Between 1977 and 1997, US employment shifted dramatically in favor of industries that used skilled labor intensively. During this same period, some cities withered while others prospered. This paper examines employment growth in 39 industries across 316 cities to evaluate the importance of learning by doing, industry-specific location fundamentals, human capital externalities, and hiring cost explanations in these geographic shifts. Growth in industries that used skill intensively was particularly sensitive to the presence of local human capital. That growth was almost always negatively related to the initial size of the industry within the city implies a limited role for learning by doing and industry-specific location fundamentals stories.
According to a recent article in The New York Times, “slowly but steadily, a hightechnology
phoenix is rising on the steel shards of Easton, Bethlehem, and Allentown,
the towns that comprise the Lehigh Valley. [. . .] Some 65 high-technology companies have
moved into the valley or been founded there in the last five years, representing 38,500
jobs—or about 12 percent of the area’s total employment. It is a striking example of the extent to which high technology is transforming some of the most damaged areas of the
Rust Belt’s industrial economy” [11].
In the last quarter of the 20th century, the composition of industrial employment
transmogrified, shifting away from goods production and towards service production.
Between 1977 and 1997, employment in the steel industry fell by nearly 40 percent from
1.2 million to 730,000; employment in transportation equipment declined from just under
1.9 million to 1.67 million; and employment in textiles and apparel from 2.3 to 1.45
million, a decline of 36 percent.1 By contrast, employment in the service industries soared
by 250 percent in business services from 2.3 to 8.1 million, and by 216, 201, and 196
percent in health, professional, and social services.
It took nearly a quarter of a century for the steel cities of Pennsylvania to recover from
the steel shocks of the 1970s and 1980s. To take one particularly hard-hit area, overall
employment in the Beaver County, PA metropolitan area declined by 20 percent between
1977 and 1997, driven largely by an 87 percent (!) decline in steel employment. Was it
inevitable that overall employment would decline in response to the decline in demand for
steel? Blanchard and Katz [5] pointed out that negative shocks need not lead to a decline
of the city as a whole. As demand for the city’s output declines, the demand for labor falls,
thus reducing wages, in response to which firms in newer, rising industries might enter.
However, their evidence indicated that job formation in regions that experience declines in
labor demand responds only weakly to movements in wages; most of the adjustment to a
decline in labor demand takes the form of out-migration.2
If, as Blanchard and Katz [5] argued, firms respond only weakly to falling wages and
rents, what do firms respond to? A number of recent theories of urban growth might help
shed light how cities adjust to shifts in the pattern of labor demand. Learning-by-doing
stories (Premer and Walz [29], Brezis and Krugman [7], Soubeyran [33]) imply that an
initial presence of an industrymay give rise to faster subsequent employment growth in that
industry. Stories of growth based on an extension of the theory of location fundamentals
(Davis and Weinstein [10]) also suggest such a relationship. By contrast, stories of growth
based on human capital externalities (Lucas [20]) suggest that cities with higher average
levels of human capital might grow faster provided that human capital—in the language of
Henderson et al. [17]—gives rise to a dynamic externality, as would be the case if the rate
of technological progress is related to the current stock of human capital.
Most theories of urban growth take the overall structure of industry demand in the
economy as constant. However, the last quarter of the 20th century was characterized by
dramatic industrial change in which resources were being freed from certain sectors and
could flow into new ones. In particular, as will be seen, industries that used human capital intensively were on the rise and industries that used unskilled labor intensively were on
the decline. A simple potential explanation for the altered patterns of urban growth is that
firms in rising industries located in cities that offered relatively abundant supplies of skilled
labor. Such supplies would have varied, depending in part upon how much such labor was
freed by the shifts in demand, as well as on variations in the quality of schooling systems
and on random location-specific factors unrelated to labor demand (e.g., amenities) that
might affect location decisions.
This paper examines empirically patterns of sectoral and industry employment growth
across 316 US cities between 1977 and 1997 to distinguish between these explanations.
This paper builds on the empirical frameworks of Henderson et al. [17], Henderson [16],
and Glaeser et al. [14], examining the determinants of employment growth during the
turbulent 1977–1997 period at the sectoral and industry levels.
The remainder of the paper is organized as follows. Section 2 briefly reviews some
theories of urban growth and development. Section 3 provides an overview of industrial
change and city employment growth. Section 4 estimates employment growth regressions
for three inclusive sectors to establish some broad patterns in the data. Section 5 extends
the analysis to the industry level. Section 6 concludes the paper with a brief summary and
directions for future research.
Between 1977 and 1997, employment in goods-producing industries stagnated, while
employment in services, particularly those that use skill intensively, rose. This compositional
shift was accompanied by equally dramatic shifts in the distribution of employment
across cities. Using data on employment growth in 39 industries across 316 cities over this
period, this paper attempted to distinguish between competing explanations for the relative
success of some cities and relative failure of others.
A number of useful findings emerged. First, a larger initial presence of an industry
in a city was associated with slower subsequent growth, particularly in the newer, skillintensive
industries that dominated nation-wide employment growth. Second, although
some researchers downplay the role of relative wages, there is some evidence here that
employment grew more slowly in cities with higher wage levels, especially in newer, more
skill-intensive industries, which finding suggests that these industries may have been more
footloose than their older counterparts. Third, cities with relatively larger manufacturing
shares experienced slower growth in newer, rising industries, a finding that is consistent
with learning-by-doing stories in which cities that specialized in older technologies have
greater difficulty adjusting to new ones.
The local abundance of college graduates had a positive effect on employment growth,
especially in skill intensive and newer industries, which is consistent with the notion that
skill-intensive industries located where human capital is relatively abundant in order to
reduce search and hiring costs. However, it is also consistent with the existence of human
capital spillovers, as well as with the notion that cities with higher fractions of college
graduates provided a wider range of intermediate goods and services that were particularly useful to firms in newer, more skill-intensive industries. More intensive patent activity was
also associated with faster employment growth, particularly in newer industries, which is
consistent with the existence of localized dynamic spillovers. Although there was some
evidence that more inventive cities were particularly attractive to firms in newer industries,
there was no evidence that more skill-intensive industries were more attracted to inventive
cities.
The findings suggest a number of possible avenues for future research. Additional work
is necessary to understand more fully the adjustment of cities. Cities with particularly
large manufacturing sectors experienced particularly large negative shocks. The amount of
unskilled and semi-skilled labor freed up by such cities should have resulted in relatively
faster growth in trade and service industries that use unskilled labor intensively. However,
the empirical work in this paper found such a result only in personal services, and not in
trade industries. Because faster growth in the so-called export sector presumably generates
faster growth in demand for trade and services, it is important to differentiate between the
effects of demand and supply when studying the adjustment of cities.
Individuals who were displaced from their jobs during the 1970s and 1980s had to
decide whether to remain in their current city or to move. Using data on displaced workers
over the period 1979–1994, Kletzer [19] found that workers who were displaced out of
the manufacturing sector were most likely to find employment in trade and services, and
experienced substantial wage losses. Her work might be extended to consider the question
of geographic as well as industrial mobility. Future work could also examine the role
of geographic mobility in determining whether a worker displaced from manufacturing
remained within manufacturing, or more generally, in alleviating wage reductions.
