Appendix A

Traffic Congestion

According to one measure developed to compare congestion on facilities throughout a metropolitan area, congestion increased by 16 percent between 1982 and 1990 in 50 of the largest U.S. metropolitan areas (Shrank et al. 1993). Congestion in this study was approximated by an index developed by researchers at the Texas Transportation Institute (TTI); as measured by this index, half of the 50 metropolitan areas tracked had reached undesirable levels of congestion by the end of the decade (Table A-1).

This index of metropolitan area congestion, however, is illustrative only. Like all other available measures of congestion based on existing data, the index is an indirect measure of the congestion occurring on metropolitan roadways. In fact, there is no good measure of urban traffic congestion that is comparable across areas and that has been collected consistently over time. Indeed, one other indirect measure of congestion that has been collected over time does not correspond with the upward trend shown by the TTI index (Gordon and Richardson, Vol. 2). As described in more detail below, aggregate trend data on the average duration of the journey to work imply that speeds experienced by the average commuter have not worsened. Gordon and Richardson (Vol. 2) hypothesize that the relative stability in journey-to-work duration indicates that, over time, motorists and businesses make rational choices to avoid delays. These choices include changing location, typically by moving from an urban to suburban or exurban site (Gordon and Richardson, Vol. 2). Nonetheless, the trends from these two data sources are inconsistent: one implies that congestion is worsening, the other implies that it is not. Which is right?



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 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Appendix A Traffic Congestion According to one measure developed to compare congestion on facilities throughout a metropolitan area, congestion increased by 16 percent between 1982 and 1990 in 50 of the largest U.S. metropolitan areas (Shrank et al. 1993). Congestion in this study was approximated by an index developed by researchers at the Texas Transportation Institute (TTI); as measured by this index, half of the 50 metropolitan areas tracked had reached undesirable levels of congestion by the end of the decade (Table A-1). This index of metropolitan area congestion, however, is illustrative only. Like all other available measures of congestion based on existing data, the index is an indirect measure of the congestion occurring on metropolitan roadways. In fact, there is no good measure of urban traffic congestion that is comparable across areas and that has been collected consistently over time. Indeed, one other indirect measure of congestion that has been collected over time does not correspond with the upward trend shown by the TTI index (Gordon and Richardson, Vol. 2). As described in more detail below, aggregate trend data on the average duration of the journey to work imply that speeds experienced by the average commuter have not worsened. Gordon and Richardson (Vol. 2) hypothesize that the relative stability in journey-to-work duration indicates that, over time, motorists and businesses make rational choices to avoid delays. These choices include changing location, typically by moving from an urban to suburban or exurban site (Gordon and Richardson, Vol. 2). Nonetheless, the trends from these two data sources are inconsistent: one implies that congestion is worsening, the other implies that it is not. Which is right?

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion TABLE A-1 Roadway Congestion Levels in 1989 for 50 Metropolitan Areas (Shrank et al. 1993) Urban Area Roadway Congestion Index Rank Urban Area Roadway Congestion Index Rank Los Angeles, Calif. 1.55 1 St. Louis, Mo. 0.99 25 Washington, D.C. 1.37 2 Cleveland, Ohio 0.97 27 San Francisco–Oakland, Calif. 1.35 3 Cincinnati,Ohio 0.96 28 Miami, Fla. 1.26 4 Norfolk, Va. 0.96 28 Chicago, Ill. 1.25 5 Austin, Tex. 0.94 30 San Diego, Calif. 1.22 6 Ft. Lauderdale,Fla. 0.94 30 Seattle–Everett, Wash. 1.20 7 Jacksonville,Fla. 0.94 30 San Bernardino–Riverside, Calif. 1.19 8 Albuquerque, N.Mex. 0.93 33 New York, N.Y. 1.14 9 Minneapolis–St. Paul, Minn. 0.93 33 Houston, Tex. 1.12 10 Memphis, Tenn. 0.91 35 New Orleans, La. 1.12 10 Fort Worth, Tex. 0.90 36 Atlanta, Ga. 1.11 12 Hartford, Conn. 0.89 37 Honolulu, Hawaii 1.11 12 Nashville, Tenn. 0.89 37 Detroit, Mich. 1.09 14 San Antonio, Tex. 0.88 39 Portland, Oreg. 1.07 15 Louisville, Ky. 0.86 40 Boston, Mass. 1.06 16 Salt Lake City, Utah 0.85 41 Dallas, Tex. 1.05 17 Columbus, Ohio 0.83 42 Philadelphia, Pa. 1.05 17 Indianapolis,Ind. 0.83 42 Tampa, Fla. 1.05 17 Pittsburgh, Pa. 0.8 44 San Jose, Calif. 