The Dangers of Non-Seasonal Data

Ed noticed that while the August release pointed to a decline in consumer credit on a seasonally adjusted basis, it increased on a non-seasonally adjusted basis during the month. Thus, the consumer was not necessarily develeraging. My response on his blog was:

Why disclude a seasonal adjustment? I am not positive as to why it would naturally jump in August, but back to school shopping seems like one obvious answer. I guess my question would be, would you want to ignore seasonal adjustments for toy sales in December too?

If you want to ignore those trends, then take the unadjusted data and look at year over year figures. August '09 was down 4.4% compared to down 4.1% in July '09 vs. July '08. In other words... consumer credit is falling.

While Ed did agree that the year over year Consumer Credit figure declined, he noted:

But looking at SA month-to-month data in isolation and saying the sky is falling completely ignores the fact that consumer credit just ticked up for the first time since December. One could explain this via cash for clunkers, but to overlook it doesn't make sense to me.

But looking at the data, it appears that overlooking the adjustment is even worse. What do we see? August was the lowest month over month change from July going back at least 40 years. And this comes on top of a massive decline in Consumer Credit outstanding on seasonal or non-seasonal basis in July from June.



So do I think consumer credit is declining at a rapid pace even though it is up on a non-seasonal basis in August? Absolutely. But it appears that Ed may have already come to the same conclusion.