On Friday the USDA will release its annual estimates for perspective plantings for the 2019/20 crop year. Estimates for perspective planting by most fundamental research groups are generally done via polls of their clients planting intentions. The USDA's own prospective planting report, according to their stated methodology, is also based primarily on surveys, conducted in the early weeks of March. As the 2016 presidential election highlighted however, polling can produce forecasts well off the mark. As a technical research company, we took an alternative approach and drew some opposing conclusions to the mainstream forecast.
Focusing on corn and beans we start by estimating a baseline for the total potential acreage of both crops. The area chart on the right is the USDA's data for plantings from 1986 to date. Last year, total corn and beans planted were 178.3 million acres broken down by 89.1 million acres of corn and 89.2 million acres of beans. The early USDA estimates for 2019 plantings show a decline of 0.7% to 177 million acres, comprising of 92 mln acres corn and 85 mln acres soybeans. The large reduction in soybeans a reflection of lower prices and the ongoing trade battle with China.
Will farmers make such a large rotation out of beans and into corn?
We are not doubting the results of the USDA polls, however polling can have sampling and non-sampling flaws. Planting intentions in March can change as market conditions or fundamentals change and this year will be no exception, especially with the timing of trade negotiations. In addition, flooding in the Midwest favors beans if early planting is delayed. As such we take a more trend based approach and provide an alternative view for your consideration.
Applying a simple linear regression (with a high r-square value of .95), forecast plantings for this year would total a record 180.7 million acres. If we just wanted to make a quick prediction, this would be a good start except that it over emphasizes the long term upward trend. If we make a simple adjustment and "de-trend" the data, we can then take a simple weighted average of the changes and produce a better estimate. Using de-trended data we estimate a net total change of +0.90 million acres from last year, producing a new benchmark of approximately 179.2 million acres. If we were to run this same process on corn and beans separately, our estimates for each would be roughly 86.5 million acres corn and 93.0 million acres of soybeans, in stark contrast to the latest USDA estimates. Note: Totals will not exactly equal 179.2 due to de-trending calculations.
This large difference in our estimate and the USDA estimate is the difference between planting trends (our estimate) and planting intentions (USDA estimate). In addition, we make no attempts to adjust plantings based on a forward projected price level.
The chart to the left shows plantings of corn and soybean acres over the same time period as above but on the same axis. Last year, plantings of soybeans topped corn by the smallest of margin, but it was only the second time in history (1983 last). As you can see from the chart, this was a combination of increasing bean acres and decreasing corn acres beginning in 2013. Certainly a contributing factor to this divergence was the price spread between soybeans and corn which bottomed in 2011.
How much is price really affecting planting decisions?
To answer this question, we need to de-trend the data once again. With the chart below, we begin to see a clearer picture as to the effect of price (soybean/corn ratio) and plantings. For comparison purposes, we indexed both corn and bean plantings to 100 at the beginning of our data. In addition, we have overlaid the ratio of average prices received for soybeans / average prices received for corn for the marketing year.
By doing this we begin to see two patterns worth noting. The first is, on a slightly delayed basis, there appears to be a small negative correlation between corn plantings and the bean/corn ratio. This of course, is what we would have expected. From 1986 to roughly 1995, the ratio narrowed (ie corn prices more attractive) and plantings for corn trended higher on a relative basis, irrespective of the overall trend in plantings. As the ratio widened and peaked in 2003 relative plantings for corn declined to the benefit of soybeans.
The second pattern of note, is what appears to be a planting cycle that lasts on average about 8 years. This is significant because in 2018 corn crossed back below soybeans. This data of course points in the opposite direction of the USDA's most recent forecast and would suggest bean planting could outpace corn for the next few years.
Additionally, when we test for correlation between the price ratio and de-trended plantings, it is not statistically significant enough to draw any solid conclusions. In fact the ratio is far more erratic as compared to the slow moving cyclical pattern in planting. This might be explained by the desire to rotate between crops irregardless of price or a long term risk management strategy of consistent plantings.
It is important to note that this report is not intended to forecast plantings but to highlight the trends versus the current surveys. Our non-survey data analysis would suggest two important distinctions between the current USDA survey. First, total acreage planted for corn and soybeans should increase if trends remain intact. Second, the de-trended data suggests a cyclical pattern that should continue to favor soybeans over corn although it might be somewhat reduced by the decline in the soybean/corn price ratio. In conclusion, while our analysis is not an attempt to forecast, it does suggest the possibility the survey forecasts may be a result of over emphasizing price impact and underestimating underlying trend.