How a change in location can change access to work
This chapter follows the mechanism connecting work from home to disability employment: how changing the place, timing, and sensory conditions of a job can alter a person's cost of participating. The focus is not on remote work as a general lifestyle choice, but on the specific prediction that occupations with larger WFH opportunities should show larger employment gains for people with disabilities. Understanding that prediction lets us separate a plausible accessibility pathway from an assumption that remote work helps everyone in the same way.
From a universal shock to unequal access
The pandemic did not create WFH as a disability program. Firms adopted remote arrangements because a public-health shock disrupted ordinary workplaces and forced them to reorganize work for the broader workforce. The IZA paper reports that before the pandemic about 7% of full workdays in the United States were worked from home. By 2023, the share had risen four-fold to 28%. The change was broad in origin, but its consequences need not have been evenly distributed.
The reason is that the old workplace was not a neutral setting. It bundled several requirements together: reach the site at a particular time, navigate its physical layout, tolerate its lighting and noise, and remain productive within a fixed schedule. Some workers could satisfy those requirements with little difficulty. Others could do the substantive tasks but faced a much higher cost of satisfying the surrounding conditions. A change in work location can therefore affect access even when the job's core tasks have not changed.
The distribution of WFH across occupations makes this possibility testable. The paper gives a sharp contrast: WFH rose by about 36 percentage points for computer occupations, such as software developers and database administrators, but by only about 4 percentage points for teachers. This is not a claim that every computer worker can work remotely or that no teacher can. It is evidence that occupations experienced different degrees of exposure to the new work arrangement. If WFH removes a barrier for disabled workers, employment should rise more where the exposure is larger, after accounting for other changes that affected occupations at the same time.
Exposure, however, is not the same as use. There are at least three links between an occupation's WFH opportunity and an individual worker's access. First, the job's tasks have to be performable away from the worksite. Second, the employer has to make the arrangement available in practice, with a schedule and tools that preserve the job's essential functions. Third, the worker has to have a home setting and personal circumstances that make the arrangement useful. A high WFH share at the occupation level tells us that the opportunity expanded; it does not tell us that every worker used it, that every employer implemented it in the same way, or that every home provided an equivalent environment.
That distinction explains why occupation is a useful but imperfect lens. Occupation-level variation can reveal whether the places where WFH expanded also saw larger disability-employment changes. It cannot by itself say which worker received a remote option, which barrier was removed, or whether a worker would have chosen a different accessible arrangement. A positive average may contain several individual paths: someone entering a job, someone remaining employed after a disability appears, someone switching hours, or someone who does not use WFH but benefits from a more flexible employer because remote work became normal. The aggregate relationship is meaningful precisely because it is a pattern across many such possibilities, but it does not narrate any one of them.
That prediction is stronger than the claim that WFH is convenient. Convenience can improve an already-employed person's preference or satisfaction without changing whether they can enter the labor force. Accessibility is about feasibility: whether the conditions of work permit a person to supply labor at all, or to remain attached to a job after a disability appears. The distinction matters because the paper's later results are concentrated in full-time employment and because its trend analysis shows employment gains coming mainly from people who had been outside the labor force, not from a large fall in unemployment.
A universal change can have an unequal effect without being designed to target a disadvantaged group. A curb cut may be useful to many people, but it matters most when a step was previously a barrier. WFH can operate in an analogous way when it changes a default condition—such as commuting or a fixed physical environment—that some disabled workers could not reliably meet. The comparison is limited: a remote job is not public infrastructure, and it does not make every task accessible. The point is about how changing a default can change participation for a group that previously had to request an exception.
The legal and organizational distinction is also important. The paper notes that reasonable accommodation can include adjustments such as remote work, flexible hours, and assistive technologies. But the post-pandemic change it studies did not arrive primarily through individual accommodation requests or a new legal ruling. It arrived as a general restructuring of work. The empirical question is consequently whether a broad change in working conditions created opportunities for disabled workers, not whether every employer has a duty to approve every remote-work request.
Two sides of the participation mechanism
The most direct pathway runs through labor supply. Imagine a worker whose substantive skills fit a job but whose route to the job is unusually expensive. Someone with a vision or physical limitation may be unable to drive or use public transit and may need a taxi or a family member to travel. The monetary cost is only one part of the burden. A commute can consume time and energy before the workday starts, introduce uncertainty into arrival, and make a full-time schedule harder to sustain. If the job can be done from home, those constraints may be reduced without changing the worker's ability to perform the job's central tasks.
