The deduction you skipped still comes out of your basis

The phrase sets the measure by which basis falls over the life of a property. Publication 946 puts it directly: you reduce basis by the depreciation allowed or allowable, whichever is greater.

Allowed covers what you actually deducted and got a benefit from. Allowable covers what you could have deducted.

Because allowed or allowable takes the greater of the two. Somebody who claimed nothing still reduces basis by everything they could have claimed.

So you lose the deduction in the years it belonged to, and lose the basis anyway at the end. That is the double cost.

Nothing about living abroad softens allowed or allowable. The property is American and the rules follow the property.

So the instinct to skip depreciation and keep the gain small does exactly the opposite of what people intend.

So the phrase is doing real work. Two words decide whether a decade of deductions exists at all.

Residential rental property runs off over 27.5 years under the general system, straight line, with a mid-month convention in the year it enters and leaves service.

Publication 527 carries the tables. On a building worth several hundred thousand dollars, the annual figure is substantial.

Only the building depreciates. Land does not, so the purchase needs splitting between the two before anything else happens.

So a decade of unclaimed deductions is a large number by the time anybody looks at it.

Improvements have their own schedules on top. A new roof does not simply join the building's run.

Take an American in Bristol who has let a house in Atlanta for nine years and never claimed a cent of depreciation. The building portion of her basis is $300,000.

She assumed her basis was still $300,000 and her gain would run from it. Allowed or allowable said otherwise.

Basis fell by the allowable amount each year regardless, so by the time she sells it has dropped by roughly $98,000 of depreciation she never took.

Her gain is $98,000 larger than she expected, and she got no deduction for any of it. The theory cost her twice over.

Sometimes, and it depends on how long the treatment has been running. A single year handled incorrectly is often an error an amended return can fix in the ordinary way.

A treatment repeated consistently over years is different, because depreciation hardens into a method of accounting once it settles into a pattern.

No amendment corrects a method. You change it instead, through a formal process with its own form.

So establish which situation you are in before filing anything. Getting that wrong wastes the correction as well as the time.

Form 3115 is the application, and a depreciation change is squarely within what it covers.

The mechanics turn on whether the change qualifies for the automatic procedure. Most depreciation catch-ups do.

An automatic change carries no user fee. The original form goes with the timely filed return for the year of change, and a signed copy goes separately to the National Office.

A non-automatic change is a different animal: you file during the year of change, pay a fee, and wait for a letter ruling.

So check which route applies before drafting anything. The two procedures share a form and little else.

This is the part that makes the exercise worth doing. A change of method ordinarily carries an adjustment reconciling the old treatment to the new one.

Where the adjustment is in the taxpayer's favor, the instructions take it into account in a single year, the year of change.

So nine years of missed depreciation can arrive as one deduction rather than dribbling out or vanishing.

An adjustment running the other way spreads over four years instead, which is a kinder timetable than it sounds.

The catch-up lands in the year of change, so it needs income to land against. A year with little American taxable income wastes it.

That is a real risk for somebody living in Britain whose credits or exclusion already shelter the rest of their income.

Doing it in the year of sale has a certain logic, because the gain gives the deduction somewhere to go.

So model the year before choosing it. The relief is only as good as what it offsets.

So choose the year deliberately rather than taking the next one. A deduction with nothing to offset is a deduction wasted.

Gain runs from the reduced basis, so the depreciation shows up as a larger gain whether or not anybody claimed it.

Part of that gain traces back to depreciation and carries its own rules rather than the ordinary long-term rates.

Our note on selling US property covers the withholding mechanics on a sale by somebody living abroad.

So the sale is where the whole question becomes concrete. Everything before it is preparation.

So compute the adjusted basis before agreeing a price. The number people carry in their heads is almost always the original cost.

Differently, and that is the point most often missed. Britain does not give depreciation on residential property in the way the American system does.

So the profit reported here and the profit reported there will not match, and neither will the gain on sale.

Our note on a US rental as a UK resident covers the British computation.

So keep one set of underlying records and two computations from it. They are measuring the same house under different rules.

The American figures are in dollars and the British ones in sterling, including the basis and the depreciation.

A consistent conversion method across the whole holding period matters more than precision in any one year.

Our note on which exchange rate to use covers the options.

So record the method alongside the figures. A rate nobody can evidence invites argument years later.

The closing statement from the purchase, which fixes the original cost. Without it nothing else can be computed.

The split between land and building, which usually comes from the assessment at the time.

The date the property first went on the market, since depreciation starts at service rather than at purchase.

Every return for the years it has been let, so the pattern of treatment is visible rather than assumed.

So start the hunt early. Closing statements from fifteen years ago take time to find.

It is not a penalty. Nobody is being punished for failing to claim; the basis rule simply runs on its own.

It is not a choice either. Depreciation on let property is not something a taxpayer elects into each year.

And it is not unfixable. The method change exists precisely for this situation.

So treat it as arithmetic rather than as a problem with blame attached. The arithmetic is what it is.

Then none of this applies, because depreciation runs from the date a property goes into service as a rental.

A house left empty while somebody decides what to do with it is not in service. Nothing accrues and nothing comes off basis.

The clock starts when the property is available to let, not when the owner moves out.

So fix that date carefully. It decides the first year and therefore every year after it.

Advertising it counts for this purpose even if nobody takes it. Availability is the test, not occupancy.

A property let for part of a year, or let for part of its area, carries a proportionate deduction rather than the full one.

The same applies where the owner occupies it for a spell between tenants. Personal use reduces what is allowable.

That proportion has to be worked out and recorded each year rather than assumed to be constant.

So keep the letting dates. They are the evidence for the only figure anybody will question.

A spell of refurbishment between tenants is its own question again, and the answer turns on whether the property stayed available to let throughout.

The first is believing that skipping depreciation protects basis. It does not, and the belief is expensive.

The second is amending years that long ago hardened into a method. The amendment does not achieve the correction.

The third is taking the catch-up in a year with nothing for it to offset. The deduction is real but it needs income.

A fourth is depreciating the land along with the building, which overstates the deduction and creates a different problem entirely.

We start by rebuilding what should have been claimed from the purchase documents, year by year.

Then we establish whether the position is an error or a method, because that decides the route entirely.

Our US federal returns work covers the computation and the form that carries it, and you can get in touch well before a sale rather than during one.

So the time to look at this is now rather than at completion. A conveyance will not wait for a method change.