Voters in Ann Arbor as well as in Ypsilanti and Ypsilanti Township will be asked sometime in 2014 to approve a transit millage – to be levied by the Ann Arbor Area Transportation Authority (AAATA). A vote by the AAATA board to place a question on the ballot is likely to come at its Feb. 20 meeting.
If it’s approved, the five-year millage will be used to pay for a range of service increases, focused on increased frequency for some existing routes, new routes, longer hours of operation and added hours on weekends.
In that context, I think it’s worthwhile to start getting a firmer grip on current ridership levels and historical trends.
In the last 10 years, AAATA fixed-route ridership – the “regular bus,” as contrasted with paratransit services – has grown from 4.2 million in fiscal year 2004 to 6.4 million rides in 2013. That’s about 50% growth.
And ridership in FY 2013 was consistent with that continued upward trend. But the trend this year was just slightly upward: The 6.4 million fixed-route rides provided in 2013 was barely up from 6.35 million rides in 2012 – about a 0.8% increase.
Two programs that offer financial subsidies to passengers – MRide and go!pass – showed a greater increase in ridership than the most recent overall trend. Rides taken on AAATA buses through the University of Michigan’s MRide program showed an increase of about 5% between 2012 and 2013 – from 2.61 million to 2.74 million rides. And rides taken with the go!pass, funded in largest part by the Ann Arbor Downtown Development Authority – increased about 4% between 2012 and 2013 – from 604,000 to 629,000 rides.
I’m going to use the term “out-of-pocket ridership” for those rides not supported through an additional subsidy from the go!pass or MRide programs. Out-of-pocket ridership actually showed a decrease of about 3% from 2012 to 2013 – from 3.15 million to 3.04 million rides.
A more detailed look at the historical ridership data by category has implications, I think, for reasonable community expectations for future out-of-pocket ridership – if a millage is approved and service improvements are implemented. And that more detailed look also has implications for a reasonable basis for negotiations between AAATA and the University of Michigan on the financial component of the MRide deal.
Here’s a quick summary of my thoughts. I think the success of the five-year improvement plan should be measured in part by increases in out-of-pocket ridership. I think the MRide agreement between UM and the AAATA should include not just a fare component but also a component to cover UM’s “local share.” And I think the go!pass program should be conceived as a tool to help increase general out-of-pocket ridership, not just serve to benefit people who have a formal affiliation with downtown Ann Arbor.
You’ll find a more detailed analysis of these issues below the fold. This column also includes more charts – and even some maps with colored dots.
What Fuels Upward Ridership Trend: Price or Quality?
Before considering Ann Arbor’s local ridership statistics, it’s worth looking at some national numbers to check for any clear trends. Chart 1 below summarizes statistics from the American Public Transportation Association (APTA). After a somewhat downward trend for total public transportation ridership from 1990 through 1996, total ridership has shown a slight but steady upward trend. In particular, the most recent nine-year period (indicated with the vertical black bar) shows year-to-year increases since 2004 in every year but two – 2009 and 2010.
From Chart 1 above, it’s also clear that bus and heavy rail are the two major categories of public transportation. It’s easy enough to scan Chart 1 to see that heavy rail (blue) shows an increasing trend since 2004. Less clear in Chart 1, however, is the slight downward trend in ridership for bus transportation. That downward trend is more readily apparent in Chart 2 below, which plots each mode separately, instead of stacked in bars.
Even while national public transportation trends for bus ridership are slightly downward for the most recent decade, public bus ridership in Ann Arbor is strongly upward. AAATA’s ridership on fixed-route buses has grown about 50% in the last decade – from 4.2 million in fiscal year 2004 to 6.4 million rides in 2013. Chart 3 shows that upward trend and reflects a two-year dip similar to the national trend:
But that upward trend is not uniform for out-of-pocket ridership and two programs that offer subsidized fares – MRide and go!pass. The University of Michigan pays a fare on behalf of its affiliates (faculty, staff, students) through the MRide program. UM affiliates can board buses simply by swiping their M-Cards in the fare boxes on AAATA buses. The go!pass program is a benefit for employees of businesses located in the Ann Arbor Downtown Development Authority’s tax increment finance (TIF) capture district. Business owners can purchase a go!pass for their employees for $10 apiece – which allows downtown employees unlimited rides for a year. The bulk of the cost to provide go!pass rides is covered by a grant from the Ann Arbor DDA to the AAATA.
Chart 4 below presents the same data as Chart 3 – but broken down by out-of-pocket ridership (blue), go!pass rides (red) and MRide rides (yellow).
