Chapter 3. Specific City Findings

The use cases provide baseline information for each location, followed by a comparison to the baseline. The Washington, DC, and Atlanta cases rely on processed output from the third-party GPS aggregator. The aggregator analyzed its data for trips and cruising trips using the system developed by the Federal Highway Administration (FHWA) and this project team. The research team specified 15,000 geographies (i.e., streets or aggregations of streets—in this case, sometimes census block groups). The limit was a contract condition of the aggregator/processor. Thus, individual results for each street face were unobtainable. Instead, street faces of particular interest were specified with other results aggregated to the census block group level. For the Washington, DC, and Atlanta cases, the data are therefore presented at the block group level with more detailed analysis into the areas where street faces have been specified.

The Chicago analysis uses raw location data that were processed by the project team. In the Chicago case, data are available at the street-face level for wherever trips have been made. The Seattle case includes processed output by the third-party GPS aggregator and analysis based on individual location data provided by a data broker. The approach in Seattle was designed to facilitate comparisons between the two different data sources.

Each case answers certain questions, but also raises questions related to data sources and interpretation that can be mediated in future analyses. For example, the Seattle case shows an increase in cruising on metered blocks after a price change. Given that the analysis is based on locations of trip ends and not cruising paths, the meaning of that finding is unclear. From the trip end locations alone, the analyst knows with certainty where the trip ends, but not the process of parking search. The following possible interpretations are listed in order of likelihood, based on professional judgment:

• Drivers cruised for parking and more quickly found a spot on a metered street because the price change affected availability as intended.
• Drivers reacted to the price increase and searched for a better bargain. Quickly realizing a bargain was not to be had, the driver accepted a vacant space on the metered block.
  • This problem might resolve itself and can be studied again after the market has had time to absorb the price shock.
  • This may be an unintended consequence of the policy change.
• The policy change had no effect on availability; other research^{21, 22} has shown that multiple price changes may be needed before a change in driver behavior can be measured.

Similarly, in the Chicago case, data were analyzed for the month of June across 3 years. Data for the third year were poor quality and were not useful for the analysis. The analysis shows potentially important differences between the 2 years for which there are good data. However, given the experiment, whether the observed differences are true differences or sampling differences is unclear. Had the analysis been strictly an analysis of Chicago, rather than a study of use cases for broad application, the recommended approach would be to acquire additional data to corroborate the findings. Instead, the analysis is presented with these caveats. While the results are potentially interesting, caution should be exercised when drawing conclusions as abstraction may not be justified on the basis of this analysis.

The following sections show baseline and comparative case conditions for Washington, DC; Atlanta; Chicago; and Seattle in that order. The Seattle case comprises multiple analyses showing a business-as-usual meter price adjustment—a special case wherein all metered pricing is revoked, and a data comparison showing the benefits and shortcomings of two different data sources.

Washington, DC: Cross-Sectional Analysis Focus on Metrorail Stations

Washington, DC, has a varied landscape with respect to density, activity type, and transportation infrastructure. The focus in this case study is cruising for parking across Washington, DC. Special attention is focused on three sports stadiums, and when the stadiums have and do not have events planned. The analysis also emphasizes cruising in the areas around Metrorail stations. Washington, DC, data span January 1–December 3, 2018. In lieu of a specific policy change that would lead to a before/after analysis, the findings here are cross-sectional, illustrating different cruising patterns for different urban forms: specifically proximate to Metrorail versus beyond the Metrorail catchment areas.

The data have been processed by a third-party data consolidator. This analysis is based on a data set of trip ends by time of day. Figure 9 illustrates the study area indicating the streets for which street-level data are available.

Source: FHWA.

Figure 9. Map. Washington, DC, study area.

Where Cruising Occurs

Cruising is higher in the areas around Metrorail stations where there is an overall higher concentration of trips relative to street length. The streets in the Metrorail catchments comprise 23 percent of Washington, DC, street length and account for 45 percent of trip ends. Thus, the trip intensity in the Metrorail catchment area is almost twice what it would be if trips were evenly distributed throughout the street network. Cruising is almost evenly split between the Metrorail catchment areas and the rest of the city (51/49), meaning that half of the cruising is concentrated on about one-quarter of the streets. In comparing trips, cruising trips are slightly disproportionately represented near the Metrorail stations (45 percent of trips and 51 percent of cruising trips) and heavily concentrated there on a street-length basis (51 percent of cruising trips and 23 percent of street miles).

Washington, DC, has an aggregate cruising rate of 5.8 percent and 6.6 percent around Metrorail stations where the concentration of all trips (cruising or not) is very high. Figure 10 shows cruising frequency by block group. To avoid the potential problem of statistical outliers, areas with fewer than 30 observations are not shown.

Source: FHWA.

Figure 10. Map. Washington, DC, cruising frequency.

The amount of time spent cruising and the distance cruised are also important urban concerns. The average length of a cruising trip is just over 2 minutes, but the time can vary substantially. Figure 11 captures and ranks cruising in Washington, DC, by showing the aggregate time spent cruising in each block group. The block group average cruise time is weighted by the number of trips for which cruising is a component; this aggregate is divided by the street length of each block group to account for differences in size (i.e., if one block group has twice as much land area and corresponding street length than another, more trips relative to the smaller block group are expected). The modified measure is average time spent cruising per block within each block group.

