New ferry report: Economists analyze data, suggest change

A new study about ferry service from Fauntleroy shows that the loading procedures instituted last summer moved slightly fewer cars than in the same time the previous year, that recurrent delays hurt the ferry system financially and that implementing a new payment system and scheduling model would be beneficial.

The study, available online and featured in The Seattle Times last Wednesday, was conducted by island resident Theo Eicher, who holds a doctorate in economics and is a professor at the University of Washington, and Jeremy Cooper, a data analytics consultant. The 44-page study, based on Washington State Ferries (WSF) data, also recommends that the state Legislature add capacity utilization — the actual number of vehicles loaded on each ferry — to the measures it evaluates at WSF. The study comes after a difficult summer of ferry travel for many islanders, who faced long lines and partially filled boats, despite the work of a task force dedicated to improving the service on the troublesome triangle route.

In a conversation last week, Eicher and Cooper stressed that they are analysts, not ferry activists.

“There is a lot going on between Vashon and Washington State Ferries, and really all we are is interested in the data,” Eicher said. “I happened to know the data was out there. I requested it, and we analyzed it, and we counted cars. In a sense it is very boring.”

Cooper, who has ridden the ferry to Vashon only a handful of times, joined Eicher in the project because he believed it would be an interesting exercise in data analytics — the kind of opportunity that he gets excited about, he said.

Both men noted the dichotomy that was evident at WSF’s public meeting on Vashon in September. There, ferry officials spoke positively about the effects of the new loading procedures, while many islanders expressed high levels of frustration.

Cooper said they thought at the time, “What does the data have to say about what is going on?”

Ultimately, Eicher received data from 2007 to 2017, which included some 18 million transactions from the electronic ticket scanners at Fauntleroy. The hard part in the process, the men said, was cleaning all the data— removing errant double reads of tickets, or ticket purchases to other destinations, so that an apples-to-apples comparison was possible.

The most striking finding, Eicher said, was the number of vehicles on ferries — capacity utilization.

“At first I thought something was wrong with the data, because these numbers seemed really low,” Eicher said. “But then you look at the Facebook page (with photos of partially filled ferries), and they corroborate the low capacity that we found.”

In fact, they found that most ferries did not load close to capacity at rush hour, with some only about 30 percent full last August. While the state Legislature evaluates WSF, in part, on on-time performance, it does not evaluate WSF on how full the boats are, even at rush hour. Currently, WSF only has a model for how many vehicles travel on each boat, but does not keep an exact tally. Eicher and Cooper recommend that this change.

“This is just a tip … to the Joint Transportation Committee that we could make things a little more efficient if they asked ferries to do that,” Eicher said.

The men analyzed further and found decreased overall demand from the early 2000s, but longer delays.

“It seems like things are moving in the opposite direction of what one would think,” Cooper stated.

WSF recently released its ridership data for 2017, indicating it was the busiest year since 2002, including an 8 percent increase in Southworth traffic. However, ridership between Vashon and Fauntleroy went down, with WSF reporting a 1.3 percent decrease in vehicle traffic to Vashon and a .5 increase in total ridership.

In the data that Cooper and Eicher cited — from 2003 through 2016 — peak travel from Fauntleroy occurred in 2005 with about 860,000 vehicles. In 2016, approximately 45,000 fewer vehicles traveled on that route.

Moreover, they found that delays have been growing progressively on the route since 2011. In that year, the cumulative delay of all ferries leaving during rush hour never exceeded two hours per day. By 2016, the average cumulative delays had grown to between one and two hours per day with several cumulative delays surpassing two hours.

The study authors conclude, “Capacity underutilization and delayed sailings produced substantially longer ferry lines and wait times.”

As for the new loading procedures instituted at Fauntleroy, Eicher and Cooper state that, based on the results of the week-long May pilot, WSF should not have implemented them. In the pilot, WSF’s goal was to move more cars during rush hour compared to the same week the year before, but they failed and moved fewer: 744 in 2016 and 739 in 2017.

