Skip to main content

As the former Deputy Director of Parking at the San Francisco Municipal Transportation Agency (SFMTA), I have been asked many times about the SFpark program. Each time the questions are the same:

  1. What is the purpose?
  2. How does it work?
  3. What are the results?

Well, the first two questions are painless to answer. SFpark is a large-scale smart parking initiative, which was developed to improve the management of on- and off-street parking in the City of San Francisco. The SFpark program includes over 12,000 off-street parking spaces (75% of the SFMTA managed off-street spaces) and 7,000 metered spaces (25% of the city’s total metered spaces). SFpark’s primary goal was to make it easy, or at least easier, to find a parking space (both on- and off-street). This was to be accomplished in three steps: 1) establishing availability goals; 2) installing electronic meters and in-street sensors that communicate data regularly to a data management system; and 3) managing demand for existing parking through demand-responsive pricing.

Through demand-responsive pricing, SFpark periodically adjusts rates to find the lowest rate possible to achieve the established availability goals (60-80%).  Basically, if SFpark establishes the ‘right’ price there should be a space available on every block face or in an SFMTA managed garage. The presumption was, if a space is always available, circling the block and/or double-parking would be eliminated and muni transit speed and reliability would increase while greenhouse gases decrease.  Additionally, an added benefit of reduced cruising would be safer streets for bikers and pedestrians.

Expanding a little more on the demand-responsive pricing model, here is the very basic structure:

  • Hourly rates for both on- and off-street parking can fluctuate by time of day and weekday or weekend;
  • Meters are assigned three time bands: (meter opening time to noon; noon to 3pm; and 3pm to close);
  • Meter rates can range from $0.25 to $6.00 per hour;
  • Meter rates can be adjusted every 30 days;
  • Garages are assigned five time bands: (Midnight to 9am; 9am to Noon; Noon to 3pm; 3pm to 6pm; 6pm to Midnight);
  • Garage rates can range from $1.00 to $10.00 per hour;
  • Garage rates can be adjusted quarterly; and,
  • Garage rates have a floor: during meter operation hours the minimum hourly charge at a particular SFMTA garage will be set as $1.00, or $0.50 less than the lowest meter rate within two blocks of the garage, whichever is lower. This directly ties SFMTA garage rates to meter rates, which is intended to incentivize drivers to go directly to the garages

Based on this model, if demand on a specific block face is low during the 9am–Noon time band the on-street rate may be as low as $0.25 per hour. However, if demand is high during the Noon-3pm time band the rate may be as high as $6.00 per hour. These rate adjustments are gradually implemented to reach the “equilibrium” rate (where demand does not exceed the availability goal).

What Are The Results?

Now to address the tough question: What are the results? In my opinion, the most challenging factor in the evaluation the program’s outcome (and consequently, the potential duplication of the program in other cities) is the program’s failure to follow one of the basic rules of the scientific method: introduction of one factor/variable at a time. During the implementation phase of the program, several variables were changed and/or implemented at or near the same time:

  • Extension or elimination of time limits;
  • Installation of electronic meters that accepted credit cards;
  • Variable rates by time bands and day of week;
  • Acceptance of pay-by-phone;
  • Providing parking data to users (parking apps); and,
  • More effective enforcement based on sensor and payment data.

The implementation of the above mentioned variables at or near the same time makes it more difficult, if not impossible, to measure the independent effect of each variable on the program’s results. Additionally, variables that are out of SFpark’s control, and were not measured, may have attributed to the program’s results. These variables include:

  • Changes in land use (e.g., commercial to residential conversion); and,
  • Economic growth (numerous infill projects and population growth).

Based on a study released in March 2014 (Assessing the Impacts of San Francisco’s Parking Pricing Experiment), the study’s authors evaluated the first two years’ worth of data and concluded that the SFpark pilot area had lower levels of cruising for a parking space compared to the control area. The study indicates that the program as a whole does show causality on parking availability and reduction in circling. According to the study, approximately 88% of the drivers in the SFpark pilot area located a space within one block of their initial block compared to 79% in the control area. However, when the study focused on a subset of highly problematic streets that likely added disproportionately to cruising, the data indicated these blocks experienced a reduction in cruising of approximately 50%. But, the extent of causality applicable to each variable is difficult to explicitly quantify. Thus, if another municipality implements a SFpark-type program, but it does not utilize all of the same variables, the program’s results may not replicate the SFpark results.

Now the question is: What will the data from the third, and final, year of the pilot reveal? SFMTA’s staff is working on the final analysis of the pilot and will provide their findings to the USDOT by July 2014. Hopefully, the program will provide more insight into the specific effects of demand-responsive pricing on parking availability, cruising, traffic congestion, greenhouse gas reduction, and transit reliability. Until then the jury is still out.