Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration
Office of Planning, Environment, & Realty (HEP)

Toolbox for Regional Policy Analysis Report (2000)

Case Study: San Francisco Bay Area, California


The most recent Regional Transportation Plan (RTP) for the San Francisco Bay Area in California includes $88 billion in expenditures to operate, maintain, and expand local transit, highway, and other transportation systems. In developing the 1998 RTP, the Metropolitan Transportation Commission (MTC) sought to analyze the fairness or equity of the plan's impacts on different population groups. To measure equity, MTC developed transit and automobile accessibility measures that were compared with and without the plan. The plan's accessibility impacts were compared between areas of "disadvantaged" population and "not disadvantaged" areas.

Further, our analysis of 38 "disadvantaged" (as defined by the Northern California Council for the Community) neighborhoods in the Bay Area found that most are located in areas where public transit service is most concentrated - lending a significant measure of equity to the overall RTP investment program.

- Metropolitan Transportation Commission Draft RTP, August 1998

The MTC's analysis found that:

MTC's analysis was useful for assessing the overall impacts of the transportation plan on regional equity. MTC's methodology could also be applied for other purposes, for example, to compare alternative sets of projects or to evaluate major project alternatives. This case study illustrates some the of strengths, limitations, and further developments that are needed in using accessibility to measure the equity of transportation investments.

Regional Setting

The San Francisco-Oakland-San Jose Metropolitan Area (the "Bay Area") is a nine-county area covering 7,179 square miles, with an estimated 1998 population of over six million. The Metropolitan Transportation Commission (MTC) is the state-designated regional transportation planning agency as well as the federally designated metropolitan planning organization (MPO) for the region.

An understanding of the Bay Area's physical environment provides a background for understanding differences in accessibility by geographic area and mode (Figure 1). The urbanized area surrounding San Francisco Bay is defined by topography. Its bodies of water and ranges of hills limit development to particular corridors between these constraints. Correspondingly, transportation networks are "linear" along these developed corridors, and bridges and tunnels form key transportation links at water and hill crossings.

The city of San Francisco, on the tip of the San Francisco Peninsula, contains the greatest regional concentration of employment in its Central Business District (CBD). The city is densely populated, supporting comprehensive transit service, and has a high number of middle- and upper-income zero-car households. The older cities on the east side of San Francisco Bay, the largest being Oakland, as well as core neighborhoods of San Jose also contain a moderately dense population and relatively high levels of transit service. Development in most the South Bay (San Jose and environs) and in the eastern and northern parts of the metropolitan area is more heavily automobile-oriented.

Transit service consists of three primary structures: local networks serving the more densely populated areas; long-distance commuter routes, including heavy rail, commuter rail, ferries, and express buses, to the San Francisco, Oakland, and San Jose CBDs; and suburban feeder buses to the commuter systems. The Bay Area Rapid Transit (BART) is a 95-mile heavy-rail system that serves the San Francisco and Oakland CBDs. Caltrain provides commuter rail service from south of San Jose to San Francisco along the peninsula.

Figure 1. San Francisco Bay Area

Fig. 1 San Francisco Bay Area

Source: U.S. Bureau of the Census.

Regional Transportation Plan

The San Francisco-Oakland-San Jose Metropolitan Area is facing an anticipated 29 percent increase in population and a 42 percent increase in jobs between 1998 and 2020. Rather than responding to growth with major new highway projects, the 1998 Regional Transportation Plan focuses on maintaining and improving existing highway and transit systems. Fifty-three percent of funding in the $88 billion plan is directed toward maintaining and operating the area's existing public transportation network, including buses and rail transit. Most of the remaining funds are directed toward roadway maintenance, expansion of the HOV-lane system, freeway and arterial management systems, and localized interchange and capacity improvements.

The complete set of projects included in the RTP is identified on a corridor-by-corridor basis in the full RTP document.


