Click for next page ( 75


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 74
Chapter 3 Market Analysis 3.2 Geographic Markets Car-sharing is overwhelmingly concentrated in the cores of the largest metropolitan regions. In the United States in 2003, 94% of membership was concentrated in eight metropolitan regions San Francisco, Los Angeles, San Diego, Portland, Seattle, Boston, New York, and Washington, DC (Shaheen, Schwartz & Wipyewski, 2004). The same picture, although to a lesser extent, is true in Canada and in Europe. While car-sharing operates in some smaller communities such as Aspen, CO, in others such as Halifax, Nova Scotia the organization has been forced to close down. In Traverse City, MI, the 20-member formal car-sharing program ended in June 2002 after two and a half years, primarily because sufficient volunteer labor could no longer be found, and the program was not large enough to support paid staff. Note that in this section, the following terms are used: Pod a location with one or more car-sharing vehicles Pod neighborhood (or pod area) the area within 1/2 mile of a car- sharing pod Current Market Settings A range of studies have identified several common neighborhood characteris- tics necessary for car-sharing to succeed (Muheim & Partner, 1998; Klintman, 1998; Brook, 1999, 2004; Bonsall, 2002; Meaton, 2003). These include: Parking pressures. Car ownership is more expensive and less con- venient in places where parking is scarce, making car-sharing a relatively more attractive option. If residents have to walk a block or two to their car, they may as well walk the same distance to a car-sharing location. Ability to live without a car. Car-sharing is not designed to meet a household's entire mobility needs, but to work in concert with other modes such as transit (see Chapter 2). The availability of good public transportation is therefore key, along with local shop- ping opportunities and a pedestrian and bicycle network. High density. Density has two major impacts on the viability of car-sharing. Firstly, it means that there is a larger customer base within walking distance of each car-sharing vehicle; doubling the density will double the number of potential customers for a given vehicle. Secondly, it means that these potential customers will have a higher propensity to join, since dense neighborhoods have lower rates of vehicle ownership and travel (Exhibit 3-10). This is partly due to the effects of density itself, since the higher the den- Page September 2005 3-26

OCR for page 74
Car-Sharing: Where and How It Succeeds sity, the greater the number of nearby destinations and the shorter the trips; and partly because density correlates strongly with other factors, such as the availability of local shopping, parking costs and the pedestrian environment. Mix of uses. Business members have been shown to have an important role in increasing utilization rates and evening out the demand cycle, since they tend to use the cars during the working day. In contrast, people using car-sharing for personal trips have a peak demand in the evenings and at weekends. The potential for this pairing of user groups with different demand patterns is greatest in mixed-use neighborhoods, where car-sharing can at- tract both business and individual members. Exhibit 3-10 Density vs. Household Vehicle Ownership San Francisco Los Angeles Chicago Source: Holtzclaw et al. (2002). A similar curve is found when plotting density against vehicle travel (vehicle miles traveled per capita) These factors are highly intercorrelated. Parking, for example, tends to be scarce in dense, mixed-use neighborhoods with good transit, while density is one of the most important factors determining the viability of high-fre- quency, high-speed transit. Other Market Settings These types of urban neighborhoods dense, mixed-use with scarce parking and good transit appear to offer the best potential for car-sharing. However, there are also other types of market setting where car-sharing has been in- troduced and appears to be viable. Three types are discussed in this section: university campuses; apartment buildings; and small towns and villages. Page 3-27

OCR for page 74
Chapter 3 Market Analysis Several potential future markets have also been suggested, such as national parks, military bases and other settings where land use and transportation decisions are controlled by a single entity. University Campuses University campuses have been one of the most fertile environments for car-sharing. They tend to have constrained parking and a highly educated community with many "early adopters" who have a desire to reduce their impact on the environment. Many campuses have requirements that parking and transportation services be self-funding through parking fees and fines and other user charges, which means that they are more likely to need to explore aggressive Transportation Demand Management (TDM) programs, including car-sharing (see, for example, Toor & Havlick, 2004). Many campuses are situated in urban centers and can be considered part of the "core" urban market for car-sharing even though they may have devel- oped partnership arrangements with a car-sharing operator (see Chapter 5). For example, Massachusetts Institute of Technology in Boston, the University of California-San Francisco, and the University of Washington-Seattle are all located in urban centers that share the basic characteristics for car-sharing viability good public transportation, high density, mixed uses and park- ing scarcity. In many cases, vehicles are likely to serve users from both the campus itself and surrounding neighborhoods. In other cases, however, campus car-sharing operates in more geographically isolated contexts, outside of the urban core. Examples include: Stanford University, CA Princeton University, NJ University of North Carolina-Chapel Hill In addition, several other campuses, while located in major metropolitan areas, are geographically separated from surrounding high-density neigh- borhoods. Examples include the University of California-Los Angeles and the University of British Columbia-Vancouver. Apartment Buildings Developers in many cities have sought to partner with car-sharing organi- zations, for a variety of reasons including parking management and pro- viding an amenity to tenants (Chapter 5). In most cases, the cars are part of the operator's regular network and function as part of the core network. Page September 2005 3-28

OCR for page 74
Car-Sharing: Where and How It Succeeds For example, Zipcar has a vehicle at the Market Commons development in Clarendon, VA, which is located on-street (albeit on a private road), and accessible to all members. City CarShare's vehicle in the 8th and Howard apartments in San Francisco is located in the apartment building's garage, but is open to all members. Other vehicle locations, however, rely on members drawn from the apart- ment building itself, and are closed to other members. This means that the neighborhood characteristics are less important although factors such as public transportation still play an important role. For example, many of Viacar's vehicles in Detroit apartment complexes are available for the buildings' tenants only. Small Towns and Villages While urban areas may offer greater potential, car-sharing programs have also been introduced in smaller cities and more rural areas. Examples include British Columbia, where the Cooperative Auto Network has vehicles in small towns in the Vancouver region, and Rutledge, MO, where the Dancing Rab- bit Vehicle Cooperative is part of an "ecovillage" development. Europe provides even more examples: Switzerland, Austria, Germany and the Netherlands all have car-sharing programs in rural areas. In Austria, for example, villages with a population as low as 1,000 people are served (Koch, 2002); in Sweden rural car-sharing cooperatives serve towns of a similar size, such as Frnebo. In the UK, the UK Countryside Agency has funded pilot projects in 13 areas (see, for example, CarPlus, 2004; The Countryside Agency, 2004). Car-sharing has also been established in many small cities, such as Aspen, CO and Kitchener, ON. While these operate at a different scale compared to major metropolitan operations, they share many of the same characteristics such as the availability of good public transportation and local services. Small-town and village car-sharing appears to be characterized by a high degree of personal involvement by the members. In some cases, this is pro- vided by volunteers, such as at the Dancing Rabbit ecovillage, or in Traverse City, MI where the withdrawal of the volunteers led the program to close. According to studies in Britain, the presence of a strong local champion is more important in making rural car-sharing feasible than factors such as good public transportation (Meaton, 2003). Page 3-29

