Content uploaded by Gordon Mulligan
Author content
All content in this area was uploaded by Gordon Mulligan on Jan 01, 2018
Content may be subject to copyright.
Chapter 2
Exploring Innovation Gaps in the American
Space Economy
Gordon F. Mulligan, Neil Reid, John I. Carruthers, and Matthew R. Lehnert
2.1 Introduction
The field of regional science emerged out of a need to understand the geographic
disposition of economic activity within and among metropolitan areas—what Isard
(1956) called the “spatial physiognomy” of development. Evolving on its own, with
little or no intervention on the part of planners and other urban policy makers, the
space economy has produced some of the most organized systems on the planet.
For example, the northeast corridor—nothing less than an American “Prometheus,”
as described in Gottmann’s classic (1961) text Megalopolis,anexplorationof
the Boston, MA—Washington, DC conurbation—is configured along an almost
perfectly linear NE !SW axis "750 kilometers in length. A stylized map of the
region, consisting of the population weighted fx,ygcoordinates of census block
G.F. Mulligan (!)
School of Geography and Development, University of Arizona, Tucson, AZ, USA
e-mail: mulligan@u.arizona.edu
N. Reid
Department of Geography and Planning, University of Toledo, Toledo, OH, USA
e-mail: neil.reid@utoledo.edu
J.I. Carruthers
Sustainable Urban Planning Program, The George Washington University, Washington, DC, USA
e-mail: jic@gwu.edu
M.R. Lehnert
Ph.D. Program in Spatially Integrated Social Sciences, University of Toledo, Toledo, OH, USA
e-mail: matthew.lehnert@rockets.utoledo.edu
©SpringerInternationalPublishingAG2017
R. Jackson, P. Schaeffer (eds.), Regional Research Frontiers - Vol. 1,
Advances in Spatial Science, DOI 10.1007/978-3-319-50547-3_2
21
neil.reid@utoledo.edu
22 G.F. Mulligan et al.
Fig. 2.1 Prometheus unbound
groups calculated by Carruthers et al. (2012), is shown in Fig. 2.1.1There are about
36,000 northeastern census block groups shown in the graph, and the correlation
coefficient between the x-coordinates (longitude) and y-coordinates (latitude) is
0.91; moreover, a linear regression of latitude on longitude, plus a constant, yields
an r2value of 0.82. Not only is the northeast region an economic titan, it is one of
the most structured objects on the planet—so structured that it might be considered
as natural as anything else created by life on earth. Though this organization
is enforced, in part, by physical geography, the overall form is one of spatial
agglomeration: economic forces acting to create a massive pattern of urbanization
that is fundamental to the prosperity of the United States (Fujita 1989; Krugman
1991a,b,1997; Fujita et al. 1999;Zenou2009; Glaeser 2011;FujitaandFrançois
2013).
At the national scale, a vast and enduring central place hierarchy has emerged
out of the same (Gabaix 1999; Ioannides and Overman 2003,2004;Gabaixand
Ionniades 2004)agglomerationprocessesthatgaverisetotheMegalopolis.The
overall pattern is shown in Fig. 2.2.Whileclearlythepatternisorderedaround
geographic advantage—for example, the transportation routes of the Great Lakes—
and disadvantage—for example, the remoteness of the Great Plains (Krugman
1991a,b,1997; Fujita et al. 1999)—the system is clearly organized (Christaller
1966;Lösch1954; Beckmann 1968;Pred1977; Berry and Parr 1988) and rank-
1For further details, see Renner et al. (2009).
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 23
Fig. 2.2 The American city system
ordered (Zipf 1949)inthemannerlongpredictedbylocationtheorists(Mulligan
et al. 2012). Nevertheless, the world has evolved since early theoretical frame-
works were advanced—and the space economy along with it. Hyperbole from the
heady days of the Internet explosion aside (see, for example, Cairncross 2001),
economic growth in advanced economies like that of the United States has become
increasingly uneven, demanding new frameworks for explaining and predicting the
processes and outcomes of economicgeography and urbanization.2The world itself
is urbanized (Angel 2012)withspatialstructureslookingalotlikethosedepicted
in Fig. 2.1 developing across the globe (Peirce et al. 2008;NelsonandLand2011;
Angel et al. 2012;Batty2013). Increasingly, these frameworks are grounded in the
concept of agglomeration economies, which give rise to specialization and patterns
of development that are often unpredictable—new economic geography models
typically predict “catastrophic” outcomes (see Head and Mayer 2004)—and highly
specialized (Behrens and Robert Nicoud 2015;DurantonandPuga2015; Combes
and Gobillon 2015;CarlinoandKerr2015).
With this history in mind, the present chapter presents an analysis examining
the fact that, over the past few decades, the United States space-economy has
separated into “have” and “have not” urban regions. Demographers, beginning with
Frey (2002), have also noticed this break and have gone so far as to suggest that
it may be irreversible. In the realm of economic geography, Florida (2014)and
Moretti (2004,2012), among others, have suggested that a permanent break has
2In the natural sciences, a good analogy is the relationship between Newton’s laws of motion,
which, to this day work well for many practical applications, but—their enduring power and
utility—were shown to be incomplete and wholly supplanted by Einstein’s theory of general
relativity.
neil.reid@utoledo.edu
24 G.F. Mulligan et al.
arisen in metropolitan America between growing areas that learn and innovate and
declining areas that do not. This notion has been reinforced in a recent case study,
by Storper et al. (2015), where the divergent paths of development taken by San
Francisco and Los Angeles are documented in great detail. Still others have noted
that a similar break has formed in the nation’s non-metropolitan areas between the
more urbanized micropolitan centers and the more isolated rural areas. In 50 years’
time, these disparities should be even starker as growth and change typically follow
dependency paths that have been established at earlier times.
Today, a number of priva te and pub lic agencie s monitor vari o us aspects o f
economic performance in the nation’s more than 350 metropolitan areas. Several
websites exist where one city can be directly compared to another using different
attributes of their labor markets: unemployment rates, average wages and salaries,
recent employment growth, and the like. Increasingly, though, attention is being
given to the notion of innovation—a term capturing the rise of high-tech industries
and the importance of those factors that sustain high-tech industries: knowledge,
creativity, advanced skills, entrepreneurship, research and development, patent pro-
duction, technology transfer, and communications infrastructure. In fact, a growing
number of observers now suggest that metropolitan areas should establish and
nurture innovation ecosystems that include key actors (corporations, universities,
etc.), service providers, venture capitalists, networks among the actors, and good
local or regional governance. However, outside of various case studies, not much is
known about how the complex features of innovation ecosystems might vary from
one metropolitan center to the next.
This chapter argues that key input, output, and contextual attributes are combined
locally to create each city’s innovation ecosystem. These attributes not only differ
from one place to another, but the manner in which they are combined varies by
location. As a start on the project 20 variables were selected in order to discern
how these innovation ecosystems currently vary across the nation’s very largest
metropolitan areas. Using multivariate techniques, these production ecosystems
were deconstructed in order to reveal the underlying dimensions of innovation
that are common to all of the nation’s 350-plus metropolitan areas. Once the
primary dimension of general innovation has been identified for all metros, the
other n #1lessimportantdimensionswillrevealhowspecificcitiesdifferentially
combine their knowledgeable and educated workforces to produce patents, engage
in entrepreneurship, and create high value-added outputs. As a consequence, the
overall innovative index of any metropolitan economy can be estimated by first
generating its score on each of the latent dimensions and then, second, adding up
those performance scores across all of the dimensions.
Today—despite advances in technology and the reduced tyranny of distance in
advanced economies—geography matters more than ever (de Blij 2012). Cities
and regions that are at the “have not” end of the spectrum described above may
be able to improve their performance by pursuing policy strategies that enhance
their connectivity around the globe. Economists increasingly emphasize competitive
cities (see Glaeser and Joshi-Ghani 2015) as a means of achieving progress and
the most recent State of the World report, published by the Worldwatch Institute,
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 25
highlights the need for urban policy to not only be environmentally sound—but also
keyed towards connectivity and inclusivity. Looking toward the future of regional
science, it seems that competitiveness, inclusivity, and connectivity on the global
stage will be fundamental to regional success, even as local innovation remains a
main driver. Even as the analysis contained in this chapter illustrates how innovation
works to establish regional productivity, the findings are used to look to a future
wherein the world is smaller and evermore interconnected.
2.2 Background
There is, not surprisingly, a large literature dedicated to trying to understand why
some cities and regions are more successful than others in building economies that
are both vibrant and resilient and which provide a high quality of life for their
residents. While no one would disagree with the premise that innovation plays a
vital role in urban well-being, there are disagreements as to how to foster innovation
within metropolitan economies (see Brueckner 2011 and Cheshire et al. 2014 for
wide-ranging and recent overviews of the state of economic thinking on cities).
One school of thought, championed primarily by Richard Florida, emphasizes
the importance of creativity, diversity and tolerance as key drivers of innovative
economies. The basic argument is that such values drive innovation and urban
growth by establishing a creative environment which attracts and retains the bright-
est and smartest—and most innovative—people (Florida 2014). Not surprisingly
perhaps, Florida and his colleagues have found evidence in support of these ideas.
