Ask what a shutdown costs and you will be given a figure for deferred pay, which is the one number in this that is certain to be recovered. Hold the question. The answer is not in any of the accounts currently being kept, and it is not denominated in dollars.
The Department of Homeland Security has been without appropriations since 14 February. Around fifty thousand transportation security officers have been working unpaid for a month. On Monday roughly one in ten did not come to work, and at some airports the figure was far higher: about thirty percent at JFK, thirty five at Houston Hobby, thirty seven in Atlanta, thirty nine in New Orleans.
The argument here is that the durable cost of this shutdown is already fixed, that it was fixed by a training requirement rather than by any negotiation, and that nothing either side agrees to in the coming weeks can now undo it.
The first test: what actually leaves
Missed pay is recovered. That is settled law and it is not in doubt: back pay is automatic, and when the appropriation arrives the arrears are paid.
What is not recovered is the person who resigned. Several hundred transportation security officers left in the opening stretch of this lapse, and the rate has not been falling.
It is worth being precise about who these people are, because the debate tends to imagine a federal workforce of career officials with savings. A screening officer’s starting salary is modest, the work is shift based, and a substantial proportion of the workforce is in the first few years of the job. An unpaid month is not an inconvenience to that household. It is a reason to take the warehouse job, which pays on Friday.
Those departures are permanent in a way the shutdown is not. The lapse ends by agreement. The resignation does not.
The second test: the number that fixes the damage
Here is the figure that should be in front of every person negotiating this and almost certainly is not.
A new transportation security officer takes four to six months to train and certify before they can staff a checkpoint alone. That is not a bureaucratic delay that could be compressed by urgency. It is a curriculum, a practical assessment and a certification, for people operating detection equipment and making judgments with legal consequences, and shortening it means certifying people who are not ready.
Apply that to a calendar. An officer who resigned in late February is replaced by somebody hired in, at best, the week the appropriation passes. That person reaches a checkpoint four to six months later. If the lapse ended tomorrow, the staffing hole it has already created persists into the late summer. Every further week of lapse pushes the recovery date out by more than a week, because the resignations continue while the pipeline has not started.
So the cost is not being incurred at the rate of one day per day. It is being incurred at a multiple, and it lands months after anybody is paying attention.
And the pipeline cannot be surged, which is the part that surprises people who assume urgency solves staffing. Before training begins there is a hiring process: a posting, a competitive assessment, a conditional offer, a background investigation and a medical. Those steps have their own durations and most of them involve queues at other agencies. A hiring surge does not shorten a background investigation; it lengthens the queue for one.
So the four to six months is the training clock alone, measured from the point a person is cleared to start. The full distance from a resignation to a replacement standing unaccompanied at a checkpoint is longer than that, and the gap is wider at precisely the airports where the resignations are concentrated, because those are the high cost cities where the job was already hardest to fill.
The third test: the callout rate is not the same thing
The ten percent who did not come in on Monday are a different phenomenon and should not be added to the resignations, because most of them will return.
An unpaid worker who does not report is usually not protesting. They are doing arithmetic about childcare they can no longer pay for, or a second job they have had to take, or a commute that costs money they do not currently have. Those are solvable the day the pay resumes.
But the callout rate does two things that matter. It concentrates the workload on whoever did turn up, at airports already short staffed, which is exactly the condition under which people decide they have had enough. And it produces the visible queues that generate the political pressure, which means the most visible symptom is the one that resolves quickly and the invisible one is the one that lasts.
What follows from all three
Put them together and the conclusion is uncomfortable for both sides of the dispute.
Whatever is agreed, and whenever, the country has already bought several months of degraded airport screening capacity in the second half of the year. Not as a risk. As an arithmetic consequence of departures that have happened and a training pipeline that runs at the speed it runs at.
That cost does not appear in any of the scoring. The Congressional Budget Office can estimate lost output. Nobody scores the replacement cost of a workforce, because the departures are recorded as ordinary attrition and the recruitment appears in a later year’s budget as ordinary hiring.
Meanwhile, the policy
One more thing belongs here, because it is the reason this is worth arguing about at all.
On 9 March a Justice Department rule took effect changing how the Board of Immigration Appeals handles cases, cutting the window to file an appeal from thirty days to ten and providing that appeals are dismissed unless a majority of the Board votes to take them.
That is a substantial change to an adjudication system, and it happened on schedule, by rule, while the funding fight over the same subject matter was in its fourth week and settling nothing.
The question
So: what does it cost?
The deferred pay, which comes back. The output loss, which is real and partly recovered. And a hole in a trained workforce that will be at its worst in the autumn, long after the lapse has ended and the story has moved on, measured in checkpoint lanes that cannot open because there is nobody certified to stand in them.
What I cannot work out is who, in the structure of this dispute, has any reason to care about that. The damage arrives after the settlement, so neither side carries it into the negotiation. It is attributable to no single decision, so nobody owns it. And it is already fixed, so conceding tomorrow would not reduce it.
Which produces a strange position. Both sides are arguing about the terms on which this ends, and the largest cost of it has already stopped depending on the answer.