It’s plain from cursory inspection of Chart 4 that from 2004 to 2005, out-of-pocket ridership (blue bars) dropped precipitously. But that’s just a function of the fact that the MRide program launched in August 2004 – so FY 2004 included just two months worth of MRide data. [The AAATA fiscal year runs from October through September.] From that drop, it’s evident that even before the MRide program launched, many UM affiliates were already riding AAATA buses, but paying their own fares.
Chart 5 below isolates the same out-of-pocket ridership data presented in Chart 4. Looking back over the last decade, the trend for out-of-pocket ridership is flattish to slightly upward. Out-of-pocket ridership increased by 11% between 2005 to 2006 – from 2.65 million to 2.95 million. But over the next eight-year span, out-of-pocket ridership increased by a total of just 3% – from 2.95 million in 2006 to 3.04 million in 2013.
In contrast to the flattish out-of-pocket ridership trend, the number of rides taken with an additional subsidy through the go!pass program has more than doubled – from 296,855 in 2004 to 628,959 in 2013. When the same go!pass ridership data from Chart 4 is isolated, as in Chart 6 below, the increasing trend is more evident:
As a fraction of total rides taken on the AAATA, go!pass rides account for about 10%. So despite the dramatic growth in go!pass ridership, the impact of that growth on total ridership is limited.
Compared to the go!pass program, UM ridership has not grown as much percentage-wise, but has still shown robust increases. From the first full year of the program, rides taken under MRide have increased by 57% – from 1,744,718 in 2005 to 2,736,848 in 2013. The MRide data is isolated in Chart 7:
Because the number of rides taken under the MRide program is such a substantial percentage of the total rides – about 43% in 2013 – the robust growth in the MRide program also results in growth for total rides.
To summarize, even though total AAATA fixed route ridership has shown a robust increase over the last decade, not much of that growth is attributable to out-of-pocket ridership. Most of the ridership growth during that period is attributable to rides that receive an additional subsidy – through the go!pass or MRide programs.
That summary provides a competing narrative to the one that the AAATA has given in connection with a recently completed analysis of peer transit agencies. That analysis was presented to the AAATA board at its Nov. 21, 2013 meeting. The peer analysis showed an 18% higher-than-average cost per service hour than peer agencies. That comparison is presented in Chart 8:
However, the AAATA calculates that the number of rides per service hour is 50% higher than the peer median. That peer comparison is presented in Chart 9:
And the greater-than-average riders per hour leads to a 17% lower operating expense per trip than the peer group median. The operating expense per trip peer comparison is presented in Chart 10:
To summarize the peer comparison analysis, even though AAATA’s cost per service hour is higher, the number of riders is also higher, which results in a lower cost per rider than the AAATA’s peer agencies. The narrative the AAATA has offered to account for this data is: “Investment in quality service pays off in ridership.” That is:
Narrative 1: Passengers are attracted to ride AAATA buses in greater numbers – because of the quality of AAATA’s service.
But a different narrative is suggested by the historical trends in ridership growth – which is concentrated in those categories that receive a fare subsidy. That alternate narrative is this:
Narrative 2: Passengers are attracted to ride AAATA buses – because a third party is paying for a large portion of AAATA fares.
Bearing in mind those narratives, let’s look ahead to the possible approval of a new transit millage and the implementation of higher quality service – like increased frequency, and extended and added hours of service. I think it’s fair to measure success of the five-year transportation improvement program by the evidence for Narrative 1 that we might see over the course of the next five years.
The kind of evidence we should expect to see, especially if Narrative 1 is true, is an increase in AAATA’s out-of-pocket ridership.
Success Statement: Out-of-pocket ridership will increase over the period of the five-year improvement plan.
In terms of the charts I’ve reviewed so far, that translates into taller blue bars in Chart 5, which I’m repeating here to save scrolling:
More on the AAATA’s Largest Customer: UM
If growth in AAATA ridership over the last decade is largely a function of increased ridership by University of Michigan affiliates, then it makes sense to ask why the university ridership has increased over this period.
At least part of the increase appears to be a function of an increased number of university employees and students. The total number of Ann Arbor campus employees (including the hospital) has increased by 20% over the last nine years – from 35,137 in 2005 to 42,277 in 2013. Most of the added jobs have been in the university’s hospital system. The employee numbers are presented in Chart 11:
Although their numbers haven’t increased as much over the same period, the student population (including graduate students) has increased by 10% – from 38,219 in 2005 to 42,348 in 2013.
