Fun with statistics

Today's retail sales figures looked strong across the board, surprising both Macro man and the market. Fixed income has not liked the data at all, with the ten year yield climbing to its highest level since Thanksgiving.

The breakdown was strong, peculiarly so (building materials up 1.8%?!?!?!?) This will no doubt lead some (David Rosenberg at Merrill, perhaps?) to suggest that the data was actually quite weak, and provide some piece of data to "prove" it.

Of course, it has often been said that you can prove anything with statistics. The state of retail in the US is no exception. Consider the chart below. It shows three separate measures of retail sales.

The orange line is at all time highs: happy days are here again!

The blue line has corrected off its highs and shows signs of basing: neither the end of the world nor party time.

The red line shows a sharp deceleration towards the lows: uh-oh!

So what gives?

The orange line is the indexed level of total retail sales in dollar terms: sales are indeed at all time highs.

The blue line shows y/y retail sales indexed: basing, perhaps. but skewed by base effects and data in the past.

The red line shows the 3 month on 3 month rate of change indexed: evidently falling off the edge of a cliff! (albeit with base effects of its own via the weak September reading)!

So choose your poison.

Are you an economy bull? Point to the orange line.

An economy bear? Point to the red line.

A soft landing aficionado? Point to the blue line.

So remember the next time that anyone shows you economic "evidence" via a ccouple of simple charts: if you look hard enough, you can prove anything with statistics.