For example, Lee et al. (2010)examinedtheroleofhumancapital,creativity
and diversity, and industry mix in explaining variations in innovation across 284
metropolitan statistical areas (MSAs). Their findings suggest that variations in
innovation across MSAs are related to differences in levels of human capital,
creativity and diversity while, at the same time, are not related to differences in
industrial mix.
The impacts of not being able to attract workers with the right skill sets are
discussed by Lautman (2011)whosuggeststhatashortageofqualifiedworkers
is creating a zero-sum labor market that results in communities “stealing” talent
from each other. This scenario is the result of two failures. First, low fertility rates
among the baby boom generation resulted in a labor shortage and second, the
education system failed to equip the available labor force with the requisite skills
demanded by the modern economy. Lautman argues that communities unable to
attract qualified workers have bleak economic futures. Florida et al. (2011)found
evidence that analytical and social intelligence skills are associated with higher
wages and that individuals with such skills tend to be located in larger metropolitan
areas. In sharp contrast, physical skills are associated with lower wages where such
skills are being concentrated in smaller metropolitan areas. Florida et al. argue
that skills are more important than either education or human capital in explaining
geographic differences in wages (Florida et al. 2011). Members of the creative
class (and by extension the communities in which they live) are more resilient than
neil.reid@utoledo.edu
26 G.F. Mulligan et al.
other workers to economic downturns (Gabe et al. 2012). In the case of the Great
Recession, for example, this greater resiliency has been attributed to the fact that
the creative classes held occupations that were less likely to be negatively impacted
by the recession, and indeed, were bolstered by post-recession structural change.
Those with creative class occupations were less likely to be unemployed during the
study period, 2006–2011, than those holding non-creative working or service class
occupations. Much of the work on the economic resiliency of the creative class
suggests that it is the possession of skill sets, rather than formal education that is
important (Gabe et al. 2012).
The role of institutions of higher education, particularly universities, has also
been examined extensively in the literature. University contributions to economic
development have been categorized into generative and developmental roles, with
the former emphasizing knowledge-driven economic development and the latter
focusing on capacity building (Gunasekara 2006). The generative role is derived
from the Triple Helix concept in which the role of universities is expanded beyond
the traditional ones of teaching and research to include economic development
(Etzkowitz and Leydesdorff 2000). Through knowledge transfer and commercial-
ization activities, in particular, universities can make important contributions to
both local and regional economic development. These activities often manifest
themselves in licensing agreements and spin-off companies. Not surprisingly,
some universities and regions are more successful than others in creating spin-off
companies (Calzonetti and Reid 2013). In regions with a track record of generating
spin-off companies, there can be considerable benefit to the local economies. For
example, the University of Utah created 188 spin-off companies between 1970 and
2010, 61% of which still had operations in Utah. These spin-offshave had a positive
impact on the state economy through the creation of a significant number of jobs that
pay well above the state average (Crispin 2010).
Knowledge spillovers have also been credited with contributing to entrepreneur-
ship and innovation within urban areas. The Knowledge Spillover Theory of
Entrepreneurship (KTSE) suggests that entrepreneurship is not exogenous but rather
the result of the presence of knowledge spillovers. In other words “entrepreneurial
behavior is a response to profitable opportunities from knowledge spillovers” (Acs
et al. 2013: 759). Opportunities for entrepreneurial activity in the form of start-up
companies present themselves when incumbent firms create knowledge that they do
not commercialize (Audretsch and Keilbach 2007). This knowledge is appropriated
by entrepreneurs who utilize it to establish start-up firms. Capitalizing on these
knowledge stocks, however, requires a particular set of “skills, aptitudes, insights,
and circumstances” that are neither ubiquitous nor uniformly distributed through the
population (Acs et al. 2009). Carlino et al. (2007)exploredtherelationshipbetween
patent intensity (patents per capita) and employment density across American
metropolitan areas.3Their findings suggest that the rate of innovation is enhanced
3It is worth noting that patent counts can be problematic for cross-country comparisons for a
number of reasons. For example Japan applied a higher standard for judging innovation with the
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 27
where employment density is higher, the local economy is more competitive (many
small rather than a few large firms), manufacturing jobs account for a larger share of
total employment, and a larger share of the adult population have a college degree.
Shapiro (2006)examinedtheroleofcollegegraduatesonMSAemploymentgrowth.
He found that a 10% increase in a MSAs concentration of college-educated residents
was associated with a 0.8% increase in subsequent employment growth. College
graduates impacted employment growth by enhancing productivity growth and the
quality of life in an MSA.
While national scale studies can be highly informative and often bring valuable
insights to the processes that underpin variations in urban economic performance,
it is also important to remember that place, particularly the uniqueness of place,
matters. All urban areas are influenced by their history and, as a result, are
susceptible to lock-in and path dependency (David 1985;Arthur1989). Each
city has its own story to tell. At the local level it is important to understand
this uniqueness and the historical backdrop of how a particular city reached the
particular position that it now occupies. For example, Reese et al. (2014) document
the negative influence of race relations anda“regimeless”governance culture in
Detroit’s economic fortunes. In an equally compelling piece on Boston, Glaeser
(2005) recounts that city’s fascinating economic history and highlights the critical
role played by human capital, institutions of higher education, and labor force skill-
sets in the city’s post-1980 renaissance.
2.3 Empirical Analysis
The main intent of the empirical analysis is to reveal the substantially different ways
that large American cities now produce their various goods and services. A mixture
of input, output, and contextual variables are used in the multivariate analysis of
352 metropolitan areas at one point in time. These groups of variables measure: (1)
the quality of the workforce, (2) the incidence of entrepreneurship, (3) the intensity
of patent production, (4) overall innovation, (5) overall productivity, and (6) the
metropolitan context for innovation.
2.3.1 Workforce
Six different variables are chosen here (1) to reflect general creativity, (2) to capture
the general benefits that flow from higher education, and (3) to highlight the
special advantages of attracting and maintaining youthful, educated workers. First,
result that it is more stringent in awarding patents than a number of other countries, including the
United States. See de Rassenfosse et al. (2016).
neil.reid@utoledo.edu
28 G.F. Mulligan et al.
Richard Florida’s widely-known twin measures of creative workers are adopted (see
martinprosperity.org/tag/creativity-index). The share of metropolitan workers in the
creative class, CCLASS, is used to capture the importance of having city workers
engaged in the fields of science and technology,design and architecture, and various
other professional areas. Alternatively, the creativity index, CREATE, is used to
measure the incidence of the 3Ts—technology, talent, and tolerance—in the local
environment. Although highly correlated (Pearson’s r D0.80), these variables are
not identical. The third variable in this cluster is the standard measure of human
capital, DEGREE, which reflects the percentage of the MSA adult population
25 years of age and older having a bachelor’s degree. This variable correlates
significantly (in the 0.55–0.60 range) with Florida’s variables, but the incidence
of college degrees is relatively high in certain places, like retirement areas, that are
not very innovative.
Three other variables are used to address the special importance of attracting
university-educatedworkers in the 18–44 age cohort.The annual Leading Locations
studies provide ordinal data regarding this youthful “prime workforce” and both
the current level and recent change in this demographic group are addressed (see
www.areadevelopment.com). So data are adopted for: (1) the percentage share of
the prime workforce, PRIMWF, in the total workforce; (2) the 3-year percentage
change, CHPRIM, in the prime workforce; and (3) the 1-year percentage rate of
in-migration, INPRIM, for this prime workforce. To facilitate the interpretation,
all the ranks are reversed so that higher scores are better. PRIMWF and INPRIM
correlate strongly with the variables in the first cluster; however, CHPRIM is a lot
more volatile and does not correlate with any other variable in the entire group.
2.3.2 Entrepreneurship
While entrepreneurship is widely known to be a key in regional growth, there
is some disagreement about how best to measure that activity. Three approaches
appear to dominate the current literature: (1) self-employment rates, (2) new product
formation rates, and (3) new business startup rates. Much research has argued that
startups, (usually) taken as those businesses opened in the previous year, are an
especially important part of Schumpterian “creative destruction” dynamics. Recent
studies have focused on how activities (STEM) of a more technical nature are
especially important in the creation of new jobs. Moreover, several observers have
argued that high-tech multipliers are much higher than other multipliers. The anal-
ysis here makes use of data provided in the report prepared by Hathaway (2013)for
the Kauffman Foundation.This report is especially valuable because it differentiates
general startups GENTRE from high-tech startups TENTRE, although the levels
do correlate highly (r D0.95 in 2010). Here startup density figures, or location
quotients using a national base, are given for each measure of entrepreneurship in
1990 and 2010. The 2000–2010changes in those concentrationmeasures, CHGENT
and CHTENT, are estimated by subtracting the 1990 densities from the 2010
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 29
densities and dividing that change in half. The degree of correlation for the density
changes (r D0.78) is somewhat lower than for the densities.
