![[Case Western Reserve University -- Toolbar]](/pix/lowpro.gif)
This paper reports on the findings of an REI study, funded by the Cleveland and Gund foundations, which took place between spring 1997 and April 1998. Its stated goal was "to develop a deeper understanding of commercialization in this region and to provide a framework for commercialization decisions that will maximize economic benefits for Northeast Ohio." This has required an ambitious effort, first, to take stock of the many components that constitute the region's commercialization system (a form of S&T report card) and, second, to make explicit the connections or paths from research to economic impact on the region.
The report is a work in progress. It builds on ten years of work by the Center to deepen our understanding of these important issues. (See Exhibit A) The next phase should use the study and its findings to shape conversations with the community's various S&T partners -- in effect, as an opportunity to define the capabilities and interactions that are most important for achieving S&T policy's primary goal: to increase the productivity of the region's industries and so to raise the standard of living of the people who live and work here.
This report incorporates the work of several contributors: Amit Sinha with Michael Fogarty analyzed technology strengths; Dilek Karaomerlioglu developed the data on the region's high-technology industries, including links between R&D labs and local production; Meg Lister Fernando and Carolyn Schnipp prepared data on university research and degrees; and Larry Goodpaster assessed the literature on science and technology intermediaries and commercialization.
We would like to thank Dorothy Baunach and Richard A. Shatten for many helpful comments and suggestions throughout the study and preparation of the report. Thanks also to Mark Overs for providing us with a picture of his beautiful sculpture, Ascending Wave 2, for use on the cover of the report.
I. INTRODUCTION: WHY INCREASE INVESTMENT IN EDUCATION, SCIENCE, AND TECHNOLOGY
II. INVEST WITH A FRAMEWORK!
III. DOES CLEVELAND HAVE AN EMERGING HIGH-TECH ECONOMY?
IV. RAISING THE CLEVELAND REGION'S EDUCATION LEVEL
V. THE CLEVELAND REGION'S UNIVERSITY RESEARCH
VI. NORTHEAST OHIO'S TECHNOLOGY STRENGTHS
VII. INTERMEDIARIES -- GETTING THE MOST FROM THE REGION'S TECHNOLOGY STRENGTHS
VIII. INVEST IN BUILDING CAPABILITIES AND PATHS FROM R&D TO INDUSTRY
IX. EXHIBITS
Note: Figures that accompany this report are not yet available.
Figure 1: Per capita income declined then stagnated, while jobs have grown steadily since 1983.
Figure 2: The region has turned around but remains disadvantaged.
Figure 3: Most of the region's relative per capita income gain appears to be due to labor force participation.
Figure 4: An economic framework for investing in the regions industries.
Figure 5: Ohio's overall productivity is low (Real Gross State Product per worker shown as a percent of the US)
Figure 6: Education, Productivity, Per Capita Income
Figure 7: S&T investments should build strong capabilities and interactions supporting the region's industries.
Figure 8: Commercialization links between the R&D base and NEO industries
Figure 9: A disproportionate share of the region's "anchor" industries are losing national share.
Figure 10: The region's economy is low-tech -- nearly 10% less high-tech than the median metropolitan region
Figure 11: Status of the Cleveland-Akron high-tech industry, 1990-1995
Figure 12: The Cleveland-Akron region's college-educated population is 6% below the median of 35 metropolitan regions
Figure 13: The Cleveland-Akron region's scale of degree programs is slightly above the median
Figure 14: Bachelors degrees, science and engineering disciplines, per capita awarded by NEO colleges and universities, by discipline, 1985-1994
Figure 15: Doctorate degrees, science and engineering disciplines, per capita awarded by NEO colleges and universities, by discipline, 1985-1994
Figure 16: Mix matters -- the scale of science and engineering bachelor's degrees in 1994
Figure 17: Mix matters -- the scale of science and engineering doctoral degrees degrees in 1994
Figure 18: The region's R&D organizations perform about $2.3 billion in R&D
Figure 19: CWRU performs about 3/4 of the region's university research
Figure 20: The region's scale of university research is low but growing relative to other regions
Figure 21: Selected metro areas -- 1994 funding indexes
Figure 22: Science and engineering federally funded research at NEO colleges and universities, by discipline, 1985-1994
Figure 23: The region's research institutions are competitive only with NIH
Figure 24: Overall S&T investment framework
Figure 25: The region's top corporate R&D labs have been shifting R&D to other locations
Figure 26: R&D to technology clusters
Figure 27: The Cleveland-Akron region is part of a worldwide innovation system
Figure 28: REI's analysis confirms the region's historic strengths in polymers and advanced materials
Figure 29: The good news is that 80% of NEO's new technology is associated with local production
Figure 30: The state and the Cleveland-Akron region have a long history of intervention
Figure 31: Tension in university -- industry: three case studies
Figure 32: Maximizing economic benefits requires new incentives and structure to encourage cooperation and a regional focus
Figure 33: Northeast Ohio's S&T system
Table 1: Status of Cleveland-Akron Regions High Technology Industries
Table 2: Ranking metropolitan regions
Table 3: Nearby and other industrial metropolitan regions
Table 4: NEO ranks in the top five in Advanced Materials
Table 5: Cleveland is ranked in the lowest quartile based on the percent of manufacturing workers with at least one year of college
Table 6: Number of degrees granted at NEO four year colleges and Universities, 1994
Table 7: Northeast Ohio schools, research and development expenditures in 1994
Table 8: Top 15 Northeast Ohio patenting organizations, 1990-1995
The broadest purpose of this study is to develop the necessary data and a new S&T investment framework for the Cleveland region. A new framework is essential for increasing commercialization of the region's estimated $2.3 billion R&D in ways that maximize its local economic benefits. Our hope is that this study will help shape a dialogue concerning S&T investments, a commercialization strategy, and the region's S&T intermediaries.
This report begins in Section II by presenting two frameworks (a productivity and S&T investment framework) for making economic development the goal and commercialization the tool. Then Section III asks, does Cleveland have an emerging high-tech economy? Section IV analyzes NEO's education level and the higher education degrees awarded by the region's colleges and universities essential to closing the region's education gap. The strengths and weaknesses of Northeast Ohio's higher education research are documented in Section V. Section VI analyzes the region's technology strengths using a unique patent database and new methodology. The section focuses on polymers, advanced materials, auto, aerospace, biomedical devices and biopharmaceutical technologies. This section also connects the region's top R&D labs with local production by industry. Section VII examines NEO's S&T intermediaries in light of new challenges. The final section, Section VIII, pulls together the report's main findings and examines implications for the region's commercialization strategy.
The Cleveland region is poised at a crossroads. There is clear evidence of a turnaround across a wide spectrum of economic conditions. (footnote 1) By several measures, the region's economy looks vigorous: It has added nearly 200,000 jobs over the past decade (a 15% gain), and its unemployment rate has dropped from about 7.5% in 1986 to 4.8% in April 19973/4 equal to the current US rate. (footnote 2) Yet the data presented in the following chart present a puzzle: If Northeast Ohio's economy has created so many new jobs, why has per capita income declined relative to other metropolitan regions since 1969?
The contrast between job growth and the decline in relative per capita income is striking. For the majority of economic development observers (civic leaders, the media, politicians, and citizens), the typical response to this picture is bewilderment. How could it be true? There is one quick, common response: Northeast Ohio is creating a lot of low-wage jobs. However, this response begs the question -- Why should NEO be producing a disproportionate share of low wage jobs? The answer is very unsatisfactory since other regions are experiencing job growth that is simultaneous with growth in per capita income.
The divergent trends, shown in Figure 1, raise another key question: Which trend should be used to assess Cleveland's turnaround? If success equals jobs, clearly we would pick the dramatically favorable employment trend. However, if successful economic development is better judged by gains in the standard of living, then we're forced to choose the per capita income trend. Amazing as it might seem, given the propensity for policy makers and media to focus attention on jobs, the vast majority of economic development specialists single out per capita income as the key development measure. (Not widely known, careful research shows that 60-90% of new jobs go to in-migrants rather than existing residents. (footnote 3) )
Why does it matter which measure we choose to reflect progress? One reason is straightforward: The factors driving the two trends differ. In other words, conditions that produce more jobs needn't improve the region's standard of living (in essence, more people are working but, on average, we're essentially no better off). In contrast, regions experiencing strong per capita income growth also create the jobs. They get both.
Per capita income in real terms (adjusted for inflation) cannot increase over the long run without productivity growth. (footnote 4) Therefore, the fundamental answer to the puzzle posed by the chart is that the productivity performance of the region's industries is low relative to other US regions. Unfortunately, NEO's long-term per capita income can never catch up to regions whose performance we target as a gauge of our own success unless the productivity growth of Northeast Ohio's industries exceeds that of the other regions -- i.e., our industries must outperform theirs.
Unfortunately, productivity data for metropolitan regions are not available, so it's not fully possible to confirm this interpretation specifically for the region. (footnote 5) This report argues that Northeast Ohio's underlying trends are consistent with the state's basic trends and our interpretation. The grounds for this interpretation are threefold: (1) economic theory -- there exists a very substantial literature on economic growth and, specifically, sources of productivity growth that points to productivity as the primary explanation for NEO's long-run per capita income trend (footnote 6); (2) It is reasonable to infer this region's productivity trend from Ohio's, which peaked in the early 1980s at roughly 98% of the US level, declined to 94% by 1990, and has stagnated at that position since then; (footnote 7) and (3) data showing Northeast Ohio's low education level and weak technology base, presented later in the report, are consistent with the productivity interpretation.
Given these supporting theory and data, placed alongside the jobs-per capita income chart shown above, it is prudent for Cleveland to commit to a productivity strategy for the next phase of the turnaround.
The Cleveland region has moved into a new phase. Broadly speaking, the world of regions can be split into two parts -- one in which markets favor a region (a net in-flow of investment and skilled workers) and a second in which the markets work against a region (a net outflow of investment and skilled workers). The two worlds are illustrated in Figure 2.
Remarkably, reflecting the good favor of markets, NEO experienced 130 years of continuous economic growth from 1840 into the early 1970s. But competitive problems with the region's industries, probably beginning as early as the 1950s, began moving Cleveland into a decline phase. At some point, Cleveland crossed the boundary, entering a period in which markets cause disinvestment.
It is this condition that produced Cleveland's crisis in the late 1970s. At that time Cleveland took on the market with sustained efforts over fifteen years. (footnote 8) The evidence from REI's Civic Vision 2000 and Beyond study, coupled with related analyses, strongly suggest the region has turned the corner. This is especially evident since 1990. However, the analysis of NEO's per capita income trend data tells us that Cleveland is still positioned in the unfavorable market range.
This assessment has two broad implications. First, without public investment, gains over the past fifteen years could quickly be reversed. Second, moving from the left side of the chart to the right side (A to B) will require a productivity strategy. One reason is that knowledge has become the key factor explaining regional fates. The region has turned around but remains disadvantaged because of its low education level and weak technology base.
A host of factors determine the long run performance of a region's industries. Our community investments affect the performance of the region's industries through several channels, including physical infrastructure, entrepreneurship, amenities, and inner city conditions. Each of these factors is important and should become an element of a broader strategy for boosting regional productivity. This study is limited to the top two factors -- education and science and technology.
Numerous economic studies have accumulated strong evidence that the lion's share of productivity growth is caused by improvements in two factors--education (which includes skills) and technology. (footnote 9) This conclusion presents the Cleveland region with a big challenge. First, for two decades, the region's education level (measured as the percent of persons over 25 with at least one year of college) has declined relative to the median of a benchmark set of 37 metropolitan regions. Not surprisingly, the trend in per capita income parallels that of education. Second, although no comparable measure for technology exists, measures using industry R&D expenditures and investment in equipment and facilities by Ohio manufacturers suggest that the region's technology base is relatively weak. (footnote 10) (footnote 11)
Technology, the second essential cause of productivity growth, comes from various sources, including universities, government R&D labs, research hospitals, and industry R&D labs. Higher education affects the technology available for the state's industries both directly by creating new technology -- and indirectly -- by providing an educated workforce that speeds up diffusion of new technology and adoption of existing technology. Technology eventually becomes embodied in our industries through private investment in new equipment and facilities. When evaluating a policy's expected economic development impact, we cannot take this step for granted. (footnote 12)
Higher education is one obvious component of a productivity strategy. Its economic importance is emphasized by The Ohio Board of Regent's estimate that each percentage point gain in a state's education level adds an estimated $590 income per person annually. (footnote 13) In Ohio, a one percentage point gain in the state's education level would amount to an estimated $6.5 billion of new income each year. The Cleveland region's estimated share would amount to $1.7 billion annually.
Such a large gain won't occur without a plan. To raise productivity, it's not enough simply to produce more graduates. The $1.7 billion annual increase would reflect the extra productivity derived from higher skills incorporated into NEO's industries.
Providing good answers to these questions will require fresh thinking and a new investment framework.
The first look at the jobs-per capita income chart suggested a recent turnaround in per capita income. Could this reflect the emergence of a productivity turnaround? It doesn't appear so. If true, Cleveland could rest fairly comfortably with some assurance that it's got in place the policies and programs necessary to improve the region's standard of living. Not only that, the payoff would be large. Seemingly small movements in relative position are worth a lot: For Cleveland, the relative gain in per capita income since 1985 added $11 billion in real terms (adjusted for inflation). However, these increases are part of a larger phenomenon affecting the Great Lakes region as a whole. For example, Pittsburgh's income gain over the same period was triple or quadruple Cleveland's -- $30-40 billion since 1985.
Digging a bit more deeply, most of the recent trend is caused by increased labor force participation -- not a better educated workforce and improved technology (i.e., productivity). In other words, the 1990s per capita income turnaround stems from a higher percentage of people working and not from greater output per worker. This is dramatically confirmed in Figure 3, which shows decline in earnings per worker coupled with a significant increase in labor force participation. Historically well below the national average, the fraction of Clevelanders working (the labor force participation rate) has been rising relative to the US. The recent labor force participation trend has been stimulated by growth in the relative importance of services industry jobs (manufacturing jobs have been declining as a percent of total employment). Therefore, the per capita income turnaround combines the two effects: earnings per worker and labor force participation (both are relative to US metropolitan regions). The trend in relative earnings per worker more closely corresponds to the trend in relative productivity growth by region. See Figure 3.
While many would view increased labor force participation as favorable, since it increases income, this trend cannot substitute for education and technology as a source of long-term gains in productivity and per capita income. In fact, as NEO's labor force participation rate approaches the nation's (i.e., Cleveland becomes more like the nation), this factor will lose influence.
This section presents two essential frameworks. The first is the productivity framework while the second is a framework for guiding S&T investments. We must have both a better framework and better data for guiding S&T investments for three reasons. First, NEO is behind, the market continues to work against the region, and economic development resources are scarce. Consequently, it's essential to make each dollar count. Second, economic growth involves complex processes. For example, connections between higher education investments and Northeast Ohio's economy are oftentimes indirect. Moreover, the important impacts of investments are frequently subtle and generally take effect only after a significant time lag, often making it easier to focus too much attention on more easily measured, short-term gains. In addition, the mechanisms linking S&T investments to economic impact are typically poorly understood. Understanding these mechanisms is crucial for making informed investments. Third, investments in science and technology occur within an innovation system, whether our policies and programs reflect knowledge of the system or not. To maximize the economic payoff from S&T investments intended to commercialize the region's $2.3 billion R&D, knowledge of the S&T system is essential.
One consequence of not having an agreed upon framework is that it's easy to pose basic questions concerning our current investments that cannot now be satisfactorily answered. Consider a few: Under what conditions does it make economic sense for Ohio and Cleveland to invest in MEMS? What would be a good economic rationale for building stronger chemistry research and graduate programs in NEO? How would we know if NEO's system of S&T intermediaries face the right incentives for commercialization that produce stronger NEO industries?
Figure 4 provides the essentials of an economic framework. (footnote 15) It suggests that investments in education and technology dominate the factors that affect productivity, thereby changing the region's standard of living. (footnote 16) Education and technology raise the standard of living primarily by increasing productivity, which boosts the competitiveness of the region's industries. For example, in industries that are closely connected through buying and selling relationships, or through reliance on the same technology, the performance of the entire industry cluster is improved. (footnote 17) Industry clusters present an opportunity to magnify the economic impact of investments by being strategic -- that is, by developing and using knowledge of commercialization and industry clusters to focus investments on the right industries (those that can be expected to produce the greatest long-run gains in per capita income). A large part of the rationale for investing in a new industry cluster (such as is the case for Cleveland's biomedical industry) is the opportunity to create sufficient critical mass that the economic impact is magnified by the interactions of the cluster.
The figure suggests that the final link in the chain of effects occurs when industries interact with people and communities, changing the number and quality of jobs, wages and salaries, poverty levels, and quality of life (health, safety, neighborhoods, etc). As mentioned earlier, the outcome for Northeast Ohio has been mixed: The region has added almost 200,000 jobs but, at the same time, its per capita income has declined relative to the US metropolitan areas. This trend should sound an alarm, warning the region to make productivity a primary focus of its policy.
At the national level, productivity growth is measured as the observed output growth -- such as growth in real Gross Domestic Product -- minus growth contributed by added inputs. By one estimate, advances in knowledge explain perhaps 60% of productivity growth. (footnote 18) In the real world, an economy's output is not easily measured. Furthermore, a firm or industry employs multiple inputs in making just about everything. Thus, sorting out the sources of growth is not a straightforward task. To complicate matters, many important inputs are provided publicly. These include roads and bridges, safety, air quality, education, and technology infrastructure. (footnote 19) (footnote 20) (footnote 21) Often, the best available productivity measure is simply output per worker, which at the national level would be GDP per worker and GSP per worker at the state level.
Productivity is one crucial missing piece of data on metropolitan regions. There are no metropolitan-level productivity data of a quality comparable to national data. However, it is reasonable to infer the Cleveland region's productivity trend from data based on Ohio's real Gross State Product (GSP). (footnote 22) See Figure 5. A crude measure of state-level productivity is simply GSP per worker. To evaluate Ohio's overall productivity performance, we can track GSP per worker relative to all states. The figure above shows GSP per worker for Ohio and three other states relative to the US from 1982 through 1994. For example, Ohio's index of roughly 95 in 1982 means that the state's productivity was 95% of the US average. Movements up and down are due to higher or lower rates of productivity growth.