Source: FHWA.

Figure 11. Map. Washington, DC, cruising impact.

When Cruising Occurs

Figure 12 shows the diurnal distribution of all trips and trips that include cruising. Cruising trip intensity lags trip making. This can be interpreted as resulting from earlier trips using available parking leaving fewer spaces open for later trips, which can show up as cruising. Other ways of looking at the data follow; the primary hypothesis holds.

Figure 13 shows the diurnal pattern of trip making with traditional morning and afternoon peaks, superimposed on the rate of cruising (number of cruising trips/number of trips). The rate of cruising is highest in the hours just past midnight, although that coincides with the time of lowest trip making and fewest absolute cruising trips (see Figure 14 for cruising trips by time of day). Cruising is lowest during the morning peak travel period and increases as the day wears on. A probable explanation for higher levels of cruising during midday and early dawn is that most people will have reached their destinations and parked at those times, leaving a relative scarcity of available spaces. This is the same phenomenon shown in Figure 12 as a lag. Early arrivers get the spaces, which leaves none for those who come after. At the same time, there is likely both more turnover and more availability during the traditional travel peaks, as some travelers are departing, others arriving, and many others en route somewhere.

Source: FHWA.

Figure 12. Graph. Diurnal distirbution of all trips and cruising trips.

Source: FHWA.

Figure 13. Graph. Cruising frequency and overall trip making.

A close look at the Metrorail catchment areas shows a similar pattern. Trip making is highest in the morning peak period, when the proportion of cruising trips is low. As more trips have been completed and more parking spaces used, the relative difficulty of parking increases and cruising trips as a proportion of all trips begins to rise. As more trips are made in the traditional afternoon peak, parking eases—demand may drop at the same time that supply increases—and cruising as a proportion of all trips declines. Cruising as a percent of all trips is relatively stable in the Metrorail catchment area, climbing between 8 and 11 a.m. and remaining between 6 and 8 percent through the afternoon and evening.

Source: FHWA.

Figure 14. Graph. Cruising frequency and total trips, Metrorail catchment area.

Source: FHWA.

Figure 15. Graph. Cruising frequency and total trips, outside Metrorail catchment.

Figure 16 shows what time of day cruising trips occur, superimposed with the frequency relative to all trip making. From the perspective of when cruising occurs in Washington, DC, the highest occurrence is during traditional business hours. Cruising trips are increasing as a proportion of total trips and in absolute numbers from 8 a.m. to about noon; thereafter, the hourly rate of cruising remains relatively steady until about 6 p.m., when it begins to drop off sharply. However, as a percent of all trip making, cruising trips decline from noon to about 5 p.m., at which time they increase again relative to all trip making. Most noteworthy is the late afternoon/early evening period when absolute cruising is near its highest but as a percentage of trips it is relatively low.

Source: FHWA.

Figure 16. Graph. Diurnal distribution of cruising and cruising as percent of all trips.

The team hypothesized that cruising on stadium event days might be more intense than on other days, but the analysis does not support that finding. All Washington, DC, stadiums are relatively easily accessed by Metrorail, and additional impacts were not detected.

Summary

Overall cruising rates in Washington, DC, tend to the national average at around 5.8 percent of trips. Both trip making and cruising are highly concentrated on the streets surrounding Metrorail stations. These streets comprise about 23 percent of street miles. However, 45 percent of trips end on streets in the Metrorail catchment and 51 percent of cruising trips end on streets in the Metrorail catchment areas. This disproportion puts the cruise rate for trips ending near Metrorail at 6.6 percent. Cruising impact is further concentrated in the area considered downtown Washington, DC.

The diurnal distribution of cruising trips is similar to findings in other cities in that there is a trip-making lag. Cruising for parking as a proportion of trips is at its lowest during the morning commute hours and steadily builds from there, peaking at midday. Cruising as a percent of all trips hits a local minimum during the afternoon commuting peak, increasing into the evening. Finally, it appears from these data that cruising is not exacerbated by events held at the Washington, DC, stadiums, which are relatively well served by transit.

Atlanta, Georgia: Longitudinal Analysis and Mixed-Use Focus

In this case study, the research team explored the geography of cruising across two different time periods. To the extent possible, the research team looked at the differences in single use versus mixed use areas. The data span from October 2019 to September 2020 and covered the areas of Atlanta located in Fulton County.

There are two defined periods within this time frame:

• Baseline: October 1, 2019–March 31, 2020, with a carve-out for the holiday period from November 25, 2019, to January 5, 2020
• Early COVID: April 1, 2020–September 30, 2020

The data are processed by a third-party data broker and results are reported at census block group level for much of the city in Figure 17). Street level outputs are available for the areas outlined and shown in green; these streets are characterized as mixed use. The data reported refer to the end locations of cruising trips. The area of detail (i.e., mixed use areas) includes downtown Atlanta and Buckhead.

Source: FHWA.

Figure 17. Map. Atlanta study area.

Where Cruising Occurs

Figure 18 shows cruising frequency by census tract. Tracts that had fewer than 30 trips were excluded from the analysis The outline delineates the areas for which street-level output is available. These subareas are examined in greater detail later in this chapter.

The baseline average weekday cruising rate in Atlanta is 4.9 percent of trips; the 75th percentile rate for block groups is 6.5 percent, suggesting very little variation. In mixed use neighborhoods, those for which the project team was able to analyze street-level data, baseline cruising averages 7.4 percent.

Source: FHWA.

Figure 18. Map. Baseline cruising Atlanta.

Not all cruising is equal, and although the mean time spent looking for parking among cruising trips is under 2 minutes, it can vary. Figure 19 shows total cruising time for each block group by weighting the average cruising time by the number of cruising trips that occur. To account for the different sizes of block groups, the result is normalized by the street length within each block group. The darkest areas in figure 3 indicate the areas of Atlanta with the most severe cruising concern.

Source: FHWA.

Figure 19. Map. Cruising impact.

When Cruising Occurs

Figure 20 illustrates cruising over the course of an average weekday in Atlanta. As occurs in other cities, cruising tends to lag trip making—the first arrivers would easily find parking. Cruising disproportionately occurs starting around 6 a.m., earlier than is seen in other cities, but possibly due to the relative paucity of on-street parking. The proportion of cruising trips drops sharply along with all trips after 7 p.m.

Cruising, as a proportion of all trips, is relatively stable, remaining between 4 and 6 percent of trips throughout the day. There is a small spike at around 1 a.m., as is documented in other cities, when trip making is at its lowest. This is illustrated in Figure 21.

Source: FHWA.

Figure 20. Graph. Diurnal distirbution of all trips and cruising trips.

Source: FHWA.

Figure 21. Graph. Cruising frequency and overall trip making.

Downtown and Buckhead

The analysis in this section is of street-level data; it provides a close up of what is happening in downtown Atlanta and Buckhead. The streets were selected for the characteristic of having mixed land use. Figure 22 exploits the block-specific trip end data to show, for the smaller area, where trips end. The highest trip density is heavily concentrated in the center of downtown. Figure 23 shows the frequency of cruising. Almost no cruising occurs where the most trips end; instead, there are small pockets of high-intensity cruising scattered throughout downtown with some of the most intense cruising hot spots located in Buckhead. The baseline average weekday rate of cruising in the sub area is 7.4 percent. The average time spent looking for parking is about 2 minutes.

Source: FHWA.

Figure 22. Map. Atlanta area of detail trip ends.

Source: FHWA.

Figure 23. Map. Atlanta area of detail cruising trip ends.

Downtown Atlanta cruising follows a similar pattern to citywide cruising, showing a slight lag in the morning. Both trips and cruising trips peak in the morning, slowly decline at midday, and reach a muted peak in the afternoon (Figure 24). The data are presented using a rolling average to allow an easier visualization in an otherwise more erratic presentation that would have been due to the smaller data set.

Source: FHWA.

Figure 24. Graph. Diurnal distribution of trips.

Figure 25 shows the diurnal distribution of trips, with a line showing the proportion of trips that are cruising throughout the day. The cruising rate in downtown Atlanta and Buckhead is relatively flat, ranging from around 6 to almost 9 percent of trips. This is shown in Figure 25.

Source: FHWA.

Figure 25. Graph. Proportion of trips cruising.

Comparison Case: April 1–September 30, 2020

After mid-March 2020, many U.S. cities instructed their residents that only trips defined as essential travel should be taken. Trip making was highly curtailed. The next section compares travel during this period with the baseline described above.

As illustrated in figures 26, 27, 28, and 29 during the comparison period, different patterns for trip destinations and cruising emerged. Even with patterns shifting, cruising remained relatively stable, at 4.7 percent, citywide (compared to 4.9 percent) and 7.9 percent in the area of detail (an apparent, but statistically insignificant, increase from the baseline 7.4 percent) in the mixed use areas of downtown and Buckhead.

Figure 26 corresponds to the baseline representation of Figure 18 to show cruising by census tract in the April–September period. Figure 27 corresponds to Figure 19 illustrating cruising impact (i.e., cruising trips weighted by the amount of time spent cruising and normalized by street length for block groups). Compared to the baseline maps, there are some visible shifts. The primary differences appear in the downtown area and in Buckhead. After a presentation of citywide trip-making and cruising summaries, a closer look is taken in these areas using the street-level data.

Source: FHWA.

Figure 26. Map. Cruising trip ends, April–September, 2020.

Source: FHWA.

Figure 27. Map. Cruising impact, April–September, 2020.

In the period of initial restricted travel, the diurnal distributions of trips and cruising trips line up quite closely. This is shown in Figure 28. Unlike the historical pattern of a morning and an afternoon travel peak, the Atlanta data show trip making increases sharply in the morning from 5 or 6 to around 8 a.m. Thereafter, trip making remains relatively steady, displaying a somewhat flat distribution until around 6 p.m., when the share of trips made declines quite steadily. Figure 33 superimposes cruising trip rates with time of trip making. Except from midnight to approximately 4 a.m., where the proportion of trips cruising is relatively high, the proportion remains relatively stable throughout the day, staying consistently between 3 and 6 percent of trips.

Source: FHWA.

Figure 28. Graph. Diurnal distribution of trips and cruising trips, April–September, 2020.

Source: FHWA.

Figure 29. Graph. Diurnal distribution of trips and rate of cruising, April–September, 2020.

Relative to the baseline, the trips in the comparison period are far more evenly dispersed across the area of detail (Figure 30). The intense destination indicated in the center of downtown in the baseline map (Figure 22) no longer exists.

Looking at Figure 31, which illustrates cruising hot spots, it appears that problematic locations for cruising in the baseline have also redistributed to the periphery of the downtown area.

Source: FHWA.

Figure 30. Map. Trip ends area of detail, April–September, 2020.

Source: FHWA.

Figure 31. Map. Cruising area of detail.

The final section in this use case looks at the diurnal distribution of trips and cruising trips, and the rate of cruising on the mixed use streets in Atlanta.

When focusing on the mixed use streets, it appears that trip making follows the citywide pattern, in that traditional peaks are blunted with one peak period lasting from around 6 a.m. to around 6 p.m. The existence of this peak does not imply congested streets for the period—as peak often implies—rather, that trips are concentrated in this time block with no apparent other peaking feature. Cruising trips are tracking exactly with all trips until about 8 a.m., after which they are more highly concentrated. At about 1 p.m. the daily share of cruising trips begins to fall below the rate of all trips, though both are declining.

Figure 32 shows a pattern of greater variation in the cruising rate, compared to the same geography in the baseline and compared to the citywide averages. Figure 33 shows the diurnal distribution of trips with cruise rate superimposed.

Source: FHWA.

Figure 32. Graph. Diurnal distribution of trips and cruising trips, April–September, 2020.

Source: FHWA.

Figure 33. Graph. Diurnal distribution of trips with cruise rate superimposed, April–September, 2020.

Summary

This section described trip making and cruising in Atlanta, focusing on mixed use areas and a comparison between the October 2019–March 2020 and April–September 2020 time periods. There is no particular policy intervention that had been implemented in the time frame. The study team looked instead at areas characterized as mixed use, relative to the rest of the city, and two distinct time periods marked by a change in direction with respect to essential travel.

Top line findings indicate:

• More cruising in downtown Atlanta and Buckhead, relative to the rest of the city
• Shifts in trip destinations during the second period with far less intensity on the downtown
• Cruising hot spots do not always correspond with areas of greatest trip making
• Cruising hot spots during spring and summer 2020 migrated away from Buckhead and redistributed around downtown

Chicago, Illinois: Year Over Year and Time of Day

For the Chicago use case, raw location data were purchased for June 2019, 2020, and 2021. From the location data, trips were inferred and analyzed as described earlier in this report. Differences are expected from the pre-COVID to the early COVID periods and additional differences to late COVID. Unfortunately, the data for June 2021—the late COVID period—were of insufficient quality to include in the final analysis. This provides an important caution for users of the tool. The deficiency was not important for illustration of a use case, but cautionary where the analysis is centered on the geography, in this case Chicago. Additional data should be acquired to perform the additional desired or required analysis. The analysis proceeds here with a focus on the 2 years for which data are of sufficient quality.

Where Cruising Occurs

With the full data set of mapmatched and cruising-identified traces, high cruising areas can be identified. Figure 34 and Figure 35 show the cruising rate for each street segment for the 2 months of data. In other words, the figures show for each trip that traveled along a street segment, what percentage of those were cruising trips. Overall, many of the same areas stand out, such as the West Loop, Pilsen, Lakeview, Hyde Park, and the area around the convention center. A couple high cruising areas located on the periphery of the city, such as the hot spots in South Chicago and Riverdale neighborhoods, appear to be noise due to a low quantity of trips in the data set.

Source: FHWA.

Figure 34. Map. Cruising 2019.

Source: FHWA.

Figure 35. Map. Cruising 2020.

The year-over-year comparison is complicated by the fact that the distribution of trips across the 2 years suggests a different geographic market penetration of the applications from which data are collected as well as differences in trip patterns. There are about 40 percent more trips in the 2020 data set (115,000 in June 2019 and 165,000 in June 2020). As Figure 36 shows, the location of trip ends is quite different. In June 2019, trips were heavily concentrated downtown, the Near North Side, and Northwest Side; all are predominantly white and affluent areas. While the pandemic may explain fewer trips into downtown or to points of interest in some North and Northwest Side neighborhoods, it would not explain why so many more trips are occurring in other neighborhoods. It is likely, therefore, that the year-over-year differences between June 2019 and June 2020 are likely biased by the mix of unknown apps present in the data. The broader distribution of trips in 2020 suggests a more distributed data collection base.

Source: FHWA.

Figure 36. Map. Comparison of trip ends 2019 and 2020.

With the caveat noted, Figure 37 shows the neighborhood changes in frequency from 2019 to 2020.

Source: FHWA.

Figure 37. Map. Change in cruising frequency.

When Cruising Occurs

Citywide, cruising remains between approximately 6.5 and 7.5 percent for most of the day. For trips ending on metered streets, which was 67 percent of trips in 2019 and 60 percent in 2020, the rate is between 8 and 9 percent between 10 a.m. and 8 p.m., with a dip to 7 percent at 1 p.m. For trips ending on non-metered streets, after a peak of 7.5 percent cruising between 8 and 9 a.m., cruising stays between 5 and 6 percent for most of the day. The volatility observed in the early morning hours is likely due to the low number of samples during those times. The higher variance noted in Figure 38, Figure 39, and Figure 40 supports that hypothesis; that figure set illustrates cruising frequency and overall trip making citywide, on metered streets and on non-metered streets, respectively.

Source: FHWA.

Figure 38. Graph. Diurnal distribution of trips and cruising trips.

Source: FHWA.

Figure 39. Graph. Diurnal distribution of trips and cruising trips ending on metered streets.

Source: FHWA.

Figure 40. Graph. Diurnal distribution of trips and cruising trips ending on unmetered streets.

Just as with the geographic distribution of trips, the differences in the diurnal distribution of trips between June 2019 and June 2020 may be genuine or may be caused by differences in the underlying data sources. In June 2019, the data show a sharp peak in overall weekday trips at 5 p.m., while in June 2020, the peak is more spread out, from 2 to 5 p.m. (Figure 41).

Source: FHWA.

Figure 41. Graph. Diurnal distribution of weekday trips.

Cruising frequency remains relatively constant throughout the day in both June 2019 and June 2020, though it is slightly higher in June 2020 (Figure 42). Despite a lower frequency of cruising trips in June 2019, however, as shown in Figure 43, the mean cruising time is consistently higher in June 2019 than in June 2020. This finding suggests that, as trips are more distributed, cruising locations are more distributed and have less intensity.

Source: FHWA.

Figure 42. Graph. Cruising frequency by time of day 2019 and 2020.

Source: FHWA.

Figure 43. Graph. Mean cruising time 2019 and 2020.

Do Cruising Patterns Change throughout the Day?

To mitigate concerns about the respective data sources for different years, the unaggregated data from Chicago allow for a high-resolution analysis of diurnal patterns within the same data set; the following analyses rely on the 2020 data set. Trips are classified into three time frames corresponding to traditional morning and evening peak periods and a midday period. The periods are: morning (7–10 a.m.), midday (11 a.m.–3 p.m.), and evening (4–7 p.m.). For the following analysis, the more evenly distributed June 2020 data are used. While some areas experienced similar levels of cruising throughout the day, in general cruising tended to be more concentrated to specific destinations in the morning, and most dispersed at midday (Figure 44). The same mapping analysis is shown below for the West Loop (Figure 45); River North (Figure 46) Hyde Park (Figure 47), and Lakeview (Figure 48).

Source: FHWA.

Figure 44. Map. Peak and midday comparison of cruising.

Source: FHWA.

Figure 45. Map. West Loop cruising peak and midday.

The West Loop, a neighborhood home to both wholesale food distributors and trendy restaurants and only one train stop, experienced some of the highest rates of cruising across the board. Parking is largely unregulated in the area, except along Madison Street, which can be identified by the series of meters running the length of the street, and in the southeast corner of the neighborhood, where cruising is lower.

Source: FHWA.

Figure 46. Map. River North cruising peak and midday.

The neighborhoods north of the Loop have some of the highest densities of retail, office, and residential space in the city. Most of these areas have metered streets, with the exception of the relatively more residential areas of Streeterville and the Gold Coast. The diurnal pattern of cruising here would be consistent with an influx of workers in the morning looking for free and/or all-day parking in non-metered areas near their workplaces, and, in the evening, Gold Coast residents returning home and competing over limited street parking.

Source: FHWA.

Figure 47. Map. Hyde Park cruising peak and midday.

In Hyde Park, the diurnal patterns show that throughout the day, cruising remains high along the northeast lakefront section of the neighborhood—an area of concentrated high-rise apartments, in contrast to the three-story multifamily buildings found throughout most of the neighborhood. The commercial center of the neighborhood around 53rd Street, East Hyde Park Boulevard, and Lake Park Avenue, also has high rates of cruising during midday and evening hours. In the morning, high cruising can be found in the southern part of the neighborhood near the University of Chicago, and in particular the University of Chicago Laboratory School, a private school serving students from kindergarten to 12th grade, which may reflect parents dropping off children, or university faculty and staff searching for parking.

Source: FHWA.

Figure 48. Map. Lakeview cruising peak and midday.

In Lakeview, cruising is highest in the evening near the busy entertainment corridor of North Halsted Street and North Broadway. Cruising is also high to the north in the Uptown neighborhood, where higher density housing is found.

Metered Streets

Chicago has metered parking along most commercial corridors throughout the city. In the 2020 data set 60 percent of trips and 55 percent of cruising trips ended on metered streets. For those trips ending on metered streets, 55 percent of cruising vehicle miles traveled were done on metered streets. For cruising trips ending on non-metered streets, only 9 percent of cruising miles accrued to metered streets. This could indicate that cruising drivers with trips ending on metered blocks are looking for open metered spaces, and not necessarily searching for non-metered spaces. Likewise, cruising drivers ending their trips on non-metered streets are likely looking for non-metered spots.