Overall Approach

The purpose of the MTC's accessibility and equity analysis was to compare the impacts, by population group, of the 1998 Regional Transportation Plan (RTP). Impacts were compared for the "RTP project" and "no project" alternatives. The project alternative included all committed plus "Track 1" projects identified in the RTP, while the "no project" alternative included only currently committed projects.

To evaluate the equity of the plan's impacts, MTC identified traffic analysis zones (TAZ) that could be characterized as having "disadvantaged" populations. These TAZs corresponded to disadvantaged neighborhoods as identified by a local group, the San Francisco-based Northern California Council for the Community (NCCC). Baseline accessibility levels were compared for disadvantaged TAZs versus other TAZs. Percent changes in accessibility with and without the RTP were also compared between the two population groups. Comparisons were made for three modes: transit, drive-alone automobile, and high-occupancy vehicle (HOV).

Definition of Disadvantaged Groups

The MTC measured equity in terms of accessibility changes to "disadvantaged" relative to "not disadvantaged" neighborhoods. Definitions of disadvantaged neighborhoods are from a 1997 report by the San Francisco-based NCCC. Using 1990 Census data, the NCCC identified 38 "disadvantaged" neighborhoods, comprised of 142 census tracts, based on median household income, public assistance income, and median gross rent as a percentage of household income. The NCCC methodology flags local neighborhoods that are 80 percent or less of each county's median household income. MTC then identified the 133 regional travel analysis zones (out of 1,099 total zones) that correspond to these 142 census tracts.

Population forecasts for 2020 prepared by the Association of Bay Area Governments (1998) indicate that 959,000 persons will reside in the 133 neighborhoods identified as disadvantaged. This compares to the 6.8 million persons expected to reside in the rest of the Bay Area in 2020 (the total population of the Bay Area is projected to be nearly 7.8 million persons.) The locations of the disadvantaged TAZs are shown in Figure 2.

Accessibility could also be compared based on other definitions of socioeconomic groups at the census tract level. Tracts containing relatively high levels of low-income or minority populations could be identified, for example, based on some locally agreed-upon threshold. Socioeconomic characteristics and mobility limitations could also be used in combination to develop a locally preferred measure of "disadvantaged" populations. Regardless of the methodology used, dividing tracts and corresponding TAZs into a "disadvantaged" group and a "not disadvantaged" group allows the difference in accessibility between these two groups to be compared. Differences among multiple groups (e.g., three or more average income categories) could also be defined and measured with appropriate statistical tests.

Figure 2. TAZs Corresponding to "Disadvantaged" Neighborhoods

Fig. 2 TAZs Corresponding to "Disadvantaged" Neighborhoods

Source: Adapted from Metropolitan Transportation Commission data.

Accessibility Measures

In its accessibility analysis, MTC computed two types of measures, each describing the accessibility of households to employment:

  1. A "threshold-based" measure, which is the number of jobs reachable within an X-minute travel time of a given zone. MTC selected a variety of thresholds, including 30, 45, 60, and 75 minutes, for comparison.
  2. A "gravity-based" measure, in which the number of jobs in each zone is weighted in inverse proportion to travel time from the zone of residence (e.g., importance diminishes as distance increases).

To compute the overall regional accessibility for a given mode, the total number of jobs reachable from each zone is weighted by the population of that zone. The accessibility measures were computed on the basis of 2020 travel times forecast by the MTC's regional travel demand model and 2020 population and employment forecasts by the Association of Bay Area Governments (ABAG).

The Montgomery County case study provides another example of the use of threshold-based measures, while the Tren Urbano case study provides another example of the use of gravity-based measures.

Calculation of Threshold-Based Accessibility

The data requirements for this computation include:

MTC uses MinUTP software as the basis for its regional travel demand model. MTC staff have developed a inNUTP-compatible Fortran program known as "ACCTMTX" that tallies the zones within a specified travel time "contour" and then tallies the employment within these zones. The output of this program is then analyzed using standard spreadsheet software (Excel); statistical software (SAS); and mapping software (MapInfo). The MTC's accessibility analysis was based on a.m. peak-period travel times computed separately for the highway, transit, and HOV 3+ networks. More information about the MTC's travel model is available through the MTC's DataMart.

