Coverage Is Not the Same as Protection
The subject here is how paid sick leave is distributed across jobs and why formal access does not guarantee usable protection. The practical question is not simply whether an average coverage rate has risen, but who can convert that coverage into time away when illness collides with work. That distinction lets us read access statistics as evidence about workers’ options—not as a direct count of presenteeism.
Averages hide the worker and the job
A coverage percentage is a ratio, but it does not tell us what sits behind the ratio. Before comparing two numbers, we need to know the unit being counted, the population included, the job to which the answer refers, and the time period. A percentage of jobs is not automatically a percentage of workers. A survey of a worker’s sole or main job is not interchangeable with an employer-based series. A current access estimate cannot be read as if it were a continuous historical measurement when the survey design has changed.
The two headline figures in the available evidence make this problem concrete. The IZA review reports that the share of U.S. private-sector jobs with paid sick leave rose from 63 percent in 2010 to 77 percent in 2023. The Bureau of Labor Statistics’ 2024–25 Leave and Job Flexibilities Module reports that 71 percent of wage and salary workers had access to paid leave at their sole or main job, compared with 66 percent in 2017–18. The figures describe different years, sources, and populations. They should not be averaged, aligned as though they form one series, or treated as evidence that one source contradicts the other.
They can still support a useful conclusion: nominal access has expanded while a large minority remains outside, uncertain about, or unable to rely on paid leave. The BLS specifically notes that workers may not know whether they can use leave until they need it. That detail matters. A benefit that is not understood at the moment of illness is less usable than one with clear rules, even if both are counted as available.
The distribution of access is also not random. In the IZA review’s pre-mandate U.S. benchmark, 76 percent of full-time employees had paid sick leave compared with 28 percent of part-time employees. Coverage was 33 percent in the bottom wage quartile and 86 percent in the top quartile. Large private firms covered 74 percent of jobs, compared with 54 percent in firms with fewer than 100 employees. These figures describe the labor market before the main expansion of state and local mandates, but they show why an average can conceal a steep economic gradient.
The BLS data show that the gap remains visible in the more recent period. Among single jobholders, 81 percent of full-time workers had access to paid leave compared with 32 percent of part-time workers. Access reached 90 percent in management, business, and financial operations, but was 45 percent in service occupations and 50 percent in construction and extraction. Public-sector workers were more likely to have access than private-sector workers, 81 percent compared with 69 percent. Workers who could work at home had 86 percent access, compared with 63 percent among workers who could not work at home.
These differences are not merely demographic trivia. They describe the circumstances in which staying home is feasible. A full-time employee in a large organization may have a leave balance, a substitute, and a predictable process for reporting an absence. A part-time service worker may have less coverage, less bargaining power over schedules, and a job that requires physical presence. The sources do not establish that any one characteristic causes presenteeism by itself; the BLS cautions that broad comparisons do not control for factors such as the distribution of full- and part-time work. But the pattern is enough to reject a one-size-fits-all picture of protection.
Work-at-home status illustrates the difference between an association and a mechanism. A person who can work remotely may have more ways to avoid close contact while still performing some duties. That can reduce a particular route of contagion. It does not mean the person is well, fully productive, or free from pressure to work. Nor does the higher access rate among remote-capable workers prove that remote work caused the difference. It may reflect occupation, earnings, employer type, or other conditions that travel with the ability to work at home.
The same caution applies to the contrast between service work and management occupations. High-contact work can make presenteeism more socially costly because a sick worker interacts with coworkers or customers. But the access numbers alone do not measure infections or the number of workers who came in sick. They identify where the option to stay home is less common, which is an important condition for understanding the later behavior.
Entitlement meets the workplace
Formal access becomes practical protection only when the worker can use it without an unacceptable loss. The BLS release separates these stages. During an average week in 2024–25, 21 percent of wage and salary workers took paid or unpaid leave, taking an average of 14.4 hours. Four percent took leave because they were ill or needed medical care. During an average month, eight percent needed to take leave but did not.
That last figure is easy to misuse. It does not mean that eight percent of workers worked sick. The group includes people who needed leave for different reasons and may have responded to unmet need in different ways. About 40 percent needed leave for their own illness or medical care, 24 percent for the illness or medical care of a family member, and 21 percent for errands or personal reasons. Some may have worked, some may have delayed care, and some may have found another arrangement. The measure is therefore evidence of a gap between need and use, not a direct prevalence estimate of presenteeism.
