NEW: 🛡️ 96.8% of stocks rated Very Safe kept their dividend intact. Read our study.
Is Invvest’s Dividend Safety Score
really reliable?
We froze the scores of more than 1,100 companies in January 2024, then tracked their dividends for two years. Here are the results.
A dividend is only worth something if it can be sustained over time.
A company can offer an attractive yield today, then sharply cut its dividend a few months later because of deteriorating results, excessive debt or simply a payout policy that has become unsustainable.
That is precisely the risk Invvest built its Dividend Safety Score to identify.
But a proprietary score is only worth something if it holds up against the facts. So we decided to put it to the test.
Above all, we wanted to do it without hindsight bias or cherry picking. The scores used in this study were calculated in January 2024, then completely frozen. We then observed what actually happened to these companies’ dividends over the following two years.
of companies rated Very Safe kept their dividend intact during the first year.
of companies rated Very Safe kept their dividend intact during the second year.
more likely to run into a dividend issue with a company rated below Safe than with one rated Very Safe.
of dividend cuts and suspensions came from companies the score had already rated below Safe.
What is the Dividend Safety Score?
Invvest’s Dividend Safety Score is an indicator designed to estimate how solid and sustainable a company’s dividend is.
Its goal is not to determine whether a stock is a good investment, nor to predict its share price.
It aims to answer a much more specific question: how likely is a company to be able to maintain its dividend over time?
To do so, the score relies on various factors related to the company’s financial health and the sustainability of its payout policy.
The higher the score, the more solid the dividend is considered. Conversely, a low score indicates a higher risk of the dividend being cut or suspended.
Test methodology
We wanted to avoid a classic bias: recalculating the scores once the outcome is already known. The protocol was therefore deliberately simple.
1. Initial sample
In January 2024, every dividend-paying US company with a market capitalization above $1 billion received a score, calculated solely from the historical financial data available at that date.
That represented 1,106 companies in the first test.
From that date on, the scores used for the study were never modified.
No new earnings releases, changes in debt, changes in profits or management announcements were taken into account.
2. Classification
- Very Safe (17 to 19)
- Safe (13 to 16)
- Borderline (9 to 12)
- Unsafe (5 to 8)
- Very Unsafe (0 to 4)
3. Tracking the dividends
- in 2024 versus 2023
- then in 2025 versus 2024
A decrease in the dividend is considered a cut, while a drop to zero is considered a suspension.
Both events count as a dividend issue in this study.
Results
First year: the lower the score, the higher the risk
A company rated “Very Safe” ran into an issue in only about 3% of cases. Conversely, nearly one in three companies rated “Unsafe” cut or suspended its dividend.
Taking the “Safe” and “Very Safe” categories together, only 5.55% of companies cut or suspended their dividend during the first year. In other words, 94.45% of companies rated “Safe” or “Very Safe” maintained their dividend.
For the few companies rated “Very Unsafe”, the issue rate even exceeds 50%, although this category has too few observations to draw a strong statistical conclusion.
Indeed, only 11 companies were rated “Very Unsafe”, because at market capitalizations above $1 billion, companies that fragile rarely pay dividends.
The real test: what happens two years later?
The second year is probably the most interesting part of this study.
We took the exact same scores from January 2024, without updating them, and looked at the dividends paid in 2025. Here is what we found.
A two-year-old score still differentiated risk
In January 2024, the model obviously could not know:
- 2025 financial results
- future board decisions
- future economic slowdowns
- or changes in payout policy
Yet the relationship between the score level and future dividend risk was still visible two years later.
This suggests that our score does not just measure a one-off situation, but also captures some of the more structural elements of a company’s financial strength.
An “Unsafe” company was about 10 times riskier
During the first year:
- 3.17% of “Very Safe” companies ran into an issue
- versus 30.61% of “Unsafe” companies.
A company rated “Unsafe” was therefore about 10 times more likely to cut or suspend its dividend than a company rated “Very Safe”.
So the score did not simply split companies into “good” and “bad”. It created several clearly distinct levels of risk.
7 out of 10 issues had already been flagged
In other words, most dividend issues did not occur among the companies the score considered the most solid.
They were concentrated in the part of the universe the model had already identified as riskier.
Does the result only hold because we analyze small companies?
We wanted to check that the results were not simply due to the presence of smaller or financially fragile companies.
So we ran the exact same test again, keeping only US companies with a market capitalization above $10 billion.
The sample included about 500 companies. And the same hierarchy appears.
First year
Second year
Among large caps, companies rated “Safe” and “Very Safe” protected their dividend in 94.87% of cases in the first year and 97.10% of cases in the second.
What about European companies?
We also extended the test to US and European companies with a market capitalization above $1 billion. The sample then reaches about 1,800 companies and the relationship remains visible.
First year
Second year
The score’s performance is nonetheless weaker in the European universe.
Part of this gap may come from the more limited availability of financial data in Europe, but also from structural differences between US and European dividend policies.
We therefore currently treat the US universe as the main test and the European results as an additional validation of the score’s overall behavior.
What this study teaches us
A financially solid company can decide tomorrow to completely change its capital allocation policy. An acquisition, an exceptional crisis or a board decision can also lead to a cut that is impossible to anticipate from historical fundamentals alone.
The question was a different one:
Does the score really distinguish companies with a low dividend risk from those with a high one?
Over the two years studied, the answer from the data is encouraging.
In every main test: Very Safe → Safe → Borderline → Unsafe
Showing that the frequency of issues increases as the score decreases.
And this relationship remains visible when we:
- look at a second year without recalculating the scores
- restrict the analysis to large caps
- extend the universe to Europe
Limitations of the study
A single generation of scores
This study is based on scores calculated in January 2024 and observed over two years.
It is an encouraging first result, but it does not yet cover several economic and market cycles. We therefore plan to repeat this study every year.
Possible survivorship bias
Some companies that left our database during the study period may no longer appear in the sample. This could slightly and artificially improve the results of some categories.
Few “Very Unsafe” companies
Only 11 companies belonged to the “Very Unsafe” category. The percentages obtained for this category should therefore be interpreted with some caution.
Data quality varies across markets
US companies generally have more frequent and standardized financial data. The European results are therefore used as additional confirmation rather than as our main validation test.
What’s next?
Every year, we will:
- freeze a new generation of scores
- track the dividends actually paid
- measure cuts and suspensions
- compare the results with the initial categories
- publish the results, whether they are good or bad
The goal is to gradually build a true track record of our score across different economic and market environments.
Special offer 🎁
To enjoy unlimited access to our Dividend Safety Scores for all stocks, we are offering you a 30% lifetime discount on the Invvest PRO annual subscription with the code:
ETUDE2026
Discount is available until October 20th 2026
The results presented in this study are based on historical observations. They do not guarantee that any dividend will be maintained in the future and constitute neither an investment recommendation nor financial advice.