The two most prominent growth states are California and Massachusetts. Reflecting the status of the US auto industry, Michigan's productivity level declined for eight years, then increased in the 1990s. Ohio's productivity level fell between 1984 and 1990; later, it seems to have stagnated at a very unsatisfactory position about 6% below the US average.
It is essential to think of education and technology jointly. Education's effect on productivity involves interaction with technical progress as well as with investment in new equipment and facilities. (footnote 23) (footnote 24) One example that also involves a region's industry clusters is education's effect on the speed of technology diffusion within an industry, state or region as well as between different regions.
One critical implication of this interdependence of education, technology, and industry clusters is that decisions about a region or industry's education level require public policy. The reason is that an individual's decision to invest in education does not take into account the full effect of the investment -- in this case, the rate at which technologies are diffused throughout an industry and region. The individual's decision is driven by an expected personal return. The potential social payoff from the interaction of education and technology, however, is likely to continue growing over time as long as knowledge is the key industry input.
Cleveland can increase its return on investment by strategically taking advantage of interactions between education and technology. If not, the result will be a regional workforce that is even less educated than the present one. The economic effects would likely be more severe for places like Northeast Ohio, because the incentives for investing in human capital will depend on the extent of job and earnings opportunities in the region. (footnote 25) The reason is that NEO's less dynamic economy gives an individual fewer incentives to undertake the costs of higher education, since the expected return would be lower. (footnote 26) (footnote 27)
One crucial element in evaluating the economic impact of S&T investments is that the effects take time. Because timing matters a lot, a long-run perspective is essential. For example, estimates of time lags between basic research and economic impact range from five to fifteen years. (footnote 28) The time frame depends on the mechanism for turning research into productivity. If the effect occurs when graduates take jobs with local companies, then it will take the six or seven years required to earn a science or engineering doctorate. If the effect occurs through faculty consulting with local firms, the lag may be considerably less. In the latter case, the timing hinges on how long it takes for the research to reach a point where commercialization opportunities exist. If a university technology is transferred through patent licensing to a NEO company, it may take longer and require additional industry R&D to commercialize the technology. (footnote 29) (The relationship between university research and the state's technology is discussed in section VI.)
Figure 6 uses Ohio data to summarize the productivity framework. The state's growing education gap, coupled with a relatively weak technology base (not pictured), largely explains the state's poor industry productivity performance and relative decline in per capita income. (footnote 30) It also suggests the potential payoff from S&T investments with a clear connection to productivity performance of the state's industries.
The potential payoff from focusing on productivity is huge. Relatively small absolute gains in productivity growth rates, after compounding, translate into massive differences in a state's income over time. For example, if Ohio's real (inflation-adjusted) rate of growth in GSP had increased at the national rate (2.2%) in 1979-92, the state's 1992 GSP would have been $31 billion larger than it was. If Cleveland got its share, roughly one-third (or about $10 billion) would have been added to NEO's economy. This state's lost income is nearly half the size of the Cleveland region's economy today or 13% of Ohio's overall economy. This sum could easily pay for the state's higher education budget several times over.
The first framework identifies the sources of regional productivity, identifying education and technology as the top two. Next we need a framework for investing in science and technology to increase commercialization and raise the odds of capturing local economic benefits.
From a broad perspective, the region shouldn't care whether technology comes from within the region or elsewhere, as long as it gets incorporated into local industries. What we want is the best technology for the region's industries, whatever its source. However, given Cleveland's disadvantaged position, what's necessary is a strategy for gaining from both internal and external sources.
Figure 7, shown below, describes the major components of the regional S&T system in a simplified form. The first point to note is that the interactions among the four parts shape the development and commercialization of new technology as well as the geography of economic benefits. We can characterize four S&T capabilities and the paths that connect them. In effect, the top two quadrants are housed in universities, while the bottom two belong to industry. (To keep the picture simple, research hospitals and government R&D labs are excluded. However, both are important sources of new technology.)
Although the sharp divisions shown in the chart between traditional basic research and interactive research don't really exist, it's helpful to draw this distinction for several reasons. (footnote 31) (For example, the chemical engineering department at CWRU does both traditional basic research and interactive research directly related to commercialization.) First, the quality of applied research leading to significant commercialization opportunities ultimately depends on the quality of basic research. Given the region's interest in commercialization, it's essential to protect and nurture the region's basic research capabilities.
Second, sources of funding for each quadrant differ. Most direct funding for traditional basic research comes from the federal government. Particularly in the case of public institutions, states fund basic research by paying for buildings, equipment, chaired professorships, and graduate programs. (Funding of university research is discussed in Section VI below.)
Third, most of the region's attention is drawn to the applied areas of research because the products are more visible and of course closer to having commercial value. It's in this quadrant that we find university-industry research centers (UIRCs), industry funding of university research, technology transfer efforts, and patents. From the outsider's perspective, this is where the action happens.
The bottom two quadrants belong to industry -- one is industry R&D (a $2 billion NEO component) and the second is industry production, which is the ultimate destination for technology. If these did not exist, of course any technology flowing from the university's two quadrants are exported to other regions. It is essential to keep in mind that local productivity gains from S&T investments require the technology to be incorporated into local production. Simply doing the R&D will have a modest effect on regional income but won't significantly raise the region's per capita income.
Seeing these components as part of a highly interactive system is crucial for policy -- a systems approach is essential for maximizing economic benefits. Also, the framework yields several critical insights about why having knowledge of the system matters. For example:
a. The second quadrant could produce competitive technology but its effects are short-lived because the first quadrant is weak and eventually quadrant two (applied research) dries up as an important source of innovation.
b. The first quadrant could produce nationally competitive basic research but, because quadrant two is weak, basic research findings produce little of potential commercial value or the results of the basic research are used by institutions in other regions. This case is not as bad as the first if graduates take positions in the region and faculty consult with local industry.
c. The two university quadrants could produce excellent research but the technology is exported because (1) there's no industry R&D network to develop university-based technology, or the R&D network is weak, and/ or (2) there's no industry destination for the technology. Even if local production exists, it's possible that the technology will be exported to the firms' other facilities located in other regions.
Regarding universities, the region's current situation indicates that interventions are piecemeal. They don't reflect sufficient knowledge of the system linking universities and industry. For example, local policy makers commonly focus on the tip of the iceberg -- the most obvious university technologies that emerge, such as FES, PDLC, or MEMS (i.e., the new products and processes, or a startup company). In reality, there exist multiple interactions between universities and industry that provide pathways for universities' influence on industry performance as well as industries' influence on university education and research. See Figure 8.
The usual perspective misses a much larger set of university interactions that affect industry productivity: graduates taking jobs, faculty consulting, and industry scientists and engineers learning from published academic papers. (A recent study of citations to academic papers on corporate patents shows increasingly close links between basic science and corporate patenting. (footnote 32) )
One clear implication of the framework is that our policies and programs operate in an open system. Knowledge derived from R&D and education can't be tied to one location. In fact, the better the research, technology, and graduates the harder it is to capture the economic benefits from S&T investments. The challenge is to produce the best and work very hard to capture the benefits.
These knowledge flows move in both directions -- in and out of the region. For example, given the scale of NEO's auto industry relative to the R&D base producing automotive technologies, a large chunk of new technology that supports this industry comes from R&D performed elsewhere. This view suggests that one key to a region's development is the strength of its internal capabilities and networks as well as its external links. A competitive S&T system means having both strong internal capabilities and paths as well as strong external networks. Ideally, the region develops and utilizes technology originating locally and draws on the world's best sources of technology for local industry.
In the next several sections, we develop a report card on the region, examining each S&T system component: NEO's high-tech industries, the university research base, education and higher education degrees, and the region's technology strengths. The last two sections give a summary status of each component, discuss S&T intermediaries, and makes broad policy recommendations.
Figure 9, shown below, presents an overview of Northeast Ohio's manufacturing industries' performance from 1990 through 1995. "Anchor" industries are three-digit industries that are highly concentrated in the region (in this case, we classify industries as "anchor" if the region's employment in an industry exceeds the region's share of all manufacturing employment). For example, if the region had 2% of employment in the US motor vehicle industry but only 1% of US manufacturing jobs, its index of concentration is 2/1 or 2.0. Consequently, it would be classified as an anchor industry.
Industries can be concentrated or not concentrated, and they can be gaining or losing US share. It's useful to think of change in a local industry's share of the industry nationally as gauge of competitiveness. If NEO is gaining national share of the particular three-digit industry, then the industry becomes more concentrated in the region. In the best of all worlds, NEO would have a large portion of its manufacturing economy in the "gaining share" quadrants -- both concentrated and not concentrated. We can think of industries positioned in the not concentrated but gaining share group as potential "emerging" industries.
The findings are mixed. Overall, nearly three-fifths of the region's manufacturing employment is in industries that are losing national share. The bulk of these (42% of total NEO manufacturing employment) are highly familiar "anchor" industries. These include industries such as metalworking, forging, motor vehicles, measuring & controlling devices, and chemicals. However, a subset of the region's "anchor" industries is gaining national share. The strongest performers are Soaps & Cleaners, Paints & Allied Products, and Miscellaneous Chemical Products. In addition, although comprising a relatively small segment of the region's economy (8%), some are potential emerging industries. These include Plastics Materials & Synthetics, Guided Missiles and Space Vehicles, and Miscellaneous Electrical Equipment. Individual high-tech industries are examined later.