The roughly 15% increase in total number of UM affiliates compares with a 57% increase in ridership under the MRide program – from 1,744,718 in 2005 to 2,736,848 in 2013. So trips made under the MRide program have increased at a greater rate than the increase in number of UM affiliates. Ridership is plotted against university affiliate numbers in Chart 12:
In any case, as the university continues to add jobs and students, it’s reasonable to expect that rides taken on AAATA buses by university affiliates will make up an increasing percentage of total rides taken on the system. In 2013 that percentage was about 43% – up from about 37% in 2005.
One consequence of that proportionally heavy university ridership is this: The seasonal pattern of university ridership is felt in the bus system as a whole.
To see that impact, let’s start with Chart 13 below, which is a year-over-year plot of monthly ridership. It reveals the seasonal pattern among the total AAATA ridership:
The basic pattern discernible in Chart 13 is a peak ridership in September and October of each year, followed by decreases in November and December. In January, ridership levels recover and are sustained through about April, when the levels decrease and remain at a lower level through July and August.
That’s not the same seasonable pattern as the one shown in Chart 14 among go!pass holders, which drops from October to November then shows a gradual rise through the rest of the fiscal year:
Given the relatively lower percentage of trips taken by go!pass riders, it’s reasonable that the go!pass seasonal pattern does not have a discernible impact on the overall seasonal plot in Chart 13.
But the seasonable pattern among university affiliates does appear to have a clear impact on the overall seasonal pattern. That conclusion is based on comparison of the seasonal pattern for UM ridership alone, which is shown in Chart 15, with the seasonal pattern by non-UM affiliates and non-go!pass riders, which is shown in Chart 16:
Comparing Charts 15 and 16 with Chart 13, it appears that the overall seasonal ridership pattern is inherited from the seasonal pattern of university riders. That hints at the possibility that the majority of rides under the MRide program are taken by the most highly seasonal component of university affiliates – students. That’s confirmed by the actual breakdown provided to The Chronicle by the University of Michigan for one period of sampling, which is presented in Chart 17:
Policy Implications for MRide Agreement: Fund the Local Share
The seasonal nature of the impact of UM ridership and the student-dominated nature of that ridership has potential implications, I think, for how the MRide deal between the university and the AAATA should be structured.
Under the current arrangement, the university pays a fare of $1 per ride to the AAATA. [I'm going to gloss over one wrinkle involving the federal grant received by UM for operation of its own blue bus system, which the university uses to defray its cash costs to cover the $1 fares for its affiliates.]
The full cash fare for a one-off ride on the AAATA is $1.50. The rational basis for the $1 per ride that UM pays is this: Riding the bus two trips every day of a 30-day period (60 rides) would result in a total full cash fare cost of $90; but someone could just as well purchase a 30-day flex pass for $58, which is roughly 2/3 the cost of paying the full cash fare each time they board the bus. Two-thirds of $1.50 is $1. QED.
I’m not persuaded that consideration of the hypothetical rides taken by 30-day flex pass holders is the best rational basis for establishing the fare per ride paid by UM to AAATA.
But rather than focusing on the amount of the fare, I think it’s more important to frame the question in terms of the typical funding model for public transportation in America – of which fares are just one component. The other three major components are local taxes, state assistance and federal assistance.
Broadly speaking, fares cover something less than a quarter of the AAATA’s expenses. Here’s a breakdown of revenues from the approved AAATA budget for the current fiscal year:
When the AAATA contemplates arrangements to provide transportation services to other jurisdictions – like Pittsfield Township, for example – that negotiation is not about fares that the jurisdiction would pay on behalf of its residents. When the AAATA reaches an agreement with the township, the amount paid by the township covers the “local share” – the darker blue wedge of the funding pie indicated in Chart 18 under POSA (purchase of service agreement). The money paid by the township to the AAATA under a POSA is analogous to the tax that Ann Arbor property owners pay. Pittsfield Township property owners don’t pay a local transit tax like Ann Arbor residents, so the township pays a POSA instead. Pittsfield Township passengers still pay their own fares.
So when the AAATA talks to UM about fares, it’s a completely different conversation from the one the AAATA has with the township. The conversation with UM is about fares. But the conversation with Pittsfield Township is about “local share.” So how is the University of Michigan different from Pittsfield Township?
In one important respect, the University of Michigan is no different from Pittsfield Township – because no property taxes are paid by UM. If UM did pay taxes on its property – which has roughly $1 billion of estimated taxable value – the roughly 2 mill Ann Arbor transit tax would generate about $2 million annually. Based on that observation, I would be eager to consider a way to add coverage of some “local share” to the MRide agreement – so that both fares and local share would be addressed. That’s reasonable, given the typical funding model for public transportation.
