Center on Urban & Metropolitan Policy Technology and Tolerance: The Importance of Diversity to High-Technology Growth “People in Richard Florida, Carnegie Mellon University, and Gary Gates, The Urban Institute1

technology Findings An analysis comparing measures of tolerance, diversity, and high-technology success in businesses 50 metropolitan areas found:

The leading indicator of a metro- Metropolitan areas with high are drawn to politan area’s high-technology concentrations of foreign-born resi- success is a large gay population. dents also rank high as technology The five metropolitan areas with the centers. Eight out of the top ten places known highest concentration of gay residents metropolitan areas with the highest are all among the nation’s top 15 percentage of foreign-born residents high-technology areas: San Francisco, were also among the nation’s top 15 for diversity Washington D.C., Austin, Atlanta, high-technology regions: Los Angeles, and San Diego. Gays not only predict New York, San Francisco, San Diego, the concentration of high-tech Chicago, Houston, Boston, and of thought industry, they are also a predictor of Washington D.C. its growth. Overall diversity is a strong indi- and open- A high concentration of artists or cator of a metropolitan area’s “bohemians” follows gays as a high-technology success. Eleven significant indicator of a metropol- metropolitan areas with the highest mindedness.” itan area’s high-technology success. levels of overall diversity (based on Ten of the top 15 “bohemian” metro- gays, bohemians and foreign-born politan areas (those with the highest people) are among the top 15 high- concentration of artists, writers, musi- technology areas. San Francisco, cians, actors, etc.) also number Boston, Seattle, and Washington D.C. among the nation’s top 15 technology are the top four high-tech regions and regions. These areas include: Seattle, rank in the top six regions on the Los Angeles, New York, Washington composite diversity index. The D.C., San Francisco, and Boston. composite diversity index also strongly predicts high-tech growth.

June 2001 • The Brookings Institution • Survey Series 1 I. Introduction pools—not, as is more frequently success in the 50 most populated argued, to gain advantages from linked metropolitan areas in the United he rules of the economic networks of customers and suppliers. States.7 development game have Related research by Spencer Glendon changed. Companies were the finds that a good deal of city growth II. Methodology force behind the old game. over the course of the entire 20th TCities and suburbs gauged their status century can be traced to levels of e constructed four by the number of corporate headquar- human capital at the turn of the indices to attempt to ters within their borders. Economic century.5 Places with talented people capture the level of diver- developers used financial and other both grow faster and are better able to sity and tolerance within incentives to lure companies to their attract other talented people. Wthe nation’s most populous metropol- communities. Now, however, people Not surprisingly, high-technology itan areas. All indices utilize the 1990 are the center of the action. High metropolitan areas contain more talent U.S. Decennial Census Public Use human capital individuals—or as we than other metropolitan areas. The Microdata Sample:8 like to call them, talent—are the key statistical correlations between the to success in this new era of economic percentage of the population with at Gay Index: This index measures the growth. Their ideas and creativity are least a college education and the over—or under-representation of the most important ingredients in the strength of the high-tech economy gay male couples in a metropolitan economic success of a firm or region. are uniformly high and significant.6 area relative to the population.9 The Firms have always located near their (See Appendix A for a more detailed index is constructed as the fraction key factors of production. In the past, discussion). of gays who live in a metropolitan companies located near raw materials, What, then, brings talented workers area divided by the fraction of the good transportation, or low costs. So, to a particular metropolitan area? How U.S. population who live in that it is not surprising that firms in today’s do they make their residential deci- area. The value takes on the proper- knowledge-based economy are increas- sions? What sets high-technology ties of an odds ratio whereby a value ingly making location decisions based centers such as San Francisco, Boston, over one says that a gay couple is on where the talent pool is located. and New York apart from other more likely to locate in the area According to the management- metropolitan areas? Why have some than the population in general while consulting firm, McKinsey and metropolitan areas—many home to values below one suggest that gays Company, the “war for talent” is the some of the nation’s most prestigious are under-represented. number one competitive issue facing university research centers and college companies in the United States and graduates—been unable to attract a Bohemian Index: Calculated in the around the world, and it remains so significant number of talented tech- same fashion as the gay index, this even though the Internet bubble has nology workers? measures the over- or under-repre- burst.2 As Hewlett Packard CEO Carly Our theory is that a connection sentation of artists and musicians in Fiorina told a conference of governors exists between a metropolitan area’s a metropolitan area. The index is recently: “Keep your tax incentives and level of tolerance for a range of based on the number of writers, highway interchanges, we will go people, its ethnic and social diversity, designers, musicians, actors and where the highly skilled people are.” and its success in attracting talented directors, painters and sculptors, The Nobel prize-winning economist people, including high-technology photographers, and dancers. Robert Lucas argues that the driving workers. People in technology busi- Regions in which these “bohemians” forces in the growth and development nesses are drawn to places known for are over-represented possess a of cities and regions can be found in diversity of thought and open-minded- milieu that favors openness to the productivity gains associated with ness. These places possess what we creativity and artistic expression. the clustering of talented people or refer to as low barriers to entry for human capital.3 Research by Harvard human capital. Diverse, inclusive Foreign-Born Index: The University economist Edward Glaeser communities that welcome gays, percentage of those within a and his collaborators provides ample immigrants, artists, and free thinking metropolitan area who were not empirical evidence of the close associ- “bohemians” are ideal for nurturing born in the United States or one ation between human capital and creativity and innovation, both keys to of its territories. regional economic growth.4 Glaeser success in the new technology. finds considerable empirical evidence This study examines the relation- Composite Diversity Index: This that firms gather in particular regions ship between our measures of diversity index is a sum of the rankings of the to gain advantages from common labor and tolerance and high-technology three individual diversity measures:

