How many sources does your revenue really lean on?

To see how many sources your revenue really leans on, split last year’s revenue by source and turn the shares into one number: how many sources of equal size would split the money the same way. That number is the effective number of sources, and it is often lower than the number of lines on the list. A farm with five sources, where soybeans bring in 48 percent of the money, leans in practice on about 3.2 sources of equal size.

What follows is the arithmetic, the bands to read it against, how to count sources that have bad years together as one, and the one-page sheet to keep with the farm’s plans.

Take a farm that closed the year with $1,000,000 in revenue from five lines: $480,000 from soybeans, $210,000 from second-crop corn, $180,000 from finished cattle, $90,000 from milk and $40,000 from renting out pasture. By their shares, those five lines carry the weight of 3.2 equal sources. Then the farmer writes down what makes each line rise and fall. The two grain lines share one cause, the two cattle lines share another, and counted that way the farm leans on 1.8 sources. The revenue concentration calculator built by Rurivia, which publishes this page, runs these figures, and the drop from 3.2 to 1.8 is the finding worth keeping.

Nobody on a farm struggles to name the activities. What is hard to say, without splitting the year’s revenue by source, is how much the farm depends on each one. Revenue arrives added up on the bank statement, and a bank statement has no column for where each deposit came from.

How do five lines of revenue become one number?

Each line’s share is its revenue divided by the total. Square each share, add the squares, and divide one by the sum.

Rurivia calculation from the revenue shares of the example farm.

One divided by 0.3166 is 3.16. Five sources of $200,000 each would give 5, and a single source would give 1, so 3.16 means this farm leans like three sources of equal size. The sum of the squares is known as the Herfindahl-Hirschman index. One minus that sum, 0.68 here, is the diversification index, also called the Simpson index, which runs from 0 for a single source toward 1 for many equal ones. The calculator shows the index next to the count.

Squaring is what gives big lines their weight. Adding the shares tells you nothing, because they always add up to 1. Squared, soybeans at 0.48 add 0.2304 and pasture rent at 0.04 adds 0.0016: the soybean line earns 12 times as much and counts 144 times as much. A small line barely moves the result, which is what you want from a measure of dependence.

Why convert to a count, when the index carries the same information? Because equal steps on the index are not equal steps in sources. An index of 0.90 means 10 equal sources and 0.95 means 20, so a gap that looks like rounding on the 0-to-1 scale is double as a count. The same thing turned up in a survey of 46 farming households in Libres, Puebla, in Mexico, where Arteaga and colleagues found two groups at 0.912 and 0.958 on this index; converted by Rurivia from their published figures, that is 11.4 sources against 23.8. That is why the count is the number to read first.

Which bands should you read the number against?

Use bands built on farm data. Ipea, the Brazilian government’s economic research institute, published a chapter that applies a classification from Sambuichi and colleagues, built on Brazilian family farm registrations, with four classes on the diversification index. Exactly 0 is a single product, up to 0.35 is specialized, up to 0.65 is diversified, and above 0.65 is highly diversified. The example farm, at 0.68, lands in the top class, which is why counting the sources that fail together matters so much.

Each cut has a reason written next to it. The specialized class covers units that, despite having more than one product, draw 80 percent or more of gross production value from just one of them. To be highly diversified, a unit had to show at least three products of similar weight in gross production value.

Both cuts check out. One source with 80 percent of revenue adds 0.64 to the sum on its own, which puts the index between 0.32 and 0.36, depending on how the other 20 percent is split; the 0.35 cut falls inside that range. Three equal sources give a sum of one third and an index of 0.667, just past the 0.65 cut. So the two cuts are counts before they are decimals: one product carrying four fifths of the value, and three products of about equal weight.

Bands published for trade between countries do not fit a farm, because they were set for trade spread over dozens of partners. On the European Commission’s scale, a sum above 0.25 signals trade concentrated in a few partners, and 0.15 to 0.25 is moderate. Four sources of exactly equal size, the best a four-source farm can do, score 0.25 there and still count as moderately concentrated.

What does the number miss when two sources have bad years together?

The effective number is built from shares alone, and shares carry no trace of which sources have the same bad year. That matters because the protection comes from sources that fail at different times. The OECD, working from farm accounts in seven countries, states that “The choice of a combination of crops whose returns are not perfectly correlated reduces the variability of the total revenue”. Two sources that always rise and fall together are one source with two names.

Where drought hits every crop at once, that protection shrinks. In the same report, the OECD found that “yield correlation is higher in Australia, implying that the failure of one crop is more likely associated with the failure of another crop”. Two farms can have the same diversification index and very different risk, depending on whether their crops fail in the same years.

