An Assessment of the Expected Impacts of City-Level Parking Cash-Out and Commuter Benefits Ordinances

Chapter 2. Literature Review

The project team reviewed available research on the effects of cash-out and changes to the price of transit and parking. While a handful of studies have been conducted to estimate the impact of cash-out policies and changes in parking and transit prices on commute mode choice, no city level policy studies were found.

Parking Elasticities

In a comprehensive analysis of eight parking cash-out programs in Southern California, Shoup (1997) found such programs were associated with a 13 percent reduction in single-occupant driving, an 11 percent reduction in vehicle trips per commuter per day, and a 12 percent reduction in vehicle miles traveled (VMT). The average price elasticity of demand for parking at the eight employer sites was -0.15. Van Hattum (2009) conducted a similar study of seven employer sites in Minneapolis-St. Paul where parking cash-out programs were implemented and found a 12 percent reduction in SOV travel. Likewise, Glascock, Cooper & Keller (2003) found a 10 percent reduction in employee parking demand resulting from parking cash-out in Seattle. Outside the U.S., De Borger and Wuyts (2009) found parking cash-out to be associated with a nearly 9 percent reduction in driving commutes and a 17 percent increase in transit use based modeled Belgian data.

Shoup (2005) reviewed seven studies conducted between 1969 and 1991 that analyzed the effect of employer-paid parking on SOV commute rates. The review found that when employers paid for parking in analyzed areas, on average, 67 percent of employees drove alone. When employees paid for parking in the same areas, the average drive-alone rate dropped to 42 percent. Price elasticity of demand for parking at the various employment sites ranged from -0.08 to -0.23, and the mean was -0.15.

Concas and Nayak (2012) conducted a meta-analysis of parking price elasticity of demand in which they reviewed 25 related articles that included 169 elasticity variables. The studies covered multiple countries, and elasticity values ranged from -6.22 to zero, with a mean value of -0.482. The authors developed a model to explain the variation in elasticity estimates based on factors such as geographic location, estimation method, and data type. Their model, applied to estimate an elasticity for the United States (using econometric techniques), yielded a parking price elasticity of -0.39.

Litman (2022a) conducted an extensive literature review of transportation price elasticities and generally found that the demand for vehicle trips with respect to parking price ranges from -0.1 to -0.3, with significant variations due to demographic, geographic, and trip characteristics. While short-run and long-run elasticities are not explored for parking pricing, Litman (2022a) summarizes other short-run and long-run price elasticities related to travel demand, where long-run values are typically two to three times short-run values. In general, commuters may be more responsive to changes in pricing over a long-time horizon, given additional time to adjust behaviour, compared to over a short-term horizon.

In a study conducted for the San Francisco County Transportation Authority (2016), researchers modeled travel demand for five parking policy scenarios, including a cash-out scenario in which drivers paid 75 percent of the parking cost. Model results showed the proportion of vehicle commuting trips fell in the Northeast Cordon (downtown business district) and citywide by 9 percent and 6 percent, respectively. Further analysis showed parking arc elasticities associated with cash-out are -0.45 for destinations within the Northeast Cordon and -0.06 elsewhere in San Francisco, which may be a result of fewer alternative mode options outside of the downtown business area.

A study conducted by Knittel and Tanaka (2019) provides new insights on price elasticities and vehicle travel, although the data used is based on drivers located in Japan. Using mobile phone data for more than 90,000 drivers, the authors found the price elasticity for vehicle kilometers traveled to be -0.30.

Shin (2020) utilized the Puget Sound Regional Travel survey to evaluate how commuting behaviors are related to commuter incentive programs. The Puget Sound region is subject to Washington State’s Commute Trip Reduction Law, which requires certain employers to have TDM programs.¹¹ Shin first examines the relationship between various benefits and commuter mode choice. Results show that transit-related benefits (i.e., free or subsidized transit passes) are associated with higher probabilities of commuting via public transport, non-motorized transport, and carpooling relative to driving alone. In contrast, employer-subsidized parking reduces the likelihood of commuting by these modes compared to driving alone.

Additionally, Shin (2020) finds that transit benefits are not only associated with lower worker commute trip VMT, but also with lower non-work trip VMT. On average, workers with transit benefits are expected to drive 3.16 miles and 1.15 miles fewer for commute and non-work trips daily, respectively. In contrast, workers with free workplace parking drive 3.13 miles and 0.99 miles more for commute and non-commute trips daily on average, respectively. When controlling for workplace transit accessibility and employment density in addition to residential built environment characteristics, Shin finds workers with transit benefits drive 2.19 fewer miles for work trips and drive 0.83 fewer miles for non-work trips on average compared to workers without transit benefits. In contrast, workers with free workplace parking are expected to drive 2.48 more miles for work trips and 0.78 more miles for non-work trips on average compared to workers without free workplace parking.

