Capital purchases dropped by 151,000 US dollars per farm in two years on the Illinois grain farms whose books the Farm Business Farm Management Association keeps, from 335,000 in 2023 to 236,000 in 2024 and 184,000 in 2025. That subtraction is ours; the three figures are the ones the August 2026 bulletin prints. The four authors read the series as a pattern rather than an accident, writing that higher capital purchases are often associated with higher net farm income, with capital purchases declining when net farm incomes decrease, often in a lagged fashion.
The dealer shows up in the year that paid, and whoever keeps the books calls in December with a number that makes a purchase look cheap. A bulletin from the same group on machinery costs states the mechanism without softening it: costs increase more rapidly during high income periods as producers make machinery investments and manage their taxable income. Two years on, when prices turn and nothing is left over, the item that was genuinely needed gets cut along with everything else, because no page ever put it ahead of the others.
What answers the question in the title is one sheet with one row per item, and five things on the row: what it is, what it costs, the year it is meant to happen, the condition that has to be true for it to leave the paper, and the name of the person who signs. The fourth is the one nobody writes and the only one that makes the sheet do work.
What decides the purchase when nothing is written
Two triggers, and both of them are about money that has already arrived. The first is cash on hand at the end of a paying year. The second is the tax bill, which is the one the machinery bulletin names when it puts investment and taxable income in the same sentence. Both triggers are real, both are dated, and neither asks whether the item was needed. A farm without a written list is not a farm without a decision rule. It is a farm running a decision rule it never agreed to, whose only input is the balance of the account.
| Year | Capital purchases per farm, US dollars |
|---|---|
| 2023 | 335,000 |
| 2024 | 236,000 |
| 2025 | 184,000 |
The same bulletin adds what the purchase keeps costing afterwards, on a narrower set of farms that it charts on its own, the high-productivity ones in central Illinois. There, in 2025, depreciation averaged $91 per acre, representing 53% of the $170 machinery-related costs for growing corn, so more than half of what an acre of corn pays for iron is the price of iron bought in earlier years. Every row that leaves the paper signs that standing charge for years forward.
The bulletin also prints the comparison almost no list carries, back on the whole enrolled group. Over the past ten years there, average annual capital purchases have exceeded average annual net farm income, with government payments included in income, and the gap was not closed by borrowing: debt rose, the value of the assets rose with it, and the average debt-to-asset ratio stayed under 20 percent. It was closed from outside the gate, by off-farm income by family members associated with the farming operation, as well as income from other businesses. That puts one more line on the sheet, under all the rows: what the rows ask for in each year, beside what the operation expects to generate that year.
When the first number is the larger one, the sheet is already saying that part of the money comes from somewhere else, and it is better for it to say so in writing than in December.
The failure is the order, not the amount
Money spent in the wrong order does more damage than money not spent. Zdenek and Lososova followed a Czech farm sample that ran between 85 and 149 holdings depending on the year, from 2003 to 2016, for the journal Agricultural Economics, published by the Czech Academy of Agricultural Sciences, and found investment flowing to land while the built structures aged: purchases of land are likely to affect the growth of the relative age of tangible fixed assets, especially the obsolescence of buildings that are not sufficiently resourced by farms. On what governs the total, they are blunter still, reporting that the volume of investment depends on available own sources of funding, access to bank loans and the volume of grants.
Land was unusually attractive in that country for reasons of its own, since a high share of the ground was rented and farms were buying it back, and the authors write that the effect on asset age is likely rather than proven. The part that travels is the shape. A list with an order on it does not stop a farm from buying land. It stops the farm from finding out, in the year the roof fails, that the roof was never on any page.
Where the year on each row comes from
From the production plan, not from the machine. What the operation intends to produce over the next few years is what decides which structure has to exist and by when, which is the work that a multi-year production plan does, and the investment list is the buying side of those same decisions. Storage is the structure row whose condition is easiest to write as a checkable number, because what a month of storage costs comes out of bills the farm already pays.
A row with no year is not an investment. It is a wish with a price on it. Erosion control is the row that most often carries neither, because it gets decided in the week after a rain and paid for out of whatever was left over, and a written soil conservation plan is what gives those rows a year before the rain instead of after it.
The budget is the other end of the same cable. An item that never appears as a line in a declared price assumption has no year in any operative sense, because nothing in the farm’s cash arithmetic is holding room for it. Writing the year on the list and never carrying it into the budget is how a plan stays true on paper and false in the account.
Why the condition is a number, and why it tightens
Because waiting has a value, and the value rises with uncertainty rather than falling. Spiegel, Britz and Finger, modeling one arable farm in northern Germany for the journal Q Open, put it as higher volatility, or a higher perceived risk level, increases both the option value and the trigger price that must be reached in order to initiate investment. Read that toward the sheet. The less predictable the price of what the farm sells, the higher the number the row should demand before the purchase leaves the paper.
That runs against what happens on most farms, where a volatile year that happened to end well is the year the purse opens widest. The study is a simulation of one farm type and one crop, so its figures are not anybody’s figures and were never meant to be. The direction is what survives the caveat: uncertainty raises the bar to be cleared, and a farm that writes the bar down before the volatile year is the only farm that can tell afterwards whether it cleared it or moved it.
Writing the bar down leaves a second thing on the table, and most lists sweep it off. Setting a condition means something else was ruled out on the same day, at a figure somebody had in mind, and that figure has a page of its own. The option you turned down is where it gets kept, with the source beside it and a date, which is what lets the next reading of this list say whether the bar moved or the alternative did.
What turns a feeling into a figure
An instrument that compares one change against carrying on. Penn State Extension defines it as a planning and decision-making framework used to compare the costs and benefits of alternative management decisions faced by a farm business, organized into four lines: the costs the change adds, the income it removes, the income it adds, and the costs it removes. Four lines are enough to state a condition somebody else can check, of the form that this row leaves the paper when a named figure reaches a named level.
The same guide draws its own limit, which is the part worth carrying. It states that partial budgeting does not account for changes in the value of money over time and sends decisions whose effects run past one or two years to a method that treats money arriving in five years as worth less than money arriving now. Most rows on an investment list are that kind of decision, which is the honest reason the condition column holds a trigger to be checked and never a projected return.
Who signs the row
A named person, and not the same one on every row. The Australian succession review that a written succession plan works from names the purchase of capital equipment, harvesters among it as one of seven operational decisions a farm has to settle explicitly, alongside choosing stock and crops, new permanent infrastructure, buying or selling land, when and where the product is sold, whether to employ labour and whether to go to off-farm income.
Turning those seven into rows with a name and a date on each is the succession page’s job. The investment list needs the same column for its own reason: without it the list describes purchases that nobody in particular is authorised to make, which in practice means whoever is standing in front of the dealer.
December is where the two sheets meet. The machinery bulletin puts making the investment and managing the taxable income in one breath, for the good reason that most farms do both in a single conversation, and the second half of that conversation runs on a choice that was usually never written down anywhere, which is exactly what the record of the tax regime sets out to fix. One page decides what to buy. The other records the rules the buying gets counted under. A farm holding only the second is running a tax plan under an investment plan’s name.
When the condition is not met and the row goes ahead anyway
Write it down twice, on the day it happens. Record the date, the person who decided, the figure the condition asked for and the figure that was actually true. A farm that buys well and a farm that buys badly look identical in the year of the purchase, and they keep looking identical unless somebody wrote the number that was supposed to be reached. Comparing the two figures afterwards is the check step of farm management, and it is the step an investment list skips more often than any other, because by then the machine is in the shed and the question feels settled.
The review date belongs on the sheet itself, beside the date it was written. A list read once a year, with the rows that fired marked and the conditions that turned out wrong rewritten, is a document. A list written once and filed is a photograph of an opinion from a year that has ended, which is the defect that costs every other document in the planning axis its usefulness first.
Where to start
Two hours, with the machinery and building inventory and the old purchase invoices on the table. One page, one row per item, five columns, and a date at the bottom.
A good year is real information and nothing here says to ignore it. What it is not is a reason. It tells the farm what it can afford this December, and it says nothing at all about what the farm needed, which is a question that had to be answered while the price was still unknown. The sheet exists so the two get asked on different days, and so the second answer can be found in writing by somebody who was not in the room when the first one was given.