It is appropriate to single out high-tech industries for two reasons: first, these industries generally exhibit a high rate of innovation and productivity, are more competitive in international trade, and pay higher wages; second, high-tech industries are a vehicle for influencing and supporting many industry clusters. For example, the electronic components industry is high-tech but it is also a member of a much larger auto industry cluster, parts of which are typically not classified as high-tech. In fact, the success of many so-called lower-tech industries depends heavily on high-tech products and services as inputs. This is one way a region gains from R&D investments focused on particular high-tech industries raise the performance of the entire regional economy and a one rationale for economic development strategies focused on industry clusters rather than individual industries. Our analysis of technologies that are vital to the auto industry, presented in section VI, makes this crystal clear.
The short answer to the question that begins this section is straightforward: the region's overall high-tech component is small but it includes several promising niches. The success of these smaller segments will be crucial to the region's overall economic performance.
This section, which presents basic data on NEO's high-tech industries, uses two primary data sources: The first is SIC based; the second is CorpTech product-based. Both sources are used to identify the region's overall position in the high-tech economy as well as to pinpoint competitive industries.
This is really about knowledge -- a radically different input
The underlying reason why NEO should focus special attention on this set of issues is that the relative importance of inputs or factors driving the performance of industries has shifted. As pointed out earlier, Cleveland is heavily invested in factors that historically explained the performance of older industries. Knowledge increasingly has become the key input for a wide spectrum of industries (like computers, aircraft, software, and biotechnology). What makes this situation so different and so significant is that, unlike a piece of equipment, many firms can use the same knowledge simultaneously. For example, Hewlett Packard's R&D investment in semiconductor technology creates "spillovers" to other firms, including competitors. (footnote 34) A spillover is simply the knowledge produced by one organization's R&D that leaks out and is therefore available to firms that are not paying for the R&D. Other firms can tap into these knowledge spillovers to improve their performance. (footnote 35)
The central implication for older industrial regions is that investments in education and S&T have become more significant as sources of industry performance and a region's standard of living. In large part, the growing significance of education and S&T stems from firms' behavior as highly interdependent industry neighbors, creating and drawing on a pool of industry-specific knowledge. What it boils down to is that regional policy makers must be increasingly alert to ways to use these investments to strengthen the knowledge base underlying NEO's existing and potentially emerging industries.C. WHO SHOULD CLEVELAND COMPARE ITSELF WITH?
The answer to this question is: It depends on the purpose. On one level, largely reflecting the basic economics of high-tech industry locations within the US, it's important to position Cleveland relative to all US regions. This perspective permits us to establish the region's competitive high-tech position by industry. (footnote 36) On a second level, it also makes sense to position Cleveland among metropolitan regions with similar industrial histories. Exhibit B lists metropolitan areas commonly used for comparison. Given the region's starting point, how good is its performance? This section examines both benchmarks.
1) SIC-based. This report relies primarily on a definition of high-technology industry developed by the Bureau of Labor Statistics (BLS). (footnote 37) BLS selects SIC industries based on two criteria: the fraction of R&D to sales and the percentage of employees classified as scientists or engineers. (footnote 38) The region's overall position is determined by adding up the positions of individual three-digit industries.
According to this definition, total US high-tech jobs declined about 11%; the Cleveland region's total declined by an even greater amount (17%). By this measure (absolute growth in high-tech employment), the region's high-tech economy is under-performing. Because its share of total US manufacturing jobs declined, however, the region's high-tech index was essentially unchanged between 1988 and 1995. (It's important to keep in mind that the nationwide employment decline should not be interpreted as a drop in high-tech's importance; rather, it largely reflects inadequacies in the usual measures.)
2) Product-based (CorpTech). An alternative high-tech measure defines high-tech industries as those producing high-tech products. We make this calculation by computing the region's share of high-tech companies, drawn from the CorpTech database.
For both definitions, NEO is benchmarked against 37 large metropolitan regions. Essentially, the two databases tell us the same "big picture" story.
Table 1 below summarizes the employment status and US share in 1995 for all NEO high-tech industries with employment exceeding 1,000. Because NEO's total manufacturing employment is roughly 1.2% of US manufacturing, any high-tech industry with more than 1.2% of US employment in that sector is considered as concentrated in NEO. For example, with employment of nearly 1,800 workers, industrial organic chemicals (SIC 281) is about 30% more concentrated in NEO than we would expect if Cleveland got its per capita share.
| Industry | Employment | US share (in 1995) | % share change |
| Electrical industrial apparatus | 6,502 | 4.0% | none |
| Measuring & control devices | 5,265 | 2.0 | -3.3% |
| Medical instruments & supplies | 4,448 | 1.7 | none |
| Miscellaneous chemical products | 3,892 | 4.7 | 9.4 |
| Special industrial machinery | 3,881 | 2.2 | -6.1 |
| Aircraft & parts | 3,197 | 0.8 | -22.3 |
| Soaps, cleaners, etc. | 2,943 | 2.5 | 64.7 |
| Paints & allied products | 2,535 | 5.0 | 9.0 |
| Electrical components | 2,307 | 0.4 | -4.0 |
| Miscellaneous electrical equipment | 2,088 | 1.4 | 214 |
| Industrial inorganic chemicals | 1,933 | 2.4 | -11.9 |
| Industrial organic chemicals | 1,786 | 1.5 | -2.7 |
| Communications equipment | 1,626 | 0.7 | -8.6 |
| Plastics & synthetics | 1,310 | 1.1 | 30.7 |
| Guided missiles, etc. | 1,236 | 1.4 | 54.2 |
Employing the SIC-based definition, NEO's high-tech industry sector is about 10% below the median of our 37 regions. See Figure 10. Moreover, in recent years the region has seemed stuck at this position with no change occurring from 1988 through 1995.
When asking how we stack up, it's natural to draw comparisons with Silicon Valley or Route 128, which represent our most visible national benchmarks. Using the most common SIC-based yardstick of high-tech presence, Cleveland's high-tech employment is roughly one-fifth the scale of Boston's and approximately one-fourth that of San Francisco. So, strictly from the standpoint of geographic concentration of high-tech industry, the Cleveland region is not a major player. If it is to become one, it will occur with specific high technology industries.
Using an alternative, product-based measure, NEO's high-technology base is a scant 15% the size of Boston and 19% that of San Francisco. (footnote 39) This calculation is based on CorpTech, a national directory database of about 55,000 companies identified by high-tech products. The Cleveland region's position is derived from a calculation of each region's share of CorpTech companies. For example, in 1994 Cleveland had 1.15% of US high-tech employment and 1.4% of CorpTech's high-tech companies.
Table 2 shows, with a few exceptions, the ranking of metropolitan regions based on 3-digit industries is consistent with the ranking based on the number of high-tech companies. For example, NEO is ranked fourteenth using the BLS measure and sixteenth overall with a total of 512 CorpTech companies. Differences in rank arise largely due to company size. For example, Seattle's rank slips when using the number of high-tech companies as a criterion because Boeing dominates the region's high-tech economy.
| Rank (SIC) | Metropolitan region | % of high-tech employment | % of high-tech products | High-tech concentration (15) |
| 1 | LA | 9.3% | 8.3% | 1.32 |
| 2 | NY-NJ | 6.85 | 11.2 | 0.86 |
| 3 | SF | 5.64 | 7.3 | 2.01 |
| 4 | Boston | 4.48 | 8.3 | 1.67 |
| 5 | Chicago | 3.56 | 4.3 | 0.92 |
| 6 | Dallas-Ft. Worth | 3.1 | 2.5 | 1.54 |
| 7 | Seattle | 2.61 | 1.3 | 1.99 |
| 8 | Philadelphia | 2.32 | 3.2 | 0.94 |
| 9 | Houston | 1.84 | 2.0 | 1.09 |
| 10 | Minneapolis | 1.71 | 2.2 | 1.23 |
| 14 | CLEVELAND | 1.15 | 1.4 | 0.91 |
Looking closer to home, NEO's scale of high-tech industry based on SIC industries exceeds Detroit, Columbus, Pittsburgh, and Baltimore but lags Indianapolis, Cincinnati, and St. Louis. The ranking based on CorpTech companies is very similar: It exceeds Buffalo, Baltimore, Pittsburgh, and Columbus but lags that of Indianapolis, Cincinnati, St. Louis, and Milwaukee. See Table 3.
So, to summarize, Cleveland is not a major national high-tech region but it ranks roughly in the middle among its closest industrial neighbors
.| Metropolitan region | % of high-tech employment | % of high-tech products | High-tech concentration |
| St. Louis | 1.47 | 1.05 | 1.29 |
| CLEVELAND | 1.15 | 1.4 | 0.91 |
| Milwaukee | 1.11 | 0.88 | 1.36 |
| Cincinnati | 0.95 | 0.73 | 1.13 |
| Indianapolis | 0.66 | 0.43 | 0.96 |
| Baltimore | 0.65 | 1.35 | 0.69 |
| Pittsburgh | 0.60 | 1.8 | 0.61 |
| Buffalo | 0.40 | 0.88 | 0.84 |
| Columbus | 0.38 | 0.53 | 0.58 |
Below Cleveland's industrial surface are several important high-tech niches. These are best identified by benchmarking each NEO industry's position against the same industry located in the best performing regions for that industry. This requires the following steps:
These calculations yield five categories, roughly organized from best to worst case:
Table 4 shows the distribution of CorpTech's 512 NEO high-tech firms by broad category of product technology. Using simply the number of companies as a measure of importance, four technologies stand out (these represent 56% of all NEO CorpTech firms). Roughly 29% of companies in this group have been founded since 1980. Not surprisingly, the region's advanced materials companies tend to be older (only 11% were founded since 1980) while software companies are mostly new (60% established since 1980). Although spawning a significant number of firms, gauged on a national basis NEO's software industry is too small to be a major player. However, there may be specialized segments of the software industry that are critical suppliers to the region's more prominent industries and, therefore, deserve special attention.
