A deep dive into recent U.S. soybean export data suggests that domestic soybean supply could be significantly lower this year than the market expects.

The U.S. Department of Agriculture will update balance sheets in its monthly supply and demand report on Friday at noon EDT (1600 GMT). The recent strong global soybean demand means that the market will be watching in particular for possible adjustments to U.S. soybean carryout for both the old and new marketing years.

Impressive domestic soybean exports in recent weeks have applied extra pressure to the U.S. soybean supply situation heading into the final month of the 2015/16 marketing year, which ends on Aug. 31.

Weekly export data indeed shows that the world’s No. 2 soybean supplier is shipping the oilseed at a fast clip, but the weekly data sets might not reflect the complete picture and in fact almost certainly under-represent it.

A close look into all government data reveals that U.S. soybean exports could be much larger than USDA’s latest projection and this could have big consequences for domestic soybean carryout moving forward.

Data divergence

There are three different sets of data issued by the U.S. government in order to track grain exports: inspections, sales and customs.

The inspections data tracks the tonnage of grain inspected by federal employees at the ports each week. The sales data is based on the sold tonnage that exporters have reported to the U.S. Foreign Agriculture Service. It also publishes a weekly exported amount but is primarily focused on reporting the sales.

The monthly customs data given by the U.S. Census Bureau is the official record of exports, but there is a lag time in the publication. For example, data through the end of June was made available on Aug. 5.

While the market waits for the customs data, it uses the inspections and sales reports for guidance on recent export activity. The inspections and sales reports are issued weekly on Monday and Thursday, respectively.

In theory, all three of these data sets would be very close to one another, and they usually are. But this year, the customs data through June implies that soybean exports are between 3 percent and 4 percent higher than what the inspections or sales data suggest. For reference, the same deviation last year was less than half a percent.

Representatives responsible for each set of data were contacted and none were able to explain why the historically large discrepancy exists. Most of the speculation pointed to a possible lag in reporting by the exporters or even the failure to report some of the shipments.

But since the customs data has the final word, it must be assumed that total soybean exports are indeed larger than what the weekly data sets suggest. And given some of the latest inspection numbers, this is very easy to believe.

Unprecedented pace

Last Friday, the customs data from the U.S. Census Bureau showed that the United States had exported 45.86 million tonnes of soybeans through June for the 2015/16 marketing year despite the slower start to the season.

In order to reach USDA’s latest figure for 2015/16 U.S. soybean exports of 48.85 million tonnes, a total of 3 million additional tonnes would need to be shipped in the final nine weeks of the season, averaging out to about 333,000 tonnes per week.

July and August are usually two of the quietest months for U.S. soybean exports, and inspection data shows that a weekly average over 300,000 tonnes during this time is historically difficult to maintain (chart).

But over the last five weeks, weekly inspection data implies that an average of about 617,000 tonnes of soybeans per week have been shipped, with the most recent week leading the way. On Monday, USDA reported that soybeans inspected for export in the week ending Aug. 4 totaled a staggering 972,000 tonnes.

Not only do these numbers suggest that soybean exports have surpassed USDA’s July estimate, but maintaining this pace in the final four weeks of the season would add to it an additional 2.5 million tonnes (92 million bushels).

And if 2.5 million tonnes of soybean exports were to be added to USDA’s July estimate of 48.85 million tonnes, this would place 2015/16 soybean exports just over 51 million tonnes. Of course this is considering the most aggressive scenario, but if such large exports were indeed realized, current year soybean exports would surpass the 2014/15 record of 50.1 million tonnes.

http://pdf.reuters.com/pdfnews/pdfnews.asp?i=43059c3bf0e37541&u=2016-08-09T100543Z_GFXEC891PDJ1I_1_RTRGFXG_BASEIMAGE.pngThe only area where U.S. soybean exports are likely to fall short in the current marketing year is in fulfilling all of the bookings. Weekly sales data through July 28 shows that there were still over 6 million tonnes in outstanding sales in 2015/16, more than double what it has ever been at this time (chart).

But unless a large amount of cancellations come in at the last minute, any of these outstanding sales will be rolled to the 2016/17 year and eventually shipped. Cancellations are unlikely at this point since U.S. soybeans are cheaper than their Brazilian counterpart.

USDA balance sheet impacts

The latest export number from customs plus the subsequent pace of implied soybean exports by inspections suggest that USDA may have to boost its estimate of 2015/16 U.S. soybean exports eventually, if not this Friday.

For one, an increase on Friday would be in line with the recent trend of USDA’s export adjustments. Since March, USDA has steadily increased current year soybean exports each month for a total increase of 105 million bushels over that time frame (chart).

As previously mentioned, maintaining extremely strong exports during August could tack on as much as 2.5 million tonnes (92 million bushels) of soybean exports to USDA’s most recent balance sheet for the 2015/16 marketing year.

On average, analysts polled by Reuters believe that USDA will place 2015/16 U.S. soybean carryout at 320 million bushels (8.7 million tonnes) in Friday’s report. This would amount to a reduction of about 820,000 tonnes from the July estimate of 350 million bushels.

The lowest trade estimate for 2015/16 soybean carryout is 254 million bushels. This would be a reduction of 2.6 million tonnes from USDA’s July figure.

Assuming that everything else on the current soybean balance sheet stays the same, these calculations suggest that the average http://pdf.reuters.com/pdfnews/pdfnews.asp?i=43059c3bf0e37541&u=2016-08-09T095444Z_GFXEC891OV87E_1_RTRGFXG_BASEIMAGE.pnganalyst estimate of 2015/16 carryout could be too high depending on how large the potential revision to exports may be and especially if August shipments are very strong.

This means that a sub-300 million bushel carryout in the current marketing year is not only possible but realistic. Even though the lowest analyst estimate for 2015/16 is much higher than 2014/15’s carryout of 191 million bushels, it is much lower than what was being considered a year ago at this time (chart).

A lower 2015/16 carryout would mean that the United States will begin the 2016/17 season on Sept. 1 with less supply. USDA’s adjustment trend to 2016/17 soybean exports have been the same as 2015/16 – they have increased each month since the initial projection in May.

Given the recent demand for soybeans and no reason to believe a slowdown is likely heading into next year, U.S. soybean exports for 2016/17 could also be bumped up on Friday.

But some offset to 2016/17 carryout is likely with an expected increase to 2016/17 soybean production. On average, analysts believe that USDA will raise soybean yield to 47.5 bushels per acre from 46.7 bushels per acre in Friday’s Crop Production report, also due out at noon EDT.

At 47.5 bushels per acre, soybean production would increase by 62.5 million bushels when assuming that USDA is unlikely to adjust planted or harvested acreage. The top yield predicted by analysts, 48.8 bushels per acre, would increase 2016/17 soybean production by 170.4 million bushels.

If yield does come in on the top end, the increase in supply could more than offset the loss from a likely export hike in both marketing years. But if analysts have nailed the soybean yield, 2016/17 soybean carryout could be on its way down as well.