USF Libraries
USF Digital Collections

How structures compare

MISSING IMAGE

Material Information

Title:
How structures compare a cost comparison of community transportation coordinators in Florida
Alternate Title:
Cost comparison of community transportation coordinators in Florida
Physical Description:
iii, 45 p. : ill. ; 28 cm.
Language:
English
Creator:
University of South Florida -- Center for Transportation Research
Florida -- Commission for the Transportation Disadvantaged
Publisher:
University of South Florida, Center for Urban Transportation Research
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Transportation -- Mathematical models -- Florida   ( lcsh )
Paratransit services -- Mathematical models -- Florida   ( lcsh )
Older people -- Transportation -- Mathematical models -- Florida   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Bibliography:
Includes bibliographical references.
Additional Physical Form:
Also available online.
Statement of Responsibility:
by Center for Urban Transportation Research, University of South Florida.
General Note:
"September 1993."
General Note:
"Prepared for Florida Transportation Disadvantaged Commission."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001929202
oclc - 36105093
usfldc doi - C01-00026
usfldc handle - c1.26
System ID:
SFS0032149:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader cam 2200265Ia 4500
controlfield tag 001 001929202
005 20080327130752.0
008 961218s1993 flua b s000 0 eng d
datafield ind1 7 ind2 024
subfield code a C01-00026
2 local
035
(OCoLC)36105093
040
FBA
c FBA
d OCLCQ
FHM
043
n-us-fl
049
FHMM
090
HE213.F6
b H69 1993
0 245
How structures compare :
a cost comparison of community transportation coordinators in Florida /
by Center for Urban Transportation Research, University of South Florida.
3 246
Cost comparison of community transportation coordinators in Florida
260
Tampa, Fla. :
University of South Florida, Center for Urban Transportation Research,
1993.
300
iii, 45 p. :
ill. ;
28 cm.
500
"September 1993."
"Prepared for Florida Transportation Disadvantaged Commission."
504
Includes bibliographical references.
530
Also available online.
650
Transportation
z Florida
x Mathematical models.
Paratransit services
Florida
Mathematical models.
Older people
Transportation
Florida
Mathematical models.
710
University of South Florida.
Center for Transportation Research.
Florida.
Commission for the Transportation Disadvantaged.
994
C0
FHM
1 8 773
t Center for Urban Transportation Research Publications [USF].
4 856
u http://digital.lib.usf.edu/?c1.26



PAGE 1

HOW STRUCTURES COMPARE A Cost Comparison of Community Transportation Coordinators in Florida ; I -/ / -------I I ' ------' ---------... ----. I --'.;-.--! I \ Florida Transportation Disadvantaged Commission --;.. ---!----+ -----\ I i I

PAGE 2

HOW STRUCTURES COMPARE A Cost Comparison of Community Transportation Coordinators in Florida Prepared for F l orida T ransportation D isadvantaged Commission By Center for Urban Transporta t io n Research University of South Florida, Tampa September 1993

PAGE 3

Transportation Dis advantaged Com mission 605 Suwannee Street MS-49 Tallahassee, Florida 32399-0450 (904) 488-6036 Executi ve D irecto r : Project Manager : Jo Ann Hutchinson Floyd G Webb, III Center for Urban Transportation Research University of So uth Florida 4202 E. Fowler Ave n ue, ENB 118 Tampa. Florida 33620 (813) 974-3120 Director : Project Director : Project Manager: Project Staff : Gary L. Brosch F. Ron Jones Rosemary G. Mathias R. Benjamin Gribbon Diane T Shannon Steven E Maas Rebecca Rahimi

PAGE 4

TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . ii LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . iii INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . I PART ONE: ISSUES AND METHODS USED TO COMPARE COSTS OF CTC STRUCTURES . . . . . . . . . . . 3 Study Issues . . . . . . . . . . . . . . . . . . . . . . 3 Adjustments for Fixed Route Trips . . . . . . . . . . . . . . . . 7 PART TWO: COMPARISON OF STRUCTURE OF CTCS BY COST-EFFECTIVENESS . . . . . . . . . . . . . . . . . . II Overview . . . . . . . . . . . . . . . . . . . . . . . . . II A Graphic Presentation of Expense Measures by Structure . . . . . . . . II Method I -Cost Differences Among Representative Networks . . . . . . 16 Method 2 Structure of the Least Expensive Systems . . . . . . . . . . 17 Method 3 Structure of the Most Expensive Systems . . . . . . . . . . 22 Method 4 Range of Costs by Structure . . . . . . . . . . . . 25 Method 5 -Median Costs by Structure . . . . . . . . . . . . . . 27 Method 6 -Average of Costs by Structure . . . . . . . . . . . . . 29 Method 7 Regression Analysis . . . . . . . . . . . . . . . . . 31 CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . 33 Summary of Finding s . . . . . . . . . . . . . . . . . . . 33 Final Analysis . . . . . . . . . . . . . . . . . . . . . . . 36 APPENDIX A: ABBREVIATIONS AND DEFINITIONS . . . . . . . . . 39 APPENDIX B: PERFORMANCE MEASURES . . . . . . . . . . . . . 4 I 1

PAGE 5

LIST OF TABLES TABLE I: Number of Coordinators by Structure and Environment in 1992 . . . . . 4 TABLE 2: Adjustments for Fixed-route Trips . . . . . . . . . . . . . . . 9 TABLE 3: Cost Efficiencies of Sample Networks . . . . . . . . . . . . . 16 TABLE 4A: Five Least Expensive Urban Coordinators . . . . . . . . . . . 18 TABLE 4B: Summary of Occurrences of CTC Structures in Table 4A . . . . . . 19 TABLE SA: Five Least Expen sive Rural Coordinators . . . . . . . . . . . . 20 TABLE SB: Swnmary of Occurrences ofCTC Structures in Table SA ............. 21 TABLE 6A: Five Most Expensive Urban Coordinators . . . . . . . . . . . 22 TABLE 6B: Swnmary of Occurrences of CTC Structures in Table 6A . . . . . . 23 TABLE 7 A: Five Most Expensive Rural Coordinators . . . . . . . . . . . . 24 TABLE 7B: Summary of Occurrences ofCTC Structures in Tab le 7A ............. 25 TABLE 8: Range of Costs for Urban Coordinator Structures . . . . . . . . . 26 TABLE 9: Range of Costs for Rural Coordinator Structures . . . . . . . . . 27 TABL E I 0: Median Co sts of Urban CTC Structures . . . . . . . . . . . . 28 TABLE II: Median Costs of Rural CTC Structures . . . . . . . . . . . . 28 TABLE 12: Average Costs of Urban Coordinator Structures . . . . . . . . 30 TABLE 13: Average Costs of Rural Coordinator Structures . . . . . . . . . . 30 TABLE 14: Summary of Unqualified Findings by Method . . . . . . . . . . 35 TABLE 15: Summary of Actual Findings by Method . . . . . . . . . . . . 37 11

PAGE 6

LIST O F FIG U RES FIGURE 1 : E x pense per Vehicle Mile by Structure: 1992 Urban CTCs . . . . . 13 F IGURE 2: Expense per R evenue Mile by Structure: 1992 Urban CTCs . . . . . 13 FIGURE 3: Expense pe r P asseng e r Trip by Structure: 1992 Urban CTCs . . . . 14 FIGURE 4: Expense per Vehicle Mile by Structure: 1992 Rural CTCs . . . . . . 14 F IGURE 5: E x pense pe r Revenue Mil e by Structure: 1992 Rural CTC s... . . . . 15 FIGURE 6: Expense per Passenger Trip by Structure: 1992 Rural CTCs . . . . . 15 Ill

