2. Urban growth and human capital: A review ofconcepts and theory

View the most recent version.

Large scale statistical research on the relationship between the educational attainment of city residents and city growth is relatively recent.¹ It is based on the fundamental idea that creativity drives growth and that the geographic concentration of more highly skilled workers creates more ideas that can then be transmitted more rapidly. As a result, an urban economy with a more highly educated workforce is expected to have higher levels of productivity and productivity growth.

Much of the literature that links educational attainment with growth uses the term 'human capital' as shorthand. This is based, implicitly or explicitly, on the premise that human capital is highly correlated with higher levels of education. Human capital can also result from experience, and this is not accounted for in this formulation. We should also note that some readers may find the term 'human capital' off-putting, because of its implied instrumental treatment of human beings. Nevertheless, we will adopt the term human capital because of its accepted meaning within the literature on which this paper relies.²

The links between human capital and growth have been formalized by Glaeser and others.³ These urban growth models suggest labour growth is a function of the change in the level of (multifactor) productivity of a city, which, in turn, depends on a set of city-specific characteristics that are treated as initial conditions. Productivity growth, it is argued, stems from both static and dynamic sources. Static sources refer to urban characteristics whose influence on productivity rises over time. For instance, rising returns to education imply the contribution of education to productivity has been rising over time. Dynamic sources are related to urban characteristics that touch off a set of productivity enhancing series of events. Jacob's (1969) contention that industrial diversity leads to the spreading application of new ideas across industries is an example of a dynamic source of productivity growth. In practise, no attempt is made to empirically disentangle these two sources of productivity growth other than through the identification of variables that are thought to be more associated with one than the other.

Much of this literature has focused on the role of human capital as a driver of long-run employment (or population) growth. Glaeser (1994), Glaeser, Sheinkmen and Sheifer (1995), and Glaeser and Saiz (2003) have found that long-run urban growth is positively associated with initial skill endowments. These results have been shown to be robust to endogeneity bias (see Glaeser and Saiz 2003, and Shapiro 2005). This link between human capital and economic growth has been corroborated in other studies (see Glaeser 2000 and Florida 2002b for discussion).

These studies often utilize fairly broad measures of human capital, such as the percentage of an urban population that has attained postsecondary degrees. This reflects a general, and seemingly uncontroversial, view that a diverse mixture of skills and competencies is required to support the growth process. Locations obtain these skills and competencies by having large numbers of educated residents at work in broad industrial and occupational cross sections of the local economy. A potential contribution of Florida's recent work on human capital (2002a and 2002b) rests with its parallel emphasis on more restrictive measures of human capital, such as 'bohemians' and 'creative classes,' that comprise more narrowly defined urban subpopulations. Florida's work has led to much speculation over whether some types of human capital are more important for urban development than others, and how different types of human capital interact. A widely cited aspect of Florida's research is the link between technological and cultural creativity—that the strength of a city's cultural and artistic environment is positively correlated with emergence of a high-technology base (Florida 2002a, 2002b).

This paper makes two contributions to the literature on growth and human capital. First, our analysis of the growth process centres on disentangling the relative contributions made by broad and specialized forms of skilled labour. Recent empirical research (see Glaeser and Saiz 2003) has evaluated the importance of skilled workers writ large, by relying on measures of educational attainment to differentiate between cities with higher- and lower-quality workforces. Florida's recent work has delved into this more deeply, effectively by asking whether certain varieties of human capital are more important for growth than others.

In this paper, we are particularly interested in the role of scientists and engineers as drivers of urban growth. This group of workers is seen by many as essential to innovation. For instance, the Progressive Policy Institute begins its analysis of changes in engineering and science employment by noting that "…[t]echnological innovation is one of the key drivers of overall economic progress, and it is fueled by a strong engineering and scientific workforce." (Atkinson and Court, 1998: 41). Similarly, the National Science Foundation notes that scientists and engineers "contribute enormously to technological innovation and economic growth, research, and increased knowledge." (National Science Board, 2004: Chapters 3, 5) Here we are interested in directly testing the link between scientists and engineers and urban employment growth.

To our knowledge, relatively little work has been done on evaluating how different forms of human capital complement one another. Our analysis of urban growth evaluates whether the contribution that different types of specialized labour make to the growth process depends on their interaction with other skilled workers. This is consistent with the view that a diverse mix of complementary skills is required for growth (Jacobs 1969). While a seemingly uncontroversial premise, these interaction effects are understudied.

If it should turn out that scientists and engineers are important drivers of growth, then we are naturally interested in what factors attract them to particular cities. That is, we are interested in the identification of the (spatially) disequilibriating forces that will advantage one particular location over another.

These forces can take the shape of pecuniary and non-pecuniary incentives. Pecuniary incentives likely relate to an economic shift that results in an advantage of one location over another. For instance, if a mix of labour emerges in a city that raises the productivity of scientists and engineers, market wages of scientists and engineers will rise in response, resulting in an inflow of scientists and engineers to take advantage of these wage increases. This wage rise may or may not be temporary, depending on the elasticity of their labour supply curve.

Alternatively, incentives can be non-pecuniary and these relate largely to amenities. Florida (2002a, b) identifies amenities as a key factor that attracts the 'creative class,' of which scientists and engineers are an important part, to particular cities. Empirically, amenities have also been identified by others as important factors driving growth in cities (Rappapore 2006, and Glaeser, Kolko and Saiz 2001). Amenities can take the form of cultural activities, climate or, as Florida argues, a social environment that has low barriers to entry for newcomers with diverse lifestyles. Regardless of the source, these amenities can act as a source of disequilibrium, even if they are fixed. For instance, Rappapore argues climate may be rising in importance as incomes rise and climate becomes relatively more important compared with other sources of utility. We estimate a partial adjustment model to account for these disequilibrating forces (see Chapter 5).

1. For a more extensive review of the literature on human capital and urban development, see Florida (2002b).

2. We considered using Florida's term 'talent', but decided it too was flawed as it implied, in the context of this paper, that those with degrees had talent and those without somehow did not.

3. See Glaeser (2000) for a succinct discussion of these models.