At a citywide scale, the relationship between cruising and metered parking may depend on the neighborhood-specific land use context. In specific neighborhoods, such as the West Loop, Streeterville, and Gold Coast, areas with high frequency of cruising can be seen in unmetered areas adjacent to metered areas. These areas are both adjacent to, or even part of, the central business district (CBD), where the majority of streets are metered, rather than just the main commercial corridors. In neighborhoods farther out, such as Hyde Park, high rates of cruising along metered streets may indicate the high demand for parking that the meters are there to regulate.

Another temporal filter can be applied to examine cruising during hours when Chicago’s parking meters are turned on and off. Chicago has different meter hours depending on the neighborhood, and parking meters in the Loop are in operation 24 hours per day. Figure 65 shows cruising frequency when trips end in places and at times meters are turned on, and in places and at times when they are turned off. When meters are turned on, cruising is highest in the two adjacent areas to the north and west of the CBD where there are fewer meters present. This could indicate people specifically avoiding metered areas and parking as close as possible to the core, or it could indicate that a lack of parking regulation in these areas is causing a higher demand than it might be if more meters were present.

When meters are turned off, the highest cruising is found in metered areas in River North and just south of the Loop that is not present when meters are turned on.

Source: FHWA.

Figure 49. Loop cruising meters on and meters off.

Summary

The Chicago case is based on raw location data that the project team analyzed directly. Having the individual trip trajectories allows a more complex study of the cruising paths taken, rather than simply basing findings on where trips end. The overall cruising rate in Chicago fluctuates between 6.5 and 7.5 percent, with higher concentrations in the denser and mixed use areas. The cruising rate is relatively stable throughout the day in both years analyzed, but shows some geographic variation across the day: it is more evenly distributed throughout the city in the morning period and more geographically concentrated in the afternoon. Finally, cruising has increased in duration in June 2020 relative to 2019.

Two important cautions arise from this part of the project. First, flexibility and adaptability in research design are a prerequisite for creating and presenting impactful analysis. The Chicago analysis was meant to be of 3 consecutive years’ worth of data. Poor data quality in the final year prevented a complete analysis. Had this been a policy study, additional data should have been sought or a different time period substituted. Second, the close analysis of raw data over a relatively abbreviated time frame indicates there are likely biases due to data collection strategy. If collection is via particular applications that have more popularity among certain demographic groups that will likely also show geographic biases in the location of trip making and trip ends. With these kinds of biases an analyst may be unable to identify the worst locations for cruising, but could get a critical sense of cruising dynamics in particular areas. Apparent missing data could inform data acquisition strategies.

Seattle, Washington: Two Meter Policies

Seattle has adopted a performance pricing policy for its metered parking streets. Using annual surveys to estimate vehicle occupancy, the Seattle Department of Transportation (SDOT) raises, lowers, or leaves the same the parking meter prices in order to hit an occupancy/vacancy target.

The first Seattle case examines cruising before and after price changes to the city’s metered parking. The business-as-usual case is examined wherein SDOT does a regularly scheduled meter price adjustment. SDOT made a price adjustment on January 28, 2020; the period of review is before and after the price change.

The second case examines cruising in the 2 weeks before and after SDOT temporarily suspended all parking meters on April 4, 2020. The city decided to suspend meters to ease the travel burden on essential workers.

For the purposes of both analyses the city was divided into metered streets, near-metered streets—on the assumption that meter policy can have spill-over effects on other streets—and non-metered areas. Due to the data agreement with the third-party processor, one of the Seattle data sources, the number of geographies was limited to 15,000, so trips not ending within 0.5 miles of metered streets were aggregated to the census block group level. Figure 50 shows Seattle’s block groups along with metered and near-metered streets.

Most of the analysis in this case uses output created by the third-party processor. Raw location data for Seattle were also obtained and used in these use cases only where noted. Additional discussion related to the use of different data sources generally, and in Seattle specifically, is provided in Chapter 4 and Appendix A.

This section discusses the following:

• Baseline cruising and trip making in Seattle using data in the period January 6–February 28, 2020
• Before-after analysis of the business-as-usual meter price change that occurred on January 28, 2020
• Comparison of the baseline to data obtained for March 21–April 18, 2020
• Analysis of cruising before and after meter prices were suspended

Source: FHWA.

Figure 50. Map. Seattle study area.

Where Cruising Occurs

Census block groups with the highest density of cruising trips tend to be in metered districts (shown in outline) in or near the commercial business district, university district, and the neighborhoods of Adams, and Fremont. High rates of cruising are also found in parts of west Seattle, neighborhoods along Martin Luther King Jr. Way South, and in Lake City. Trips in these areas tend to be of shorter duration as proximate parking is likely easier to find (Figure 51).

Source: FHWA.

Figure 51. Map. Cruising frequency.

Not all cruising trips are the same. A cruising trip could take drivers a little out of their way or far out of their way; thus, time spent cruising should be considered along with cruising frequency. Figure 52 shows the aggregate level of cruising as aggregate time spent cruising for trips that end in each block group, normalized by the length of street in each block group. When considering total time spent cruising the problem areas align well with metered areas, suggesting that Seattle has appropriately placed its meters where pricing is needed as a curb management strategy. In this case, the map reflects where trips end, not the paths taken by drivers; though it appears counterintuitive, a plausible interpretation is that trips with a circuitous component are ending on metered blocks with greatest frequency. This could be that drivers are finding vacant spaces on metered blocks where they had not on other blocks. This is an affirmation of the meter policy that could be abstracted to the blocks where people prefer to park. It could also mean that drivers are looking for bargain parking, and when they cannot find it, drivers take paid parking as a second choice. Alternatively, this could be a result of cruising occurring in the metered areas where, policy goals notwithstanding, meter rates may have been too low to bring demand down to the level of supply.

