Monday, April 26, 2021

Beefing with the Population Research Center: Forecasting at the MPO

The Policy Committee for our local Metropolitan Planning Organization zooms on Tuesday the 27th, and they will be asked to sign on to a letter containing a critique of the latest population forecasts from the Population Research Center at PSU.

This might seem like a minor technical matter, but the memo paints it as a big deal, and from here the problems are emblematic of and ingredient in the way our transportation planning is pretty messed up. 

The agenda item: forecasts look too low

In more detailed memo inside the meeting packet, MPO staff say that the new forecasts issued this year show, relative to the last forecast from 2017, a "sharp decrease in growth projections" and this delta, this change, they "find...very alarming."

 "sharp decrease...alarming"

Just in neutral terms, sure, a change in forecast is an interesting matter and analysts would be interested in learning more about why there is this change. Trying to understand the change better is, in and of itself, not weird or objectionable.

But why is the lack of growth so alarming? Is there some reason we have to assume greater growth? Strong Towns might say that the "growth Ponzi scheme" depends on outrageous projections of growth, and default planning assumptions for growth might be worth a closer look. Does debt service on bonds and the sales of bonds require these assumptions? Formulas for Federal funding? (And politics and redistricting? What all is at risk in this? If we are adjusting forecasts every 4 years, what gets out of whack over the four year interval?)

There might be an interesting subtext here.

And there is more. A section in the memos talks about the "importance of having reliable forecasts for long-range transportation planning."

Ostensible interest in "reliable forecasts"

It's well and good for the MPO to say they must insist on "reliable forecasts," but they have never publicly assessed their own forecasts' reliability.

Year 2000 projected traffic counts
for Front Street Bypass
(actuals from City)

Here are circa 1980 forecasts for the Front Street by-pass and downtown streets, some of which were wildly off, and the MPO or City have never discussed the "reliability" of that forecasting. The MPO might have an interest in reliability from PSU, but not from the MPO itself!

There is a difference in scale, sure, the variance on individual streets is going to be wider than the variance on a whole metropolitan population.

But what is lacking is the idea of variance and range.

The whole problem here is that we are asking a forecast for one number instead of treating a forecast as a range of probable outcomes.

A bell-ish curve from actual traffic forecasts

If we treated the forecasting as such a range, it would be easier to absorb and discuss a disagreement. The MPO could issue a range of forecasts with the PSU numbers and one or more forecasts with adjusted numbers, and explain why they made the adjustment. But because we suppress any discussion of uncertainty, and apparently wish to maintain the illusion of fine accuracy for our forecasting, committing to one single number rather than a range of probable outcomes, we have this beef.

Instead of beefing with PSU, the MPO should redirect their analytical efforts inwards and spend time developing a feedback assessment loop of their own forecasting and make it public.

(See previous notes with criticism of our prevailing approach to traffic forecasting here.)

Meeting and zoomy details

The committee zooms at noon on Tuesday the 27th.

No comments: