The Details and the Devil Therein
My senior year of college, which is getting to be an alarming count of decades in the past, I finally got around to taking “Public Speaking,” a freshman-level course. As a college veteran impatient to wrap things up I had none of the trepidation that an introvert usually feels about public speaking. I always volunteered to go first and get it over with.
The others followed according to preset schedule, and what happened next was comic and given human nature, entirely predictable. Early in the order a student with a more, say, relaxed view of academic achievement wouldn’t be ready. Next in line was suddenly in the frame, only they thought they were speaking tomorrow—not ready either. The rest fell like a row of dominoes, including students who otherwise wouldn’t miss a deadline, ever.
There’s an allegory in here somewhere.
“Averages hide more than they reveal.”—most statisticians
FINBIN is a helpful database if you want to get a more granular look at the state of agriculture. One is still confined to averages, or at least, to data populations large enough to preserve the anonymity of individual farms. It is possible, with any data group, to produce forecast scenarios testing variations in key inputs, especially, yields, crop prices, crop inputs, and rents. Thus far, we have found that on average (!), almost any way you slice FINPACK data (size, region, livestock, crops, etc.), working capital remains very positive and debt:assets low relative to 25-year history available through FINBIN.
After the unprecedented-in-recent-history profits 2020-2022, one would hope that ordinary cyclicality in agriculture would still leave farms with ample working capital and moderate debt-to-asset ratios. Indeed it has. Our own forecasts show that regardless of the persistent high costs matched with very bearish crop markets, farms on the balance should finish 2025 with still-strong balance sheets relative to history.
In a conversation with Matt Bialick a few weeks ago, we talked about the problems of averages and how in real life all the action happens at the fat tails [of the normal bell curve distribution]. It inspired me to slice the FINPACK data another way: by overall debt levels (sadly one can’t easily parse it by liquidity).
I sorted for crop farms in MN from I94 & south, 1,500 acres and up, segregating those with <40% debt: assets from those with less. This filters out part-time crop enterprises and gives us a more homogenous group of operations.
By way of summary:
Rather more farms are “high” debt than low, 63% of our sample.
The high group averages (there’s that word again) 44%, the low, 19%. It’s worth noting that the trend is down for both classes (not shown in the reduced summary above).
Lower debt historically coincides with higher yields and better crop marketing performance.
Lower debt operations own more land (615 vs. 387 acres).
Lower debt operations are more liquid, 323 days’ expense in working capital vs. only 87.
Disturbingly, land valuations with high debt operations are 30% higher, $8,444 vs. $6,132, which too easily looks like window-dressing to shore up reported balance sheets and maintain status quo.
That’s the negative, and there’s little here that bucks our expectations. Here is some more positive counterpoint:
High debt operations are growing faster (2000-2024). Ergo, higher debt is funding capex needed to expand acreage under management.
Return on assets is almost the same regardless of debt level.
Related to the above, EBIT in 2024 is slightly higher (or losses less) in high debt enterprises. It’s the higher interest cost (due to higher principal and higher rates) that eats up the difference. It’s almost as if to say, all things being equal, isn’t it nicer to be richer. Or, necessity is the mother of invention.
Anecdotally, the poorest-performing operation I ever ran analysis on was 4,000 acres, all owned and debt-free. Return on assets, however, was in the tank. Had they sold the lot and put the proceeds in passbook savings it would’ve increased their net revenues and decreased their risk.
Ergo, debt levels alone are not predictive of performance. In the current environment they do indicate serious fragility. Current conditions might produce a loss for the low-debt operations this year, but they could be left with $1.4 million in working capital, a slight reduction from 2024. The high debt operations opened with $460,000 and closed with a slight deficit. With balance sheets under stress and (perhaps) the $8k land valuation under scrutiny, their lenders might look askance at terming out short-term debt to restore working capital.
On average everyone’s fine, but splitting into just two groups, the larger of the two is under significant stress. There are enough dominos lined up where the weakest might knock into the stronger ones adjacent and send them falling where otherwise they might have stood.
I said there was an allegory in there.
A lot could happen between now and 2026 loan renewal, but even slightly peeling back the curtain we call “averages” as I’ve done here reveals significant stress. I don’t propose that 63% of farms are likely to fail, even in the worst case. The last time we had major value reset, not coincidentally, concurrent with a USD crisis of lesser magnitude than the one now imminent, the stress looked like this:
The jump in exits in the 80s is unmistakable, but also, didn’t even crack 2% annually and only just exceeded the background level of exits in the “boom” years of the 70s. And yet, land prices dropped by half or more and took 20 years to recover. A minority of small dominos can have a major impact.
History rhymes and doesn’t repeat, as I’ve repeated myself too often. At the same time, since I’m resorted to cliches, this time is never different.