Evidence from the metropolitan area of Milan
In this chapter we estimate the effects of spatial agglomeration on KIBS vertical disintegration with reference to the metropolitan area of Milan. Relying on a rich firm-level dataset, we first geo-referenciate each KIBS firm by employing a GIS routine; then we define a set of distance-based density indicators counting the number of neighbouring firms located within a series of concentric rays from each KIBS firm location. Our estimates show that KIBS tend to be more disintegrated as the number of neighbouring firms increases, without showing any clear process of spatial decay within five kilometres. We also find that this result is particularly strong for creative KIBS, whereas the correlation for traditional professional and technology-related ones is weaker.1
In recent years, next to the outsourcing of material inputs and the low skill-intensive stages of production, firms have begun to externalize business services and high skill-intensive tasks. Improvements in communication technology as well as the increasing globalization of information software (Morgan, 1994) have enabled business services to be broken down into modules, that do not need to be developed internally, and can be produced almost anywhere in the world. Thus, firms are able to contract out services, ranging from routine call centre work to higher value software programming and research and development (R&D) activities (Prencipe, 1997; Willcocks and Lacity, 2006). Despite the relevance of this topic, little empirical research has been conducted on service outsourcing, and, particularly, on the outsourcing of knowledge-intensive business services (KIBS) (Earl, 1996; Barthelemy, 2001; Amiti and Wei, 2005, 2009; Bengtsson and Dabhlikar, 2009; Windrum et al., 2009; Rodriguez and Nieto, 2012).
The key role of KIBS in driving regional competitiveness and growth is well recognized in the regional economics literature (Fujita and Thisse, 2002; Acs et al., 2008; Piras et al., 2009; Doloreux et al., 2010; Ciarli et al., 2012). The benefits brought by a high concentration of KIBS on the level of development of regions range from the generation of knowledge spillovers, to a higher rate of innovativeness and entrepreneurship, or a higher level of productivity in non-KIBS activities. Despite this, little empirical research has been conducted on the determinants of KIBS location, as well as on the determinants of KIBS boundaries (Windrum et al., 2009).
Identifying which local attributes attract new activities, or tend to push existing firms to contract out business-related activities is of paramount importance for understanding the drivers of local entrepreneurship and of regional competitiveness. In this respect, the aim of this work is to assess the impact of one of these attributes, namely the spatial concentration of industry, on the degree of vertical disintegration of KIBS firms.2 In supplying knowledge-intensive services to other firms, KIBS also demand for knowledge-intensive activities (see Figure 10.1).
A larger and denser market not only favours contacts and face-to-face interactions during the production process, but also makes the accessibility to local inputs easier. Therefore, ‘the argument of distance-sensitiveness applies to both input and output markets’ (Andersson and Hellerstedt, 2009: 105), as pointed out also by transaction cost theory (Klein et al., 1978; Pirrong, 1993; McLaren, 2000), according to which specialized suppliers, i.e. firms selling to other firms, are expected to be more frequent in thicker markets in order to reduce hold-up problems related to asset specificity.
With this framework in mind, we aim at estimating if, and to what extent, the degree of vertical disintegration of KIBS is affected by the number, and the distance, of neighbouring KIBS located in a specific metropolitan context. In particular, relying on a rich firm-level dataset, we estimate the impact of the urban
concentration of industry on the share of purchased business services by KIBS headquarters located and operating in the metropolitan area (MA) of Milan (see Figure 10.2).
The choice of Milan is not only driven by data availability. In fact, other than being the economic engine of Italy, Milan is considered ‘a European capital of advanced services and a fashion and design global capital’ (OECD, 2006: 87), and the ‘centre of Italian business management and strategy’ (OECD, 2006: 93). In particular, the ‘core of Milan is progressively becoming a technology and service hub’ (OECD, 2006: 87–8). Despite its prestigious industrial history, Milan is now highly specialized in high-tech service activities. For instance, in 2002, in front
of an average of 32 per cent of employees engaged in high-knowledge activities in Italy, and 39 per cent in the Lombardy region, Milan accounted for 46 per cent. Interestingly, while the employment shares in high-tech manufacturing were 10.5 per cent for Italy, 14.8 per cent for Lombardy and 15.1 per cent for Milan, the shares for knowledge-intensive services were much higher, namely 21.6 per cent for Italy, 24 per cent for Lombardy and almost 31 per cent for Milan (OECD, 2006: Table 1.3). In addition, according to census data, Milan is now one of the most densely populated agglomerations in Italy, with more than 7,000 inhabitants per square kilometre (see Figure 10.3).
The rest of this chapter develops as follows. After defining KIBS in the next section, we go on to review the literature related to the effects of spatial agglomeration on the vertical disintegration of firms. Then we describe our dataset and the variables involved in the estimates. In the following section we first discuss the methodology adopted for geo-referencing our data and computing the distance-based agglomeration variables, then, we discuss our estimation results. In this respect, we should swiftly acknowledge that the presence of non-linearity in the geographical localization of firms and of potential endogeneity in the link between agglomeration and vertical disintegration claim for a bit of caution in interpreting standard OLS (ordinary least squares) estimates. For these reasons, we consider our econometric evidence as just exploratory. The final section concludes and discusses some directions for future research.