June 2001 • The Brookings Institution • Survey Series 2 the gay index, bohemian index, and III. Findings foreign-born index. Table 1: A. The leading indicator of a metro- Tech Ranking We compare our diversity measures politan area’s high-technology and Gay Index. to a measure of high-technology success is a large gay population. industry concentration and growth Gays can be thought of as canaries of Milken Gay developed by the Milken Institute. In a the knowledge economy because they Tech-Pole Metropolitan Index July 1999 study titled America's High- signal a diverse and progressive envi- Ranking Area Ranking Tech Economy, an index called the ronment that fosters the creativity and 1 San Francisco 1 “Tech-Pole” was calculated to rank innovation necessary for success in 2 Boston 8 metropolitan areas (using 1998 data) high tech industry. Gays are frequently 3 Seattle 6 based on a combination of two factors: cited as harbingers of redevelopment 4 Washington D.C. 2 (1) the output of an area’s high-tech and gentrification in distressed urban 5 Dallas 19 industries expressed as a percentage of neighborhoods. Studies also suggest 6 Los Angeles 7 the output of the nation’s high-tech that the presence of gays in a metro- 7 Chicago 15 industries; and (2) a ratio of the politan area provides a barometer for a 8 Atlanta 4 amount of a metropolitan area’s broad spectrum of amenities attractive 9 Phoenix 22 output from high-tech industries to to adults, especially those without 10 New York 14 the amount of the nation’s output children.13 To some extent, the gay and 11 Philadelphia 36 from high-tech industries.10 lesbian population represents what 12 San Diego 5 The first measure favors large might be called the “last frontier” of 13 Denver 10 metropolitan areas; the second favors diversity in our society. 14 Austin 3 small areas with large technology Utilizing 1990 Census data, we 15 Houston 21 sectors. By multiplying them, the constructed a measure of the concen- 36 Cleveland 47 Milken Index creates a measure that tration of gays in a community relative 37 Miami 12 favors neither. The term “tech-pole” to the population in general.14 We call 38 Rochester 13 refers to the relative technology gravi- this measure the Gay Index. Eleven of 39 Albany 30 tational pull that a metropolitan area the top 15 high-tech metropolitan 40 Nashville 28 exerts on high-tech industries. This areas also appear in the top 15 of the 41 Greensboro 46 measure is used throughout the study Gay Index (see Table 1). The five 42 Oklahoma City 27 to compare a metropolitan area’s tech- metropolitan areas with the highest 43 Las Vegas 48 nology prowess with our measures of concentration of gay residents are 44 Norfolk 37 talent, tolerance, and diversity. all among the nation’s top 15 high- 45 Richmond 29 We also compare our measures with technology areas: San Francisco, 46 Buffalo 50 the Milken Institute’s Tech-Growth Washington D.C., Austin, Atlanta, and 47 New Orleans 24 Index.11 This measures growth in San Diego. 48 Honolulu 20 output of high-tech industries within In all of our statistical analyses, the 49 Memphis 33 metropolitan areas from 1990 to 1998 Gay Index does better than other indi- 50 Louisville 42 relative to the national growth rate in vidual measures of social and cultural output of high-tech industries during diversity as a predictor of high-tech Source: Milken Institute “Tech-Pole” the same period. location (see Appendix B). The corre- Ratings; Gay Index constructed by Richard All 50 metropolitan areas, ranked by lations are exceedingly high and Florida and Gary J. Gates using data from our indices and the Milken measures, consistently positive and significant.15 1990 U.S. Decennial Census Public Use are shown in Appendix B. Appendix C The results of a variety of multivariate Microdata Sample (5%) details statistical techniques and regressions analyses back this up. The modeling.12 Gay Index is positively and signifi- cantly associated with both the ability of a region to attract talent and to 1998 rank in the top ten for the Gay generate high-tech industry.16 Index. In addition, the correlation Gays not only predict the concen- between the Gay Index (measured in tration of high-tech industry, they are 1990) and the Milken Tech-Pole Index also a predictor of its growth. Five of calculated for 1990-2000 increases the cities that rank in the top ten for over time, as shown in Figure 1. high-technology growth from 1990 to Figure 1 also suggests that the bene-

