"It is not sufficient to observe that each assumption by itself, is reasonable. Too often, many of these assumptions are 'reasonable, though maybe on the optimistic side.' It is therefore necessary to evaluate the set of assumptions and show that, as a whole, it represents a realistic portrayal of future conditions and avoids the risks associated with compound optimism." FEIS Transit Ridership Forecasting Technical Report, Oct. 1999, page 4
The above quote from Sound Transit’s Forecasting Report certainly raises a good point. A review of Sound Transit materials raises the question of whether ST has paid heed to this warning. The following list identifies particular areas of concern. In an effort to focus on questions that can be verified without too much difficulty, the points listed deal primarily with inputs to the Sound Transit (ST) model, rather than how the model itself is structured.
1) Transit Fares. The ST Forecasting Technical Report includes a table (4.1e) which lists the 1992 and future year transit fares assumed in ridership forecasting. ST has assumed that fares will increase only at the rate of inflation. A comparison with current transit agency fares shows that in fact fares have increased much faster than inflation, in some cases doubling or even tripling from the levels shown in the table. Much of this can be attributed to the impacts of I-695, but whatever the reason, the fare assumptions used in the forecasts are no longer valid. Since ST is assuming that 40% of light rail passengers will transfer from other public transit, this will have an impact on Link ridership. The fare increases of the last few years should not be viewed as one-time occurrences. King County Metro’s recently adopted budget assumes an additional fare increase for the coming year.
2) Rail Speed. The FEIS for the Link project indicates the light rail line is expected to average between 24 and 28 mph. This may not seem like a particularly optimistic assumption, but a look at other systems reveals that 24-28 mph is above the average achieved by light rail lines in the U.S. It appears that ST has arrived at this figure by assuming average station dwell times of only 20 seconds at most of the stops, and 30 seconds at the five most heavily used stations. A check of the TRB’s Transit Capacity and Quality of Service Manual does not list any systems that have dwell times of such brevity. As a comparison, the Vancouver Skytrain, which has comparable ridership to that forecast for Link, has dwell times that average just over 30 seconds, and that is achieved with smaller cars than ST proposes to use. If ST were to use more realistic dwell times the light rail speed would be slower and ridership would be correspondingly decreased. ST has also made assumptions about the effectiveness and reliability of signal priority systems intended to expedite rail operations on surface streets. The FEIS describes two possible methods of controlling traffic signals. The performance of these priority systems in actual operation is not yet known.
3) Rail Service Frequency. The FEIS indicates that 2010 ridership forecasts were based on four minute headways. Elsewhere ST has indicated that five minute headways were anticipated in the first year of service. In the final draft Fleet Management Plan (July 2000) a range of possible peak hour service frequencies are given including 5.5 minute headways. The ST New Starts submittal (September 1999) indicates that weekday peak hour headways will be 6 minutes for the MOS. From this conflicting information it is not possible to determine what year 2010 light rail headways will be. The ridership forecasts should be carefully checked to verify the assumed headways are consistent with ST’s most current operating plan (and budget). Less frequent service will result in longer wait times, slower transfers between modes, and lower ridership.
4) Income and Auto Ownership Costs. The ST forecasts are based on the PSRC model which includes an estimation of income by FAZ. The ST Forecasting Report states, "The increasing trend in income and car ownership in the past four decades has adversely affected the per capita transit ridership during this period. Furthermore, as real wealth increases, the propensity to use the less expensive mode decreases." It is not apparent from the document how recently this input to the model has been updated. It is obvious that since 1990 there has been an enormous increase in incomes in the Puget Sound area. It is reasonable to expect this will result in some shift in mode choice. If ST’s forecasts are based on data that is nearly ten years old, it may no longer accurately represent consumer mode choice.
The ST Forecasting Report also states that the PSRC mode choice model includes auto costs. These costs were reduced as a result of I-695’s repeal of the MVET. This decrease in auto costs, which is ongoing, increases the attractiveness of SOV (and car-pool) use relative to light rail and other transit modes. Therefore, if ST’s model has not been updated to reflect the I-695 changes it will tend to overstate light rail ridership.
5) Capacity and Passenger Comfort. The ST ridership forecast does not take qualitative differences into consideration. Though there are many qualitative factors, one of the most important is whether riders are seated or forced to stand. Currently, more than 95% of bus riders are seated. In contrast, ST forecasts over 30% of peak hour light rail passengers will be forced to stand. The effect this loss of comfort will have on ridership is difficult to forecast, but experience suggests that discretionary riders are highly sensitive to changes in service quality. The failure to adjust for the large increase in standees probably results in optimistic ridership forecasts.
6) Bus Service Levels. Transit agency budgets were decreased by over $200 million per year as a result of I-695. This loss of revenue has resulted in service cutbacks and substantial revisions to long range service plans. More recently, King County Metro gained approval for an increase in the sales tax, however, this new revenue does not fully offset the lost MVET. Historically, sales tax growth has been significantly slower than MVET growth. Unless ST has made appropriate corrections to the transit network coded in the model, the service levels and riderhsip will also be overstated.
7) Bus Service Speed and Reliability. Conversion of the downtown bus tunnel to rail operation will necessitate evicting all buses from the tunnel. Currently that totals over a thousand bus trips per day. Because the proposed rail line is short and serves relatively few stations, it will replace very little of the existing bus service. Therefore the routes that now provide fast and reliable service through the tunnel will be forced to operate on the congested surface streets. Slower and less reliable service will result. This impact is not inconsequential. Even with light rail in operation ST forecasts three out of four transit passengers will be bus riders. The City of Seattle and Metro have been examining plans to improve bus routing through the Seattle CBD, but these improvements have not yet been implemented. Even under the very most optimistic scenario, bus speeds on the surface streets will be significantly slower than bus service through the tunnel. Metro has budgeted 40,000 annual service hours to compensate for the additional running time.
Since bus routing and scheduling issues are still being worked out it is unclear whether ST has fairly and completely accounted for adverse service impacts in their ridership forecasts.