Five days off isn't a retention policy. Six or more is where resignations start to drop.

You have a resignation letter on your desk, and you are about to look at the benefits sheet for the first time in a while. Five paid days off. You tell yourself it is in line with what other companies in your market do, and that people who want to leave will leave regardless.

A paper published on January 27, 2026 in the Journal of Strategy and Management puts a number on that assumption. It is titled "Does one week now prevent two weeks notice later? A longitudinal study of paid time off and employee retention." The authors are Patricia Stoddard Dare and Vasilios D. Kosteas, both of Cleveland State University; Candice Vander Weerdt of Cleveland State's Department of Management; and LeaAnne DeRigne of Florida Atlantic University.

Here is how far I got. The paper is behind a paywall, so I read the published abstract through Crossref's metadata, plus HR Dive's January 29, 2026 coverage of it. I did not read the full text. Everything below about method and findings comes from that abstract. Anything about what the paper does not say is limited to what the abstract does not say.

The method, per the abstract: a longitudinal sample of employees from the National Longitudinal Survey of Youth, with more than 32,000 observations over 18 years, analyzed with logistic and fixed-effect models, framed by Conservation of Resources theory. The finding that matters:

Logistic regression results reveal that a low number (1–5) of paid days off per year has little effect on voluntary resignations. Yet there is a meaningful reduction in resignation for employees who received a moderate (6–10) or a high (11 or more) number of paid days off annually.

Abstract, Journal of Strategy and Management, as published via Crossref

The abstract adds that the pattern showed up in separate samples of men and women, and that women responded significantly more to higher levels of paid days off than men did. One note on sources: HR Dive's write-up says the 6–10 day benefit was especially strong for men. The abstract I read says women responded more. I have gone with the abstract, and I could not reconcile the two without the full paper.

The limits, which are mine and not the authors': this is observational. It shows that people with more paid days off resigned less, not that adding days to your policy would cut your resignations by a given amount. The abstract does not say which employers, industries or ages the sample covers beyond the survey's name, and I cannot tell you how closely it resembles your people. The 32,000 figure counts observations over time, which is not the same as 32,000 different workers. Those are things I could not check from the abstract.

Why this lands differently for the person at the top. In a company like yours, the PTO number is probably one you set, or inherited and never revisited. It rarely appears on a dashboard. Nobody is measured on it. It does not generate a complaint; it can show up as a quiet exit with some other reason attached. A cost that never carries its own label is easy to leave alone. The finding here is that the first five days do very little, which means a policy can sit in the range where you are paying for the benefit and getting almost none of the effect.

The part the research doesn't say, but that I'll say plainly: the person who decides how much time off the company offers is often the person least bound by it. If you can take a week whenever you choose, the difference between five days and eight can look trivial. For the person who has to ask, it may be the difference between a real break and a long weekend. That is my reasoning, not a result in the paper, and the abstract does not test it.

Also note the shape of the finding. It is a threshold, not a slope: little effect at 1–5, meaningful reduction at 6–10, and a meaningful reduction at 11 or more. The abstract does not say that 11 is better than 8, only that both beat the low range. If your budget allows one move, the abstract points to getting out of the bottom tier rather than chasing the top one.

One thing to do this week. Pull the voluntary resignations from the last eighteen months. Next to each name, write the paid days that person had. If you offer one policy for everyone, the question becomes how many days people actually used before they left, and whether anyone left within a few months of a year in which they took almost none. Do not draw a conclusion from a handful of exits. Just see whether the pattern is there before you decide the benefits sheet is not part of the problem.

Retention is a wiring question as much as a policy one: what you offer, and how the person at the top is wired to see it. If you want to see where your own blind spots sit, the Leadership Profile is built for exactly this.