A farm needs two depreciation figures and not one. The figure on the return was built to work out taxable income, and the FAO handbook on cost of production statistics states what happens when the same figure is asked to do a second job: policy goals drive the tax version, and “the two methods provide similar results only by coincidence”.
The figure already exists on your farm, and somebody outside the gate chose it. The accountant applied the rule the country requires and applied it correctly, because that number is there to settle tax. It carries that purpose with it when it crosses into the cost of a cycle. A combine harvester deducted in one good year then shows as costing nothing in the years after, the lowest price the farm can accept drifts down, and nobody decided it should drift. Two cycles later the machine has to be replaced, the money is not there, and no sheet anywhere on the farm ever said how many years that machine was meant to last or what those years were costing.
Kentucky’s agricultural economists put the consequence in a single line. Ellis, Shockley and Shepherd write that “If tax depreciation is used for financial statements (e.g. income statements), farm performance measurements, including profitability, will be inaccurate”. The same publication draws the line this page stays behind: “Decisions about appropriate depreciation methods should be discussed with a tax professional”. What follows is not that decision. It is the farm writing down how long it plans to keep what it owns.
Why the filed figure cannot do this job
The filed figure was built to move income between years, not to measure what a machine lost. Ellis and Shockley give the cleanest case of the two pulling apart: a new grain truck can be deducted in full in the year of purchase under the first-year rule the United States allows, and if the owner sold it twelve months later it “would still have considerable value”. The truck lost something real that year. The return says it lost everything, and the two statements are both correct, for different questions.
Why the line gets left blank anyway is the first thing the Kentucky publication says, and the reason it cannot stay blank comes in the same breath: depreciation is easy to overlook on a farm and still “plays a major role in both taxes and profitability”. It is not small enough to leave out while the rest of the cost sheet gets built. Which rule produced the figure now on your return is written down on the record of the tax regime, and that record is where the first number keeps living.
Who decides how many years, and where that number comes from
You decide, and the definition is short enough to copy onto the sheet. Ellis, Shockley and Shepherd define it as ownership rather than engineering: “The economic life of new machinery is equal to the length of time you plan to own that machine before selling or replacing”. Nobody at a desk can answer that. The dealer cannot, the accountant cannot, and the manufacturer never tried to.
Defaults exist and they are honest about being defaults. The same authors record that the state’s farm business management program “uses a standard 10-year economic life on most farm machines”, and in the same sentence that economic life varies drastically depending on the operation. A default you adopt knowingly and a default that arrived by itself produce the same arithmetic and are not the same sheet. That is why the years column has a second column beside it, holding one short phrase: your own habit of replacement, what a neighbor gets out of the same machine, or the published table you read. The phrase is what the next person can argue with.
The hours hiding behind the years
Every published life in years carries an assumption about hours per year, and that assumption is usually invisible. Conab, the Brazilian public body that publishes the country’s cost of production series, prints both columns for the same machine: a tractor at ten years and 15,000 hours, a combine harvester at ten years and 5,000 hours, with residual values of 25 and 20 percent of the new price. Divide one column by the other, which is arithmetic done here and not a statement by Conab, and the ten years for a tractor assume 1,500 hours a year. A tractor that runs 400 hours a year is not ten years old in that table’s sense when it turns ten.
Agricultural economists have a name for the ratio. The Faculty of Agricultural Sciences at the Universidad Nacional de Asunción, in Paraguay, calls it the equalization point, the hours of useful life divided by the years of useful life, and works a tractor rated at 12 years or 12,000 hours, which puts the point at 1,000 hours a year. Above that line the machine wears out before the calendar runs out and the loss behaves as a cost of running. Below it, the machine is not taking the kind of wear that justifies replacing it, and what is eating the value is age and obsolescence rather than use.
The choice between the two is old, and it is a choice. Robb, Smith and Ellis, describing a whole farm machinery cost program in 1998, split the loss into “loss in value as a function of usage” and loss to technological obsolescence, note that depreciation is a critical element of machinery ownership cost and that the calculation method can matter, and put the decision on the input screen as a switch the user has to set, time or use. Hours per machine per year is the one input this sheet borrows from somewhere else, and it is the same reading that produces the repair bill for one machine.
