ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
lost in commuting, and they search for residences closer to remains significant in explaining the variation of commute 
jobs. range. See the following regression result: 
(2) Poor Job Accessibility for Some Inner-City Low-wage 
Workers 
The job accessibility values in Table 2 confirm similar trends 
between job access and workers’ mean wage rates. For job 
accessibility, the mean wage of $25,000 is the turning point 
(note that the difference between the category of $22,500- 
25,000 and that of $25,000-30,000 is only minimal). In 
general, among TAZs with mean wage rates less than 
$25,000, job accessibility declines with higher mean wage 
rates; among TAZs with mean wage rates above $25,000, 
job accessibility increases with higher mean wage rates. The 
only divergence from this trend is the group of TAZs with the 
lowest mean wage rate (below $15,000). This group has 
poorer job accessibility than the group of TAZs with mean 
wage rates between $15,000-17,500. 
Examining the spatial variation of job accessibility in Figure 5 
reveals an astonishing observation. The general pattern is 
that TAZs closer to the CBD enjoy better job accessibility. 
However, some of the inner-city TAZs suffer the worst job 
accessibility. Shen (1998, 1999) finds a similar pattern in 
Boston, and attributes the problem to poor vehicle availability 
and stiff competition for suitable jobs among low-wage 
workers. The critical role of automobile ownership is also 
recognized by Taylor and Ong (1995). This research 
suggests an additional factor that makes the problem worse: 
congestion caused by high residential and employment 
densities. 
(3) Workers’ Mobility Handicapped by Lower Wages 
Both job proximity and accessibility indexes measure 
potential, not actual, mobility. An assessment of actual 
mobility calls for analyzing the actual commuting pattern. 
One way to gauge the mobility of resident workers is to see 
how far they travel to their workplaces. Based on the 
journey-to-work data, the index of average commute range 
(R) is constructed to measure the actual mobility of resident 
workers: 
where T^ is the actual number of commuters from TAZ i to j, 
dij is the aerial distance between them, and Wi is the number 
of resident workers at i. In other words, resident workers at i 
commute a radius of R, on average. From Table 2, average 
commute ranges increase steadily from 7.51 km among 
TAZs with a mean wage rate below $15,000 to 12.83 km 
among TAZs with a mean wage rate $25,000-30,000. That is 
an increase of 71%. Although the commute range declines 
slightly for TAZs with mean wage rates higher than $30,000 
(consistent with the trend of job proximity values), it remains 
around 12 km. 
From a planning perspective, mobility may be viewed 
positively or negatively desirable (Salomon and Mokhtarian, 
1998, p.131). Urban land use planning policies (e.g., mixed- 
use developments and job-housing balance) attempt to 
reduce the social inefficiency of travel or mobility. On the 
other side, mobility is the freedom of an individual to move 
and, in our case, the ability to obtain job opportunities. Less 
mobility of lower-wage workers limits their search ranges for 
jobs as well as for housing. For workers in the lowest wage 
group, their mobility is mainly handicapped by their financial 
resource and thus the dependence on slower public transits 
(or bicycles or walking). 
The shorter commute range (R) by lower-wage workers 
cannot be explained by their proximity to jobs (D) alone. 
Even when the job proximity is controlled for, the wage rate 
R, = 5.0247+0.2088Di+0.0723MeanWage, 
(16.60) (18.52) (7.86) 
with R 2 =0.355 (n=808). The t-values for corresponding 
variables are presented in the parentheses. In other words, 
given the same job proximity, lower-wage workers tend to 
take jobs closer to their homes. This may be a result of 
narrower job search radii or concerns of financial burden of 
excess commuting. 
Surprisingly, the mean commute time is relatively stable 
among TAZs of various mean wage rates ranging from 20.82 
to 22.19 minutes. The monetary or psychic cost or the 
opportunity cost of time spent commuting is valued probably 
as important by low-wage workers as by high-wage workers. 
The hourly wage rate varies to a great deal, but the portion 
of commuting cost out of total wage is probably similar 
among workers of different wages. Workers with better 
financial resources may outbid others for residential 
locations with easy access to highways and be able to 
maintain similar commute times while traveling farther. The 
stability of commute time is also evident across different time 
periods (Gordon et al., 1991; Cervero and Wu, 1998). At the 
personal level, commute time may be viewed as a negativity, 
but commute range as a plus. 
(4) The Classic Monocentric Model Revisited 
By a visual examination of Figure 3, jobs are scattered 
regionwide in Cleveland, and the distribution pattern can be 
hardly characterized as monocentric. However, downtown 
Cleveland is by far the largest job center and exerts 
dominant effects on forming the patterns of job proximity and 
accessibility. Figure 4 shows that, with only a few 
exceptions, the job proximity improves towards the CBD. 
Areas around the CBD enjoy proximity of less than 15 km to 
jobs, whereas areas on the urbanized periphery are more 
than 23 km from jobs. A simple regression indicates that a 
one-kilometer increase in distance from the CBD leads to an 
average loss of 0.665 km in job proximity (R 2 as high as 
0.846 for n=808). Figure 5 shows that the job accessibility 
pattern also conforms to the monocentric model, but to a 
less degree. Divergence from the monocentricity occurs in 
(1) some inner-city TAZs with the poorest accessibility, and 
(2) some TAZs near several suburban job centers with the 
best accessibility. A regression indicates that a one- 
kilometer farther from the CBD is associated with an average 
drop of 0.0096 in job accessibility (R 2 =0.394 for n=808). 
However, as illustrated previously, workers with various 
wages respond to job proximity and accessibility differently. 
The residential mean wage rate only increases marginally 
with distance from the CBD (a regression has a weak 
R 2 =0.023 for n=808). As distance increases from the CBD, 
workers travel farther (a regression between commute 
ranges and distances from the CBD yields R 2 =0.29 for 
n=808), but spend little more times (the regression between 
commute times and distances from the CBD has R 2 =0.02 for 
n=808). Again, faster travel speeds in areas farther away 
from the CBD explain the difference. 
SUMMARY AND DISCUSSION 
This research uses two indexes, job proximity and 
accessibility, to measure resident workers’ locational 
advantage with respect to their suitable job markets. Job 
proximity is designed to capture the spatial separation 
between residents and jobs. Job accessibility measures 
one’s ability to overcome such separation that may be 
affected by transportation means, road network, congestion 
and intensity of competition for jobs among workers. Several 
findings are noticeable and have important implications for 
public policies. 
294