When should the Christmas decorations go up?

Tim Harford asks whether it is efficient to have the Christmas decorations go up earlier each year.  Partly for traditional and aesthetic reasons, he would prefer a shorter Christmas season. 

In my basic model, suppliers with P > MC put up decorations to stimulate more buying and gift-giving.  Bringing on Christmas early is a form of associative advertising.  A shorter Christmas season would mean less overtime work, less spending (more saving), and spending spread more smoothly throughout the year.

What’s the benefit of having all that consumer spending clumped together at Christmas season?  I can think of three hypotheses:

1. It yields "thick market" externalities, such as allowing people to do most of their shopping when inventories are highest.

2. It yields a "do or die" season, in which new products can be rapidly and ruthlessly evaluated.  Don’t we already have the verdict on Zune?

3. Suppliers are inefficiently "fishing" for early "capture" of consumers as part of a common pool problem.  If a given supplier doesn’t grab that consumer’s attention now, someone else will.  Of course there can be no property rights in "the attention of consumers," so that attention is consumed inefficiently early.

I believe in #1, #2, and #3, but I suspect #3 is the operative force, leading me to side with Tim.  Christmas is fun, but perhaps we have just a little too much of it.