1
Let’s switch up gears and talk about pork economics at the producer level. There are many interesting and revealing ways to understand what is going on economically on your farm, but you are likely missing much of it because you have sifted down to a small set of indicators of performance that you look at all the time (to the exclusion of others). These are all fully blessed and approved by your consultants, industry press and other experts. Many things which become significant and costly problems, even though preventable often for many months, never show up in these overused and miscalculated metrics. In addition, there is information available that could reduce cost or increase revenue and it is not found in any standard record-keeping program or scheme.
So where would you begin to develop new and valuable information about your farm’s profitability (or lack thereof) that you are not monitoring now? Certainly, one of the most important ones is to visualize your data in a new way. The very first thing I do when I want to see the basic financial health of the farm is to plot the individual weights as a histogram (not a bar chart) from animals sold over a period of time. You can choose any period if you realize the embedded effects of seasonal, cyclical and trend factors which are largely due to the animal’s interaction with the environment (anything from day-length to temperature to measures of “comfort” which can include drafty spaces and even abuse by other pigs or humans) are all in there. Disease of course will be well represented in your graphs in a way you have probably never seen it.
I will demonstrate this and its importance in a moment but first let’s address one of the most important pitfalls regarding swine weight data. Keep in mind there may be reasons to disregard some of my warnings if you have other purposes in mind with your metrics or have a unique flow. If you use the slaughter data for a single building, plotting all of the individual animal’s weights at harvest, you have a couple realizations that must be acknowledged. First, make sure all of the animals were really from only that building and only from a single turn. It is not unusual for a truck to fill one compartment with the last pigs from another barn in order to be efficient. If your slaughterhouse records a tattoo/ear tag etc. that you place on each animal, you can have the computer sort that out.
If you market a building by truckloads over a period like 2-3 weeks, allowing slower growing animals to reach more profitable weights between truck visits, you have really created several different groups (of differing age pigs at slaughter) within your single building data once the building is emptied. If you visualize data from the entire barn after closeout, it will have potentially deceptive patterns which can be less useful. Depending on how long you wait between load outs for this building, the animals in each group begin to take on substantially different inputs and are no longer from the same or very similar “process” (things such as possibly a feed change, different weather and temperature, different stocking density, changes in pen competitiveness, in-pen environment differences like drafts, wetness etc., have all shifted).
What does it mean when I lump those animal groups together and analyze them as a single group even within a single barn? Would it affect your current understanding of the meaning of building level ADG and FCR on your farm? You may think that this extra two-three weeks is negligible but I urge you to remember that during the last weeks of growth, pigs are putting on more weight per day than at any other period. At ending growth rates averaging only 1.25 kg/day (2.75 lbs.), that’s 17.5 kg, or a little over 38.5 lbs. of body weight acquired in “different” environments over the last 14 days but still in the same building and lumped together in your single metric.
If you visualize the market weights on the chart I have provided you will notice that the distribution is bi-modal and that the left side is extended farther to the left from the approximate mean for the whole group than from the right. When you observe a bi-modal distribution (two modes, i.e., “humps”, or high points), this indicates that there were at least two separate processes or production functions underway in a single barn. This can be from a group of sick pigs, pigs of differing genetics or sex, maternal barrows mixed in, some pigs were added to the building at a different time, some portion of the building maintains a different environment (fans not working, wet spots etc.)
When comparing the different colored weight distributions of the marketing loads, notice that all the groups except the orange-colored distribution seem to have achieved a similar mean sale weight by the time of their marketing but also note how the total weight variation grows for those groups each time they are given extra time to achieve the goal weight. As this happens the mean becomes a potentially deceptive single descriptor for the group or subgroup. Observe how much wider the distributions become as their total days are extended.
We don’t know the circumstances which created the first “modal group” (in orange). Could be they were selected as a group at lower weights to go to a different buyer who pays more for lighter weight pigs or they were sick or had a defect that caused the operator to want to get them away from the rest of the group so they would not compete for limited resources within the building. We note an additional 12 days was given to the barn after the orange group was marketed to achieve better average weights. Lots of stuff is going on in there.
What are we missing by creating an evaluation of the group through averaging the entire group’s metrics, like average ending weight, average growth rate (ADG) and FCR’s for instance? We can look at that by working through the idea of subpopulations next time.