OCR for page 74
Chapter 3 Market Analysis Other programs, however, have had success through sharing administration with a "parent" car-sharing organization. The Cooperative Auto Network has five rural locations in Tofino, Nanaimo, Courtenay, Cortez and Whistler, operated through its Vancouver headquarters. It will place cars anywhere that 16 "committed pioneers" are willing to both purchase shares in the co- operative, and actively pursue other members. A similar approach is used by Mobility Switzerland. It will open a new location where 20 members are already signed up, and where at least five new customers can be recruited during the first year. Other criteria include the availability of reasonably priced parking, proximity to transit, and good lighting for personal security (Mobility Switzerland, 2004). Analysis of Existing Locations The studies discussed in the previous section were largely qualitative in nature, assessing the broad characteristics of neighborhoods with car-shar- ing. This section provides more quantitative data on the market settings for car-sharing, through an analysis of census data. These detailed neighborhood characteristics are critical to the success of car-sharing, not least since the distance of a car-sharing pod from members' homes is strongly correlated both with the propensity to use car-sharing (Katzev, Brook & Nice, 2000), and with member satisfaction. This satisfaction related to distance from a pod covers not only convenience, but surprisingly also reliability, car avail- ability, ease of use and cleanliness (Lane, 2004). Use of Census Data: An Example from Madison Census data have been used by many operators in determining where to locate new pods, and the feasibility of starting service in a particular city. For example, in Madison, WI the car-sharing feasibility study used this source to determine which neighborhoods to take forward for a market study (Grossberg & Newenhouse, 2002). The researchers analyzed four variables, selected based on a literature review, for each census tract within the city limits: Percentage of workers commuting by non-auto modes Average vehicles per household Residential density Percentage of population aged 16-24 The initial screening was undertaken using the commute mode split vari- able, and 12 tracts with the lowest auto mode splits taken forward. These Page September 2005 3-30

OCR for page 74
Car-Sharing: Where and How It Succeeds 12 tracts also had low vehicle ownership rates. Three tracts were eliminated at the next stage, because they were located near the university campus and more than 50% of residents were aged 16-24 and would not be eligible for the service (Community Car requires at least five years driving experience). While two of the remaining tracts were low density on average, they were retained since they incorporated high density areas. This example shows how census data can play an important role in deter- mining the feasibility of car-sharing in different settings. However, it raises several questions, particularly regarding the choice of variables. Intuitively, commute mode split, vehicle ownership and density (which tend to be closely correlated) are likely to be strong indications of the fertility of the ground for car-sharing. However, car-sharing has been successfully established on several university campuses, raising doubts about the importance of age-related demographic variables. More importantly, there has been little quantitative research into the existence of any thresholds, and whether dif- ferent variables may play an explanatory role. Methodology This section documents the results of a GIS-based analysis of the mar- ket settings of car-sharing pods in various cities. Census data were analyzed for all 13 US cities that have significant car-sharing operations Aspen, Boston, Chicago, Denver-Boulder, Los Angeles, Madison, New York, Philadelphia, Portland, San Diego, San Francisco, Seattle, and Washington DC. Programs with fewer than four vehicles (such as Ann Arbor) and those on university campuses outside metropolitan regions (e.g. Chapel Hill) were excluded from the analysis. Full technical details of the GIS-based analysis are found in Appendix B. In contrast to the Madison example discussed above, which used census tracts, a much finer grain of analysis was used for the GIS analysis census block groups. In the City of Madison (population 208,000), for example, there are 153 block groups but just 63 tracts. Sixteen variables (see Exhibit 3-11) were analyzed at two different scales: One-half mile radius from every pod considered the typical dis- tance people are willing to walk to a pod Regional averages, for comparison purposes (for all variables ex- cept intersection density and residential density) . For an initial analysis of six cities, data were analyzed for a -mile radius and -mile radius, and results were found to be similar. Page 3-31

OCR for page 74
Chapter 3 Market Analysis The analysis looked at a range of census variables that may have an influence on the viability of car-sharing. These variables encompass demographics, commute mode share, vehicle ownership and neighborhood characteristics. Exhibit 3-11 compares the results for pod neighborhoods to the regional averages. This comparison helps identify the characteristics of pod neigh- borhoods that differ from other parts of the region. Exhibit 3-11 Summary of Demographic and Neighborhood Characteristics Pod Neighborhood Average Vehicles Cities Regional Weighted Evenly* Weighted Evenly** Average*** Difference 1 2 3 =1-3 Demographics % 1-person households 51.8% 51.0% 27.2% 24.6% % households with children 12.5% 12.5% 32.4% -19.9% % of rental households 71.5% 70.5% 39.6% 31.8% % households earning > $100,000 18.2% 16.7% 17.9% 0.3% % with Bachelor's degree or higher 54.6% 52.4% 34.0% 20.6% Commute Mode Share % drive alone to work 33.0% 39.3% 69.4% -36.4% % carpool to work 6.6% 6.7% 11.6% -5.0% % take transit to work 30.8% 23.7% 8.8% 22.0% % bike to work 2.1% 3.1% 0.8% 1.3% % walk to work 21.9% 21.1% 4.4% 17.5% Vehicle Ownership % households with no vehicle 40.0% 34.7% 11.3% 28.7% % households with 0 or 1 vehicle 82.0% 76.9% 46.0% 36.0% Average vehicles per household 0.84 0.97 1.66 -0.83 Neighborhood Characteristics Housing units per acre 21.7 17.1 Intersections per acre 0.37 0.34 % units built before 1940 43.6% 34.9% 16.9% 26.7% * Mean of data for all individual vehicles, meaning that pods with more vehicles will be weighted more strongly. ** Mean of means for each city, i.e. each city is weighted the same regardless of car-sharing fleet size. *** Mean of means for each region. Household and Neighborhood Characteristics Almost without exception, pod neighborhoods in all 13 cities have distinctly different characteristics compared to their surrounding regions. Even the least dense pod neighborhoods with the lowest transit use still have higher Page September 2005 3-32

OCR for page 74
Car-Sharing: Where and How It Succeeds densities and higher transit usage than the regional norm. Some of the main differences include: Household Size and Composition and Education. One-person households are far more common in the areas surrounding pods. The presence of children is noticeably less likely as well. Residents living in pod-areas are also far more likely to rent and hold a Bachelor's degree or higher. Income. Surprisingly, income was not a noticeable factor in the resident profiles of pod neighborhoods in the 13 cities. On aver- age, pod-area residents' income levels are within 1% of region- wide averages, but there are substantial variations from city to city. Mode to Work. Residents in pod neighborhoods are far more likely to take transit and walk to work, rather than drive, com- pared to their regional counterparts. The high mode share for walking is also indicative of mixed-use development. Vehicle Ownership. Residents of car-sharing neighborhoods own substantially fewer vehicles compared to the regional average, and are more likely to be car-free. Neighborhood Characteristics. Car-sharing vehicles in most cities (Aspen, Chicago, Denver-Boulder, and Los Angeles are excep- tions) tend to be located in older, historic, neighborhoods, which are likely to be more walkable and have less off-street parking. Car-sharing neighborhoods also tend to have higher densities; in most cities, they fall into the range of 7 to 25 housing units per acre. Explaining Variations in Car-Sharing Service The previous section analyzed the fundamental characteristics of car-shar- ing neighborhoods. This section takes the analysis further, by analyzing the amount of car-sharing the level of service that different neighborhoods can support. The level of service concept is often used with other modes, such as automo- biles and transit. For this study, a "car-sharing level of service" indicator was defined to indicate the total amount of service i.e., the number of car-shar- ing vehicles in a given neighborhood. This allows analysis of the amount of service that can be supported by neighborhoods of different types. The car-sharing level of service was calculated for each pod based on the total number of vehicles within the half-mile radius. Exhibit 3-12 shows an example of how the level of service was calculated. In this example, the level of service for the pod located in the center of the circle is 10 because Page 3-33

OCR for page 74
Chapter 3 Market Analysis there are a total of 10 vehicles in various pods within the half-mile buffer. The variables were tested for the entire data set as a whole, and individu- ally for the eight cities with a medium-sized to large car-sharing operation (25 vehicles or more). Exhibit 3-12 Level of Service Calculation The results of the correlation analysis are shown in Exhibit 3-13. An asterisk indicates a strong relationship between the variables (statistically significant at the 5% level); two asterisks indicate a very strong relationship (statistically significant at the 1% level). For all the cities analyzed, level of service correlated negatively with drive alone to work and average vehicles per household in other words, neigh- borhoods with lower drive-alone and vehicle ownership rates tend to have more car-sharing service. Level of service also correlated positively with households with no or one vehicle and households with no vehicle. Other variables with consistently statistically significant correlations (negative or positive) with car-sharing level of service include the percentages of one-person households, households with children, and rental households; commute mode share for walking and carpooling; intersection density; and residential density. Given that most variables have a high degree of correlation, it is interesting to look at which do not correlate either for the data set as a whole, or for certain cities. These variables include transit commute mode share, which Page September 2005 3-34

OCR for page 74
Car-Sharing: Where and How It Succeeds Exhibit 3-13 Correlation with Car-Sharing Level of Service Pearson Correlation with Car-Sharing Level of Service Variable Boston Los Angeles New York Philadelphia Portland San Francisco Seattle Washington DC All Records % 1-person households .619(**) 0.124 .699(**) .679(**) .822(**) .236(*) .758(**) .441(**) .478(**) % households with children -.548(**) 0.106 -.593(**) -.627(**) -.729(**) -.552(**) -.646(**) -.303(**) -.412(**) % of rental households .198(**) 0.317 .230(*) .404(*) .760(**) .317(**) .653(**) .383(**) .301(**) % households earning > $100,000 .356(**) -0.15 0.148 0.145 -.308(**) 0.037 -.425(**) -.308(**) -.066(*) % with Bachelor's degree or higher .210(**) -.483(**) .381(**) .573(**) -0.028 -0.055 -.472(**) -0.04 0.063 % drive alone to work -.441(**) -.620(**) -.406(**) -.627(**) -.851(**) -.480(**) -.758(**) -.653(**) -.431(**) % carpool to work -.503(**) 0.338 -.414(**) -.596(**) -.715(**) -.608(**) -.708(**) -.340(**) -.363(**) % take transit to work 0.033 .492(**) 0.043 -.626(**) .607(**) .477(**) .277(**) .198(**) .104(**) % bike to work -.149(*) -.425(*) .202(*) 0.109 0.005 -0.046 -.318(**) .688(**) -0.003 % walk to work .374(**) 0.337 .376(**) .718(**) .915(**) .281(*) .850(**) .538(**) .512(**) % households with no vehicle .427(**) .661(**) .551(**) .667(**) .902(**) .361(**) .832(**) .681(**) .399(**) % households with 0 or 1 vehicle .522(**) .485(**) .400(**) .735(**) .793(**) .422(**) .770(**) .633(**) .488(**) Average vehicles per household -.495(**) -.620(**) -.497(**) -.722(**) -.839(**) -.405(**) -.819(**) -.680(**) -.458(**) Housing units per acre .751(**) -.445(*) .379(**) .843(**) .636(**) .656(**) .671(**) .890(**) .174(**) Intersections per acre .374(**) 0.114 -.259(**) .577(**) .710(**) .475(**) .642(**) .519(**) .290(**) % units built before 1940 .311(**) -0.024 -.208(*) -0.26 0.144 .583(**) 0.142 .475(**) .223(**) * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). Page September 2005 3-35

OCR for page 74
Chapter 3 Market Analysis correlated positively in most cities but not in Boston, New York, and Phila- delphia. Income, education, bicycle commute mode share and the percentage of units built before 1940 were other variables that did not have a consistent correlation with car-sharing level of service. The correlation analysis shows that as level of service increases, so does the proportion of rental households, one-person households, households with low vehicle ownership, and transit and walking mode shares. Similarly, as level of service increases, the proportion of households with children, commuters who drive alone or carpool, and average vehicles per household decreases. Through multiple regression analysis, several models were tested for their ability to predict the level of car-sharing service for a neighborhood. It should be noted that New York appears as a case unto itself it has very high residential density and very low vehicle ownership rates, and was therefore excluded from the regression analysis. For the other 12 cities, the best models were found to use vehicle ownership rates combined with walk mode share. This model predicts almost 50% of the variation in car-sharing between different neighborhoods. In other words, these characteristics of a neighborhood are half of the explanation for car-sharing success. Full details of the multiple-regression analysis are provided in Appendix B. The walking mode share variable suggests that car-sharing level of service is higher in areas that have a mix of residential and employment uses and areas that are more pedestrian friendly. Commute mode share for walk- ing has the strongest correlation with car-sharing level of service of any of the variables examined. The average vehicles per household variable has an intuitive connection to car-sharing success; in neighborhoods with lower vehicle ownership, more households are able to fulfill their daily needs without a car. While this formula provides a partial explanation of car-sharing success, there are clearly other factors that combine with these neighborhood characteristics to fully explain where car-sharing will succeed, such as the amount of capital that operators have to expand to the fullest market potential. Member Perceptions of Neighborhood Type Another source of quantitative data on market settings comes from the survey of car-sharing members. In addition to the information on demo- graphics (discussed earlier in this chapter) and social and environmental Page September 2005 3-36

OCR for page 74
Car-Sharing: Where and How It Succeeds impacts (Chapter 4), various questions explored the types of neighborhood in which car-sharing members live. This enables issues such as parking availability to be explored, along with subjective impressions of the qual- ity of transit and the pedestrian environment none of which are available through census data. Respondents to the internet survey furnished a good deal of information about the settings in which they lived. Most of them were center city resi- dents. They described their living environments as shown in Exhibit 3-14. As can be seen, these findings serve to confirm the results from the census data. Exhibit 3-14 Locational Information for Car-Sharing Members Strongly Locational descriptors Agree Agree My neighborhood has a good walking environment 46.2% 40.3% My neighborhood has good public transit service 48.5% 37.9% It's easy for me to walk to a grocery store 37.2% 29.5% More than once, I have spent a long time looking for a 26.3% 21.2% parking spot in my neighborhood Nearly 60% of all respondents lived in a home that had a driveway, garage, or other off-street parking space, but nearly half of those persons (29% of total respondents) did not use that parking space. Of those who did use such a parking space, only 13% paid for its use. Combined with the fact that less than half of respondents report difficulties parking in their neighbor- hoods, this suggests that parking difficulties are just one of many factors influencing the success of car-sharing in a given neighborhood. Summary of Results One of the main conclusions that can be drawn from this analysis is that car-sharing users are not necessarily representative of the neighborhoods surrounding car-sharing pods. For example, as discussed in the earlier part of this chapter, 83% of members surveyed have a Bachelor's degree or some post-graduate work. In contrast, 55% of residents living close to pods have a Bachelor's degree, higher than the regional average of 34% but still far below the 83%. Most importantly, although this variable has explanatory power in some cities, it is not consistently related to car-sharing success. In Portland, San Francisco and Washington, DC, there is little relationship Page 3-37

OCR for page 74
Chapter 3 Market Analysis between education levels in a neighborhood and the amount of car-sharing service. Another indication comes from income. As discussed in the earlier part of the chapter, there is a wide income spread among car-sharing members. The pod neighborhoods in Chicago, however, have some of the highest propor- tions of high-income households in any of the cities examined, even though the car-sharing program there has targeted low-income households. These differences between member and neighborhood characteristics are not unexpected, given that car-sharing's member base consists of such a small proportion of residents. Instead, it seems that car-sharing is appealing to a large number of highly educated, but not necessarily high-income, gentrifiers and young professionals. They are living in urban neighborhoods which are characterized by a high proportion of rental housing; single-person and childless households (even though car-sharers may live with a partner or children themselves); pedestrian friendliness; and relatively high density. This suggests, then, that the most rewarding path for analysis is to focus on neighborhood and transportation characteristics that promote car-sharing, rather than on finding neighborhoods that match the individual demo- graphic characteristics of car-sharing members. For example, even though high education levels are one of the hallmarks of car-sharing members, the neighborhoods with the highest percentages of college graduates may not be the most fertile turf for car-sharing. Indeed, both Flexcar and City CarShare have been forced to close pods in Palo Alto, CA home of Stanford University and one of the most highly educated communities in the United States. Instead, certain transportation characteristics may be the most important to identifying potential markets for car-sharing. Variables such as commute mode split, household composition and in particular vehicle ownership seem to be the best proxies for the types of neighborhoods where car-shar- ing succeeds. They indicate places where transit and walking are realistic alternatives, and where a car is not needed for everyday travel. They also indicate places that attract a high proportion of single, childless households. Specifically, average vehicles per household and number of people who walk to work within a half mile of a pod location appear to be the most important variables for predicting car-sharing success as determined in the multiple-regression analysis. The percentage of households with no or one . Aspen, Boston, Denver-Boulder and New York also have over 20% of households earning more than $100,000 per year. Page September 2005 3-38