The resulting measure, Aik, is the sum of all jobs accessible within an X-minute travel time of zone i by mode k. This is computed for each zone. To compute overall regional accessibility, Ak, for mode k, the employment accessible from each zone is weighted by the population of the zone, as follows:


As an example, the average Bay Area resident can reach 71,722 jobs within 30 minutes travel time by taking transit during the a.m. peak period.

Calculation of Gravity-Based Accessibility

The gravity-based accessibility measure is computed as follows:


Where Aik represents the employment accessibility for zone of residence i and mode k. Total regional accessibility by mode k is then computed as for the threshold-based measure.

In addition to the data required for the threshold measure, a "distance decay" parameter or exponent for the travel time function (b) is required. MTC selected a value for b of 2.0 as typical for home-based work travel (the methodology has since been updated using the gravity model coefficient from MTC's trip distribution model). A different value could also be used that is consistent with an area's local travel demand model.

A technical complication in computing the gravity-based measure is that some zones will not have transit service, and thus travel time to these zones is not defined. This problem can be solved in two ways:

  1. Select an arbitrarily large number for travel time that will produce a very small (but non-zero) number when raised to the given power; or
  2. Remove these zone pairs from the computation of Aik for each zone.

Comparison Between Groups

In addition to simply looking at differences in accessibility, MTC conducted statistical tests to determine whether these differences were significant. The tests used in this analysis include paired t-tests and the standard error of difference between means.

The paired t-test is used to compare the project with the no-project alternative. This test examines, on a zone pair basis, whether the difference between the project and no-project accessibility measure is significantly greater than zero. If the t-statistic on a paired t-test is greater than 1.96, then there is at least a 95 percent probability that there is a difference between the two alternatives.

The standard error of difference between means is used to compare accessibility between "disadvantaged" and "non-disadvantaged" sets of TAZs. The difference between means is divided by this standard error of difference between means to obtain a t-statistic. If the t-statistic is greater than 1.96, then it is concluded that there is a significant difference (or, there is less than a one-in-20 probability that there is no difference between disadvantaged and advantaged neighborhoods).

Percent differences are also reported in the analysis. These are fairly intuitive measures to describe the magnitude of differences between alternatives and between subsets of neighborhoods.

Level of Effort

MTC staff estimate that perhaps one to two person-weeks were required to conduct the accessibility analysis, including writing the ACCMTX program, computing and processing the accessibility measures, and documenting results. Updating the analysis would take about two to three person-days, including analysis and documentation time.

As noted, the analysis utilizes travel demand model data as well as programming, statistical, spreadsheet, and GIS software that most MPOs would have in-house. Some programming skills are required to analyze the travel demand model output. In addition, the comparison of means for statistical significance requires a basic understanding of how to apply various statistical tests.


Baseline Accessibility

Table 1. compares the number of total jobs, on a per capita basis, accessible within 30, 45, 60 and 75 minutes travel time by transit, driving alone and by carpool, during the a.m. peak period. The first thing to note is the much higher number of jobs accessible to drive-alone users compared to transit users. This makes sense given the faster door-to-door speeds of driving alone relative to taking transit. Carpool accessibility is even greater than drive alone, given the presence of high-occupancy vehicle (HOV) lanes. Also, as would be expected, the number of opportunities that can be reached increases significantly the farther a person is willing to travel.

Table 1. Total Jobs, per Capita, Within Travel Time by Means of Transportation

RTP Project Alternative

Travel Time Transit Drive Alone Carpool
30 Minutes 71,722 702,394 795,502
45 Minutes 256,705 1,377,907 1,689,108
60 Minutes 535,217 2,240,825 2,657,197
75 Minutes 884,520 3,041,925 3,449,408

Source: MTC, 1998.

Figure 3 illustrates the baseline level of transit accessibility on a regionwide basis. As would be expected, accessibility is highest in the core urban areas of San Francisco and Oakland, where there is extensive bus and rail service. Accessibility is also relatively high in the inner areas of San Jose, as well as in suburban zones served by BART and CalTrain.