June 2001 • The Brookings Institution • Survey Series 3 amenity-rich city “the Entertainment Figure 1. Pearson Correlation of the Milken Tech-Pole Index Machine.”20 The city as an entertain- over time and the 1990 Gay Index. ment machine is defined by a whole new cluster of amenities mainly 0.8000 revolving around a vibrant street level culture—from outdoor cafes and hip 0.7750 restaurants to art galleries and a pulsating music scene. To get at the relationship between 0.7500 creativity and technology, we examine a metropolitan area’s receptiveness to 0.7250 bohemians and artistic creativity. Again, we construct a new measure of 0.7000 our own.21 We call it the Bohemian Index—or the “Boho Index.” The Correlation with Gay Index 0.6750 index is based on the number of writers, designers, musicians, actors 0.6500 and directors, painters and sculptors, photographers, and dancers found in 0.6250 a metropolitan area. We believe this 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 measure offers considerable improve- Milken-Tech Pole Year ment over traditional measures of amenities (such as restaurants, Source: Milken Institute “Tech-Pole” Ratings; Gay Index constructed by Richard Florida and Gary museums, symphonies and the like) in J. Gates using data from 1990 U.S. Decennial Census Public Use Microdata Sample (5%) that it provides a direct measure of the producers of those amenities. Metro- politan areas that are over-represented fits of diversity may actually This increases our confidence in the by these “bohemians” are those with compound over time by increasing a strength of the concentration of gays an appreciation of amenities that region’s high-tech prosperity. as a predictor of high-technology support and showcase creativity and We also examine how the concen- concentration and growth. artistic expression. tration of gays in combination with While our findings on the link So what do we find? There is a other factors affects high-tech growth. between the Gay Index and high-tech strong correlation between our Boho A metropolitan area’s percentage of growth cannot be viewed as conclu- Index and the concentration and gay residents provides the only signifi- sive, the results are quite consistent strength of high-tech industry.22 Ten of cant predictor of high-tech growth in a with our theory that social and the top 15 bohemian metropolitan region when we factor in other cultural diversity attracts talent and areas also number among the nation’s regional characteristics such as talent, stimulates high-tech growth.19 These top 15 high-technology areas, notably foreign-born residents, bohemians, findings support the view that encour- Seattle, Los Angeles, New York, Wash- several measures of lifestyle amenities, aging diversity and lowering barriers to ington D.C., San Francisco, and and population.17 entry can help to attract human Boston (see Table 2).23 Thirteen out of Statistically, one might be capital and generate technology-based the bottom 15 high-tech metropolitan concerned that the influence of San growth. areas fail to appear in the top 15 of Francisco (which ranks unusually the Boho Index. high, first in fact, on both the high- B. High concentrations of creative tech and gay indices) may create a people or “bohemians” are also an C. Metropolitan areas with high false association between the two indicator of a metropolitan area’s concentrations of foreign-born measures. To check for this, we high-technology success. residents rank high as technology removed San Francisco from the data It has long been thought that talented centers. and repeated the analyses. The basic people are attracted to amenities or a The growth and development of great findings remain virtually the same.18 In better quality of life. Terry Clark and cities comes from their ability to fact, the influence of the Gay Index on Richard Lloyd argue that amenities are harness diversity, welcome newcomers, high-tech growth strengthens slightly an increasingly important part of what and turn their energy and ideas into when San Francisco is not included. cities have to offer, dubbing the new innovations and wealth. The Milken

June 2001 • The Brookings Institution • Survey Series 4 highly as regions of technology and Table 2: Table 3: technological growth. Tech Ranking Tech Ranking and Leading high-tech centers are and Boho Index. Foreign-Born Index. places where people from virtually any background can settle and thrive. Milken Boho Foreign- In Silicon Valley, the world’s leading Tech-Pole Metropolitan Index Milken Born high-tech center, nearly a quarter of Ranking Area Ranking Tech-Pole Metropolitan Index the population is foreign born; and 1 San Francisco 8 Ranking Area Ranking almost one-third of the Valley’s high- 2 Boston 9 1 San Francisco 4 tech scientists and engineers hail 3 Seattle 1 2 Boston 9 from foreign countries, according to 4 Washington D.C. 6 3 Seattle 15 research by Annalee Saxenian of the 5 Dallas 15 4 Washington D.C. 10 University of California at Berkeley.25 6 Los Angeles 2 5 Dallas 16 Roughly one-quarter of new Silicon 7 Chicago 20 6 Los Angeles 2 Valley businesses established since 8 Atlanta 13 7 Chicago 7 1980 had a Chinese or Indian-born 9 Phoenix 24 8 Atlanta 31 founder, according to Saxenian’s 10 New York 3 9 Phoenix 21 study. That figure increased to more 11 Philadelphia 35 10 New York 3 than 30 percent between 1995 and 12 San Diego 18 11 Philadelphia 25 1999. And, these figures may actually 13 Denver 14 12 San Diego 6 underestimate the extent of immi- 14 Austin 10 13 Denver 29 grant influence, because firms started 15 Houston 30 14 Austin 19 with a non-Asian co-founder are not 36 Cleveland 47 15 Houston 8 counted. 37 Miami 27 36 Cleveland 22 Economists have long argued that 38 Rochester 31 37 Miami 1 diversity is important to economic 39 Albany 45 38 Rochester 24 performance. Generally speaking, they 40 Nashville 4 39 Albany 30 have been talking about the diversity 41 Greensboro 21 40 Nashville 47 of firms or regional industrial struc- 42 Oklahoma City 49 41 Greensboro 48 tures. John Quigley argues that 43 Las Vegas 11 42 Oklahoma City 36 regional economies benefit from the 44 Norfolk 37 43 Las Vegas 13 location of a diverse set of firms and 45 Richmond 26 44 Norfolk 32 industries.26 In the knowledge- 46 Buffalo 48 45 Richmond 39 economy, ethnic, social, and cultural 47 New Orleans 41 46 Buffalo 27 diversity is likely to be even more 48 Honolulu 17 47 New Orleans 26 important. 49 Memphis 40 48 Honolulu 5 To look more closely at the role of 50 Louisville 33 49 Memphis 46 ethnic diversity in high-technology 50 Louisville 49 concentration, we examine the rela- Source: Milken Institute “Tech-Pole” tionship between high-tech industry Ratings; Bohemian Index constructed by Source: Milken Institute “Tech-Pole” and the percent of a region’s popula- Richard Florida and Gary J. Gates using Ratings; Foreign-Born Index constructed by tion that is foreign born. Nine of the data from 1990 U.S. Decennial Census Richard Florida and Gary J. Gates using top 15 (and eight out of the top ten) Public Use Microdata Sample (5%) data from 1990 U.S. Decennial Census metropolitan areas in terms of foreign- Public Use Microdata Sample (5%) born residents were also among the nation’s top 15 high-technology areas: Los Angeles, New York, San Francisco, Institute identifies immigration as one San Diego, Chicago, Houston, of the two most powerful demographic Boston, Washington D.C., and Seattle trends reshaping the nation’s cities fornia and Texas, as well as Chicago, (see Table 3). Twelve of the bottom and regions. Its list of “Melting Pot Washington D.C., and New York City 15 high-tech areas do not rank in the Metros” ranks the most ethnically are also hot spots of economic top 15 on the Foreign-Born Index diverse regions in the country. Sure growth.24 Not surprisingly, these (although Miami ranked number one). enough, highly ranked regions in Cali- metropolitan areas are also ranked The statistical correlation between

