European cities in advanced producer services and real estate capital flows : a dynamic perspective

As emphasized in Chapter 1, the implications of economic globalization for Europe have been a major concern of the European Union (EU) and its constituent member states for over a decade. As discussed in Chapter 5 the advanced global economy now functions as complex horizontal networks of firms, made possible by developments in information and communications technology (ICT) since the 1970s. A consequent paradigm shift in the spatial organization of the global economy has led the geography of territorial structures such as the EU, its nation states and their borders, to be overlaid by more fluid relational spaces generated by business networks in which cities have a critical functional role. Amongst these networks, city-located APS (advanced producer services), finance and linked business services, have come to have rising prominence in the world economy from the latter years of the twentieth century onwards, as recognized in the EU Lisbon Strategy to boost Europe's economic competitiveness in a global context (EC, 2000). Hence, in this chapter, we focus attention on the role and position of European cities in the evolving APS network economy. Advanced producer services were first identified by Friedmann (1986) as having a key role in the process of capitalist 'world city' formation and by Sassen (1991, 1994) as having complementary geographic dispersion and concentration dynamics which result in a new economic role for 'global' cities.


Introduction
As emphasized in Chapter 1, the implications of economic globalization for Europe have been a major concern of the European Union (EU) and its constituent member states for over a decade. As discussed in Chapter 5 the advanced global economy now functions as complex horizontal networks of firms, made possible by developments in information and communications technology (ICT) since the 1970s. A consequent paradigm shift in the spatial organization of the global economy has led the geography of territorial structures such as the EU, its nation states and their borders, to be overlaid by more fluid relational spaces generated by business networks in which cities have a critical functional role.
Amongst these networks, city-located APS (advanced producer services), finance and linked business services, have come to have rising prominence in the world economy from the latter years of the twentieth century onwards, as recognized in the EU Lisbon Strategy to boost Europe's economic competitiveness in a global context (EC, 2000). Hence, in this chapter, we focus attention on the role and position of European cities in the evolving APS network economy.
Advanced producer services were first identified by Friedmann (1986) as having a key role in the process of capitalist 'world city' formation and by Sassen (1991Sassen ( , 1994 as having complementary geographic dispersion and concentration dynamics which result in a new economic role for 'global' cities. But it was Castells' theorization of a 'space of flows' constituted by a 'multi-edged network ' (1996, pp. 75-6) which drew attention to the way in which APS business organization now interconnects cities worldwide in an economy that is 'informational, global, and networked' (1996, p. 77;Beaverstock et al., 2000). As the operational 'nodes' for information and financial flows in worldwide office networks, cities have become interlinked across territorial borders, characterized by specialized 'functional transnationality'. Indeed, globally linked cities are the specific geographical locations in the world for global APS network centralities (Sassen, 2002). So although Castells recognized that a space of flows was 'transcending' the politically constructed, territorial scales of the 'space of places ' (1996) this has not diminished the importance of contemporary cities as geographically situated, governed places (Hoyler and Pain, 2002).
In spite of predictions that developments in ICT heralded the 'end of geography' and urban business clustering, in-depth interviews held with senior APS actors working in Northwestern Europe have found that the most important international business relationships, skills, knowledge flows, innovations and transactions still occur in offices in major global cities and this clustering dynamic is expected to continue (Taylor et al., 2003;Hall and Pain, 2006;Hoyler et al., 2008). Hence the digitization and financialization of 'invisible' trade has not lessened APS network centralities and the spatial concentration of specialized high complexity, high-value-added functions. Moreover, global firms say that increasingly competitive global APS markets require their presence in the deep infrastructure of the world's major global cities such as London and New York.
It follows from these parallel operational requirements for co-location and functional concentration that the global space of APS inter-city flows is intrinsically dependent on the availability of appropriate physical city infrastructures such as state-of-the-art office real estate developments, ICT and multi-modal transportation, including international airport hubs (Pain, 2011). In other words, the material form of cities and their supply of commercial office real estate are interdependent with their functional role in these important sectors of the global economy. Furthermore, office real estate has been financialized and 'repackaged' as an international, tradeable financial asset (Coakley, 1994;Lizieri, 2009) leading to functional integration between global finance and real estate in global and globalizing cities.