On the other hand, it’s inconceivable to me that the AAATA should or would contemplate asking a church – a tax-exempt entity just like UM – to pay a “local share” in addition to whatever fares are paid, so that the cost of bus rides taken by its congregants to attend church would be covered.
To find some kind of middle ground – between UM-as-township and UM-as-church – I think it’s worth being more thoughtful about specific UM riders. Instead of thinking about “UM affiliates” as an unanalyzable block, it would be useful to think about the funding issue in terms of rider profiles.
Consider Dan, a university employee who lives within the city limits. I’d be willing to count the “local share” for Dan’s MRide trips as covered by the transit taxes he or his landlord pays. So UM could reasonably contend that only Dan’s fare needs to be a part of the MRide funding negotiation. Analyzed in the same way would be a non-city resident, Fran, who commutes to work at a facility leased by the UM from a private entity. That is, the property taxes paid by the UM’s landlord would count as covering the ”local share” of Fran’s MRide trips. Only Fran’s fares would be subject to negotiations on the MRide agreement.
And there’s nothing special about students per se when considering rider profiles in this way. It’s not important whether a UM affiliate is a student, but rather that their place of residence generates tax revenue. If UM student Stan rents a house, then the transit tax paid by his landlord can count as the “local share” for his MRide trips. Only Stan’s fare needs to be a part of the MRide agreement negotiations.
But UM students who live in dormitories should be analyzed differently, I think. If Mary lives in a UM dormitory, which doesn’t generate any tax revenue, then her trips taken under MRide don’t currently have a funded “local share.” That’s what makes UM different from a church: Churches don’t maintain large residential complexes with a population that need services. And that’s one reason why it’s reasonable to treat UM differently from a church.
It’s also reasonable to treat UM differently from a church because of the sheer volume of UM affiliates, the majority of whom are students, who ride AAATA buses. Given that the volume of UM bus riders has a significant impact on the seasonal ridership patterns overall, it’s reasonable to treat UM differently from a church.
The current agreement between AAATA and UM covering funding for MRide was ratified by the AAATA board on Sept. 16, 2010 and runs for five years, through 2015. So it’s not too early to start thinking about how the negotiations on an extension of the agreement might be framed.
Policy Point 1: The MRide agreement between the University of Michigan and AAATA should include a component to acknowledge “local share” in addition to fares.
Usefulness of Ride-Level Data: go!pass Data Illustration
To implement an MRide agreement based on the kind of profiles I’ve contemplated above would, I think, be straightforward – even if it would require some work. That work would include making sure that the relevant profile data for the basis of MRide local share funding is encoded in all UM affiliate M-Cards. The fare boxes used on AAATA buses would collect that profile data from passenger M-Cards when they board AAATA buses and swipe them through the fare box.
The GFI Genfare fare boxes used by the AAATA can collect all sorts of data, including longitude and latitude of the boarding location.
I’ll illustrate the usefulness of this kind of information with boarding data from the go!pass program – partly because that’s the data I have. But the go!pass data is interesting in its own right.
I plotted out a set of go!pass boarding locations – collected from November 2012 through October 2013. Locational data was missing from some of the boardings, but the dataset included 571,792 rides with valid lat-long coordinates. Those coordinates correspond to the location where the boarding took place. As a coarse-grained technique to collect the locations into groupings, I truncated the lat-long values at four decimal places (reduced from 8 places) and colorized the dots based on frequency of boardings. That’s the plot in Map 19 [link to dynamic map]:
First, the plot illustrates the fact that the AAATA bus service already lives up to the part of the authority’s new name – the Ann Arbor Area Transportation Authority. It’s clear that downtown employees are boarding the bus at myriad locations outside the city of Ann Arbor.
If you click through to visit the dynamic version of the map, you’ll notice that it takes several seconds to load. First to load are the boarding locations east of the intersection of Washtenaw Avenue and US-23. About 30% of go!pass boardings are located east of US-23.
Let’s assume that in most cases a boarding east of US-23 is by a non-city resident inbound toward Ann Arbor, and that the east-of-US-23 boarding is paired with a downtown Ann Arbor boarding on the same day. From those assumptions, it’s plausible to conclude that more than half (30% times two) of the bus rides taken under the go!pass program are by passengers who live outside the city of Ann Arbor and commute to work in downtown Ann Arbor.
In Map 20, the Blake Transit Center’s location emerges in the plot of boarding locations between Fourth and Fifth Avenues north of William. But clearly the BTC is not the only place people get on the bus in downtown Ann Arbor:
At an Ann Arbor Downtown Development Authority retreat on Jan. 30, 2014, board member Al McWilliams described a grand vision, which he allowed was unattainable, of Ann Arbor as a carless zone. Visitors who arrived to Ann Arbor would leave their private vehicles on the periphery and take public transportation into the city.
