Curious about the gap between Sound Transit’s ridership models and reality, I had a discussion with ST planner Bob Harvey. He sent me this ST Transit Ridership Forecasting Interim Report (warning! 2 MB pdf) that explains the model ST uses for all its estimates across all modes, although the actual document is focused on North Link.
How it was computed: The Sound Transit model takes ridership data from existing Metro or Sound Transit routes in 2004. It takes PSRC data on population, demographics, employment, and highway congestion from 2004 and extrapolates them forward to 2010 or so based on PSRC assumptions. Then it applies the ridership improvements associated with Central Link’s level of service (speed, headway, etc.) to come up with the current ridership estimate.
What went wrong: Prior to Link opening, projected headways and travel time increased, cause ridership estimates to deteriorate somewhat. Once Link opened, the models overestimated regional employment. They also suffered from a lack of good ridership data for Rainier Valley-Seatac trips, a transit market that really wasn’t served before. And of course, no Metro route is completely replaced by Link.
Additional potential factors include the introduction of ORCA, transfer policies, and charging fares in the tunnel. Mr. Harvey, who didn’t work on the Central Link estimates, also notes that planners made “optimistic assumptions” about bus service realignment in Southeast Seattle, the segment that is “most definitely underperforming” compared to projections.
The estimates are actually worse than they first appear, given that they are geared to undercount ridership driven by sporting events and the airport.
The recent update replaces the insufficient bus data with Link data, but is still a victim to obsolete PSRC regional projections from 2004 data. Later this year, ST will release new projections based on a fully updated data set. It’s reasonable to assume that applying a standard growth rate to a baseline that’s well below trend will cause a general decrease in ridership estimates.