2.3.3 Patents
Patent production has been studied for years, especially in the more advanced
economies. The analysis here uses the figures disclosed for utility patent grants
in all classes, as provided annually by the United States Patent and Trademark
Office (2015). Here the metropolitan location is attributed to the residence of the
first-named inventor. The website www.uspto.gov provides figures for the patent
volumes, which correlate highly with city size, and these figures must be standard-
ized to address local specialization. The raw patent figures PATENT for 2010 and
2000 were transformed into patent densities LQPATE for each year by computing
location quotients using the local and national population figures. Next, the change
in patent density CHLQPA was computed as the 2010 location figure minus the
2000 location quotient. Patent densities correlate significantly with entrepreneurship
densities (r is approximately 0.50 in both instances), but there is virtually no
correlation between the alternative measures of change. Finally, a composite
variable ENTPAT was computed as GENTRE*LQPATE to capture the possible
synergy effects between entrepreneurial activity and patent production in 2010.
2.3.4 Overall Innovation
Acomprehensiveviewofmetropolitaninnovationisgivenbythecompositevari-
able, INNOVA, which is calculated by Statsamerica (see www.statsamerica.org).
This index is comprised of four distinct sub-indices—human capital, economic
dynamics, productivity and employment, and economic well-being—where each
sub-index is in turn comprised of several performance indicators. The human capital
sub-index comprises 30% of the total score and, as considered above, this reflects
both the age and the educationof the workforce. The economic dynamicssub-index,
which comprises a further 30%, is different from all of the other variables used here
because it addresses venture capital and broadband connection. The other two sub-
indices are to a large degree captured in the variables that are considered below.
The basic problem with this composite variable is that many of the input data are
known to be highly correlated (redundant),but it remains unclear where this double-
counting actually occurs.
neil.reid@utoledo.edu
30 G.F. Mulligan et al.
2.3.5 Overall Productivity
The nation’s metropolitan areas vary tremendously in terms of their per capita
productivity. In fact, gaps in urban productivity have been growing steadily for
decades. To capture this aspect of the nation’s metropolitan landscape, the analysis
makes use of productivity figures provided each year by the Bureau of Economic
Analysis (see www.bea.gov). Highly innovative cities generally have high levels of
gross domestic product per capita, GRDPPC, but not all highly productive cities
are in fact highly innovative. This occurs, in part, because some cities remain
productive in high value-added industries that are in the declining part of their
product cycle. In our case, the degree of correlation between GRDPPC and INNOVA
is r D0.64.
2.3.6 Metropolitan Context
Four other variables are included in the multivariate analysis in order to provide
suitable context for the different variables that have been discussed above. The first
of these is metropolitan population, POPULA, where it is known that the largest
cities, like New York and Chicago, have very different production ecosystems—
with highly beneficial internal and external scale economies—than the smallest
cities, like Yuma and Dothan. A second variable, WAGSAL, addresses average
wages and salaries across the cities, and is adopted to recognize that qualitative
differences exist in the metropolitan workforces. Moreover, new patent production
and entrepreneurship tend to be promoted in those areas that already enjoy higher
wages and salaries. Two other variables are chosen to control for the highly
variable composition of both metropolitan income and employment in the United
States. One variable, EARRAT, measures the ratio of earned income to total
income and thereby distinguishes retirement places and depressed areas, with high
amounts of non-earnings income, from the nation’s most vibrant and productive
cities. A second variable, EMPRAT, measures the ratio of total employment to
population and distinguishes those productive places with high (gross) labor force
participation rates from those unproductive places with low participation rates.
These variables are all taken from or manufactured from BEA data for the year
2010.
2.3.7 Empirical Results
Table 2.1 shows the factor loadings for each of the 20 variables across 10 latent
dimensions (factors) after applying a Varimax rotation. These numbers basically
represent the degree of correlation of the variables with the different orthogonal
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 31
Tab l e 2 . 1 Loadings for the ten orthogonal factors
F1 F2 F3 F4 F5 F6 F7 F8 F9 F10
CCLASS 0:65 0:32 0:07 0:13 0:10 0:42 0:02 0:00 0:01 0:35
CREATE 0:75 0:20 !0:04 0:21 0:01 0:32 0:03 0:04 0:15 0:34
DEGREE 0:81 0:15 !0:01 0:20 0:20 !0:08 0:05 0:06 0:29 !0:13
PRIMWF 0:66 0:19 0:08 0:18 0:26 0:56 !0:06 !0:10 !0:01 !0:06
CHPRIM !0:02 !0:02 0:05 0:02 0:12 0:09 0:02 !0:99 0:01 0:01
INPRIM 0:86 0:09 0:15 0:11 0:18 !0:07 0:05 !0:01 !0:03 0:05
GENTRE 0:45 0:44 0:38 0:16 0:18 0:17 !0:02 !0:02 0:58 0:16
CHGENT 0:04 0:10 0:93 !0:13 0:05 !0:09 !0:09 !0:05 0:12 !0:07
TENTRE 0:45 0:41 0:39 0:12 0:17 0:14 !0:01 !0:02 0:60 0:12
CHTENT 0:06 !0:09 0:92 !0:05 0:05 0:09 !0:02 !0:05 0:12 !0:07
PATE N T 0:14 0:55 !0:13 0:75 0:12 0:05 0:09 !0:05 0:05 0:13
LQPATE 0:30 0:88 !0:05 0:05 0:10 0:14 0:14 0:04 !0:04 0:11
CHLQPA 0:03 0:17 !0:04 0:03 0:00 !0:02 0:98 !0:01 0:00 0:02
ENTPAT 0:15 0:90 0:07 0:08 0:11 0:02 0:10 !0:01 0:28 0:04
INNOVA 0:60 0:46 !0:03 0:21 0:21 0:23 !0:10 0:03 0:22 0:32
GRDPPC 0:24 0:21 0:01 0:26 0:76 0:21 0:05 !0:04 0:13 0:39
POPULA 0:17 0:04 !0:09 0:95 0:03 0:11 !0:02 !0:06 0:07 0:06
WAG SAL 0:27 0:35 !0:16 0:44 0:35 0:13 0:06 !0:01 0:15 0:60
EARRAT 0:08 !0:10 !0:02 0:13 0:42 0:80 !0:02 !0:13 0:11 0:12
EMPRAT 0:30 0:09 0:16 !0:03 0:85 0:26 !0:02 0:01 0:05 !0:06
neil.reid@utoledo.edu
32 G.F. Mulligan et al.
dimensions. The ten rotated factors, identified later, are arranged in order of their
declining importance in accounting for the variance in the 20 !352 data matrix;
together the ten factors account for more than 92% of that variance. Many variables
load heavily and few lightly on Factor 1: hence, it is recognized as the primary
underlying dimension of the factor analysis. In fact, this single factor accounts for
nearly 20% of the variance in the data set. The frequent large loadings reveal that
this factor represents youthful, educated, and talented workers who are very active
in entrepreneurship, patent production, and overall innovation.
The other latent dimensions of the analysis are identified in Table 2.2,where
the amount of the total variance accounted for by each factor is also shown.
Here the second factor, F2, which accounts for 14.5% of the variance, captures
those creative workers that are very involved in patent production despite the
fact that these workers have lower levels of university education than in the
first factor. Factor scores, which are standardized, are computed in order to see
how the various metropolitan areas perform on each factor. Table 2.3 shows that
Ithaca (3.02) and Ann Arbor (2.57), both university-dominated cities, are ranked
highest on F1, while San Jose (13.49) and Boulder (6.10), both well-known
high-tech centers, are ranked highest on F2. The various loadings in Table 2.1
indicate that while Ithaca and Ann Arbor have somewhat younger workforces,
San Jose and Boulder are much more productive in patents. Looking down
the top-15 lists for F1 and F2, two places—Ann Arbor and Corvallis—actually
appear twice. So these two cities are identified as being especially innovative in
that entrepreneurship, patent production, and GDP per capita are all rated very
highly.
The revealed factors or dimensions basically deconstruct the various metropoli-
tan economies in ten separate ways. In a sense, then, the city’s score on each
dimension reveals something unique or important about the innovation ecosystem
of that metropolitan area. Some metros are productive but not so entrepreneurial,
others are entrepreneurial but not so active in patent production, while still
Tab l e 2 . 2 Interpreting the ten orthogonal factors
Factor %Variance Interpretation
F1 19:7 You n g , ed u cat e d, a n d hig h ly t a l en t ed in n ovat o r s
F2 14:5 High patent production with high entrepreneurship
F3 10:7 Ascending entrepreneurial
F4 9:8 Large cities with diverse patents and entrepreneurs
F5 9:6 Young w o rke r s i n ver y h igh va l ued - a dd e d ind u str i es
F6 7:6 Young , t al e nte d w o rke r s in hi g h va lu e - ad d e d i n dus t ri e s
F7 5:2 Ascending in patent production
F8 5:1 Significant rise in young workers
F9 5:0 High entrepreneurship but low patent production
F10 4:8 Tal e n t e d w o r kers w i t h m o d erate entrepreneurshi p and patents
F8 is flipped (reversed) for interpretation reasons
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 33
Tab l e 2 . 3 Top - 1 5 metro areas on each factor
Fac t or 1 Fac t or 2 Fac t or 3 Factor 4 Fac t or 5
Ithaca San Jose Boulder New York Midland TX
Ann Arbor Boulder Ft Collins Los Angeles Casper
Charlottesville Boise City Cheyenne Chicago Bridgeport
Corvallis Corvallis Missoula San Francisco Sioux Falls
Olympia Burlington VT Corvallis Philadelphia Ocean City
Gainesville FL San Francisco Champaign Boston Columbus IN
College Station Santa Cruz Bend Dallas Charleston WV
Naples Rochester MN Dover Houston Dubuque
Durham Austin Sioux Falls Miami Napa
Tucson Seattle Seattle Seattle San Jose
Lawrence Bremerton Colorado Sp San Jose Fargo
Iowa City Rochester NY Grand Junction Atlanta Lake Charles
Prescott Fort Collins Boise City Detroit Atlantic City
Athens Ann Arbor Kansas City Was hington Sioux Falls
Was hington Raleigh Crestview Phoenix Madison
Fac t or 6 Fac t or 7 Fac t or 8 Fa ctor 9 Fac t or 10
Manchester Bremerton Coeur d’Alene Boulder Bridgeport
Provo Corvallis Harrisonburg Manchester Durham
Wor cester San Jose War ner Robins Provo Trenton
Burlington VT Seattle Florence Denver Washing t on
Ames Burlington VT Lawrence Salt Lake City Midland TX
Rochester MN Rochester MN Hot Springs Washing t on Hartford
Des Moines Bloomington IL Casper Dallas Anchorage
Columbus OH Reno Kingsport Fort Collins Kokomo
Raleigh Santa Fe Bay City Colorado Sp San Francisco
Morgantown Champaign Spartanburg Phoenix Hinesville
Springfield IL Spartanburg Ogden Las Vegas Kennewick
Atlanta Iowa City Michigan City Orlando Palm Bay
Akron Portland ME Nashville Ann Arbor Boston
Santa Cruz Tucson Grand Forks Huntsville Seattle
Greenville NC Provo San Angelo Cheyenne Huntsville
Factor F8 has been flipped for interpretation
others are simply large and perform about average in both patent production and
entrepreneurship.