Only three technologies stand out. NEO ranks in the top five only in Advanced Materials. (This finding is consistent with our analysis of new technology based on patents.) However, two additional product technologies emerge -- Factory Automation and Subassemblies (ranked eighth and eleventh, respectively). Unfortunately, because the data are limited to 1996, we can't determine whether NEO is gaining or losing ground. (footnote 40)
| Technology | # Companies | % founded since 1980 | Rank of 37Metro Regions |
| Advanced Materials | 85 | 11% | 5 |
| Factory Automation | 73 | 23% | 8 |
| Subassemblies | 70 | 19% | 11 |
| TOTAL | 512 | 29% | 16 |
This section has briefly examined the status of the region's high-tech industries. This focus is justified because it's a crucial leverage point for the region's S&T investments. First, these industries generally exhibit a higher rate of innovation and productivity growth, are more competitive internationally, and pay higher wages. Second, high-tech industries influence the performance of all of the region's prominent industry clusters. This sector supplies much of the nation's new technology to all industries. Third, even though underrepresented, local high-tech industries are outperforming the bulk of the region's manufacturing industries. In summary, policies and programs that leverage this sector give NEO an opportunity to raise productivity and, consequently, the region's standard of living.
At the beginning of the 1980s, which was Cleveland's worst period of industrial decline since the Depression of the 1930s, the business community had an explanation for the region's dramatic loss of manufacturing -- high labor costs! There's no doubt that high costs of doing business in the region became an important competitive problem. But the explanation for Cleveland's industrial decline has two sides. Labor costs become a competitive problem only when not offset by high productivity. A region can live with high labor costs if it also has a high rate of innovation and productivity growth. However, if productivity growth is low, high labor costs reduce employment, retard the openings of new firms, and decrease investment. (footnote 41) (footnote 42) This was the underlying condition that undermined Cleveland's industries and caused decline in relative per capita income.
As discussed earlier, a large part of the solution to this problem is an investment strategy focused on factors that raise productivity. Almost without notice, Cleveland's overriding concern with high labor costs began to shift increasingly to education and technology as the solution to competitive problems. Picking just one measure to characterize the region's future, that measure would have been 'wages' in 1980; today it's 'education.' The key drivers of regional performance have shifted.
Moreover, this shift parallels a shift by leading magazines' assessment of the business climate of US states and metropolitan areas. Most major business climate indexes, such as those of Fortune magazine, have dramatically reduced the weight given to cost factors (such as wages and taxes) while raising such factors as education, quality of life, and entrepreneurship to the top. (footnote 43) This shift reflects the growing awareness that knowledge has become the key determinant of economic progress.
Figure 12 below describes NEO's education level relative to the median of the 35 metropolitan regions over two and a half decades (1970-96). For this purpose, education level is measured with percent of persons over 16 years of age with four or more years of college. The 1996 number was estimated from the Cleveland region's portion of the Current Population Survey and, consequently, has a large error. The bar next to the 1996 number shows the estimated standard error at the 95% confidence level. It's reasonable to assume that the true number is anywhere between 84% and 101%. The best guess is 94%.
Ignoring other regions, the Cleveland region's education level seemed to be improving. For example, the percent of NEO's adult population with four or more years of schooling increased from 10.5% to 19.9% between 1970 and 1990. However, examined relative to 35 competitor metropolitan regions, its education level was low and declining for two decades (1970-90). (footnote 44) Given this trend, no one should be surprised to discover that NEO's per capita income fell from a position 5% above the median of the 37 regions to 4% below the median over the same period. (footnote 45) (footnote 46)
Good News! The 1996 estimate suggests some very good news. Following two decades of decline in the region's education level, it's possible that a stronger economy over the past decade, coupled with economic development efforts, have stopped the "brain drain" hemorrhage.
The previous data describes the entire workforce. Education is also a key ingredient of competitive manufacturing. Viewed nationally, comparative advantage in international trade is predominantly in high-technology goods and services. Workers in these industries are increasingly better educated than workers in industries that do not export. (footnote 47) Furthermore, some evidence at the state level substantiates the significance of manufacturing worker education for raising productivity growth rates. (footnote 48) (For manufacturing, education level is measured with the percent of workers with at least one-year of college.) For reasons tied to Cleveland's historically dominant industries, the region ranked in the lowest quartile in 1996 based on manufacturing workers with some college. (footnote 49) (See Table 5 below.)
To summarize, Cleveland does not have a well-educated manufacturing workforce. This finding is consistent with both the poor overall performance of the region's manufacturing industries and below-average scale of high-tech industry. Without change, Cleveland's manufacturing industries will have increasing difficulty competing internationally. The region will be most successful if it builds on industry strengths.
| Top quartile | San Francisco, Seattle, Rochester, Atlanta, Houston, Denver |
| Second quartile | St. Louis, Phoenix, Milwaukee, Minneapolis, Grand Rapids, Dallas |
| Third quartile | Detroit, Boston Cincinnati, Pittsburgh, Los Angeles, New York |
| Lowest quartile | San Diego, Philadelphia, Kansas City, Chicago, Cleveland, Miami |
Achieving the median education level isn't good enough if the region's goal is relative gain in per capita income.
The gap between Cleveland's current position and the education level of the top quartile of US regions is more than double the median gap -- the gap is about 15% for both four or more years of college and at least some college. (Metropolitan regions in the top quartile, by estimated rank for four or more years of college, include: San Francisco, Seattle, Washington, DC, Denver, Minneapolis, Dallas-Fort Worth, Atlanta, Hartford, and San Diego.) Cleveland's top quartile gap is calculated as the ratio of Cleveland's percent college or better to the percent of the lowest metropolitan area in the top quartile.
These data raise a fundamental concern for the region's future. First, for 1992, REI estimated that it would be worth an estimated $2 billion annually in 2000 to achieve the best education level for Great Lakes metropolitan areas. This is an amount equivalent to writing one combined check for three annual budgets -- the City of Cleveland's, the City School District's, plus Cuyahoga County's annual budget. Achieving the average of the top five US metropolitan areas would more than double the additional income (an estimated $5 billion in that year alone).
Second, even if achieving the median were good enough, the 6% gap would require graduating four to five times the current number of persons all in one year and retaining 100% of the graduates for NEO industry. This large gap highlights the necessity to be strategic with investments in degree programs -- i.e., selectively building higher education for important clusters of industries in order to maximize the long-run effect on productivity.
Third, achieving the median isn't enough for most industries. NEO is shooting at a moving target because education is increasing everywhere. Therefore, the region should consider setting more ambitious goals -- e.g., building education for industries to compete in the top five of metropolitan regions with the same industry cluster.
The two ways that a region's education level changes are: (1) local higher education institutions graduate students who stay in the region and (2) in-out migration. Northeast Ohio's four-year higher education institutions currently award about 17,000 degree annually. This is NEO's most powerful economic development resource. To the extent that these individuals stay in Northeast Ohio, they raise the region's education level. The size of the long-run economic payoff, through productivity gains, depends on several factors, including the proportion taking Cleveland jobs, mix of degrees (e.g., economics versus electrical engineering), quality of graduates, and industry destination. If net migration is zero (the outflow of college graduates equals the inflow), then NEO is completely dependent on the scale and mix of higher education to raise the region's education level. In the worst of all possible worlds, the 17,000 graduates leave NEO and no college graduates move to NEO from other regions.
Migration can be a second powerful factor altering the region's education level. It can work for us or against us. Until recently net migration of college graduates was negative (more college graduates exited the region than entered). Why? Essentially, migration is primarily driven by local industries' demand. Ultimately, whether Ohio is a net importer or net exporter of college graduates depends on the strength and mix of our industries.
So what we have is a 'chicken-egg' problem since net migration is really determined by the conditions that cause industries to be strong or weak 3/4 namely, factors such as education and technology -- the very same things that we need to build strong industries.
To complicate matters a bit more, industry demand conditions interact with supply (the workforce). For example, for a given industry demand, individuals will be more readily drawn to regions with a rich mix of amenities and a high quality of life. (footnote 50) Less attractive states and metropolitan regions must pay a premium for highly skilled, mobile workers. From a region's perspective, growth from new inputs results largely from investment and critical flows of people and physical capital into and out of the state. This condition should put a premium on knowing a lot about the determinants of regional migration patterns and private investment. Both are essential for economic development policy. (footnote 51) Strong industries will selectively draw in-migrants; however, NEO may be able to speed up the process by investing in amenities that are attractive to educated, skilled workers. (footnote 52)
To sum up, both forms of investment (human and physical capital) intermingle because investment usually includes newer technology -- and migration typically involves younger, better-educated workers. The intermingling of the two components strengthens the argument for coordinating public investments in education and technology. Given the region's recent population growth, it's likely that the "brain drain" of the late 1970s and early 1980s is over; however, while some net in-migration may be occurring, it's not likely to be large relative to the region's population size.
