Informa caused a few sparks (nice pun and it's not even 10 am yet) Friday with their corn production estimate. Although they pegged yields at 164.8 bu/acre, higher than Allendale and FCStone's estimates from earlier in the week, they suggested that things could ultimately prove to be significantly lower.

Their 164.8 bu/acre estimate implies a 13.349 billion bu crop using the USDA's August harvested acres, but their 158.5 bu/acre "downside" suggests a crop of only 12.839 billion bu. The latter number being well below the USDA's 13.365 billion bu estimate from last month.

That was plenty of fuel to light a fire under a market that already almost universally expects that last month's USDA yield estimate for corn of 165 bu/acre is way too high. Indeed many are now thinking 160 bu/acre is a best case scenario. Informa, you see, are normally famously on the high side of trade expectations.

A quick flick back through my records shows that in September 2009 Informa's yield estimates were pegging the US corn crop at 13.304 billion bushels, it ultimately ended up at 13.110 billion. In September 2008 they were suggesting 12.406 billion, against final production of 12.101 billion.

It seems unlikely to me that the USDA are going to drop 5 bu/acre in their yield estimate in just one month when they come out with their prediction on Friday. I'd guess that 162-163 bu/acre might be nearer the mark. Even so, the trade will probably quickly discount that as being albeit a step in the right direction, still too high again.

Footnote: Although Informa are sometimes regarded as being regularly too high, there are others who appear to be frequently on the low side. When bundled all together we should have a pretty accurate mix you'd have thought, especially this late in the season, right? Well not that accurate actually, coming into the USDA's September report last year the average trade estimate for corn was 12.901-12.932 billion bushels (depending on who's survey you picked) and for soybeans 3.249-3.253 billion (ditto). Both those estimates finally turned out to be significantly on the LOW side. In the case of corn 178-209 million bushels too low. In the case of beans 106-110 million bushels, beneath final output.