Monday, October 7, 2013

Is 30% of traffic actually searching for parking?

Is 30% of traffic actually searching for parking?
San Francisco Examiner headline asked that question recently.  (Hat tip Parking Today blog)

Should you care if you are not in San Francisco? Yes! That motorists searching for parking can add greatly to congestion is a very widely used argument for improved on-street parking management, especially performance pricing.

To be precise, the Examiner article questions "the statistic that 30 percent of all congestion in the City is caused by frustrated drivers circling the block for that elusive parking space."

Err. Thirty percent of traffic is not the same as thirty percent of congestion.

But, either way, does the SF Examiner article have a good point? Is this a gotcha moment for that 30% thing?

Wait a minute! Let's take a step back first.

My first reaction upon skimming the article was to say the article is holding up a straw man to easily knock him down.

Is anyone really claiming 30% of urban congestion (or traffic) is cruising for parking? 

Not me. Not Donald Shoup.

How much of this traffic (in Seoul in this case) is actually searching for parking?

Donald Shoup is the original source of this "30% of traffic" talk. Did he claim that 30 percent of congestion is due to parking search traffic?  Not exactly. Here he is in 2011:
Sixteen studies conducted between 1927 and 2001 found that, on average, 30 percent of the cars in congested downtown traffic were cruising for parking.  [my emphasis]

I can see how this morphed into a more general claim. But notice that he qualifies the "30% of traffic" with "of congested downtown traffic".

Why did he do that? Because the studies he is citing have that specific focus.

They can't and don't claim to measure the share of ALL traffic or all congestion that arises from parking search. As Shoup says in Chapter 11 of his magnum opus on parking policy reform, The High Cost of Free Parking (p.291):
The studies are selective because researchers study cruising where they expect to find it - on streets where curb parking is under-priced and overcrowded. 

From the sixteen studies, there were six measurements of the percentage of traffic that is seeking parking. The results ranged from 8 to 74%, for that average of 30%.   [By the way, thirteen of the studies generated another number that highlights the cumulative costs of cruising for parking, the average time spent by each motorist on the parking search (an average of 8.1 minutes in case you were wondering). Some studies yielded both numbers.]

All six estimates were from major business districts in urban cores:
  • Detroit, USA 1927 A:  DOWNTOWN location between 2 and 6 pm (presumably on a weekday). Found 22% of traffic was 'cruising for parking').
  • Detroit 1927 B:  DOWNTOWN location between 2 and 6 pm. Found 34% as cruising percentage.
  • New Haven, USA 1960: CBDs of New Haven and Waterbury, Connecticut, with three study times (quiet summer period, average November period, and pre-Christman rush period). Calculated that at least 17% of CBD traffic was parking search traffic. 
  • Freiburg, Germany 1977: CENTRAL Freiburg, with 74% of 800 tracked cars estimated to be searching for parking. 
  • Cambridge, Mass., USA 1985: Harvard Square BUSINESS DISTRICT, 10 am to 3:30pm (hence covering the lunch peak), estimating that 30% of cars were cruising fro parking. 
  • New York, USA 1993: MID-TOWN, West Side, 8-10am and 11am-2pm, finding 8% of traffic was searching for parking. 

The point of all of this is NOT the 30% average! 

The point is to highlight that, parking search traffic is potentially a very big deal in CERTAIN IMPORTANT CIRCUMSTANCES.  

What circumstances? When on-street parking is badly managed and saturated, so that we get lots of search traffic (and waiting and double parking too).

That is how I have always understood the talk about parking search traffic being a significant percentage of traffic.

Shoup himself was cautious in the book about making too much of the specific numbers from the studies he cites (p.291):
But these studies dating back to 1927 are mainly of historical interest. The data were probably not very accurate when they were collected, and the results depended on the time of day, the specific place, and the season when the observations were made. 

Nevertheless, these studies do show that parking search is potentially an enormous source of congestion and unnecessary traffic in busy business districts when on-street parking is mismanaged so that it is often totally full.

As Shoup goes on to say (p.291):
But because curb parking is underpriced and overcrowded in the busiest parts of most of the world's big cities, the sun never sets on cruising. 

More recent investigations in New York City commissioned by Transportation Alternatives confirmed that parking search CAN be a huge share of traffic in busy areas. 

In 2006, on Prince Street in Soho (pdf) on Tuesday, Friday and Saturday afternoons/evenings, they found an average of 28% of intercepted motorists (at traffic lights) were looking for parking in the area. On Saturdays the share was 41%!

In 2007, in Park Slope, Brooklyn (pdf), they looked at four periods (weekday peak, two weekday moderate times, and weekend peak) and found overall 45% of traffic was cruising for parking.

So, like the studies cited by Shoup, both of these NYC studies focused on problem places at problem times.

We can't extrapolate to a general claim about shares of traffic. But again they highlight how extreme parking search traffic CAN BE.

But let's get back to the San Francisco Examiner. Does the article have a good point or not?

Sadly, maybe it does. I thought it was bashing a straw man, but no.

There really has been some fast and loose talk about the 30 percent thing.

It was challenging SFPark claims in particular. So I looked for some and found this:
“Circling for parking accounts for approximately 30 percent of San Francisco’s congestion,” said Tom Nolan, Chairman of the SFMTA Board of Directors. 

SFPark is San Francisco's Performance Pricing trial. It has not been as controversial as many expected. Nevertheless, questioning a key SFPark claim makes good news copy.

SFPark said the claim was based on Donald Shoup's book. There hasn't been a specific study of this in San Francisco.

So I have to agree that it is a misleading claim.

What is true is that, based on the Shoup-cited studies, parking search is very likely to account for a lot of San Francisco congestion.

But no-one currently knows how much of San Francisco's congestion (or traffic) arises from cruising for parking.

Sorry SFPark. 

And I gather SFPark is not alone. Various others may have been getting a bit too attached to that specific 30% statistic, taking it as gospel, and using it in sloppy ways.

By the way, it is easy to see how this happens. Quoted statistics, especially startling ones like this, tend to harden with time and distance from their origins.

Donald Shoup's statement that parking search traffic was found in certain studies "to be on average 30% of congested downtown traffic" or my statement in a recent presentation, that parking search traffic is "often 30% or more" could easily become misleading if quoted beyond their original contexts discussing situations with badly mismanaged on-street parking.

And with repeated mentions of that 30% figure, it seems to have morphed into a misleadingly precise and general claim about the percentage of overall traffic or congestion being due to parking search. And then, even more misleadingly, it was then claimed to apply to a specific place again, like San Francisco. Oops.

What is the bottom line here?

It would be annoying if debunking the sloppy use of this 30% average undermines the important point that:
cruising for parking CAN cause a huge mess in busy areas, at busy times, when parking is mismanaged. 

This point is still valid and not undermined by the discussion above. Parking saturation certainly causes traffic congestion (and other problems).

How much exactly? Well, it depends. Do your own little study in the context you are interested in.

But potentially it is a LOT.  Much more than was widely realized before Donald Shoup came along.

Pity about that 30 percent figure, though.