Misreading misleading charts

March 26, 2009 in Finance,Math

This chart caught my eye because it is potentially misleading (click to zoom).  It shows the year-over-year change in hotel occupancy rates, from 2001-2009.

Hotel Occupancy Rates 20091.jpg

My first impression, on viewing the small chart, was that we haven't hit the low of the last recession.  But (as the box very clearly points out) the low of the last recession was on 9/11, a dramatic outlier not necessarily representative of the trend.

So what?

The problem with Y/Y charts is that any useful information  depends entirely on the level one year ago.  The graph above shows a dramatic boost in occupancy rates in September 2002.  Were people really flocking to hotels?  No, it just happened to be 1 year since the outlier low of 9/11.  Anyone who looked at that datapoint and missed the spurious cause of the high positive number would have thought occupancy rates were through the roof!  In fact, looking at the numbers in late 2002, the September datapoint is +10% and the 9/11 datapoint is -20% -- so rates are likely net lower! This is borne out by the early '03 rates being low compared to the early '02 rates, themselves lower than early '01.

Which brings us to the reason Y/Y charts are so compelling: slow, drawnout downturns don't appear as serious as they actually are, and rapid recoveries look even better than they are.  If rates slowly drop for more than a year, then the drop in rates during the second year won't look as serious because it's measured against an already-reduced level. The first (potentially false) sign of recovery will show up as a series of positive numbers, despite actual rates being depressed.  A rapid recovery from a shock - like late 2002 - is even better because the increase in rates is compared to a low level.  This is not to say that such numbers aren't useful, but that they are meaningless out of absolute context.

To tie this to more immediate numbers, this should give anyone pause who looked at the very-positive February M/M change in durable goods orders and thought "durable goods orders have increased dramatically! The recession is over!"  The orders only increased RELATIVE to an absurdly low January, which was the 6th month in a row to see a significant drop.  Y/Y down over 20% and - again - that's measured against the already-reduced levels in early 2008.

via Calculated Risk: U.S. Hotel Occupancy Rate at 58.5%.

Leave a Comment

{ 1 trackback }

Previous post:

Next post: