Model shows: cut cows and make more money
The GSL Linear Programming Model is showing dairy farmers how to reduce herd numbers but, in most cases, produce slightly more milk.
The model has been designed to optimise resource allocation on farms to maximise profits, showing how a farmer can get the most milk from the least amount of inputs. It can tell a farmer if it makes sense to build a feedpad, extend the season to pick up milk, drop to once-a-day (OAD) milking or buy their neighbour's farm.
It has analysed operations on 20 or so farms in recent months and "we haven't found a single one that is operating anywhere near an optimal resource allocation," said Peter Fraser, of Ropere Consulting.
Potential gains have ranged from 32c/kg milksolids (MS) on the most efficient farm to $1.45/kg. The average was 87c/kg.
Environmental gains will result, too. Fewer dairy cows reduce a farm's greenhouse gas (GHG) footprint and will reduce its adverse environmental impact.
An analysis of three Waikato farms using 2009/10 data indicated the ability to achieve large and positive improvements in farm income of up to 53 percent. There were also reductions in costs of up to 30 percent, useful decreases in carbon emissions of up to seven percent and production increases of up to 6.5 percent.
"So the story that farmers can't do anything is basically rubbish," Fraser said.
"We are showing you can."
The Ministry of Agriculture and Forestry (MAF) has used the model to find out the impact of emission charges on farm profitability and to test whether dairy farmers are producing before, at, or beyond the tipping point where the application of additional inputs (cows, additional feed, and so on) erodes profitability (see shaded box page 114).
But Barrie Ridler, of Grazing Systems, who developed the complex thinking around the systems within the model, said optimising profits is up to the farmers.
"You have to find what the farmer's real objectives are. They are not always straight money, so maybe we can work other options which may not generate as much money, but make life easier or whatever."
He became interested in modelling dairy farms while teaching at Massey University in the late 1970s and early 1980s. Linear models, aimed at profit optimisation, in those days needed a mainframe. Most modellers therefore used spreadsheets, which work on averages and benchmarking.
In the 30 years since then, computer power has grown enormously, programming methods have changed and linear modelling can be done on a laptop.
"The other models tell farmers what to do, and they do it, then they tell them something else to do, and they do it," Ridler said.
"With my model, you set the farm up and you run it, and suddenly the model is pulling you along and saying here are the constraints and how you overcome the constraints.
"It's inviting you to think about the new ideas that are pushing up, rather than the other models whereby you push the models and you don't learn anything new out of them."
Federated Farmers Dairy chair Lachlan McKenzie, who milks 660 cows near Rotorua, said no other programme could model both the physical production systems - the amount of grass grown and numbers of cows eating it - and the financial side of the business, then allow farmers to ask "what if?" questions.
He credits the model with understanding in a mathematical way the complexities of a biologically based farming system and the inter-relationship between the physical things done on the farm and the economic outcomes from that.
"And it can come up with sensible answers."
Adjusting cow numbers has ramifications for a complex range of issues on a farm.
When changes are made, Ridler's model doesn't look only at costs/cow - numbers of vet visits and AB costs - but at "a whole cascade of changes in the cost structure".
McKenzie described what happened on his farm. First, Rider had to be satisfied that the profit estimates and other key data accurately represented his farm.
"Then they pushed the button to optimise production and it told me to do some things that I wouldn't have normally thought was the thing to do".
The model suggested McKenzie reduce his cow numbers, for example, and increase his summer cropping.
"It challenges your normal thinking about your farm system and showed me other ways, and then I could go and delve into it, and ask why it is telling me to do certain things," he said.
He tends to change his farm practices when new information comes though, and saw the model as another tool to help find better ways of doing things.
"You've still got to understand the physical nature of your business," he said.
"There are still risks - we've just had two days of heavy rain in the Bay of Plenty.
"You can then use the model to see what happens if you have half your farm underwater or no growth, and it will tell you the ramifications and you can check the ‘what ifs' and the best options for getting out."
Fraser, who set up his business after leaving his job at MAF's dairy desk earlier this year (Dairy Exporter, February, page 20), is helping to commercialise Ridler's model.
"It's an economic model, not an accounting one," he said.
Because farming is a biological system, the optimum won't be reached in real life, "but you can get it quite close".
He gave some examples, keeping farmers' names confidential.
One farmer was considering converting to once-a-day milking (OAD). The model showed a reduction in net farm profit of 85c/kg MS if that idea was implemented.
"Such a loss of income could have been catastrophic."
Another farmer wanted to reduce debt, so was advised to use additional nitrogen (N) to increase his stocking rate.
The model showed that 40t of extra nitrogen produced enough dry matter (DM) to graze 50 extra cows. After costs were taken into account, however, he would be just $6000/year better off.
The corollary is that the farmer could get rid of 40t of N and be only $6000 worse off, which meant converting to Fonterra's organic milk programme might be a better way to lift farm profitability.
Fraser said the model can generate some interesting conversations around the kitchen table.
He cites the case of a farmer who believed bigger is better. His son had different ideas.
The model not only showed the farm was carrying 70 cows too many, but optimising it would increase net profit by more than $300,000/year.
Critically, fewer cows mean there would be no need for another labourer, saving on wages and extra staff housing.
Instead of facing a $200,000 bill for another house, revenue could be generated from the sale of 70 unneeded cows.
"The base problem is that some farms are over-stocked so the cows are under-fed," Fraser said.
"Farmers are therefore filling the gap with supplements, be it palm kernel or maize silage, or whatever.
"But this gap is artificial - it's created through over-stocking."
Some farmers will counter that they use kernel only to cover feed deficits, so a herd of 600 cows gets supplements for two months/year but runs on grass for the other 10 months.
Fraser disputes this, arguing they are in effect feeding 500 cows on grass and 100 cows on kernel, both for the entire year.
"That's effectively what they've got, so if they drop 100 cows, they can drop the kernel because the surplus cows aren't competing with the others for tucker."
This allows farmers to raise output /cow rather than raising output through more cows.
It does not mean OAD is wrong in all cases or that the model is hostile to palm kernel or maize.
The modelling shows circumstances change from farm to farm.
That's why the GSL model does not favour averages, benchmarks and key performance indicators, said Fraser.
"Let's say you have a dry spell in January and February, and your grass growth rates come down," he said.
"The model will calculate if it is better to start to cull cows, or dry off early, or bring in feed.
"If it's better to bring in feed, it will work out what's the least-cost feed source."
The bigger story is the link between farm profitability and environmental outcomes, Fraser enthuses.
With just one exception - to optimise profits the model has reduced herd sizes with obvious implications for reducing greenhouse gases (GHG) and so on.
This fortifies the results in work done for MAF Policy around three years ago based on MAF farm monitoring data.
The study showed that if approximately 250,000 cows could be taken out of the Waikato, farmers would make the same or slightly more money and milk production would be much the same or slightly higher, assuming that they had the skills to obtain higher production/cow.