Figure 3. Baseline Transit Accessibility, 2020

Fig. 3 Baseline Transit Accessibility, 2020

Source: Adapted from Metropolitan Transportation Commission data.

Beyond these baseline comparisons, the results of the accessibility analysis are designed to answer two questions:

  1. First, is accessibility (by mode) significantly increased in the project alternative compared to the no-project alternative?
  2. Second, is accessibility better or worse for disadvantaged neighborhoods, as compared to not disadvantaged neighborhoods?

Project Versus No-Project Alternative

Table 2 compares transit accessibility under the project versus no-project alternatives. Table 2 shows the total change and percent change in accessibility, as well as a t-statistic indicating whether this difference is significant. (A t-statistic greater than 1.96 indicates that there is at least a 95 percent probability that the change is significant.) Results are presented for the 30, 45, 60, and 75-minute travel time contours as well as for the weighted (gravity-type) index. Table 3 repeats these results for auto (drive-alone) accessibility.

Table 2.
Transit Accessibility, RTP Project versus No-Project Alternative

Average per Capita Total Employment Within X Minutes Peak Transit Travel Time

Neighorhood No-Project Project Difference Pct. Diff. t-statistic
30 Minutes
Not Disadvantaged 58,900 61,689 2,789 4.7% 9.61
Disadvantaged 141,729 144,595 2,866 2.0% 4.64
TOTAL 68,924 71,722 2,798 4.1% 10.53
45 Minutes
Not Disadvantaged 229,256 243,264 14,008 6.1% 13.58
Disadvantaged 345,766 354,330 8,564 2.5% 5.79
TOTAL 243,356 256,705 13,349 5.5% 14.43
60 Minutes
Not Disadvantaged 502,733 529,266 26,533 5.3% 15.41
Disadvantaged 566,936 578,432 11,496 2.0% 5.56
TOTAL 510,503 535,217 24,714 4.8% 16.04
75 Minutes
Not Disadvantaged 855,326 886,607 31,281 3.7% 15.34
Disadvantaged 850,614 869,357 18,743 2.2% 7.61
TOTAL 854,756 884,520 29,764 3.5% 16.34

Weighted Transit Accessibility
(Index = 100.0, No-Project Index for All Neighborhoods)

Neighorhood No-Project Project Difference Pct. Diff. t-statistic
Not Disadvantaged 96.9 101.3   4.5% Significant
Disadvantaged 138.1 142.1   2.9% Significant
Total 100.0 104.2   4.2% Significant

Source: Metropolitan Transportation Commission, 1998.

Table 3.
Highway Accessibility, RTP Project versus No-Project Alternative

Average per Capita Total Employment
Within X Minutes Peak Highway Travel Time

Neighorhood No-Project Project Difference Pct. Diff. t-statistic
30 Minutes
Not Disadvantaged 672,087 712,686 40,599 6.0% 26.92
Disadvantaged 592,799 627,646 34,847 5.9% 14.30
TOTAL 666,492 702,394 35,902 5.4% 29.36
45 Minutes
Not Disadvantaged 1,309,940 1,400,037 90,097 6.9% 35.12
Disadvantaged 1,131,717 1,217,172 85,455 7.6% 14.04
TOTAL 1,288,371 1,377,907 89,536 6.9% 37.75
60 Minutes
Not Disadvantaged 2,070,462 2,273,468 203,006 9.8% 32.24
Disadvantaged 1,797,151 2,003,736 206,585 11.5% 14.28
TOTAL 2,037,386 2,240,825 203,439 10.0% 35.06
75 Minutes
Not Disadvantaged 2,872,647 3,079,769 207,122 7.2% 31.61
Disadvantaged 2,526,254 2,767,058 240,804 9.5% 16.30
TOTAL 2,830,727 3,041,925 211,198 7.5% 34.98

Weighted Highway Accessibility
(Index = 100.0, No-Project Index for All Neighborhoods)

Neighorhood No-Project Project Difference Pct. Diff. t-statistic
Not Disadvantaged 100.8 105.5   4.7% Yes
Disadvantaged 94.2 98.6   4.7% Yes
Total 100.0 104.7   4.7% Yes

Source: Metropolitan Transportation Commission, 1998.

