Just when analysts thought they had figured out the pattern in the U.S. Department of Agriculture’s U.S. soybean forecasts, the agency’s latest estimate broke form and created even more uncertainty in the market.
In its monthly U.S. crop production report on Oct. 19, USDA increased harvested area of the oilseed by 740,000 acres but lowered the yield by 0.4 bushel to 49.5 bushels per acre.
The area boost was somewhat expected by the trade, but the yield move was not. The average market guess was that USDA would increase yield to 50 bu./acre from 49.9 in September, but only one out of the 20 polled analysts placed yield at 49.5 bu./acre or below.
In any other scenario, this seemingly small move would not be as shocking or consequential. But in recent years, USDA’s soybean yield estimates have tended to build over time even when weather was less than perfect, leading many to believe the agency might be lowballing from the start.
The adjustments to U.S. soybean area and yield cancelled each other, and the production peg ended up the exact same as last month’s figure, an all-time high of 4.431 billion bushels. This meant that the only change to the balance sheet was the 44-million-bushel cut in old-crop stocks, which the market already knew about from USDA’s Sept. 29 stocks report.
But a steady harvest volume must have felt like a smaller harvest volume to traders, as benchmark November CBOT soybean futures jumped nearly three per cent the day of the report. Follow-through buying the following day broke the $10-a-bushel mark (all figures U.S. funds) for the first time since Aug. 1.
The yield reduction has left many market participants stumped as to where the estimates go from here, as USDA’s uncharacteristic move has created concern that the crop could get even smaller as more harvest results roll in.
USDA had not lowered soybean yield in its October report since 2011. Further, the agency had not reduced soybean yield in any report from August through November since 2013.
An October reduction may have been in store in 2013 considering sizable yield cuts in the two preceding months, but the agency did not release that month because of a government shutdown, leaving 2011 as the most recent October yield cut.
However, soybean yields and characteristics appeared to have changed in 2012, in which the soybean pod weights topped 2009’s record by five per cent. Ever since then, pod weights have only been higher.
Generally, higher pod weights can be justified with a slightly lower pod count, and vice versa, unless the weather is absolute perfection. But this criterion might not exactly fit the more recent years, as pod counts have been all over the board.
In 2013, the lowest soybean yield that USDA printed was 41.2 bu./acre in September. The November number came in at 43 bu./acre and the final at 44, which at the time tied 2009’s record. But USDA’s initial moves that year do not align with those of 2017, so this somewhat eliminates 2013 as a comparable year in terms of the agency’s tendencies.
What all this suggests is that we truly might be in uncharted waters with 2017’s soybean harvest, as the past does not offer too many clues.
Some analysts wonder if USDA’s October soybean forecast was perhaps less meaningful than usual given that harvest pace is at a three-year low, thus reducing the amount of results the agency could use to comprise its yield estimate.
The fear is that if earlier-planted beans fared better than the later-planted ones, national yield could drop even further in November or January as those results are realized.
Earlier-planted crops are often at less risk because they can go through pollination and fill under less stressful weather conditions, but this cannot be generalized as the performance of early versus late varies both regionally and year by year.
Early-October harvest progress does not necessarily indicate the likely movement in USDA’s yield estimates going forward, but data since 2000 suggests that reductions after October took place when this pace was ahead of average, not behind.
Final January yield landed lower than the October peg three times (2000, 2003, 2010) and early-October harvest was ahead of average in those years. Final yield came in below USDA’s November number five times since 2000, and in four of those times the same harvest pace was quicker than normal.
However, the low amount of data points in this analysis is certainly limiting, or perhaps it just proves how rare it would be if final 2017 yield ends up lower than 49.5 bu./acre.
Despite the potentially shrinking U.S. soybean crop, domestic and global soybean supplies are still comfortable, for now. But like the U.S. yield situation, analysts know that soybean carry-outs are highly subject to change throughout the marketing year, and this may continue to support the uncertainty going forward.
USDA projects 2017-18 U.S. ending stocks at 430 million bushels, higher than the year-ago 2016-17 prediction of 395 million. Final 2016-17 carry-out came in lower than earlier ideas at 301 million bushels, and interestingly, USDA’s initial May 2016 figure for last year’s ending stocks of 305 million bushels was the closest of any other estimate to the final.
Similar to the U.S. stocks, USDA has also had a tendency to reduce the global soybean carry-out estimates over time.
In the report, USDA placed global carry-out in 2017-18 at 96.05 million tonnes, which is the amount of soybeans the world is predicted to have approximately one year from now. This would be an all-time high, narrowly topping last year’s 94.86 million.
However, USDA’s global trimming trend was massively violated in 2016-17 as carry-out has swelled nearly 40 per cent since the initial numbers were published, the result of substantially higher production volumes worldwide.
But last year’s U.S. carry-out slimmed down anyway, as much of the global supply has built in South America. When the local 2016-17 soybean marketing years end in March 2018 for Argentina and January 2018 for Brazil, the former will be holding a landslide record amount of soybeans and the latter’s inventory will hit a six-year high.