June 2001 • The Brookings Institution • Survey Series 5 percentage foreign-born and high- find that our diversity measure has a technology success was quite strong.27 positive and significant effect on high- Table 4: While the relationship between tech growth from 1990 to 1998 (see Tech Ranking and immigrants and high technology is Appendix C). These results offer Composite Diversity Index. encouraging, it is often said that diver- strong evidence of the importance of sity in high-tech industry is somewhat the combined effects of social, Milken Composite narrow and that it does not include cultural, and ethnic diversity for both Tech-Pole Metropolitan Diversity members of traditional racial minori- high-tech location and growth. Ranking Area Ranking ties, such as African-Americans. To get 1 San Francisco 2 at this issue, we explored the relation- VI. Key Findings and Policy 2 Boston 6 ship between high-tech industry and Implications 3 Seattle 5 the percentage of population that is he bottom line of our analysis 4 Washington D.C. 3 non-white. The results here are frankly is basic: tolerance and diver- 5 Dallas 15 disturbing. We find no significant sity clearly matter to 6 Los Angeles 1 correlation here whatsoever.28 high-technology concentra- 7 Chicago 11 Ttion and growth. Of our individual 8 Atlanta 14 D. Social, cultural, and ethnic measures of diversity and tolerance, 9 Phoenix 21 diversity are strong indicators of a the top factor that correlates with a 10 New York 4 metropolitan area’s high-technology metropolitan area’s high-technology 11 Philadelphia 32 success. success is a large gay population. 12 San Diego 7 Our argument about diversity is simple Other significant factors include high 13 Denver 17 and straightforward. Diversity of concentrations of foreign-born people 14 Austin 8 human capital is a key component of and bohemians. When we use a 15 Houston 18 the ability to attract and retain high- composite measure of overall social, 36 Cleveland 43 technology industry. To demonstrate cultural, and ethnic diversity, the 37 Miami 10 this, we constructed a Composite results are even stronger. 38 Rochester 22 Diversity Index (CDI) measure based We do not mean to imply that these 39 Albany 36 on the three diversity indicators that results prove that a large gay popula- 40 Nashville 25 we discussed earlier—the gay and tion or concentration of bohemians 41 Greensboro 42 bohemian indices and the percent of directly causes the development of a 42 Oklahoma City 39 foreign-born residents. We ranked our technology industry; the theory is that 43 Las Vegas 24 50 metropolitan areas by each indi- people in technology businesses are 44 Norfolk 37 cator (low to high) and summed the drawn to places known for diversity of 45 Richmond 30 three rankings. thought and open-mindedness, and 46 Buffalo 48 The CDI strongly correlates with that our measures potentially get at a 47 New Orleans 27 the Milken Tech-Pole Ranking. As broader concept of diversity and inclu- 48 Honolulu 12 Table 4 demonstrates, the top 11 siveness. 49 Memphis 44 metro areas on the CDI are also In this study, six metropolitan areas 50 Louisville 47 among the top 15 Milken Tech-Pole ranked in the top 15 for each of our regions. The statistical correlation three individual indices of diversity. Source: Milken Institute “Tech-Pole” between the Milken Tech-Pole rank- Four of these six metro areas were the Ratings; Composite Diversity Index ings and CDI rankings is higher than top ranked high-tech regions on the constructed by Richard Florida and Gary J. the same correlation with rankings by Milken Index—San Francisco, Boston, Gates using data from 1990 U.S. Decennial any of our individual diversity meas- Seattle, and Washington D.C. The two Census Public Use Microdata Sample (5%) ures, or by simpler measures such as remaining metros—Los Angeles and the percentage of college graduates in New York—also ranked in the top ten the population.29 on the Milken Index. Even more compelling, the CDI While more research is certainly strongly predicts high-tech growth. required to get at the causal linkages human capital helps to attract talent When we estimate the effect of the between diversity and high-tech and that talent is in turn associated CDI on high-tech growth and factor in industry, we feel confident in the with high-technology industry and the percentage of college graduates in nature of the relationship our analysis regional growth. We certainly welcome the region, population, and measures has uncovered. We are convinced that more research designed to address this of culture, recreation, and climate, we tolerance and low entry barriers to important issue.