PAGE 7

INTRODUCTION The Florida Legislature, tluough Chapter 427 of Florida Statutes, established the Transportation Disadvantaged (TD) Commission with the assigned responsibility of accomplishing coordination of transportation services provided to persons who are transportation disadvantaged. Through the Commission, an organization in each county is designated as community transportation coordinator (CTC) and charged with the responsibility of ensuring coordinated transportation services are provided to the transportation disadvantaged population. More specific responsibilities of the coordinators are established by Florida Rule 41-2, F AC.' The structure of each CTC varies. The organization can be a private non-profit entity, a private for-profit company, a government agency, or a transit agency. The coordinator's service delivery network also may vary. The coordinator may act as the sole-provider in the county, providing all of the trips itself; -it may p rovide some trips and broker others to contract transportation operators, making the CTC a partial brokerage; or the coordinator can broker all of the trips as a complete brokerage The scarcity of resources and the diversity of organization and delivery approaches has resulted in a desire to determine whether any of these approaches consistently pro vides more cost effective service. The TD Commission has contracted with the Center for Urban Transportation Research (CUTR) at the University of South Florida to determine whether a particular type of service delivery network is more cost-effective. In addition, CUTR recognized a need to determine whether a particular organization type is consistently more cost-effective. This report explores the issues involved in determining the most cost-effective CTC structure, compares the cost of different CTC structures, and presents CUTR's conclusions regarding CTC structures. A detailed explanation of the issues and the methods used follows. 1 See also the Florida Five-YeN TransJ)()rtatlon Disadvantaged Plan, prepared for the Transportation Disadvantaged Conunission by the Center for Urban Transportation Research. I

PAGE 8

This is a blank page

PAGE 9

PART ONE ISSUES AND METHODS USED TO COMPARE COSTS OF CTC STRUCTURES A number of important issues must be considered when comparing the co st of CTC structures Many of these issues, particularly those that defme the approach of this study, are discussed below. Other issues, primarily those related to interpreting the cost comparisons, are discussed later in the document as each concern arises. The actual comparison and analysis begins in Part Two. Study I ssues Definitions. A coordinator's overall structure includes two important aspects that are considered in this study. These aspects of structure are organization type and network type Organization types are categorized as private non-profit, private for-profit, government, or transit agency. Networ k types are categorized as sole source, partial brokerage, or complete brokerage. Organization type and network type correspond to the "coordinator type" and "coordinator system" reported by each CTC in annual operating reports submitted to the TD Commission. Other definitions also are consistent w ith the TD Commission reporting system and Rule 41-2, F AC > Abbreviations used in this report are listed with other definitions in Appendix A. Environment The acrual cost of paratransit operations can be influenced by a surprisingly large number of variables. Many of these factors have to do with the environment in which each entity operates, which can be categorized as either urban or rural. The most appropriate structure may be different for urban and rural environments. Because the environment seems to be a factor that affects cost-efficiency, separate conclusions should be drawn for urban and rural settings. Still, i t may be difficult to use this study's conclusions to recommended a structure for some coordinators because the separation between urban and rural e n vi ronments is somet ime s subtle. In order to consider separate operating environments, urban CTCs are defined as those that include a populat ion center of more than 50 000 in the service area. Many CTCs, however, hav e a mix of urban and rural operating environments. l See lhe Transporlatlcn Disadvantaged Co mmission ln.struclions for Completi on of A_nnual Operating Report, Form Rev. 2/24/92; Transporunion Disadvantaged Cammission (TDC) Glossary of Temrs, Rev. 3/3/92; and Rule 41-2.002, FAC, "De finitions. 3

PAGE 10

S amp le Site. I n comparing the costs of coordinato r structures, a certain set of systems must be examined. The reporting requirement s for CTCs mean that a s ignificant amount of infonnation is available on all CTCs in Florida. This has enabl e d CUTR to cons ider I 00 percent of the CTCs existing in 1992. T h e t otal number of CTCs, however, was only 48, whic h me an s the actua l sample s ize is quite s mall even though it includes all the existing CTCs. Statistically, s m aller sample siz es yield lower confidence l evels and lower levels of accuracy. Sample size becomes even more of a prob l em if the 48 CTCs are divided int o smal ler categories. To iso l ate the effect of structure on cost, the 48 CTCs must be exam ined in smaller groups, dividing them by organization type, network type and/or operating environment. Because there are so few CTC s it is not feas i ble to work with peer groups that are defined by organization type, network type, and operating environment at the same time because the data sets beco m e as small as o n e or two C T Cs, if any. Table I shows the total number of coord i nator s by structure and by environment indicating maximum sample sizes for each peer group. TABLE 1 Numbe r of Coordinaton b y S tructure and Environment in 19 92 .: . : .. : ; : ; ._-::...::>_._:. It: V-' > v Sole . Piortial Comp lete To tal' I ; ., Organlution T ype. : . ;, Sour .. llfokO
PAGE 11

Mean.s of Comparing Costs. The general approach used for comparing costs can vary. Researchers may examine available information, collect new information from all coordinators or a sample of coordinators by means of interviews or survey s, or examine several systems very closely. Data collection costs and accuracy are important factors in choosing a means of comparing cost. CUTR bas chosen to use two approaches. This study makes use of the operating and financial information that all CTCs are required to report annually, which is least costly to collect and likely to be the most accurate, as discussed below. The researchers have also examined three CTCs in more detail to provide examples and possible insight about why one structure may appear to be less expensive than another. Availability of Data. The cost of collecting data from more than one CTC and organizing that information into a form that can be used for comparison among all CTCs is high The research team therefore bas lo oked to available uniform data. The annual Statewide Operations Report (SOR) contains recent information on all CTCs. This information is comparable because the data are reported in a standard format and the TO Commission provides standard instructions and definitions. Furthermore, this information will continue to be collected regularly, allowing for the possibility of follow-up studies. In addition, CUTR collected more detailed information on three CTCs, in an effort to determine what data were readily available. No significant cost information was found in comparable formats except for the information available in the SOR. Accuracy of Data. The study depends upon the accuracy of data provided through the coordinators' annual operating reports (AORs) and the information collected at the sample sites. If each CTC bas reported data accurately in its A OR, one can be confident in conclusions based on the SOR. Accuracy could vary due to improper cost allocation poor record-keeping, or improper reporting. There is no clear evidence that any of these errors have occurred in reporting When the SOR was prepared, CUTR attempted to verify or correct suspect data by calling CTCs if subtotals did no t add up to totals, if large differences in revenue and expense existed, or if large changes in operating measures appeared. Because instruction s are provided for the completion of AORs, and because of the data collection verification process, the AOR and SOR information is the most reliable information available. Perspective in Determining Cost Determining the lowest-cost structure for coordinators begs more specific questions. What costs? Cost to whom? The cost-effectiveness measures this study examines are based on total operating expenses, which should be fully allocated. Precise ly how costs and invisible costs (such as in-kind services) are allocated can depend on perspective. For the purpose of this study, it bas been assumed that the TO Commission wishes to advise each 5

PAGE 12

CTC about which network structure may be most cost-effective, given the constraints of the local environment, so that funding from the TD Trust Fund can be most effective. It is assumed, therefore, that the information collected by the TD Commission is appropriate data for the problem posed. Cost Measures. Several standard measures provide information about the cost of a system on a per-unit basis. Per-unit information is needed to make comparisons. The Statewide Operations Report provides information for calculating three cost measures that will be used to study the cost of coordinator structures. These measures are: Expense per passenger trip; Expense per vehicle mile; and Expense per revenue mile. Expense s includes all operating expenses (including depreciation but not capital), in accor dance with the AOR Instructions and SOR calculations. Passenger trip, vehicle mile, and revenue mile are defined in Appendix A of this report. Medians, Averages, and Ranges. In examining a peer group such as all rural CTCs or all brokerages, it is necessary to look at the cost-effectiveness of the group. For this purpose, medians, averages, and ranges are considered. This study examines these statistics by grouping all CTCs with the same operating environment and structure. Examining the range of costs involves identifying the lowest and highest unit costs for each peer group, and seeing which peer group has the lowest and highest costs. A median is the middle value of a series that has been sorted in ascending or descending order. Average costs can be computed several ways. The method most useful for this study will be to compute an average based on the unit costs of each CTC, rather than using an average based on the total expenses and total units of the group. Time Period. The most recent data available on CTCs is the 1991/92 operating year (July I, 1991 -June 30, 1992), and, therefore, this is the information relied upon in this study. Examining multiple periods (e.g., the costs of service structures over several years) is another approach that could be used. The number of CTCs changes each year, however, which could result in comparing mixed sets of coordinators. One approach would be to ignore the CTCs that Spe<:i6c infonnation on how these measures are collected and calculated can be found in the /990191 & /991192 SOR and the Transportation Disadvantaged Instructions for Completion of the Annual Operations Report. 6