Like WSF, Cooper and Eicher analyzed the first month of implementation, but the men could not replicate WSF’s findings. WSF counted ticket redemptions between 3 and 6 p.m. Monday through Friday from the third week of June to the third week of July. WSF reported that the number improved just slightly, from 635 in 2016 to 638.5 in 2017. The study authors, however, found different totals — and a slight decrease — during the summer trial rush hour windows: from 582.5 in 2016 to 580.8 in 2017.

They also state that WSF should expand what it considers peak time in the summer. The 1:40 p.m. sailing regularly fills to capacity, as does the 6:35 p.m. ferry. Therefore, summer rush hour should be defined as 1 to 7 p.m., not simply 3 to 6 p.m.

They conclude: “WSF justifications and evaluations for new loading procedures were based on incorrect data analysis.”

They also recommend that WSF seek outside data analysis experts for projects such as this.

“Future data analysis used to motivate operational changes should be conducted by independent consultants engaged by Washington State. The outcome of this analysis should be reported to the public in a transparent, replicable fashion,” they state.

As economists, Eicher and Cooper say that the toll of ferry delays extends beyond people arriving home late and carries true financial costs. They found on the triangle route that in an almost 1 to 1 ratio, as ferry capacity utilization declines, the vehicle demand for ferry service falls. To a lesser degree, there is a similar correlation with passengers. The result, Eicher and Cooper state, is that the state loses out on an estimated $120,000 in revenue in each of the summer months.

To improve service, the men suggest that WSF implement GoodtoGo!, which WSF officials have not implemented in part because that system would only capture drivers, not passengers, and fares would increase to make up for the difference. Eicher and Cooper, however, state that efficiency gains would outweigh costs and that savings in wages could substantially offset the loss of passenger revenue.

At WSF, a recently compiled list of questions and answers addresses GoodtoGo! It states that there are two ways that system could be used in the future: as a payment option at the tollbooth keeping WSF’s existing fare structure or as the exclusive method of payment, which would require that WSF’s fare structure be simplified. The same document notes that WSF’s long-range plan, due to the Legislature in September, will include recommendations for new technology options.

Eicher and Cooper recommend a new model for the triangle route schedule that they believe would eliminate terminal bottlenecks. They suggest ferries leave Fauntleroy in 30-minute intervals, with each vessel serving dual destinations and carrying 70 percent and 30 percent Vashon and Southworth cars respectively. They believe the Fauntleroy dock would clear with each sailing if a traffic light and a dedicated tollbooth for preticketed passengers were in place to facilitate loading and unloading. They add that adjustments could be made for crew changes and other needs and still adhere to the fundamentals of the schedule, which would have more runs to both Vashon and Southworth during rush hour.

“A fresh approach, rather than modifications or simple service cuts to the existing schedule, may provide improved efficiency,” the report states. “It is unlikely that altering an already suboptimal schedule through minor service changes or further reductions in service will result in greater WSF efficiency.”

WSF has frequently cited the size of the Fauntleroy dock as a problem with loading boats efficiently, but Eicher and Cooper indicate that while a new dock is slated for 2025, it is not certain to be bigger. If it is not enlarged, they expect problems to persist for the triangle route.

“Under current conditions, it seems unrealistic to expect less congestion with larger vessels and further population growth in Southworth if the dock size is held constant,” they state.

At WSF, spokeswoman Hadley Rodero said the Fauntleroy project is a seismic replacement of the trestle and transfer span to replace the deteriorating timber piling. Dock size has not been determined.

“We have not begun preliminary engineering, pre-design or the environmental review process that will determine the full scope and design of the project,” she said in a recent email.

The men note that they shared their they findings before their public release with WSF leadership, the Joint Transportation Commission, state senators and representatives and Triangle Taskforce members whose emails they could find online.

WSF’s Rodero said ferry officials read the study, as they do with all public comments and also shared it with the task force.

“The study’s recommendations are also things we’ve heard in other public comments and during our rounds of public outreach,” she said.

While Eicher and Cooper’s say they have made recommendations aimed at improving ferry operations, they reiterate their work is about numbers, not the people of WSF.

“We have nothing but the utmost respect for the ferry workers who work their hearts out come rain or shine to allow us to travel back and forth to the island 24/7,” Eicher said.