Shin found that the availability of free workplace parking benefits for a given worker significantly impacted the VMT of the other members in the workers’ household differently than their own. Namely, other members of the same households were expected to have lower work and non-work VMT (such a reverse relationship was not discovered for transit benefits). The differing impacts on free workplace parking make intuitive sense as workers with parking may add additional VMT by adding stops to their commute trips while other household members would then have fewer trips they would need to make for the household. Shin also notes one possible explanation is related to residential choice: “[w]orkers, on average, are found to live closer to their workplaces if their household members (except for themselves) are offered employer-sponsored parking benefits; however, average commute distance is statistically significantly longer for workers who have their own parking benefits than for those without” (p. 15). Overall, Shin’s work demonstrates spillover effects of various commuter benefits, into both non-work VMT for a given worker, as well as total VMT for others in their household. The differing directional relationships introduce ambiguity into the aggregate effect on VMT and thus no adjustments are made to the analysis here due to Shin’s research.

As such, policies implementing employer-subsidized transit benefits may offer additional VMT reductions (along with subsequent congestion, crash, and emissions reductions). The extent of these reductions is dependent on many factors, including how transit benefits are offered to employees (e.g., as a monthly pass or not). For example, a member of the peer review group convened for this study noted her employer in Los Angeles paid a per-trip fee associated with employee transit passes. As such, employees had to log their trips and were asked to only use their employer-paid benefits for work trips. Conversely, another member noted that in Boulder, CO, there is evidence that employees with access to free transit passes are more likely to use transit for non-work trips and are also more likely to bike to work more.

In a study of German commuters, Evangelinos et al. (2018) found parking cash-out offerings significantly reduce the probability of commuting by car, even when restricting the sample to vehicle-only commuters. For these commuters, parking cash-out values offered were equivalent to the cost of a transit pass rather than the cash value associated with the parking space.

Brueckner and Franco (2018) develop a theoretical model based on a simplified version of a city divided between a suburban and central zone, connected by a roadway and transit line. Assuming the optimal allocation of resources, the authors’ theoretical model results reveal the percentage of commuters traveling by car goes from above 80 percent to around 50 percent as the share of parking costs covered by employers goes from zero to 100 percent. Using data from a stated choice experiment of commuters in Nanjing, China, Ding and Yang (2020) found a 25 percent increase in parking cost was associated with an 8.7 percentage point reduction in auto mode share, while 50 percent, 75 percent, and 100 percent increases in parking costs were associated with 14.3, 19.1, and 19.9 percentage point reductions in auto mode share, respectively. It should be noted, however, that commute locations were to Nanjing’s CBD and starting mode choice probability for commuting by car was only 26.2 percent, compared to 24.6 percent for bus and 49.2 percent for rail.

In a study using the 2012 California Household Travel Survey, Khordagui (2019) modeled the decision to drive in a hypothetical scenario where all commuters pay for parking or take an equivalent parking cash-out incentive based on the average paid parking price in the workplace zip code. Model results show a 10 percent increase in the price of parking (and subsequently the cash-out value) is associated with a reduction in the probability of driving alone to work of one to two percentage points, with the lower end result related to the “parking opportunity cost scenario,” which attempted to explore the opportunity cost of free parking based on the prevalent parking price in each geography. The marginal effects vary with prices and the relationship is not linear, but the study notes these findings correspond to a parking price elasticity range of -0.13 to -0.26. It suggests that this result is generally in line with, and at the lower end of, some prior studies, such as a study by Su and Zhou (2012) reporting an elasticity of -0.23 and Washbrook et al. (2006) who report an elasticity of -0.30.

Travel and Transit Elasticities

Beyond elasticities associated with parking price, travel behavior may also be examined through transit price elasticities (reflecting expected changes in transit use in response to transit price). Here, elasticities range widely depending on a number of factors, including the type and level of transit service, user type, and time of use. Litman (2015) suggests the ranges for peak commute transit elasticities with respect to prices should be -0.15 to -0.30 in the short term, and -0.40 to -0.60 in the long term. However, Litman (2015) notes the elasticities could be as high as -0.8 to -1.0 for suburban commuters, which could make up a significant portion of the employee population taking a transit benefit. A study by Gillen (1994) provided disaggregated transit elasticities and identified an elasticity range of -0.10 to -0.19, which is specific to work trips.