Traditionally, KIBS are defined as ‘services that involved economic activities which are intended to result in the creation, accumulation or dissemination of knowledge’ (Miles et al., 1995: 2), or as ‘private companies or organizations that rely heavily on professional knowledge, i.e. knowledge or expertise related to a specific (technical) discipline or (technical) functional-domain to supply intermediate products and services that are knowledge-based’ (Den Hertog, 2000: 505). Broadly speaking, KIBS are activities primarily concerned with the provision of knowledge-intensive inputs to the business process of other organizations, both from the private and the public domain (Muller and Doloreux, 2009). In addition, KIBS are characterized by their heavy reliance on professional knowledge, both codified-explicit and tacit-implicit. They can be considered a primary source of information and external knowledge; they can use their knowledge to produce intermediary services for their clients' production processes; and they are typically supplied to business through strong supplier-user interactions (Miles et al., 1995; Muller and Zenker, 2001).
The literature has also traditionally identified two types of KIBS: (i) advisory services, primarily involving legal activities, bookkeeping, auditing, business and management consulting, marketing, advertising and other administrative tasks; and (ii) technical services, such as computer services, engineering and design, technical testing and analysis (Koschatzky and Zenker, 1999).
Miles et al. (1995) provide the following distinction between (i) traditional professional services which are likely to be intensive users of new technology (marketing, advertising, training, design, financial services, office services, building services, management consultancy, accounting, legal services, environmental services), and (ii) new technology-based services (telematics and computer networks, training in new technologies, design involving new technologies, technical engineering, R&D, IT-based building and environmental services, and so on).
More recently, a finer disaggregation led to the identification of a third class of KIBS, namely C-KIBS (or creative KIBS), which include activities developing creative and cultural knowledge such as advertising, graphic design, etc. These activities are no longer involved in providing knowledge support to customers, but they are primarily providing solutions to business problems of their clients (Bettiol et al., 2012; Miles, 2012).
Recent developments in regional and urban economics has identified in the spatial clustering of firms a crucial factor for determining the degree of vertical disintegration of firms. This relationship dates back to Stigler (1951), and rests on the ideas that, on the one hand, spatial proximity to suppliers per se reduces transport, search and managerial costs, leading to higher vertical disintegration (Goldstein and Gronberg, 1984), while, on the other, it reduces the scope for opportunistic behaviour by increasing mutual visibility and reciprocal trust (Helsley and Strange, 2007), particularly when dealing with innovation processes and complex transactions (Love and Roper, 2001).3
On this principle, Holmes (1999), using census data on USA manufacturing plants in 1987, finds that the intensity of input purchased by a plant is positively correlated with the level of employment of neighbouring plants in the same industry. In this case, a positive relationship arises between outsourcing and industrial specialization, whereas a null or even negative correlation emerges with respect to the employment level in neighbours operating in related industries. Finally, mixed evidence characterizes the relationship between vertical disintegration and urbanisation: positive up to a certain level of population density and negative afterwards.
Similar results are also obtained by Ono (2001, 2006) for the USA and Li and Lu (2009) for China. Using data from the 1992 Annual Survey on Manufacturers, the former finds that, once controlled for firm characteristics, a final producer is more likely to outsource when it is located in thick markets, as measured by the level of employment in MAs. This effect derives from an increase in the local demand for outsourcing, which, in turn, increases the localization of specialized suppliers, thus decreasing the price of the service. Then, a lower price will increase the potential benefits for the manufacturing firm, thus increasing the likelihood to outsource. In particular, she finds that a doubling of US local demand for white collar services increases the probability to outsource by an amount between 7 per cent and 25 per cent.
Looking at Chinese manufacturing in 2002 and 2003, Li and Lu (2009) also find a positive effect of the geographic concentration on the degree of vertical dis-integration, even when controlling for endogeneity. In particular, a 1,000 increase in own neighbouring employment increases the firm purchased-input intensity by 0.017 percentage points when endogeneity is not controlled and simple OLS are used, and up to a 0.098 percentage points when instrumental variables are used.
The positive correlation between spatial clustering of firms also arises in the Turkish textile and engineering industries (1993–2000), as well as in the Spanish electronic industry (1995–97). In the former case, Taymaz and Kiliçaslan (2005) find that the increase in the number of firms operating in the same sector and province has a positive relation with the firm propensity to both subcontract offering and receiving. In the latter case, Rama and Calatrava (2002) and Rama et al. (2003) find that the probability to establish stable outsourcing relationships among entrepreneurs increases, among other factors, with the geographical and professional proximity of firms, being particularly relevant within industrial districts.
