Impact of Transit Fare Increase on Ridership and Revenue

Impact of Transit Transit Fare Increase Inc rease on Ridership and Revenue Metropolitan Transportation Authority, New York City Robert L. Hickey The introduction of each fare incentive resulted in an increase in MetroCard market share, as well as an increase in the proportion of customers paying fares with discounted MetroCards (1-day, 7-day, and 30-day passes and bonus MetroCard). Customers continued to shift to discounted fare options between 1999 and 2002, so that by October 2002, nearly three-quarters of all trips were made with discounted MetroCards, as shown in Figures 1 a to 1d . As a result of customers’ switching to lower-priced fare options, the nonstudent average fare declined from $1.38 in 1996 to $1.04 in 2002. To make the purchase of MetroCards more mo re convenient, the agency began to install MetroCard vending machines (MVMs) at stations in 1999, and by September 2002, 20 02, all stations in the subway system were equipped with at least one MVM. The MVMs allow customers to purchase MetroCard s either with cash or with credit or debit cards. A new single-ride ticket, available for $1.50 in cash, also became available at all MVMs, but not at token booths. The May 4, 2003, fare change was the first NYCT fare increase implemented with the multiple-fare structure made possible by MetroCard. For the first time, the agency had the ability to increase fare media prices at different rates. Although that provided greater flexibility in developing a new fare stru cture, estimating the ridership and revenue effects of different fare scenarios became more complex. While there are considerable data available on the direct ridership effects of past NYCT fare increases, the agency had no previous experience with either the direct ridership effects of unlimited-ride pass price changes or the shift of customers when prices of different fare media increase at different rates and customers can shift from one fare medium to another.

On May 4, 2003, the Metropolitan Transportation Authority’s New York City Transit subsidiary (NYCT) raised its subway and bus fares for the first time since discounted and unlimited-ride MetroCards (electronic fare cards) were introduced between July 199 7 and January 1999. Before the 2003 fare increase, the agency did not have any experience with either the direct ridership effects of price changes for unlimited-ride passes or the shift of customers when prices of different fare media increased increased at different rates and customers cou ld shift from one fare medium to another. Partly on the basis of work done by other transit agencies, NYCT developed a spreadsheet model that used direct fare elasticities to estimate absolute ridership loss and used trip diversion rates (similar to crosselasticities) elasticitie s) to estimate the likelihood that passengers would shift from a fare instrument with a larger percentage increase to one with a smaller increase. The actual shift of customers between fare instruments after the fare increase was greater than projected, with the result of a lower than expected systemwide average fare. The negative revenue impact from the lower average fare was mostly offset by a lower than expected ridership decline. The estimated fare elasticities were below the average of historical NYCT fare elasticities and suggest that customers using unlimited-ride passes were less sensitive to fare changes.

On May 4, 2003, the Metropolitan Transit Authority’s New York City Transit subsidiary (NYCT) raised its subway and bus fares for the first time since November 1995, when the base fare rose f rom $1.25 to $1.50. In the period since the 1995 fare increase, NYCT completed systemwide installation of automated fare collection (AFC) equipment and expanded the acceptance of MetroCard, an electronic fare card capable of storing time or value, to all subway stations and bus routes. Between July 1997 and January 1999, NYCT introduced the following MetroCard fare incentives:

RIDERSHIP AND REVENUE TRENDS BEFORE FARE INCREASE

Free transfers from bus to subway and from subway to bus starting July 1997 (previously, (previo usly, a full fare was charged for these transfers); A 10% bonus value added on all value-based MetroCard purchases of $15 or more starting January 1998 [e.g., a customer spending $15 received a $16.50 MetroCard ($15 plus the 10% bonus of $1.50, equal to one free ride)]; Starting July 1998, 7-day and 30-day unlimited-ride passes; and Staring January 1999, 1-day fun pass. •

The 2003 fare increase came amid a period of unsettled subway and bus ridership resulting from a weak New York City economy and the impact of the September 11, 2001, World Trade Center attack. The ridership trend during the 2 years before the fare increase was in sharp contrast to the significant ridership growth that occurred between 1996 and 2000, when average weekday nonstudent subway ridership increased by 24% and weekday nonstudent bus ridership increased by 50%. The 1996 to 2000 ridership growth r esulted mainly from the introduction of the MetroCard fare incentives, as well as a robust New York City economy, as measured by 10.5% NYC employment growth over the same period. A NYCT paper on the impact of the AFC fare incentives was presented at the January 2000 TRB meeting (1).

•

• •

Office of Management and Budget, New York City Transit, Metropolitan Transit Authority, 2 Broadway, Room D17.81, New York, NY 10004. Transportation Research Record: Journal of the Transportation Research Board, No. 1927, Transportation Research Board of the National Academies, Washington, D.C., 2005, pp. 239–248.

239

2 40

Transportation Research Record 1927

30 Day Pass 7%

MetroCard 11%

Cash/Token/SRT 28% 7 Day Pass 20%

Regular MC 14% Cash/Token 89%

Bonus MC 31%

(a ) 30 Day Pass 11%

(b )

Cash/Token/SRT 19%

30 Day Pass 14%

Cash/Token/SRT 15%

Regular MC 11% 7 Day Pass 27%

Regular MC 13% 7 Day Pass 30%

1 Day Pass 3%

Bonus MC 25%

1 Day Pass 5%

Bonus MC 27% (c)

(d)

FIGURE 1 Histor Historical ical fare fare media market market share share (discounted (discounted fare media media shaded): shaded): (a ) October 1996, ( b ) October 1998, ( c ) October 2000, and ( d ) October 2002. (SRT single-ride ticket; MC MetroCard.) 



