with heads of couple households, aged 19 to 64 and participating in the labor-market. Due to
data availability on previous occupation, we deleted individuals who never worked, that is only
18 individuals. The final sample contains 10,473 individuals, all of them being males.
3.2 Neighborhood typology
The agglomeration of Lyon presents a well-marked spatial structure, with some parts of the city
characterized by a concentration of disadvantaged communities. Figure 1 maps the percentage
of unemployed workers among labor-force participants. In most American cities, central neigh-
borhoods exhibit higher unemployment rates than peripheral neighborhoods. In Lyon also, the
neighborhoods with the lowest unemployment rates are found in the far periphery, but Figure
1 shows that the highest unemployment rates are found in the close periphery of Lyon’s munic-
ipality and not in the center.12 As seen in Figure 2, in some of the neighborhoods displaying
the highest unemployment rates, more than 50% of households (and even more than 70% for
some of them) are housed in the public renting sector. This pattern is very typical of French
cities and reveals the role that the public housing projects built in the 1970’s had in spatially
concentrating low-income households. The unemployment spatial structure is also quite related
to the distribution of education levels and professional statuses as well as to the distribution
of ethnic minorities. As a consequence, one can suspect the existence of neighborhood effects
affecting labor-market outcomes of public housing tenants and of other individuals located in
these neighborhoods.
Our typology of neighborhoods is aimed at reflecting for each neighborhood its social
composition and the neighborhood effects that might potentially affect job search and unem-
ployment. Therefore, it is built on the basis of the following variables: distribution of population
by education levels, percentage of executives and blue-collars in labor force, percentage of unem-
ployment, long-term unemployment and youth unemployment, percentage of household heads
of foreign nationality, and percentage of lone-parent families. Each of these neighborhood char-
acteristics is likely to affect individual unemployment propensity: low income levels (proxied
by professional status distribution) may decrease the global investment in human capital and
human capital spillovers; high unemployment rates as well as high rates of foreigners decrease
information on job opportunities and may give rise to statistical discrimination; low education
levels give low incentives for youths to invest in education and, together with high proportions of
12 Preliminary tests of the impact of time distance on unemployment probability did not however reveal any
empirical support for the spatial mismatch hypothesis, which was therefore left aside.
11