There is a peculiar assumption underlying much of the debate over Donald Trump’s second administration: that whatever one thinks of his transformation of the federal government, a future president can simply change it back.
A regulation can be rewritten. An agency can be reopened. Scientists can be rehired. Federal property can be reacquired. Grants can be restored. Alliances can be repaired. Environmental protections can be reinstated. Congress can pass new laws.
In other words, if Americans eventually decide that the Trump experiment went too far, we can just hit the reset button.
But governments do not have reset buttons.
They have people, buildings, laboratories, databases, contracts, leases, institutional knowledge, relationships, court precedents and complicated legal obligations. Destroying those things costs money.
Reconstructing them costs money again.
That raises a question that deserves far more attention than it has received:
If a future administration tried to restore the federal government to approximately the condition in which Trump found it on January 20, 2025, what would that cost taxpayers?
My rough estimate, based on the scale of changes visible through August 2026, is that the extraordinary federal reconstruction bill could already be on the order of $100 billion to $200 billion, above the ordinary cost of running the restored government.
A reasonable midpoint today might be around $140 billion.
If the transformation continues through January 2029, a future administration attempting a serious restoration could conceivably face a bill of $150 billion to $300 billion.
And that still would not put everything back.
Once broader economic losses and permanently destroyed institutional capacity are considered, the total consequences of dismantling and subsequently rebuilding parts of the federal government could ultimately reach several hundred billion dollars.
The important point is not whether every dollar of federal spending before Trump was justified. Obviously it wasn’t. Every large organization accumulates obsolete programs, unnecessary offices and inefficiencies. Government should constantly be looking for ways to operate better.
The important distinction is between reform and demolition.
If I renovate my kitchen because it is inefficient, I may save money.
If I bulldoze my kitchen, discover that I still need a kitchen, and then hire someone to build another one, I should not congratulate myself for having reduced kitchen expenses during the months when I couldn’t cook.
Yet something remarkably similar can happen in government accounting.
The Incredible Disappearing Savings
Imagine eliminating 100,000 government jobs.
During the first year, the payroll savings look impressive.
Then imagine that four years later another administration determines that 70,000 of those jobs actually performed necessary functions.
Those positions must now be recreated.
Recruiters must find replacements. Background investigations must be performed. Security clearances may have to be issued. Computers must be purchased. Offices must be staffed. New employees must be trained.
Some of the former employees will return.
Many won’t.
A 55-year-old engineer with 27 years of government experience who accepted retirement in 2025 may not be interested in beginning another federal career in 2029.
A young scientist whose grant disappeared may now work for a pharmaceutical company.
A cybersecurity specialist may have doubled her salary in private industry.
A Foreign Service family that uprooted itself after a program disappeared may not be willing to do it again.
And the most valuable thing those people possessed may have been something the government cannot purchase from a contractor:
memory.
They knew why a regulation contained an obscure sentence.
They remembered what happened the last time an agency attempted a particular policy.
They knew whom to telephone at another department.
They understood a computer system that had been modified repeatedly for 20 years.
They knew which contractor always underestimated costs.
They knew which congressional committee cared about which issue.
They knew where the bodies were buried—not literally, one hopes, but bureaucratically.
Organizations accumulate this knowledge slowly.
They can lose it overnight.
That makes the destruction of institutional knowledge a particularly deceptive form of government cost because it rarely appears anywhere in the federal budget.
There is no line item labeled:
Loss of 437,000 years of accumulated employee experience: $_____.
The number is nevertheless real.
Rehiring Isn’t the Same as Rebuilding
Suppose a future administration ultimately restores hundreds of thousands of federal positions.
The salaries themselves should not all be counted as reconstruction costs.
If those employees had remained in government, taxpayers would have been paying those salaries anyway.
That is simply the normal cost of operating the federal government.
The reconstruction expense is everything necessary to recreate the capability that disappeared.
Recruiting.
Hiring bonuses.
Relocation.
Training.
Security investigations.
New contracts.
Temporary contractors filling gaps.
Consultants reconstructing lost processes.
IT restoration.
Duplicated work.
Management time.
Litigation.
And years of reduced productivity while inexperienced employees learn jobs previously performed by people who had spent decades doing them.
If an agency saves $500 million by eliminating a program and subsequently spends $900 million reconstructing it, the government did not save $500 million.
It spent an additional $400 million while temporarily depriving Americans of whatever service the program provided.
That is the accounting we should eventually perform.
The Government May Have to Buy Back What It Sold
Federal property creates an even easier example.
Imagine that the government owns an office building worth $100 million.
The administration decides it doesn’t need the building and sells it for $80 million.
The Treasury receives $80 million.
Success!
Then five years later the government determines that it actually needs roughly that amount of office space.
Unfortunately, commercial property prices have risen.
The building—or its equivalent—now costs $120 million.
The taxpayers effectively sold a $100 million asset for $80 million and purchased its replacement for $120 million.
The government’s temporary $80 million windfall eventually produces a $40 million net capital loss compared with simply keeping the building, before transaction costs are even considered.
Leases can produce similar problems.
So can federal land transactions, mineral rights, energy leases and infrastructure.
Some environmental decisions can be reversed relatively inexpensively because the United States still owns the land.
A future president can restore protections to federal land much more easily than the government can recover property that has actually passed into private ownership.
Once an asset has been sold, the government cannot simply announce:
“We changed our minds. Give it back.”
The Fifth Amendment has something to say about that.
The government may have to purchase it.
And taxpayers write another check.
Science Is Worse
Government laboratories and research programs expose the biggest flaw in the theory that everything can simply be restarted.
Science operates in time.
Consider a hypothetical 10-year study following 20,000 children from birth.
Researchers collect medical, environmental and developmental information every year.
After six years, funding disappears.
The research team disperses.
Participants stop reporting.
Samples are destroyed because nobody pays to store them.
Four years later Congress restores the program.
The government can restore the funding.
It cannot restore the missing four years.
Those children grew up.
There is no appropriation Congress can pass that makes them four years younger.
Agricultural experiments have planting seasons.
Astronomical observations depend upon celestial events.
Climate measurements depend upon continuous records.
Biomedical studies depend upon patient cohorts.
Ecological studies sometimes depend upon populations that subsequently disappear.
Graduate students build careers around research programs. Eliminate the program and they do not sit quietly beside their microscopes waiting for America to change presidents.
They go somewhere else.
Some go into industry.
Some leave science.
Some leave the United States.
That is why cancelling a $10 million scientific program does not necessarily mean restoring it later costs $10 million.
Sometimes $20 million cannot restore it.
Sometimes $100 million cannot restore it.
Sometimes the information is simply gone.
Regulations Can Be Rewritten. Time Cannot.
There is another misconception about federal regulation.
A new president cannot simply sign an executive order declaring:
“Everything is now exactly as it was on January 19, 2025.”
The United States has something called administrative law.
Agencies generally have to develop rules through legally prescribed procedures.
Evidence must be gathered.
Economic impacts must be analyzed.
Proposed rules must be published.
The public must be allowed to comment.
Agencies must consider those comments.
Final rules must be justified.
Then someone sues.
That lawsuit moves through federal court.
Perhaps another rule is required.
Then another lawsuit begins.
Consequently, an environmental regulation that took six years to construct and six months to eliminate may require another several years to reconstruct.
The taxpayer cost of the bureaucratic process itself might be only a few billion dollars spread across the entire federal government.
The larger cost is lost time.
If pollution that would have been prevented occurs during those years, restoring the regulation does not remove that pollution from someone’s lungs.
If habitat is destroyed, rewriting a regulation does not instantly reconstruct the ecosystem.
If workers are injured under weakened protections, a future Labor Department cannot travel backward through time and prevent the injuries.
Public policy has hysteresis.
That is a useful engineering concept that deserves wider application to government.
The path back is not necessarily the reverse of the path in.
Then There Are the Courts
Here we reach the portion of the Trump transformation that no future president can simply reverse.
Federal judges are not agency administrators.
Article III judges generally serve for life.
A president elected in 2028 cannot fire Trump-appointed judges because he dislikes their jurisprudence.
Congress can change statutes.
Agencies can promulgate new regulations.
Future cases can distinguish or sometimes overturn earlier precedents.
Congress can create additional judgeships.
Constitutional amendments are theoretically possible.
But there is no legitimate constitutional mechanism by which President X can simply order:
“Restore the federal judiciary to its January 2025 configuration.”
Those judges may still be deciding cases in the 2040s, 2050s and perhaps beyond.
The Supreme Court presents an even more consequential version of the problem.
Its constitutional interpretations can constrain presidents who have not yet been born.
Ironically, judges themselves are relatively inexpensive.
