Nobody can answer that from a table, and the people who print the tables say so before anyone else does. Mississippi State University Extension publishes the engineering factors farmers reach for when they have nothing else, and it puts the order of preference in the same paragraph as the factors: repair costs are best estimated “using detailed history for the individual piece of equipment because of differences in manufacturers, management factors, and working conditions”. The estimate is the fallback. Your own paperwork is the answer.
That paperwork is in four places at once, which is the actual reason the number is missing. Some invoices went to the bookkeeper, because somebody entered them. Some sit in the office drawer, because they were paid in cash and never traveled further. For part of it no invoice stayed on the farm at all, and what exists is a line on a bank or card statement.
The largest ones never reached the farm: the dealer did the work, emailed the file, and it is still in an inbox nobody searches. When the machine starts giving trouble and somebody asks whether it is worth keeping, the conversation runs on impressions. One person remembers a fortune. Another remembers one transmission. Neither can put a figure on the table, and an impression does not survive a lender’s question or a buyer’s.
What ends this is one sheet for one machine, and it is built from documents that already exist.
Why the published table cannot answer for your machine
Three separate publishers of machinery cost estimates rank the farm’s own record above their own tables, and they give the same reason. The University of Wisconsin-Extension guide to field machinery costs concedes the ranking on its first page, before it prints a single table: “The most accurate method of determining machine costs is complete records of the actual costs incurred. Estimating costs is an alternative”. Its repair section gives the reason: “Repair costs depend on hours of annual use, and are difficult to predict because operators differ greatly in the levels of repair and maintenance they do”.
Robb, Smith and Ellis, writing in the Journal of Natural Resources and Life Sciences Education about a whole-farm machinery costing model, open the same subject with one short sentence: “Maintenance and repair costs are difficult estimations”. Their model offers the user a choice between a calculated repair cost and one the user types in, and the authors attach a condition to their own feature. The typed-in figure “should only be used if actual records are available”, or if the calculated one looks wrong.
Mississippi State closes its publication on the same note: “Actual farm records are the preferred source of data for calculations”, and the engineering factors are what you use when records are not available. Read those three together and the ranking is not in dispute. What is in dispute is whether your farm can produce the record, and that is a question about where paper sits, not about arithmetic.
Where the expensive half of the record actually lives
The work a farm’s own paperwork most reliably misses is the work that was not done on the farm, and the clearest evidence for that comes from a study that had to go and collect it. Calcante, Fontanini and Mazzetto set out to recalculate the machinery repair parameters for Italian conditions, using 100 four-wheel-drive tractors and 20 self-propelled combine harvesters. They could not close the data from owners alone. They used two sources: filling in forms with the owners, and querying the databases of dealers and authorized workshops.
The second source is the finding. Those dealer databases “represented the most complete source of repair and maintenance activities”, and the activities they were most complete about were the scheduled and the extraordinary ones, which the authors describe as work “that are rarely performed in farms”. A farm workshop handles grease, oil and the small repair. The clutch, the rebuild and the harvest breakdown go somewhere else, and the record goes with them.
Argentine work reaches the same shape from the other direction. Frank, in the Argentine national agricultural research institute’s journal RIA, splits machine upkeep into scheduled tasks, defined as those whose timing the manufacturer’s manual states explicitly, and unscheduled ones. Accumulated spending on the scheduled tasks tracks a straight line and barely leaves it. Accumulated spending on the unscheduled ones, he writes, is plainly random and can spread widely around the expected figure. The spending nobody can forecast is the unscheduled kind, and both it and the scheduled work leave the property. That is why the sheet does not close without what the dealer holds.
One warning belongs with those numbers before anything else is done with them. In the Italian data, labor, parts and total accumulated repair cost varied so widely from one machine to the next that no average describes any single machine, and the authors name what drives the spread: whether the scheduled maintenance plan was actually followed, how powerful the machine is, how hard and how continuously it is worked, and the operator. Nobody’s average is your number. That is the argument for building your own, not against measuring at all.
The four places the paper is, in the order they pay off
Work the sources in the order that returns the most paper per hour spent, and stop only when the fourth is done, because the fourth is the one that changes the total. Each of these is a document the farm can already reach today.
| Where to look | What it gives you | What it will not have |
|---|---|---|
| The bank and card statements for the cycle | Every payment that left an account, with a date and a payee | Anything paid in cash, paid by barter, or paid from another account |
| The bookkeeper’s file | Entered invoices, with amounts, dates and supplier names | Everything that was never sent to them |
| The office drawer and the glovebox | Cash invoices, the roadside repair, the small workshop | Nothing, and no order either. This is where sorting costs the most time |
| The dealer’s service history, pulled by serial or chassis number | Scheduled and major work, parts and labor separated, usually with the hour reading at the time of the job | Work done by any other workshop, and anything you fitted yourself |
The fourth row is a phone call, and the number that unlocks it is stamped on the machine. It is also the row that most often arrives as a list the farm has never seen, covering jobs the owner remembers approving and had no figure for.