The OECD also found that “Price risk tends to be more systemic so that higher coefficients of correlations are found between prices than between yields”. In plain terms, two crops have poor harvests in different years more often than they have poor prices in different years. A second crop protects you against a bad harvest much more than against a bad market, and a farm whose problem is price usually needs another kind of protection, such as a hedge or a forward contract.

In the calculator you can account for this. Next to each source’s name and amount there is a group column, where you type a short word for what makes that source rise and fall, such as “grain” or “cattle”. Sources with exactly the same word are added together, and the result is the grouped effective number, 1.8 in the example. Type the word identically every time: “grain” and “grains” become two groups.

Guessing the groups from memory is risky, and the check is cheap. Write revenue by source for the last five to eight years, with sources in rows and years in columns, and mark every year a source came in below its own average. Two sources marked in the same years belong in one group, whatever they are called. Three years of records are too few for this check, and eight are enough. A source that has existed for only two years cannot be tested this way, so its group is a judgment, and it is worth writing that down next to it.

For each pair that shows up, ask whether both depended on the same rain or on the same market, because the answer changes what helps. A weather cause calls for an activity that fails at a different time of year or in a different kind of year. A market cause is often not solved by adding an activity, since the new one may be sold in the same year of low prices.

Fill in the group for every source or for none. A source left without a group counts as a group of its own, which assumes it is independent of everything else. With two groups typed and three sources blank, the grouped effective number comes out high because nobody checked those three, not because they are independent.

One limit has no fix. The Ipea chapter notes that the diversification index can hold still across two different periods even when the production mix has changed a great deal, which happens when the changes leave each product’s share of the value where it was. Replace every crop on the farm, keep the same shares, and the number does not move.

What does diversifying cost you?

Spreading revenue over more lines is not free. Mzyece and Ng’ombe used a national survey of rural households in Zambia and measured diversification by the land given to each crop. They found that crop diversification improves income stability but reduces technical efficiency, meaning less output from the same land, labor and inputs. They trace it to more competition for resources among the different crops: the same land, machines and attention stretched over more activities. The data come from Zambian households, so the size of the effect may not carry over to a larger farm, but the mechanism is familiar to anyone running several activities.

Concentration can also be a strength. A farm often concentrates on the activity it does best, and a second activity pays off most when it shares machinery, labor and know-how with the first, which economists call economies of scope.

Insurance and hedging can do part of what a second activity does. In OECD simulations, higher government support prices for cereals led farmers “to concentrate more on wheat production that generates higher return, even if with higher variability”. A farm that insures its crop or hedges its price buys some of the steadiness a second activity would have given, without the second activity’s demand on management.

How much steadier does a mix get? The OECD measured how much revenue per hectare swings from year to year, relative to its average, which statisticians call the coefficient of variation. Comparing single crops with the mix farms grow, it reports that “the size of the reductions in the coefficient of variations varies among the countries and it can be as high as one-half”. In the German figures, single crops swing by 0.16 to 0.31 while the crop mix German farms grow comes in at 0.12, steadier than its steadiest crop. Those German figures are marked in the report as simulated rather than taken from farm accounts, so read the size of the gap, not the decimal.

So the trade is steadier revenue against some efficiency, and weighing it is a judgment only the person running the farm can make.

Why is share of revenue not share of margin?

The calculator works on revenue, because revenue is the number a farm can take straight from its own records. It leaves out profit. A source can be a third of your revenue and none of your profit: cattle finished on bought feed can turn over a lot of money and keep little of it, while a small line with almost no cost can carry the year. So the effective number answers what happens to the money coming in if a line stops, not which line pays.

Profit per line is a separate calculation, covered in three other articles on this site: setting a target margin before the cycle starts, the lowest price that still covers your costs and what a crop should leave you when the cycle ends. That split between revenue and profit is Rurivia’s reading of what the arithmetic can carry, not a finding from the sources cited here.

You can run the same calculator on margin. For each source, take what was left after paying the costs that exist only because of it, and type those amounts in place of revenue. If the effective number on margin comes out lower than on revenue, the farm depends on fewer sources than the first reading suggested, and the largest line on margin is the one that carries the year. If it comes out higher, a line that turns over a lot and keeps little was making the revenue reading look safer than it is. A source that lost money cannot go into this run, and that loss is already the answer about it.

The number also leaves out time. It covers twelve months that already closed, so a contract signed in November and a line that ends in March weigh the same in it, though one adds to next year and the other drops out. Nor does it say when money arrives: two equal sources that both pay in June leave the farm short from July on. To check that, run the calculator a third time with twelve lines, one per month. From the effective number of months you can see whether your cash comes in across the year or in one or two months.

What do you do when the number comes out bad?

“Diversify” is not a move by itself, and the studies cited here show it has a cost. The moves that follow are ordered by what they cost, and the first two are often skipped because they seem too cheap to count as a decision.