There is also rich literature on short-run and long-run price elasticities of demand for gasoline consumption. Litman (2022a) compiles several such studies placing the short-run price elasticity of demand ranging between -0.11 and -0.27 compared to between -0.58 and -0.71 in the long-run. This aligns with earlier findings from Espey (1996), who found a short-run elasticity of -0.26 and a long-run elasticity of -0.58.

Greenberg and Evans (2015) conducted a review of travel price elasticity data as part of an effort to estimate the impact of cash-out, pay-as-you-drive car insurance, and the conversion of State and local sales taxes on newly purchased vehicles to mileage taxes. The base of the analysis is the overall variable driving cost, which focuses on per-mile fuel costs. With an initial parking price of zero, the percentage increase in the price of driving is derived by summing the new parking price (which could be presented through cash-out as an “opportunity cost”) and the pre-existing fuel price and comparing it with the pre-existing fuel price on its own. According to Greenberg and Evans, converting results from studies of the elasticity of demand for VMT with respect to fuel price to an elasticity of VMT with respect to the per-mile price of driving yields elasticities ranging from -0.22 to less than -0.50. Such a conversion is performed by first recognizing that because part of the response to higher fuel prices is mileage shifting to more efficient vehicles and more fuel-efficient driving, the per mile price of driving experienced by drivers rises, on average, by a lower percentage than the fuel price. Citing the compendium of studies in Litman (2022a), Greenberg and Evans justify attributing fuel savings from higher fuel prices evenly between reduced mileage and better fuel economy. The change in the per mile cost of driving, then, is essentially assumed to be half of the change in the cost of fuel when calculating the price elasticity of demand for VMT. Greenberg and Evans settle on an elasticity of demand for VMT with respect to the per mile price of travel as -0.30 and also note that three other major studies use this same value.

Considerations in Applying Elasticities

The use of parking price and per-mile travel cost elasticities to estimate the impacts of parking cash-out policies could raise some concerns for analyses covering relatively short time periods. This is because the manner in which commuters respond to cash-out payments tied to the loss of a parking space may be different than how they respond to increases or decreases in direct parking charges or per-mile travel costs.

Research from Thaler (1980) notes an “endowment effect,” which could apply here by considering that the aggravation that some consumers may experience from giving up a parking space they had already begun to use could exceed the pleasure they would receive from accepting a cash-out. Thaler concluded that “[t]he aggravation that one experiences in losing a sum of money appears to be greater than the pleasure associated with gaining the same amount” (p. 43). He also discussed a well-known economic theory called the “endowment effect,” noting that it causes “a certain degree of inertia [to be] introduced into the consumer choice process since goods that are included in the individual’s endowment will be more highly valued than those not held in the endowment” (p. 44).

While loss aversion may affect current holders of parking spaces, it would not affect new employees. However, discussions with the peer review group for this study highlighted that the endowment effect would not be in play for new employees, who would not have experienced free parking prior to choosing between parking and cash. This statement is supported by Thaler who, as noted above, said that the endowment effect applies to items that are “held” by the individual. This means that if the endowment effect does impact behavior as it relates to parking cash-out, that effect is likely to decrease over time as employees turn over. For this reason, the project team determined that it is acceptable to model cash-out related behavior change using travel and parking cost elasticity data.

Considerations for Monthly and Daily Parking Cash-Out

The above studies primarily focused on the impact of monthly cash-out policies, but responses to cash-out offers may vary depending on how a program is implemented. Two survey-based studies were conducted in Dublin to determine employee preferences for different cash-out program parameters. Commuter surveys administered by both Farrell et al. (2005) and Watters et al. (2006) found that, when presented with scenarios under which commuters would have to give up their parking space, surveyed participants overwhelmingly opted for daily cash-out options over annual, monthly, or one-time payouts. This was true even in the absence of monetary values associated with cash-out options in the Farrell et al. (2005) survey and even when daily cash-out resulted in the lowest potential monetary gain in the Watters et al. (2006) study.

In Minneapolis, researchers worked with FHWA, the Minnesota Department of Transportation, Metro Transit, and the City of Minneapolis to target monthly contract holders in city-owned parking lots, offering a variety of cash-out offerings, including daily cash-out where monthly parkers received some financial renumeration when not parking (Lari et al., 2014). In one phase of the study, the effects of two different daily-parking cash-out offerings were evaluated. For users who opted for a free daily transit pass and $2 rebate on days where transit was taken instead of driving, the lowest average monthly SOV rate across the nine months of the study was 68 percent, compared to 75 percent at the end of the study and 83 percent prior to the study. Users who opted for the same transit rebate scheme plus a $7 daily rebate when neither parking nor transit was used exhibited a 72 percent SOV rate prior to the study and 60 percent SOV rate at the end of the study. This offering is very analogous to parking cash-out.

Equity

Lower-income households are less likely to own and have access to a private vehicle than moderate and higher-income households (FHWA, 2020). Additionally, low-income households are increasingly dependent on walking, biking, and transit for their travel. As such, free parking is a financial benefit that many lower-income employees cannot access. This challenge is expected to increase as a result of especially high vehicle prices, topping $47,000 for the average new vehicles (Cox Automotive, 2022a) and $28,000 for the average used vehicles (Cox Automotive, 2022b) in mid-2022. It follows that parking cash-out and related commuter benefits policies can enhance equity by providing cash or an alternative benefit for employees who may not be able to use free/subsidized parking.

This consideration is certainly nuanced, as low-income households who do primarily rely on vehicle travel may have less of an ability to shift to other, non-vehicle modes, particularly if they are not located near high-quality transit or supportive active transportation infrastructure. Dong et al. (2012) make a strong point regarding cost responsiveness to auto use by income. Namely, the authors note that higher income households might be expected “to be less price sensitive than lower income ones, but… realize that poorer people spend a good deal more of their travel budgets on necessary trips, such as commuting. So they may have little choice but to pay the extra travel cost imposed: or they may have to stay home if the trip is of a more discretionary nature. Either way, they are often dependent on their current travel option. Higher and middle-income households, in contrast, may find it easier to drop some discretionary trips if fuel or other prices rise sharply. Over the longer run they also have more opportunities to adapt their lifestyle to absorb additional travel expenditures” (p. 11).

The team looked at whether price elasticity for car travel among lower-income workers, and specifically for car commuting where such data may be available, is different than for higher-income workers. This is an equity issue because if price elasticity were to be higher for low-income workers then they would be more likely to shift behavior and subsequently benefit financially more than others. So far, the data seems inconclusive, indicating that while fewer low-income workers drive to work in the first place, they do not seem to alter their behavior noticeably more than others when the price of driving does change. One recent Swedish study of app-based travel incentives supports this; in addition to income level, homogenous responses to pricing were found across age and education levels, although there were differences observed by gender (Axhausen et al., 2021). A report published by the Victoria Transport Policy Institute identifies patterns found in pricing and transport research, including that higher-income travelers are less sensitive to pricing than lower-income travelers (Litman, 2022c).

In contrast, a study by Gillingham (2013) finds heterogeneity across income groups in their VMT response to (gasoline) prices; however, results from this work indicate higher income households tend to be more responsive to pricing than lower income households. Gillingham suggests some potential explanations for this, including more observed discretionary driving trips, higher levels of vehicle ownership and within-household vehicle switching, and potential to shift from driving to flying for some trips in wealthier households. Gillingham notes this finding is in contrast to other work in this area. West (2004) and West and Williams (2004) find that lower-income households are more responsive to gas price changes. Inconsistent findings related to price responsiveness here may be due to differences in commuting and work-based travel experienced across income and industries. It is difficult to make a clear inference from these results, not only due to their somewhat contradictory results, but also because gas price responsiveness may not be fully aligned with responses to parking pricing.

Literature examining responsiveness to parking pricing by income is sparser than that looking at responsiveness to fuel costs. A Transit Cooperative Research Program report (Vaca and Kuzmyak, 2005) notes that “model-derived analyses…suggest that parking pricing impacts, as measured by SOV trip reduction, may be as much as eight times greater for trip makers in the lowest income quintile as for travelers in the highest quintile” (p. 13-6). These analyses, however, rely on simulated (versus empirical) data based on the 1995 Census Public Use Microdata Sample (PUMS) for Sacramento and Los Angeles. Further, they examine price influences on the proportion of work trips taken by SOV, not VMT.

Despite limitations in clear available data on parking price responsiveness by income level, equity of the current analysis can be evaluated through comparison to other data sources looking at auto ownership levels and the provision of benefits across income levels. A later section of this report presents such an evaluation.