Saturday, December 12, 2009
Yesterday, in comments, DataMunger suggested using the EIA's weekly product supplied series, instead of the refinery production numbers. This morning, I tackled that, using the same seasonal adjustment approach (easier the second time, as one can cut and paste into the same spreadsheet). Firstly, here's the raw data:
Interestingly, it doesn't look strongly seasonal, and indeed that is confirmed by computing the seasonal adjustment factors (along with yesterday's series):
It's not really clear there is too much seasonal adjustment here, versus just noise. I find this a little puzzling - the product supplied is not seasonal, but, at least the subset of refinery products I picked on is? Anyone have any insight into that? The most obvious back and forth in the situation would be summer gasoline demand versus winter heating oil demand, but both of those are included in my "refinery production" sum.
Anyway, putting that series, with the same kind of 27 week centered moving average, on the same chart as the other, you can see it's a similar story:
The "product supplied" numbers, present a very slightly more optimistic picture - the series stabilized a couple of months earlier, and now looks to be in a weakish (if bumpy) recovery, though looking at the earlier behavior of the series, it's probably not statistically different from "flat".