There's been a lot of talk of downturns lately in Pricing and Revenue Management (PRM) circles. It was a major theme of the recent NAA Maximize conference and was also covered at NMHC OPTECH last week. While no one knows when the downturn is coming, everyone seems to agree that it's a good idea to plan for it. While occupancy and rent growth are still strong, the sheer length of the recovery since the last recession has everyone wondering how much longer this bull run can last.

Recent discussions got me remembering some things about the last recession. Many of you may know (and many may not) that the National Bureau of Economic Research (NBER) is the body that officially declares when recessions begin and end. According to them, the Great Recession began in December of 2007 and ended in June 2009, a whopping 19-month long period of economic decline.

The challenge with the NBER is that they typically don't declare a recession until a year or more past when it started (the NBER declared this last recession on December 1, 2008, which was almost exactly one year past when it started). That is useful for academics doing historical research but not of much use to business people trying to make decisions.

At the time, I was still responsible for pricing and revenue management at Archstone. I remember thinking that it would have been nice to know we were in a recession a year ago so we could have changed our strategy appropriately. 2007 had been such a great year that no one thought we were in trouble in the first half of 2008 (it wasn't until the Lehman collapse on September 15, 2008, that we realized we were in some deep trouble). Rents had been rising and occupancy was stable through the first half of 2008, so we were pretty comfortable (or so we thought).

Then it hit me: "I wonder what our PRM system was seeing." I knew that there was a key statistic, de-seasonal demand, that might shed some light on the economic conditions. For those not familiar with the term, de-seasonal demand is the algorithm's forecast of lease demand stripped of the seasonality at the time. The system (LRO in Archstone's case) calculated and updated this core statistic each week (LRO measures demand by week). The system had a full history of that calculation going back several years.

I pulled the appropriate data from the database (it's not available in the user interface, so it required a back-end query) and looked at de-seasonal demand from January 2006 all the way through December 2008—and there it was. De-seasonal demand grew throughout 2006-07, showing a peak and then a clear decline driven by the recession. And when was that peak? Right in December 2007—exactly when the NBER said the recession started.

I was a bit blown away. We might have been thinking things were fine in the first half of 2008, but LRO recognized it. Sure, rents were rising in the first half of 2008; but that was due to seasonality, not market strength. And with LRO, they weren't increasing as aggressively as they would have since the software was seeing the slow decline characteristic of the early part of that recession. It was reacting to changing cyclical market conditions that we humans were not yet able to detect!