Set what shipped against what you declared, and not against the best price the year offered. The peak is a test almost nobody passes: over the 1995 to 2004 corn crops, the professional advisory programs tracked by the University of Illinois priced in the top third of the price range between 17 and 25 percent of the time, with by far the largest share landing in the middle third, between 58 and 63 percent.
Soybeans came out much the same, with the top third between 17 and 19 percent and the middle third between 67 and 69. Judged against the high, a farm fails almost every year and learns nothing from failing. Judged against its own declaration, it gets a number it can act on.
The cycle ends and the impression is that the selling went well, or badly, depending on which sale stayed in the memory. The share that was written down beforehand, part of the volume by one date and the rest after it, was never checked against what actually left, because the sales sit in loose settlement documents, in bank statements and in two people’s heads. Nobody added up the volume by tranche, and nobody worked out the price the farm actually received. Without that figure the next conversation starts from an impression, and the plan gets quietly rewritten until it fits what already happened, which makes it impossible to break and therefore useless.
What is the number at the bottom of the table?
One figure, and it is the volume-weighted average price the farm actually received. Multiply the volume of each settled sale by the price of that sale, add the results, and divide by the total volume sold. It is not the average of the prices, which would count a single truckload and a contract covering the whole cycle as equal weights, and the difference between the two arithmetics is usually larger than any argument about whether the market was kind.
That figure has a name outside the farm gate too, and knowing the name is what stops the number being dismissed as home accounting. FAO defines producer prices as prices received by farmers, determined at the farm gate or at the first-point-of-sale when farmers participate in their capacity as sellers of their own products, and collects them once a year from 179 countries for 212 crop and livestock products, and monthly from 113 countries for 200 products. Every one of those national figures is built from farms doing, at their own scale, exactly the calculation described in the paragraph above.
The back half of the table works even with nothing declared. A farm that never wrote a plan can still recover every settled sale from the paperwork it already keeps and produce the weighted average, and that number becomes the first standard it has ever declared. The front half, the part with a target share against a date, is the page built in a written selling plan, and a reader who does not have one yet should build it there rather than here.
One thing the figure is not: it is the price received, before the cost of having waited for it. Any lot that sat in a bin between harvest and settlement carried interest, shrinkage and handling that no settlement document shows, and the weighted average flatters that lot against one sold off the combine. Work the holding cost out in what a month of storage costs and write it as its own line beside the table, because the farm that skips it is comparing a gross number against a plan it meant in net terms.
Why is the best price of the year the wrong line to be judged against?
Because it converts a decision into a lottery result after the fact. The Illinois report opens by naming the belief, the adage that farmers market two-thirds of their crops in the bottom third of the price range, and its own measurements do not support it. Over 1995 to 2004 the benchmark built from prices farmers actually received landed in the top third of the range 10 percent of the time in corn and 18 percent in soybeans, in the middle third 53 percent and 63 percent, and in the bottom third 38 percent and 20 percent.
The longer record points the same way, with a caveat the authors put in themselves. The report cites an earlier study covering 1973 to 2003 with nearly identical methods, which put farmer prices in the top third 15 percent of the time in corn and 22 percent in soybeans, and the authors note that these differ only marginally from the advisory programs measured over the shorter window. They add, in the same paragraph, that comparing across the two studies calls for caution, because the two cover very different stretches of years.
The middle is where the selling lands, for professionals and for farms alike, and a yardstick that calls the middle a failure will be abandoned by the person holding it.
Read those figures for what they are. The advisory programs were not picked at random, the marketing assumptions behind every price were built to describe a representative corn and soybean farm in central Illinois in the United States, and the report says plainly that it is inappropriate to make performance inferences for an individual advisory program from the aggregate. What travels across a border is not the percentage. It is the shape: hitting the high is rare, hitting the middle is normal, and any test built on the high produces the same verdict every cycle.
Two comparison lines, and they do not measure the same thing
There are two families of yardstick and they answer different questions. The Illinois report specifies both. A market benchmark should measure the average price offered by the market over the marketing window of a representative farmer, and the authors built two versions of it, over 24 months and over 20, in order to test the sensitivity of performance results to different market benchmark assumptions. A behavioral benchmark, by contrast, should measure the average price actually received by farmers for a crop.
The gap between the two is not decoration. For the 1998 corn crop in central Illinois the 24-month market benchmark came to 2.24 dollars per bushel while the farmer benchmark built from market prices came to 1.92 dollars per bushel, the same crop and the same year, with the line to beat moving by thirty-two cents depending on which family was chosen. The weighted average your table produces belongs to the second family, because it is a price received. Holding it against a series of prices offered compares two different animals and will flatter or punish the farm for no reason connected to any decision anybody made.
Which line the cycle will be judged against is a choice made before the first sale, and it belongs on the plan rather than here. Where the offered prices are recorded day by day, with a source and a time, is the price record. This page needs only that the two are not confused with each other at the moment of the comparison.
How much actually went out, and by when?
Far less than habit suggests, and it moves violently between years. McNew and Musser tracked six Maryland grain marketing clubs through a real-time pricing exercise from 1994 to 1998 and found that the amount priced at harvest ranged from a low of 10 percent in 1997 to a high of 68 percent in 1994, with the all-club peak of 67 percent reached in August 1998 against a normal level of around 45 percent. The same groups, the same region, the same five-year stretch, and the priced share nearly seven times apart between one year and another.
Their comparison against a separate survey is the part worth keeping. In August 1995 the average advisory service had priced 35 percent of the corn crop while the average marketing club had priced 40 percent, which the authors call a remarkable similarity. Two independent groups, measured the same month, agreeing to within five percentage points. The priced share is a measurable quantity, it is comparable across farms and across advisers, and a farm that cannot state its own is missing something everyone else in the conversation can produce.
The limits are the authors’ own. These were paper transactions in a marketing game rather than money changing hands, only five marketing years were available, and no transaction costs were charged. The paper does not claim the clubs made money from forward pricing. What it does establish is that the share priced by a given date is not stable, not habitual and not something anybody should expect to recall correctly a year later.
The public series is an estimate. Your table is the fact.
Reverse the usual hierarchy here, because it is backwards. The published series is the approximation and the farm’s own settled sales are the exact figure. FAO says so about its own data: while the aim is to stay as close as possible to the farm gate concept, wholesale prices or even prices at local markets may be appropriate proxies of farm-gate prices when the marketing chain is very limited. Nobody has to approximate the price your own farm received. It is written on the settlement documents.
What the public series is good for is a dated line to stand beside your own. The University of Illinois publishes a database of the average price received by United States farmers from 1960 onward, for corn, soybeans, wheat, barrows and gilts, steers and heifers, calves and milk, built from the National Agricultural Statistics Service, and it carries the warning this whole page rests on: the prices represent the average price actually received by farmers, and therefore, may be different from average prices offered by the market. A reader outside the United States looks for the equivalent series from the statistics agency of his own country, which FAO’s country coverage says exists in most of them.
Why does the plan get rewritten to fit the selling?
Because memory moves toward the outcome, and it moves further the more there is to remember. Kaida and Kaida asked 63 volunteers to answer numerical estimation questions, showed them the correct answers, waited three and a half hours, then asked them to write down what they had originally answered. Recalled estimates had shifted toward the correct answers rather than away from them, and the drift landed on more items when there were 50 answers to hold than when there were 20.
Read the ordering, and stop there. Those were general knowledge questions answered by volunteers in a single day in a laboratory, not sales decisions made across a cycle, and no figure from that study describes what happens in anybody’s office. The direction is what transfers, along with the load effect: the more items a person is carrying, the more the remembered version drifts. A cycle of selling is a great many items, spread over months, each one attached to a price that later turned out to be good or bad.
This is the whole reason the declared share has to be copied from the page that was written first rather than typed from memory at the moment of reconciliation. A plan reconstructed after the outcome is known will always show good adherence, because it has been adjusted, without dishonesty and usually without anyone noticing, to describe what already happened.
What the table looks like
One row per settled sale, and one block of columns underneath. Fill it from the settlement documents and the bank statements, not from recollection, and stop at sales where the money has actually settled.
| Column | What goes in it |
|---|---|
| Date | The date the sale settled, not the date it was agreed |
| Volume and unit | The quantity that left, in the unit the buyer paid on |
| Price per unit | The net price on the settlement document, after deductions |
| Buyer or channel | Who paid, so the channel can be totaled separately |
| Tranche | Which declared tranche of the plan this sale belongs to |
| Total volume | The sum of the volume column |
| Weighted average price | Volume times price, summed, divided by total volume |
| Declared share | The percentage written for each tranche before the cycle |
| Executed share | Volume of that tranche divided by total volume, as a percentage |
| Gap | Executed share minus declared share, in percentage points |
A commitment that has not settled does not belong on this table. Volume sold forward and not yet delivered, and volume delivered without a price yet fixed, are both real and both live on the sheet of the volume already priced until the day money changes hands. Mixing the two lists is the most common way the executed share comes out wrong, and it always comes out wrong in the flattering direction.
What changes in the next declaration
Two lines of prose under the table, and they are the only part of this that is not arithmetic. The first names what explained the largest gap. The second says what changes in the next declaration because of it, and it has to be a change to the page rather than a resolution about behavior.
The correction is almost never a better forecast. The Illinois report notes, among the findings from its full version, that it is difficult to predict the pricing performance of advisory programs based on past performance, which is a hard sentence for anyone whose plan for next cycle is to listen to somebody sharper.
What can be changed is the declaration: a tranche whose trigger never fired because nobody had authority to act on it, a date set in a month when the farm is in the field, a share that was never realistic against the delivery capacity available. Each of those is a line to rewrite, and rewriting it before the cycle starts is legitimate in a way that rewriting it afterwards is not.
The gap also feeds the margin. The realized income half of a declared margin target depends on the weighted average price this table produces, and using a remembered price there quietly corrupts a second document. Booking the reconciliation is the same move as a scheduled review date on the goals: without a date in the calendar and a name beside it, the meeting where somebody would notice the gap never gets called.
Where to start
Two hours, once per cycle, with the settlement documents on the table. No platform, no adviser, no price forecast.
The farm that does this once discovers something uncomfortable and useful in the same movement: the price it received is a fact it never possessed before, and it is now stuck with it. Every future argument about whether the selling went well has a number in it, and the number belongs to the farm rather than to the market. That is what the marketing axis is for, and it is the fourth movement of farm management doing the only job it has, which is to send a finding back to the page where the next decision gets declared.