1.04 20 Oklahoma City,Okla. 0.79 45 Denver, Colo. 1.03 21 Charlotte, N.C. 0.78 46 Phoenix, Ariz. 1.03 21 El Paso, Tex. 0.74 47 Sacramento, Calif. 1.02 23 Kansas City, Mo. 0.74 47 Baltimore, Md. 1.01 24 Corpus Christi, Tex. 0.72 49 Milwaukee, Wis. 0.99 25 Orlando, Fla. 0.72 49 NOTE: Based on the index used by Shrank et al. (1993), a score of 1 indicates that an area has undesirable levels of areawide congestion.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion The apparent inconsistency between the TTI index and trend data on commuting trips can be explained in three ways: (a) neither the TTI index nor the journey-to-work data measure congestion directly, (b) they are actually measuring different things (traffic volumes on specific facilities versus durations of work trips), and (c) the origins or destinations, or both, of work trips are changing over time in response to congestion on facilities. Because of these differences, one would not expect the two trends to be entirely consistent. A brief overview is provided here of these three explanations for the apparent inconsistency between the two measures of congestion. As background for this review, the broad trends affecting travel demand in metropolitan areas are summarized first. Second, alternative measures of congestion are reviewed. Finally, the reasons for the apparently paradoxical findings between these alternative, but indirect, measures of congestion are summarized. TRAVEL DEMAND Total metropolitan area travel in the 1980s grew sharply, which has led to considerable concern about growing traffic congestion. Between 1980 and 1990, vehicle kilometers of travel in urban areas grew almost 50 percent, whereas urban population grew only 12 percent (Gordon and Richardson, Vol. 2). Many metropolitan areas have shown double-digit population growth over the last decade, but even in areas without increased population, more women have joined the work force, job growth has increased, and population and employment growth has continued to shift to outlying areas of metropolises (Pisarski 1987; Gordon and Richardson, Vol. 2). The amount of traffic growth engendered by these changes has increased faster than population in almost all of the largest metropolitan areas (Table A-2). (The measure of increased traffic growth per lane-kilometer in Table A-2 takes into account the expansions of highway capacity.) Even metropolitan areas with little population growth, such as Chicago, New York, Philadelphia, and Boston, show double-digit increases in traffic as more Americans live in suburban residential areas, work in other parts of the metropolitan area, and increasingly rely on the automobile to travel between them.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion TABLE A-2 Trends in Congestion in the 10 Largest Metropolitan Areas (Hanks and Lomax 1990; Shrank et al. 1993) Metropolitan Areaa Congestion Ranking in 1990 Travel Growth, 1982–1990 (% change) Population Growth, 1982–1990 (% change) DKT per Lane-Kilometer, 1982–1990 (% change) New York 9 31.3 0.7 16.5 Los Angeles 1 46.2 15.4 27.2 Chicago 5 49.4 6.1 25.9 Philadelphia 17 48.1 3.7 22.6 Detroit 14 12.1 5.0 -2.4 San Francisco–Oakland 3 47.5 10.5 35.8 Washington, D.C. 2 57.5 27.0 28.1 Boston 16 35.8 3.9 26.1 Houston 10 33.9 19.5 -4.1 San Diego 6 83.6 29.2 61.8 NOTE: DKT = daily vehicle kilometers traveled. 1 km = 0.6 mi. a In order by 1990 population.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Work Trips During the 1970s and into the 1980s, the suburban rings around metropolitan areas experienced dramatic increases in workers, and the subsequent increases in commuting trips greatly increased peak-period travel (Cervero 1986; Pisarski 1987). Between 1970 and 1980, for example, Houston's population grew 40 percent and its suburban employment mushroomed 90 percent (Pisarski 1987). These employment increases were fueled by the rapid development of office and industrial parks outside urbanized areas—a phenomenon that has occurred all over the country. “Already, two thirds of all American office facilities are in [outlying suburban areas], and 80 percent of them have materialized in the last two decades” (Garreau 1991). The “edge cities” of New Jersey that abridge New York City, for example, have more office space than Manhattan (Garreau 1991). Rapid job growth in outlying areas of central cities quickly resulted in more jobs in the suburban rings than in the urban core—and in much more dispersed locations. By 1980, commutes from suburb to urban areas—the popular notion of the typical journey-to-work trip—represented only 18 percent of commute trips for metropolitan areas with 500,000 or more population. Suburb-to-suburb trips accounted for 40 percent of commute trips by 1980 and were growing at over twice the rate of suburb-to-central-area trips (Pisarski 1987, 43). These trips are also shorter in duration than other commute trips within metropolitan areas (Gordon and Richardson, Vol. 2, Table 2). The trips may be longer in distance but are completed at higher speeds than commute trips within central cities or commute trips from suburbs to central cities. Preliminary data from the 1990 census on mode choice and average travel times within individual metropolitan areas indicate a continued strong increase in the choice of the automobile for the journey to work and the growing preference for driving alone—even in highly congested areas like Los Angeles, San Francisco, and Washington, D.C. (Pisarski 1992). The percentage of workers driving alone to work increased from 64.4 percent in 1980 to 73.2 percent in 1990 (DOT 1993). The continued popularity of automobile commuting, even when more time is apparently being spent in congested conditions, traces to several trends. In addition to the increased suburbanization of jobs, other important changes include the continued preferences for the privacy and convenience of automobile travel and lower-density suburban locations for housing (Downs 1992). The declining real cost of gasoline during the last decade has made the

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion choice of automobile travel all the more attractive, as have subsidies that encourage use of the automobile. The deep subsidies that come in the form of employer-provided parking, which is usually costless to the employee, encourage solo driving (Shoup, Vol. 2). The provision of infrastructure through property taxes and other nontransportation revenue sources further masks the true cost of driving automobiles, which thereby encourages their use (Hanson 1992). Nonwork Trips The previous discussion of travel demand focused on peak-period travel and work trips. The continued suburbanization of jobs and residences has also coincided with a pronounced increase in nonwork travel (Gordon et al. 1988). Moreover, nonwork trips during the peak are growing much faster than any other category. Between 1977 and 1983, for example, total trips increased by 16.3 percent, but nonwork trips by private automobile in the morning peak period increased by a whopping 42 percent (Gordon et al. 1988). This growth in nonwork peak travel by automobile is most pronounced among suburban residents and is most closely associated with two-worker households and families with children. Most surprisingly, about half of peak-period travel is now made up of nonwork trips.1 Summary Underlying the sharp increases in travel demand in recent years are some important trends. The entry of women into the work force has caused travel demand to outpace population growth. The suburbanization of jobs and continued suburbanization of residences have caused metropolitan 1   As argued by Gordon et al. (1988), given that more than half of peak-period trips are not work trips, the case for congestion pricing is all the stronger. Presumably the demand for nonwork trips is more elastic than that for work trips. Hence a substantial fraction of trips could be shifted with modest prices. Although this argument is quite plausible, it is not clear whether peak-period nonwork trips are actually occurring on the most congested facilities likely to be subjected to congestion pricing. For example, Giuliano (Vol. 2) notes from her survey research on very congested facilities that most of the travelers are indeed work travelers. Nonetheless, even if only some of the peak-period travelers on congested facilities are making discretionary trips, pricing would most certainly alter their behavior.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion areas to spread. Suburbanization has caused a dispersal of automobile travel over a wider, less dense area in which the automobile is the only convenient alternative for work trips. A much faster rate of growth has occurred in suburb-to-suburb trips than in suburb-to-central-area trips. Substantial growth has occurred in nonwork trips during the peak period. MEASURING TRAFFIC CONGESTION Although any motorist is aware of traffic congestion when he or she experiences it, available data to provide a precise measure of congestion are another matter. Available data provide only indirect measures of congestion. The alternatives are to infer congestion from a) traffic volumes on facilities or (b) changes in trip durations over time. Facility-Based Measure The Highway Capacity Manual, the most widely used reference in highway planning and design, does not define congestion on facilities explicitly (TRB 1985). Rather, it lays out a definition of capacity, which must be understood relative to the desired or expected level of service (LOS). The Highway Capacity Manual defines the maximum capacity of a facility as “the maximum hourly rate at which persons