Where the Money Actually Goes

#Fifty-five per cent of your variable spend is one bag

Ask a room of South African livestock farmers what they argue about most and you will hear fencing, water and the vet. Now look at what an extensive beef enterprise actually spends its variable money on.

In the Land Bank's beef 18-month budget — 1 000 LSU, 625 cows, 80% weaning, 240 kg weaners — winter lick alone is 55.2% of total variable cost. Repairs to troughs, fences, windmills and dams are 0.9%. The vet is 0.2%.

The three things farmers argue about hardest are, between them, about one rand in a hundred.

#The decomposition, line by line

Source: Land Bank Livestock Enterprise Budget 2023/24 v2. These are 2023/24 input prices and must be re-priced before you use them; the shape of the pie is what travels, not the rand figures.

Read the corollary hard. Deferred camps carrying standing forage into winter. Foggage. A late-cut hay reserve. Moving calving so that peak cow demand meets spring growth instead of fighting it. Every one of those attacks a line worth 55% of your variable spend. A new water trough attacks a line worth 0.9%.

#The costs that are not in a gross margin

A gross margin is deliberately incomplete. Four things sit outside it and all four can close a farm.

Land. Rent forgone, or interest on the purchase. In peri-urban Gauteng this is not a grazing valuation — the land is priced for development, which makes it excellent collateral and terrible economics.

Interest. Around July 2026 SA prime was reported at 10.50% against a repo of 7.00%, though that figure comes from a secondary aggregator and should be checked against the Reserve Bank before it goes in a plan.

Labour. The national minimum wage rose to R30.23 per ordinary hour on 1 March 2026, with farm workers on the full general rate (Labour Guide). At 45 hours a week that is roughly R5 895 a month before UIF, housing, COIDA and the other statutory add-ons. That number decides whether a herder-based system is affordable — and a herder is also a theft control and an animal-health observer, not only a grazing cost.

A drought reserve. This is the provision most South African farmers never make, and it is a financing line rather than a virtue. The reason is that multi-peril insurance has never worked here: insurers were not making a profit on it, drought risk is systemic rather than area-specific the way hail is, and South Africa — unlike most countries — offers no state subsidy or assistance. That assessment is from June 2019, and nothing published since was found to say a state-backed scheme has appeared, so treat it as almost certainly still true and not as this week's fact. Get your own broker's exclusions in writing. Then assume you are self-insuring the drought and budget for it.

#Stock theft is a line item, not a tail risk

In the 2024/25 financial year South Africa recorded 190 000 head of livestock reported stolen across 25 606 complaints — 80 600 sheep, 55 700 goats and 53 600 cattle (gov.za, 27 June 2025). The R880 million direct-loss figure attached to that statement is an AfriForum calculation, not a SAPS number, and the ministerial statement puts the reporting rate at roughly 30%. A separate 2025 study found 82% of respondents did not report theft at all, believing police would not act (SA Crime Quarterly 74). Use the range. Anything between one in five and one in three incidents reaching a docket is consistent with the published work.

The best quantified series is the Eastern Cape: 209 235 animals worth R715 million across the 2018–2024 financial years (Ndzungu, Ngoshe & Jaja 2025). Its structure is the part to memorise, because it is counter-intuitive:

• Sheep were 57.1% of the animals stolen (119 404), goats 24.8%, cattle only 14.7%.
• Cattle were 56% of the money (R399m of R715m), sheep R148m, goats R104m.
• Recovery ran at 17.2% of animals — and only 6.9% of the value. Put the other way, 93.1% of the rand value was never recovered.

Your theft exposure is not proportional to your head count. It is proportional to what each head is worth, which is why prevention spend should be weighted towards cattle even on a farm where sheep are the animals disappearing.

#Predation, costed honestly

The number everyone quotes is R1 390 453 062 a year in national small-stock predation losses (Van Niekerk 2010). Quote it with all three of its problems attached or do not quote it: it is self-reported by telephone with no carcass verification, it is an extrapolation from five provinces, and it is sixteen years old in 2010 rands and has never been repeated. It is the best number available and it is a weak number.

Its predator attribution — jackal 65%, caracal 30%, stock theft 3%, vagrant dogs 1%, with a mean loss of 6.14% of small livestock and 13% of lambs aged nought to six months — is a Northern Cape sample. That is not the Highveld profile. On a peri-urban Gauteng smallholding the realistic threats are free-roaming domestic dogs and opportunistic theft, and a free-roaming dog is somebody's property, which makes it a civil claim rather than a hunting decision.

On what control costs, the cleanest SA figures come from eleven Eastern Cape farms: total cost per head fell from USD 23.41 in the lethal-control year to USD 9.60 and then USD 5.92 in two non-lethal years, with mean stock loss falling from 13.6% to 3.7% (McManus et al. 2015). Teach the roughly 4:1 ratio, not the dollar — and teach the caveat with it: eleven self-selected farms compared to themselves across successive years, with no control group.

#The capital numbers nobody publishes

Now the honest gap. This course can give you a variable-cost benchmark. It cannot give you a capital benchmark, and it will not invent one.

There is no current published South African figure for agricultural fencing per kilometre, water reticulation, boreholes, solar pumping, handling facilities, abattoir throughput fees for small lots, or soil laboratory pricing. That is not an oversight; it was searched for and not found.

What exists is worse than nothing if you misuse it:

• The one agricultural fencing figure located is R4 033 per kilometre for a 5-strand barbed-wire fence — in April 2000 (KZN DARD). Useful as a ratio between fence types. Useless as a 2026 price.
• Electric fencing in South Africa is sold overwhelmingly into the urban security market, where quoted rates of roughly R95–R395 a metre installed are normal. Those are not agricultural rates and must never enter a farm budget.
• One Gauteng installer's game-fencing list page, accessed July 2026, ran from R145 a metre for basic high-tensile to R625 a metre for premium heavy-duty. That is one company's web page, not an industry benchmark.

The substitute discipline is not a rule of thumb. It is paperwork.

#What changes on Monday

Pull your last twelve months of variable spend and rank it. If bought feed is not around three-quarters of it, either you have found something real or your records are incomplete — and both are worth an evening to establish.

Then add the two lines you almost certainly do not have: a theft provision priced at replacement cost, and a drought reserve. Neither is pessimism. They are the two costs that arrive whether or not you budgeted them.