statistics

Day range$234.15 – $237.30

52W range$145.87 – $258.30

Volume1.0M

Avg. volume1.9M

Dividend amount$2.06

P/E ratio, GAAP32.51×

Forward P/E19.51×

Beta0.52

Market cap$54.69B

Cardinal Health is the clearest example of the first row in that table. Its card above carries a price, a yield of 0.88% and a dividend of 2.06 USD a share. Where the score would be, there is nothing.

Its payment record is not the problem. It has a 42-year run and a 30-year record of increases, both reconciled against the company's own filings and its SEC data. Its most recent reported earnings per share is 7.23 USD, comfortably positive against that 2.06 USD dividend.

What is missing is the published payout ratio, which our score strikes over an average of three complete fiscal years rather than a single one. When that window cannot be assembled, the payout measure does not resolve, and without it the score does not get written. The dividend looks entirely affordable on the figures we do have. "Looks affordable" is not the same as "measured as affordable", and the score only reports the second.

Why we would rather publish nothing

The alternative is worse, and it is the more common choice in this category.

A score computed from whatever inputs happen to be available, without saying so, is indistinguishable from a score computed from all of them. The reader cannot tell a firm read from a guess, and the guess is more likely to be wrong precisely where the data is thinnest, which is where they most need it to be right.

We already publish a confidence level for exactly this reason, and our guide to how a dividend safety score works explains what high, medium and low mean. The blank is the same principle taken to its end. Below a certain amount of information, the honest output is not a low-confidence number. It is no number.

A dividend site's entire value rests on its figures being checkable. Publishing a score for Cardinal Health that quietly skipped the payout pillar would buy completeness at the cost of the thing the number is for.

How common is this elsewhere?

Worth setting the number in context, because one in ten sounds high until you ask what the alternative looks like.

Most dividend sites publish a safety or quality rating for close to everything they cover, including funds. That is achievable in one of two ways: score the things that cannot be scored, or fill the gaps with defaults and do not mention it. Neither is visible to a reader comparing two sites, because a number and a number look alike whatever produced them.

Our 78 blanks are the visible cost of refusing both. Of them, 41 are funds we will never score. The remaining 37 are mostly companies that have stopped paying a dividend, which each page states plainly elsewhere. Framed that way, the population of "companies that pay a dividend, that we could score, and do not" is small: on these figures it is the two lending vehicles above, plus the handful whose payout window has not resolved.

That is the number worth watching, and it is the one we would rather report than a coverage percentage that flatters itself by counting ETFs.

What to do when you hit a blank

• Check whether it is a fund first. If it is, the blank is permanent and correct, and the score was never the right tool.
• Treat it as unmeasured, not unsafe. A blank carries no verdict in either direction. Cardinal Health's 42-year record is real whether or not we have scored it.
• Look at what is published instead. The payment history, the growth rates and the yield are all there on the page, and for a long-record payer they answer most of the question on their own.
• Come back later. Most blanks are an input that has not resolved rather than one that cannot. A later filing often fills them without anything about the business having changed.

The five-step check on whether a dividend will hold is the manual version of what the score automates, and it works fine on a company with no score. You can also filter by score in the screener, where the unrated simply do not appear, which is its own small reason to know they exist.