PAGE 13

do not have data available for each year; however, that would further reduce the number of cases examined. One of the measures examined; expehSC per revenue mile, is unavailable before 1990/91. The most recent data are likely to be the most accurate, as coordinators have had more time to become familiar with AOR reporting requirements. DaJa. The base data from the 1991/92 SOR, which this study relies on, is shown in Appendix B. Methods. As indicated above, the limited number of CTCs reduces the confidence level with respect to the complete accuracy of conclusions. There are simply too few CTCs to be statistically confident in any pattern seen in the data. Therefore, CUTR has used several methods to analyze which network and organization types are least expensive. Although some of the methods employed may be better than others, the use of multiple approaches might confirm patterns if the same conclusions are found in several methods. This srudy examines the cost effectiveness of each of the sample systems studied. The structure of the l east and most expensive coordinators in the system also are studied. The ranges, medians, and averages of specific cost measures for different peer groups also are compared. Each of these methods is described and employed in the following section. Type of Service. For this study, the cost of paratransit services without fixed-route transit trips should be examined because the use of fixed-route transit services can significantly lower costs for reasons that have little to do with the structure of the coordinator. The exclusion of the trips carried on fixed-route transit in the calculation of expense performance measures is discussed in greater detail below Adjustments for Fixed-route Trips One data comparability issue is the inclusion o f fixed-route trips in the calculation of expense measures. Of the 48 coordinators, ten report fixed-route trips that are included in the total number of trips; the cost of these trips to the coordinator is included in expenses These trips are typically very inexpensive to CTCs because their cost is only the cost of a discounted transit ticket or pass, plus administration. If fixed-route transit trips account for a portion of a CTC's total trips, the total cost per trip will appear to be lower. Fixed-route transit mileage is not included in the CTC operating statistics collected in AORs, so the fixed-route trips have little effect on a coordinator's cost per mile performance. While it is true that a CTC that coordinates many TO trips on fixed-route service may be coordinating in a cost-effective manner, it is 7

PAGE 14

difficult to compare the costs of organization and network structures if unit costs include fixed route trips. Of the ten coordinators with fixed-route trips, six include Sectio n 18 fixed-route trips.' These Section I 8 trips are more like paratransit trips, both in nature and cost, than the regular fixed-route transit TO trips that are inexpensive to coordinators The other four CTCs with fiXed route trips include Section 9 fiXed-route trips For Brevard County, many of the fixed-route trips also are more like paratransit trips. The CTCs in Dade, Duval, and Hillsborough counties include significant numbers of fixed-route transit trips in their total coordinated trips. For these three systems, it is useful to count only paratransit trips and expenses so that these systems can be included in comparisons. Clzn the fu;ed-route trips be removed from expense per trip measures? Expense per passenger trip is calculated by dividing the total expenses by the total number of trips. Using data reported in the AORs, CUTR subtracted the fiXed-route trips from the total number of trips for the three coordinators affected. By examining the reported expenses for fixed-route trips and by talking with the coordinators, CUTR can closely estimate the amount that each coordinato r pays for each fixed-route trip. This cost is $0.55 per trip for Dade $0.60 for Duval, and $0.00 for Hillsborough County's CTC. (In Hillsborough County, transit passes are given to the C T C at no cost.) Multiplying these costs by the number of fixed-route trips allows us to subtract the fixed-route purchase expense from the total expenses A new expense per passenger trip can then be calculated The administrative costs of coordinating these fixed-route TO trips cannot be estimated and removed, however, which could result in somewhat overestimated cost per trip measures. Effect of fu;ed-route adjustments. The method described above for adjusting expense information for fixed-route trips results in the changes to base information shown in Table 2. The adjustment for fixed-route TO trips alters the range, median, and simple average performance measures for two network types and three organization types. Section 18 and Section 9 refer to Seai ons 18 and 9 of the Urban Mass Transit Act, which provide feder11l funding to rural and wban areas re$peelively for public transit. Trips that arc coordinated through these transportation mechanisms are reponed sepntely in the coordinators' annual operating repons (AORs). 8

PAGE 15

Dade Duval P,U,PT C,U,PFP P,U,G $1. 52 $1.92 TABLE2 NA $1.51 $1.92 $2.04 $2.27 $1.92 NOTE : P = Partia l brokerage V = Urban PT = Public nnsit brokerage R = Rul81 PFP = Private G Go\oemment $1.95 $225 $1.92 $7.46 $12.04 $7.39 $15.90 $13.44 $7.67 rtec1 u Section 9 removed. as Seelioo 18 txed-tOUte ll'lps Vehicle reponed, ao Dade is not considered for this perfonnance:::=:-: ...... =:::.=,..-. ____________ _,_ ... Removsl of Hillsborough County bed-route bipa has no effect on cos1fmile. Hil&borough County ha& no fate a&&Oeiatsd with ftxed-routa trips, and fixed-route mileage is not oolleaed in the AOR prooeu. The effects of the adjustment on cost per mile measures is small resulting in a slightly smaller range of costs and a negligible effect on medians and averages for certain structures. The effect on cost per trip figures is more noticeable, raising the range and averages and in some cases the medians, for some structures. The adjusted numbers are appropriate for the comparison of coordinator structures. The following section presents the actual comparison of costs by structure using all the methods described in this section. 9 7

PAGE 16

This is a b l ank page

PAGE 17

PART TWO COMPARISON OF STRUCTURE OF CTCS BY COST-EFFECTIVENESS This section swnmarizes the approach described in Part One, presents a graphic look at expense measures by structure, then compares ere cost-effectiveness by structure us ing seven different methods. Overview This section compares the per unit costs of ere structures The structures examined are organization type and network type. The study includes all cres existing in the 1991/9 2 fiscal year. The unit costs compared are expense per revenue mile, expense per vehicle mile, and expense per passenger trip, based on the 1991192 SOR The CTCs in urban and rural environments are exam i ned separately. Organization type and network type also are examined separately because of the small sample sizes To help compensate for the limited sample size, multiple methods are employed. This section examines three sample CTCs, looks at the structure of the least and most expensive rural and urban systems, and compares the range, median, and average unit costs for each structure. A brief look at comparing costs via regression analysis is also included. Th e issues and methods involved in comparing the costs of CTC structures are discussed in greater detail in Part One. A Graphic Presentation of Expense Measures by Structure Before analyzing the cost-effectiveness of various structures using different methods it is useful to take a quick look at the costs we are analyzing. The following series of graphs shows the unit costs for each ere as a pair of single points. The points are clustered fust by organization type, then by network type on the same graph. There are three graphs for urban coordinators, showing expense per vehicle mile, expense per revenue mile, and expense per passenger trip. The series is repeated for rural coordinators. Each is discussed only briefly, as the actual numbers will be analyzed in more detail in the methods that follow. II

PAGE 18

Figure 1 shows expense per vehicle mile for all urban coordinators, clustered by structure. At a glance, this graph shows a number of things First, it shows that most urban CTCs are private non-profits (13) or government entities (nine); only two are private for-profit CTCs and two are public transit CTCs.' The points, which represent the cost per vehicle mile of a coordinator are for some structures clustered for some spread evenly over a range, i n some cases both. Private for-profit coordinators, for example, are clustered around $1.50 per vehicle mile. Some ranges are quite large however. Government CTCs seem to range from least expensive, at $0.27 a mile, to most expensive at $3.01 per mile, while most are clustered between $1.50 and $2 00 per mi l e. In Figure I, as well as Figures 2 and 3, each data point represents a different urban CTC. Each urban CTC is shown twice in each figure, once for organization type and one for network type The data include al l reported TD trips and operating costs, except for Section 9 fi x ed-route transit trips. Figure 2 p r esents similar infonnatio n for expenses per revenue mile. Because the measures are similar, the graph looks much the same as in Figure I. Figure 3 plots the expense per passenger trip for urban coordinators grouped by structure. Using this measure we see different structures with the highest costs. Some structures are mor e clustered and some span a greater range Figures 4, 5 and 6 present the same set of infonnation for rural CTCs I n these three figures, each data point represents a different rural CTC. Each CTC is shown twice in each figure, once for organization type and once for network type. A review of these six graphs does not reveal any obvious patterns re l ative to the cost effectiveness of different CTC structures In order to interpret the graphs, and the costs of different structures, one needs to compare lowest val ues, highest values the full range of costs or where the costs are clustered. This study examines the costs associated with each structure using each of these methods. Averages and medians are used to estimate the value at which costs are clustered. Additional graphs that combine range, medi an and average performance measures are included in Appendix B 12