Figures 2 and 3 show the ridership and NYC emp loyment trends from January 2001 to April 2003. In early 2001, New York City employment growth began to slow, and starting in May 2001, employment began to decline slightly from the same month in 2000. Sub way ridership continued to increase, though not at the previous year’s rate, with January to August 2001 weekday subway r idership increasing 4.9% from the previous year, compared with 7.2% growth during the same period in 2000. Bus ridership remained strong between January and August 2001, with weekday bus ridership increasing by 7.5% from the previous year. In the 11 months following the September 11, 2001, World Trade Center attack, New York City lost nearly 145,000 jobs (3.8% decline), and weekday subway ridership declined by 2.6% from the previous 11-month period. The subway ridership loss was mainly at Manhattan stations (particularly lower Manhattan), with ridership still increasing in the boroughs outside Manhattan. Bus ridership continued to grow during that period but at a diminishing rate as the year went on. It is possible that some subway riders were initially diverted to buses

because of subway line reroutings or unscheduled service disruptions that occurred after September 11, particularly in lower Manhattan. Some customers may have also shifted to th e bus from fear of another anoth er attack. While subway ridership was lower than for the same period of the previous year throughout the 11 months following September 2001, most of the drop occurred in the months immediately after that time, time, and subway ridership stabilized and remained near the immediate postSeptember 11 level through most of 2002. Although it was expected that employment would increase in 2003, New York City jobs continued to decline, and subway and bus ridership both were declining slightly from the previous year through April 2003 (partially from severe weather in January and February). Despite the drop-off in ridership after September 11, 2001, combined subway and bus ridership rose 35.8% between 1996 and 2002. However, passenger revenue reven ue rose by only 4.7% over the same period because of the lower average fares resulting from the MetroCar d fare incentives.

Hickey

24 1

% Change from Prior Year 15%

10%

5%

0%

-5%

-10%

-15% Jan Jan Feb Feb Mar Mar Apr Apr May May Jun Jul Jul

Aug Aug Sep Oct Nov Nov Dec Jan Feb Mar Mar Apr Apr May May Jun Jul Jul

2001

2002

NYC NYC Empl Employ oyme ment nt

FIGURE 2

Aug Aug Sep Oct Nov Nov Dec Dec Jan Jan Feb Feb Mar Mar Apr Apr 2003

Subwa Subway y Avg. Avg. Wkdy. Wkdy. Ride Ridersh rship ip

Averagee weekday subway ridershi Averag ridershipp and NYC employm employment, ent, 2001–20 2001–2003. 03.

% Change from Prior Year 15%

10%

5%

0%

-5%

-10%

-15% Jan Jan Feb Feb Mar Apr May May Jun Jul Jul 2001

Aug Aug Sep Sep Oct Oct Nov Nov Dec Jan Feb Mar Mar Apr Apr May May Jun Jul Jul 2002

NYC NYC Empl Employ oyme ment nt

FIGURE 3

Aug Aug Sep Oct Nov Nov Dec Dec Jan Jan Feb Feb Mar Mar Apr Apr 2003

Bus Bus Avg. Avg. Wkdy Wkdy.. Rider Ridershi ship p

Average weekday bus ridership ridership and and NYC employ employment, ment, 2001– 2001–2003 2003..

2 42


NEW FARE STRUCTURE While the impact from lower average fares was offset by the large ridership growth in previous years, projections of lower ridership growth, coupled with slower growth in tax-based subsidies and rapidly increasing agency expenses, were expected to lead to budget deficits in 2003 and 2004. Thus, the primary goal in setting the new fare structure was to generate sufficient revenue to maintain a balanced budget through 2004, as required by state charter. A secondary goal was to develop a fare structure that would advance the agency’s business strategy of enhancing the convenience of purchasing MetroCards while minimizing revenue support and maintenance expenses. To that end, the agency sought to structure the fare options to do the following: Encourage customers to purchase higher value cards to reduce the number of sales transactions, shorten lines at MVMs and booths, and reduce the amount of card stock that must be purchased; Encourage even-bill denomination purchases to shorten transaction times and reduce the number of machine jams and service calls; and Encourage debit or credit purchases to reduce the amoun t of cash that must be processed (Because a large proportion of 30-day pass sales were debit or credit purchases, a fare structure that moved customers to the 30-day pass wou ld likely increase debit or credit purchases). •

•

•

Table 1 shows the fare structure that was approved by the MTA boardfor impl implemen ementati tation on in May 2003.The fare faress for disc discounte ounted d highe higherr value cards (7-day and 30-day passes and bonus MetroCard) increased at a lower rate than did the 33.3% increase ($1.50 to $2) for nondiscounted lower-value cards (single-ride tickets and MetroCards with no bonus value added) and for cash. The bonus v alue added to MetroCards was increased from 10% to 20%, for an effective bonus fare of $1.67, a 22.2% increase over the former effective bonus fare of $1.36. As a further inducement to attract customers to the bonus MetroCard, the minimum threshold thresho ld for receiving the bonus was lowered from $15 to $10, so that customers could get a bonus trip when purchasing only five trips. Beforethe Bef orethe far faree inc increa rease,the se,the $4 pri pricefor cefor the1-da the1-day y pas pass,with s,with a bre breakakeven amount of only three trips (based on the effective bonus fare), was considered low compared compa red with similar passes in other cities, and it was purchased by customers beyond the intended market of visitors to the city. Therefore, the 1-day fun pass was the only fare category that increased at a higher rate than the base fare, rising by 75% from $4 to $7.

TABL T ABLE E1

A new $33 7-day express bus plus pass, valid on express and local buses and the subway, was introduced to replace the $120 30-day express bus plus pass. Final Fin ally,at ly,at thesametim thesametimee itappro itapprovedthe vedthe ne new w far faree str struct ucture ure,, th thee MTA board also approved a resolution to discontinue acceptance of the token, for more than 40 years the sole means of paying the subway fare.