This is a change in the cost of participation, not necessarily a change in skill. Consider two versions of the same job. In the first, the worker must repeatedly solve a transportation problem and adapt to a building that may be difficult to navigate. In the second, the essential work is performed from a controlled home environment. The task may be identical, but the total effort required to deliver it is different. A worker who was previously absent from the labor force may now find the job feasible; a worker already employed may find it easier to remain. The mechanism predicts a change in who can participate even if the employer does not lower the job's formal standard.
The working environment can impose a similar hidden cost. At home, a worker may have more control over desk height and placement, lighting, background noise, or the layout of a bathroom and kitchen. Control does not guarantee comfort or productivity; it changes the worker's ability to configure the environment around a limitation. For someone sensitive to noise, a quieter setting may reduce the effort required to concentrate. For someone with mobility restrictions, eliminating movement through an inaccessible office may matter more than the job title suggests. These examples explain how WFH can increase the supply of labor by lowering the per-hour disutility or practical cost of working.
Time is another part of the mechanism. WFH may make it easier to adjust work times or hours around treatment, fatigue, pain, or fluctuating functional limits. The paper does not measure each of these experiences separately, so they should be treated as plausible pathways rather than observed components of the estimated effect. Their common structure is what matters: a worker's capacity is conditional on the environment, and a different environment can change whether participation is sustainable.
There are plausible employer-side pathways too. A physical workplace may require ramps, elevators, automatic doors, or other modifications. If work can be done elsewhere, some of those site-specific costs may be avoided. Remote hiring may also change how visible a disability is during an interview, which could affect selection or discrimination. Finally, a worker may be more productive when they can control the conditions around them. The paper gives the example of a worker with a hearing disability who can work more effectively without background noise.
These employer-side channels differ from the worker-side story even when they lead to the same observed outcome. If WFH mainly lowers the cost of supplying labor, more people become willing or able to enter or remain in the market. A larger supply can increase employment while putting downward pressure on wages, at least relative to comparable work. If WFH mainly raises employer demand by lowering accommodation costs or increasing productivity, employment could rise alongside wages. The paper uses wage patterns later to weigh these possibilities; at this stage, the distinction tells us what evidence a convincing explanation must include.
The channels can also interact. An employer may offer remote work because it lowers the cost of a physical-site accommodation, while the worker accepts it because it eliminates a costly commute. A hiring process may be less revealing of disability, but the worker's ability to stay employed may still depend on a quiet room or flexible hours. Conversely, an employer can make WFH nominally available while setting rigid meeting schedules or providing inaccessible software, so the formal option does not remove the relevant barrier. These combinations are why the label “remote” is not itself a mechanism. The mechanism is the change in conditions that the worker actually experiences.
The mechanism can be represented as a sequence rather than a list of benefits:
The sequence has an important gate: the occupation and the job's actual organization determine whether a remote arrangement can remove the relevant barrier. WFH cannot eliminate the physical presence required by every task. Nor does a nominal policy guarantee that a worker can use it effectively. An arrangement may be technically remote but still impose rigid hours, inaccessible software, excessive video demands, or performance expectations that recreate the original barrier in another form. The sources do not measure these implementation details, so they belong to the boundary of the mechanism rather than to the estimated result.
The home environment can also redistribute rather than eliminate costs. A worker may need suitable space, equipment, reliable internet, privacy, or help with care responsibilities. The paper does not estimate those resources or their distribution. This matters for both interpretation and practice: the average employment response can be positive while some workers gain little, incur new costs, or prefer a well-designed accessible workplace to working from home. A serious accessibility judgment asks which constraint changed, not simply whether the arrangement was labeled remote.
The same conditional logic applies to disability categories. A quieter home environment may matter especially for some sensory limitations; eliminating a commute may matter especially for some mobility limitations; flexible timing may interact with several conditions. The paper's baseline study population is a selected set of physical-disability categories, and its strongest causal estimates should not be extended automatically to cognitive disabilities or to every person who answers “yes” to a CPS question. The mechanism is a reason to investigate heterogeneity, not permission to assume it.
This is why WFH can be called infrastructure only in a qualified sense. It changes a default work condition at scale, potentially lowering the need for each worker to negotiate the same exception. But it is infrastructure with a task boundary, a resource boundary, and a worker-choice boundary. The physical site may still need to be accessible. A worker may still need assistive technology, altered duties, a flexible schedule, or a different arrangement. And an employer still has to organize remote work so that the opportunity is real rather than symbolic.
The mechanism now gives us a concrete empirical expectation: if WFH is changing access, the employment response should be larger in occupations where WFH expanded more, and the response should appear in labor-force participation or retention as well as in job matching. But a prediction is not proof. The pandemic changed labor demand, disability reporting, and many other parts of work at the same time. The next question is how the researchers separate the occupation-specific WFH relationship from those concurrent changes.