There's been a good thread on the Institute of Transportation Engineers traffic engineer's listserv about forecasting traffic. The big question is whether or not the miles driven in the United States will grow, plateau, or go down in the future.
I pulled data out of the Texas Transportation's 2010 Annual Urban Mobility Report for the vehicle miles traveled on an average day in Minneapolis/St. Paul (based on my cursory review of a lot of different datasets, this trend appears to be representative of vehicle miles traveled around the country over the last +/- 30 years).
The vehicle miles traveled per day are shown in the the graph below, along with trendlines between 1982 and 2007 using a 5.69% straitghtline growth rate as well as a 3.60% compounded growth rate. All three trendlines meet at the same values in 1982 and 2007.
A lot of traffic forecasting done in the business relies on trendlining past data forward using a growth rate (straightline or compound as shown in the chart). The question facing forecasters: is the dip/plateau in traffic from 2007 to 2009 an anomaly or the new reality. No one knows. The improving employment numbers vs. rising gas prices are muddying the crystal ball. The oil embargo in the 70's raised the same question, yet traffic grew at a big clip between 1982 and 2007.
As a conservative traffic engineer, I'll continue to trendline traffic up over the next few years until the new trendline solidifies. I'll also continue to say "if I could forecast the future twenty years from now, don't you think I'd be an investment banker instead of a traffic engineer?"
I'm pretty confident my near term traffic forecasts are good enough to decide where we need turn lanes and if we should add a traffic signal/roundabout to an intersection. Basically, the improvements needed right now to make sure traffic can get in and out of a development. I'm also comfortable with the idea of preparing long term forecasts (conservatively) to ensure enough public right-of-way is preserved for future road expansions that may be needed. However, I'm not comfortable spending a lot of money on big improvements today to serve a need forecasted to be there twenty years from now.