PAGE 19

FIGURE 1 Expense per Vehicle Mile by Structure: 1992 Urban CTCs 0roanl%81ton Networt( 1<2.5C --------------------------li( -----------------------------------------,.---------------------* 1<2-0C ----- ----------------ii[ ---------------! jji X + :g $1.5( -----i"------------=F -------------------------------------"3t-------------.,------. -t---j!f $1.0( $0.5( Coordinators (grouped by IYJMI) FIGURE 2 Expense per Revenue Mile by Structure: 1992 Urban CTCs $3.50 X $3.00 ------------------------------------------------------------------------------+ + ;tE X !I 4'2.()0 ----------------------------)!1-----,.-------J I % A ll $1.50 -----:-----------------:. ________ _________ ___ 1 $1.()0 : ------------------------------------------------------- ---- -----------------------------------------------------------------------------Coo
PAGE 20

j > l l FIGURE3 Expense per Passenger Trip by Structure: 1992 Urban CTCs $14 ---------------------:t: -------------------------$12 X $10 _____ >(: ________ _______ ,.. _____________ ____ ____ $8 ---- :----------------------------------------------.: .-----. -------_ j ___ $8 ______ j[ __ ___ ------------------ll 'lE X f $4 ........... --------------... ....... ......... -------......................... ----... ----. ... .......... -$2 $1 FIGURE 4 Expense per Vebicle Mile by Structure: 1992 Rural CTCs NX -----.... ---..................... -----* ----1---....... ------------------------------------------ I $1 .0CII . --. ------................ --------!................... --* ------------.---------* X so.sot------------------------*"----+------'"------------------1 1"4

PAGE 21

FIGURE 5 Expens e per Revenue Mile by Structure: 1992 Rural CTCs ............ .. X I .... $1.60 .............. . .. ............. I .... $1.00 . JL ................................... "' .... .... $0.00 FIGURE 6 Expense per Passenger Trip by Structure: 1992 Rural CTCs Organization NIMW<>rk .. ...... ..................... .......... ... .. ........... ....... ............ .............. ............. X $10 ....... f ..... ................................... >..<. ..... ............. ............. .. ........... ... lii! i I I Coordlna10ra (grouped by type) IS

PAGE 22

Method 1 Cost Differences Among Representative Networks For background information, three CTCs were studied, each having a different network structure. The Manatee County Board of Commissioners serves as CTC for Manatee Count}' and represents a government sole source coord i nator; the Comsis Corporation operation in Duval County represents a private for-profit complete brokerage; and Big Bend Transit, Inc.,' represents an urban private non-profit partial brokerage. In-depth analysis of these three systems revealed two important po ints First the specific expenses and local ciR:urnstances of each coordinator vary greatly according to a number of factors, reaffirming that a particular siruc/ure can, at most be only one factor of many that determine the costs and cost-effectiveness of a CTC. Second, the visits revealed that there are many slight differences in operating practices and record-keeping. These differences mean that, on a practical level, the only comparable information availaiJle will be that which is collected regularly with detailed ins/ructions by the TD Commission. Based on the annual operating reports submitted for fiscal year 1991192, the three sample coordinators have the cost performance measures shown in Table 3 TABLE 3 Cost Efficlenties of Sa mple Ne tworks* ;( 2_, . A!-' Expense :.,.. Expe11Dper CJ'Q :: . ,. . Vehlcle ftll!a Revenue MIIe Paooenger Trip Manatee County Sol e Source $3.01 $3.16 $9.83 Govemment Comsls-Duvar Complete Brokerage $1.5 1 $225 $ 13.44 Private Fo r-p
PAGE 23

Manatee and Duval Counties axe among the most expensive in their peer groups, while Big Bend Transit is among the average and less expensive coordinators in its peer group. The next step, : . therefore, is to identify the structure of the coordinators that are least expensive. Method 2 Structure of tbe Least Expensive Systems One method for determining the most cost-effective structure is to identify the structure of the lowest cost coordinators. After examining e)(pense performance measures for all 48 CTCs, i t appeSlS that the l east e"pensive coordinator, in all three cost measures, is an urban coordinator. This coordinator is a government entity operating with a complete brokerage. The least e)(pensive rural coordinator in terms of the two cost -permile measures is also a government entity operating a complete brokerage.' In terms of cost per trip, however, the least e)(pensive rural coordinator is a government entity acting as a sole source provider. No conclusions may be drawn based on this information about the most cost-effective service delivery network. For rural SleaS, two very different network types (sole source and comp lete brokerage, depending on the cost measur e) appear to be least e)(pensive. The least e)(pensive coordinators in both environments are government entities, but there is no clear reason why these coordinators appear to be least expensive. The structure may or may not be a determining factor. Also, differences in reporting practices by one of these systems may cause one to appear to be least e"pensive. If, however the next most ine)(pensive coordinators are also e)(amined and a p3lticular network and/or organization type occurs consistently, then a relationship between structure and cost might be inferred. Therefore, the next step is to examine the five least e"pensive coordinators in rural and urban 3leaS to look for patterns Table 4A lists the five least e"pensive urban CTCs in each of three e)(pense categories, with their network and organization types shown. 6 CTC for Orange, Osceola, and Seminole counties; SOR. CTC for De Soto, Hardee, Highlands, and Okeechobee counties; SOR CTC for Columbia, Hamillon, and Suwannee counties. 17

PAGE 24

TABLE4A Five Least Expensiv e U rban Coord inators 1 Orange. Oseeola, Seminol e 2. Marion 3 Brevard 4 Volusia 1. Orange, Osceola. Seminole 2. Brevard 3. Volusla 4 OkaJoosa 5 C lay 1 Orange, Osceola, Seminole 2. Pinellas 3. Brevard 4. Volusi a 5. Bay $0.27 $0. 69 $0.75 $0 95 $1.10 $0.37 $0. 87 $1.02 $1.21 $1. 30 $2.87 $3.83 $3. 87 $4. 67 $5.12 Compl ete Brol
PAGE 25

TABLE4B io Table 4A Percentage ot:Urban COotdlnators Network Type Sole Source 2 13% 7 26% Partial Brokerage 9 60% 12 44% 4 27% 8 30% Orga ni zatio n Type Private Non-Profit 8 53% 13 48% PriVate For-Profit 0 0% 2 7% Governmen t 3 2 0% 9 33% Pub lic Tra n s i t 3 20% 3 11% percentage of coordinators. Where the shares are not similar some rel ationship between structure and cost-effectiveness is suggested The ranking of shares are compared rather than the exact percentages because the small sample sizes can vary the percentages dramatically. In thi s case, the distribution of occurrences is similar to the number of each urban network type. Of all urban systems there are more partial brokerages than complete brokerages, which there are more of than sole source providers; this is also true among the least expensive systems. In tenns of urban organization typ e s, the pattern is similar Privat e non-profit entities occur most commonly among the least expensive systems shown in Table 4A, but among urban systems there are more private non-profit entities tban government e ntities. There also are somewhat fewer public transit agencies and very few private for-profit entities. This pattern is similar to the pattern seen in the l east expensive systems P u blic transit coordinators occur slightly more often among the least expensive systems than among the total number of CTCs, but the occurrence is not enough to be significant. Because the distribution of structures among least expensive systems and all systems are similar this method suggests that CTC structure is not a factor that determines the least expensive systems for urban environments 19

PAGE 26

T a ble 5A li sts the five least expensi v e rora l CTCs in each of three e x pense cat egories, with the i r n etwork and organization type s TABLE SA Fiv e L east Rural C oordin a tors 1. De Soto, Hardee $0 .50 Com plete Brok e r Government Highlands, O keechobee 2. Di
PAGE 27