After examining all of the top-15 lists in Table 2.3, it becomes apparent that
some metropolitan areas appear much more frequently than others. For instance,
San Francisco appears on the top-15 lists for F2, F4, and F10 and Seattle appears
on the top-15 lists for F2, F3, F4, and F10. So these two large metros demonstrate
neil.reid@utoledo.edu
34 G.F. Mulligan et al.
astrongsimilarityintheiroverallinnovativeecosystems.Infact,avectoroftheten
ranked scores (where lower is better) provides a simple signature for the innovation
ecosystem of each and every metropolitan area. In the case of San Francisco, this
vector is (181, 6, 89, 4, 75, 229, 15, 137, 87, 9), and in the case of Seattle this
vector is (185, 10, 10, 10, 117, 104, 4, 91, 25, 14). Here the only significant rank
difference is seen on F6 where Seattle has substantially younger workers. The
pattern of ranks in both of these vectors is very different from that of a moderately
innovative place like Tucson, AZ (10, 135, 316, 121, 324, 267, 13, 83, 270, 131) or
a weakly innovative place like Dalton, GA (351, 131, 113, 193, 169, 241, 278, 102,
45, 84).
Earlier in the chapter it was noted that an innovation index can be constructed by
assessing performance on each dimension and then aggregating this performance
across the different orthogonal dimensions. Factor scores, FS, which are standard-
ized across the cities, can be used to gauge performance on each dimension and then
these factor scores can be summed. However, there is some disagreement about
whether or not these scores should be weighted by the importance of each factor
before the summation occurs. So in the first instance the unweighted index INNOV1
is calculated as:
INNOV1DFS1CFS2C$$$CFS10
while in the second instance the weighted index INNOV2 is calculated as:
INNOV2Dw1FS1Cw2FS2C$$$Cw10FS10
where the weights are calculated by taking either the square root of each dimen-
sion’s (rotated) eigenvalue or the square root of the percentage of the variance that
is accounted for by that dimension. So the first index treats all factors equally, while
the second index places greater emphasis on those factors that are found to be more
important in the analysis. However, this added importance might only reflect the fact
that the redundancy of having more highly correlated variables is being extracted by
those factors with larger eigenvalues.
An examination of the results indicates that our methodology penalizes those
metros that have experienced downturns in either entrepreneurship or patent
production in recent times. If a metropolitan area has a large negative score on
Factor 5 or Factor 7, then its overall performance can be severely compromised.
Considerable drops in entrepreneurship depress the overall innovation scores for
Riverside, CA and Provo, UT while considerable drops in patent production depress
the overall innovation scores for Austin, TX and Boise City, ID. So if the analyst
preferred an index that did not address change, then both INNOV1 and INNOV2
could be recalculated with the dimensions F5 and F7 completely removed from the
formula.
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 35
Tab l e 2 . 4 Top 1 0 % i n n ovators: unweighted and weighted sc o r e s
Rank City INNOV1 City INNOV2
1Boulder 17:68 Boulder 53.21
2San Jose 14:94 San Jose 50.43
3Seattle 14:11 Seattle 37.34
4San Francisco 11:74 San Francisco 35.74
5Was hington 11:71 Wash i ngton 32.97
6Denver 9:30 Corvallis 30.23
7Corvallis 9:11 New York 29.19
8Durham 8:20 Denver 23.29
9New York 8:00 Durham 22.77
10 Portland OR 7:11 Boston 21.49
11 Boston 6:28 Fort Collins 20.01
12 Kansas City 6:27 Burlington VT 19.90
13 Salt Lake City 6:26 Raleigh 19.39
14 Des Moines 6:11 Trenton 17.34
15 Cheyenne 5:78 Portland OR 17.30
16 Raleigh 5:61 Des Moines 17.07
17 Chicago 5:52 Kansas City 16.81
18 Burlington VT 5:48 Cheyenne 16.70
19 Cedar Rapids 4:91 Los Angeles 15.45
20 Los Angeles 4:87 Atlanta 15.10
21 Sioux Falls 4:82 Huntsville 15.04
22 Atlanta 4:79 Chicago 14.77
23 Huntsville 4:43 Colorado Sp 14.72
24 New Orleans 4:42 Champaign 14.67
25 Charlotte 4:39 Bremerton 14.61
26 Philadelphia 4:16 Philadelphia 14.34
27 San Diego 4:09 Salt Lake City 13.92
28 Baltimore 4:04 San Diego 13.46
29 Jacksonville 4:02 Richmond 13.45
30 Provo 3:96 Rochester MN 13.18
31 Casper 3:95 Missoula 12.87
32 Bridgeport 3:94 Bridgeport 12.52
33 Richmond 3:86 Minneapolis 12.48
34 Minneapolis 3:83 Provo 12.35
35 Portland ME 3:81 Santa Fe 12.33
INNOV2 is calculated using the square root of the percentage of the variance extracted by each
factor
In any case, Table 2.4 shows these innovation indices for the top 35 metropolitan
areas. Figures 2.3 and 2.4,which accompany the table, show the cities that make
it on the list twice and once, respectively. Here the scores on Factor 8 have been
inverted or flipped as was noted earlier. The lists are much the same, but there are
neil.reid@utoledo.edu
36 G.F. Mulligan et al.
Fig. 2.3 Metropolitan areas listed twice in Table 2.4
Fig. 2.4 Metropolitan areas listed once in Table 2.4
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 37
seven metros (20% of the total) on one list that are not on the other; in fact, the
degree of correlation between the two indices across all of the metropolitan areas is
rD0.92. So 28 cities qualify as a top 10% innovator on both indices, while a further
14 cities qualify as a top 10% innovator on one of the two indices.
When using INNOV1 Boulder, CO stands out alone at #1 (its score is 17.68), San
Jose and Seattle then group closely together at #2 and #3 with scores between 14.0
and 15.0, and then San Francisco and Washington group together at #4 and #5. After
that the scores slowly descend until reaching the two least innovative cities: Naples,
FL (#4.95) and Punta Gorda, FL (#5.79). Alternatively, when using INNOV2,
Boulder (53.21) and San Jose (50.43) distance themselves from the remainder of
the pack; five other places have scores ranging between 29.0 and 38.0, and only
then does a pattern of steady decline appear in the weighted index. Now the two
least innovative cities in the nation are identified as being Sumter, SC (#14.11) and
Sheboygan, WI (#14.20).
Although this can only be a rough indicator of geographic differences in
metropolitan innovation, the average index values for the 117 Sunbelt cities are
calculated to be INNOV1 D#1.14 and INNOV2 D#0.52, while the average index
values for the 235 Snowbelt cities are INNOV1 D0.53 and INNOV2 D0.26. So
the Sunbelt appears to be marginally less innovative than the Snowbelt. Moreover,
separate OLS regressions show that this spatial distinction is significant for INNOV1
at the 0.05 level and for INNOV2 at the 0.15 level. However, once these OLS
regressions include metropolitan population as a second independent variable, the
Snowbelt-Sunbelt distinction becomes significant at the 0.05 level in both instances;
moreover, the new results indicate that a strong positive relationship exists between
innovation and metropolitan population in both cases. The nation’s most innovative
metropolitan areas are also mapped out in order to discern whether or not these
metros follow some sort of geographic pattern.
One of the major shortcomings of indices is that they often remain descriptive
tools and are infrequently used to make testable inferences—in this case about other
attributes of metropolitan labor markets. Regional science should become interested
in how the two constructed innovation indices correspond to the most widely-used
measures of health in labor markets: those being the current unemployment rate,
recent employment growth, recent wage growth, and recent per capita productivity
growth. Moreover, the field should be interested in how the performances of our
two indices compare to the performances of the three input variables CCLASS,
CREATE, and INNOVA that were used in their construction.