The reasons for non-use reveal why the gap matters. Among workers who needed leave but did not take it, 29 percent said they had too much work. Fourteen percent cited fear of negative employment consequences or a denied request. Thirteen percent could not afford the loss in income. These are not all the same barrier. Too much work points to a coordination problem: the worker may believe that absence will leave a team or shift unable to function. Fear or denial points to the relationship between a nominal right and its enforcement. Affordability points to the remaining price of absence even when leave is technically available.
The same nominal entitlement can therefore produce different behavior in two workplaces. In one, a worker can report an illness, use a paid balance, and hand off the shift because staffing is designed to absorb short absences. In another, the worker may technically have the same balance but expect a reduced schedule, a difficult performance conversation, or a coworker left alone with the workload. The survey cannot identify every step in that chain, and it does not prove that the second worker came to work sick. It does show why a legal or administrative measure of access is only a starting point for understanding behavior.
This is also why the job itself belongs in the analysis. A remote-capable employee may be able to avoid close contact while continuing a limited amount of work, whereas a food-service or construction worker may have no comparable arrangement. The two workers could report the same need for leave but face different consequences from not taking it. For the onsite worker, the choice may immediately affect customers or coworkers; for the remote-capable worker, the main consequence may be reduced productivity or slower recovery. Paid sick leave addresses the income and employment side of that choice, but it does not make the underlying job constraints disappear.
The barriers can reinforce one another. Too much work can make a request feel risky, which makes the worker less likely to ask whether leave is available. A partial wage replacement can turn that hesitation into a financial calculation, while uncertainty about the rule can delay the decision until the worker is already scheduled to appear. These interactions are not separately measured by the BLS release, so they should not be presented as a quantified causal model. They are the practical reason to treat the reported reasons as pathways that require different responses rather than as interchangeable explanations.
Together, they make “usable protection” more precise. It requires at least four conditions. The worker must be eligible. They must know that the benefit exists and understand how to claim it. The replacement must be sufficient for their household budget. Finally, the workplace must make the request credible in practice by allowing the absence without a prohibitive penalty. A policy can improve one condition while leaving another untouched.
This is why access gaps and unmet need should be treated as contributing conditions, not as a single diagnosis. If part-time workers have lower coverage, the first problem is eligibility or benefit design. If a covered worker cannot leave because the shift cannot be staffed, the problem includes workplace capacity. If a worker fears losing hours or being judged unreliable, the formal rule may be stronger than the practical protection. If the benefit replaces only part of wages, the worker faces a financial constraint. Each pathway can lead to work while sick, but the remedy would not be identical.
The evidence also cautions against assuming that non-use is always a preference. Economic models often begin with workers weighing the value of rest against the cost of absence. That is useful, but the observed choice may reflect a constrained decision rather than a free one. A worker who stays at work because missing a shift threatens rent payment is not revealing that they value work more than health in some uncomplicated sense. They are responding to the price and risk structure around the decision.
At the same time, the opposite mistake is also possible. Not every worker who takes leave is responding to contagious illness, and not every absence is inefficient. The BLS figures include vacation, family care, errands, and personal reasons. The IZA review’s broader economic framework therefore asks us to consider both the value of time away and the costs imposed when leave is used unnecessarily. Protection is not measured by maximizing use; it is measured by making the safer and more productive choice feasible when illness makes working a poor option.
That standard changes how we interpret inequality. A higher average access rate can be a real improvement and still leave the workers most exposed to contact, low wages, or inflexible schedules with the fewest usable options. It also explains why remote work should not be treated as a substitute for paid sick leave. Remote work may lower one kind of transmission risk, but it can preserve pressure to work through illness and is unavailable in many service and site-based jobs.
The practical audit is straightforward. When someone cites an access number, ask: Who is counted? What kind of job do they hold? What happens when they request leave? The first question checks the denominator. The second checks whether the worker can avoid physical presence or absorb the absence. The third tests whether formal coverage survives contact with workload, fear, denial, and income loss. Only after those questions are answered can we interpret a change in leave use or infection activity as evidence about presenteeism.
Coverage, then, is a necessary condition for protection but not a sufficient one. It tells us that a policy has placed an option somewhere in the system. It does not tell us whether the option reaches the worker who needs it, at the moment it matters, at a cost they can bear. The next step is to ask what happens when mandates expand that option—and whether the resulting change is lower contagious work, higher legitimate leave use, or both.