NYCT FARE MODEL Shortly after the introduction of the MetroCard fare incentives was completed in 1999, NYCT began to explore techniques to estimate the ridership and revenue effects of future fare increases. Several transit agencies were contacted to obtain information on their experience with past fare increases and their methods for estimating ridership and revenue changes caused by changes in fare stru cture. At the time, the Metropolitan Atlanta Rapid Transit Authority (MARTA) had developed a model to identify shifts among fare payment methods and to forecast revenue and ridership changes resulting from a fare change (2). The Southeastern Pennsylvania Transportation Authority (SEPTA) developed a similar approach that used factors to determine the ridership shift between fare categories in response to a change in the fare ratio of two fare instruments. NYCT staff developed a fare model that used the approach developed by MARTA and SEPTA, S EPTA, and they modified it to reflect NYCT’s fare structure and ridership patterns. Base ridership and average fares, by fare category, are entered along with the proposed fares. The model uses a two-stage spreadsheet that applies direct fare elasticities to estimate the ridership loss from the absolute fare change within each fare category, and then it applies a ridership d iversion rate (similar to a cross-elasticity) for each pair of fare instruments to estimate the likelihood that passengers will shift from a fare instrument with a larger percentage increase to one with a smaller increase. The direct elasticities used in the model are based in part on res ults from the nine NYCT fare increases that occurred between 1972 and 1995. As shown in Table 2, subway average weekday ridership point elasticitiess with respect to the fare (on the basis of average weekday elasticitie ridership for a full year before and after the fare increase) averaged 0.09 between 1972 and 1995. Weekend subway ridership elasticities were higher ( 0.12 for Saturdays and 0.21 for Sundays due to the more discretionary nature of weekend subway ridership. Bus average weekday ridership elasticities averaged 0.37 over the same period (Table 2). Saturday and Sunday Sund ay bus ridership elasticities were close to the weekday elasticities ( 0.38 for Saturdays and 0.39), because bus ridership is generally more discretionary (i.e., less work −

−

−

−

−

−

Faree Struct Far Structure ure Appro Approved ved May May 4, 2003 2003 MetroCard Bonus Cash/SRT/ Tokena

Regular MetroCard

Percent

Threshold

Express Bus Effective Fare

7-Day Pass

30-Day Pass

1-Day Pass

Base Fare

30-Day Passb

7-Day Passb

Old fare

$1.50

$1.50

10

$15

$1.36

$17

$63

$4

$3

$120

NA

New fare

$2.00

$2.00

20

$10

$1.67

$21

$70

$7

$4

NA

$33

—

—

75%

33%

NA

NA

% change

33%

33%

SRT single-ride ticket. a Token was eliminated under new fare structure. b Also valid on subway and local bus. 

22%

24%

11%

Hickey

24 3

TABLE 2 NYCT Fare Increas TABLE Increases: es: Average Average Weekday Riders Ridership hip Change Change and Point Elasticities Subway Date of Fare Change 1/5/72

Change in Fare

Ridership Change

17%

−

9/1/75

43%

−

6/28/80

20%

−

7/3/81

25%

−

1/2/84

20%

−

1/1/86

11%

1/1/90

15%

1/1/92

9%

11/12/95

20%

Average

20%

Point Elasticity

4%

−

5%

−

3%

−

3%

−

1%

−

1%

0.24 0.12 0.13 0.11

2%

−

17%

−

5%

−

11%

−

7%

−

3%

−

6%

−

4%

−

8%

−

7%

−

−

0.29

−

0.04

−

0.04

−

0.09

−

−

0.38

−

−

−

−

1%

6%

0.12 −

Point Elasticity

−

−

4%

−

Ridership Change

0.07

0%

−

oriented) than subway ridership. The elasticities shown in Table 2 do not account for changes in New York City employment. The NYCT fare model is illustrated in Figure 4 for subway and Figure 5 for bus. Part 1 in each figure shows show s the model inputs, including base revenue and ridership, old and n ew fares, direct elasticities, elasticities, and ridership diversion rates. A higher direct elasticity was used for low-value fare instruments (regular MetroCard, cash, and single-ride ticket), while a lower elasticity was used for high-value fare instruments (bonus MetroCard, 7-day and 30-day passes). From experience at other agencies, it was assumed that customers who purchase fare media in smaller increments ride the system less frequently and are more discretionary in their trip making than are others. The ridership diversion rates reflect estimates of the likelihood of customers switching between pairs of fare instruments, according to experience in other cities and knowledge of NYCT customers. The highest trip diversion rates were set for the 7-day pass–bonus MetroCard at 0.40, and the bonus–regular fare MetroCard at 0.30. It was assumed that there would be no shift from MetroCards (including single-ride tickets) to tokens and cash or from MetroCards to single-ride tickets. Part 2 of Figures 4 and 5 shows the model output. Section A of Part 2 shows the ridership loss for each fare instrument resulting from the application of direct elasticities (ridership times percentage fare increase times elasticity). Sections B through E of Part 2 in Figures 4 and 5 show the calculation of the ridership shift between fare instrum ents. As an example, the shift from 7-day pass to bonus MetroCard in the subway model, Figure 4, is highlighted. The model mod el first calculates the relative change in average fare between each pair of fare instruments on the basis of current and proposed fares. Under the fare structure in effect before the May 2003 fare increase, the ratio of the 7- day pass average fare to the bonus MetroCard average fare was 0.688 (Section B). Under the May fare increase, with the 7-day pass price increasing by 23.5% and the bonus MetroCard price increasing by 22.2% (shown in Part 1), the ratio of the 7-day pass average fare to the bonus MetroCard average fare becomes 0.696 (Section C), for a relative change of 1.1% (Section D) [i.e., (0.696 0.688)/0.688]. The model calculates the number of shifted trips as a percentage of ridership for the fare instrument that is becoming relatively more expensive. Thus, on the basis of a 7-day pass–bon us MetroCard diversion rate of 0.40, the 1.1% increase in the 7-day pass to bonus Metro−