TABLESB Network So l e Source 7 46% 11 52% Partial Brokerage 4 27% 7 33% Comp l ete Bro k erage 4 27% 3 14% Private Non -Profit 11 73% 1 7 81% Government 4 27% 4 19% For rural systems, sole source coordinators occur mo s t commonly among the t op five leas t expensive systems; however, some complete or partial brokerages are more cost-effective than some of the most cost-effective so l e source CTCs. Here, too w h ile there are slightly more examples of sole sources among the most efficient coordinators, there are also more sole source providers among total rural CTCs. C omple t e b r o k e rages are somewhat more common among the least expensive systems than they are among total rural CTCs, b u t are not more common than other network types. In terms of organization type private non-p ro fit coordinators occur most freque n tly as the least expensive coord i nators. There are, however, many more private non-profit coordinators (17) than government coordinators (four) in rural areas. T he distrib u tion of rural structures is about the same among the least expensive systems and total rural CTCs. Both for ur ban and rural CTCs, the most cost-effective structures seem t o vary dependi n g on the performanc e measure; i.e., whether one looks at cost per trip, co s t per vehicle mile or cost per revenue mile. Furthermore, the distribution of structure s among the least expensive coordinato r s is similar to their distribution among al l C T C s. This method suggests that structure is not a factor rela ted to the least expensive urban or rural coordinators 21

PAGE 28

Method 3 -Structure of the Most Expensive Systems While an examinati o n of the least expensive coordinators does not indicate that any one structure is least expensive a similar examination of the most expensive coordinators could reveal a particular structure that tends to be among the most expensive. Again, the urban and rural coordinators are examined separately. Table 6A lists the five most expensiv e urban CTCs for each of three cost measures. It also shows the network and organization type of those CTCs TABLE6A Five Most Urban Coordinators ,;! T '"URBAN; '' ' ... ii"} :;c;' t ; ''I "!'.N etwork Type I .:: Cos t : : ; r ' . 1 Manatee $3 .01 Sole Source Government 2 Pasco $2.46 Partia l Brokerag e Government 3. Palm Beach $224 Complete Brokerage Government 4.Martin $2.06 Sol e Source Private Non.Profit 5 Hill sborough $1.92' Partia l Gove-rnment . 1 Manatee $3 .16 Sol e Source Government 2. Polk $ 2.73 Partia l Brokerage Govemmen l 3. Duval $225' Complete Brokerage Private ForProfit 4 Escambla $2.11 Complete Brokerage Private For-Profi t 5. Pinellas $2.10 Complete Brokerage Govemment . ,, .; .r 1. Dade" $15.90' Partial Brokerage Transit Authority 2 Duval $13.44' Complete Brokerage Private For-Profit 3 Polk $12.45 Partial Brokerage Government 4 Cottier $11.1 1 Sole Source Private Non .Profit 5 Pasco $9.87 Partia l Brokerage Government Expenses tot Dade. Ovvs l and Hilt:aborough COYnMt .,. to eXClUde fixed-route tran si t trips EJcpense per vehlc::le mile los unavdeble for Dade County. 22 .

PAGE 29

A review of Table 6A shows that all n etwork types and all organizatio n types are represented among the most expensive co6fdif!AtOrs. Although some structures appear more than others they are more common structures in gene ral. Tabl e 6B compare s the distribution of structures among the most expens i ve CTCs with that of all urban CTCs. TABLE6B Summ ary o f Occurre nc es of C T C Structur es in Tab l e 6A Network Type Sole Souroe 4 27% 7 26% Partial Brokerage 6 40% 12 44 % Complete 5 33% 8 30% OrgaoiJ;at i on Tyee Private N on-P rofit 2 13% 13 48% Privat e For-Profit 3 20% 2 7% Government 9 60% 9 3 3 % P ublic Tran sit 1 7% 3 1 1 % Table 6B shows that the distribution of network types among the mos t expensive urban systems is very sim i lar to the d istribution of network types for all urban coordinators. Comp lete brokerages, for examp l e represent one-third of the urban CTCs and one-third of the mos t expensive CTCs. This s u ggests that net work type is not a factor related to the most expe nsive coordinat o rs. With regard to o r ganization type, the table s hows that governments and private for-profits are overrepresented among the most expensive coo r di n ators. Public transit c o ordinators are repre se nted about the same in both group s Privat e non-profits are under-represe n ted among the most expensive urban CTCs, whic h suggests that they are l ess li k ely to be very expensive. The 21 rural coordinators are exami ned separately. Table 7A lists the five most exp e nsive rural CTCs in each of three expense categories, with their network and organization types shown. 23

PAGE 30

TABLE7A Five Most Expensive Rural Coordi n ators 1. Columbus. Hamilton, $1.88 Sole Source Government SliW'annee 2. Putnam $1.77 P811lal Brokerage Private Non Profit 3. C i trus $ 1 .48 So l e Source Government 4 Lafayette $1.26 Sole Source Private Non-Profit 5 Saint Johns $1.21 Sole Source Private Non Profit 5 Union $121 Source Privat e Non-Profit 1. Putnam $2.21 Pal1la l Brokerage Private Non Profit 2. Columbus. Hamilton, $1. 94 Sole Source Government Suwannee 3. FrankHn $1. 82 Pal1lal Brokerage Private Non-Profit 4 Bradfot'd $1. 62 Complete Brokerage Privat e Non-Profit 5. Union $1.60 Sol e Source Private Non Profit 1. Franklin $58.64 Pal1la l Brokerage Private Non-Profit 2. Gutr$33. 64 Pal1la l Brokerage Private Non Profit 3. Bradford $23.70 Comp lete Brokerage Private Non-Profit 4 Lafayette $20.43 Sole Source Private NorW>rofit 5 .Unton $16.78 Sol e Source Private Non-Profit R ural are all govemmeot or privata non-pn>fit organization typet.. -Stint JoMt and Union Counties both have the aame cost per ve h icle mile ; ltlere a,., t 8 obSMYtlion:t. in Tabl e 7A. ExpenM per revenue m ile fOt Gutr Col.lnty, Table 7A shows that all rural network and organization types are represented among the most expens i ve coordinators, so any structure has the potential to be expensive How representation among the most expensive rural coordinators compares to the representation of different structures among all rural CTCs can indicate any propensity for a structure to be very expensive. Table 7B sununarizes Table 7A, showing the number of times each network and organization type occurs among the most expensive systems. I t also shows the occunences as a percentage and compares that to the structure's representation among all rural CTCs. 24

PAGE 31

TABLE7B Sol e Source 9 56% 11 52% Partial Brokerage 5 31% 7 33% 2 13% 3 14% Organ i zation Private Non-Profit 13 81% 17 81% Government 3 19% 4 19% Table 7B shows that the distribution of network-and organization types an1ong the most expensive coordinators is almost exactly the same as the distribution of types among all rural coordinators. This strongly suggests that no particu lar structure is associated with the m o st expens ive rural systems In summary, a look at the most expensive systems suggests tha t private non-profit coordinators are less likely to be expensive in urban environments, and no p a rticular n e twork type is associated with the most expensive urban systems. No particular network or organization type is associated with the most expensive rural systems M et hod 4 Range of Cos ts by S tructur e Method 2 examined the structure of the least expensive coordinators and Method 3 e xam ined the structure o f the most expensive coordinators; but some structures are am ong th e l east and most expensive. These two methods do not consider the full range of unit costs exhibited by the CTCs associated with each structure. By examining the range of costs associated with each structure, it can be sh o wn how inexpensive and how expensive each structure can be, based on the least and most expen siv e CTC of each type Table 8 shows the range of costs associated with each urban structure. 25

PAGE 32

TABLES Range of Costs for Urban Coordinator Structures Sole Source (7) Partial Brokerages (12)" Private Non-Profit (13) Private For-Profit (2)" Government (9)" Public Transit (3)" $1.15-$3.01 $0.69 $2.46 $0.27 $2.24 $0.69-$2.06 $1.45-$1.51 $0.27 $3.01 $0 .75- Adjusted to exclude fixedroute trand trips, $1.21-$3.16 $0.87 $2.73 $0.37 $2.25 $1.02-$2.10 $2.11 $2.25 $0.37 $3.16 $0.87 $1.96 $5.12-$11.11 $3.87$15 .90 $2.87 $13 44 $4.67 $11.11 $8.59-$13.44 $2.87 $12 45 $3.87 $15.90 "The upper limit of this r ange is not avaiable; estimates 'ugge.t an upper limit higher than that of private for13rofit. T able 8 shows that the range with the lowest minimum values in all cost measures are for complete brokerages among urban networks, although this infonnation already was revealed using Method 2. Complete brokerages also have the lowest upper limit for cost per mile. In tenns of cost per trip, however sole sources have the lowest upper limit. This is interesting because in cost per mile measures, sole sources were consistently more expensive The network with the highest cost measure is different for each cost measure. With regard to the organization type of urban systems, government coordinators include both the least and most expensive coordinators in cost per mile measures. The private for-profit or public transit coordinators seem to have the lowest upper limit of cost per vehicle mile. The group of public transit coordinators has the lowest upper limit of costs per revenue mile, while the group of private non-profit coordinators has the l owest upper limit of cost per trip. Examining the range of costs produces mixed results for urban coordinators Similar infonnation can be examined for the structure of coordinators in rural areas, as shown in Table 9. 26