OCR for page 74
Car-Sharing: Where and How It Succeeds vehicle also appears to have a strong, non-linear relationship with car-shar- ing success (see Appendix B). Surprisingly, physical factors such as density, intersection density and age of housing do not stand out as primary indicators. The role of density is discussed in more detail in the next section. Role of Density The results provide some conflicting suggestions about the overall im- portance of residential density. This variable is clearly important in some manner for car-sharing. As noted above, it is an indication of the potential customer base for a pod doubling the density will double the number of customers within walking distance. It also serves as a good proxy for the auto-orientation of a neighborhood. Holtzclaw et al. (2002), for example, found that residential density served as the best predictor of vehicle travel, explaining 63%-86% of the variation in vehicle miles traveled in San Fran- cisco, Los Angeles and Chicago. However, the density levels for pod neighborhoods are far below what might be expected from a review of other research. For example, 25% of pod neighborhoods have a density of 8.5 households/acre or less. For comparison, single-family "sprawl" often clocks in at around three units to the acre, while San Francisco Bay Area data suggest that transit ridership increases noticeably at 10 households per residential acre (Holtzclaw, 2002). A threshold of 15-25 units per acre is often cited as a desirable minimum for transit oriented development, while 4-6 units/acre appears to be the minimum for even basic hourly frequencies (for a broader discussion, see Kuzmyak et al., 2003; Dittmar & Poticha, 2004). One explanation may be that many pods are situated close to rail stations with large amounts of surface parking, which lowers gross densities, or are in mixed-use centers with lower residential densities but a large daytime population. Certainly, relatively high walking rates (22% on average for all pod neighborhoods) suggest a predominance of mixed-use development. However, it is also possible that density is not as dominant in explaining car-sharing market settings as it is, for example, in the case of transit. Car-Sharing Thresholds In summary, then, how can a current or would-be car-sharing operator, or a transit agency or other partner organization, assess the types of neigh- Page 3-39

OCR for page 74
Chapter 3 Market Analysis borhoods where car-sharing may be viable? Some guidelines, based on the analysis in preceding sections, are shown in Exhibit 3-15, which shows two sets of thresholds: low service, where car-sharing may be viable but where limited growth can be expected, and high service, where car-sharing is likely to flourish. These thresholds are not precise requirements. Rather, they are intended as guidelines to show the approximate neighborhood characteristics that help to sustain car-sharing. There are certainly examples of successful car- sharing operations that do not meet these thresholds, particularly in the special niches discussed earlier in this chapter. However, these guidelines can assess the extent to which neighborhoods do have supportive charac- teristics. Combined with the other considerations discussed in Chapter 8, such as support from partner organizations, they can help determine the likelihood of success. Exhibit 3-15 Guidelines for Where Car-Sharing Succeeds Level of Service Variable Low High* Demographics % 1-person households 30% 40%-50% Commute Mode Share % drive alone to work 55% 35%-40% % walk to work 5% 15%-20% Vehicle Ownership % households with no vehicle 10%-15% 35%-40% % households with 0 or 1 vehicle 60% 70-80% Neighborhood Characteristics Housing units per acre 5 5 * High service roughly equates to 10 or more car-sharing vehicles within a half-mile radius. Note: For most variables, the values are the suggested minimums that are needed to achieve a given level of car-sharing service. For the "% drive alone to work" variable, the values are the suggested maximums. . These values were approximated from analyzing percentiles and scatter plots for each variable. Page September 2005 3-40