Bus

0.40 0.26 0.42 0.35 0.30 0.37 0.41 0.41 0.37

Card ratio would decrease 7-day pass ridership by 0.43% (0.40  1.1%) or 1,624,636 trips, and it would increase bonus MetroCard ridership by the same amount (Section E). When passengers switch to using passes, it was assumed that they would increase their trip making by 20%. The estimated percentage of generated pass trips is based in part on NYCT’s ridership change after the introduction of passes in 1998. The 20% figure is also in line with results in other cities after the introduction of passes. Pass users in Atlanta increased their trip making by 14% after a monthly pass was introduced in 1979 (3), and 7-day pass users in Chicago Ch icago increased their transit travel 31% after a 7-day pass was introduced in late 1998 ( 4 ). Using the ridership loss from Section A and the trip diversions in Section E (including the 20% factor for passes), ridership after the fare change is calculated (Section F). Revenue is calculated by multiplying the new average fares for each fare medium (current average fare increased by the percent increase in far e and pass price) by the ridership after the fare change. The trip diversion rates used in the model were conservative in that they result in small shifts from fare instruments with larger increases to those with smaller increases. Therefore, there was a risk that revenue would be lower than projected if mor e customers switched to lower priced fare media. There was also the risk of lower revenue if the actual ridership loss were higher than projected pr ojected or if the average pass fares were lower than projected. Because of the risk that deviations in a series of factors (such as lower than expected average fare accompanied by higher than expected ridership loss) could compound the revenue variance from for ecast, many of the fare options that were examined were tested with alternate scenarios that measured the effect of changing one or more of those factors. The various scenarios provided a range of revenue projections to help identify the risks in modeling a given fare structure. Aside from the inherent risks in projecting pro jecting ridership loss and diversions, the model also has limitations that, in some cases, required adjustments to the model results. The main limitation of the NYCT model is that it cannot accommodate introduction of new fare media (e.g., the 7-day express bus plus pass) or elimination of existing fare media (e.g., tokens). The potential market share of the new 7-day express bus pass was estimated outside the model by obtaining records of all MetroCard trips made by express bus users over a typical week and by using

2 44


Part 1: Model Inputs Inputs Base Ride Ri ders rshi hip p Reve Re venu nue e Avg Fare

Token 146,18 146, 188, 8,04 044 4 $212 $2 12,1 ,197 97,4 ,407 07 $1.45

Single_Ride 13,5 13 ,585 85,6 ,671 71 $20, $2 0,37 378, 8,50 506 6 $1.50

Regular MC 163,96 163, 969, 9,10 100 0 $222,04 $222, 046, 6,45 456 6 $1.35

Bonus MC 7-Day 30-Day 356,68 356, 686, 6,14 140 0 379,7 379 ,758 58,6 ,664 64 21 216, 6,73 735, 5,33 332 2 $439 $4 39,0 ,059 59,7 ,78 86 $321 $321,7 ,750 50,9 ,952 52 $197 $197,0 ,098 98,6 ,645 45 $1.23 $0.85 $0.91

1-Day 81,644 81,6 44,1 ,141 41 $77, $7 7,85 859, 9,61 619 9 $0.95

Old Fare New Fare % Change New Avg Fare

$1.50 $2.00 33.3% $1.94

$1.50 $2.00 33.3% $2.00

$1.50 $2.00 33.3% $1.81

$1.36 $1.67 22.2% $1.50

$17 $21.00 23.5% $1.05

$63 $70.00 11.1% $1.01

$4 $7.00 75.0% $1.67

Direct Elasticities

Token -0.15

Single_Ride -0.15

Regular MC -0.15

Bonus MC -0.10

7-Day -0.10

30-Day -0.10

1-Day -0.15

Token --0.25 0.25 0.25 0.20 0.10 0.05

Single_Ride 0.00 - -0.25 0.20 0.05 0.05 0.05

Regular MC 0.00 0.00 --0.25 0.05 0.05 0.05

Bonus MC 0.00 0.00 0.30 --0.30 0.20 0.10

7-Day 0.00 0.00 0.25 0.40 --0.20 0.10

30-Day 0.00 0.00 0.10 0.15 0.20 --0.05

1-Day 0.00 0.00 0.05 0.05 0.05 0.05 ---

A. Ridership Loss (from application application of direct elasticities) Token Single_Ride Regular MC Rid ide ers rsh hip Loss -7,3 -7 ,30 09,4 ,402 02 -67 -6 79, 9,2 284 -8,1 -8 ,198 98,4 ,45 55

Bonus MC -7,92 -7,9 26,3 ,35 59

7-Day -8,9 -8 ,93 35,4 ,49 98

30-Day -2,4 -2 ,40 08, 8,1 170

1-Day -9,1 -9 ,18 84, 4,9 966

Ridership Diversion Rates Token Single-Ride Reg MC Bonus MC 7-day 30-day 1-day

Part 2: Model Outputs Outputs

B. Old Fare Ratios Token --0.968 1.072 1.179 1.713 1.596 1.522

Single_Ride 1.033 --1.108 1.219 1.770 1.649 1.573

Regular MC 0.933 0.903 --1.100 1.598 1.489 1.420

Bonus MC 0.848 0.821 0.909 --1.453 1.354 1.291

7-Day 0.584 0.565 0.626 0.688 --0.932 0.888

30-Day 0.627 0.606 0.672 0.739 1.073 --0.954

1-Day 0.657 0.636 0.704 0.775 1.126 1.049 ---

Token --0.968 1.072 1.286 1.849 1.915 1.160

Single_Ride 1.033 --1.108 1.329 1.911 1.979 1.198

Regular MC 0.933 0.903 --1.200 1.725 1.787 1.082

Bonus MC 0.777 0.752 0.833 --1.437 1.489 0.901

7-Day 0.541 0.523 0.580 0.696 --1.036 0.627

30-Day 0.522 0.505 0.560 0.672 0.965 --0.605

1-Day 0.862 0.834 0.924 1.109 1.595 1.652 ---

D. Change in Ratios (positive means fare is relatively more expensive) Token Single_Ride Regular MC Bonus MC Token --0.0% 0.0% -8.3% Single-Ride 0.0% --0.0% -8.3% Reg MC 0.0% 0.0% ---8.3% Bonus MC 9.1% 9.1% 9.1% --7-day 7.9% 7.9% 7.9% -1.1% 30-day 20.0% 20.0% 20.0% 10.0% 1-day -23.8% -23.8% -23.8% -30.2%

