CityFixer

Does San Francisco's Smart Parking System Reduce Cruising for a Space?

By most measures, yes. By one big one, maybe not.

Image
San Francisco's SFpark pilot program raises questions about whether setting flexible meter rates is really the best way to free up parking spaces. (Flickr/Carlos Gomez)

Last week, the San Francisco Municipal Transportation Agency released yet more evidence that the city's SFpark system has been a great success. SFpark changes meter rates based on parking demand to maintain an average occupancy between 60 and 80 percent: When parking on a street is too full (or too empty), the hourly price goes up (or down) to free up (or fill up) spaces. The goal is to distribute parking more evenly and, more critically, to reduce time spent cruising for a space.

By SFMTA's latest measures, SFpark hits all of its key marks. Prepare for the chart parade: City blocks with SFpark met their target occupancies much more often than they did before the system went into effect, especially compared with control blocks that aren't part of the system:

Drivers found a spot quicker on SFpark blocks—about five minutes quicker than before the system went into effect:

Vehicle miles traveled also decreased around SFpark, another sign that cars spent less time searching for a space:

And local sales tax revenue went up, too, suggesting that people spent less time in their cars and more time in San Francisco businesses:

That's great news from one of the world's most progressive parking programs, especially one already serving as a model for other cities to follow. But it's not the whole story. Another recent analysis, conducted by transportation scholars Daniel Chatman of the University of California, Berkeley, and Michael Manville of Cornell University, found that on one critical measure of success—parking availability—SFpark may not be meeting its goal after all.

On the surface, occupancy rates and parking availability might seem like the same thing. Actually, they can be quite different. Just because an occupancy rate on a street is less than 100 percent, that doesn't necessarily mean that a space is always open when a driver cruises by.

Let's take, for example, a block with 10 parking meters observed over a three-hour period. If cars were parked there for 1,200 minutes of the 1,800 available minutes, that would indicate a 67 percent occupancy rate—which is within the SFpark target window. But those 1,200 minutes could be spread evenly over the three hours, or they could reflect all 10 spots being full for two straight hours. In the latter case, no parking spaces would be available during these two hours, and street cruising would no doubt rise.

To tease out that difference, Chatman and Manville oversaw street observations on nearly 50 city blocks at three separate times between May 2011 and May 2012. Some of the blocks had SFpark, some didn't. All told, the research team observed more than 13,400 cars. The study collected data on occupancy and availability, as well as several other key measures that should be influenced by meter pricing, including parking duration, vehicle turnover, and carpooling.

The first thing they found, in keeping with the latest SFMTA report, was that SFpark prices did control the occupancy rate of a block quite effectively. On average over the year of the study, when meter rates rose $1 per hour, occupancy fell 10 percent (and when rates fell $1 per hour, occupancy rose 7 percent).

On parking availability, however, the system didn't perform nearly as well. Initially, higher meter prices led to more time with spots open, but after a while that trend reversed. Averaged out over the full year of observation, Chatman and Manville found no statistically significant relationship between meter rates and available spaces. In other words, raising the cost of parking at meters on a block didn't necessarily ensure a space was always open. Their results on average duration and carpooling were similarly insignificant.

They conclude in this month's Research in Transportation Economics:

Over SFpark’s first year, price increases on the blocks we examined were associated with reductions in average block occupancy. In this respect the program worked as intended. Yet these moves toward lower average occupancy did not appear to yield SFpark's desired policy outcome. The price increases that improved average occupancy did not consistently improve parking availability.

The study raises a good question about whether occupancy or availability is a better measure of smart parking success. From a driver's perspective, availability is almost certainly more important. Far better to know that a spot will always be open, whatever the price, than to leave the house wondering.

At the same time, the research has its limitations. To name two: It only captures a window at the very start of the SFpark program, and it measured blocks by eye rather than by sensors. Its chief lesson is also limited. Chatman and Manville conclude that meter rates might not always be high enough to discourage parking on certain blocks, despite reaching $6 an hour at times. But for political reasons—namely, realizing how much the public hates to pay for parking—San Francisco caps meter prices even on the most congested streets.

All things considered, it might be best to think of SFpark as an imperfect system, yet also the best system an American city can realistically achieve. Until we find a publicly palatable way to let the price of parking truly meet demand, that still seems like progress.

 

About the Author