Kub's Den

Corn and Fertilizer Prices: Correlation Doesn't Mean Causation

Elaine Kub
By  Elaine Kub , Contributing Analyst
Over the long term, corn prices and nitrogen-based fertilizer prices seem to display a relationship. (DTN chart using USDA data)

When we look at a chart of fertilizer prices plotted alongside a chart of corn prices, we immediately spot a pattern and assume there is a strong, predictable relationship. Prices were relatively high for all of the above when the U.S. planted upwards of 95 million acres (ma) of corn. From July 2012 to July 2013, nearby corn futures prices averaged $7.22 per bushel while dry urea fertilizer had a price tag around $588 per ton, liquid anhydrous fertilizer fetched $842 per ton and UAN28 was $387 per ton.

Prices then dwindled in a mostly synchronous fashion while the profitability and popularity of planting corn dwindled. In 2017, when U.S. farmers barely planted more than 90 ma of corn, the calendar-year average for nearby corn futures came to $3.59 per bushel, dry urea was $336 per ton, liquid anhydrous fertilizer was $458 per ton and UAN28 was $230 per ton. In other words, corn prices dropped by roughly half over those four-and-half years, while fertilizer prices simultaneously dropped 40% to 45%.

Interestingly, fertilizer prices have recently rallied. DTN collects roughly 1,700 retail fertilizer bids from 310 locations weekly, and from a low last fall (September/October) to the start of the 2019 planting season, urea prices are now 13% higher and anhydrous prices are 24% higher, moving from $480 per ton to $597 per ton (U.S. average) last week. Liquid anhydrous ammonia is consistently the cheapest way to purchase nitrogen. On a per-pound basis, those anhydrous prices have moved from 29 cents per pound of nitrogen to 36 cents per pound of nitrogen.

Corn prices haven't followed along with that fertilizer rally of the past six months. Nearby futures were roughly $3.60 last fall and they're roughly $3.80 now, a bump of about 5%. Yet when we look at that chart of the two markets moving together, we see the long-term pattern and our brains start to think, "Hmm. Shouldn't we expect corn prices to catch up with the shift in fertilizer prices?"

Unfortunately, no -- not necessarily. These two markets (corn and nitrogen-based fertilizer) are indeed strongly correlated. But correlation does not equal causation. Even without getting into the math, we can already logically figure out that fertilizer prices are likely dependent on corn prices (over the long term) and not the other way around. When the demand for corn-growing inputs rises, the price of nitrogen fertilizer likely rises (all else being equal). However, a rise in the price of any input -- whether it's nitrogen fertilizer, seed, land or borrowing money -- does not necessarily follow-through to a higher price received for corn.

Then there's the math itself. The first problem is spurious correlation. Just looking at a chart of two variables and noticing a pattern doesn't mean much -- the human brain can draw a pattern based on any chance relationship or no relationship at all. Maybe a chart of the price of green beans in Sweden would look like it rises and falls in synchronicity with the monthly marriage rate in Mexico (or whatever). Spurious correlation may also arise when there's not necessarily a direct relationship between one variable and another, but when there is a relationship between each of those two variables to a third variable. I've seen a classic example showing that human height is positively correlated with the number of words in a person's vocabulary, but I'm 5 feet 3 inches tall and I promise you I have a sufficiently large vocabulary, thank you very much. The underlying relationship between those two variables is the age of the humans, with babies (20 inches tall or so) throwing off the curve.

I suspect that's a big part of what's happening here on the corn and fertilizer chart. Both sets of prices are quoted in dollars (dollars per bushel or dollars per pound of nitrogen) and all dollar-denominated commodity price tags will, by default, rise when the U.S. dollar weakens and fall when the dollar strengthens (all else remaining equal).

One way to screen out the effects of spurious correlation is to calculate correlation not between the actual prices from one week to the next, but between the changes in prices from one week to the next. Corn prices gained 1.33% last week, for instance, and average fertilizer prices gained 0.07% during the same timeframe. Quoted that way, it doesn't matter that the underlying unit is dollars.

The correlation between the weekly returns in corn futures prices and the weekly returns in average retail fertilizer prices is positive 0.11, which doesn't sound like a lot, but actually, it's considered meaningful in statistics-terms (significant at the 0.005 level of significance).

There are more disclaimers to make. All this math was done on nearby corn futures, which track the market for old-crop corn (physical grain already in existence), while the fertilizer market is a reflection of expectations in the new-crop corn market (production inputs for corn plants that are intended to grow in the future). Also, both of these markets reflect supply and demand dynamics beyond the borders of the United States, but the prices used here were U.S. prices. In addition, nitrogen-based fertilizers are used by farmers for many crops besides just corn: notably wheat, other small grains, canola, sugar beets, hay crops and even soybeans (although not in the quantities required for high-volume corn yields).

Nevertheless, despite all the disclaimers, I think we can see this clear relationship between corn prices and fertilizer prices and then draw some conclusions about causation. Demand for fertilizer, or any corn production input, is ultimately determined by the acreage planted. In 2019, the acreage that's going to be planted to corn is generally expected to be relatively large. The USDA will release its annual Prospective Plantings survey on Friday, March 29, and there's a feeling that the survey number could show around 92 ma of corn-planting intentions. However, those intentions will ultimately be subject to spring weather -- including flooding, which could limit access to Corn Belt fields in the timeframe best suited to planting corn.

So yes, corn prices and fertilizer prices may both display a relationship with U.S. corn acreage. But this year, the unusual thing about that corn/fertilizer price relationship is that the corn acreage is being driven, not by profitability and price of corn itself, but by the unprofitability of other crops.

***

To track fertilizer price movements as planting season continues, watch for DTN Retail Fertilizer Trends, released weekly on Wednesdays by DTN Staff Reporter Russ Quinn, whose help on my analysis for this article was much appreciated.

Elaine Kub is the author of "Mastering the Grain Markets: How Profits Are Really Made" and can be reached at elaine@masteringthegrainmarkets.comor on Twitter @elainekub.

(BE/BAS)

Elaine Kub