For assessing (and not predicting) recent growth, the time period 2007–2013
was selected, in order to include several years both before and after the year 2010.
Table 2.5 shows some Pearson correlation coefficients between the two innovation
indices calculated above and these four important labor market attributes. The table
also shows correlation coefficients for the three input variables just mentioned.
Evidently, a strong relationship holds between all the measures and the 2010
unemployment rate across the nation’s metropolitan areas, where metros with more
talent or higher innovation rates enjoyed significantly lower unemployment rates.
Very clearly, too, the five measures are strongly correlated with recent employment
neil.reid@utoledo.edu
38 G.F. Mulligan et al.
Tab l e 2 . 5 Correlations between innovation measures and labor market attributes
Index Un rate 2010 Job growth Wage gr owth GDP growth
INNOV1 !0.29** 0.29** 0:13"0:11"
INNOV2 !0.36** 0.32** 0:12"0:08
CCLASS !0.33** 0.20** 0:01 0:06
CREATE !0.18* 0.20** !0:11" !0:05
INNOVA !0.29** 0.21** !0:01 0:02
Growth is during 2007–2013; ** significant at 0.01 level; * significant at 0.05 level.; n D352
growth, although the weighted index INNOV2 outperforms the others by a fair
amount. However, the two new indices are clearly superior to the three input
variables in the areas of wage growth and per capita GDP; in fact, CREATE actually
has a significant negative relationship with growth. Evidently, innovation is now a
determining factor in various aspects of metropolitan growth and change, but the full
extent of its effects will have to be established by more sophisticated econometric
models. But, for present purposes, it is worth stating that OLS regressions, using
logarithms, indicate that a 1% increase in either innovation index decreases the
unemployment rate by approximately 0.15% and increases employment growth by
approximately 0.13%.
2.4 Challenges to Regional Science
The multivariate analysis undertaken above clearly indicates that U.S. metropolitan
areas currently exhibit a great amount of heterogeneity in the composition of
their high-tech economies. This means, among other things, that the prospects
for future metropolitan growth are highly polarized where, very clearly, there
will be winners and losers. These findings suggest that regional science is now
confronted with several theoretical problems and empirical issues that must be
addressed before policymakers can discern the degree to which the metropolitan
space-economy will be divided into two broad welfare camps. These topics can
be roughly categorized as follows: (1) understanding how urban creativity leads
to innovations; (2) understanding how these innovations create new industries; (3)
understanding how these new industries are transformed into old industries; and (4)
understanding how global connectivity can capitalize on local agglomeration forces.
At the same time, it will be important for regional science to understand how human
capital, startups, and patent production, among other factors, interact in order to
drive city growth during all three of these stylized phases or time periods.
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 39
2.4.1 Creativity
Human creativity is uncharted territory for regional science because, outside of a
few contributions by people like Ake Andersson (2009), little has been written about
why new ideas often appear only at certain places and only at certain times. Here,
the discipline of psychology has much to offer and regional science should begin
thinking—much like the field of behavioral economics has—about how to extend
standard models in psychology to incorporate features of the economic landscape
like population density and the age-gender composition of the population.
AgoodplacetobeginiswiththeresearchofDeanSimonton(1999)on
Darwinian approaches to creativity. In one paper Simonton (1997)outlinesatwo-
step model that addresses time-specific rates for both ideation (new ideas) and
elaboration (their application). This leads to the construction of a productivity curve
P(t) for each person where:
P.t/DcŒexp.-at/#exp.-bt/!
Here, a is the ideation rate and b is the elaboration rate and both of these rates have
negativesigns because the fixed stocks (representing all combinatorial possibilities)
of new ideas and their elaborations are both being continuously exhausted over
time. The constant c Dabm/(b #a) where m represents the maximum number
of ideational combinations the person can conceive in a lifetime. So the genius or
long-lived person can conceive of many more possibilities than can the average or
short-lived person and m can be set much higher for these cases. So the stylized
productivity curve rises quickly at first, then peaks a number of years later (usually
at 20–25 career years), and eventually falls, but at an ever-decreasing rate. The rates
for ideation and elaboration (0 < a,b < 1) will both vary from one activity (industry)
to another so the two-step model actually exhibits a fair bit of flexibility.
It would be a useful—but not entirely a straightforward—exercise to expand
this individual model to include several matters of interest to regional scientists.
One important step would involve identifying where so-called geniuses prefer
to live. Our current thinking is that superstars prefer living in highly urbanized
environments in order to enjoy interaction with other highly endowed people and
to maximize their opportunities to find a compatible partner. However, there are
indications that many of these highly talented people actually prefer some mixture of
urban and non-urban lifestyles. At the very least, there is a need to better understand
how superstars selectively demand the key attributes provided by these two very
different residential settings. Simonton’s model could be modified to portray a
hierarchy of creativity by using the shift parameters in his model while recognizing
that large cities have more industries than small cities. Another consideration
would be some appreciation of demographic heterogeneity. Metropolitan areas have
different numbers of persons in various age-gender cohorts and Simonton’s model
could be adapted to address this fact. So the city-wide productivity curve for new
ideas and elaborations would necessarily aggregate across all the various individual
neil.reid@utoledo.edu
40 G.F. Mulligan et al.
productivity curves for each resident of the city. One more important consideration
would be population density. A lot of the thinking in the NEG framework is based
on the notion of externalities arising from agents that are located close together in
space. Presumably Simonton’s model couldbemodifiedsothatbothoftheratesa,
bwould rise with higher density and this in turn would raise the city’s aggregate
productivity curve to a higher level.
2.4.2 New Industries
Economists now regularly differentiate between those innovations that are more
important for subsequent growth and those that are not. Following Schumpeter,
the former—and much more influential group of commercial applications—are
usually called disruptive or radical innovations. Uber, the new taxi service, is an
example of a deliberately disruptive business model that seeks to transform an
industry as old as vehicles themselves through innovation and networking (Cramer
and Krueger 2016). Such innovators, though often unwelcome, are usually not only
rewarded but often transform existing practices and rules of business organization
and industrial/service production. The emerging literature on this topic reminds
one of Thomas Kuhn’s vision of scientific progress, where normal science and
revolutionary science become separated by entirely different paradigms
It is increasingly apparent that some firms, cultures, nations, and cities are simply
more open and receptive to new ideas and practices than others (Hofstede 1997).
Furthermore, there are variations in the ability of a firm to take new ideas and
utilize them—in other words absorptive capacities vary across individuals, firms,
industries, and cultures (Cohen and Levinthal 1990;Spithovenaetal.2011). The
field needs a better understanding of the determinants of absorptive capacity at the
scales of both the individual and the firm (Schmidt 2005). Of course, this position
has been established in the metropolitan setting by Richard Florida who has long
maintained that the tolerance exhibited by the creative class is very important for
sustained urban growth. But mainstream economists now recognize the importance
of such factors as openness and tolerance and have begun to develop models where
agents actually choose between incremental and radical innovations depending upon
their innate attributes, including their age, level of human capital, and the attitudes
they hold toward risk (Acemoglu et al. 2015).
Radical innovations, which can be stymied by both pecuniary and non-pecuniary
means, are absolutely critical for continued economic success in our cities, and
these are the innovations that will increasingly differentiate the fortunes of our
metropolitan areas well into the future. Along these lines, Storper et al. (2015)
have argued that one of the reasons Greater San Francisco has outpaced Greater
Los Angeles in recent decades is because the private and public officials of the
Bay area were generally open to new ideas and practices, and were also able to
cooperate with one another in a constructive way. In other words, innovations
are fostered when openness occurs in both the firms and the wider institutions
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 41
(networks) of society. While Social Network Analysis (SNA) has been utilized by
social scientists (mainly sociologists) since the 1940s (Freeman 2011), it is only
more recently that geographers and regional scientists have started to use SNA
to enhance our understanding of innovation and regional economic development
(Fritsch and Kauffeld-Monz 2010;TerWalandBoschma2009). SNA holds
considerable potential to unlock significant insights into the role of networks in the
innovation process.
An obvious challenge to regional science is to develop models that shed light
on how the location-specific balance between normal and disruptive innovations
is reached. As in the handful of firm-level innovation models that already exist,
visible attributes like age and education of the workforce will clearly be important
drivers of both the generation and acceptance of radical innovations. These agent-
based models should also allow key personnel to move upwards in the same firm
or, alternatively, move sideways to a different firm in order to accomplish their
disruptive innovations. As a result, externalities will likely once again play an
important role in these sorts of urban innovation models. The overall risk-avoidance
attitude of the urban population will also be an important factor in this process.
Moreover, adopting a learning-by-doing approach, those cities that already enjoy
successful high-tech industries will be more open to radical innovations than others,
and this factor alone should introduce an element of circular and cumulative change
to the future geography of the most important innovations.