7-Day -7.4% -7.4% -7.4% 1.1% --11.2% -29.4%

30-Day -16.7% -16.7% -1 -16.7% -9.1% -10.1% ---36.5%

1-Day 31.3% 31.3% 31.3% 43.2% 41.7% 57.5% ---

7-Day 2,784,534 64,694 780,805 -1,624,636 - --8,488,723 1,700,920 - 4, 4,782,406

30-Day 3,508,513 163,028 1,967,629 8,560,467 8,488,723 --2,347,269 25,035,630 25

1-Day 0 0 -1,275,690 -1 -1,762,771 - 1, 1,700,920 -2,347,269 ---7,086,650

Bonus MC 7-Day 30-Day 352,30 352, 309, 9,50 504 4 36 366, 6,04 040, 0,75 759 9 239 239,3 ,362 62,7 ,792 92 $530,044 $530 ,044,04 ,048 8 $383, $383,099, 099,840 840 $24 $241,86 1,862,23 2,232 2

1-Day 65,372 65,3 72,5 ,525 25 $109,098 $109 ,098,948 ,948

Token Single-Ride Reg MC Bonus MC 7-day 30-day 1-day C. New Fare Ratios Token Single-Ride Reg MC Bonus MC 7-day 30-day 1-day

E. Trip Diversions (includes increase in trips by new pass users) Token Single_Ride Regular MC Token --0 0 Single-Ride 0 --0 Reg MC 0 0 --Bonus MC -3,322,456 -247,012 -3,726,570 7-day -2,320,445 -53,911 -650,671 30-day -2,923,761 -135,857 - 1, 1,639,691 1-day 0 0 1,275,690 Total Shif t -8,566,662 -436,780 -4,741,243 F. Post Fare Increase Increase Ridership and Revenue Revenue Token Single_Ride New Ridership 130, 13 0,31 311, 1,98 980 0 12,4 12 ,469 69,6 ,607 07 New Revenue $252,20 $25 2,203,61 3,610 0 $24,939 $24 ,939,214 ,214

F IG IG UR UR E 4

F a re re m od od e ls ls f o r su su b wa wa y. y.

Regular MC 151,02 151, 029, 9,40 402 2 $272,698 $272 ,698,075 ,075

Bonus MC 3,322,456 247,012 3,726,570 --1,624,636 -7,133,723 1,762,771 1, 3,549,723

Hickey

24 5

Part 1: Model Inputs Base Ride Ri ders rshi hip p Reven Re venue ue Avg Fare

Cash/Token 139,91 139, 915, 5,66 660 0 $132,8 $13 2,856, 56,63 632 2 $0.950

Old Fare New Fare % Change New Avg Fare

$1.50 $2.00 33.3% $1.36

$1.36 $1.67 22.2% $1.14

$17.00 $21.00 23.5% $1.04

$63.00 $70.00 11.1% $0.98

$4.00 $7.00 75.0% $1.67

Cash/Token -0.30

Single Regular MC -0.30 -0.30

Bonus MC -0.20

7-Day -0.20

30-Day -0.20

1-Day -0.30

E. Ridership Dive Diversion rsion Rates Rates Cash/Token Token --Single-Ride 0.25 Reg MC 0.25 Bonus MC 0.25 7-day 0.20 30-day 0.10 1-day 0.05

Single Regular MC 0.00 0.00 --0.00 0.30 --0.30 0.25 0.10 0.05 0.05 0.05 0.05 0.05

Bonus MC 0.00 0.00 0.30 --0.30 0.20 0.10

7-Day 0.00 0.00 0.25 0.40 --0.20 0.10

30-Day 0.00 0.00 0.10 0.15 0.20 -- 0.05

1-Day 0.00 0.00 0.05 0.05 0.05 0.05 ---

A. Ridership Loss (from application application of direct elasticities) elasticities) Cash/Token Single Regular MC Ride Ri ders rshi hip p Lo Loss ss -13, -1 3,99 991, 1,56 566 6 -68, -6 8,33 331 1 -6 -6,8 ,807 07,6 ,657 57

Bonus MC -6,578 -6,5 78,5 ,518 18

7-Day -9,9 -9 ,952 52,2 ,243 43

30-Day -1,6 -1 ,657 57,5 ,535 35

1-Day -9,1 -9 ,122 22,4 ,491 91

Direct Elasticities

$1.50 $2.00 33.3% $1.27

Single Regular MC Bonus MC 7-Day 30-Day 1-Day 683,31 683, 311 1 68, 68,07 076, 6,57 572 2 148, 148,01 016, 6,66 662 2 211 211,4 ,485 85,1 ,15 59 74,5 74 ,589 89,0 ,086 86 40,5 40,544 44,4 ,403 03 $1,02 $1 ,024,9 4,924 24 $6 $69,6 9,648 48,22 ,221 1 $137,6 $137,641, 41,48 488 8 $178, $178,45 459,1 9,174 74 $65,96 $65,966,2 6,281 81 $38,6 $38,664, 64,93 932 2 $1.500 $1.023 $0.930 $0.844 $0.884 $0.954 $1.50 $2.00 33.3% $2.00