or vehicles can reasonably be expected to traverse a point or uniform section of a lane or roadway during a given time period under prevailing roadway, traffic, and control conditions.” LOS is defined as “a qualitative measure describing operational conditions within a traffic stream, and their perception by motorists and/or passengers.” Six levels of service are described, from LOS A, which is free flow, to LOS F, which is characterized by stop-and-go traffic (see accompanying text box). The definition of congestion for any facility using the LOS approach depends on the quality of service “expected,” which may vary between designers and users and even among users. In measuring congestion with the LOS approach, the judgment of the designer or analyst plays a large part in what is defined as congestion. Hanks and Lomax (1990), researchers at TTI, developed a widely cited index estimating metropolitan area congestion that is based on such judgments. The index is one of the few measures that can be consistently applied across metropolitan areas (Hanks and Lomax 1990; Shrank et al.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion LEVELS OF SERVICE A: Free flow. Drivers unaffected by others in traffic stream, free to select speed and maneuver. B: Still free flow, but with more notice of other drivers. Slight loss in freedom to maneuver. C: Stable flow, but the beginning of the flow in which operations become significantly affected by the presence of other drivers. Declining comfort and convenience. D: High density but stable flow. Maneuvering restricted. E: At or near capacity. Speeds reduced to slow but uniform flow. Maneuvering extremely difficult. Flow unstable. F: Stop-and-go flow. Queues develop behind breakdowns in flow. 1993).2 This index estimates congestion on the freeways and principal arterials within an urbanized area by comparing daily traffic per lane-kilometer with a judgment regarding the traffic level at which congestion begins. The estimated congestion is weighted by total travel on each system such that a value of 1.0 indicates the beginning of congestion for an urban area. The TTI roadway congestion index (RCI) is (1 mi = 1.6 km) where VMT is vehicle miles traveled. Hanks and Lomax base their estimate of congestion on the proportion of traffic volumes on freeways and principal urban arterials that exceeds LOS C (Hanks and Lomax 1990, A.7-A.11). According to their analysis of 2   The congestion index is updated annually. Shrank et al. (1993) provide 1990 data for 50 urban areas.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion traffic volumes and delays on Houston-area roads over the 1970 to 1985 period, The maximum freeway service flow rate for level-of-service C is 1,550 passenger cars per lane hour (volume/capacity ratio equal to .77 for a 70 mph design speed facility) [(TRB 1985)]. Using average values for K factor (the percentage of daily traffic volume during the peak hour) and directional distribution, and including some adjustment for trucks, these values can be interpreted to indicate that 15,000 vehicles per lane day is an estimate of the beginning of level-of-service D operation. . . . The use of the boundary between level-of-service C and D as the beginning of congestion is consistent with reports by the Department of Transportation to Congress on the status of highways in the United States (congestion begins at a volume/capacity ratio of .8) and the AASHTO Policy on Geometric Design of Highways and Streets. (Hanks and Lomax 1990) Although their analysis leads Hanks and Lomax to define 15,000 vehicles per lane-day for freeways as a point at which congestion is becoming critical, in their index they use 13,000. They define 13,000 vehicles per lane-day as “a measure of approaching congestion,” which is consistent with a measure used in other FHWA and Texas Department of Transportation reports. The TTI index is based on counts of daily traffic volumes rather than measures of peak-period traffic volumes because the daily traffic count on most major routes is an available statistic, whereas peak-period traffic volume is not. The selection of daily traffic levels at which congestion occurs is based on estimates of the average percentage of travel occurring in peak periods. Traffic congestion, however, is specific to individual routes, corridors, or intersections. The use of average peaks gives, at best, an indirect gauge of the congestion that motorists encounter on their trips. Moreover, as noted above, the percentage of work and nonwork traffic occurring during peak periods appears to be changing over time, with peak-period traffic growing faster (Gordon et al. 1988). As a final point, the aggregation of the traffic data to a metropolitan-wide average, although useful as an indicator for the metropolitan area, may mask conditions that are much worse on some facilities and much better on others. Despite these weaknesses, the increases in the TTI measure of metropolitan congestion over time are plausible because the index is based on measures of traffic volumes compared with estimates of highway capacity, which is not increasing nearly as fast as traffic. As long as these conditions continue, congestion must increase as demand approaches capacity.