Creating dozens of additional federal judgeships would barely register in a federal budget measured in trillions.
The financial expense isn’t the obstacle.
The institutional expense is enormous.
If Democrats eventually respond to a conservative judiciary by dramatically expanding federal courts, Republicans would inherit exactly the same power when they returned to office.
Fifty new judges become 50 more.
Nine Supreme Court justices become 13.
Then perhaps 17.
Soon judicial independence begins to resemble another congressional election.
A government can spend money rebuilding an agency.
It cannot purchase its way back to an earlier constitutional culture.
International Relationships Have Memory Too
The same problem exists overseas.
Suppose America cancels an international development program.
Employees leave.
Local offices close.
Foreign partners find other sources of funding.
China steps in.
Four years later Washington returns carrying checks and announces:
“Good news. America is back.”
The people on the other side of the table have learned something.
American commitments expire every four years.
That changes their behavior.
They demand stronger guarantees.
They diversify relationships.
They make agreements with China, Europe, Russia, regional powers or private investors.
They hedge against the possibility that the United States will disappear again after the next election.
Restoring the State Department’s budget is easy.
Restoring confidence in American continuity is much harder.
Trust is another government asset that does not appear on the federal balance sheet.
It nevertheless has enormous value.
There Is Also the Cost of Paying People to Stop
Some government reversals create an especially strange form of expenditure.
Government first spends money encouraging an activity.
Then government spends money terminating it.
Then another government may spend money restarting it.
Energy policy provides an obvious example.
Imagine Washington leases federal waters for offshore wind development.
Companies spend billions preparing projects.
A subsequent administration decides it doesn’t want the projects and pays billions resolving or terminating those arrangements.
A later administration again decides that offshore wind is important.
It must auction leases again.
But developers now understand that an American federal lease may not represent a stable long-term commitment.
Investors price political risk.
They demand higher returns.
Financing becomes more expensive.
The government may have to provide better terms to attract them.
The taxpayers can therefore pay three times:
Once to create the program.
Again to dismantle it.
And a third time to reconstruct it.
This isn’t primarily a Democratic or Republican problem.
It is a policy-volatility problem.
Imagine building an interstate highway halfway across South Dakota.
A new administration decides highways are wasteful and pays contractors to tear it up.
Four years later another administration decides the highway was necessary after all.
Nobody would describe the resulting reconstruction expense as prudent fiscal management.
Yet when the infrastructure is institutional rather than concrete, the destruction is harder to see.
Government Is Infrastructure
This may be the fundamental conceptual error.
Americans understand physical infrastructure.
We understand that neglecting a bridge eventually costs money.
If painting a steel bridge costs $10 million and we refuse to spend the money, that looks like a $10 million saving.
Twenty years later replacing the corroded bridge costs $400 million.
Nobody would seriously argue that the government saved $10 million.
Institutional infrastructure works similarly.
The Centers for Disease Control and Prevention are infrastructure.
The National Weather Service is infrastructure.
Federal scientific laboratories are infrastructure.
The diplomatic corps is infrastructure.
The air traffic control system is infrastructure.
The people who inspect food are infrastructure.
The engineers who regulate nuclear reactors are infrastructure.
The employees who understand Social Security’s ancient computer systems are infrastructure.
The analysts who monitor financial institutions are infrastructure.
Government capability is a capital asset even when accountants do not put it on a balance sheet.
Destroying capability therefore creates a liability.
Some Trump Changes May Be Worth Keeping
Any serious reconstruction effort should acknowledge something else.
A future administration should not blindly restore everything.
That would be foolish.
Perhaps Trump eliminated programs that deserved elimination.
Perhaps some agencies really were overstaffed.
Perhaps buildings genuinely weren’t needed.
Perhaps regulations imposed costs exceeding their benefits.
Perhaps federal processes became more efficient.
A competent successor should preserve successful reforms regardless of who created them.
If Trump discovers that an agency can perform the same mission with 7,000 employees instead of 10,000, the next Democratic president shouldn’t hire 3,000 people simply because 10,000 existed under Biden.
Government isn’t a historical reenactment.
The objective should be restoration of capability, not restoration of every organizational chart.
This distinction also prevents the argument from degenerating into partisan arithmetic.
The question isn’t:
“How much money did Trump cost?”
The better question is:
What government capabilities were eliminated, what benefits resulted, what costs resulted, and what would taxpayers have to spend if the country subsequently decided those capabilities were necessary after all?