2.4.3 Transformation of Industries
One of the greatest challenges to regional science over the next 50 years will be to
understand the ever-changing nature of industries as they first appear, then mature,
and eventually become declining industries The traditional capitalist model, based
on the notion of the product cycle (Vernon 1966), is that new industries generally
arise only in those places with favorable incubation properties—high human capital,
support services and more—and then diffuse outward to other locations with
less costly labor as they become sunset industries and the production process
is rationalized. By the 1990s this model, originally envisaged to serve domestic
purposes, had been nicely adapted to explain the global dispersal of industry from
the core manufacturing regions of Anglo-America and Western Europe to the
world’s more peripheral developing regions. However, this stylized model has now
proved to be somewhat inadequate for several reasons.
Mainly, manufacturing has simply become less important—relatively speaking—
in advanced economies where more jobs are now available in trade and the services.
In fact, most top-20 lists of future occupations will mention a litany of service
jobs, particularly those that are health-related, across the various private and public
service industries (see for example United States Department of Labor 2015).
It seems, too, that industry numbers might rise or fall at certain locations but
occupation numbers remain a lot more stable.
neil.reid@utoledo.edu
42 G.F. Mulligan et al.
More theoretically, as captured in the thinking of Dixit-Stiglitz-type models,
consumers have shown a great preference for more and more variety in the different
goods and services that national economies now provide. In fact, this demand
for variety will likely drive the next industrial revolution in goods production as
identity-minded consumers strive to differentiate their personal products as much
as possible. There is evidence that many consumers are primarily interested in
goods and services that they consider authentic—with authenticity being defined in
terms of products that are made by small-scale independent specialist producers,
often using high quality inputs (Kovács et al. 2014;LewisandBridger2000).
The explosive growth of the craft beer market in the United States and elsewhere
exemplifies this trend (Moore et al. 2016). Twenty-one percent of beer sales in
the United States in 2015 were accounted for by the craft segment (Brewer’s
Association 2016). The rising popularity of craft beer and other craft products
creates a market opportunity that has also been explained in terms of resource
partitioning theory (Carroll et al. 2002)andhasledtowhatJovanovic(2001,
108) refers to as “variety proliferation”. A complex interplay of factors seems to
be driving the demand for these products including the demand for customized
goods and services, anti-mass production sentiments, and the status conferred by
being seen to be the owners/consumers of certain products (Carroll et al. 2002).
These changing market dynamics may mean challenging times ahead for some
large-scale producers. In the case of the brewing industry, large-scale brewers have
already been jolted into action: AB InBev’s $104 billion takeover of SABMiller is
partly a response to declining market share in the U.S. and other western markets.
Purchasing SABMiller gives AB InBev access to the former’s production and
distribution networks in future growth markets, such as Africa (Shadbolt 2015;
Forbes, 2015). Changing consumer pattern is partly driven by growing importance
of the millennial demographic. Described as “confident, self-expressive, liberal,
upbeat and open to change” (Pew Research Center 2010:1),Millennialsnumber
75.4 million and are the largest age cohort in the United States (Fry 2016).
Understanding the millennial consumer will be critical to future corporate success—
as consumers, millennials are highly dependent upon social media for information,
feel that there is too much power vested in large corporations, prefer to purchase
from companies that support solutions to specific social issues, and often seek the
opinions of peers (often via social media) before making a purchasing decision
(Barton et al. 2012; Fromm et al. 2011; Cone Communications 2013; Winograd
and Hais 2014).
Finally, nearly all goods production has become highly roundabout where vast
international supply chains—networks—are normally involved in the production of
final consumer goods. But, when communication and transportation allow just-in-
sequence assembly, the pervasive logic of the product cycle model will invariably
assert itself. However, the dispersal of new industries to the periphery might be
delayed for some time whenever skill-intensive, batch production in the center still
provides cost advantages over unskilled, Fordist production in the periphery. This
fact is evident in the fashion industry where the production of women’s clothing
is only now departing from Los Angeles which, in contrast with New York, chose
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 43
to specialize in low-quality, fast fashion clothing. In fact, New York’s alternative
choice on the ‘fashion divide’ has served to encourage strong jobber-contractor
relationships that in turn led to problem-solving cooperation among the different
agents in that industry. So, at the very least, regional science must find ways of
recognizing that different industries will mature or evolve in somewhat different
ways. Perhaps a typology for industry maturation is needed, where it is recognized
that location is critical to production but only at certain junctures in each industry’s
complex product cycle. Moreover, regional science must improve our understanding
of how successful, healthy regions are able to reinvent their export bases along the
lines already noted by Douglass North and Hans Blumenfeld some 60 years ago.
2.4.4 Modeling Interconnectivity Agglomeration
This chapter opened by presenting what is perhaps the most productive urban
agglomeration in the history of humanity—namely the American Megalopolis—
and the system of cities in which it is situated. Looking toward the future, regional
science must find ways of modeling technological agglomeration, or the manner
in which such dense urban agglomerations are bound together across the globe.
This is especially true in an economy that is dominated by creative industries in
metropolitan areas. The cross-sectional analysis developed earlier in the chapter
is very useful for providing snapshot insights into the nation’s metropolitan space
economy but more sophisticated longitudinal models are needed to uncover the
long term trajectories of both economic and population growth across those
hundreds of metropolitan areas— and how they are woven together nationally and
internationally.
Perhaps the simplest way to begin is to draw an analogy with the research on
regional production functions and hypothesize that productivity growth across cities
depends upon both patent production and business startups, along with controls
for various initial conditions. This, of course, raises various issues related to
measurement and specification (see below). In any case, a straightforward linear
regression model has been estimated whose reduced form is as follows;
PRODtDf.PRODt-1;PATN t-1;PROPt-1IMt-1/
where PROD is productivity (real GDP per capita), PATN is the volume of
patents, PROP is the number of proprietors operating single-person businesses,
and M is a vector of initial conditions, including metropolitan population, climate
attributes, and a Sunbelt versus Snowbelt distinction. The three key variables were
all standardized to make them scale-free: city productivity was changed into a ratio
using the national average for productivity (in the appropriate years) while both
patents and proprietors were converted into densities (by dividing each by its city
population) before being changed into ratiosusingtheappropriatenationaldensity
figures.
neil.reid@utoledo.edu
44 G.F. Mulligan et al.
Over the 12-year period, 2001–2013, proprietorship (0.885) has a coefficient
that is nearly twice that for patents (0.415), so there is a prima facie case for
startups being a much more important contributor to GDP growth than trademarks,
copyrights, and the like. However, when a system of three equations is solved for a
Carlino-and-Mills adjustment model, then over the very long run—as convergence
in the coefficients takes place—patents are seen to be about four times as important
as proprietorship. Seven of the nine coefficients in this 3 by 3 adjustment model
are significant and positive; however, with the lag of 12 years, proprietors exhibit
anegativeimpactonpatentgrowthandpatentsexhibitanegativeimpacton
proprietorship growth. So, as the direct and indirect effects unfold over the very
long run, it seems that the importance of patents eventually overshadows that of
startups in the metropolitan productivity model.
The estimation was repeated with shorter lags in order to assess the pattern of
coefficients before, during, and after the events of the Great Recession. During
2001–2005, patents and proprietors both had a positive effect on one another but,
during both 2005–2010 and 2010–2013, the two negative effects identified above
were once again prominent. So is it possible that productivity growth since the
Recession has taken on a different form where many metropolitan areas, largely
in the Snowbelt, are specializing in patents while others, mostly in the Sunbelt,
are specializing in business startups? Only more extensive research will answer
that question. In addition, the two direct effects on productivity growth that were
mentioned earlier were apparent in the first and third time periods where, if anything,
the effects of startups were even stronger in the later as opposed to the earlier years
(the ratio of coefficients climbed from approximately 2:1 to approximately 5:1).
However, during the second time period, proprietorship was negatively related to
productivity growth, probably because the recessionary events of 2007–2009 were
so much more pronounced in the cities of the Sunbelt states.
This preliminary analysis sheds light on several issues that will challenge
regional science over the upcoming years. To begin, analysts will have to decide
if the specification of this simple model is appropriate or not; in other studies it
has been found that entrepreneurs seem to occupy an intermediary position between
patent activity and the creation of jobs by incumbent firms (Tsvetkova 2015). In
the same way, some patent activity might translate directly into higher productivity
through incumbent firms, but other patent activity might require new proprietors
in order to see the light of day. And, of course, there is the standard observation
that all patents are not equal. Besides, the use of proprietors to represent startups in
the estimation is by itself problematic; other data sources will have to be mined to
get a more accurate picture of annual business startups in these metropolitan areas.
Finally, econometric experimentation should clarify what the appropriate time lags
are for these or other sorts of metropolitan productivity growth models.
The dimension of global connectivity can be modeled in the spirit of contagion
in financial markets (see, for example, Kolb 2011). Just as financial contagion
can be broken down within and across regions—whether regions defined within a
nation or among them—so, too, can innovation. Regional scientists (see Kelejian
et al. 2006;Hondroyiannisetal.2009)havealreadystartedmakinggooduse
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 45
of spatial theory and methods to study contagion and similar frameworks would
extend to innovation in a natural way. Ultimately, both research on and, indeed, the
practice of innovation must recognize that regions—whether established titans like
the megalopolis or emerging ones like those in China—can leverage advantages
derived from spatial agglomeration globally. This is perhaps the next large question
horizon for regional science in terms of the matter of this chapter: how can localized
spatial agglomeration be transferred across networked global space? Some thinkers
(Khanna 2016)seemtobelievethattheanswerstothisquestionhaveimportant
implications for the future of the global economy. If so, regional science is a field
uniquely positioned to help explore, understand, and predict what is yet to come.
