In constructing a sample of pairs for our analysis, we apply two additional critieria,
selecting all pairs that (i) reside in the same block group within the Boston metropolitan area and
(ii) do not belong to the same household. Overall, the sample contains 2,037,600 pairs that meet
all of the above criteria. The first column of Table 1 characterizes this sample of matched pairs,
reporting the percentage of pairs that fit the description in the row heading: at least one member
of roughly 72 percent of the pairs has children; about 15 percent of pairs match two single
individuals.15
Examining the characteristics of the sample of pairs shown in Table 1 highlights three
key dimensions of heterogeneity in which our study will be limited due to the small size of the
corresponding sample in the Boston metro area. In particular, (i) only 0.53 percent of all pairs
reflect a match between two high school dropouts, (ii) only 1.59 percent of all pairs reflect a
match between two non-white workers, and (iii) only 1.92 percent of all pairs reflect a match
between two immigrants. Given the nature of the sample, it is not surprising that the our analysis
tends to be more precise in other dimensions of individual heterogeneity including age, the
presence of children, education (aside from high school dropouts), gender, and marital status.
For the second stage of our analysis, which examines the impact of neighborhood
characteristics on labor market outcomes including labor force participation and employment, we
add those prime age (25 to 59) individuals that are not currently employed; this sample has
163,594 observations.16 Table 2 reports summary statistics for this sample. The first column
reports the sample frequencies for each individual characteristic, while the remaining five
columns report labor market and commuting information: the fraction of individuals that are
currently employed, average weeks worked in the previous year, average hours worked per week
in the previous year, average earnings for the sample of individuals that were fully-employed in
the previous year, and average commute for those that are currently employed.17 College
graduates, married males, and whites display the strongest attachment to the labor force, with
respect to employment rates as well as hours and weeks worked. These groups also tend to work
the farthest away from home. On the other hand, high school dropouts and married females tend
to have weak labor force attachment and work close to home when employed.
15 It should be noted that the sample contains only a small fraction of Asians and Hispanics and so these
two groups are combined. Specifications where these groups are separated yield very similar results.
16 We again limit the sample used in each labor market outcome equation to individuals for which the
corresponding dependent variable has not been imputed.
17 The Census provides information on current employment and labor force participation as well as the
location of current workplace at the time of the survey in April 1990. Information on earnings, hours, and
weeks are reported for the previous year. Fully-employed in 1989 refers to any individual who worked at
least 40 weeks and at least 30 hours per week.