A Note on the Importance of Frequency in Regional Transit

Apologies for the long periods between posts. I’ve been caught up with school, work, and the Jewish holidays, so time for blogging has been infrequent. That being said, here’s a short post on something that caught my eye as I was doing research for a paper.

Anyone interested in planning, economic, or transportation issues should be aware of a series of papers authored by Richard Voith, a former economic advisor to the Philadelphia Fed, Wharton School professor, and member of the SEPTA board. His writing covers topics like capitalization of transit access, urban-suburban real estate dynamics, and transit efficiencies. The last topic is the subject of a 1994 paper titled “Public transit: Realizing its potential,” published in the Philadelphia Fed Business Review. The paper is a general argument, but it also includes some interesting data on Philly transit systems circa 1994, which I thought it would be interesting to present here.

In a kind of appendix to the paper, Voith presents two charts (modified here for readability). SEPTA refers to the Regional Rail (commuter) division ONLY, not the urban rail subway or trolley lines or bus operations. Remember, this is 1994 data; I’m sure the numbers have changed since then:

Fare $3.25 $1.60
Trains per Line 7 33
Annual Ridership per Mile of Railroad 75,312 785,261


Cost per Rider $7.32 $2.26
Subsidy per Rider $4.37 $0.89
Annual subsidy per mile of railroad $329, 698 $699,225

As Sunny Zheng pointed out on Twitter, these numbers don’t capture absolutely everything; SEPTA has (and presumably had) pass products that would have lowered the average fare charged to most riders. I don’t know whether the subsidy numbers account for those lowered fares; the SEPTA fares quoted are peak, Zone-2 (a distance that roughly corresponds to PATCO’s length).  Still, the numbers say a lot. As Voith wrote:

As shown in the figure, PATCO’s fare is less than half of SEPTA’s. PATCO runs almost five times as many rush-hour trains on its single 14-mile line as SEPTA runs on its average commuter line. PATCO also runs much more frequent off-peak service. The net effect of the lower-price, higher-quality service is that PATCO carries over 10 times more people per mile of railroad than SEPTA does.[b] Thus, for very similar suburban markets and the same destination, ridership levels are dramatically different. The level of current SEPTA ridership doesn’t necessarily reflect transit’s potential.

I see three possible contributing factors to the ridership differences. One, and likely the most powerful factor, is frequency. As Jarrett Walker has repeatedly stressed, frequency is the single best way to attract riders to a transit service. Another is the low fare, which was well below that of SEPTA. A third is that transit demand from South Jersey to Philly is relatively inelastic, because driving a car into the city requires paying a toll on one of only a few congested bridges.

It’s possible that the low fare helped to attract riders to PATCO, and the pricing of car access to Philadelphia closer to true cost certainly helped the situation. But, especially accounting for the passes, SEPTA fares in Zone 2 were’t really that much higher than those of PATCO. I think it’s pretty clear that this data validates everything Jarrett Walker has ever said about frequency. In fact, ridership on PATCO was so high that had the agency been willing to raise fares $.89–$1.43 today, arguably a raise that a relatively wealthy suburban clientele could have afforded–it could have eliminated subsidies altogether.

There were downsides to the frequency with which PATCO was blessed, of course. The subsidies per mile of railroad were more than double those of SEPTA’s regional rail. But one thing is very clear:

When you invest in frequent transit, you get what you pay for.

Ridership follows investment in operations. We can argue about the necessity of subsidies; I’m guessing that the value to Philadelphia of keeping all of the cars used by PATCO commuters in 1994 out of Center City was far greater than the $.89/passenger subsidy. This is yet another argument for the rapid-transitization of the SEPTA Regional Rail network (and others!), a case that has been made repeatedly by transit advocates and some professionals. Frequency is expensive in absolute and per-mile-of-rail terms, and probably unworkable without reworking of the labor agreements that American commuter rail systems currently operate under. But in terms of “buying” passengers, frequency is cheap–almost certainly cheaper than anything else transit agencies can do.

I don’t want to shortchange the place that the bridge tolls hold in PATCO’s success; pairing realistic car pricing with high-quality transit is the future of American urban transportation policy. But driving into and parking in Center City is expensive regardless of where you’re coming from, and the success of PATCO’s high-frequency service is apparent. There are many, many places in the US where more-frequent regional rail service would be feasible; looking back 20 years, we can see the economic justification for what has become a popular idea among transit advocates.