2.5 Summary and Conclusion
This chapter set out several specific goals, in the context of the history of city
systems and advanced urban economies. First, it argued that key input, output,
and contextual attributes are combined locally to create each city’s innovation
ecosystem. These attributes not only differ from one place to another, but the manner
in which they are combined varies by location. The project then settled on 20
variables in order to discern how these innovation ecosystems currently vary across
the nation’s very largest cities. Next, using multivariate techniques, production
ecosystems were deconstructed in order to reveal the underlying dimensions of
innovation that are common to all of the nation’s 350-plus metropolitan areas. Once
the primary dimension of general innovation was identified for all metropolitan
areas, the other n #1lessimportantdimensionswereusedtorevealhowspecific
metros differentially combine their knowledgeable and educated workforces to
produce patents, engage in entrepreneurship, and create high value- added outputs.
As a consequence, the overall innovative index of any metropolitan economy could
then be estimated by first generating its score on each of the latent dimensions and
then, second, adding up those performance scores across all of the dimensions.
The results produced a clear set of innovation centers, spread evenly across the
United States. Not only do these findings square with contemporary theory on
agglomeration economies, as explained by Mulligan et al. (2012), they line up nicely
with older, less behaviorally motivated theories of central place hierarchies.
To conc lude, it is wo r th point i ng out severa l areas f or futur e researc h. First,
while satisfactory for present purposes—as is often the case—more and better data
are key to refining this project; ideally, the work could be replicated using micro
data on firms. Microdata would facilitate the development of behavioral models
of firm agglomeration, which could then be productively used to explore gaps in
innovation and performance both across and within metropolitan areas. Looking
within metropolitan areas would be especially interesting to better understand the
kind of mega-regions(see Nelson and Land 2011)identifiedintheintroductionand
that are increasingly relevant within the policy frame. Identifying the wage gradients
that motivate firms to decentralize (Brueckner 2011)wouldbeespeciallyusefulfor
neil.reid@utoledo.edu
46 G.F. Mulligan et al.
understanding the processes that give rise to the “fine structure” of contemporary
land-use patterns. Clearly, such evolution needs to take advantage of time series
data, and look at two or more points in time in order to chart the trajectory of
growth and change within regions as their fortunes rise and fall. Further, future
research would do well to trace the contours of the actual business world by looking
specifically at Fortune-500 companies, and their respective headquarter-cities—plus
by examining occupational (as opposed to employment) related data. Finally, such
work might also look at particular industrial outcomes, including patents and/or
other products of innovation.
Acknowledgements The authors thank Randy Jackson and Peter Schaeffer for their helpful
comments on a previous draft of this chapter.
References
Acemoglu D, Akcigit U, Celik MA (2015) Young, restless and creative: openness to disruption and
creative innovations. SSRN Electron J. doi:10.2139/ssrn.2392109
Acs ZJ, Braunerhjelm P, Audretsch DB, Carlsson B (2009) The knowledge spillover theory of
entrepreneurship. Small Bus Econ 32:15–30
Acs ZJ, Audretsch DB, Lehmann EE (2013) The knowledge spillover theory ofentrepreneurship.
Small Bus Econ 41:757–774
Andersson A (2009) Economics of creativity. In: Karlson C, Andersson A, Cheshire P, Stough R
(eds) New directions in regional economic development. Springer, New York
Angel S (2012) Planet of cities. Lincoln Institute of Land Policy, Cambridge, MA
Angel S, Parent J, Civco DL, Blei AM (2012) Atlas ofurbanexpansion.Lincoln Institute of Land
Policy, Cambridge, MA
Arthur WB (1989) Competing technologies, increasing returns, and lock-in by historical events.
Econ J 99:116–131
Audretsch DB, Keilbach M(2007)Thetheoryofknowledge spillover entrepreneurship. J Manag
Stud 44(7):1242–1254
Barton C, Fromm J, Egan C (2012) The millennial consumer: debunking stereotypes. The Boston
ConsultingGroup,Boston,MA
Batty M (2013) The new science of cities. MIT Press, Cambridge, MA
Beckmann MJ (1968) Location theory. Random House, New York
Behrens K, Robert-Nicoud F (2015) Agglomeration theory with heterogeneous agents. In:
Duranton G, Henderson JV, Strange W (eds) Handbook in regional and urban economics.
North-Holland, Amsterdam
Berry BJL, Parr JB (1988) Market centers and retail location. Prentice-Hall, Englewood Cliffs, NJ
Brewer’s Association (2016) Small and independent brewers continue to grow double digits—
Brewers Association. In: Brewers Association. https://www.brewersassociation.org/press-
releases/small-independent-brewers-continue-grow-double-digits/.Accessed1Aug2016
Brueckner JK (2011) Lectures on urban economics. MIT Press, Cambridge, MA
Cairncross F (2001) The death of distance: how the communications revolution is changing our
lives. Harvard Business Press, Brighton
Calzonetti FJ, Reid N (2013) The role of universities insupportingnewtechnologyindustries
through commercialization spin-off activities. Stud Reg Sci 43:7–23
Carlino GA, Chatterjee S, Hunt RM (2007) Urban density and the rate of invention. J Urban Econ
61:389–419
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 47
Carlino G, Kerr WR (2015) Agglomeration and innovation. In: Duranton G, Henderson JV, Strange
W(eds)Handbookofregionalandurbaneconomics.North-Holland,Amsterdam
Carroll GR, Dobrev SD, Swaminathan A (2002) Organizational processes ofresourcepartitioning.
Res Organ Behav 24:1–40
Carruthers JI, Hepp S, Knaap G-J, Renner RN (2012) The American way of land use: a spatial
hazard analysis of changes through time. Int Reg Sci Rev 35:267–302
Cheshire PC, Nathan M, Overman HG (2014) Urban economics and public policy: challenging
conventional wisdom. Edgar Elgar, Northhampton, MA
Christaller W (1966) Central places in southern Germany. Prentice-Hall, Englewood Cliffs, NJ
Cohen WM, Levinthal DA (1990) Absorptive capacity: a new perspective on learning and
innovation. Adm Sci Q 35:128
Combes PP, Gobillon L (2015) The empirics of agglomeration. In: Duranton G, Henderson JV,
Strange W (eds) Handbook of regional and urban economics. North-Holland, Amsterdam
Cone Communications (2013) 2013 cone communications social impact study: the next cause
evolution. Cone Communications, Boston, MA
Cramer J, Krueger AB (2016) Disruptive change in the taxi business: the case of Uber. Am Econ
Rev 106:177–182. doi:10.1257/aer.p20161002
Crispin JE (2010) The economic impact of startup companies and invention licensees originating
from research at the University of Utah. Utah Bus Econ Rev 70:1–7
David PA (1985) Clio and the economics of QWERTY. Am Econ Rev 75:332–337
de Blij HJ (2012) Why geography matters: more than ever. Oxford University Press, Oxford
de Rassenfosse G, Jaffe A, Webster E (2016) Low-quality patents in the eye of the beholder:
evidence from multiple examiners. NBER Working Paper. doi:10.3386/w22244
Duranton G, Puga D (2015) Urban land use. In: Duranton G, Henderson JV, Strange W (eds)
Handbook of regional and urban economics. North-Holland, Amsterdam
Etzkowitz H, Leydesdorff L (2000) The dynamics of innovation: from National Systems and
“Mode 2” to a Triple Helix of university–industry–government relations. Res Policy 29:109–
123
Florida R (2014) The creative class and economic development. Econ Dev Q 28:196–205
Florida R, Mellander C, Stolarick K,Ross A (2011) Cities, skills and wages.JEconGeogr12:355–
377
Forbes (2015) Forbes. In: Forbes. http://www.forbes.com/sites/greatspeculations/2015/12/15/how-
the-potential-ab-inbev-sabmiller-deal-focuses-on-africa/#1bebe5713a6e.Accessed1Aug2016
Freeman L (2011) The development of social network analysis—with an emphasis on recent
events. In: Scott J, Carrington P (eds) The Sage handbook of social network analysis. Sage,
Los Angeles, CA
Frey WH (2002) Three Americas: the rising significance of regions. J Am Plan Assoc 68:349–355
Fritsch M, Kauffeld-Monz M (2010) The impact of network structure on knowledge transfer: an
application of social network analysis in the context of regional innovation networks. Ann Reg
Sci 44:21–38
Fromm J, Lindell C, Decker L (2011) American millennials: deciphering the enigma generation.