Hours, not the year, is the denominator
Divide the total by hours run, not by the calendar, because a machine that sat idle did not consume its share of anything. Mississippi State gives the operation in one line: total cost per hour comes from dividing total annual cost by the estimated annual use in hours. Wisconsin gives the same instruction for a tractor, where the repair cost per hour must be calculated for the entire year and then divided by the total annual use in hours before any of it is charged to the implements that tractor pulled.
The Italian study went further and is worth copying on this point. “Unlike other papers, where R&M costs were grouped on an annual basis”, they tied each cost to the working hours on the machine at the moment of the job. That is a one-column change on your sheet, the hour reading written next to each invoice, and it is what later tells you whether the spending is climbing with age or was one bad month.
Two readings of the hour meter close the arithmetic, one at the start of the cycle and one at the end. The difference between them is the denominator, the total added up from the invoices is the numerator, and what that machine cost per hour to keep running is one divided by the other, in whatever currency the farm keeps its books in. That is the whole calculation, and the same pair of meter readings is already the denominator of fuel bought against hours worked. The standard that the money was spent against, or spent instead of, is the factory service interval.
Worked through once, in the currency the farm keeps its books in: the meter read 4,180 hours at the opening of the cycle and 4,610 at the close, so the denominator is 430 hours. Three workshop invoices and one line from the dealer add up to 12,900. That is 30 per hour. The number is worth nothing against a neighbor’s and is not meant to be, because his hours were counted on another machine doing other work. It is worth everything against the same machine next cycle, which is the only comparison the sheet was built to make.
The published estimate exists, it is public, and it is coarse
Knowing what the fallback looks like is what stops anyone treating it as the answer. It has one shape, and it is easy to recognize: a fixed percentage of the new list price, charged against a block of use hours. Mississippi State publishes the factor that way, split by machine type and power band, and works an example through to a figure for the year. Wisconsin publishes the same shape for implements, as an hourly cost for each thousand dollars of list price, on a table that moves with annual hours and with the working life you assume for the machine.
Argentina publishes a parameter of exactly the same shape. Frank defines it as the ratio between a machine’s average hourly repair and lubrication spending across its working life and its price as new, which is the same move: a percentage of what the machine cost new, spread over hours. Two continents, one method, and both of them describe a number built to fill a gap in somebody’s cost survey.
Notice what that parameter is for. It belongs to whoever has to publish a cost of production for a region and cannot phone every farm. Yours is for a decision about one machine, and the two jobs need different accuracy. Notice also which half of the machine it covers: Mississippi State keeps ownership costs and operating costs in separate columns, and repairs are on the operating side. What the machine costs to exist, before it turns a wheel, is a different sheet, built from purchase value and from the useful life you declared, and this one does not attempt it.
The line under the total, which is what makes the total honest
Write what is not in the number, in its own line, or the number quietly claims to be complete. Four things fall out of a first sheet almost every time. Tires bought inside a larger order with no separate line. Filters and oil bought at the depot on a mixed invoice. The day an employee spent under the machine instead of in the field. Work paid by barter or settled against something else. None of those is a failure of record keeping. Each is a real cost sitting in a document that describes something else.
Grouping is the recognized way out, and it is written into international guidance for exactly this case. The FAO handbook on agricultural cost of production statistics lists it among good practices: when machinery expense cannot be tied to one product, the statistician could group “fuel, lubricants and repair and maintenance expenses” and then split the group by a stated key, naming as candidates the proportion of land cultivated, hours used, or the number of machine field passes for each crop.
The rule that matters for your sheet is the order: the total comes first, whole and per machine, and the split between activities is a second decision, made against a declared split rule written before anyone sees the total, not a way of tidying the total afterward.
The finished sheet is a supplier to other calculations, and that is where it earns its afternoon back. One machine’s cost per hour is a line that the break-even price receives already worked out rather than estimating, and the rest of what it feeds sits in Finance.
Where to start, and what closing it looks like
Pick one machine, the one you have been arguing about. Allow three to four hours for the first machine in the first cycle, most of it waiting on the dealer’s service history, and under an hour in the cycles after that, because by then the line gets written when the invoice arrives instead of a year later.
What you now hold is one number with a boundary written around it, which is a different object from a number. It can be handed to a lender, put in front of a buyer, or set beside the same sheet for the same machine next cycle, and in each case the list underneath tells the reader what it does not cover. The person who signed it can be asked about it.
The harder work starts when the sheet is finished, because a figure with nobody attached to it changes nothing. Somebody has to decide what happens next: the machine goes back on the scheduled interval, or the next cycle gets a running sheet instead of an archaeology project, or the question of keeping it goes on a calendar with a date. Deciding and then changing the standard is where the four functions of management close, and a repair figure produced after the decision has already been made is a fine piece of bookkeeping that nobody used.