What goes in the residual column, and how exact it has to be
An estimate, expressed as a share of the new price, and the sheet survives being wrong about it. The FAO handbook sets out the straight-line method in one sentence: the difference between the purchase price of the asset and its estimated value at the end of its expected life, divided by the number of years of expected service life. Two of those three numbers are on documents you already hold. The third is a guess, and the handbook is candid that when no reliable market data exists the statistician has to make assumptions about the service life of the asset and the rate at which it loses its service capacity. A national statistician guesses this. So can you.
Robb, Smith and Ellis say the same thing from the other side, calling the estimate of current and future salvage values a challenge in a paper whose whole subject is calculating machinery cost precisely.
Four published tables, and the question each was built for
Published references exist for the guess, and each was built for a different question, which is the reason none of them is the answer to yours.
| Who publishes it, and where | What it gives | What it was built for |
|---|---|---|
| FAO, international | The method: purchase price less estimated end value, over the years of expected service life | Making national cost estimates comparable between countries |
| University of Kentucky, United States | Salvage value as a percentage of purchase price, by implement type, machine age and annual hours | Farm financial statements, enterprise budgets and net farm income |
| Conab, Brazil | Useful life in years and in hours, plus residual value, for machines, implements, buildings and irrigation sets | The public cost of production series for Brazilian crops |
| Article 31 of Decree 6105/2021, Paraguay, as read by the Universidad Nacional de Asunción | Ten years of useful life and 20 percent residual value for machinery and implements | Working out the deduction the taxpayer may claim |
Three of those four were built by public bodies for a purpose that is not tax, and the fourth is the tax rule itself, sitting in the table so the difference is visible. The Paraguayan authors add the warning that matters most to a reader holding a table: published lives are a reference, and they can be replaced by local data such as the farm’s own records.
Why the origin of the number is a column and not a footnote
Two cycles of the same farm stop being comparable without it, and the people who do this for a living hit the same wall. Johnson, Huang and Grant screened 372 studies of farm-level financial decisions published between 1960 and 2024 and kept 61, and found the treatment of fixed costs varying from study to study, with depreciation and rent among the fixed costs that gross margin analysis leaves out entirely. Their reading of what that costs is blunt: many studies do not give enough methodological detail, which limits how far anyone can replicate or apply them.
This is a review of published research on one enterprise, cow-calf, and it describes how those 61 papers were written rather than how farms keep books. It transfers on one point only, and that point is the whole reason for the column: when the criterion behind a cost figure is not written next to the figure, nobody downstream can tell whether two numbers are comparable or merely similar. A farm reading its own sheet two cycles later is downstream of itself.
What the annual figure is for once it exists
The annual charge goes into the outer column of the floor price, and it goes against the investment plan. It is one of the ownership lines that the break-even price receives already worked out and never calculates on its own, and it is the sum that a written investment plan should have been setting aside every year for the replacement it lists. Both of those documents fail quietly when this line is missing, because the arithmetic still completes.
Worked through once, in the currency the farm keeps its books in: a tractor on the purchase invoice at 480,000, a useful life the farm declares as ten years with the dealer’s replacement advice written beside it as the origin, and a residual estimated at 25 percent, which is 120,000. Subtract the residual from the invoice and 360,000 is what the declared years have to absorb.
Divided by ten, the annual charge is 36,000. Declare twelve years instead, on the same machine and the same invoice, and the charge falls to 30,000. Nothing about the tractor changed between those two lines. What changed is a number somebody on the farm chose, which is the whole reason it belongs in a column with a date and a name and not in somebody’s head.
What the sheet produces on its own is smaller than either of those documents and harder to argue with: a number per year, per machine, that somebody signed. Declaring what a machine is expected to do, checking the result against it and rewriting the declaration with a name attached is the last of the four functions of management applied to equipment, and it is the same movement the rest of the finance axis runs on.
Where to start
Two to three hours, with the folder of purchase invoices on the table. Nobody outside the gate has to be called, because all four inputs are decisions of whoever operates the machine.
The sheet is worth most in the year you are wrong. A machine kept twelve years when the sheet said eight is not a failure of the sheet. It is the only way anyone finds out that the replacement habit and the replacement plan were two different things. Nothing on the return will ever tell you that, because the return was never asked.