PAGE 33

Sole Source (11) $0 .78$1.88 $0.92$1.94 $4.05 $20.43 Partial Brokerages (7) $0.75$1.77 $0.86 $2 .21 $4.70 $58.64 $0.50 $0.86 $0.67 $1.62 $6.50 -$23.70 Private Non-Profit (17) $0.53 s1 n $0.86 $2.21 $4.70 $56.64 Government (4) $0.50 $1.66 $0.67$1.94 $4.05 $6.73 In rural areas, the range of complete brokerages has the lowest minimum and the lowest maximum cost per mile measures and the range of sole source networks has the lowest minimum and the lowest maximum cost per trip measures. For organization type, the lower limit of the range of costs for government entities is the lowest in all three measures. The range of government entities also has the lowest maximum cost per trip and cost per revenue mile while the set of private non-profit coordinators has the lowest maximwn cost per vehicle mile. The highest costs are found among par1ial brokerages and among private non-profits in terms of cost per revenue mile and cost per trip but the sole source and government coordinators include the highest cost per vehicle mile. Clearly, there are several ways to interp ret these ranges. Extremes in a range caused by unusual systems, inaccurate data, or inappropriate use of fixed-route information may cause one structure to appear to have the potential for being least expensive. The median, however is not affected by those extremes. Method 5 Median Costs by Structure This method of comparing the expenses of coordinator structures considers all of the coordinators in each peer group, rather than only the least and most expensive. By examining median performance measures, the costs of the "typical" CTC for each structure can be compared The median performance measures for each structure, after adjusting numbers to exclude fixed route transit trips, are shown in Table I 0. 27

PAGE 34

TABLE 10 Median Costs of Urban CTC Stnu:tures Sole Sources 7 $1.39 $1.72 $ 8.71 Partial Brokerages 12 $1.31 $1.47 s 8 .7 8 8 $1.48 $ 1.86 $7.63 Private No nProfit 13 $1.30 $1.42 $6.97 Private For-Profit 2 $1.48 $2 18 $11. 0 2 Government 9 $1.88 $1.94 $ 7 .67 Public 3 nos$ 1.59 $ 9.10 Adjusted to exeludt transil trips, -The median could not be computed because onty two CTCs reported veh iCle mile; estimates suggest a med ian of $1 .42. In urban areas, among the "typical" coordinators for each structure, a private non-profit organization type and partial brokerage network type are the l east expensive, both on a per trip and per mile basis. The difference in medians is not dramatic, however. The median costs of rural coordinator structures are shown in Table II. TABLE 11 Median Costs of Rural CTC Structures ' .1992;M8.<11an P.erfoimance Meaaures " '< "' ,., < Ex'*'". per .. Expense per V:ehiCI4!;,N,IUe Passenger Trip . Sole Source 11 $1.03 $1.18 $ 8.73 Partial Brokerages 7 $1. 11 $ 1.1 7 $9.72 Complete Brokerages 3 $0.53 $1.12 $14.38 Private Non-Profit 17 $1.02 $1.18 $9.72 Government 4 $1.22 $1.28 s 6.23 In rural systems, the stru cture with the least expens ive median depends on the performance measure In examining cost per mile, the complete brokerage and private non-profit median coordinators appear to be least expensive. Based on cost per trip, however, the sole source and 28

PAGE 35

government median coordinators appear to be least expensive. Furthermore, the structures that are least expensive by one measure are most expensive by another measure. The comparison of median costs suggests that private non-profit organization types and partial brokerage networks tend to be slightly less expensive among urban coordinators. but does not suggest a particular structure for rural environments. The median does not give any weight to the fact that some structures might have a potential for extreme costs (high or low). A compromise approach is to look at the average costs for each structure. Method 6-Average of Costs by Structure The average cost for each structure may be obtained by adding the cost performance measures for all the CTCs of the same structure type and dividing by the number of CTCs This average treats all CTCs equally and does not give larger CTCs a greater influence on the average cost for that structure. Unlike the median values, averaging costs does give some weight to extreme values but does not look specifically at the extreme values as the range does. This method of averaging is not used to yield the aggregate cost performance measure for each trip or mile delivered by that group of CTCs. It does, however, give a cost performance measure for a hypothetical "average" CTC of a group. Using averages is, in effect, a compromise between examining ranges and medians. The average of urban CTC cost measures, computed by structure, are compiled in Table 12. Based on these averages, in urban counties partial brokerages have the best cost per mile performance, while complete brokerages have the best cost per trip performance. In terms of organization type, public transit coordinators seem to have the lowest costs on a cost per revenue mile basis, while private non-p rofits have the lowest costs on a per trip basis. These advantages are not dramatic, however, as the costs of the organizat io n types are very close in some cases. The private organizations have almost the same average cost per trip as the government agencies, and almost the same cost per revenue mile as public transit agencies. 29

PAGE 36

TABLE12 Average Costs of Urban Coordinator Structures -'j ' '.. . ;EXpenSe .,.,,,, ' T . .' .. -ngerL nP ., Sole Source (7) $1,67 $1.89 $8.59 Partial Brokerages (12)" $1.40 $1.60 $7 .91 Complete Brokerages (8) $1.46 $1.67 $7.50 Private Non-Profit (13) $1.28 $1.48 $7.45 Private For-Profit (2)" $1.46 $2.18 $11 .02 Government (9)" $1.87 $1.97 $7 .46 Public Transit (3) $1.47 $9.62 Adjusted to exclude fi:xed40ute transit trips. This was not computed because only two coordinatots reported vehicle m ile$; no organization has been marked In bcd as lest expensive without this figure. The average costs for rural CTCs, according to structure and performance measure, are shown in Table 13. Among Mal networks, the set of complete brokerages have the lowest average cost per mile, while the set of sole source netwo rks have the lowest average cost per trip. With regard to organization type, the indication is also split. In terms of cost per vehicle mile private non profits have the lower average, while the government coordinators have the lower cost per trip. Note that the average cost per revenue mile is nearly the same for the two organization types. TABLE 13 Average Costs of Rural Coordinator Structures Sole Source (11) $1.16 $1 32 $8.68 Partial Brokerages (7) $1.13 $1.37 $18.n Complete Brokerages (3) $0-63 $1.14 $14.86 Private Non-Profit (17) $1-04 $1.31 $14.60 Government ( 4) $1.21 $1.211 $5.81 30

PAGE 37

An evaluation of average per unit costs for each type of structure provides complex answers The cheapest structure depends on which measur e is used Thus no structure appears to have an obvious advantage. a . Method 7 -Regression Analysis CUTR also considered using multi-variate regression analysis to analyze costs The basic thrust of regression modelling is to explain as much of the variation in the dependent variab le (e.g., cost) as possible by the explanatory variables in a regression model. In addition to the qualitative variables for structure, regression analysis requires several appropriate quantitative variables. The regression model specification proposed was based a standard cost function: where cost (C) is a function of the price of rolling stock (R), labor {L), fuel (F) and output as measured in vehicle miles (Q!. Miscellaneous costs that are unique to each ere would be captured in the error term of the regression equation To determine whether network or organizational structure has any significance for total cost of a ere, the dependent variab le, total cost, is regressed on the aforementioned factors along with qualitative variables for network and organizational type. A qualitative variable is used to indicate the presence or absence of an attribute or quality, such as "1" for urban "0" for rural. Statistical significance of the qualitative variab l e is represented by a t-statistic," where t-values that exceed a critical value would indicate that structure is important in explaining the variation in total cost. The validity of any regression model relies on the collection and estimation of the appropriate data Because eres often contract for transportation services, many costs are included in the category of "purchased tranSPOrtation" as an aggregate number, and thus, the allocation among the cost factors necessary to estimate this model are no t available. That is some of the labor and fuel costs of each coordinated system are inseparable from purchased transportation. Estimates from a model that is incorrectly specified may exhibit SPUrious correlation among the variables and result in serious misinterpretation of the underlying relationship. Other variables from the AORs could not be substituted because they were inappropriate, qualitative, or not comparable. The exploration of other models and variables reveals similar problems with variables or the size of study sets. This is not to say, however, that 31