Barkley, Kansas City, MO
Fry R (2016) Millennials overtake Baby Boomers as America’s largest generation. In: Pew
Research Center RSS. http://www.pewresearch.org/fact-tank/2016/04/25/millennials-overtake-
baby-boomers/.Accessed1Aug2016
Fujita M (1989) Urban economic theory: land use and city size. Cambridge University Press,
Cambridge
Fujita M, Krugman P, Venables A (1999) The spatial economy: cities, regions and international
trade. MIT Press, Cambridge, MA
Fujita M, François TJ (2013) Economics of agglomeration: cities, industrial location, and
globalization. Cambridge University Press, Cambridge
Gabaix X (1999) Zipf’s lawforcities:anexplanation. Q J Econ 114:739–767
Gabaix X, Ionniades YM (2004) The evolution of city size distributions. In: Henderson JV, Thiesse
JF (eds) Handbook of regional and urban economics. North-Holland, Amsterdam
neil.reid@utoledo.edu
48 G.F. Mulligan et al.
Gabe T, Florida R, Mellander C (2012) The creative class and the crisis. Camb J Reg Econ Soc
6:37–53
Glaeser E, Joshi-Ghani A (2015) The urban imperative toward competitive cities. Oxford
University Press, Oxford
Glaeser EL (2005) Reinventing Boston: 1630–2003. J Econ Geogr 5:119–153
Glaeser EL (2011) Triumph of the city: how our greatest invention makes us richer, smarter,
greener, healthier, and happier. Penguin Press, New York
Gunasekara C (2006) Reframing the role of universities in the development ofregionalinnovation
systems. J Technol Transfer 31:101–113
Hathaway I (2013) Tech starts: high-technology business formation and job creation in the
United States. Kauffman Foundation Research Series: Firm Foundation and Economic Growth.
www.kauffman.org. Accessed 1 Aug 2016
Head K, Mayer T (2004) The empirics of agglomeration and trade. In: Henderson JV, Thiesse JF
(eds) Handbook of regional and urban economics. North-Holland, Amsterdam
Hofstede GH (1997) Cultures and organizations: software ofthemind.McGraw-Hill,NewYork
Hondroyiannis G, Kelejian HH, Tavlas GS (2009) Spatial aspects of contagion among emerging
economies. Spat Econ Anal 4:191–211. doi:10.1080/17421770802625965
Ioannides YM, Overman HG (2003) Zipf’s law for cities: an empirical examination. Reg Sci Urban
Econ 33:127–137
Ioannides YM, Overman HG (2004) Spatial evolution of the US urban system. J Econ Geogr
4:131–156
Jovanovic B (2001) Fitness and age: review of Carroll and Hannan’s demography of corporations
and industries. J Econ Lit 39:105–119
Kelejian HH, Tavlas GS, Hondroyiannis G (2006) A spatial modellingapproachtocontagion
among emerging economies. Open Econ Rev 17(4):423–441
Khanna P (2016) Connectography: mapping the futureofglobalcivilization. Random House, New
York
Kolb RW (ed) (2011) Financial contagion: the viral threat to the wealth of nations. Wiley, New
York
Kovács B, Carroll GR, Lehman DW (2014) Authenticity and consumer value ratings: empirical
tests from the restaurant domain. Organ Sci 25:458–478
Krugman P (1991a) Increasing returns and economic geography. J Polit Econ 99:483–499
Krugman PR (1991b) Geography and trade. MIT Press, Cambridge, MA
Krugman P (1997) Development, geography, and economic theory. MIT Press, Cambridge, MA
Lautman M (2011) When the boomers fail. Logan Square Press, Albuquerque, NM
Lee SY, Florida R, Gates G (2010) Innovation, human capital, and creativity. Int Rev Public Admin
14:13–24
Lewis D, Bridger D (2000) The soul of the new consumer: authenticity—what we buy and why in
the new economy. N. Brealey Pub, London
Lösch A (1954) The economics of location. Yale University Press, New Haven
Moore MS, Reid N, McLaughlin R (2016) The location determinants of microbreweries and
brewpubs in the United States. In: Cabras I, Higgins D, Preece D (eds) Brewing, beer and
pubs: a global perspective. Palgrave McMillan, London, pp 182–204
Moretti E (2004) Human capital externalities. In: Henderson JV, Thiesse JF (eds) Handbook of
regional and urban economics. North-Holland, Amsterdam
Moretti E (2012) The new geography of jobs. Houghton Mifflin Harcourt, New York
Mulligan GF, Partridge MD, Carruthers JI (2012) Central place theory and its reemergence in
regional science. Ann Reg Sci 48:405–431
Nelson AC, Land RE (2011) Megapolitan America: a new vision for understanding America’s
metropolitan geography. APA Press, Chicago, IL
Peirce NR, Johnson CW, Peters F (2008) Century of the city: no time to lose. Rockefeller
Foundation, New York
Pew Research Center (2010) Millennials: a potrait of generation next. Pew Research Center,
Was h ington, DC
neil.reid@utoledo.edu
2ExploringInnovationGapsintheAmericanSpaceEconomy 49
Pred A (1977) City systems in advanced economies. Halsted Press, New York
Reese LA, Sands G, Skidmore M (2014) Memo from motown: is austerity here to stay? Camb J
Reg Econ Soc 7:99–118
Renner RN, Carruthers J, Knaap GJ (2009) A note on data preparation procedures for a nationwide
analysis of urban form and settlement patterns. Cityscape J Policy Dev Res 11:127–136
Schmidt T (2005) What determines absorptive capacity. In: DRUID tenth anniversary summer
conference 2005—dynamics of industry and innovation, Denmark
Shadbolt P (2015) Brewerybattle:ABInBevandthecraftbeerchallenge.BBCNews,13Oct.
http://www.bbc.com/news/business-34383721
Shapiro JM (2006) Smart cities: quality of life, productivity, and the growth effects of human
capital. Rev Econ Stat 88:324–335
Simonton D (1997) Creative productivity: a predictive and explanatory model of career trajectories
and landmarks. Psychol Rev 104:66–89
Simonton D (1999) Origins of genius. Oxford University Press, Oxford
Spithoven A, Clarysse B, Knockaert M (2011) Building absorptive capacity to organise inbound
open innovation in traditional industries. Technovation 31(1):10–21
Storper M, Kemeny T, Makarem NP, Osman T (2015) The rise and fall of urban economies.
Stanford University Press, Stanford, CA
Ter Wa l AL, Boschma RA (20 09) Apply ing soc i al network analy sis in econ omic geo graph y:
framing some key analytic issues. Ann Reg Sci 43(3):739–756
Tsvetkova A (2015) Innovation, entrepreneurship, and metropolitan economic performance:
empirical test of recent theoretical propositions. Econ Dev Q 29:299–316
United States Department of Labor (2015) Fastest growing occupations. Occupational outlook
handbook, 17 Dec. https://www.bls.gov/ooh/fastest-growing.htm
United States Patent and Trade Office (2015) Calendar year patent statistics: listing of all United
States metropolitan and micropolitanareas,totalutility counts, 2000–2013. www.uspto.gov.
Accessed 1 Aug 2016
Vern o n R (1 9 66) I n te r nat i ona l i nv e s tm e nt a n d i n te r n at i ona l t ra d e i n t he pr o duc t c y c le . Q uar J E con
80:190–2017
Winograd M, Hais M (2014) How millennials could upend wall street and corporate America.
Governance studies at brookings, Feb. https://www.brookings.edu/wp-content/uploads/2016/
06/Brookings_Winogradfinal.pdf
Zenou Y (2009) Urban labor economics. Cambridge University Press, New York
Zipf GK (1949) Human behavior and the principle of least effort. Addison-Wesley, Cambridge
Gordon F. Mulligan is professor emeritus, School of Geography and Development, University
of Arizona. His primary research interests are in economic geography, with special interests in
urbanization, housing, location theory, regional development, and spatial behavior. He is past book
review editor of the Annals of Regional Science and past co-editor of the Journal of Regional
Science, and has been an RSAI Fellow since 2011. Previous faculty positions were atthe University
of Washington, Queen’s University, and Flinders University. Dr. Mulligan earned the Ph.D. in
geography from the University of British Columbia in 1976.
Neil Reid is Professor of Geography and Planning and Director of the Jack Ford Urban Affairs
Center at the University of Toledo. His primary research interests are economic geography with
special interests in industrial location and local economic development. He serves as the Executive
Director of the North American Regional Science Council, Editor for the Americas for Regional
Science Policy and Practice, and Book Review Editor for Economic Development Quarterly. He
has held the position of Visiting Professor in the European Faculty of Engineering at Cz˛estochowa
University of Technology in Poland where he taught courses in industrial location and regional
development. He holds a Ph.D. in Geography from Arizona State University.
neil.reid@utoledo.edu
50 G.F. Mulligan et al.
John I. Carruthers is the Director of the Sustainable Urban Planning Program at the George
Was hington Universi t y. H i s areas of r esearch includ e urban and r egiona l e c onomics; sus t a inable
urban development; environmental quality of life; economic geography; geospatial and economet-
ric analysis. He holds an adjunct faculty appointment in the Department of Urban Planning and
Engineering at Yonsei University, in Korea. Dr. Carruthers has served as an elected member of
the RSAI Council (2009–2011) and the North American Regional Science Council (2011–2014)
and is in the Board of the Western Regional Science Association. He earned is MS and doctorate,
respectively, at the University of Arizona and the University of Washington. Regional Science is
his home field.
Matthew R. Lehnert is a PhD student in Spatially Integrated Social Science (SISS) at the
University of Toledo. His primary research interests are in international development, with special
interests in the spatial dimensions of human development, poverty, and education in Morocco.
He has served in adjunct positions at Owens Community College and Al Akhawayn University.
Currently he holds a research assistantship with the North American Regional Science Council.
neil.reid@utoledo.edu