Looking at Italian manufacturing firms over the period 1998–2003, Antonietti and Cainelli (2008) find that the outsourcing of KIBS also is driven by the interplay between R&D and firms localization within dense local production systems. This finding reinforces the Love and Roper (2001) results in stressing the role of geographic proximity, knowledge spillovers and closer interaction among agents in simplifying the management of complex transactions and in increasing firms' competitiveness even in the face of increasing globalization of production.
On the same line, Antonietti et al. (2012), using data collected by the national Fiscal Authority, find that the propensity by small firms, operating in the machine tools industry of Emilia Romagna region in Italy, to fully outsource production activities to external suppliers increases with the density of neighbouring employment in related three-digit sectors.
Finally, relying on a wide dataset on Italian business groups in 2001, Cainelli and Iacobucci (2012) find that a higher vertical related variety, i.e. a higher degree of technological similarity between two industries, reduces the need for firms to integrate activities, as it increases the opportunity to find specialized inputs within the region of firm location.
All these studies find that, as the local concentration of employment increases, firms tend to increase the purchase of material inputs or business services from external suppliers. This relationship is generally estimated with respect to manufacturing firms, whereas little attention is given to the service outsourcing activities made by service firms, despite the fact that ‘outsourcing within service sectors accounts for the majority of all outsourcing’ (Windrum et al., 2009: 203).
The recent literature has also emphasised the role that spatial proximity plays in driving the performance of KIBS industries. The local demand for intermediate goods and business services, the need for frequent interactions with clients, but also the presence of local institutions and universities – as well as research centres and firm incubators – typically located within MAs or regional innovation systems, not only foster the creation of KIBS, but also their innovative capability and growth (Doloreux et al., 2010; Di Maria et al., 2012; Shearmur and Doloreux, 2012).
Since KIBS are the firms which mostly demand business-related services, we attempt to answer the following research questions:
(1) Does the spatial concentration of KIBS industry in the metropolitan region of Milan affect the degree of vertical disintegration of individual KIBS activities?
(2) Does this effect depend on the nature of KIBS, i.e. traditional professional versus technology-related versus creative KIBS?
In so doing, we attempt to fill a twofold gap in the urban economics literature: on the one hand, works have been primarily involved in studying the economic behaviour of manufacturing activities; on the other, empirical studies have failed to analyse the agglomeration-disintegration relationship with respect to the KIBS context. In this respect, some works have considered the role of KIBS in regional clusters and the role of spatial proximity in shaping their economic performance (Smedlund and Toivonen, 2007; Doloreux et al., 2008; Doloreux and Shearmur, 2012; Shearmur and Doloreux, 2012), but few studies have analysed the relationship between spatial agglomeration and firm disintegration, i.e. the degree of specialization in the service context (some exceptions are Baumgardner, 1988; Antonietti and Cainelli, 2008, 2012; Anderson and Hellerstedt, 2009).
Finally, we focus our analysis on a novel urban context, i.e. the city of Milan in Italy, which, although being of the major European KIBS hub, has never been investigated before with respect to firm location and vertical disintegration.
Data are drawn from AIDA, a commercial database collected by Bureau Van Dijk. This large data set of Italian joint stock companies gathers information on balance sheet data, as well as on the sector of activity and on the geographical location of each firm. After cleaning of missing values in the variables of interest, we come up with a sample of almost 12,000 KIBS (headquarters) firms located in the MA of Milan in 2008.
We consider as KIBS those firms belonging to the ‘professional, scientific, and technical activities’ industry (according to the standard ATECO 2007 classification), and, in particular, to the following two-digit sectors: 62 – computer programming, consultancy and related activities; 63 – data processing, hosting, and related activities; 69 – legal and accountability activities; 70 – head offices and management consultancy activities; 71 – architectural and engineering activities, technical testing and analysis; 72 – scientific research and development; 73 – advertising; 74 – other professional, scientific and technical activities.
Table 10.1 shows the distribution of firms by industry and by location inside or outside the city of Milan. The three most diffused activities in the MA concern management consultancy, advertising and software production. Outside Milan, instead, we find a higher concentration of computer programming activities and architectural/engineering studios. Looking at the three broad classes of KIBS, we observe that traditional professional and creative KIBS are more likely to locate within the urban core of Milan, whereas technology-related KIBS seem to prefer a more peripheral location.