The analysis shows that the project alternative provides significantly greater accessibility than the no-project alternative. This holds true for all modes and all measures, and for both disadvantaged and not disadvantaged neighborhoods. Overall transit accessibility is increased by 3.5 to 5.5 percent, depending upon the measure. Total drive-alone auto accessibility shows a larger increase of 4.7 to 10.0 percent.

Disadvantaged Versus Not Disadvantaged Neighborhoods

Table 4 shows how accessibility differs under the RTP project alternative for disadvantaged versus not disadvantaged neighborhoods. Table 4 also indicates whether or not this difference is statistically significant.

Table 4.
Employment Accessibility for Disadvantaged versus Not Disadvantaged Neighborhoods, RTP Project Alternative

Average per Capita Total Employment Within X Minutes Peak Travel Time

Travel Time Contour Difference Between Means* Std. Error, Diff. Between Means t-stat. Is Difference Significant?
30 Minutes Transit 82,906 16,875 4.91 Yes, disadvantaged more accessible
45 Minutes Transit 111,066 30,021 3.70 Yes, disadvantaged more accessible
60 Minutes Transit 49,166 40,015 1.23 No, not significantly different
75 Minutes Transit -17,250 49,346 -0.35 No, not significantly different
Highway (Drive-Alone)
30 Minutes Drive -85,040 -36,667 -2.32 Yes, not disadvantaged more accessible
45 Minutes Drive -182,865 -66,293 -2.76 Yes, not disadvantaged more accessible
60 Minutes Drive -269,732 -95,822 -2.81 Yes, not disadvantaged more accessible
75 Minutes Drive -312,711 -120,752 -2.59 Yes, not disadvantaged more accessible

*Disadvantaged - Not Disadvantaged.

Source: Metropolitan Transportation Commission, 1998.

The analysis shows that disadvantaged neighborhoods have significantly higher transit accessibility, at least for jobs within a 45-minute travel time. There are over 100,000 more jobs reachable within 45 minutes from the average disadvantaged neighborhood than from the average non-disadvantaged neighborhood. (Differences for 60- and 75-minute transit travel times, on the other hand, are not statistically significant.) The weighted accessibility index is about 40 percent higher for disadvantaged than for non-disadvantaged neighborhoods.

Automobile accessibility, in contrast, appears somewhat greater for not disadvantaged neighborhoods than for disadvantaged neighborhoods. Over 180,000 more jobs (13 percent more) are accessible to not disadvantaged neighborhoods than to disadvantaged neighborhoods within a 45-minute travel time. The differences for all travel time contours are statistically significant. The difference in the weighted index, however, is relatively small - only six to seven percent lower for disadvantaged neighborhoods than for not disadvantaged neighborhoods. Furthermore, this difference is not statistically significant.

Figure 4 and Figure 5 illustrate changes in accessibility for disadvantaged and not disadvantaged neighborhoods, as a result of the plan. (These differences are also shown in Tables 2 and 3.) Not disadvantaged neighborhoods benefit from greater improvements in transit accessibility although they are starting from a lower baseline than disadvantaged neighborhoods. In contrast, disadvantaged neighborhoods appear to benefit slightly more in terms of automobile accessibility (but again, are starting from a lower baseline). One interpretation of these findings is that the RTP decreases the relative differences in accessibility between disadvantaged and not disadvantaged neighborhoods, for both highway and transit modes.

