The purpose of this study was to provide an in-depth review of the Balanced Transportation Analyzer (BTA), a detailed spreadsheet model designed to forecast the traffic and revenue impacts of various proposed cordon tolling schemes for Manhattan. This review, performed at the request of Move NY and the Blue Marble Project, involved an evaluation of the following:
• Soundness. A sound model is one in which the internal calculations are performed accurately and for which all cell references are linked correctly. Stated another way, a sound model must have all of its internal “plumbing” connected properly. HNTB reviewed hundreds of key linkages within the model to ensure the model’s soundness.
• Robustness. A robust model is a model that is able to evaluate a wide range of scenarios without crashing or yielding irrational answers. HNTB performed numerous “stress tests” on the BTA to assess its ability to handle significant changes in a wide array of key inputs.
• Reasonableness. A reasonable model is a model whose high-level results are consistent with what one observes in the real world. To evaluate the model’s reasonableness, HNTB continually posed the following three questions during its analysis: First, do the model’s most fundamental outputs—namely, change in average vehicle speed and increase in net revenue—make sense under a given set of assumptions? Second, do the fundamental outputs change in a reasonable way when key assumptions are varied? And third, if the changes do not appear to be intuitive, is there a reasonable explanation based on a more detailed review of the model’s internal function?
The model’s soundness was evaluated through a detailed review of the model’s structure, its key formulas, and the relationships among the various tabs that comprise the model. Both soundness and robustness were tested through an evaluation of various scenarios and through sensitivity analysis. Finally, the model’s reasonableness was tested through a detailed look at the BTA’s projections of key results. This included a review of the model’s forecasts of revenue, of projected traffic volumes and speeds, of county and borough incidence of tolls, and of benefitcost calculations.
This assessment of the BTA proceeds in the following manner:
• Section 2 provides an overview of the proposed cordon tolling program.
• Section 3 provides a summary of the work that HNTB performed during the course of
• Section 4 summarizes HNTB’s general comments on the BTA.
• Section 5 documents HNTB’s validation of traffic data employed by the BTA.
• Section 6 contains an assessment of the BTA’s projections of key results.
• Section 7 provides an analysis of alternative scenarios as defined and requested by Move NY.
• Section 8 contains a sensitivity analysis of selected variables to evaluate their impact on key model results.
• Section 9 provides a brief comparison of the BTA’s Baseline forecasts with cordon tolling experiences in other cities.
• Section 10 provides a series of recommendations and areas for further study.
Cordon Tolling Overview
The BTA is designed to forecast the traffic and revenue impacts of various cordon tolling proposals for Midtown and Lower Manhattan. Though the specifics of the various proposals vary, they generally contain the following basic elements:
• The imposition of a cordon toll for all automobiles (not including taxis) driven into
Manhattan’s Central Business District (CBD). The CBD is defined as all points on Manhattan south of 60th Street.
• The imposition of a surcharge on all medallion taxis operating south of 96th St. These surcharges include a per-trip (or “drop”) charge, a surcharge on the per-mile rate, and a surcharge on the wait time rate.
• A reduction in tolls on six “Outer-Borough” bridges.
The BTA operates on the premise that the revenue to be generated by the cordon tolling program (unless designated otherwise) will be poured back into the transportation system. The investments may be differentiated between the roadway network and the transit network. The model makes further estimates regarding the extent to which the transit investments could in turn improve service and attract more users.
Figure 1 provides an overview of the layout of the proposed cordon tolling program. It should be noted that the BTA, while equipped to evaluate the program outlined in Figure 1, is actually capable of assessing a broad array of toll and fare scenarios. For example, it enables the user to charge taxis a cordon toll as opposed to a CBD surcharge. It allows the user to evaluate various toll discount levels on the Outer-Borough bridges and various levels of tolls on the Queensboro Bridge upper roadway. Moreover, the BTA is flexible and allows for cordon tolls to vary by time of day as well as by type of day (i.e. weekdays vs. weekends). It also allows for the cordon toll to be differentiated between automobiles vs. commercial vehicles. In short, the BTA has built-in versatility to allow it to do more than is documented in this memorandum.
Summary of Work to Date
HNTB’s approach to reviewing the BTA involved five basic steps.
• Step 1 was to learn the model’s structure. The BTA is an Excel-based model with 62 unique worksheets (or “tabs”), most of which are interconnected. HNTB took a significant amount of time to understand both the fundamental components of the model and the relationships among these components. This process was aided by a conference call at the beginning of the project, in which Mr. Charles Komanoff, the developer of the BTA, provided an overview of the model’s structure and purpose.
• Step 2 was to study the model’s mechanics. This involved looking at each tab and reviewing how it functioned.
• Step 3 was to identify the key assumptions and to examine how the model’s key results responded to changes in these key assumptions.
• Step 4 was to conduct an ongoing dialog with Mr. Komanoff in order to ask questions and provide feedback. This interaction was critical in order to (a) improve HNTB’s understanding of the model, and (b) provide Mr. Komanoff with a stream of information that could support the ongoing process of revising and updating the model. Over the course of the study, the dialog between HNTB and Mr. Komanoff resulted in six unique updates to the BTA. Each update was made by Mr. Komanoff and was reviewed by HNTB.
• Step 5 was to repeat Steps 2 through 4 as necessary to test and confirm the validity of any changes made to the BTA.
The modifications made to the BTA over the course of the study included the following:
• Some errant cross-references within the model were identified and corrected. Examples of such corrections included cell references pertaining to volumes on the upper roadway of the Queensboro Bridge (QBB) and cell references between the “MTA Crossings” and the “User Inputs” tab.
• The methodology for estimating the reduction of traffic on the QBB upper roadway was changed. Previously, the model used a fixed percentage; this was changed to a formuladriven percentage based on the proposed toll rate and the assumed toll elasticity of demand. This change provided more realistic results, since the actual traffic reduction would be dependent on the magnitude of the toll rate and the responsiveness of drivers to that toll rate.
• Some minor discrepancies concerning bus revenue calculations were fixed. Previously, some calculations were based on 2012 actual numbers while others were based on 2013 projected numbers. All calculations were subsequently changed to be based on 2013 projections for consistency purposes.
• Several points of annotation and documentation were updated and clarified. For example, the new BTA more clearly documents the rationale for estimating the share of George Washington Bridge traffic that ultimately enters the Central Business District (CBD), defined as the section of Manhattan south of 60th Street (encompassing Midtown and Lower Manhattan).
• Many parts of the taxi tab were overhauled. The new version accomplishes the following:
o It accounts for an apparent change in the existing fare structure, as posted on
the New York City Taxi & Limousine Commission’s website.
o It clarifies the geographical breakdown of Medallion taxi trips (e.g. the distinction
between intra-CBD trips, trips between the CBD and points north of 96th Street, and trips completely outside the CBD).
o It revises formulae used to convert speeds and travel times into taxi fares. Prior
to the revision, the formulas tended to underestimate the number of miles driven when the taxi was traveling at speeds greater than 6 mph.
o It provides a more consistent treatment of fare-trips versus person-trips. Prior
to the revision, the model didn’t properly account for the fact that the average
occupancy in an automobile was different than the average occupancy in a taxi.
The latest version of the BTA (dated October 8, 2013) incorporates all changes resulting from HNTB’s research and analysis. The net result of all the changes was a projected reduction in annual net revenue2 of about $30 million (or roughly 2%) compared to the July 15th version that was current when HNTB’s study began.