The internal structure of cities: the economics of agglomeration, amenities and accessibility

[this is the summary of my dissertation; click here for the Dutch summary]



We have seen a vast increase in the number of people residing in urbanised areas in the last two centuries. It rose from about 7.5 percent in 1800 to more than 50 percent nowadays (Huriot and Thisse, 2000; Glaeser and Gottlieb, 2009; Glaeser, 2011).  The number of urban megapolises is steadily rising: cities like New York, London, Tokyo, Mumbai and Rio de Janeiro become an increasingly common sight. The foremost question for urban economists is then why cities exist and why they keep emerging. This intellectual quest has been pursued for already more than a century by a wide variety of scholars. Marshall (1890) was the first to argue that the most important reason why firms agglomerate is a reduction in transport costs. Firms can lower the costs of moving goods, people and ideas by agglomerating in an urban area (Glaeser, 1998; Glaeser and Gottlieb, 2009; Ellison et al., 2010). More specifically, firms may locate near suppliers and customers to save shipping costs. In the nineteenth century many cities have evolved around transport nodes like railway stations and waterways. For example, the industrial concentration in Manhattan’s midtown was mainly around the central station (Glaeser, 2008). Firms may lower the costs of moving people by locating in a city, because large urban area often implies a thick labour market. Large labour markets like Los Angeles and Paris are offering a wide array of different jobs and give workers the opportunity to specialise. For specialised workers it is easier to switch between employers and for employers it is easier to find specialised workers. A third source of agglomeration economies is the exchange of ideas. Firms learn from each other and increase the rate of innovation. A classical example is Silicon Valley’s computer cluster where the ICT revolution started. Fallick et al. (2006) show that employees in Silicon Valley often switch between jobs, which increases the likelihood that knowledge acquired in one firm is used in another. Marshall (1890) suggests that these external economies are particularly important for firms of the same sector. Jane Jacobs (1969) argues that especially diversified urban environments may stimulate the exchange of ideas between different firms leading to new product combinations. Hoover (1936; 1937) already made the distinction between urbanisation economies, which are external economies occurring between firms irrespective of sector, and localisation economies, occurring between firms of the same sector.

Although the benefits of reducing transportation costs of goods are at the heart of the new economic geography models (Krugman, 1991a; b), it may be argued that this agglomeration force is not so important anymore, as the costs of transporting physical goods have declined with more than 90 percent in the last century (Glaeser and Kohlhase, 2004). Nevertheless, the agglomeration advantages that are based on reducing transport costs for people and ideas remain and are thought to be the main reason for the growth of cities like San Francisco and Tokyo (Glaeser, 2008). Indeed, it has been argued that face-to-face contacts have become the most fundamental force of agglomeration (Storper and Venables, 2008). Deal-making, relationship adjustment, finding a new employer and exchanging ideas are all heavily dependent on face-to-face contacts.

It has been widely recognised that the raison d’être of contemporary cities cannot only be explained by productivity advantages. Urban economists consider cities also as centres of consumption (Roback, 1982; Glaeser and Gottlieb, 2006). Dense urban areas generate a critical mass for the provision of many consumption amenities like theatres, cinemas and cafés and restaurants, and guarantee a wide diversity of consumer goods (Glaeser et al., 2001). For example, the theatre district in London’s West End could only evolve because of the enormous number of potential customers. In the future, as people seem to become increasingly rich, the quality of life is likely to become an even more important determinant of location choices of households.

Cities not only generate positive externalities for firms and its inhabitants, a concentration of people and firms may also lead to negative externalities. The largest metropolitan areas in the world are for example all plagued by traffic congestion. Scholars and policy makers therefore have offered a plethora of solutions to these problems. For example, policies may opt for road expansion or congestion pricing (Small, 1992; Verhoef et al., 1996; Hymel, 2007). Another solution is suggested by urban planners (e.g. Cervero, 2004). They argue that one should aim for urban development around public transport nodes, e.g. railway stations. It is then argued that a good access to public transport and mixed land use around nodes will reduce traffic congestion and increase the liveability of neighbourhoods. Advocates of this ‘smart growth’ movement claim economic benefits, but the benefits of transit-oriented mixed-use development have not been well understood (Jenks et al., 1996; Burton, 2000). Crowding may also imply negative externalities in the form of air and noise pollution and higher crime rates (Glaeser and Sacerdote, 2006; Bajari et al., 2012). Although the latter issues are potentially relevant and interesting, we will not study them in this dissertation.



Each chapter in this thesis analyses at least one of these three economic forces (agglomeration, amenities, accessibility) that shape the urban spatial structure of Dutch cities. In Chapters 2 and 3 we pay attention to agglomeration economies. It is shown that doubling of agglomeration leads to an increase in commercial rents of about 3.5-10 percent. It appears that especially retailers and local governments prefer to locate in the dense parts of cities. Chapter 4 analyses high rents in extremely tall buildings. We hypothesise that high rents in tall buildings are caused by the possibilities to interact with other workers in the building (within-building agglomeration economies) and a landmark effect that is only relevant for the tallest buildings. Conditional on location attributes, we find that buildings that are 10 meters taller are about 4.2 percent more expensive, so within-building agglomeration economies seem to be very important. We also show that the landmark effect is economically significant: it is about 2.8-5.5 percent of the rent for a building that is 5 times the average height (about 140 meters in the Netherlands). We analyse the tendency of business services and multinational enterprise to co-agglomerate within cities in Chapter 5. For example, many headquarters of multinational corporations as well as lawyers, accountants and consultants are located in and near the Amsterdam South-axis. We show that business services tend to co-agglomerate with multinational enterprises in the northern part of the Randstad. In addition, we present empirical evidence that multinationals stimulate the births of new business services firms because of agglomeration economies.

The second part of this dissertation analyses location choices of households. Chapter 6 uses house prices to proxy land prices in the city of Rotterdam. We then test the implications of the theory by Lucas and Rossi-Hansberg (2002), who developed a model where the urban spatial structure is determined by the external benefits of agglomeration and the commuting costs for workers. We show that in mixed urban areas, agglomeration is an important determinant of the rent, while in predominantly residential areas accessibility significantly impacts rents. This is in line with the theory of Lucas and Rossi-Hansberg (2002). Chapter 7 analyses mixed land use in a more sophisticated way. We show that households are willing to pay about 2.5 percent more for a house in a mixed neighbourhood compared to a house in a monofunctional neighbourhood. Again, we analyse heterogeneity and it appears that especially apartment occupiers tend to prefer mixed neighbourhoods, likely because they are willing to pay more for urban amenities. In Chapter 8 we focus on historic amenities and how these amenities cause sorting of households within cities. We show that there is a (conditional) price difference at the conservation boundary of about 3 percent, which we interpret as the willingness to pay for external historic amenities. Listed buildings are about 6 percent more expensive, which we interpret as the willingness to pay for internal historic amenities. Moreover, it is shown that rich households sort themselves in conservation areas and in listed buildings, because they have a higher willingness to pay for historic amenities. The results contribute to an explanation for the substantial spatial income differences within cities. We analyse policies to protect historic amenities in Chapter 9. It is argued that these policies may have adverse side-effects as they limit the possibilities of house owners to change their houses. We show that regulatory costs may be as large as 10 percent of the house price. So, house owners’ benefits due to protection should be substantial to compensate for the regulatory costs. In Chapter 10 we analyse the effects of improved access to public transport on house prices. We do not find a house price increase due to station openings. We also show that in other chapters public transport accessibility seems of limited importance in the location choices of firms and households.

To summarise, agglomeration, amenities and accessibility are all found to be relevant economic forces that shape the urban spatial structure. It is also shown that agglomeration and the presence of amenities seem to be more important than accessibility. Despite the vast decrease in transportation costs of goods in the last century and the dramatic reduction in communication costs, agglomeration economies still seem to be a main driver of firms’ location choices. This is likely caused by the need for costly face-to-face contacts that are needed in deal-making, relationship adjustments and exchanging ideas. Consumer amenities are likely to become an even more important driver of the urban economy: as households tend to become increasingly rich, it may be expected that the quality of life becomes the backbone of location choices of households within cities. Likely because of the (relatively) superb accessibility of many Dutch cities by car and public transport, we find limited effects of increased public transport accessibility.


Policy implications

Our results also have several implications for place-based policies. We start with the evidence on agglomeration economies. A vast number of subsidies is given to attract firms to certain regions. An example is the policy to attract and maintain headquarters of multinational enterprises in the Netherlands. These policies are justified if there are substantial agglomeration externalities, so that the money invested to attract one firm will have substantial wider economic benefits. Our results in Chapters 2, 3, 4 and 5 show that agglomeration economies are important. This may suggest that policies that aim to attract firms improve welfare because of agglomeration economies, but this conclusion is difficult to verify: i.e. if a certain municipality attracts a certain firm, another municipality will lose that firm and the related benefits. Nevertheless, we can learn something from the evidence on agglomeration economies. First, in Chapter 3 we showed that agglomeration economies seem to be much more pronounced in urban areas. This suggests that subsidies aiming at attracting firms to deprived rural density may be inefficient because positive agglomeration externalities are absent in these places. Second, Chapter 4 shows that within-building agglomeration economies are likely important. This suggests that if municipalities should stimulate construction, they should encourage the construction of relatively tall buildings. Municipalities should also reconsider and relax limitations on building heights. Third, in Chapter 5 we highlight that policies that aim at attracting multinational enterprises may yield external benefits through the births of knowledge intensive business services, although the effect is relatively small. Fourth, we show in Chapters 2 and 7 that household density is generally valued negatively, likely because of negative crowding effects (e.g. less access to open space). This is bad news for planning policies, such as Transit Oriented Development and Smart Growth, which aim at increasing (household) densities near nodes of public transports.

Our evidence on the importance of amenities also leads to policy recommendations. First, from Chapter 8 it may be concluded that the physical side of cities, and therefore history, leads to sorting of households over space. Policy makers should be aware of the fact that long-term national policies that stimulate preservation of historical buildings in specific cities may have large spatial effects and may cause problems related to social segregation, as high income households are attracted disproportionally by historic amenities. Chapter 9 also highlights that the house owners’ costs of regulatory restrictions may be large. Policies aiming at protecting historic amenities should be aware of this and should only protect cultural heritage that yields sufficient benefits for house owners. Third, we showed that the urban fabric is valued by house owners: planning policies that stimulate mixed land use may increase house prices, given that land uses are compatible.

This dissertation also analyses the impact of accessibility, in particular access by public transport. Many planning policies aim to fight traffic congestion and reduce automobile dependency by investing in public transport. For example, Transit Oriented Development aims at a development of a high-quality public transport system together with development of mixed use neighbourhoods around stations. We find little evidence for positive accessibility benefits of new stations, which makes it harder to justify huge investments in the opening of new stations. The evidence also questions the strong focus of planning concepts like Transit Oriented Development on new development around public transport nodes.