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Trip-Based Measure Gordon and Richardson (Vol. 2) compared survey data on average work trips from different national surveys with the areas ranked most congested by facility-based measures. Their analysis shows no correlation between the average duration of the journey to work and congestion in the cities Hanks and Lomax identified as the most congested (see Gordon and Richardson, Vol. 2, Table 1). Los Angeles, for example, which is ranked first by the TTI index, is placed in the middle of metropolitan areas on the basis of average work trip times. Hartford, which ranks at the bottom of the TTI list, has almost the longest average work trip times of the areas studied. Moreover, despite the increased congestion levels over time indicated by the TTI index, the average work trip duration between 1983 and 1990 has hardly changed (Gordon and Richardson, Vol. 2, Table 4). Gordon et al. characterize the disparity between these findings this way: The commuting paradox reflects the apparent contradiction between perceptions of worsening traffic congestion and evidence of either declining or stable commuting times. However, not only is there no contradiction but the two phenomena are causally related. Rational commuters will, sooner or later, seek to escape congestion by changing the location of their homes and/or their jobs. This type of adjustment is easier to make in large, dispersed metropolitan areas with alternate employment subcenters and a wide variety of residential neighborhoods. The process is facilitated by the decentralizing location decisions of firms seeking to move closer to suburban labor pools. (Gordon et al. 1991) Although it is plausible that congestion brings about the suburbanization of jobs and residences, it may be premature to state that average commute trips are not increasing. The estimates of travel times used by Gordon and Richardson are based on surveys in which respondents are asked to estimate the duration of their work trips. As the authors acknowledge, one would expect that respondents would round off trip times in 5-min increments. Thus it would be difficult to discern small changes in the average over time. Moreover, respondents may have difficulty accounting for the increased variability of trips. In other words, respondents may be encountering congestion more frequently than in the past. Although the average commute on a “good” day may still be, say, 20 min, the frequency of “bad” days may be increasing. The subjective measure of average trip times would tend to make it difficult to account for such changes.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion An additional shortcoming of this measure is that in order to develop an average commute time for the entire metropolitan area, these estimated averages are then averaged again. In short, this is a highly aggregated measure of a phenomenon that is localized and specific. Many travelers within a metropolitan area may be experiencing sharply increased congestion, even though in aggregate these time losses are being offset by the growth in shorter durations in expanding parts of the same area. There is also a problem of discrepancies among individual surveys. The national census data relied on by Gordon and Richardson, for example, do not correspond with recently released local travel surveys of the Los Angeles and San Francisco metropolitan areas. Preliminary census data for Los Angeles County report an average commute time of about 27 min, but a recent survey by the Southern California Association of Governments (SCAG) reports an average for 1991 of 29.2 min (Pisarski 1992; SCAG 1993). For the San Francisco Bay Area, the discrepancies are much larger. In the Bay Area's 1981 survey, average commutes were 9 percent higher than those reported to the census, and in 1990 the average commutes were 18 percent higher (Purvis 1994). In contrast to the relative stability in average commute times reported in census data, local surveys for both Los Angeles and the San Francisco Bay Area over a 20- to 30-year period indicate that reported average commutes are increasing. Between 1967 and 1991, average commutes in Los Angeles County increased by 21 percent, and between 1960 and 1990, average commutes in the San Francisco Bay Area increased 17 percent (SCAG 1993; Purvis 1994). Despite these data discrepancies, the argument