As dependent variable, because of the high presence of missing values in the sales and output value data, we cannot rely on the standard costs/output index of vertical disintegration (as in Holmes, 1999). In order to overcome this issue, we follow the literature on (international) outsourcing (Feenstra and Hanson, 1996, 1999; Bengtsson and Dabhikar, 2009) measuring the degree of vertical
Industry (2-digit) | No. | % | % Milan | % Out Milan |
62 – Computer programming, consultancy | 2,111 | 18.21 | 16.48 | 24.74 |
63 – Data processing, hosting, and related activities | 1,277 | 11.01 | 9.47 | 16.85 |
69 – Legal and accountability activities | 758 | 6.54 | 6.93 | 5.04 |
70 – Head offices and management consultancy activities | 2,998 | 25.86 | 27.98 | 17.80 |
71 – Architectural and engineering activities | 1,134 | 9.78 | 8.93 | 13.01 |
72 – Scientific research and development | 233 | 2.01 | 2.06 | 1.82 |
73 – Advertising | 1,884 | 16.25 | 17.48 | 11.61 |
74 – Other KIBS | 1,199 | 10.34 | 10.66 | 9.13 |
Professional KIBS (69 + 70 + 749) | 4,203 | 36.25 | 38.67 | 27.10 |
Technology-related KIBS (62 + 63 + 72) | 3,621 | 31.23 | 28.02 | 43.41 |
Creative KIBS (71 + 73 + 741 + 742 + 743) | 3,770 | 32.52 | 33.32 | 29.49 |
Total | 11,594 | 100.0 | 79.12* | 20.88* |
* The percentage is computed on the total amount of firms (12,008).
disintegration of the firm by the share of purchased services over total production costs:
(10.1) |
where CSit is the value of purchased services by firm i in period t, TCit is total production costs defined as the sum of CSit, labour cost, depreciation and other costs including energy and material costs, transport, reimbursements, training, advertising, and so on. This purchased-inputs variable allows accounting for the fact that ‘many business services are likely to be exactly the kind of locally produced intermediate input that producers in localized areas will have greater access to than producers in isolated areas’ (Holmes 1999: 316).
Table 10.2 shows the average value of the DIS variable by industry and geographical area (inside or outside Milan).
From Table 10.2 it is evident that, first, the average degree of vertical disintegration is higher than 50 per cent for almost all KIBS industries. This evidence is in line with input-output data (see Figure 10.3) provided by the Italian
Industry (2-digit) | Mean | Median | Milan | Out Milan |
62 – Computer programming, consultancy | 0.500 | 0.488 | 0.513 | 0.465 |
63 – Data processing, hosting, and related activities | 0.444 | 0.415 | 0.451 | 0.430 |
69 – Legal and accountability activities | 0.532 | 0.539 | 0.545 | 0.463 |
70 – Head offices and management consultancy activities | 0.560 | 0.585 | 0.567 | 0.521 |
71 – Architectural and engineering activities | 0.563 | 0.577 | 0.584 | 0.508 |
72 – Scientific research and development | 0.530 | 0.537 | 0.542 | 0.475 |
73 – Advertising/marketing | 0.587 | 0.622 | 0.600 | 0.512 |
74 – Other professional KIBS | 0.525 | 0.527 | 0.532 | 0.496 |
Professional KIBS (69 + 70 +749) | 0.552 | 0.573 | 0.559 | 0.511 |
Technology-related KIBS (62 + 63 +72) | 0.482 | 0.462 | 0.494 | 0.452 |
Creative KIBS (71 +73 + 741 + 742 +743) | 0.567 | 0.590 | 0.582 | 0.503 |
Total | 0.535 | 0.541 | 0.549 | 0.483 |
Institute of Statistics (Istat), in which the average share of purchased services by knowledge-intensive service industries in 2005 is around 75 per cent, with respect to an average of 33 per cent by the manufacturing industry. Second, creative KIBS (0.567) are on average more disintegrated than traditional professional KIBS (0.551), which, in turn, are on average more disintegrated than technology-related KIBS (0.509), these differences being significant at 1 per cent level after a t-test. Third, firms located in the city of Milan are on average more disintegrated (0.549) than units located in the periphery (0.483), regardless of their specialization (this difference is significant at 1 per cent level after a t-test). We interpret this as a sign that the spatial concentration of activities is one key determinant of firm boundaries, as it can reduce transport and search costs, as well as facilitating the knowledge and information exchange among agents (Goldstein and Gronberg, 1984; Holmes, 1999; Love and Roper, 2001; Antonietti et al., 2012).
Finally, as the dependent variable is bounded between 0 and 1, the model we estimate must be consistent with this constraint. This is the case, for instance, of a logistic functional form, in which:
(10.2) |
Taking logarithms and rearranging terms, we estimate the following model using OLS:
(10.3) |
Following the urban economics literature on the determinants of outsourcing at the firm level (Love and Roper, 2001; Ono, 2001, 2006; Antonietti et al., 2012), as a vector of explanatory variables X, we include those described in Table 10.3: firm age (AGE), size (SIZE) and its squared value (SIZE2), the one-year lagged value of the dependent variable, which captures persistency effects in vertical disintegration (DIS2007), a set of distance-based spatial agglomeration variables, which should capture the effect of the local concentration of industry on KIBS firms vertical disintegration and a set of industry-specific dummies which should capture unobserved characteristics at the industry level.
Variable | Description |
AGE | 2008 – start-up year |
SIZE | Output value, 2008* |
SIZE2 | Squared output value, 2008 |
DIS2007 | Lagged vertical disintegration variable, 2007 |
TOT 0–1 km | Total number of KIBS located within a 1-km ray from firm i location |
TOT 1–3 km | Total number of KIBS located between 1 and 3 km from firm i location |
TOT 3–5 km | Total number of KIBS located between 3 and 5 km from firm i location |
TOT 5 km | Total number of KIBS located within a 5-km ray from firm i location |
Rel_Tot 0–1 km | Total number of firms located within a 1-km ray from firm i location with respect to the total number of firms in the MA of Milan |
Rel_Tot 1–3 km | Total number of firms located between 1 and 3 km firm i location with respect to the total number of firms in the MA of Milan |
Rel_Tot 3–5 km | Total number of firms located between 3 and 5 km from firm i location with respect to the total number of firms in the MA of Milan |
Rel_Tot 5 km | Total number of firms located within a 5-km ray from firm i location with respect to the total number of firms in the MA of Milan |
Industry dummies | Dummy = 1 if the firm belongs to two-digit industry s, 0 otherwise |
* Because of the presence of a high amount of missing or zero values, we cannot rely either on employment variables or on labour cost variables.
As our dataset contains information on the spatial coordinates (latitude and longitude) of each observation, we proceed to firm geo-referenciation by using a GIS
routine. In so doing, we clearly identify the position of each firm in the MA of Milan, and we can calculate the Euclidean distance among them. In this respect, Figures 10.4(a), (b) and (c) show the distribution of KIBS firms by two-digit industry, which suggests firm density follows an inverted U-shaped path. This holds both for neighbouring firms belonging to the same two-digit industry and for firms belonging to the other KIBS industries as a whole, with respect to the industry of reference. This evidence is interesting, as it shows that the number of neighbouring firms follows a non-linear path that reaches a maximum at a distance between two and three kilometres from each firm.
With this picture in mind, and following Arzaghi and Henderson (2008), we define a series of three rings moving out in increments of one and two kilometres, reaching up to a boundary of five kilometres (i.e. from 0 to 1, from 1 to 3, and from 3 to 5 km). Differently from Arzaghi and Henderson (2008), we do not have census tracts, but, knowing the exact position of each firm, we can define rings starting from this location, instead of using the centroid of each tract.4
For each ring, we count the number of existing KIBS firms. Ring 1 is the count (in logarithm) of existing firms located within a ray of one kilometre from firm i (TOT 0–1), which actually is the own ring. Ring 2 is the count of firms located between one and three kilometres (TOT 1–3) from firm i, and Ring 3 is the count of
firms located between three and five kilometres (TOT 3–5). Finally, we also include the total count of neighbouring firms within a five-kilometre ray (TOT 5 km).
Next to absolute counts, we also calculate relative measures of spatial agglomeration by dividing the number of neighbouring firms by the total number of KIBS firms located in the MA of Milan. So, we obtain the relative amount of neighbouring firms within a one-kilometre ray (Rel Tot 0–1), between one and three kilometres (Rel Tot 1–3), and between three and five kilometres (Rel Tot 3–5). Again, we include the total relative count of neighbouring firms within five kilometres (Rel Tot 5 km).
Tables 10.4 to 10.7 report our OLS estimates. Because of the high correlation among spatial agglomeration variables (all above 0.8 and statistically significant at 1 per cent level), we first include them separately. Tables 10.4 and 10.6 show the OLS estimates on the whole sample, while Tables 10.5 and 10.7 report the estimate results for the sub-samples of professional (P), technology-related (T) and creative (C) KIBS, respectively.
From Table 10.4, it emerges that a higher relative degree of vertical disintegration is associated with a lower age, whereas the relationship with firm
size is non-linear. Second, the variable DIS2007 is highly significant with a high coefficient: we interpret this as a clear sign of persistency in the purchase of business services by KIBS firms (Antonietti and Cainelli, 2012). These three results are also robust across all the different specifications.
When looking at our distance-based agglomeration variables, we first observe that the degree of vertical disintegration of each firm increases with the number of neighbouring firms clustered within a ray of five kilometres. In particular, a 1 per cent increase in neighbouring firms location (which means adding an average of 60 new KIBS) increases the level of vertical disintegration by almost 0.2 per cent. Second, the effect of the local concentration of industry on vertical disintegration increases with distance, instead of showing a clear path of spatial decay. One possible explanation for this result is statistical in nature, as it concerns the length of the ray we have chosen for data geo-referenciation. In this respect, it can be that a ray of five kilometres is too small for capturing the real process of spatial decay (if any) of the effects of the geographical concentration on KIBS boundaries. If this is the case, one possible solution is to extend the ray and analyse the process of spatial decay over wider distances.
Third, when we split the sample according to the type of KIBS, our previous results seem to hold particularly for the creative KIBS sub-sample, whereas the effect for P and T-KIBS is weaker. This result is a clear sign that KIBS tend to behave differently, according to their nature. In this respect, creative KIBS are not only the least integrated type of KIBS, but they are also the most present within the urban core of Milan. The need to rely on trusty and stable relationships seems to be more important for creative KIBS, i.e. for those activities which are more likely to develop complex, less standardized and high skill-intensive front-office activities.
Results from Tables 10.6 and 10.7 confirm and reinforce the previous findings. In this case, we find a positive correlation between vertical disintegration and the share of KIBS neighbouring firms, with much higher estimated coefficients with respect to the ones in Table 10.3. In this case, a 1 per cent increase in the share of neighbouring KIBS is related to an average 13 per cent increase in the degree of vertical disintegration, this effect ranging from a +7 per cent (significant at 10 per cent level) for T-KIBS to a +18 per cent for C-KIBS.
This chapter is an attempt to explore the relationship between the local concentration of industry and the vertical disintegration of KIBS firms. The main novelty of this relationship is derived from the possibility to geo-referenciate a large number of KIBS and define distance-based measures of urban density in the service industry. In this respect, we first localize each firm in the dataset, and then we define a series of rings within which we count the number of neighbouring firms. These variables are finally used as covariates in the OLS estimates of the degree of vertical disintegration, measured as the share of purchase of business services over total production costs.
Our estimates show that urban density matters in explaining KIBS firms' boundaries. In particular, we find that, once controlled for firm age, size and for persistency effects in input purchasing, a 1 per cent increase in the number of neighbouring KIBS within five kilometres is associated to an average 0.2 per cent increase in KIBS vertical disintegration. This effect also increases with distance: in particular, the effect of a 1 per cent increase in the absolute number of neighbouring KIBS passes from an average 1.2 per cent within the first one-kilometre ray to 1.4 per cent within the second ray (1–3 km), up to 2.2 per cent within the third ray (3–5 km).
Results are even stronger when we measure spatial agglomeration in relative terms, i.e. counting the number of neighbouring firms with respect to the total number of firms in the MA of Milan. In this case, a 1 per cent increase in the relative number of neighbouring firms within a ray of five kilometres is related to a 13 per cent increase in the share of purchased business service inputs. We find this effect to be particularly strong in the 3–5 km ray, where the marginal effect reaches an average value of almost 30 per cent.
Moreover, when we look at the knowledge content of KIBS, we interestingly find that these results particularly hold for creative KIBS, whereas the correlation for traditional professional and technology-related KIBS is weaker. This means that, when looking at the agglomeration-disintegration relationship within the advanced service industry, one should carefully account for the type of KIBS activity: in our case, we argue that the most creativity-based front-office service activities are the ones that mostly rely on an urban location in order to reduce transport costs and increase the probability to establish valuable face-to-face contacts.
The presence of non-linearities in the geographical localization of firms, of spatial correlation among observations, and of potential endogeneity issues in the OLS estimates, calls for a bit of caution in interpreting these results. For these reasons, we consider them as a starting point for the development of a future, deeper, analysis. To this purpose, one direction for future research could involve the definition and calibration of a spatial density function synthesizing the process of spatial decay at the urban level (Arbia et al., 2012). Second, the use of panel data would allow better control for the presence of (industry-specific and/or location-specific) fixed effects as well as for reverse causality issues and for factors affecting the location choice of KIBS.
1. We acknowledge with thanks Eleonora Di Maria and the participants to the 2011 Uddevalla Symposium for useful comments. We also thank Italo Mairo for technical support and three anonymous referees whose suggestions heavily contributed to improve the paper.
2. This chapter focuses on the relationship between increased fragmentation of production and the local trade of intermediates, thus excluding the more general issue related to the international fragmentation of production and the global dispersion of production activities (Jones and Kierzkowski, 2005).
3. On the theoretical ground, Baumgardner (1988) develops an interesting model in which the degree of specialization, i.e. the number of tasks or activities an individual provides, is explained by the level of the local demand, as proxied by the level of population in the local labour market area.
4. Although recently adopted in the quantitative economic geography literature, the idea of geo-referencing economic agents and then defining concentric circles around them for measuring the degree of spatial concentration is not new. Its roots have to be found, for instance, in the Chicago School of Urban Sociology (Park, 1925; Burgess, 1925) and in the so called ecological approach of the city (for a review see Graziano, 1996). We thank an anonymous referee for providing us these details.
Acs, Z. J., Braunerhjelm, P., Audretsch, D. and Carlsson, B. (2008), ‘The Knowledge Spillover Theory of Entrepreneurship’, Small Business Economics, 32: 15–30.
Amiti, M. and Wei, S. J. (2005), ‘Fear of Service Offshoring: Is it Justified?’, Economic Policy, 20(42): 308–47.
Amiti, M. and Wei, S. J. (2009), ‘Service Offshoring and Productivity: Evidence from the US’, The World Economy, 32(2): 203–20.
Andersson, M. and Hellerstedt, K. (2009), ‘Location Attributes and Start-Ups in KIBS’, Industry and Innovation, 16(1): 103–21.
Antonietti, R. and Cainelli, G. (2008), ‘Spatial Agglomeration, Technology and Outsourcing of Knowledge Intensive Business Services. Empirical Insights from Italy’, International Journal of Services Technology and Management, 10: 273–98.
Antonietti, R. and Cainelli, G. (2012), ‘Urban Density and Vertical Disintegration: The Case of Service Firms in Milan’, in Di Maria, E., Grandinetti, R. and Di Bernardo, B. (Eds), Exploring Knowledge-Intensive Business Services. Knowledge Management Strategies, Basingstoke, UK: Palgrave Macmillan: 214–30.
Antonietti, R., Ferrante, M. R. and Leoncini, R. (2012), ‘Local Spillovers, Production Technology and the Choice to Make and/or Buy. Empirical Evidence from Emilia Romagna Mechanical Industry’, Regional Studies, DOI: 10.1080/00343404.2012.657169.
Arbia, G., Espa, G., Giuliani, D. and Mazzitelli, A. (2012), ‘Clusters of Firms on an Inhomogeneous Space: The High-Tech Industries in Milan’, Economic Modelling, 29(1): 3–11.
Arzaghi, M. and Henderson, J. V. (2008), ‘Networking off Madison Avenue’, Review of Economic Studies, 75: 1011–38.
Barthelemy, J. (2001), ‘The Hidden Cost of IT Outsourcing’, MIT Sloan Management Review, 42(3): 60–9.
Baumgardner, J. R. (1988), ‘The Division of Labor, Local Markets, and Worker Organization’, Journal of Political Economy, 96(3): 509–27.
Bengtsson, L. and Dabhikar, M. (2009), ‘Manufacturing Outsourcing and its Effect on Plant Performance – Lessons from KIBS Outsourcing’, Journal of Evolutionary Economics, 19: 231–57.
Bettiol, M., Di Maria, E. and Grandinetti, R. (2012), ‘Creativity versus Standardization in Knowledge-Intensive Business Services: A Real Trade-Off?’, in Di Maria, E., Grandinetti, R. and Di Bernardo, B. (Eds), Exploring Knowledge-Intensive Business Services. Knowledge Management Strategies, Basingstoke, UK: Palgrave Macmillan: 120–36.
Cainelli, G. and Iacobucci, D. (2012), ‘Agglomeration, Related Variety and Vertical Integration’, Economic Geography, 88(3): 255–77.
Ciarli, T., Meliciani, V. and Savona, M. (2012), ‘Knowledge Dynamics, Structural Change and the Geography of Business Services’, Journal of Economic Surveys, 26(3): 445–67.
Den Hertog, P. D. (2000), ‘Knowledge-Intensive Business Services as Co-producers of Innovation’, International Journal of Innovation Management, 4(4): 481–528.
Di Maria, E., Grandinetti, R. and Di Bernardo, B. (2012), ‘Introduction’, in Di Maria, E., Grandinetti, R. and Di Bernardo, B. (Eds), Exploring Knowledge-Intensive Business Services. Knowledge Management Strategies, Basingstoke, UK: Palgrave Macmillan: 1–12.
Doloreux, D. and Shearmur, R. (2012), ‘Collaboration, Information and the Geography of Knowledge-Intensive Business Services’, Journal of Economic Geography, 12(1): 79–105.
Doloreux, D., Amara, N. and Landry, R. (2008), ‘Mapping Regional and Sectoral Characteristics of Knowledge-Intensive Business Services: Evidence from the Province of Quebec’, Growth and Change, 39(3): 464–96.
Doloreux, D., Freel, M. and Shearmur, R. (2010), Knowledge-Intensive Business Services. Geography and Innovation, Farnham, UK: Ashgate.
Earl, M. (1996), ‘The Risks of Outsourcing IT’, MIT Sloan Management Review, 37(3): 26–32.
Feenstra, R. C. and Hanson, G. H. (1996), ‘Globalization, Outsourcing, and Wage Inequality’, American Economic Review, 86(2): 240–5.
Feenstra, R. C. and Hanson, G. H. (1999), ‘The Impact of Outsourcing and High-Technology Capital on Wages: Estimates for the United States: 1979–1990’, Quarterly Journal of Economics, 114(3): 907–40.
Fujita, M. and Thisse, F. (2002), Economics of Agglomeration – Cities, Industrial Locations and Regional Growth, Cambridge, UK: Cambridge University Press.
Goldstein, G. and Gronberg, T. (1984), ‘Economies of Scope and Economies of Agglomeration’, Journal of Urban Economics, 16: 91–104.
Helsley, R. and Strange, W. (2007), ‘Agglomeration, Opportunism, and the Organization of Production, Journal of Urban Economics, 62: 55–75.
Holmes, T. (1999), ‘Localization of Industry and Vertical Integration’, Review of Economics and Statistics, 81: 314–25.
Klein, B., Crawford, R. and Alchian, A. A. (1978), ‘Vertical Integration, Appropriable Rents and the Competitive Contracting Process’, Journal of Law and Economics, 21(1): 297–326.
Koschatzky, K. and Zenker, A. (1999), ‘The Regional Embeddedness of Small Manufacturing and Service Firms: Regional Networking as Knowledge Source of Innovation’, Regions and Firms Working Paper No. R2/1999, ISI.
Li, B. and Lu, Y. (2009), ‘Geographic Concentration and Vertical Disintegration: Evidence from China’, Journal of Urban Economics, 62: 55–65.
Love, J. H. and Roper, S. (2001), ‘Outsourcing in the Innovation Process: Locational and Strategic Determinants’, Papers in Regional Science, 80: 317–36.
McLaren, J. (2000), ‘Globalisation and Vertical Structure’, American Economic Review, 90(5): 1239–54.
Miles, I. (2012), ‘KIBS and Knowledge Dynamics in Client–Supplier Interaction’, in Di Maria, E., Grandinetti, R. and Di Bernardo, B. (Eds), Exploring Knowledge-Intensive Business Services. Knowledge Management Strategies, Basingstoke, UK: Palgrave Macmillan: 13–34.
Miles, I., Kastrinos, N., Flanagan, K., Bilderbeek, R., den Hertog, P., Huntink, W. and Bouman, M. (1995), ‘Knowledge Intensive Business Services: Their Roles as Users, Carriers and Sources of Innovation’, Report to DG13 SPRINT-EIMS, PREST, Manchester, March.
Morgan, K. (1994), ‘The Exaggerated Death of Geography: Learning, Proximity and Territorial Innovation Systems’, Journal of Economic Geography, 4(1): 3–21.
Muller, E. and Doloreux, D. (2009), ‘What We Should Know About Knowledge-Intensive Business Services’, Technology in Society, 31: 64–72.
Muller, E. and Zenker, A. (2001), ‘Business Services as Actors of Knowledge Transformation: The Role of KIBS in Regional and National Innovation Systems’, Research Policy, 30: 1501–7.
OECD (2006), OECD Territorial Reviews: Milan, Italy, Paris: OECD.
Ono, Y. (2001), ‘Outsourcing Business Service and the Scope of Local Markets’, Federal Reserve Bank of Chicago, Working Paper No. 09.
Ono, Y. (2006), ‘Market Thickness and Outsourcing Services’, Regional Science and Urban Economics, 37: 220–238.
Piras, G., Dall'Erba, S. and Percoco, M. (2009), ‘Service Industry and Cumulative Growth in the Regions of Europe’, Entrepreneurship & Regional Development: An International Journal, 21 (4): 333–49.
Pirrong, S. C. (1993), ‘Contracting Practices in Bulk Shipping Markets: A Transaction Cost Explanation’, Journal of Law and Economics, 36 (2): 937–76.
Prencipe, A. (1997), ‘Technological Competencies and Product's Evolutionary Dynamics: A Case Study from Aero-Engine Industry’, Research Policy, 25: 1261–76.
Rama, R. and Calatrava, A. (2002), ‘The Advantages of Clustering: The Case of Spanish Electronics Subcontractors’, International Journal of Technology Management, 24: 764–91.
Rama, R., Ferguson, D. and Melero, A. (2003), ‘Subcontracting Networks in Industrial Districts: The Electronics Industry in Madrid’, Regional Studies, 37: 71–88.
Rodriguez, A. and Nieto, M. J. (2012), ‘The Internationalization of Knowledge-Intensive Business Services: The Effect of Collaboration and the Mediating Role of Innovation’, The Service Industries Journal, 32 (7): 1057–75.
Shearmur, R. and Doloreux, D. (2012), ‘Is There a Connection Between Geographic Clustering and KIBS Innovation?’, in Di Maria, E., Grandinetti, R. and Di Bernardo, B. (Eds), Exploring Knowledge-Intensive Business Services. Knowledge Management Strategies, Basingstoke, UK: Palgrave Macmillan: 193–213.
Smedlund, A. and Toivonen, M. (2007), ‘The Role of KIBS in the IC Development of Regional Clusters’, Journal Of Intellectual Capital, 8 (1): 159–70.
Stigler, G. J. (1951), ‘The Division of Labor is Limited by the Extent of the Market’, Journal of Political Economy, 59: 185–93.
Taymaz, E. and Kiliçaslan, Y. (2005), ‘Determinants of Subcontracting and Regional Development: An Empirical Study of Turkish Textile and Engineering Industries’, Regional Studies, 39 (5): 633–45.
Willcocks, L. P. and Lacity, M. C. (2006), Global Sourcing of Business and IT Services: Technology, Work and Globalization, Basingstoke, UK: Palgrave Macmillan.
Windrum, P., Reinstaller, A. and Bull, C. (2009), ‘The Outsourcing Productivity Paradox: Total Outsourcing, Organisational Innovation, and Long Run Productivity Growth’, Journal of Evolutionary Economics, 19: 197–229.
18.216.24.36