Source: FHWA.

Figure 52. Map. Cruising impact area.

There is also a potentially important boundary effect. With the exceptions of cruising hot spots that are completely contained within the metered area, proximate to Pike Place Market and south of Lake Union, most of the commercial business district cruising is at the boundary of the metered and non-metered areas. This observation could warrant further thought regarding expansion of metered areas to discourage bargain hunting that may occur.

Source: FHWA.

Figure 53. Map. Cruising for parking boundary effects.

When Cruising Occurs

Citywide, cruising trips follow a similar diurnal distribution to overall trips (Figure 54), with morning and evening peaks. Cruising trips, as a proportion of all trips, ranged between 4.5 and 8.5 percent throughout the day, with peaks at midday and in the early evening toward the end of the evening peak travel period (Figure 55). Both peaks follow the two peaks in the overall trip numbers, suggesting that as morning commuters park near their workplaces and evening commuters park near their homes or near evening activities, available parking spaces fill up, leading to higher rates of cruising for parking.

Source: FHWA.

Figure 54. Graph. Diurnal distribution of trips and cruising trips.

Source: FHWA.

Figure 55. Graph. Diurnal distribution of trips and cruising as percent of trips.

In the metered and near-metered areas shown in Figure 56, the cruising lag is more pronounced than it is across the city. As shown in Figure 57, the rate of cruising is at its highest around noon, remaining high (between 9 and 12 percent of trips) until around 10 p.m.

Source: FHWA.

Figure 56. Graph. Diurnal trips and cruising on metered and near-metered streets.

Source: FHWA.

Figure 57. Graph. Diurnal distribution of trips and cruising frequency on metered and near metered streets.

For trips that end more than 0.5 miles from metered streets (i.e., non-metered areas hypothesized to be largely immune to meter pricing policy and the demands of metered and near-metered areas), the midday and evening peaks observed in other areas do not materialize. Between 9 a.m. and midnight, cruising frequency on non-metered streets hovers around 6 percent, varying only by just over 1 percent (Figure 58).

Source: FHWA.

Figure 58. Graph. Diurnal distribution of trips and cruising frequency in non-metered areas.

A disproportionate number of all trips, as well as cruising trips, end on metered streets (Table 2). Metered streets make up only 2 percent of overall streets in Seattle, yet account for 14 percent of trip ends and 18 percent of cruising trip end points. Near-metered streets are 10 percent of all streets, 19 percent of trip ends, and 24 percent of cruising end trips. Calculating the ratio of cruising trips to all trips, metered and near-metered streets show similar ratios of 1.24 and 1.22, respectively. This is unsurprising since metered streets are located where there is a higher density of destinations and a greater demand for parking.

As shown in Table 3, vehicles cruising in the non-metered area, a relatively rare event, spend, on average, under a minute searching for parking. On the other hand, a cruise trip which is more prevalent in the metered areas, takes longest there, with near-metered areas running a close second.

Metered and near-meter streets have average cruising times 2–3 times higher than cruising trips that end on non-meter streets.

Comparison: Before versus after Business-as-Usual Meter Price Change

Consistent with SDOT policy, the January 28, 2020, meter price change entailed either a 50-cent increase, 50-cent decrease, or no change, depending on the observed occupancy on a particular block.²³ This case examines whether this price change had an effect on cruising behavior. The date range representing the before condition is January 6–27, 2020. The period of analysis representing the after condition is February 5–28. 2020.

Figure 59 shows the frequency and changes in frequency of cruising on metered, near-metered and non-metered blocks. Metered blocks are divided according to whether the price was increased, decreased, or held constant. Near-metered streets were classified for price increases, decreases, or being held constant based on the policy treatment of the metered block to which they are nearest.

Figure 60 shows the time spent cruising in the metered, near-metered, and non-metered areas. The working hypothesis for non-metered areas (i.e., those beyond half a mile of a parking meter) is that travel behavior would not be affected by meter price changes. Supporting that hypothesis, Figure 60 shows there is no change in frequency of cruising trips ending on blocks in non-metered areas across the price change.

The changes in cruising on streets where prices either increased or decreased were small. Where meter prices were reduced in Seattle and on near-metered streets proximate to streets with metered price drops, the frequency of cruising trip ends also decreased. The decreases are small, and likely due to sampling error. The error bars in Figure 59 show the potential statistical overlap. On streets where the price was increased—in theory, creating more vacancies—the number of trips cruising also increased. This should not be interpreted to mean that price increases stimulated additional cruising, but rather that more cruising trips are ending on streets where prices were increased because there is more turnover and related availability on those streets. In both cases (i.e., whether meter prices were increased or decreased), the amount of time a driver spent cruising decreased. Limitations of the data only allow observation of where the trip ends. A plausible interpretation is that with increased vacancy, more trips that would have been cruising before the price adjustment no longer were—those cruising trips would have ended on other blocks or would have taken a longer time. The increase in cruising trips that ended on metered streets with price increases is statistically significant; this suggests a real change had been measured. Streets proximate to metered streets that had price increases show a decrease in cruising trips, which, again, suggests parking demand is being better accommodated on metered blocks.

Source: FHWA.

Figure 59. Graph. Change in cruising by meter price change and area type.

Source: FHWA.

Figure 60. Graph. Time spent cruising by meter price change.

As the differences are small, albeit consistent, it can take several adjustments before a travel behavior change registers. A study of San Francisco showed a change in cruising behavior only after eight price changes.²⁴ Although the change in cruising frequency is not significant, the fact that cruising frequency is higher in areas that had a price increase confirms SDOT’s information on where and when there is higher demand for parking.

Cruising Pattern Changes in Response to Meter Decommissioning

In the earliest days of the pandemic, travel patterns changed substantially. In the pre-pandemic period, trips followed the expected diurnal distribution with pronounced peaks in the morning and late afternoon. By late March–April 2020, however, travel was substantially curtailed and it also showed minimal peaking. Trips were low at night and in the early morning, but relatively flat between 8 a.m. and 3 p.m., dropping off sharply from there. In response to the COVID pandemic, SDOT temporarily suspended all parking meters and time limits, except for residential permit zones. This section focuses on an analysis of cruising and travel patterns in the 2 weeks before and after the meter suspension on April 4, 2020. To set the context for the meter decommissioning, this section first covers the changes in cruising behavior after the initial COVID-19 lockdown in March, 2020.

Figure 61 shows the diurnal distribution of trips before the Government’s guidance to social distance and restrict travel and the distribution of trips after that guidance. Figure 62 shows the traditional morning and evening peaks for cruising in the before period and the reduced and flattened pattern in the after period.

Source: FHWA.

Figure 61. Graph. Diurnal distribution of baseline trips and early lockdown trips.

Even at one-third of pre-pandemic trips, cruising frequency was curiously consistent. The percentage of trips that included a cruising component in the January–February 2020 time period ranged from 4 to 8 percent, with an average daily rate of 7.1 percent. During the early pandemic, frequency of cruising trips ranged between about 4 and 8 percent, with an average daily rate of 6.1 percent. This represents a statistically significant decline, but only of 1 percent.

Source: FHWA.

Figure 62. Graph. Cruising frequency in early 2020 and in early spring 2020.

The disaggregated data allow for a detailed spatial snapshot of cruising patterns, and how they shifted between winter and spring 2020. Figure 63 shows the rate of cruising along street segments in the winter time period (January 6–February 28) and in the spring time period (March 21–April 18). Mirroring the aggregated output shown above, cruising rates are only slightly lower in the spring, despite the large decrease in overall trips. While many residential areas, in addition to areas near the University of Washington and east of downtown (home to several hospitals), maintained similar cruising rates, neighborhoods with more jobs than residents (downtown and the heavily industrial areas south of downtown) saw cruising decrease significantly.

Source: FHWA.

A. Seattle cruising hot spots,
January–February 2020.

B. Seattle cruising hot spots,
March–April 2020.

Figure 63. Comparison of cruising hot spots.

On April 4, 2020, SDOT temporarily suspended all parking meters as a relief measure for essential workers. SDOT also temporarily lifted existing time constraints on parking at the meters and generally throughout the city. The exception was in residential permit parking areas where time limits continued to be enforced.

Source: FHWA.

Figure 64. Graph. Cruising rates before and after meter suspension.

The frequency and time spent cruising before and after the suspension were compared. Citywide, the decision to decommission the parking meters in this period had no discernible impact on cruising frequency. As before, there was no change in average cruising time in the non-metered areas, which serves as a control. Average cruising time decreased slightly on both metered and non-metered streets once the meters were decommissioned. However, the difference is not statistically significant (Figure 65). It is possible that travel and cruising behavior were not responding to meter prices during such an acute point in the early pandemic.

Source: FHWA.

Figure 65. Graph. Average time spent cruising before and after meter suspension.

Summary

Two parking meter policies were analyzed in this section. First, a business-as-usual case was considered in which SDOT adjusted meter prices according to an annual occupancy survey. Second, meter suspension was implemented as part of the city’s response to the pandemic. In the first case, there was a consistent decrease in the amount of time a cruising vehicle spent looking for parking after the meter rates had been adjusted. In the second case, even though trip-making rates had been substantially lowered, compared with the baseline, cruising was consistent, and differences before and after the meter suspension were indiscernible.

The Seattle analysis relies on aggregated data outputs of the third-party processor supplemented with raw location data from the data aggregator. The two approaches are taken to showcase the relative strengths of each. The third-party data are more abundant providing greater statistical confidence. The location data allow for more complex analyses of paths taken. Each has a potential function. A comparison of the data sources is provided in appendix A.

Finally, with respect to the apparent lack of difference in trip making and in cruising following the meter policy changes, it is possible that a 2- or 3-week period before and after a policy change may be insufficient time in which the market adjusts and differences can be observed.