Traffic Forecasting - Past Performance Does Not Guarantee Future Results

I'm working on a traffic study where the reviewing agencies asked us to prepare 20 year forecasts in addition to looking at the build out year.  Typically, we look at data trendlines on nearby roads and throughout the county to determine a "typical growth rate" for traffic in the area.  This has historically been an annual growth rate of 1 to 3%.  This often leads us to factoring traffic up 50% or more over 20 years and then layering on the traffic from the proposed development.  Factoring existing traffic volumes up approximately 50% is also how traffic forecasts are often prepared for road design projects.

I'm strongly reconsidering this approach.  Consider Figure 1 below from the Federal Highway Administration's Office of Highway Policy Information website.  From 1986 to 2006, traffic on all of our highways did fit the model of growing by about 2 to 3% a year (or 50% to 60% over the 20 years).  But since 2005, we've had a significant drop in traffic and the trendline sure makes it look like traffic growth has plateaued. 

I've also been reading through the Texas Transportation Institute's 2011 Urban Mobility Report.  They talk about the terrible state of congestion.  I'm not going to argue that congestion doesn't suck, but take a look at Exhibit 1 from their report.  Compared to 1982, congestion is indeed terrible now compared to 30 years ago.  But looking at the 2000 to 2010 data shows congestion peaked in 2005 and we're not much worse off than we were a decade ago. 

Also consider the August 2011 to August 2010 data below from the Federal Highway Administration.  From their website, "Travel on all roads and streets changed by -1.7% (-4.6 Billion vehicle miles) for August 2011 compared with August 2010."   

But maybe Minnesota's different...... NOPE.  Below is Figure 1 from Mn/DOT's Vehicle Miles of Travel Trends in Minnesota:  1992-2010 showing traffic in Minnesota has followed the national trend in plateauing.

Based on national and local trends, my conclusion is that it is very reasonable to think traffic growth has plateaued.  The punchline for traffic impact studies: the "no-build" traffic forecasts should be the same as the existing traffic volumes.  We don't need to do opening day forecasts and 20 year forecasts because they can reasonably be expected to be similar. 

And given our huge budget shortfalls, this should also mean a policy of fixing the infrastructure we have.  NOT expanding our transportation system to add capacity.

Comments

[Reuben]

As always, this can be tied back to money. As long as outside funding sources (e.g. state, federal) continue to reward agencies with inflated no-build volumes, there is little benefit to agencies for projecting more realistic no-build volumes.

In addition, many agencies rely on inflated no-build volumes to justify the "need" for a project as required by federal environmental documents.

[Scot]

I would very much like to see a realistic look at these, especially with regard to peak hour traffic. Growing peak hour traffic is functionally impossible past a certain technical threshold. As the road's capacity is reached the peak hour spreads out, it can't get "taller". But too often I am asked to apply agressive growth rates for 20-30 years. What is a very busy but manageable road at 2,400 peak hour vehicles in the main direction of a 6-lane arterial is just ridiculous when that is doubled.

We did see a noticeable and sometimes significant drop in daily volumes around here (DFW) corresponding to the economic downturn of 09-10. That has now reversed and most counts in built out areas are about where they were in 2007 before the decline. Regarding your point, here the overall trending is mixed, as the region is still growing at a decent pace, so some roads are in the growth stage as more uses and connections are added, while others have indeed plateaued. Luckily, the growing cities are generally decent about realizing that the measured 16% annual growth (4 years of counts, not just two data points) won't continue forever.

[DJ]

Traffic growth is a result of many factors. These are way too numerous to mention and their interactions are far too complicated to understand, let alone forecast out 20 year into the future with any degree of accuracy.

A recent study of past MPO level forecasting efforts found (rough numbers) that about half had predicted growth accurately …but only half of those did so for the right reasons.

75% of professional forecasting efforts are misleading.

Who wants to try and convince others that they know what energy prices will do next year, let alone 5, 10, or 20 years out. Even if you could, what makes you think that you can translate that knowledge into any real understanding of trip making?
Who thinks that they can predict what technology will do to trip making? The iphone is responsible for the elimination of millions of trips. Where’s that covered in our forecasting? This gets us back to the old access vs. mobility debates that took place in the 90s. The jokers that used to debate that issue are leading places like the Seattle area MPO. What good have they ever done? When growth was rapid, they invented Washington’s Growth Management Act. Now that trip growth and the economy are sucking wind, those same people are promoting growth. The only thing these people are good at is figuring out which way the wind is blowing and setting their sails to it. They couldn’t predict the future any better than my cat.

Anyway, urbanization is a continuing trend. What effect will future urbanization have on trip making? Buildable land, farmable land, population growth/decline, employment, relative wealth. It goes on and on.

The things we can’t predict are tremendous and their impacts on future trip making is just as tremendous – and we just don’t have that crystal ball.

Past forecasting efforts merely copied past travel trends during times when things were relatively stable. No such luck now.

Better to be a traffic analyst or micro simulation expert than a travel demand forecaster. Why, better to simply use fake numbers than to be in the business of creating the fake numbers.