Therefore, for now it makes sense to assume that the potential for raising the region's overall education level hinges on our degree programs -- specifically, the scale, mix and quality of graduates produced by higher education institutions plus local industry's capture of these graduates. (footnote 53) The principal reason for Ohio's net loss of college graduates in the previous decade was the weaker demand for scientists and engineers by the state's technology-based industries. (footnote 54) A superficial glance at these statistics might tempt politicians to cut higher education programs. In other words, since NEO seems to be a net exporter of college graduates, why not simply cut the scale of degree programs? From a broader perspective, however, this would be a serious mistake because it takes a passive view of industry demand. Knowledge-based industries are drawn to concentrations of educated, skilled workers. Therefore, cutting programs that are well matched to Ohio's industry clusters would hasten the movement of knowledge-based industries away from Ohio, constrain the diffusion of technology locally, and limit the growth of emerging high-tech industries.
NEO's four-year colleges and universities grant nearly 17,000 degrees annually. Table 6 gives raw numbers of NEO degrees for each institution by level: bachelors, masters, doctorate, and professional. Some differences by level of degree are worth noting. For example, bachelors increased 20%. At the same time, doctorates increased 22%, although this number is well below the US growth in doctorates (32%). Masters, which include MBA degrees, increased 48%; in contrast, professional degrees declined 6% locally while increasing 2% nationally.
| Academic | Doctorate | Professional | Master's | Bachelor's | Associate's | Total |
| CWRU | 191 | 390 | 1,032 | 704 | 0 | 2,317 |
| CSU | 27 | 273 | 1,015 | 1,787 | 0 | 3,102 |
| John Carroll | 0 | 0 | 243 | 787 | 0 | 1,030 |
| KSU, all | 140 | 0 | 1,006 | 3,206 | 818 | 5,170 |
| NEO University College of | 0 | 105 | 0 | 0 | 0 | 105 |
| Notre Dame | 0 | 0 | 2 | 119 | 2 | 123 |
| Oberlin | 0 | 0 | 6 | 673 | 0 | 679 |
| Univ. of Akron, all | 110 | 182 | 809 | 2,548 | 758 | 4,407 |
| NE Ohio | 468 | 950 | 4,113 | 9,824 | 1,578 | 16,933 |
| NE Ohio, percentage of | 22 | 29 | 25 | 19 | 8 | (100) |
| OHIO TOTAL | 2,166 | 3,240 | 16,232 | 51,080 | 20,059 | 92,777 |
Source: CASPAR Database System, Quantum Research Corporation and the National Science Foundation
One way to evaluate the scale of degree programs is simply to compare degrees awarded with population (degrees per capita). The per capita index makes a crude adjustment for scale of the local economy. Using the per capita index for the region as a whole, measured relative to the 37 metropolitan regions, NEO's four year institutions awarded about 4% more bachelors, 5% more masters, and 2% more doctoral degrees in 1994. Nearly three-fifths were bachelors. The bachelor's share has remained approximately constant over the ten years while masters increased from 20 to 24% of all degrees (this category includes MBAs). Doctorates make up only 3% of all NEO degrees granted.
How do we know what's the right scale? Can we simply conclude that the region's scale and mix are fine. The answer is: No we can't, for several reasons: First, at the economy-wide scale, the appropriate scale depends on the pattern of in-out migration. For example, if there is net in-migration of college graduates, then a lower scale will work. In contrast, if net in-migration is negative, then a larger scale is necessary. So, ideally, comparisons of Cleveland with the other 36 metropolitan regions should adjust for in-out migration.
Second, at the industry level, industry mix matters a lot. For example, given the region's concentration of chemical-related industries (ignoring migration), a per capita scale of chemical engineering graduates exceeding 1.0 would be necessary to satisfy the chemical industry's requirements. This suggests a need to adjust degrees awarded for industry mix. (This calculation is not straightforward because particular degrees are employed by more than one industry.)
Third, desired scale should also take into account expected growth. For example, NEO's degrees awarded increased by 20% 1985-1994 while the region's index of degrees awarded per capita changed very little. Setting goals for scale of degree programs should adjust for expected job growth by occupation and industry over the next five to ten years. See Figure 13.
One way to identify potential strengths and weaknesses in NEO's degree programs is to group degrees granted into four categories based on two calculations: (1) The extent of local concentration, which is simply NEO's share of all degrees awarded by discipline, adjusting for population. An index of 1.0 means that NEO grants its per capita share of degrees; an index exceeding 1.0 means that the region grants more than its per capita share of degrees. (2) Growth in the index -- A number exceeding zero tells us that NEO is gaining relative to the US on a per capita basis (scale is increasing); a number less than zero indicates that NEO is losing share relative to the US. The two figures below show these calculations by discipline for both doctorates and bachelors degrees. See Figures 14 and 15.
The two figures hint at NEO's strengths and weaknesses by discipline. However, an assessment of strengths and weaknesses relative to NEO's existing and future industries is beyond the scope of this study.
Several bachelors degree disciplines stand out. To note a few: Despite the increasing importance of computers and information technology, the region awards a very low quantity of computer science degrees. Electrical engineering is somewhat below average scale and lost ground over the past decade. At the doctoral level, several disciplines are noteworthy, including materials engineering, chemical engineering, and chemistry. Strikingly, NEO has virtually no scale for computer science doctoral programs.
NEO appears to have begun reversing the decline of its education position relative to competitor regions. Although gaining slowly, we are still far behind -- especially if we set more ambitious targets, like reaching the top quartile of regions based on education. Second, the region would have to grow the region's scale of degree programs just to keep up with industry's growing requirements for college graduates. Third, mix matters a lot, so it's necessary to expand the base selectively. See Figures 16 and 17. The low number of computer science graduates at all levels is the most obvious gap. A clear picture of gaps requires an industry-level assessment of degree requirements. Fourth, policy, interacting with a state's demographics, determines the scale and mix of higher education. For the most part, Ohio can decide how many college graduates we want to produce by adjusting such things as entrance requirements, tuition (including in- and out-of-state differentials), scholarships, and the mix of programs (such as graduate versus undergraduate). Ohio has chosen to make the tuition price high, which limits the number of residents who choose to go to college. The quality of programs, relative to tuition, also affects the attractiveness of Ohio's colleges and universities for talented entrants. An important question to consider is: How large an effect does Ohio's relatively high tuition have on the region's ability to raise its education level?
R&D is big business. R&D performed by NEO's institutions currently totals an estimated $2.3 billion. Roughly $2 billion R&D is performed by industry (footnote 55); universities are responsible for about $174 million (University Hospitals' research support gets mingled with CWRU's medical school), NASA-Lewis $70 million and the Cleveland Clinic $54 million. (footnote 56) See Figure 18.
Two reasons explain the region's stake in university research: the first is its potential for producing new technology of consequence for local industries; second, and likely more important, is the know-how developed through research. The region has a stake when graduates feed into local industries or when faculty consultants solve technology problems or give important ideas to regional companies. (footnote 57)
A number of national studies provide strong evidence that the nation's investments in academic research produce a high social rate of return (perhaps 25-50%).(footnote 58) (footnote 59) But a high national social rate of return is not sufficient to justify a state or region's investment in research. The reason is very simple: Knowledge spillovers are available to everyone. (footnote60) (footnote 61) In fact, states and local areas face a dilemma, namely that first-rate research programs are necessary to produce globally competitive technology but the best research with the greatest commercial potential will quickly become known throughout the world. Therefore, at best, proximity confers only a temporary advantage to a region and its industries.
Most basic, (yet frequently overlooked), calculating the rate of return to state and local research investments requires tracing benefits to local industries. If university research is to affect the Cleveland region's productivity growth, it must create local startups or influence development of new technology by its R&D labs, and benefiting firms must have production located in the region. Moreover, entrepreneurship is an essential ingredient for capturing the economic benefits of technology created by university research. The policy challenge is to figure out ways to create a highly effective commercialization process.
Finally, research and graduate education go hand in hand. In fact, the destination of graduates will substantially affect any calculation of payoff from state and regional investments in research. (footnote 62)
The combination of large investments and complexity demands that evaluation become a more serious component of the investment strategy. (footnote 63) Evaluation metrics also require a framework. A strategy for maximizing the economic payoff will support the appropriate research, encourage commercialization, and create mechanisms for capturing the economic benefits.
This section of the report focuses on the region's university research. A separate section will examine the most competitive industry sources of new technology.
Nearly a decade ago, an REI report to Cleveland's Technology Leadership Council set a goal to increase the region's research scale from 60% of the median per capita to 80-90% by the year 2000. (footnote 64) This was to be accomplished by selective investments in area universities. Our estimate was that making this happen would require increasing the real growth rate of funding from 4.3% in 1980-88 to 5-7%. REI's study identified several necessary conditions for achieving this goal:
These conditions still apply!