PAGE 38

changes in annual reporting methods would necessarily allow multi-variate regression analysis to identify one structure as clearly most cost-effective. This method suggests that the least expensive structure cannot be determined based on the information that CTCs normally collect and report. The following section swnmarizes the results of all seven methods. 32

PAGE 39

CONCLUSIONS No single method is best able to identify a coordinator structure that is the most cost effective. Instead, several methods were used to see if one structure could be identified consistently as cost-effective The methods included examining: sample systems; the least expensive systems; the most expensive systems; the range, median, and average costs of each structure; and regression analysis. Each method considered three measures of cost-effectiveness and evaluated urban and rural systems independently. As described below, different methods gave different results. Summary of Findings Method I examined the cost-effectiveness of three representative systems, using per unit cost measures and on-site visits. Of the three systems, the coordinators with sole source or partial brokerage networks appeared, at first, to be the most cost-effective. The case studies demonstrated that some operating data may not be comparable among coordinators. More importantly, however, these case studies demonstrated the complexity of the question. Site visits showed that many other operating factors strongly influence the actual cost-effectiveness of individual systems. Method 2 examined the least expensive coordinators to determine their structures At first it revealed that the urban CTC that was most cost-effective was a government-operated complete brokerage. The most cost-effective rural CTC was a government-operated sole source or complete brokerage, depending on the cost measure used. The five least expensive CTCs in each cost measure and environment included almost all structures. The distribution of the occurrences of each structure was similar to the distribution of structures among all CTCs, suggesting that costs are, in fact, independent of structure. Method 3 examined the most expensive coordinators to determine their structure. Although there are more private non-profit organizations than other types in urban environments, they rarely appeared among the most expensive systems. This suggests that private non-profits are unlikely to be very expensive in urban environments. The distribution of the occurrences of each structure otherwise were similar to the distribution of structures among all CTCs, suggesting that 33

PAGE 40

costs are, in fact, independent of network type in urban environments and independent of any structure i n rural environments Method 4 looked at the entire range of unit costs for each structure to see which had the lowest upper-end cost. Method 4 results differed according to the cost measure and environment but suggested that, for urban systems, complete brokerages or sole source networks operated by public transit or private non-profit organizations have the lowest maximum costs. For rural systems, this method suggested that a government complete brokerage or sole source network has the lowest maximum costs. Method 5 looked at the median costs for each structure. This method suggested that the most cost-effective s tructure for urban systems is a partial brokerage ope r ated by a private non profit organization For rural systems, this method did not suggest a particular organization type is least expensive for rural coordinators. Method 6 examined the average costs for each structure Because average costs are similar to median costs, the results were nearly the same as for Method 5. Method 6 however, also suggested comp l ete brokerages as one of the most cost-effective network types in urban environments. Graphs that show range, average, and median performance measures combined are presented in Appendix B. Method 7 explored the use o f regression analysis to identify the least expensive structures, and found that information collected and reported by CTCs was not designed to provide enough quantitative and comparable detail to perform this analysis. There are additional methods that could be employed to evaluate relative cost effectiveness among different types of CTCs, but they require data that is not cWTently collected by the CTCs. Given the multiple approaches used in this study, there is no reason to believe that the use of additional methods would result in a different conclusion. The overall approach of this study has been to employ multiple methods and look for consistencies in the results The structure that each method would suggest as least expensive, without interpretation, is shown in Table 14 by cost measure. 34

PAGE 41

TABLE 14 Summary of Unqualified Findings by Method 1. Representative SVstems The structure of the least expensive of the systems that were examined. 2a. Least Expensive Structure of the least expensive of all coord i na t ors. 2b. Least Expensive The most oommon strudllre among the top five least expenslve coordinators. 3 Most Expensive The most common structure among the top five m0$1 expensive coordinators. 4 Ranae Structure with the lowest upper limit when CTCs are grouped by structure. 5 Median Structure with the lowest median costs when CTCs are grouped by structure. 6. Structure with the lowest average (total per costs divided by CTCs in group) 7. Regression Structure with a strong correlation to low total costs. Vehicle Mile R evenue M i l e P,PNP P,PNP C,G C G e.G Vehicle Mile CorP, G or PNP Revenue Mile P PNP Passenger Trip P PNP P PT Vehicle M ile S or P G Revenue MUe C. G Passenger Trip P, G Combined ... Vehicle MiJe Revenue M i le Trip Vehicle MUe Revenue Mile Vehicle M i le Revenue Mile P, PNP C.PT C.PT S PNP P PT P.PNP P.PNP P,PT P, PT C.PNP NA NOTE: P Panial btoke1'8Q8 U = Urbtn PT Public transit C Complete I:WC*erage R : Rural PfP .. Privl forproit See .Awendix A i>r key to abbr$\lletiOns of ooetdinator s1l'Uctures. No Nni CTCs were examined as reprnenta:IMI $)'$1tmt 8aMd on ratio ol octurT'OnCOS to total numbet or C TC.s. -Data not available. 35 NANA NA C,G C,G S G S ,PNP C,PNP S orP, PNP e.G S,PNP S or P. PNP S or P PNP G G NA

PAGE 42

This table demonstrates the variety of conclusions that might be drawn at fust glance. The table shows that the most cost-effective structure suggested varies depending upon both the method and measure employed. Furthermore, nearly all structures are represented among the suggested types. There are no major consistencies or patterns. A study of Table 14 does show some patterns although not strong. Private for-profit networks do not occur as least expensive using any of the methods and measures. This is not surprising or conclusive, however, given that there are only two such coordinators in the Florida coordinated transportation system. Other structures occur infrequently. In urban environments, sole source coordinators are suggested by fewer methods than brokerages. In rural environments, partial brokerages are suggested by fewer methods than other networks. But otherwise every organization type and network type is suggested as potentially least expensive, prior to analysis and interpretation. Final Analysis Each method employed did require analysis and interpretation to draw conclusions. Table 15 sununarizes these conclusions. When the initial results of each method are artalyzed, several methods suggest independently that no particular structure is a primary factor affecting cost-effectiveness. For urban systems, private non-profit and public transit organization types are suggested as least expensive by some methods, while others indicate that the apparent cost-effectiveness is not related to structure. Given this inconsistency and based on the study approach used, it must be concluded that no particular organization type is clearly l east expensive for urban systems. All three network types are suggested as potentially least cost-effective, suggesting that cost effectiveness is not really predictable based on network type for urban environments. For rural coordinators, only one method suggested any particular structures as least expensive which does not establish any consistency. Furthennore, that method identifies multiple structures as potentially most cost-effective All other methods applied to rural coordinators find that no particular structure is least expensive. For both rural and urban coordinators, all structures include examples of expensive and inexpensive CTCs. 36

PAGE 43

TABLE lS Summary of Actual Findings by Method 1 Representative Systems 2. Least 3. Most Expensive 4. Range 5&6. Median and Average 7. Regression a. Stl\lcture Is only one factor affecting the cost-eflectiveness. b. Differences in increase the of a. Cost-effectiveness Is of structure. a. Private non-profit organization is a. Cost-effectiveness is independent least of stl\leture b Cool-et!ectiveness is independent of network a. or sohl-source networks are least b. Public transit or private non-profit is lea-st b. Government organizations may be more cost-effective. a. Partial or complete brokerage is a. No structure is obviously least expensive. b. Private non--profit organization is least expensive. a. Inconclusive with available data. The approach of the study was to search for consistent results across multiple methods ; the inconsistency in results suggests that no one structure is inherently least expensive overall In addition, several methods also reach this conclusion independently. While it is unrealistic to conclude that structure has no effect on cost, it has become apparent that structure is at most only one factor which affects cost, and it is not at all the most important. Furthennore, which structure is most advantageous depends on local circumstances. 37

