Research Request - Transit Agency Peer Groups

[KatrinaMKaiser]

Research Question

Single sentence description: How can analysts define standard agency peer groups based on their similar characteristics?

Detailed description: Transit agencies' service and performance should be compared to their peers. Who are agencies' peers? Currently analyses generally use geography (CT district, RTPA) or urban/rural (based on 5311 receipts or NTD definitions).

TTTF March 11 TDA Proposal from Juan Matute

Remove farebox recovery and CPI escalation clauses from TDA

Replace with

Allocation of STA revenues based on passenger boardings to operators meet at least one of the following conditions (assessed during triennial audit):

Meet ridership growth targets versus peer group
Targets to be negotiated - may be 25%, median, or higher
Peer group: groups of a minimum of 5 operators to be established based on customer base characteristics (socioeconomic factors), type of service (eg bus vs commuter rail), and urban form/land use pattern of service area.

(or)

Have implemented state-identified measures to increase transit ridership:
Realtime info quality (GTFS-RT)
Open loop payments
Meet interagency service connection standards (to be developed based on user experience/user-based design perspective)
Maintain high-quality transit services identified in RTP/SCS or receive waiver from MPO (only for operators within MPO area)
Implement other initial recommendations for transit agencies identified by SB 125 Transit Transformation Task Force Report
[note: list is updated over time by State]

(or)

Exceed performance targets from customer surveys or third-party assessments (e.g.: mystery shops)
Target to be negotiated, should be 75th percentile or higher for peer group (or larger)

If an operator fails to meet these above thresholds, then the RTPA can take actions at the discretion of the RTPA:

RTPA assumes service planning
RTPA assumes fare policy coordination
RTPA prescribes corrective actions to be taken by operator
RTPA absorbs or merges operator

Optional performance-based funding incentive

RTPAs receives funding bonus when:
Service speed ratios (80% to 20% speed for corridor) improve or meet state targets
Rationale: creates RTPA/CTC/CMA motivation for transit priority investments.

[note: list is updated over time by State]

How will this research be used?

This research will be used to create standard groupings for presenting current and future transit analyses for policy audiences. More actions pending the actions of the Transit Transformation Task Force and the Legislature.

Stakeholders & End-Users

@KatrinaMKaiser is currently asking the question. The audience is policymakers setting guidelines around transit performance who want to compare agencies to similar agencies.

Metrics

• Assess the example characteristics proposed (these may require sub-issues and sub-research tasks)
  • customer base demographics
  • type of service
  • service area land use
  • service pattern
• Explore using K-Means Clustering algorithm or other similar statistical methods to group agencies based on their characteristics

Data sources

• Cal-ITP data sources: GTFS Schedule, NTD

• External data sources: ACS (for demographics)

• Remaining data source questions: Research needed: are there standard typologies for "service pattern"? Refer to Jarrett Walker and similar.

Deliverables

Initial exploratory notebook(s)
List of Clusters and their member agencies

Timeline of deliverables

Estimated start date TBD

[KatrinaMKaiser]

Could support #1348 - classifying small/big agencies in MTC area

[KatrinaMKaiser]

Had discussion with Brian Taylor, Jacob Wasserman at UCLA regarding proposed bus network redesign evaluation research. They suggested using propensity-score matching to identify agencies similar to agencies that had performed frequency-oriented bus network redesigns, to statistically estimate the benefit of the redesign while controlling for other local factors.

[KatrinaMKaiser]

From Juan 5/29:

In a more advanced version, operators would be clustered into groups of 10 or more based on mode of service, vehicles available, population density of service territory, job density of service territory, and, perhaps, service area overlap with other transit operators (a GTFS spatial analysis exercise). An agency scoring in the bottom 1 or 2 of the cluster would get some remedial help in their triennial audit. Or face consolidation (FWIW, I like the BC Transit model for consolidation starting in 1979). Several large transit operators, especially those operating rail, wouldn't be candidates for consolidation and wouldn't fit this clustering method and would instead rely on a triennial audit, where I would expect trends over time for GTFS-RT quality, customer experience metrics (Transit App surveys or mystery shops) and several of these metrics to be considered holistically.

And perhaps agency costs would be adjusted for regional consumer price index maintained by California Department of Industrial Relations. Either that or they'd just be clustered with regional peers.