The community's efforts have produced a major success story: The region's research scale increased significantly from 1985 through 1994. In other words, NEO's universities are gaining R&D support at a faster pace than the average of all US universities. See Table 7. Total research funding has more than doubled in real terms (adjusting for inflation) over the past decade. This growth significantly exceeds the average of all US universities, which was 62%. Federal support to NEO universities increased by somewhat less (76%); this amount exceeds the comparable US growth rate of 55%.
| Academic Institution | Total R&D Expenditures | Federal R&D Expenditures | State & Local R&D Expenditures | Industry R&D Expenditures | Own R&D Expenditures | Other R&D Expenditures |
| Case Western Reserve U. | 133,272 | 97,302 | 3,513 | 5,708 | 12,468 | 14,281 |
| Cleveland State University | 10,570 | 4,620 | 2,531 | 701 | 2,718 | 0 |
| John Carroll University | 0 | 0 | 0 | 0 | 0 | 0 |
| Kent State University | 11,312 | 6,005 | 899 | 130 | 3,449 | 829 |
| Northeastern Ohio Universities | 2,446 | 1,136 | 23 | 175 | 673 | 439 |
| Notre Dame College | 10 | 0 | 0 | 0 | 10 | 0 |
| Oberlin College | 0 | 0 | 0 | 0 | 0 | 0 |
| University of Akron (all) | 16,783 | 2,331 | 2,331 | 2,331 | 5,869 | 353 |
| NE Ohio Total | 174,393 | 9,297 | 9,297 | 9,297 | 25,187 | 15,902 |
| Ohio Total | 626,033 | 50,399 | 50,399 | 50,399 | 103,571 | 62,221 |
Source: CASPAR Database System, Quantum Research Corporation and the National Science Foundation.
Analysis by: Center for Regional Economic Issues, Weatherhead School of Management, Case Western Reserve University.
Research scale is simply the ratio of NEO's research dollars per capita to research dollars per capita for the median of the 37 metropolitan regions. An index value of 1.0 means that the region's R&D share is equivalent to the sample's on a per capita basis.
Relative to the US, the region's university research scale increased consistently from .66 in 1985 to .75 in 1990 and to .81 in 1994. (The index is calculated as NEO's federal research dollars per capita divided by US federal research dollars per capita x 100.) See Figure 20.
With a few exceptions, Cleveland's university research is overwhelmingly Case Western Reserve University's. For example, CWRU accounts for more than three-fourths of the region's total expenditures on research and 86% of federally funded university research in 1994. (NSF treats much of the research support of CWRU and University Hospitals as one unit.)
For regional economies, however, it's not sufficient to make national gains; it's also necessary to make gains against competitor regions. Relative to the 37 metropolitan regions, NEO's scale increased from .68 to .78 between 1985-90; then Cleveland lost some of the ground it gained in the last half of the 1980s and its position declined to .76 by 1994.
The two charts above tell two stories: The first is that NEO's academic research programs are becoming more competitive nationally (gaining share within the US). The second tells us that the region has lost ground relative to competitor regions since 1990. Since Cleveland competes against specific places and not the US generally, the latter trend is the more relevant to assessment of economic implications.
Measured with respect to either the US or the 37 metropolitan regions, Cleveland's university research scale is low (nearly one-fourth below the median per capita amount for the 37 metropolitan regions). Figure 21 compares NEO's research scale with eight other metropolitan regions selected from the 37. Since 1.0 is the median of the 37 regions, the chart quickly suggests that a goal to achieve the median, though challenging, may not be good enough. For example, Pittsburgh's research scale is 21/2 times Cleveland's. Baltimore's scale exceeds Cleveland's by a factor of 31/2. Closer to home, Columbus has a research scale that is double this region's.
None of this discussion is meant to argue that a region's scale of university research by itself drives a regional economy. Perhaps unfortunately, no one factor determines a region's economic performance. It takes a combination of factors. Furthermore, as previously discussed, university research can affect the region's performance only if it increases the technology and education incorporated into local industries.
Despite impressive gains, major gaps in research capabilities exist. See Figure 22. To summarize:
Federal support of research is the best indicator of a program's competitiveness. Alarmingly, the share of total federal support of NEO's engineering research declined from 1.35% to 1.01% between 1985 and 1994. Within engineering, only chemical engineering gained relative to competitor regions.
The region's success in gaining scale relative to other regions stems from success gaining an increasing share of a large, growing NIH budget for R&D. Two-thirds of growth in federal support of academic research over the past decade came from NIH. See Figure 23. NEO's share increased from 1.21% in 1985 to 1.40% by 1994. Life Sciences overall increased its share from .88 to 1.13% of total US support of life sciences. It is striking that all other federal sources of R&D support to NEO's higher education institutions declined over the same period. (Not included in these figures is the Cleveland Clinic. The Clinic received $24.5 million in research support from NIH in 1996.)
Chemical engineering shows the biggest jump, significantly increasing its share of US support of this field from .72% to 3.6% over the decade.
The federal share of total university research has declined from 75% to 65% of total R&D support. Simultaneously, funding from other sources has grown. In 1994, the state and industry invested a combined total of $20 million in NEO's academic research. The growth in state support signals states' growing interest in universities as an important source of new technology and economic development. In 1994, state and local governments contributed over $9 million to NEO's university research. State support is predominantly focused on state institutions.
This category of support is important for several reasons. First, it is an expanding component. Together with industry's growing influence on university research, state funding represents increased attention to commercialization and efforts to strengthen university-industry connections for economic development purposes. Second, state support has the potential to help jump-start research programs, eventually helping to develop competitive areas of research. However, the evidence concerning the effect of state support of research isn't clear -- at least regarding NEO's public institutions. The effect is not obvious from either data on relative gain in federal R&D support or for changes in the National Academy of Sciences (NAS) ratings of doctoral programs between 1988-1993. (footnote 65) Third, changes in the structure of funding for research and graduate programs alter the mix and quality of these programs. No evidence of these effects exists to guide future decisions.
Industry's support of research at NEO's institutions grew significantly from $7.2 to $10.9 million between 1988 and 1994 (about 50% in nominal terms). This growth parallels a national trend. This source is often viewed as being especially significant because it is seen as an indicator of industry's interest in the commercial potential of university research -- at particular universities. A somewhat surprising finding is that growth in the state's support of research hasn't leveraged relative gains in industry support. In the 1990s, industry's funding of research by NEO's universities declined relative to institutions nationwide. Although industry funds have increased significantly in the 1990s, the region's growth is less than that of the U.S. (118% versus 125%). Unfortunately, without digging into each institution's proprietary data, we can't know the discipline-level breakdown, since NSF doesn't provide industry-funding data by discipline.
One hypothesis: It is beyond the scope of this study to learn more about industry's support of specific research and graduate programs in this region. However, one hypothesis is that industry funding of academic research is increasingly targeted to the strongest programs, wherever they may be. If so, without any change, the trend away from NEO's university research programs may continue.
Universities are also funding a growing piece of their own research. For example, NEO's universities funded over $25 million in 1994. This self-funding grew from 11% to 14% of the total over the decade and now exceeds the combined total of industry and S&L funding of research.
Cleveland has shown that, with commitment and a plan, it can gain research scale. The region is on target for achieving its goal of 80-90% of the median scale of 37 metropolitan regions by the year 2000 (the scale in 1994 was .76, or 76% of the median). However, such relative gains require increasingly competitive research programs: Other institutions are also investing in research so, we're aiming at a moving target.
Therefore, TLC's challenge for the region's university research base is twofold: First, it should continue investing in the region's research and graduate education capacity. Second, because gains have been made on a narrow base, largely in the biomedical sciences, it is a good time to take stock of priorities. (With the exception of mechanical engineering, all NEO doctoral programs that ranked in the NAS' top quartile were located at CWRU and were in medical and biological sciences.) In particular, the region should consider broadening the base of competitive research programs -- especially in engineering fields.
R&D is the region's stealth sector. It quietly produces a nearly invisible product -- knowledge. One reason for caring about the health of the region's R&D base is that it's a large industry. It generates a lot of income. However, the fundamental argument for caring about local R&D starts with the firm. Evidence clearly shows that the portion of a firm's output devoted to R&D increases the firm's productivity. (footnote 66) But what about states and regions? We mostly care about the R&D base because of its potential to create new technology for local industries and, therefore, raise productivity and thus the region's standard of living. Most state and local S&T policies are at least indirectly based on the belief that R&D is an important determinant of regional economic performance. Yet there's no guarantee of a one-to-one correspondence between company or industry R&D and local economic growth. For instance, a company's R&D and its production facilities are often located in different states or regions. So the technology can quickly leave the state.
Even though in reality each component interacts with other components, the Figure 24 suggests a simplified commercialization path from invention to economic impact.
Moreover, technology moves increasingly freely among companies and regions of the world. Since this is the case, why not simply let someone else's region do the R&D. We could then adopt other regions' technology. There are several reasons. First, local R&D sources of new technology probably speed up commercialization and technology. Earlier adoption creates a head start and early adopters gain an advantage. Second, a large, influential R&D base creates opportunities for new entrepreneurial companies and industries. Entrepreneurial startups are not guaranteed but local R&D increases the likelihood of startups. Third, without a set of R&D-performing local companies, there would not exist any chance to commercialize university technology such as MEMS (microelectro-mechanical systems) technology for NEO.
The firm-level evidence connecting R&D to firm performance is solid. But, largely because of data limitations, the evidence connecting local R&D to local economic performance is on less firm ground. If firms located all of their R&D and production facilities in one region, then we'd clearly expect a strong correlation between local R&D, productivity, and per capita income growth. But the reality is that R&D and production sites don't perfectly coincide and, furthermore, the innovation system is leaky. Knowledge of technologies moves quickly among labs and regions. The beneficiaries need not reside in the same region with R&D.
