Paying People to Leave Is an Old Authority

Offering federal employees money to resign sounds like an invention of the past two years. It is an authority Congress wrote in the 1990s, and the rules it comes with explain most of what has happened since.

The statute that authorises the payments is short. An agency may offer a voluntary separation incentive to an employee who then resigns or retires, and the payment may not exceed $25,000.

The cap is the first thing to notice, because it has not moved since it was set. A sum that was a meaningful fraction of a federal salary three decades ago is now a fraction of a fraction, which is why a buyout alone persuades few people to leave a career post.

the definitions section immediately before it sets out who may be offered one and for what purpose: to reduce the number of employees, or to restructure the workforce by reshaping the mix of skills an agency holds.

Those two purposes are not the same, and the difference is the whole argument. Reducing headcount is arithmetic. Restructuring is a claim that the agency knows which skills it needs and is trading the ones it has for them.

The statute also attaches a condition that reveals what Congress was worried about. An employee who accepts a payment and returns to federal employment within five years must normally repay the entire amount.

That clause exists because the obvious abuse is obvious. Without it an agency could pay a worker to leave in September, book the saving, and rehire the same person as a contractor in March, having spent money to accomplish nothing.

OPM's guidance to agencies adds the procedural layer: a plan submitted in advance, identifying the positions and functions to be cut, the number of employees involved, and how the agency will operate afterwards.

The plan requirement is the part that turns a buyout from a cash offer into a workforce decision. It forces an agency to say, before any money moves, which work it intends to stop doing.

None of this describes what happened in 2025. The deferred resignation programme was not run as a set of agency-by-agency separation incentives under this authority; it was a government-wide offer, made centrally and quickly.

Its mechanism was different too. Rather than a lump sum capped at $25,000, employees who accepted were placed on the administrative leave statute's paid status and continued to draw full salary and benefits for months before separating.

That is a larger payment than the buyout statute permits, delivered in a form the buyout statute does not govern. An employee on months of paid leave at a senior salary receives well beyond the cap, without the cap ever applying.

It is also, in the ordinary sense, a separation incentive. The employee is paid, does no work, and leaves at the end. The difference is which statute is cited on the paperwork.

The administrative leave provision was written for a narrower world: investigations, building closures, weather, the interval while a personnel matter is resolved. Congress amended it in 2016 specifically to stop agencies parking employees on paid leave indefinitely.

That amendment imposed a limit of ten days a year for the investigative category and required agencies to record what they were doing. It did not anticipate a voluntary programme in which the leave was the inducement rather than the consequence.

The practical result is a programme that is very likely lawful and sits outside every safeguard built for the thing it resembles: no advance plan, no position-by-position identification, no cap, and no repayment obligation on anyone who returns.

The repayment clause is the omission that will matter longest. Under the older authority, a rehired employee writes a cheque back to the Treasury. Under this one, nothing is owed.

There is a defensible case for moving fast. Agency-by-agency plans take months, and a government intent on shrinking quickly will reach for the instrument that does not require them.

But the safeguards being bypassed were not decoration. Each was written in response to a specific way that paying people to leave had previously gone wrong.

Which is why the number to follow is not the cost of the programme but the rehiring rate — how many of those posts are filled again within five years, and by whom. Under the older statute that question had a financial answer. Under this one it has only a political one.