Due to financial deregulation and the creation of new international investment funds since the latter part of the twentieth century, APS foreign direct investments (FDIs), occupation and the ownership of city office space have all risen sharply, as have inter-city capital flows (Lizieri and Kutsch, 2006;Lizieri et al., 2011). The result is an 'interlocking' between APS location drivers, office occupation and financial investments; hence, centralities in global financial and linked business networks are replicated and reinforced by international real estate investment flows. This begs the question whether the geographies of real estate capital flows can be regarded as a proxy for those in global financial networks and, if so, whether the world's most prominent global cities are especially vulnerable to volatility in the global financial system as a consequence (Lizieri, 2009).
As noted by Lizieri (2009) the physical infrastructure of cities partially 'locks down' and 'fixes' cross-border capital flows in international office markets, hence, in the context of neoliberalization, urban administrators have come to regard APS inward investment as a vital source of finance 'to mobilize city space as an arena both for market-oriented economic growth and for elite consumption practices' (Brenner and Theodore, 2002, p. 21;Peck and Tickell, 2002;Swyngedouw et al., 2002;Knox and Pain, 2010). Understanding the intertwined relations between finance and real estate flows is therefore of major relevance for Europe 2020 priorities for economic growth and for the issue of where development occurs, which is the primary concern of the territorial cohesion objective (EC, 2000;CEC, 2010CEC, , 2011CEC, , 2012. This chapter addresses this concern through an empirical study of APS office networks and international office real investment flows. First we describe the bespoke data and innovative analytical methodologies employed in our research. Second we examine the changes in the APS global space of flows, and the position of Europe within this, from the year 2000 (the time of the launch of the EU Lisbon Strategy) to 2010. Third, we consider the impact of the global financial crisis on international office real estate investment flows. We compare their network centralities with those for financial services networks in order to establish whether real estate capital flows can be considered a proxy for global city financial concentration and we examine whether Europe's international financial centres are vulnerable to real estate office market falls in periods of financial instability by examining flow data for the years immediately before and after the crisis. Finally, we draw attention to the main implications of our results for Europe 2020 and territorial cohesion policies.

Data, analyses and network visualizations
Unique primary data collected by the Globalization and World Cities (GaWC) Research Network 1 in four census periods between 2000 and 2010 2 were used in our APS analyses. 3 Taylor's interlocking network model 4 was employed to calculate the global 'connectivity' that is conferred on cities worldwide by APS firms and which generates a functionally interlinked 'world city network' (WCN) (Taylor, 2001(Taylor, , 2004Taylor et al., 2011). The size of APS offices and their functions in the global network of an individual firm gives rise to intra-firm linkages between the cities where they are located and hence to inter-city functional linkages.
Calculating the total connectivity generated by all the global firms present in a given city thus shows the position of that city in the APS world city network. By applying the interlocking network model in our research we have thus been able to plot the connectivities, positions and functional interlinkages of European cities in the world city network between 2000 and 2010. We have adopted a time-series approach to analysis in order to shed light on changing dynamics during this period and we have also undertaken specific sectoral analyses in order to inform understanding of the role of financial and other sectors in shaping European spatial relations in the network economy.
While comparisons between relative connectivity values suggest a vertical hierarchical ranking of cities, it is important to understand that what is being measured in GaWC APS analysis is how well a city is connected to all others in the worldwide network of global cities, hence the inter-city relations and relative network positions of cities represent cross-border functional complementarities between them and not competitive relations. However, as we will explain in due course, world city network connectivity and inter-city office real estate capital flows are highly integrated; hence there are important territorial outcomes from a city's network positionality.
In order to be able to make the time series comparisons for the GaWC census periods, it was necessary to standardize the four datasets which contained different numbers of firms, cities and APS sectors. 5 Standardization was undertaken at the level of city pairs (dyads) 6 1 See www.lboro.ac.uk/gawc. 2 The 2010 data have been added to our analyses since the completion of the original ESPON funded research and so are not included in the official final report. 3 The sectors studied were banking and finance, together with specialized legal, accountancy, advertising and management services, which have complementary business relations in global markets. 4 For full technical details about the model and measurement method, see Taylor (2001Taylor ( , 2004Taylor ( , 2012. See also Borgatti et al. (2002), for an explanation of the software used in the network analysis. 5 The data standardization method is explained in detail in Tiger Final Report Working Paper 3 . It differs from standardization procedures employed by Taylor and Aranya (2008), for a total of five sectors: accountancy, law, advertising, management consulting and financial services for the following numbers of firms and cities worldwide: A subset of 285 cities present in all four datasets was then used for further analysis.
Real estate investment flows not only between, but also within, cities worldwide were analysed using Real Capital Analytics (RCA) data on the top 1000 commercial property transactions in each year from 2007 to 2010. Information on recorded deals checked and triangulated by RCA was used to compile a dataset suited to our analyses. Information on office property location, price, and buyer identity and location, was used to trace all recorded deals geographically, resulting in a dataset similar to an origin-destination matrix. We identified both the location of office property sold and the location of the purchaser of that property, defined as the headquarters of the beneficial owner, for every commercial deal in the real estate database in order to elicit the capital flows generated by international investors. In some instances the purchaser location was unclear. In the case of 'off-shored' or tax-haven purchaser headquarters we researched the location of the effective operational headquarters. Where financial firms have acted as asset managers for a wide range of investors whose identity is unknown, we have used the location of the principal office of the fund manager. While this may differ from the head office of the parent company, for example, when a Swiss bank manages its real estate funds in London, this latter location reflects both the origin of the capital going into the funds and also the location of the asset managers generating the associated capital flows. In deals involving multiple investors in joint ventures, in the absence of clear capital shares, we divided the acquisition price into equal parts. In this way we produced a final dataset comprising 297 cities worldwide showing global networks of major financial investments in the commercial office real estate market for the 2007-10 period. Both aggregated and annual data 7 are referred to in this chapter. The aggregated data provide a robust overview of the deals taking place between 2007 and  and Hanssens et al. (2011), which explains minor differences in ranks of cities compared to those previously reported. 6 The term 'dyad', as referred to here, indicates the relations between a pair of cities that are generated by the APS networks which are interlinking them in the world city network . 7 Annual fluctuations are short-term however these results are controlled for by data aggregation over the longer-term 2007-10 time period. 2010, 8 whereas annual data allow us to examine changing capital flows before and after the global financial crisis.
In order to examine the respective network structures for financial services and real estate investments, Pearson and Spearman Correlation and Quadratic Assignment Procedure (QAP) tests were applied to comparable data sets for a subset of 141 cities (Krackhardt, 1988). 9 Because the central purpose of this chapter is to examine the position of European cities in these networks, mappa mundi analyses (Vinciguerra et al., 2010) 10 were used to visualize these network structures when geographical distance is not considered and 'dyad' interlinkages between specific pairs of global cities. In some cases we also group cities territorially either at the level of the nation state or the EU to show how the space of flows maps onto the space of places. 11 In this way we depict the functional relationships between European and other global cities in an evolving business and financial network space.

A global overview
The results from APS time series data analyses show that London, New York, Hong Kong Nevertheless, there has been a general decline in city APS global connectivity between 2008 and 2010 in the aftermath of the financial crisis and the onset of world economic recession with some notable exceptions. London has retained its 2008 connectivity level but the real surprise has been that Dubai has experienced a huge recent connectivity increase, which has also changed its world city network position from 48th in 2008 to 10th in 2010. 12 Shanghai's world city network position continued to rise to 6th in 2010.
Inter-city dyad links both for general APS network connectivity and for financial network connectivity have remained strongest between London and New York during the whole 2000-10 period. However, links between many other city pairs have strengthened, many of which involve cities in Europe.
We have used a mappa mundi analysis to visualize the territorial implications of these relational changes (Vinciguerra et al., 2010). Considering the global financial services networks, we see that the UK appears to be in a peripheral world city network position in spite of the stable top global position of London in each census period, due to the relatively weak global connectivity of other UK cities (Figures 1 and 2). On the other hand the US has retained a relatively central network position due to the size of its national territory and its larger number of cities with APS offices than is the case in the UK. Country size can thus be relevant in a competitive territorial paradigm. We see that China is rapidly gaining world city network centrality principally due to the upsurge in connectivity of Shanghai and Beijing and the stable leading position of Hong Kong. The development of other globalizing Chinese cities can be expected to accelerate this centralizing dynamic (Derudder et al., 2013).  Principal components analysis 13 reveals the inter-city relations generated by the European strategies of these firms for six factors in the year 2008, benchmarking the way firms were using European cities before the economic recession hit: 1. The '"Outer" European Capitals Strategy' -firms focusing principally on cities other than London and Paris, especially cities in Eastern and Southern Europe such as Warsaw, Budapest, Prague, Bucharest, Athens and Lisbon.
2. The '"Primate" (London) Strategy' -firms focusing only on London for their European work. 13 A standard Varimax rotation was applied and the output scores allowed analysis of city relational outcomes generated by 160 firms. For six factors, output scores could be interpreted as representing the strategies of the global firms using cities in the European territory (see Table 24, Pain et al., 2012).

Global network connectivity and position conferred on cities by different APS sectors
The issue of which specific APS sectors are driving contemporary city connectivity changes and the city functional linkages generated by them is clearly important to inform EU and member state policies to support further city economic growth in a European and global context. Madrid, Paris and Brussels -with New York being its fourth dyad. Interestingly, the three top nodes for intra-city, self-investment flows are the three 'global cities' originally identified by Sassen (1991) -New York, Tokyo and London -demonstrating both the investment concentration within these mature financial centres and the importance of critical mass in city capital and real estate markets.
Looking at the worldwide rank order of city in-, out-, and self-investment flows, other interesting results are those for Hong Kong, which ranks just 43rd for in-flows and 16th for out-flows compared with a self-investment rank of 7 (only just behind Paris [4], Seoul [5] and Moscow [6]), suggesting that Hong Kong may be less active as a conduit for Asian investment flows than might be expected. Very many other cities worldwide -236 in allare also nodes for in-flows and 122 cities are nodes for out-flows but, by comparison, only 66 cities are nodes where intra-city (self-) investment is occurring, which may be indicative of their less established office property markets. In general, it is mature cities that are seen to dominate the higher in-flow, out-flow and self-investment rankings, and top ranked global cities -London, New York and Tokyo -have significantly higher overall flow levels, suggesting that they are the leading global spaces for both office real estate finance and APS centralities.
Within Europe, Paris demonstrates a similar distribution of investment flows between 2007 and 2010 to that of London, ranking 3rd as a node for in-flows, 27th for out-flows and 4th for self-investment. Looking at other European investment geographies, Munich has been the 4th city for out-flows within Germany with cross-border links to multiple other European cities, as well as extensive dyad relations across the world. Frankfurt has been Europe's 5th outward investor followed by Dublin (6th) Mappa mundi visualization shows that European countries, the UK (ranked 1st), followed by France and Germany, have been ahead of the US and China, respectively, in their volume of in-flows for the whole 2007-8 period. Within Europe, the UK ($36.17 billion) has been especially important having had more than twice the in-flows of France ($15.11 billion) and significantly higher volumes of in-flows than Germany. At the same time, out-flows from the UK (ranked 3rd / $12.01 billion) were significantly lower in volume than those of Germany (2nd / $25.75 billion) and the US 1st / $34.09 billion). Many cities have contributed to 2007-10 US outflow dominance over in-flows compared to the UK. Eastern European countries, the Czech Republic, Hungary, Poland and Romania have had no self-investment or out-flow scores but they have ranked higher for in-flows suggesting that their internal capital markets have not yet developed sufficiently for their investors to participate in major global real estate transactions.

Global finance and real estate interdependencies
Focusing on the value of office real estate investment deals for city dyads, we see the amounts of finance capital passing between any two cities during the 2007-10 period. This allows us to identify which specific cities are the originators of outbound, and the recipients of inbound, investments in inter-city relations. In this case, we focus specific attention on the roles of mature financial centres in articulating the in-and out-flow investments at different geographical scales. The 2000-8 GaWC data for financial services network connectivity identify London, New York, Tokyo, Hong Kong, Singapore and Paris as the six major world financial centres with a consistent top ranking during the past decade. Next we consider the position of these cities in international real estate investment flows at the time of the financial crisis.
Financial services and aggregated real estate data reveal that cities with high global real estate investment in-flows also have high financial services network connectivity scores ( Figure 4). There is thus a strong inference that interdependencies between global financial services networks and real estate office markets are constructing intersecting network centralities in global cities. As many as 141 cities in the world are interconnected through both global finance and real estate investment networks, albeit their connectivity rankings vary substantially. As might be expected, London (rank 1) and New York (rank 2) are the best-connected world cities for both financial services network connectivity and for real estate in-flows; however, as already discussed, London has nearly three times the real estate in-flows of New York which may reflect its supreme role as an international financial services centre.  (Krackhardt, 1988) revealing the extent to which networks are similar through correlation, or whether one network structure can be explained by other network structures through regression. The outcomes demonstrate that financial services and real estate networks are correlated by almost 36 per cent. These results confirm the significance of the interrelationship between the geographies of real estate investment capital flows and global city financial services network connectivity. Direct causal relationships cannot be proved due to the limited availability of real estate time series data. However, international office investments are clearly strongly concentrated in major global cities during the period surveyed. To some extent this is a function of the scale and value of the office space available in those markets -but the availability of that space helps to preserve the role of the major financial cities in global capital networks and the value of the space reflects the economic effect of their network centrality.

Global cities at risk?
Given the interdependence demonstrated between global city financial services network connectivity and real estate investment flows, it is clearly important to consider the extent to which the concentration of capital flows in global financial services networks and real estate markets represents a risk of contagion for the world's major global cities in financial crises.   http://www.rcanalytics.com/ Significantly then, in the aftermath of the crisis, the position of the EU in global real estate flows looks strong in comparison with that of the US, reflecting its higher representation of international financial centres. Whereas US cities suffered more severely from the impacts of the crisis, cities in the European Union and London in particular proved to be more attractive locations for real estate investment by the year 2010. Hence, the surprising feature of the crisis has been that the sustainability of international financial clusters, both globally and within Europe, appears to have increased rather than decreased in spite of their exposure to global capital market volatility and the sharp capital value falls generally experienced.

Conclusions: implications for European policy
In conclusion, we find that two interconnected globalization processes are relevant for policy to promote the position of the EU territory in the advanced global economy (EC, 2011). First, a process of deepening APS concentration prior to the global financial crisis, has led many globalizing cities across the world and in Europe to gain increased connectivity, linking city regions and national economies to the global space of flows. London has developed an especially important role in linking Europe to North America through New York ('the New York-London axis', see also Wójcik, 2013) and, more recently, to the emerging Pacific Asia region through Hong Kong, Shanghai and Beijing. Understanding which specific sectors are making the links between cities is clearly an important consideration in policies both to promote Europe's economic growth in a global context and more balanced development across the EU. Second, increasing interdependencies between financial services networks, office real estate transactions and capital flows, have led to financial functional integration in major global city markets such as that of London. Real estate capital flows can thus be confirmed as being a proxy for city integration in global financial networks, demonstrating the significance of world city network position for capital flows into urban infrastructure which contributes to economic resilience.
It has been hypothesized that a strong interrelationship between capital flows in office markets and depth of a city's financial integration might make global cities like London especially vulnerable to global investment swings associated with financial crisis. However, we have found that in practice there has been an international 'flight to liquidity' by investors, which has favoured the largest real estate markets with higher unit prices and greater transaction volumes. The correlation between financial services network connectivity scores and aggregated volumes of real estate capital flows, suggests that capital has actually flowed to higher-value city locations in the recent financial crisis. We speculate that global financial firms and linked APS, generate agglomeration economies and higher profits from clustering in global cities so that higher rents per square metre paid are capitalized in transaction prices which are reflected in real estate investment levels (see Lizieri and Pain, 2013).
These results have an important bearing on the tension between European strategic spatial and economic growth priorities introduced in Chapter 1. Our results indicate that spatial policy objectives for urban polycentricity to rebalance territorial development at metropolitan to EU-wide scales (European Commission, 1999;CEC, 2011CEC, , 2012