That audit should eventually happen.
And it should be ruthless in both directions.
If a Trump reform saved taxpayers money without materially reducing government performance, keep it.
If a program was eliminated and nobody misses it, don’t resurrect it.
If an agency can operate efficiently with fewer people, leave it smaller.
But where dismantling created false savings, those savings should be identified as such.
The $140 Billion Question
Through August 2026, I would tentatively place the extraordinary cost of reconstructing something resembling the pre-Trump federal institutional structure at approximately $100 billion to $200 billion.
Call the midpoint $140 billion.
That estimate is necessarily speculative.
Nobody knows which Trump policies will survive.
Nobody knows what Congress will do.
Nobody knows how many employees will ultimately leave.
Nobody knows which properties will be sold.
Nobody knows which contracts will be terminated.
Nobody knows how courts will rule.
And, most importantly, Trump still has roughly two and a half years remaining in his term.
If the current trajectory continues, I can imagine a 2029 reconstruction program costing $150 billion to $300 billion in extraordinary federal expenditures.
The larger economic consequences could be considerably greater.
Several hundred billion dollars would not strike me as implausible once lost research, disrupted businesses, cancelled grants, workforce disruption, delayed projects, litigation and other secondary effects are included.
But even that number fails to capture everything.
How much is a lost scientific discovery worth?
How much is 30 years of expertise possessed by a nuclear safety engineer worth?
What is the price of an international ally deciding that an American promise lasts only until the next presidential election?
How much is a four-year hole in a climate dataset worth?
What is the monetary value of legal precedent that survives for 40 years?
How much does it cost to reconstruct public confidence in an institution?
There are no satisfactory entries for these things in the federal budget.
The Lesson Goes Beyond Trump
And this may ultimately be the most important lesson.
America needs to become much better at distinguishing reducing government spending from reducing government capability.
They are not synonymous.
Sometimes eliminating capability is exactly what taxpayers want.
But when we eliminate an institution that the country later discovers it still needs, we haven’t necessarily saved money.
We may simply have deferred the bill.
This principle should apply regardless of which party occupies the White House.
A future Democratic administration could make precisely the same mistake in reverse—rapidly constructing enormous institutions that a Republican administration subsequently dismantles.
Then Democrats rebuild them.
Then Republicans dismantle them.
Each party claims savings when it tears down what the previous party built.
Meanwhile taxpayers finance both construction crews and demolition crews.
That is madness.
A wealthy country can afford disagreement.
It cannot indefinitely afford to reconstruct its government every four years.
The Final Invoice
Someday the Trump era will end.
Whether that happens politically in 2029 or whether its institutional influence continues for decades is another question.
When it does, Americans should resist two equally simplistic narratives.
The first will say that Trump permanently fixed a bloated government and saved enormous sums of money.
The second will say that the next administration merely needs to restore everything Trump eliminated.
Neither is likely to be true.
Some reforms will probably prove valuable.
Some damage will prove reversible.
Some damage will be expensive to reverse.
And some things will never be restored.
People retire.
Scientists change careers.
Buildings are sold.
Companies disappear.
Relationships change.
Research windows close.
Habitats vanish.
Court precedents survive.
Knowledge is forgotten.
Time passes.
A government can appropriate another $200 billion.
It cannot appropriate another 2025.
That is the real problem with treating the federal government as though it were a collection of unnecessary expenses that can be switched off and switched back on whenever political control changes.
Government is not merely spending.
Government is accumulated capacity.
America spent generations building some of that capacity. It certainly built too much in some places, built it badly in others and occasionally continued programs long after their usefulness expired.
Reforming those institutions is legitimate.
Eliminating unnecessary ones is legitimate.
But demolition should never be confused with savings simply because the demolition occurs in one fiscal year and the reconstruction appears on somebody else’s budget four years later.
If the next administration eventually spends $200 billion rebuilding institutions dismantled during the Trump years, that money should not be remembered simply as the new president’s spending.
Part of it belongs on the historical ledger of the decision to dismantle those institutions in the first place.
Because when someone knocks down a house, the price of rebuilding it is part of the cost of knocking it down.
The United States may eventually discover that lesson on an extraordinary scale.
And when the final invoice arrives, the most sobering line may not be what Donald Trump spent.
It may be what America has to spend putting itself back together.
Leave a comment