that Gordon and Richardson make—that individuals and businesses relocate to avoid congestion —has partial corroboration from other sources. Other data cited above indicate the substantial increase in the suburbanization of jobs during the 1980s and the continued faster growth in suburban and exurban population than in central-area population. Thus there is supporting evidence that changes in jobs and residences within metropolitan areas are altering the commuting behavior of many workers. Whether these changes are caused by or are large enough to offset growing congestion, however, is not certain. SUMMARY Two different indirect measures of congestion appear to tell different stories. One indicates that areawide congestion is worsening; the other

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion indicates that behavioral changes keep congestion from worsening because people make rational choices to avoid delay. It can be argued that both stories are true. Consider this example. Drivers become frustrated with the length of time spent on a congested route to work; over time they change their residence to shorten the commute time. [Some may choose to move closer to work in order to reduce travel times; some may choose to move both job and residence to outlying areas to escape congestion (Wachs 1993).] Despite the change to a different route for most of the trip, the congestion on that facility might increase because of increased trip making by others. The time savings may offset the growing time losses of others using the increasingly congested facility. Thus it is possible to have both increased congestion on facilities and a stability in average commute times. On the other hand, both of the indirect measures of congestion reviewed here have weaknesses. Both are highly aggregated when congestion is local and specific. Moreover, the facility-based measure assumes average peaking characteristics that, in practice, would tend to vary widely across routes and that may be changing over time. The trip-based measure could have biases because of its subjective nature, and there is some unexplained discrepancy with other data sources. In the context of the issues being discussed in this report, whether congestion on facilities is increasing or whether congestion is being held in check by locational changes is almost a moot point. Traffic congestion is an important social problem because it wastes time and consumes resources that could be invested more productively. The theoretical underpinnings of pricing to reduce congestion are outlined in Appendix B. REFERENCES ABBREVIATIONS DOT U.S. Department of Transportation SCAG Southern California Association of Governments Cervero, R. 1986. Suburban Gridlock. Center for Urban Policy Research. Rutgers University, Brunswick, N.J. DOT. 1993. Journey-to-Work Trends in the United States and Its Major Metropolitan Areas: 1960 to 1990. Office of Highway Information Management, Federal Highway Administration . Downs, A. 1992. Stuck in Traffic: Coping with Peak-Hour Traffic Congestion.The Brookings Institution, Washington, D.C. Garreau, J. 1991. Edge City: Life on the New Frontier.Doubleday, New York. 546 pp.

OCR for page 104
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Gordon, P., et al. 1988. Beyond the Journey to Work. Transportation Research, Vol.22A, No. 6, pp. 419–426. Gordon, P.,et al. 1991. The Commuting Paradox. Journal of the American Planning Association, Vol.57, No. 4, Autumn, pp. 416–420. Hanks, J.,andT. Lomax. 1990. Roadway Congestion in Major Urban Areas, 1982 to 1988. Report 1131-3. Texas Transportation Institute. Hanson, M. 1992. Automobile Subsidies and Land Use: Estimates and Policy Responses . Journal of the American Planning Association, Vol.58, No. 1, Winter, pp. 60–71. Pisarski, A. 1987. Commuting in America: A National Report on Commuting Patterns and Trends.Eno Foundation for Transportation, Inc., Westport, Conn. Pisarski, A. 1992. New Perspectives in Commuting Based on Early Data from the 1990 Decennial Census and the 1990 Nationwide Personal Transportation Study (NPTS) .U.S. Department of Transportation, July. Purvis, C. 1994. Changes in Regional Travel Characteristics and Travel Time Budgets in the San Francisco Bay Area: 1960–1990. To be published in Transportation Research Record, TRB, National Research Council, Washington, D.C. SCAG. 1993. Summary Findings: 1991 Southern California Origin-Destination Survey February. Shrank, D., et al. 1993. Estimates of Urban Roadway Congestion, 1990. Research Report 1131-5. Texas Transportation Institute. Transportation Research Board. 1985. Special Report 209: Highway Capacity Manual.National Research Council, Washington, D.C. Wachs, M. 1993. Learning from Los Angeles: Transport, Urban Form, and Air Quality . Transportation, Vol. 20, No. 4, pp. 329–354.