A budget and a guess are made of the same figures. What separates them is the line written beside each figure saying what price it assumes, where that price came from, and on what date it was read.
That line is missing on most farms, and its absence costs in a specific way. The account of what comes in and what goes out over the next three years lives in one person’s head and gets rebuilt in every conversation, with a different price in mind each time. When it finally goes on paper for the lender, the figures arrive alone.
Six months later nobody can say whether the plan broke because the price came in different, because the volume came in different, or because the arithmetic was wrong from the start. Since the assumption was never written, it cannot be corrected either, and the next budget is built on the same hunch without anyone knowing it is the same one.
A budget of this kind is published every year with its assumptions attached, and anyone can read them. USDA Agricultural Projections to 2035, prepared by an interagency committee and released in February 2026, states that “The scenario presented in this report is not a USDA forecast about the future. Instead, it is a conditional, long-run scenario of how markets would evolve under current conditions, existing laws, normal weather patterns, and underlying trends.” It dates its own inputs as well: data through the November 2025 supply and demand estimates, with macroeconomic assumptions completed in September 2025. A projection covering an entire country declares what it stands on. A single farm has less reason to keep that hidden, not more.
What a whole-farm budget is actually for
Comparing plans against each other, not predicting the year. Penn State Extension puts the purpose in one sentence: “A whole-farm budget is used to estimate the expected income, expenses, and profit of a given farm plan, to compare the profitability of alternative farm plans”. The unit of the exercise is the plan, and there is always more than one of them, including the plan of carrying on as before.
A figure that exists to be compared with another figure does not have to be right the way a weight on a scale has to be right. It has to be built the same way in both plans. That is only checkable if the assumption is visible in both, which is why the assumption column is not clerical work: two plans priced off two different assumptions cannot be compared at all, and the comparison will still produce a winner, quietly, decided by the assumption instead of by the plan. The same is true of any goal with a number and a deadline stated in money, and of every other document in the planning axis that ends in a figure.
Why the assumption belongs in a column of its own
Because the objective and its assumptions fail together. The Manual de Organización y Gestión de la Empresa Agropecuaria, published for the agricultural schools of Buenos Aires province by its agriculture ministry with content coordinated by INTA, the Argentine national agricultural research institute, states the dependency directly: objectives are the results the business wants to reach within a given period, with given resources, inside a set of assumptions, and if those assumptions do not hold, the objective does not hold either. Defining the assumptions, it argues, is of the first importance, because they are what set the scenario the business will be operating inside.
USDA says the same thing about its own figures in one line: “Changes in any of these assumptions can significantly affect the projections, and actual conditions will alter the outcomes.” A budget with no such line next to it still has assumptions. It keeps them in someone’s memory, where nobody can argue with them and nobody can correct them. On any farm running more than one activity the column carries a second question as well, because an expense that serves two of them arrives at the budget already divided, and what it was divided by is a declared split rule with a date of its own, which belongs beside the assumption and not in a different folder.
Which line of the budget breaks first
The income side, not the expense side. The Argentine manual is blunt about where variances originate: any change in the variables involved moves the expected result, and the main variances show up at the level of income rather than expenses, because price and yield stay variable right up to the end of harvest.
The manual attaches a condition to that, in a parenthesis of its own, and the condition matters: it holds where the farm has not priced any of the crop forward. Expenses land sooner and move much less in either case. The reasoning holds anywhere the sale price is left to settle after the cost has already been spent, which covers a pen of cattle as squarely as a field of grain.
That ordering has a consequence for the table. Most of the care a farm puts into budgeting goes into the expense rows, which are the stable half, while the two figures that decide the outcome, price and volume, get one number each and no note. Penn State supplies the way to make the price assumption testable: the break-even price is projected total costs divided by expected yield, and it is “the minimum price per unit required to cover all projected costs at the expected yield”.
Written into the assumption column, it turns a hunch into a figure that a delivery note can later be held against, and it is the same figure that belongs in a written selling plan. If the two documents carry different numbers, one of them is wrong.
What a single average figure hides
The downside. A review published in the Journal of Applied Farm Economics in 2025 gathered 61 studies of farm-level financial methods in cow-calf operations. Just over a third of them built variation into their figures. The rest ran on single fixed values.
One of the reviewed models shows what that choice costs. Run on average values for costs and yields, it made flexible grazing look far better than the conservative alternative. Run again with variation in cattle prices and forage yield, the range of financial outcomes widened and exposed downside risk the averages had covered up.
The reviewers read the pattern as leaning on average conditions understating the economic risk of drought and variable forage while overstating the gains. This is one review of one production system, and a model is not a herd, so what carries over is the shape of the error and not its size. A single central figure in the price column is not a neutral choice. It is an optimistic one, by construction.
How far out the figure still means something
Accuracy falls as the horizon lengthens, and the fall has been measured. Fang and Katchova, writing in Q Open in 2023, compared OECD-FAO and USDA baseline projections for rice and wheat in the European Union, the United States and China, and reported that the accuracy of those projections, for most crop variables in all three markets, falls as the projection horizon lengthens.
The misses lean, and they lean the same way for years at a time. USDA’s own Economic Research Service, in a report published in February 2021, went back over its ten-year projections from 1997 to 2017 and found the baseline consistently too high on wheat harvested area and too low on soybean area.
Seven years out, the average miss ran about 10.4 percent high for wheat, 7.3 percent low for soybeans and 3.3 percent low for maize. Those misses repeated in the same direction instead of cancelling each other out, and repetition is exactly what makes an error correctable. A farm that never wrote its price assumption down has no way of discovering whether it has a direction of its own.
University of Maryland Extension sets the working horizon: “It is important to project 3-5 years of financial statements for your farm business depending on a loan application, long-term goals or a new enterprise decision.” Three years is the short end of that range, and the short end is where a first table belongs, with the third year’s assumption carrying the widest band.
Where the rows come from
From documents the farm already holds. Maryland is explicit about the starting point: “The best way to start making any projections is to review your enterprise budgets and financial statements.” The arithmetic that follows is stated just as plainly, since the projected income statement is built “by multiplying the income and expenses in the various enterprise budgets by their respective total number of units and adding them together”. The unit counts on the right of that multiplication are the volumes that a multi-year production plan already lists cycle by cycle.
Published budgets are a starting point and not an answer. Penn State: “you should think of the sample budgets as a first approximation and then make appropriate adjustments using the ‘Your Estimate’ column to add, delete, and adjust items that reflect your specific production situation.” That instruction is the argument for the assumption column in a single sentence. A published figure carries somebody else’s situation inside it, and once it has been copied into your table with no note attached, nobody can tell whose situation is sitting in the row.
Two documents, and the one the lender reads
Both are needed, because they answer different questions. Maryland separates the projected income statement, which asks whether the plan makes money, from the cash flow budget, which asks whether the money is in the account in the month the bill arrives. Of the second it says the cash flow statement “is a critical component of the business plan and will be reviewed by lenders.”
| Projected income statement | Cash flow budget |
|---|---|
| Cash income | Beginning cash balance |
| Less cash expenses | Plus cash income |
| Less depreciation | Less cash expenses |
| Equals profit or loss | Plus capital sales, less capital purchases |
| Plus loan receipts, less loan principal payments | |
| Plus non-farm receipts, less withdrawals | |
| Equals ending cash balance |
The right-hand column is where the machinery shows up, so the year a purchase falls in is a budget decision and not an afterthought, which is the link between this table and a written investment plan. Depreciation appears on the left and not on the right, and how the farm accounts for its result decides which side a line lands on and in which year, which is settled by the record of the tax regime rather than by the spreadsheet. One price assumption feeds both columns in different months, and a price with no month attached fills the profit line while leaving the cash line unanswered.
The date the table gets opened again
Written on the table before it is filed, not decided later. Maryland describes the comparison and what it is for: “compare the actual and projected cash flow statements to see if things are going as planned, to devise remedies for unforeseen problems or to take advantage of opportunities not anticipated”, and at the end of the year, “use the actual cash flow statement to estimate the projected cash flow for the next year.”
The Argentine manual gives the same instrument two jobs and names them apart. Used before the fact, the gross margin is for budgeting. Used after the fact, on results that have already happened, it is for control. One table, two moments, and the second moment is where an assumption stops being a hunch and becomes a reading with a correction attached to it and a name against the correction.
The name belongs to whoever reports the actual figure for that row, and that person has to be able to open the sheet without asking anyone, or the second moment never happens. None of the second moment works if the first left nothing to compare against, which is the reason the column exists at all and the point at which this table joins the four functions of management instead of decorating them.
Where to start
Three hours, with twelve months of delivery notes and expenses on the table.
A price assumption that turned out wrong, written down with its date and its source, is worth more than one that turned out right and was never recorded, because only the first can be examined next year for a direction. No lender will ever ask to see it. A projection covering a whole country missed the same way for two decades and changed its method because it had the record to prove it. A farm that keeps the number in its head gets a different answer in every conversation and never finds out whether the answer has been leaning the same way the whole time.