June 2001 • The Brookings Institution • Survey Series 6 The basic message to city leaders That is a message other cities and and economic developers is clear. regions should heed. Talent powers Tech Ranking Talented people go to places that have economic growth, and diversity and and Talent Index. thick labor markets, are open and openness attract talent. Companies tolerant, and offer a quality of life they remain important, but no longer call Milken Talent desire. Places that attract people the shots. The location decisions of Tech-Pole Metropolitan Index attract companies and generate new people are just as important—poten- Ranking Area Ranking innovations, and this leads to a tially more important—than those of 1 San Francisco 3 virtuous circle of economic growth. firms. In fact, companies increasingly 2 Boston 2 Cities must begin to combine their will go where talented people are 3 Seattle 6 goal of providing a better business located. 4 Washington D.C. 1 environment with strategies aimed at Building a vibrant technology-based 5 Dallas 10 improving their diversity and toler- region requires more than just 6 Los Angeles 23 ance. investing in R&D, supporting entre- 7 Chicago 13 Austin is at the cusp of an inte- preneurship, and generating venture 8 Atlanta 5 grated approach to capture capital. It requires creating lifestyle 9 Phoenix 35 high-technology talent. The region has options that attract talented people, 10 New York 9 made significant investments in and supporting diversity and low entry 11 Philadelphia 20 research and development, higher barriers to human capital. These 12 San Diego 14 education, and business incubation attributes make a city a place where 13 Denver 7 with major facilities like Sematech and talented people from varied back- 14 Austin 4 MCC. But alongside this, the region grounds want to live and are able to 15 Houston 12 has also made considerable invest- pursue the kind of life they desire. 36 Cleveland 42 ments in its lifestyle and music 37 Miami 46 scene—from the clubs and bars of Appendix A: The Talent Index 38 Rochester 19 Sixth Street to Austin City Limits and 39 Albany 15 the South-by-Southwest film and welve of the top 15 metropol- 40 Nashville 29 music festival. When a high-tech itan areas in the Tech-Pole 41 Greensboro 41 company, Vignette, recently expanded Index ranking are among the 42 Oklahoma City 27 into a new facility in downtown top 15 areas with the highest 43 Las Vegas 50 Austin, a part of its deal was to estab- Tconcentration of talent. San Fran- 44 Norfolk 45 lish a million-dollar fund to support cisco, Boston, Washington D.C., and 45 Richmond 11 the local music scene. Atlanta are among the top cities for 46 Buffalo 44 Austin Mayor Kirk Watson has been technology concentration and high 47 New Orleans 34 the driving force between a powerful human capital. Seven areas (Cleve- 48 Honolulu 16 and progressive strategy that aims to land, Miami, Greensboro, Las Vegas, 49 Memphis 26 capitalize on the “convergence” Norfolk, Buffalo, and Louisville) that 50 Louisville 47 between technology, talent, and toler- rank in the bottom 15 of our Talent ance, as he puts it. Watson was Index also rank in the bottom 15 of Source: Milken Institute “Tech-Pole” recently re-elected with 84 percent of the Tech-Pole rankings. Our estimates Ratings; Talent Index constructed by the vote. “Austin has benefited from a suggest that a 1 percent increase in Richard Florida and Gary J. Gates using convergence between technology and the percentage of college graduates in data from 1990 U.S. Decennial Census our laid back, progressive, creative, a metropolitan area would result in an Public Use Microdata Sample (5%) lifestyle and music scene,” he says. increase of its Tech-Pole ranking by “The key is that we continue to nearly two places. Using the Milken preserve the lifestyle and diversity, Institute’s Tech-Growth Index, we also which enables us to lure companies examine the relationship between and people from places like Silicon talent and the growth of high-tech- Valley. We are building the habitat to nology industry. The results suggest do that.” that talent is associated with growth of high-technology industry, but not as strongly as its association with the Tech-Pole index.30

June 2001 • The Brookings Institution • Survey Series 7 Appendix B: Rankings of Top 50 Metropolitan Areas by Various Indices Rankings Milken Foreign Talent Index Milken Tech-Growth Composite Gay Born Bohemian % College Metropolitan Area Tech-Pole Index Diversity Index Index Index Degree San Francisco 1 10 21483 Boston 23668992 Seattle 3 20 5 6 15 1 6 Washington D.C. 4 24 3 2 10 6 1 Dallas 5 91519161510 Los Angeles 650172223 Chicago 7 13 11 15 7 20 13 Atlanta 8 5 14 4 31 13 5 Phoenix 9 3 21 22 21 24 35 New York 10 37 4 14 3 3 9 Philadelphia 11 27 32 36 25 35 20 San Diego 12 25 7 5 6 18 14 Denver 138171029147 Austin 14 1 8 3 19 10 4 Houston 15 7 18 21 8 30 12 Portland 16 2 16 23 23 5 17 Indianapolis 17 40 40 34 44 34 31 Kansas City 18 11 34 35 42 22 24 Minneapolis 19 29 19 17 35 7 8 St. Louis 20 45 49 45 43 38 36 Orlando 21 38 9 11 17 12 32 Sacramento 22 6 20 9 14 39 22 Detroit 23 33 26 44 20 25 48 San Antonio 24 4 31 32 12 50 40 Pittsburgh 25 26 46 39 37 46 39 West Palm Beach 26 34 13 16 11 16 33 Tampa 27182318183249 Columbus 28 28 29 25 38 29 21 Salt Lake City 29 19 28 41 28 23 25 Birmingham 30 35 50 49 50 42 38 Baltimore 31 44 38 31 33 44 18 Cincinnati 32 46 33 38 40 19 28 Charlotte 33 12 45 43 41 36 37 Dayton 34 39 41 26 45 43 43 Milwaukee 35 43 35 40 34 28 30 Cleveland 36 49 43 47 22 47 42 Miami 3730101212746 Rochester 38 31 22 13 24 31 19 Albany 39 41 36 30 30 45 15 Nashville 40 17 25 28 47 4 29 Greensboro 41 14 42 46 48 21 41 Oklahoma City 42 42 39 27 36 49 27 Las Vegas 43 21 24 48 13 11 50 Norfolk 44 15 37 37 32 37 45 Richmond 45 22 30 29 39 26 11 Buffalo 46 48 48 50 27 48 44 New Orleans 47 32 27 24 26 41 34 Honolulu 48 47 12 20 5 17 16 Memphis 49234433464026 Louisville 50 16 47 42 49 33 47 Appendix C: Statistics

orrelations utilized both a statistic measures the correlation Pearson and a Spearman between the relative rankings of the rank order correlation. The two variables. As such, the Spearman Pearson statistic measures correlation tends to be less influenced Cthe correlation between the values of by outliers in the data. All regression two variables while the Spearman analysis utilized Ordinary Least Squares estimation techniques:

Correlations Utilized in the Analysis

Pearson Correlations Between Milken Tech-Pole and Growth Measures with Talent and Diversity Measures.a

Composite High-Tech % College Diversity Boho % Foreign Gay Growth Graduates Indicator Index Born Index Milken Tech- 0.23 0.72 0.68 0.62 0.43 0.77 Pole (0.10) (0.001) (0.001) (0.001) (0.002) (0.001) High-Tech - 0.25 0.27 0.19 -0.04 0.31 Growth - (0.08) (0.06) (0.19) (0.78) (0.03) aSignificance level shown in parenthesis. Bold cells are significant at 0.10 level or higher.

Spearman Rank Order Correlations Between Milken Tech-Pole and Growth Measures with Talent and Diversity Measures.a Composite High-Tech % College Diversity Boho % Foreign Gay Growth Graduates Indicator Index Born Index Milken Tech-Pole 0.30 0.60 0.63 0.54 0.48 0.60 (0.03) (0.001) (0.001) (0.001) (0.001) (0.001) High-Tech Growth - 0.20 0.23 0.24 0.07 0.26 (0.15) (0.10) (0.09) (0.63) (0.07) aSignificance level shown in parenthesis. Bold cells are significant at 0.10 level or higher.

Regression Estimation Utilized in the Analysis

OLS Estimation of the effects of various metropolitan area traits on the Milken Tech-Pole Index and Ranking.a Independent variables: (1) (2) (3) (4) Dependent Variable: Dependent Variable: Dependent Variable: Dependent Variable: Milken Tech-Pole Milken Tech-Pole Ranking Milken Tech-Pole Milken Tech-Pole % College graduates 27.2* 1.9* 21.5* (7.25) (5.46) (4.99) Bohemian Index 4.4* 2.3** (5.43) (2.90) R-squared 0.52 0.38 0.38 0.59 N=50 aAbsolute value of t-statistics are shown in parentheses. *Significance < 0.001 level. **Significance < 0.01 level.

June 2001 • The Brookings Institution • Survey Series 9 160. Glaesar, E.L. 1997. “Learning in OLS Estimation of the effects of various metropolitan area Cities,” NBER working paper, 6271. traits (measured in 1990) on high-technology growth from Glaeser, E.L., J.A. Sheinkman, and 1990 to 1998.a A. Sheifer, 1995. “Economic Growth in a Cross-Section of Cities.” Journal of Independent variables: Dependent Variable: Monetary Economics, 36, 117-143. High-tech growth Composite diversity index 0.005*** - 5 Glendon, Spencer, 1998. “Urban Life (2.91) Cycles,” working paper, Harvard University. Gay Index - 0.15* (1.70) 6 The analysis indicated a Pearson correla- % College Graduate -0.41 -0.24 tion between the Milken Tech Pole and (0.29) (0.15) the population with a college degree at % Foreign Born - -0.28 0.72. The Spearman rank order correla- (0.34) tion was 0.60—that is the relationship Bohemian Index - 0.26 between the rank number of a region on (1.23) the Tech Pole and the rank on the Culture -.00002 -0.00005 percentage of the population with a BA or (0.27) (0.69) above. Interestingly, talent explains more Recreation -0.0002** -0.0001 than 50% percent of the variation in high- (2.56) (1.54) technology concentration. Climate -0.0001 -0.0004 (0.27) (1.13) 7 We combine any MSAs that are also Population 0.00 0.00 part of a Consolidated Metropolitan (0.80) (0.001) Statistical Area (CMSA) as defined by R-squared 0.32 0.28 the U.S. Census Bureau. As a result, N=50 the following areas are constituted as a aAbsolute value of t-statistics are shown in parentheses. single metropolitan area: *Significance < 0.10 level. San Francisco: San Francisco, **Significance < 0.05 level. Oakland, San Jose ***Significance < 0.01 level. Los Angeles: Los Angeles, Anaheim, Riverside Miami: Miami, Fort Lauderdale Endnotes tion—capital, people and land –and land New York: New York, Bergen County, is always far cheaper outside cities than Newark, Middlesex County, Nassau 1 Richard Florida is Heinz Professor of inside… It seems to me that the ‘force’ we County, Suffolk County, Monmouth Regional Economic Development and need to postulate account for the central County Director of the Software Industry Center role of cities in economic life is of exactly at Carnegie Mellon University. Gary Gates the same character as the ‘external human 8 The Census Bureau releases two public is a Research Associate at the Urban Insti- capital’…What can people be paying use samples of decennial census person- tute in Washington D.C. Manhattan or downtown Chicago rents level data, the 1 percent and 5 percent for, if not for being near other people?” PUMS. The 5 percent PUMS is a repre- 2 Chambers, Elizabeth G., et al, 1998. “The sentative sample (3 in 10) drawn from the War for Talent,” The McKinsey Quarterly. 4 Glaeser, E.L., 2000. “The New Economics 1 in 6 sample of people who filled out a Number 3. of Urban and Regional Growth”. In census long-form, which amounts to 5 Gordon Clark, Meric Gertler, and percent of the U.S. population or approxi- 3 Lucas, Robert E., 1998. “On the Maryann Feldmen (eds). The Oxford mately 12 million observations. The Mechanics of Economic Development,” Handbook of Economic Geography. Oxford: sample is representative of the entire U.S. Journal of Monetary Economics, p. 38-9. Oxford University Press , 83-98. Glaeser, population and also is considered repre- Lucas says: “If we postulate only the usual E.L. 1999. “The Future of Urban sentative within metropolitan statistical list of economic forces, cities should fly Research: Non-Market Interactions,” areas, the unit of analysis utilized for this apart. The theory of production contains working paper, Harvard University. work. nothing to hold a city together. A city is Glaeser, E.L. 1998. “Are Cities Dying?” simply a collection of factors of produc- Journal of Economic Perspectives, 12: 139-

June 2001 • The Brookings Institution • Survey Series 10 9 Black, Dan, Gary Gates, Seth Sanders, 14 Several cautions must be noted regarding 18 The only qualitative difference was that and Lowell Taylor. 2000 “Demographics of the Census data and gays and lesbians. We the Pearson correlation between percent the Gay and Lesbian Population in the are measuring only individuals in same-sex college graduate and the Tech-Pole index United States: Evidence from Available unmarried partner relationships. As such, was slightly higher than the same correla- Systematic Data Sources,” Demography these figures do not take into account tion with the gay index. However, the 37:2, 139-154. They demonstrate that non-partnered gays. In addition, we esti- Spearman rank order correlation with the unmarried same-sex partners in the mate that the Census only captured Tech-Pole index was higher for the gay Census are comprised primarily of gay and approximately 35% of all gay/lesbian part- index. lesbian couples. nerships (see Black, Gates, Sanders, and Taylor, 2000). 19 One question raised by this strong connec- 10 DeVol, Ross C., Perry Wong, John tion between gays and high-technology is Catapano, and Greg Robitshek, 1999. 15 The Pearson correlation between the Gay the extent to which gays and lesbians are “America’s High-Tech Economy: Growth, Index and the Tech-Pole index was 0.77 over-represented in the industry. If gays Development, and Risks for Metropolitan while the Spearman rank-order correlation and lesbians make up large fractions of Areas,” Milken Institute. was 0.60 (both are significant at the 0.001 this industry, then it could be that the level). The growth index measures change location of high-technology firms brings 11 DeVol et al (1999.) in high-tech output within metropolitan about a larger concentration of gays in a areas from 1990 to 1998 relative to region. To look at this, we analyzed 1990 12 In addition to statistical research on the national change in output during the same Census data to assess the extent to which 50 metropolitan areas, the study was period. gays and lesbians are over-represented in informed by interviews, focus groups and some high-technology fields and indus- case studies the authors have conducted 16 The high correlation observed between the tries. Gay men are about 1.3 times more separately or together over the past several gay concentration and the high-technology likely to be scientists and engineers than years. index could in part be a result of a limited the population in general. Lesbians are as sample of the 50 largest metropolitan likely as the rest of the population to be in 13 There are a number of reasons why this areas in the country. Therefore, further these occupations. If the gay men and conclusion makes sense. Gay males are studies examined if this positive associa- lesbians are combined, the result shows less likely to have children. In addition, if tion remains when a broader cross-section that they are 1.2 times more likely than they couple with another male, their of metropolitan areas is utilized. Indeed, the population to be scientists and engi- household income will be on average while the correlation does weaken when neers. We also examined those employed higher than the income of male/female 242 metropolitan statistical areas are in the computer and data processing serv- households since males on average earn examined, it is still positive and significant ices industry. Gay men are 2.3 times and more than females in our economy. With at 0.23. lesbians are 1.3 times more likely than the no children in the household and relatively population to be employed in this industry. high incomes, gay couples can devote 17 The overall F-statistic for the model is also Together, gays and lesbians are 1.9 times larger portions of their income to the only significant at the 0.07 level. This more likely than the population to be purchase and development of amenities. means that the total predictive power of employed in the computer services See: Black, Dan, Gary Gates, Seth the combined variables is significant, but industry. While some of the correlation Sanders, Lowell Taylor, 2001. “Why Do somewhat weak. There is also a strong between gays and high-technology might Gay Men Live in San Francisco,” forth- relationship between the concentration of result from their over-representation in coming, Journal of Urban Economics. gays in a metropolitan area and other the industry, it seems difficult to explain Research has shown that gay male couples measures of diversity, notably the percent how their over-representation would live in some of America’s most sought after of foreign-born residents: San Francisco, predict growth. To do so would be to urban areas and are more likely to live in Washington D.C., and Los Angeles all suggest that that gays and lesbians are distressed areas within cities and gentrify ranked highly. These results reiterate the somehow on average more productive or their surroundings more than any other impact of diversity on a broad section of entrepreneurial than their heterosexual household type. See: Gates, Gary, 2000. society. Low barriers to entry seem to be counterparts. “Essays on the Demographics and Loca- good for the whole economy, not just high- tion Patterns of Coupled Gay Men,” skilled labor. 20 Clark, Terry, and Richard Lloyd, 2000. Doctoral dissertation, H. John Heinz III “The Entertainment Machine,” University School of Public Policy and Management, of Chicago. Carnegie Mellon University.

June 2001 • The Brookings Institution • Survey Series 11 21 Originally, we undertook a series of 25 Saxenian, Annalee, 1999. “Silicon Valley’s analyses using various measures of ameni- New Immigrant Entrepreneurs,” Public Acknowledgements: ties. We looked at things like climate, and Policy Institute of California. professional sports, and arts and culture, The Brookings Institution and many others. In most of this work, we 26 Quigley, John Q., 1998. “Urban Diversity Center on Urban and Metropol- found only a loose relationship between and Economic Growth.” Journal of itan Policy would like to thank amenities, talent, and technology. A large Economic Perspectives. 12:2, 127-138. Surdna Foundation, Inc. and part of the reason, we came to believe, was The Fannie Mae Foundation due to the measures themselves. They 27 The Pearson correlation between percent for their generous support were based on combinations of different foreign born and the Tech-Pole index was of our work on central city factors and were not very consistent or 0.43 (significant at the 0.01 level) and the competitiveness. reliable. And many of the things we would Spearman rank-order correlation with have liked to measure—such as a city’s 0.48 (significant at the 0.001 level). We Elizabeth Currid, Sam Youl Lee, music or art scene were simply unavail- did not find a similar correlation with and Ji Woong Yoon assisted with able. high-technology growth. this article. The authors bene- fited from discussions and 22 The two measures are correlated at 0.62 28 The Pearson correlation was –0.11 and helpful comments provided by using a Pearson correlation and 0.54 using the Spearman rank-order correlation was Ashish Arora, Mark Kamlet, and a Spearman rank-order correlation. Both –0.20. Neither were statistically signifi- Ben Margolis. are significant at the 0.001 level. cant.

23 While an association between the 29 The Spearman rank order correlation Bohemian index and the Tech-Pole index between the Milken Tech-Pole and the is observed, a similar connection between composite diversity measure was 0.63. See growth and this index is not noted. Alone, Appendix C for other correlations. the Boho Index can explain nearly 38% of For More Information: the variation in high-tech concentration. 30 The Pearson correlation between the Richard Florida, Ph.D The combination of the Boho Index and percentage of college graduates in a region H. John Heinz III Professor of our measure of talent account for nearly in 1990 and growth in high-technology Regional Economic 60% of the high-tech concentration from 1990 to 1998 was 0.25 (significant at Development measure. the 0.10 level). Heinz School of Public Management 24 Frey, William, and Ross DeVol, 2000. Carnegie Mellon University “America’s Demography in the New Pittsburgh, Pennsylvania 15213 Century: Aging Baby Boomers and New Phone: (412) 268-8784 Immigrants as Major Players,” Milken Email: [email protected] Institute. Gary J. Gates Research Associate The Urban Institute 2100 M Street, N.W. Washington,, D.C. 20037 Phone: 202-261-5750 Email: [email protected]

For General Information: Brookings Institution Center on Urban and Metropolitan Policy Phone: (202) 797-6139 Website: www.brookings.edu/urban

June 2001 • The Brookings Institution • Survey Series 12