Figure 4. Change in Transit Accessibility, RTP Project Versus No-Project (2020)

Fig. 4 Change in Transit Accessibility, RTP Project Versus No-Project (2020)

Figure 5.
Change in Highway Accessibility, RTP Project Versus No-Project (2020)

Fig. 5 Change in Highway Accessibility, RTP Project Versus No-Project (2020)

Spatial Distribution of Impacts

Figure 6 illustrates the spatial distribution of accessibility changes between the project and no-project alternatives for the transit mode. Figure 7 illustrates the same changes for the auto (drive-alone) mode. The greatest percentage benefits for transit occur in the suburbs of the South Bay that are served by new light rail extensions and express bus/HOV lanes, and to the more distant areas to the north and east served by U.S. 101 north, I-80, and S.R. 4 that also benefit from HOV lanes and longer-distance bus service. These corridors to the north and the east also appear to benefit most in terms of relative automobile accessibility.

While percentage changes are greatest in outlying areas, it should be noted that these areas are starting from a much lower baseline level than the existing core urbanized areas in the region. Absolute changes in accessibility (not shown) are more evenly distributed throughout the region. The primary exception is the eastern part of Santa Clara County that shows much higher transit gains as a result of the new rail and bus transit projects in the area.

Figure 6.
Change in Transit Accessibility, RTP Project versus No-Project Alternatives

Fig. 6 Change in Transit Accessibility, RTP Project versus No-Project Alternatives

Source: Adapted from Metropolitan Transportation Commission data.

Figure 7.
Change in Auto Accessibility, RTP Project versus No-Project Alternatives

Fig. 7 Change in Auto Accessibility, RTP Project versus No-Project Alternatives

Source: Adapted from Metropolitan Transportation Commission data.

Overall Findings

MTC draws the following conclusions from their analysis of the 1998 RTP:



Some noteworthy strengths of the MTC's application of accessibility measures include:

General Limitations

The MTC analysis represents a state-of-the-practice comparison of the distributional implications of alternative regional transportation plans. While the results provide valuable insights, there are nonetheless certain limitations that need to be recognized. These limitations are common to other accessibility analyses, such as those documented in the Montgomery County and Tren Urbano case studies. Specifically:

Lessons Learned

The MTC suggested a number of lessons drawn from their experience that might be useful for other areas conducting a similar equity analysis:

Further Development

The MTC analysis illustrates an important improvement in analysis capabilities. Opportunities for further development in the application of accessibility measures are possible. For example:


Published Documents

Association of Bay Area Governments (ABAG's Projections ‘98), Oakland, CA. Internet:

Northern California Council for the Community. A Guide to the Bay Area's Most Impoverished Neighborhoods - By County: Bay Area Partnership for Building Healthy and Self-Sufficient Communities for Economic Prosperity. San Francisco, CA (1997).

Purvis, Chuck. "1998 Regional Transportation Plan: Equity & Accessibility Analysis." Internal memorandum, Metropolitan Transportation Commission (June 24, 1998).


Organization Person Phone
Metropolitan Transportation Commission Chuck Purvis 510-464-7731

Appendix: Occupational Matching


"We conclude that the very purpose of tracking changes in accessibility is to provide feedback on the degree to which resource allocation decisions in the urban transportation field are helping to redress serious inequities in accessibility to jobs, medical facilities, and other important destinations." - Cervero, Rood, and Appleyard (1995)

Cervero, Rood, and Appleyard (1995) used census transportation planning data to study spatial and temporal trends in job accessibility, with the San Francisco Bay Area serving as a case context. The authors refined a standard measurement of accessibility to include "occupational matching." This approach accounts for the consistency between employed residents' skills and employment roles within specific neighborhoods and labor force occupational characteristics in employment zones.

The authors' use of occupational matching led to significant findings beyond those of an analysis without occupational matching. For example, inner-city neighborhoods tended to score well on job accessibility indices without occupational matching. In contrast, the greatest job opportunity mismatches tended to be found in some of the region's poorest neighborhoods, including a number of inner-city neighborhoods. The highest match effect tended to be in well-to-do residential neighborhoods where median household incomes are well above the regional average.


