Tuesday, September 29, 2009

CORRECTION: 18 state total information

I receive information that the 18 state condition line is weighted by last years planted area numbers. While the simple average is close this week, that is coincidence (and honestly makes much more sense). So the national total weighted by last years planted area changed little, but when you break the states up, there are some declines in major states, and the only increases are in realtively minor or lower yielding states. I still cant determine why we get such a small national change while I see declines in yields for 18 states, increased yields in 6 states and no change in 17 states. The only think I can figure out is that the two biggest changes in percentage terms are 3% plus increases in GA and CO, with a 3% decline in AL. But I'm showing IA and IL and NE yields falling arround 1% from last week.

Last week (VP, P, F, G, Ex)
IL 2-7-29-49-13
IA 2-6-17-49-26
NE 2-5-14-52-27

This week (VP, P, F,G Ex)
IL 2-8-29-49-12
IA 3-6-16-52-23
NE 3-5-13-52-27

So in IL we had a one unit shift from EX to P, not a huge change but enough for a 1% decline in yields
In IA we had 3 units move from Ex to G and 1 from F to VP, enough for 1% decline in yields
in NE it was only one unit from F to VP that gave me 0.8% decine in yields

I use every condition catagory and I do not force some spread betwen the parameters on each condition, I only inforce that as you go up the scale of conditions, the paramter must be >1% than the previous category.

While I would like to think that state by state numbers would provide additional accuracy, I need to investigate how much of a difference it would make. Parsimony has always been one of the objectives of doing this. In this case it the condition movements are dependent on the differences between the condition parameters with a state and across states.

No comments:

Post a Comment