Part 2: Model Outputs Outputs

B. Old Fare Ratios Cash/Token --0.633 0.928 1.021 1.125 1.074 0.996


Bonus MC 0.979 0.620 0.909 --1.102 1.051 0.975

7-Day 0.889 0.563 0.825 0.907 --0.954 0.885

30-Day 0.931 0.590 0.864 0.951 1.048 --0.927

1-Day 1.004 0.636 0.932 1.026 1.130 1.078 ---

Cash/Token --0.633 0.928 1.114 1.215 1.288 0.759


Bonus MC 0.898 0.568 0.833 --1.090 1.157 0.681

7-Day 0.823 0.521 0.764 0.917 --1.061 0.625

30-Day 0.776 0.491 0.720 0.865 0.943 --0.589

1-Day 1.318 0.834 1.223 1.468 1.601 1.698 ---

D. Change in Ratios (positive means fare is relatively more expensive) Cash/Token Single Regular MC Bonus MC Token --0.0% 0.0% -8.3% Single-Ride 0.0% --0.0% -8.3% Reg MC 0.0% 0.0% ---8.3% Bonus MC 9.1% 9.1% 9.1% --7-day 7.9% 7.9% 7.9% -1.1% 30-day 20.0% 20.0% 2 0.0% 20 10.0% 1-day -23.8% -23.8% -23.8% -30.2%

7-Day -7.4% -7.4% -7.4% 1.1% --11.2% -29.4%

30-Day -16.7% -1 -16.7% -16.7% -9.1% -10.1% ---36.5%

1-Day 31.3% 31.3% 31.3% 43.2% 41.7% 57.5% ---

E. Trip Diversions (includes increase in trips by new pass users) Cash/Token Single Regular MC Bonus MC Token --0 0 3,179,901 Single-Ride 0 --0 18,636 18 Reg MC 0 0 --1,547,195 Bonus MC -3,179,901 -18,636 -1,547,195 --7-day -2,220,883 -5,423 -270,145 904,749 30-day -2,798,313 -6,833 -680,766 -6 -2,960,333 1-day 0 0 633,506 875,391 Tota To tall Shif iftt -8,1 -8 ,19 99,0 ,098 98 -30 -3 0,8 ,89 92 -1 -1,8 ,86 64,5 ,59 99 3,5 ,56 65,5 ,53 39

7-Day 2,665,060 6,508 324,174 -904,749 ---4,727,315 844,675 -1,,79 -1 791 1,6 ,64 48

30-Day 3,357,976 8,200 816,919 3,552,400 4,727,315 --1,165,652 13,62 13,6 28,4 ,46 61

1-Day 0 0 -633,506 -875,391 - 844,675 -8 -1,165,652 -1 ---3,5 -3 ,51 19, 9,2 223

Token Single-Ride Reg MC Bonus MC 7-day 30-day 1-day C. New Fare Ratios Token Single-Ride Reg MC Bonus MC 7-day 30-day 1-day

F. Post Fare Increase Increase Ridership and Revenue Cash/Token Single Regular MC Bonus MC 7-Day 30-Day 1-Day New Ridership 117,7 11 7,724 24,99 ,996 6 584,08 584 ,088 8 59,404 59,404,31 ,315 5 14 145,0 5,003 03,68 ,682 2 19 199,7 9,741 41,26 ,269 9 86,56 86 ,560,0 0,012 12 27 27,90 ,902,6 2,689 89 New Revenue $149,047 $149 ,047,376 ,376 $1,168,1 $1,1 68,127 27 $81,034,3 $81,034,337 37 $164,804 $164,804,080 ,080 $208,2 $208,207,8 07,879 79 $85,059 $85,059,243 ,243 $46,566 $46,566,162 ,162

F IG IG UR UR E 5

F a re re m od od e ls ls f or or b us us .

2 46


those data to create a trip frequency matrix of express bus users. All users making enough weekly express bus, local bus, and subway trips to break even or save money (based on the difference between the $33 pass price and the cost of the trips at the bonus fare) were assumed to shift to the new pass. The new fare media choice of token users was based on surveys conducted at selected token booths that measured the percentage of token purchase transactions by denomination (e.g., single-token purchases accounted for 54% of token sales transactions). It was assumed that 80% of the single-token purchasers would switch to the single-ride ticket and all other token users would switch to the regular MetroCard [no shift from token to cash (on buses), bonus MetroCard, or any of the unlimited-ride passes]. Another limitation of the model is that, while it calculate calculatess the passenger shift between the bonus MetroCard and other fare instruments in response to a change in the bonus percentage (and the resulting change in average fare), it cannot estimate the effect of lowering or raising the purchase threshold for receiving additional bonus value. The effect of changing the bonus threshold was calculated independently, on the basis of data on MetroCard sales transactions by purchase amount as well as the token purchase survey. It was assumed that customers purchasing amounts near or above the threshold amount would take advantage of the bonus. Finally, because the model uses aggregate data, average fares are used within each fare category. As a result, the model assumes that customers who shift between fare categories previously paid the same average fare, as did all customers in the old fare category, and they will pay the same average fare as existing customers do in the new fare category. In reality, the number of trips made per month or week by customers who shift from value-based cards to passes, for example, is likely closer to the break-even amount than with the number of trips made by the average pass user. Therefore, there is a risk of understating revenue for customers who shift from value-based fare cards to lower-fare passes, because they likely make fewer trips and would therefore have a higher average fare than did existing pass users. Conversely, there is a risk of overstating revenue when a pass has an extremely large increase (e.g., 1-day pass), because the average n umber of trips made by the remaining pass users would likely increase, which would result in a lower average fare.

Average Fares On the basis of rate of increase in each fare category and the new market shares resulting from the shift of customers among fare categories, the model projected that the combined subway and bus nonstudent average fare would rise from $1.04 to $1.30, a 25.3% increase. The subway average fare was projected to increase by 24.6%, and the bus average fare was projected to rise by 25.8%.

Ridership Loss The model projected a 4.1% total nonstudent ridership loss, or a 0.16 ridership elasticity based on a 25.3% fare increase. The projected subway nonstudent ridership loss was 2.9% ( 0.12 elasticity based on a 24.6% increase), and bus ridership was projected to decline by 6.5% ( 0.25 elasticity based on a 25.8% increase). The ridership loss resulting from application of direct elasticities elasticities was partially offset by the assumed 20% increase in trip making by customers’ shifting to 7-day or 30-d ay unlimited-ride passes. −

−

−

Revenue Gain As a net result of the average fare increase partially offset by the ridership loss, total revenue was projected to increase by 20.0%. The full year impact (2004) was projected at $430 million, and the 2003 impact (May through December) was projected at $286 million.

ACTUAL RESULTS

MODEL RESULTS

The three key components in comparing projected and actual results from the May 4 fare increase are the following:

Fare Media Market Share Table 3 compares projected fare media market share from the base 2003 forecast (excluding fare increase) with the market share yielded by the model for the approved fare structure, which is adjusted for th e elimination of the token, introduction of the 7-day express bus pass, and change in bonus threshold. The fare categories with the lowest

TABLE TA BLE 3

percentage fare increase (bonus MetroCard and 30-day pass) were projected to have the largest percentage market share increase. The largest percentage decrease was in the category of token, cash, and single-ride ticket, owing mainly to the elimination of tokens and the assumption that most token customers would shift to regular MetroCard. As a result of the shift of token users, the regular MetroCard market share was projected to increase, despite having a 33.3% fare increase. The 7-day pass market share was essentially unchanged due to trip shifting from cash and token and regular MetroCard, offset by customers’ shifting to the 30-day pass.

1. Avera Average ge fare, fare, by fare fare medium medium;; 2. Actual market market shares shares and the resulting resulting systemwide systemwide average average fare; and 3. Actua Actuall riders ridership hip loss.

Fare Media Media Market Market Share and Average Average Fare: Base and Project Projected ed Fares Market Share

1-Day

7-Day Express

Nonstudent Average Fare

75.0%

NA

—

Cash / SRT/ Token

Regular MetroCard

Bonus MetroCard

7-Day

30-Day

Increase in fare

33.3%

33.3%

22.2%

23.5%

11.1%

Base

14.7%

11.4%

24.7%

29.0%

14.3%

6.0%

NA

$1.04

7.5%

13.1%

28.4%

29.0%

16.6%

4.8%

0.5%

$1.30

Projected

Hickey

24 7

Together, the three components determine how close the actual revenue increase is to the model projection.

the ridership change after the fare increase, with the difference in the two variances assumed to reflect the fare increase impact. That method yields a 0.5% subway ridership increase ( 1.9% May through December 2003 subway ridership variance minus the 2.4% March and April variance) and a 4.2% bus ridership decrease ( 4.1% May through December bus ridership variance minus the 0.1% positive March and April variance). The second method compares the ridership change from 2002 to 2003 in the 2 months before the fare increase with the change after the fare increase. That method yields a 0.9% decline in subway rider ship and a 4.7% bus ridership loss. Taken together, the results of these methods suggest that the ridership loss due to the fare increase was lower than expected. While it is unlikely that there was actually a positive impact on subway ridership as inferred by comparing ridership to b udget, the previous year comparison (supported by other estimates) yields a subway ridership loss of approximately 1%, well below the projected 2.9% loss. The bus ridership loss was somewhere in the range of 4% to 5%, compared with the projected 6.5% loss. The resulting combined subway and bus ridership loss is approximately 2% compared with the pr ojected 4.1% ridership loss. −

−

Average Fares by Fare Medium

−

As mentioned earlier, there ther e was a risk that the actual average fares by fare medium could be lower than projected. However, except for the 1-day pass, the average fares by fare medium were close to, and in many cases higher than, the projected average fares. The 1-day pass average fare was 6% lower than projected. With a 75% increase in the 1-day pass price, the number of trips requir ed to break even increased from 3 to 5. As a result, the average number of trips per 1-d ay pass increased from 4.2 to about 4.5 per pass. Thus, the increase in the 1-day pass average fare was less than it would have been had the number of trips per pass remained constant. Despite the lower 1-day pass average fare, the average fares by fare medium had little effect on the systemwide average fare.

Fare Media Market Share and Systemwide Average Fare As shown in Table 4, the shift of customers to the 30-day pass and the shift away from the 1-day pass were greater than expected. In addition, although the overall pay-per-ride MetroCard market share was close to the projection, the proportion of pay-per-ride trips made with bonus MetroCard was higher than expected. The underestimate of market share for the fare media with the lowest averag e fares (30-day pass, bonus) yielded a $1.26 systemwide average fare, 2.5% lower than the projected $1.30 average fare. The finding wo uld suggest that, if the expected ridership loss occurred, total revenue would be 2.5%, or about $64 million, lower than projected.

Ridership While it is a fairly simple task to calculate the ridership change in relation to budget (or previous previou s year) after the fare increase, it is not so simple to separate the impact of the fare increase from that of other factors, such as the economy or the u nderlying ridership trend. The analysis was further complicated by lower ridership in January and February 2003 from unusually severe winter weather. As a result, March and April 2003 were used to estimate the ridership trend before the fare increase. Table 5 shows the results from two methods used to isolate the fare increase impact on ridership. The first method compares the ridership change from the 2003 base ridership budget (i.e., excluding the fare increase impact) in the 2 months before the fare increase with

TABLE TA BLE 4

Revenue It appears that most of the negative revenue impact from the lower than expected average fare was offset by the positive impact from the lower than expected ridership loss. Depending on the methodology used, the revenue impact from the fare increase was between $6 and $20 million below the projected impact. With a similar approach to the method as shown comparing ridership to budget, a comparison of the revenue variance from the budget before the fare increase in the periods before and after the fare increase yields a 20.2% revenue increase. Applying that percentage to the revenue budget before the fare increase for May 4 through December31 Dece mber31 (adj (adjuste usted d for the theMarc March h and andAprilbudget Aprilbudget vari variance ance)) yiel yields ds a revenue increase of approximately $280 million, about 2% lower than the projected $286 million increase. Alternatively, a comparison of the revenue change from 2002 to 2003 in the periods before and after the fare increase yields a revenue increase of approximately $266 million, nearly 7% (or $20 million) below the projected increase.

Elasticities The $1.26 systemwide (subway and bus combined) averag e fare after the fare increase represented a 21.2% increase incr ease over the full year 2002 average fare of $1.04. Under the assumption that total subway and

Fare Media Media Market Market Share Share and Average Average Fare: Project Projected ed and Actual Actual Fares Fares Market Share

Increase in fare Projected a

Actual a

Nonstudent Average Fare

Cash / SRT/ Token

Regular MetroCard

Bonus MetroCard

7-Day

30-Day

1-Day

7-Day Express

33.3%

33.3%

22.2%

23.5%

11.1 %

75.0%

NA

—

7.5%

13.1%

28.4%

29.0%

16.6%

4.8%

0.5%

$1.30

7.5%

10.8%

32.0%

27.8%

20.0%

1.7%

0.2%

$1.26

July–Decemberr 2003 results. July–Decembe

2 48


TABL TABLE E5

Ridersh Rid ership ip Com Compar parison isonss Ridership Change from Budget Subway

March–April

2.4%

−

a

May–Dec. Expected

−

4.1%

−

4.2%

−

6.5%

−

−

0.5%

−

2.9%

−

−

Total

0.1%

1.9%

−

Difference

Bus

Ridership Change from 2002 to 2003 Subway

1.5%

−

1.3%

2.6%

−

1.1%

−

4.1%

−

Bus

Total

0.3%

−

4.4%

−

4.7%

−

6.5%

−

2.1%

−

0.9%

−

2.9%

−

0.7% 2.9% 2.2% 4.1%

Excludes August 15–18, 2002 & August 14–17, 2003, because of blackout on August 14, 2003.

a

bus ridership declined 2%, the resulting systemwide fare elasticity is 0.10, compared with the 0.16 elasticity yielded by the model. The subway average fare increased 20.5% from $1.11 in 2002 to $1.34 after the fare increase. Assuming a subway ridership loss of 1% yields an elasticity of 0.05, compared with 0.12 from the model. The bus average fare increased 22.4% from $0.92 in 2002 to $1.12 after the fare increase. On the basis of a bus ridership loss of 4% to 5%, the bus elasticity is between 0.18 and 0.22, compared with the projected 0.25 elasticity. −

Outliers

−

−

−

−

−

−

CONCLUSIONS Absent available market research data on customer preferences or historical data on fare media shifts after past fare increases, the fare model provided a reasonable reason able estimate of the revenue impact from the fare increase. From the actual results, there are several modifications that will likely improve the model’s ability to forecast the impact of future NYCT fare changes.

As a result of fare increase of 75% in the 1-day pass from $4 to $7, the market share for the 1-day pass dropped to 1.7% compared with the projected 4.8% market share, as shown in Table 4. It is clear from that large market share decrease that, when the rate of increase for a particular fare medium is disproportionate to all other fare categories, the standard trip diversion rates, while valid for smaller increases, may not be sufficient to project the shift away from that fare instrument. In such cases, it may be prudent to estimate the market share shift independently, on the basis of available data or general knowledge of customer behavior.

ACKNOWLEDGMENTS The author thanks Larry Hirsch of the revenue analysis unit of NYCT Office of Management and Budget for his constructive suggestions. Special thanks go to David Jordan (retired from the revenue analysis unit) for his suggestions and for his efforts in developing the earlier versions of the NYCT model.

Direct Elasticities The actual elasticities estimated above are well below the average of the historical elasticities shown in Table 2 and the direct elasticities used in the model. It is likely that customers using unlimited-ride passes are less sensitive to fare changes because, onc e they have purchased a pass, there is no incentive to reduce the number of trips made. In fact, the cost per trip declines with each additional trip made with a pass. In future versions of the model, lower direct elasticities will be used, particularly for unlimited-ride passes.

Ridership Diversion Rates From the greater than expected shift among fare media, it is clear that the ridership diversion rates used in the model were too low. An analysis of the actual results revealed that if the diversion rates used in the model were doubled, the resulting market share projection would have closely matched the actual shift between fare media.

REFERENCES 1. Hirsch, Hirsch, L. R., J. D. Jordan, Jordan, R. R. L. Hickey, Hickey, and and V. Cravo. Cravo. Effects Effects of Fare Transportation on Research Incentives on New York City Transit Ridership. In Transportati Record: Journal of the Transportation Research Board, No. 1735, TRB, National Research Council, Washington, D.C., 2000, pp. 147–157. 2. Harris, A. A. E., R. Thomas, Thomas, and D. Boyle. Boyle. Metropolitan Metropolitan Atlanta Atlanta Rapid Transit Authority Fare Elasticity Model. In Transportation Research Record: Journal of the Transportation Research Board, No. 1669, TRB, National Research Council, Washington, D.C., 1999, pp. 123–128. 3. McC McColl ollum, um, B. B. E., and and R. H. Prat Pratt. t. TCRP Report 95: Traveler Response to Transportation System Changes. Chapter 12—Transit Pricing and Fares.

Transportation Research Board of the National Academies, Washington, D.C., 2004. 4. Foote, P. J., and D. G. Stuart. Stuart. Impacts Impacts of Tansit Fare Fare Policy Policy Initiative Initiative Under an Automated Fare System. Transportation Quarterly, Vol. 54, No. 3, 2000, pp. 51–66. The Public Transportation Marketing and Fare Policy Committee sponsored publication of this paper.

Impact of Transit Fare Increase on Ridership and Revenue

Recommend Documents