Accessibility Measure

The authors use a gravity-type accessibility measure that describes the accessibility of residents to jobs. The basic formula used is:

Equation 1  (Equation 1)


AIi  = Accessibility Index for residential zone i.
Ej  = Employment - number of workers in zone j.
dij  = Distance (in miles) - highway network distances between zonal centroids, for all i-j interzonal pairs less than 45 miles.
g = Empirically derived impedance coefficient, set at -0.35 for commute trips in the San Francisco Bay Area.

While network highway distances were used for spatial impedance in this study, network travel times could easily be substituted.

Occupational Matching

To incorporate occupational matching, the accessibility formula was refined as follows:


(Equation 2)

Where the additional variables are:

pik  = Proportion of employed residents in zone i working in occupational class k (see below).
Ejk  = Number of workers in zone j working in occupational class k.

The five occupational classes used in this index (values of k) are:

  1. Executive, professional, managerial;
  2. Sales, administration, clerical;
  3. Services;
  4. Technical; and
  5. All others, excluding all non-civilian positions.

The refined accessibility measure is known as an occupational match accessibility index. For any residential zone i, proximity to jobs in zone j will contribute positively to the accessibility index based on the degree to which occupational roles of employed residents in zone i match the occupational opportunities in zone j.

Subtracting the standardized base accessibility index (equation 1) from the standardized occupational match index (equation 2) provides a "match effect" - an indication of the relative importance of occupational matching as an input into the calculation of job accessibility. Large positive values indicate that residents are generally well qualified for nearby jobs, while a large negative value indicates the existence of a serious job mismatch.

Calculation for Employment Centers

The authors conducted a similar set of analyses with reference to the region's 22 largest employment centers. The base and occupational match equations take a similar form as for residential accessibility, except that:

  1. The AI is calculated for the employment center j (combinations of contiguous census tracts); and
  2. pjk is the proportion of workers in the employment center by occupational class;

Ejk is replaced by Rik, the number of employed residents in each residential zone i working in occupational class k.

Data Requirements

All the necessary data inputs for this analysis can be obtained from the Census Transportation Planning Package (CTPP) and from regional travel model network data. The required data include:

If traffic analysis zones (TAZ) are not equivalent to census tracts, network data may only be available for TAZ pairs. In this case, two options are available:

  1. Tract-level employment data can be converted to TAZ-level employment data using tract-TAZ equivalency files. This option was used in the Bay Area analysis.
  2. A census tract travel time matrix can be created by relating each tract to the nearest TAZ and looking up values from the TAZ travel time matrix.


Residents to Jobs

The accessibility of residents to jobs was compared with and without occupational matching. Findings include:

The authors conclude that these match effect findings likely reflect several dynamics. First, there was probably more fluidity in housing markets and greater residential mobility among well-educated, higher-salary workers over the 1980s, enabling them to more easily sort themselves into locations reasonably close to job opportunities. Additionally, leading Bay Area firms also tended to locate with reference to potential pools of professional and executive employees during the 1980s. At the other end of the income spectrum, however, poor households anchored in often declining inner-city neighborhoods appear to have found themselves less and less accessible to jobs which they qualify for over the course of the 1980s.

Employment Centers to Workers

The findings using this measure were similar to the findings based on accessibility of residents to jobs. In particular:


The addition of occupational matching to regional accessibility analysis can add considerable richness to the findings. In the current case, it helped to demonstrate differences in accessibility that were not apparent from a generalized accessibility analysis.

The technique does suffer from some limitations. For example:

A future research topic is to develop alternatives to standard gravity-based measures of accessibility. Particularly important is to stratify accessibility indicators along socioeconomic and other qualitative dimensions. While residents of a neighborhood might be close to a lot of job opportunities, this accessibility is meaningful only if they have the skills or education to qualify for those jobs.


Cervero, Robert; Timothy Rood and Bruce Appleyard. Job Accessibility as a Performance Indicator: An Analysis of Trends and Their Social Policy Implications in the San Francisco Bay Area. Institute of Urban and Regional Development, University of California, Berkeley, CA UCTC Working Paper No. 366 (1995). Internet:

Updated: 10/20/2015
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000