Split the buyer before splitting the crop. Two sources with different names and one buyer are one source. A second buyer for the same product changes nothing in the field and removes the risk that one buyer stops paying and the account runs dry. It does nothing against price, because both buyers price off the same market, so be clear about which of the two risks you have removed. The buyer for each sale belongs in a selling plan written before the cycle starts, next to the price you expect, and that price comes from a record of local prices with their source.

Move the calendar, not just the product. A source that pays in a different month helps the bank account even when it belongs to the same group in the calculator. The effective number will not change, but the overdraft will.

Prefer a source whose bad year has a different cause. This is the only move that raises the grouped effective number, not just the plain one. Test any candidate: write down the one event that would make it fail, and check whether that event is already written next to your biggest group. A crop that fails in the same drought as your main crop adds a line to the list and nothing to the grouped effective number.

Resize what you already sell before adding something new, and do it in a few large moves. The example farm’s 3.16 is short of the 5 that five equal lines would give, so there is room inside them. Moving 8 points of revenue from soybeans to milk and pasture rent, 4 each, raises the number from 3.16 to 3.85 with no new activity. A new activity worth a quarter of revenue raises it to 4.16, but it is a whole new activity. Three new lines of 3 percent each raise it to 3.78, nearly twice the gain per dollar moved, but the Zambian study links many small lines to lower efficiency. All these figures are Rurivia’s, on the example farm.

Consider buying the protection instead of growing it. When your concentrated activity is the one you do best, the alternative to a second activity is a price contract or insurance on the first, as the OECD simulations suggest. Two articles on this site cover the price side: what price you actually keep after you hedge and whether locking in the price now is worth what it costs.

Decide changes for the next cycle, not this one. The revenue mix is set at planting, not at selling. That decision goes in a written plan for the next three production cycles, and the price behind each line goes in a farm budget with its price assumption written down.

When the index is already good, the risk is drift. In the Brazilian state of Espírito Santo, a study by Galeano for the state’s rural research agency found the index, published as an effective number, fell from 9.12 in 2013 to 6.21 in 2022 as coffee went from 38.86 percent of production value to 50.81 percent. That is a whole state, not a farm, but a farm’s mix can drift the same way, one good year of its main crop at a time, so the number needs a review date, not a single reading.

What goes on the one-page sheet?

On one dated page, kept with the farm’s plans, you write nine things the calculator cannot work out, since it only takes a name, an amount and a group for each source. The page belongs with the farm’s other planning records, each covered in the farm planning articles on this site.

The twelve months measured, and what in them will not repeat. Use the same twelve months for every line. Beside them, list what came in that year and will not come next year: a machine sold, an insurance payout, a payment from the year before that arrived late, two harvests sold in one year because one was delayed. Each of these makes one source look bigger or smaller than usual, and the arithmetic cannot tell. Without that list, one year’s mix passes for the farm’s usual mix.

One line per source, with the buyer named. Not “grain”, but the company that paid. This is where two sources with one buyer turn up.

The month each source actually pays. The payment date, not the contract date.

The last bad year of each group, and whether two groups share it. The cause you wrote is what you believe; the year is what happened. When two groups point at the same year, either they share a cause and are one group, or that year was bad for two reasons at once. Either way, it changes what you plant.

The largest buyer, added up across sources. This differs from the largest source, and on farms that sell several products to one cooperative it is often larger.

What your sources share besides price and weather. Two lines that need the same combine in the same three weeks, the same crew at the same peak, the same shed or the same collateral at the bank can fail together. That holds even when their market and weather causes differ. The calculator has no place for this, so you only see it if you write it here.

The next move, and why it does not share a cause with the biggest group, plus the month it would first bring in money. A candidate that fails this test is dropped.

The trigger for redoing the sheet, written as a threshold, for instance any group passing 60 percent of revenue, plus a calendar date for redoing it anyway. Drift goes unnoticed without a date.

Who keeps the sheet, because a record nobody owns tends to stop being updated in the second year.

Write each group’s cause in the farm’s list of risks too; ranking that list is covered in the article on which farm risks deserve attention first. In the example, the grain group is 69 percent of revenue (soybeans 48 plus corn 21). A cause that can hit that much in one year belongs at the top of the risk list, not only on a revenue sheet.

Where to start

Take last year’s bank statements and write next to every deposit where it came from. That step alone, before any arithmetic, is what many farms are missing, and it often turns up a line nobody had been counting as revenue. Add up each source, run the numbers in the calculator, and then do the part the calculator cannot do for you: write, next to each line, the one event that would make it fail. When two lines get the same answer, give both the same group word. Knowing what the farm depends on before a bad year tests it is one piece of running the operation, and the farm management page on this site covers the rest.