The dynamic described by McWilliams is already a reality in a very limited way – through the AAATA’s park-and-ride lots. Results from a recent survey of downtown commuters indicated that 8% of the people who responded said they drive part way and then take the bus the rest of the way to work. About that 8% of respondents, the report from CJI Research concludes that the point from which they take the bus is “presumably from a park and ride location.” Data from go!pass boarding swipes makes it possible to start validating that assumption.
At least some go!pass riders take advantage of officially designated park-and-ride lots. That’s illustrated in Maps 21, 22 and 23:
But the dataset I plotted suggests that there could be other locations that are potentially being used as informal park-and-ride lots by go!pass commuters. The year’s worth of boarding data showed Meijer at Carpenter and Ellsworth recorded about 8,000 boardings. And Fountain Plaza near Washtenaw Avenue and Golfside recorded more than 4,000 boardings. Neither of those locations are part of the AAATA’s park-and-ride program. High frequency boarding locations are shown in Map 24 [link to dynamic map]:
It’s also possible that those locations are not necessarily being used as informal park-and-ride lots for work commuting purposes – but rather are simply shopping destinations for some go!pass holders.
That’s a possibility that highlights the conditions for use of go!passes. Namely, they’re not just for commuting to work. Using a go!pass to run shopping errands is completely within the parameters of the program.
And that’s an opening to make a point I’ve written about before: The DDA should consider extending the benefits of go!passes more broadly than just to downtown employees. At the DDA board retreat on Jan. 30, 2014, board chair Sandi Smith floated the idea of a go!pass for downtown residents as well as employees. I count as progress the fact that any DDA board member is reflecting on what kind of transportation program the DDA would like to fund.
In a column written over a year ago, I argued for extending the go!pass benefit generally to Ann Arbor transit taxpayers – but reducing the amount of the subsidy. I won’t repeat those arguments here. But some readers might wonder why I think the DDA could legally fund activity outside the downtown district. It’s because the roughly $500,000 annual go!pass subsidy is funded out of revenue collected in the public parking system, which the DDA manages. That is, the go!pass is not funded with tax increment finance (TIF) revenue.
So from a legal perspective, if it’s legal for the DDA to turn over 17% of the gross revenue from the parking system to the city of Ann Arbor to use wherever in the city it likes – as the DDA does now under the terms of its contract with the city – then it’s legal to fund a city-wide transportation pass with public parking revenue.
Or if the DDA is committed to providing an additional transportation subsidy only to those people who have a connection to downtown, then that connection could be defined simply as “boarded the bus in the DDA district.” Imagine a go!pass distributed with every tax bill – a transportation pass for which users could purchase additional rides by using an online interface. And imagine that every month, when a user logs on to review their account, there’s a message that reads something like: “You boarded the bus 15 times in the DDA district. This month the DDA is able to pay for 10 of those rides.” That’s a program that could support not just downtown workers, but also downtown residents and downtown shoppers.
Ultimately, I think that programs similar to the go!pass could be used as a tool to help boost out-of-pocket ridership. And that’s important, because I think that’s part of the way we should measure the future success of AAATA service.
Policy Point 2: The DDA should consider using Ann Arbor public parking system revenue to fund a transportation pass with broader eligibility than just downtown employees.
A Big Pile of Data
Around this time of year, The Chronicle has historically published a transit data roundup, adding the most recent set of ridership data to the set we’ve been gradually compiling. For this column, I’ve left out Amtrak ridership – because the five-year transit improvement plan is not supposed to involve any rail-based service. And the five-year transit improvement plan is focused strictly on the immediately adjacent geographic vicinity of Ann Arbor.
And this year I’ve written the annual data roundup as a column, because data mostly does not speak for itself. But data does repeat itself, and so will I. Again, here’s what I conclude based on this big pile of data.
I think the success of the AAATA’s five-year improvement plan should be measured in part by increases in out-of-pocket ridership. I think the MRide agreement between UM and the AAATA should include not just a fare component but also a component to cover UM’s “local share.” And I think the go!pass program should be conceived as a tool to help increase general out-of-pocket ridership, not just serve to benefit people who have a formal affiliation with downtown Ann Arbor.
The Chronicle could not survive without regular voluntary subscriptions to support our coverage of public bodies like the Ann Arbor Area Transportation Authority. Click this link for details: Subscribe to The Chronicle. And if you’re already on board The Chronicle bus, please encourage your friends, neighbors and colleagues to help support The Chronicle, too!