PAGE 44

This is a blank page

PAGE 45

APPENDIX A ABBREVIATIONS AND DEFINITIONS Abbreviations The following abbreviations are used in the text. AOR CTC CUTR FAC SOR TD Annual Operating Report Community Transportation Coordinator Center for Urban Transportation Research Florida Administrative Code Statewide Operations Report Transportation Disadvantaged The following abbreviations are used to describe the structure of coordinators. Environment R u Network Tvpe c p s Organization Type G PFP PNP PT Rural operating Urban operating Complete brokerage Partial brokerage Sole source Government Private for-profit Private non-profit Public transit agency 39

PAGE 46

Definitions Complete Brokerage: System in which the CTC does not provide any of the on-street transportation services itself, but contracts with transportation operators to provide all on street transportation disadvantaged services. Partial Brokerage: System in which the ere provides som e of the on-street transportation services and contracts with one or more other transportation operators to provide the remainder of the on-street transportation disadvantaged services. Passenger Trips: The number of one-way passenger trips actually provided through the CTC s coordinated system during this reporting period A one way passenger trip is defined as a unit of service provided each time a passe n ger enters the vehicle, is transported, then exits the vehicle. (If the passenger travels from home to the doctor, then to a store then home, the total number of one-way passenger trips would be three.) Passenger trips includes both those within and outside the service area. Revenue Miles: The total of School Bus Revenue Miles and ere and Other Transportation Operator Revenue Miles. These figures are unduplicated mileage counts and are also referred to as Loaded Miles. (If 10 passengers rode 10 miles, there would be 10 revenue miles.) Bus Revenue Miles: The total number of mil es driven while TO passengers are actually on the school bu s CTC and Other Contract Operator Revenue Miles: The total number of miles driven while T O passengers are actually riding on the vehicles. Section 9 fixed route/fixed schedule revenue miles are excluded. Sole Source: System in which the CTC provides all the on-street transportation disadvantaged serv ice. Vehicle Miles: The total of School Bus Vehicle Miles and ere and Other Transportation Operator Vehicle Miles. &boo! Bus Vehicle Miles: The total number of miles driven by school buses while providing transportation services to TO passengers under a school bus utilization agreement; this includes deadhead, maintenance, and non-revenue mil es. CTC and Other Transportation Operator Miles: The total number of m il es driven by the CTC and all transportation operators under contract with the CTC, while involved in the transport ofTD passengers; this includes deadhead, maintenance and non-revenue mi l es, but excludes "Charter" vehicle miles. Section 9 fixed-route/fixed schedule vehicle miles are excluded. Only purchased service vehicle miles that are attributable to TD services are included. Transportation D4advantaged tnstnJCtions for Completion of Annual Operating &port, Form Rev. 2f24 1 92 40

PAGE 47

APPENDIXB . PERFORMANCE MEASURES The following table shows the performance measures on which much of this study is based. These measures, as well as this table, are extracted from the 1991-92 Statewide Operations Report, adjusted slightly to exclude Section 9 fixed-route transit trips, as explained in Part One. The table is followed by six graphs The range, median, and simple average of all three cost measures expense per vehicle mile, expense per revenue mile, and expense per passenger mile are shown on the graphs separately for urban and rural structures. 41

PAGE 48

u PNP R G u PT c '} A c R c u R u PNP u R u : R Ft R PNP R u G u PNP ,, s u PNP R R PNP I) .... . u G p Q > u u G u G R PHI' R PNP $1.1$ NR :-"}A;t ;5':1,'S1.3t ... .' $1.26 $0. 1 5 . suo s;.o2 $0.91 $1.4{ s1.o3 .. $ l > 28 ". S0.75 < '-"> $2.00 1\fa .-.A S0.78 S1.15 S0.27 : : $2'@.': $2. .. SU2 $U4 .,.. St. 77 St.21 suo $1 ( M ''52.0.. -'_;-; tlf.e'T . .. _,... S t.12 ; $W'' u.n Sf.09 $1.82 $1.62. $1.34 NR NR $U2 $0.92 S1. ,_. ... $1 .60, ; C brokttag.t R R.urat PFP Pttvlle for.proftt G GO'IIenwnent Adjusled to exclude Section 9 nnlil bips AI': (no 1n2 AOR WM tho QNnty) HR: AOR was submitted, bUt i'ldNiddl data irem was 42 ' 110.92 S4.70 15,95 $6.21 $9.22 ,,. $4.87 $2 .87 $7.02 sa.et $3.83 $ 12 .45 $7.8% $5.29 .....

PAGE 50

Range of Costs fo r Urban CTC Structures Expense per Passenger Trip* :::: : . .. \ .,:: ... .. ;: ; : . .-., . . . :::: ; : .;, :. : : .. : T .. . . . . $10.00 -::: 4 : : ............ .: :: .... ... . . . . $&.00 : . . :. : ,. : :: . .. ': . .. : . . : .... . . . $&.00 .... .... ... ; . ..... .. .... 00 _,$$. t2 . . . . . ........ $11:87 ... . ... ... ... $3.47 . $3.57 $2.00 H.t7 . . . . s p c PNP PFP G PT Number 7 12 1 3 Median $8.71 $&.78 $7.&3 $& ,97 $11.02 $7,$1 $0.10 Simple Average ..... $7.91 $7.$5 $7.45 $11.02 $7.4a $0.&2 Network Structure Organization Adjusted to exc lude ttxed route tn!H; .. SSot provl'der, PzPattial Brokerage, C=ComJ*te Brokera ge. PHP Private Non-Profit. PFPPrivete For-P ro fit GGovemment PT Public: T r a n sit I High I Low Med i a n X Average Range of Costs for Rural CTC Structures Number Median Expense per Vehicle Mile ..... r-..-..--------,-----------,n,r-; 1. St.17 11 .77 . $ 1 .50 ... . .... . ''00 : .... ... 'io:eo .0.75 : I so&o $ 0 .50 $0.83 $0.50 so.oo s p c PNP G tt 7 17 $1.03 1 1 '11 $0.,3 11.02 11.22 Si.mple Av.raga $1.10 11.13 ... $1,04 $1.21 Network Structure Organization S Sole provider, PcPartil.l B r okerage, C Complett PNP PriYate Non-Profit G GovemrMnt 44 I High I low Median X Simple Average

PAGE 51

Range of Costs for Rural CTC Structures Expense per Revenue ..... r.--. ... "". -..... ,. ---,..,.,-. ""::-. :"' ."'" . -$2-.2-,----,., ... 12.00 . . . "" .. .. :. . ,:::_ ...... : $1,00 . ... ...... . ' . .... $0.84 so.so .. . . i : :\, .... ..... : : . .... . .... s9 ,o7_. . . :. . : ., .. ,. , ; . :: ..... ,. ,:" $0.00 .. s p c PNP G Numbr 7 17 Median $1.18 $1.17 suz sua $1.28 Simpl Average $1.37 $1.14 $ 1 31 $1.29 Number Median Network Structure Organization Expense per Passenger Trip 170.00 r----.. C". """' .. :.:. ----:--.. ,-. ------, . .. ...... . . . .. .... ... . . . . :. S&O.OO ; ;::' : -. . : $40.00 . ', : .': .J :'. : .. : .. : : : . : ... -.... ;.. ... : : m .oo . ': -. . . . . . . . . . . . . . . . . . . $20.00 ... $10.00 . . .. : : : : : .: ... : .... . ....... ......... . $0.73 .. o .... : r. .. I . ... . .. .. . ... .... . . . . ........ . . . . . :::M:: ..... S P C PNP G $8.23 S i rnp l t Av1ragt 11 te.73 s.e, ea 7 $a.72 $18.77 3 $14.35 S14, M 1 7 $0.72 $14.00 1$,81 Network Structure Organization S=Sole provider P PartlaJ Brokerage, C Compltte 8rok.ef8ige. PNPPriv&te NonProflt, 45 I High I t.ow Median X Simple Average I High I Low Median X S l mpl t Average

PAGE 52

CUI A Center for Urban Transportation Research College of Engineering University of South Florida Tampa, Florida