Is there any correlation between local R&D and economic performance? We have two sources of evidence on this point. First, using formal productivity analysis methods, we statistically evaluated the contribution of a state's R&D and manufacturing worker education to state-level manufacturing productivity growth. (footnote 67) The evidence is strongly consistent with firm-level studies. In addition, the evidence also shows that interaction of R&D and education can become a source of "increasing returns" (the returns to R&D and education are magnified, causing a disproportionate gain in manufacturing productivity). The latter finding is consistent with new theories of economic growth which single out "knowledge" as the key factor.
Second, in a previous study, in order to confirm the potential benefits from local R&D, we statistically evaluated the relationship between average state-level industry R&D over the years 1976-83 and state per capita income in 1986 for the 22 states for which these data are consistently available. If R&D boosts productivity growth, then we should expect per capita income gains to follow. On average, industry R&D was 1.4% of Ohio's Gross State Product (GSP) from 1976 through 1983. The regression shows a strong, positive, and statistically significant association between state per-capita income and lagged industry R&D as a percent of GSP. (The lag attempts to examine causation.) The results suggest that a one-third increase in a state's R&D would, on average, be associated with 4% additional personal income. (footnote 68)
Beyond the obvious interest in R&D as a large industry sector, it's reasonable to think that geographic concentrations of R&D affect industry productivity and per capita income. It's also prudent to include R&D as a key ingredient of a local economic development strategy focused on increasing the region's standard of living. R&D is a critical component of the local S&T/ innovation system. However, the estimated 4% higher per capita income resulting from a one-third increase in R&D is just an average. The contribution of R&D to productivity and per capita income will vary from region to region depending on two factors (1) the strength of the technology-specific R&D network, and (2) the extent to which local R&D is connected to local production.
The scope of this study did not permit analysis of all possible technologies of significance to the Cleveland region -- especially because, in using a new methodology, Cleveland is operating in new territory. Therefore, to make the task more manageable, it was necessary to limit the examination to several existing and emerging technologies. (footnote 69) There were selected in concert with Cleveland Tomorrow's Technology Leadership Council and the Greater Cleveland Growth Association:
In addition to the five selected broad technologies, REI has analyzed one important narrower technology to illustrate the significance of investment issues involving university research and graduate programs -- MEMS (microelectro-mechanical systems). MEMs is discussed in a later section of this report.
Our methodology for identifying regional technology strengths starts with patents. A patent -- a property right in the commercial use of a device -- is granted only if the invention is evaluated as non-trivial. This means that the device would not appear obvious to a skilled practitioner of the technology. Patents are inventions that represent "potential" economic value; innovation means commercializing the invention by incorporating it in a new product or process. (footnote 70)
Working in collaboration with the National Bureau of Economic Research, supported by the National Science Foundation, the Center for Regional Economic Issues has developed a unique US patent database. The database contains nearly 3 million patents extending from 1969 through 1996. Inventor location information is available starting in 1975. The US system currently generates about 100,000 patents annually (up 41% since 1986). Nearly half of US patents are assigned to foreign companies. The Cleveland region currently produces 1.12% of all domestic patents. NEO's share has fallen dramatically over the past decade (from 1.72% in 1986 to 1.12% in 1995, or one-third). These statistics are based on R&D location -- not the assignee (headquarter) location. R&D location is determined from information in the patent on the inventor's address. Therefore, our results will differ from other patent analyses relying on assignee location.
1. NEO's top 15 patenting organizations from 1990 through 1995 produced 1,894 patents. Including independently owned patents, the region added nearly 2,800 new inventions over this period. (Given the magnitude of new patents assigned to individuals, it would be important to examine individual inventors more closely in future research on potential entrepreneurs.)
| Company | Number of Patents |
| BF Goodrich | 266 |
| Standard Oil | 144 |
| Goodyear Tire and Rubber | 289 |
| Lubrizol | 191 |
| GE | 149 |
| Picker | 131 |
| Nordson | 97 |
| NASA-Lewis | 86 |
| University of Akron | 67 |
| Babcock & Wilcox | 54 |
| Eltech Systems Corp. | 52 |
| Bridgestone/ Firestone | 51 |
| Allen-Bradley | 50 |
| PPG Industries | 42 |
| Lincoln Electric | 42 |
| TOTAL | 1,894 |
2. Table 7, shown above, lists the top 15 NEO patenting organizations and might suggest a ranking of organizations based on importance -- but it doesn't. As everyone involved in patenting knows, very few patents are really important. In fact, an entire literature and cottage industry has developed around the use of patent data to measure importance. This issue is discussed later with the methodology for identifying technology strengths. Ultimately, the importance of the patented technology hinges on what it contributes to the economy.
3. Cleveland is becoming a less important center of new technology. NEO produces somewhat more than its per capita share of patents. However, one disturbing finding is the region's loss of share of all US patents over the past two decades. From 1986 through 1995 NEO's share of US domestic patents declined from 1.72% to 1.12%; the share of all US patents (including foreign) declined from 0.92% to 0.62%.(footnote 71) This loss of share can be interpreted as deterioration in the region's R&D capabilities relative to the nation by roughly one-third. Cleveland is becoming a less important center of new technology.
There are at least two possible explanations. One possibility is that the region's industries are relatively old and, consequently, are not producing the quantity of new technology characteristic of new industries -- such as information technology. A second possibility is that Cleveland corporations are shifting their R&D to other locations. This would show up as a decline in the percent of the region's top companies' patents that are produced here.
We tested the second hypothesis. First, we calculated the total number of patents awarded to each of the region's 52 corporations with 20 or more patents over the period 1985-95. Second, we computed the percent of each corporation's patents produced in NEO. With a few ups and downs, the region's overall share of patents awarded to the 52 corporations as a group declined by roughly 40% (from a 19% share in 1985 to 11% in 1995).
This invisible drift of corporate R&D away from the Cleveland region raises a critically important issue: Why is this happening and what can the region do to strengthen NEO as a location for corporate R&D? How can NEO stem further losses?
One inventor or one firm acting alone never creates technology. It truly is a cooperative venture, relying on an innovation system that links inventors, firms, technologies, regions, and countries. Often those cooperating are not aware of knowledge flows based on their R&D. Other times, a firm will invest a portion of its R&D to actively acquire external knowledge. (footnote 72)
REI's new database provides a powerful resource for research on R&D networks and technology strengths. The database of nearly 3 million patents includes highly detailed information on patented technology. The specific variables are: 400 patent classes; the name of the organization to which the patent is assigned; the inventor(s) with address (allowing for identification of R&D location, including county); claims; country (today roughly 1/2 of all US patents are assigned to a foreign company); type of organization, including corporation, university, hospital, government lab, or independent inventor; and cited patents and other cited sources, such as academic publications.
Patents leave a "paper trail" by including citations to other patents and, frequently, non-patent citations, such as academic papers. Cited patents identify the building blocks of a new technology. Cites make it possible to analyze the diffusion of technology over time and by organization, technology, and location. For example, the citation information permits us to trace knowledge flows both within and between states and metropolitan regions (for example, between Silicon Valley and Los Angeles or Cleveland) by technology and by type of R&D institution. In principle, a citation of Patent X to Patent Y means that X represents a piece of previously existing knowledge upon which Y builds. (footnote 73) (footnote 74) (footnote 75)
The simplest measure of an "important" patent is the number of citations -- i.e., the number of subsequent patents that cite a patent as prior art. Kent State University's liquid crystals patents have received an astounding 439 patent citations to date. (The average patent receives fewer than three or four citations.) An analysis of the KSU citations reveals the rapid diffusion of KSU's technology to other regions, including Japan, California, and Texas. They also reveal the absence of the technology's use by Ohio companies. Significant local diffusion of the PDLC technology would show up as KSU's patents being cited by a network of corporate R&D labs in Ohio.
Patent citations can also be used to evaluate the importance or influence of particular organizations and regions by technology (i.e., those organizations and regions that have the greatest impact on the development of new technology).
Patents provide the most comprehensive source of information on new technology. Our methodology, which is described more thoroughly below, uses a new systems approach to evaluate NEO's technology strengths. The methodology identifies "clusters" of closely-related technologies using REI's patent data and 400 patent classes. What makes this a "systems" approach is that the technologies are derived from networks of R&D organizations (companies, universities, and government labs) working on specific technologies with R&D facilities located in various regions of the world. Because the technologies are "clusters" of individual patent classes representing broader sets of technologies, the coverage of NEO's technology strengths is very extensive. In other words, although the study focuses on a limited set of technologies, the clusters include the influence of a high percentage of all significant technologies embodied in the full set of 400 patent classes. Future work to pinpoint other technologies or to analyze "enabling" technologies can build from this work.
Our technology clusters provide a very significant advantage over analyses based on individual patent classes. The reason is that narrow technologies, such as those represented by individual patent classes, are highly interdependent. For example, polymers is not simply one patent class. It consists of a set of closely related patent classes (technologies), which form the different ingredients of many individual polymer technologies. In effect, it's equivalent to a chemist's mixing of different chemical ingredients under various conditions to produce a material with new properties. Our database and methodology give us the means to identify clusters of closely connected technologies surrounding a core technology, such as polymers. The methodology is briefly described below.
Broadly, we define a regional technology strength as a technology possessing several characteristics: