G L O B E L I C S. Working Paper. The development of aerospace clusters in Mexico. Javier Martínez Romero

Transcription

1 G L O B E L I C S W O R K I N G P A P E R S E R I E S T HE G LOBAL N ETWORK FOR E CONOMICS OF L EARNING, I NNOVATION, AND C OMPETENCE B UILDING S YSTEM The development of aerospace clusters in Mexico Javier Martínez Romero Working Paper No ISBN:

2 GLOBELICS The development of aerospace clusters in Mexico Javier Martínez Romero PhD candidate, Associate researcher at the Canada Research Chair on the Management of Technology, Université du Québec à Montréal

3 Abstract Introduction: High technology companies tend to cluster around knowledge-producing institutions (Braunerhjelm and Feldman, 2006). Aerospace follows this pattern, and there are well known examples of aerospace clusters located around large prime contracting aerospace producers (Niosi and Zhegu, 2005). A recent trend that is taking momentum is the setting up of manufacturing facilities by leading aerospace companies in Mexico. Objectives: The objective of this study is to investigate the attracting factors of Mexican clusters and to examine some of the country policy measures to incentive the aerospace sector, in order to assess the potential of Mexico to be a relevant part in this world-class high technology industry. Methods: A survey applied to selected aerospace firms in Mexico and interviews with regional governments are the information source. Results: Agglomeration forces in Mexican aerospace clusters are strongly related with manufacturing advantages. In addition, policy measures seem insufficient to encourage firms to undertake more complex activities. Conclusions: Even though advantages not related with innovation may explain the presence of firms in Mexico's aerospace clusters, these advantages should not be minimized given the tough quality standards of the aerospace industry, and if properly managed they may form the base for future development. 2

4 1. Introduction High technology economic sectors are often associated with economic success. Particularly, a common claim is that high-technology sectors are the main driving force of industrial growth in an advanced economy. These are thought to be economic engines able to positively impact other more traditional technology sectors. Thus, the importance of a high technology sector is not just its own contribution to GDP, employment, revenue or exports, but the way it impacts other industries in the economy. Given the importance of high-technology sectors, scholars are interested in uncovering the economic, institutional and social dynamics that characterize their functioning, while policy makers are eager to obtain practical knowledge related to these dynamics in order to set policy measures able to foster the creation and growth of this kind of activity. The opening of a manufacturing plant by the Canadian aerospace firm Bombardier at Querétaro, Mexico, was a highly publicized event by Mexican government officials. National and regional politicians announced the entry of Mexico into the world aircraft manufacturing industry. The fact is that Bombardier took a step that had already been taken by some other foreign aerospace firms: namely, transferring part of their activities to Mexican soil. It seems that this delocalization trend is gaining momentum. Apparently this implies that Mexico has the ability to enter a high technology industry and enjoy the benefits of that kind of sector. Nevertheless, high technology companies have requirements that are not easily met by developing countries. Specifically, high technology companies tend to cluster around knowledge-producing institutions (Braunerhjelm and Feldman, 2006). This is the case in ICT, biotechnology and other knowledgeintensive sectors. Aerospace is no exception to this phenomenon, and there are well known examples of aerospace clusters like Montreal in Canada, Toulouse in France, and Everett (WA) in the United States (Niosi and Zhegu, 2005). These aerospace clusters exhibit particular characteristics, like for instance, the presence of large prime contracting aerospace producers that act as anchor tenants (Niosi and Zhegu, 2010); in other words, they are a magnet for other aerospace firms. This trend of foreign aerospace companies setting up manufacturing facilities in Mexico, and the 3

5 promotion by national and regional governments of the localities where these firms establish as aerospace clusters arise the question about the possibility of a developing country like Mexico to develop a high technology industry like aerospace. Since aerospace activity tends to cluster, and requires specialized inputs and a strong government support, it is important to uncover if those conditions are met in the Mexican case, and if they seem to be conductive to master aerospace technology. The purpose of this paper is precisely to know if those conditions exhibit similarities with what happens in countries with an important aerospace industry. The specific questions addressed in this paper are: 1) Do Mexican aerospace clusters exhibit similar centripetal forces found in other well known aerospace clusters? 2) Are there firms that might be considered anchor tenants? 3) Are there differences among Mexican localities, regarding their attraction factors? And 4) Does Mexican government policy towards the sector, exhibit similar measures taken by other successful countries? We think these questions will give us some understanding about the potential of Mexico's aerospace clusters to be a relevant part in the value chain of this world-class high technology industry. A survey applied to selected aerospace firms in Mexico, interviews with regional governments, and government publications are the information source of this study. The rest of the paper goes as follows: Section 2 reviews the characteristics of clusters, agglomeration forces, and policy measures prevalent in successful countries in the aerospace sector. Based on the previous discussion and the conditions of Mexico's innovation system, general and empirical hypotheses are proposed. Section 3 shows data, methodology, and the results obtained. Finally, Section 4 presents conclusions and some policy implications. 2. Theoretical framework Several approaches have been used to analyze aerospace clusters. For instance, Beaudry (2001) studied the firms' rate of entry and the complementarities among these firms in aerospace clusters in England. Niosi et al (2005) put the accent in the institutional support given by the government in Canada in order to foster innovation capabilities to transit from a cluster into an innovation system in the case of Montreal. Others like Smith and Ibrahim (2006) propose a typology of aerospace clusters in the case of Wales in England. Using the notion of technological knowledge, Ehret and Cooke (2009) analyze the outsourcing patterns in British aerospace 4

6 clusters. More recent works like Niosi and Zhegu (2010) use the concept of anchor tenant to explain the rise and fall of aerospace clusters in the United States. Benzler and Wink (2010) advocate that in the near future, relocation of aerospace activity will be influenced by the presence of technological platforms in existing or potential clusters. There are at least two basic concepts underlying these approaches: first, the concept of cluster as an entity that is capable to attract and improve the performance of firms (Beaudry, 2001; Smith and Ibrahim, 2006; Niosi and Zhegu, 2010); and second, the notion of technological knowledge, in the sense of a complex learning process that requires specialized inputs (Niosi et al, 2005; Ehret and Cooke, 2009; Benzler and Wink, 2010). The approach taken here is to use the cluster concept as a guide, and couple it with other relevant approaches related to technological knowledge and innovation. In order to do so, we are going to briefly present the deconstruction of the cluster proposed by Martin and Sunley (2003) and how the resulting analytical dimensions can be enriched by coupling other concepts related with technological knowledge and innovation. Based on the deconstruction of the cluster concept made by Martin and Sunley (2003) it can be argued that any concept that attempts to understand the agglomeration of economic activity should include at least three relevant dimensions: 1) industrial boundaries, 2) geographical scope, and 3) the socio-economic dynamics that take place in the agglomeration. However, given that we are dealing with a high technology sector in a developing country, we propose to transform those analytical dimensions in the following way: 1) technological characteristics of the firms, 2) agglomeration forces, and 3) innovation and technology policy (ITP), and the institutional environment. Literature review regarding these analytical dimensions is presented in next three subsections Technological characteristics of aerospace firms This subsection has three objectives: 1) to discuss the identification of aerospace firms; 2) to classify these firms according to the technological intensity of their activities; and 3) to characterize aerospace clusters depending on the type of firms within them. 5

7 In terms of industrial classification, the aerospace industry has clear criteria. Firms that are registered with codes of aerospace activity are considered as aerospace firms. Nevertheless, it is important to note that sometimes firms are active in other industries besides aerospace. This situation is common for small metal shop firms that are suppliers to a range of industries. It is also possible to have system and sub-system integrator firms active in other sectors, like for instance, the Canadian firm Bombardier who is also involved in terrestrial transportation equipment, or the American General Electric, who has a wide range of product outside of aerospace turbines. However, most integrator firms have clear aerospace divisions. Source: Niosi and Zhegu (2005: 8) In terms of the technological intensity, the aerospace pyramid (see Figure 1) classifies all these aerospace firms in layers, the highest of which consists of the systems integrators (the ones that assembly the whole aircraft), the most important firms in the industry in terms of technological intensity 1. As firms go down in the pyramid their technological intensity is expected to diminish. 1 These subsystems are also called modules. According to Brusoni and Prencipe (2001), Niosi and Zhegu (2005) and Frigant and Talbot (2005), the aircraft is a complex system in which the system is the whole aircraft and it is made up of different subsystems or modules arranged in a hierarchical fashion. These subsystems are in turn made up of other subsystems. 6

8 Although this classification is very useful, attention should be paid when classifying subsidiaries in foreign countries. The mere existence of a modular architecture implies the standardization of interfaces. While it is true that new technological developments have been taking place in the production of airplanes (Benzler and Wink, 2010), it is also true that some of the processes are standardized and well understood (Esposito, 2004; Kehayas, 2007). According to the logic of the Product Life Cycle, - pioneered by Vernon (1966) - when standardization occurs it opens the door to re-localization of activities to cost-efficient locations. Proximity to the US, a leader in aerospace industry, plus the manufacturing experience and infrastructure of Mexico, makes it a likely candidate for foreign firms to relocate their most codified processes. For this reason, it is important to take care when classifying foreign subsidiaries; two situations could arise. First, the subsidiary may indeed be property of the parent company, but its activities may pertain to products or activities located in a lower Tier level compared to the parent's products and activities. Indeed, Ehret and Cooke (2010) stress that even if outsourcing to third parties has been introduced in the aerospace sector, some big firms, like Airbus, continue to manufacture many components that are non-core and non-strategic (p. 312). The other situation is when the subsidiary is in fact engaged in products related with the Tier of the parent company, but while the latter is concentrated in knowledge-intensive activities (like R&D) the former is concentrated in manufacturing. Therefore, it will be misleading to consider a subsidiary as a Tier 1 firm, just by the sole fact of the parent company to be so. For this reason, it is mandatory to introduce a knowledge related concept like the anchor tenant developed among others by Philips (2002), Feldman (2003), and Agrawal and Cockburn (2003). According to the last authors, the anchor tenant is a large, locally present firm that is (1) heavily engaged in R&D in general and (2) has at least minor absorptive capacity in a particular area (p. 1229). Niosi and Zhegu (2010) apply that concept to explain how aerospace clusters in the US have flourished or banished depending on the ability of their engine firm or firms to act as anchor tenant. These authors show how the acquisition of Lear Jet (Located in Wichita, Kansas, US) by Bombardier, meant the transfer of R&D activities to Montreal, Canada; deprived from R&D-intensive activities, the system integrator was no longer able to exert the function of anchor tenant as it used to (p. 272). 7

9 The presence of an anchor tenant firm heavily influences the type of firms that may establish in an aerospace cluster. Niosi and Zhegu (2010) found that there was a high correlation between the existence of an anchor tenant and the presence of other innovator firms in the cluster (p. 274). Also, the specialization of these other innovator firms may be influenced by the anchor tenant firm (Niosi and Zhegu, 2010). It is important to mention that different sub-system or modules of the aircraft (like wings, fuselage, turbines, avionics, etc) require different capabilities. Thus, when a cluster has an anchor tenant firm dedicated to a specific aircraft module, these other cluster firms may be specialized according to the type of module the anchor tenant produces. The anchor tenant has influence not just in attracting to the cluster other innovator firms and their respective specialization, it actually can promote the creation of research laboratories and university programs dedicated to aerospace (Niosi et al, 2005) 2.2. Agglomeration forces in aerospace clusters This subsection has two objectives: 1) to describe the nature of the more prevalent centripetal forces at play in aerospace clusters; and 2) to discuss the spatial scope of these forces, or in other words, the geographical limits of an aerospace cluster. The idea that some economic activities derive a number of benefits from being geographically concentrated is not new. A pioneering formulation of the phenomenon was made by the British economist Alfred Marshall. According to Marshall (1890), there are three main agglomeration economies that firms operating in one sector achieve by locating together: the first one has to do with attracting of specialized suppliers; the second with attracting specialized workers; and the third with the creation of an industrial atmosphere in which innovative ideas are easily transmitted among local agents. Implicit in the argument is that these economies would not be attained if firms were geographically scattered. According to Niosi and Zhegu (2005), the creation of a labour pool by an anchor tenant firm is the main Marshallian agglomeration force in aerospace clusters. Contrary to the idea put forward more recently by Italian scholars working on industrial districts (e.g. Brusco, 1990; Pyke and Sengenberger, 1990), and the work of Michael Porter (1990; 2000) in clusters, on which firms within a cluster are complementary to each other in terms of the 8

10 product, the aircraft modular architecture allows firms working on different modules to be located in different clusters. In this way, while the final assembly of an Airbus aircraft takes place in Toulouse, France, the different modules (like wings) can be (and actually some of them are) produced in other places. Therefore, the idea of firms profiting from knowledge spillovers just by the fact of being located in a cluster is flawed in the case or aerospace, or for that matter for any other sector. As Giuliani (2007) underlines, even for technological mature sectors, the ability of a firm to profit from spillovers depends on its absorptive capacity. Also, it is important to differentiate between business and knowledge networks. The former are the normal links that a firm establishes to exchange inputs and outputs, while the latter are channels by which knowledge may flow. Giuliani and Bell (2005) argue that big firms with a developed absorptive capacity usually are the gatekeepers of the knowledge networks. This means that they mediate between the cluster external knowledge and the way in which that knowledge diffuses within the cluster. Thus, within a cluster, different firms have different abilities to profit from knowledge spillovers. This is more so given that a lot of business and knowledge flows take place among firms located in different clusters (Niosi and Zhegu, 2005). Therefore, the main attracting forces in aerospace cluster have to do with the presence of a specialized labour pool and the presence of big firms. These are the ones that generate critical knowledge and the ones that are able to bring to the cluster external knowledge. Moreover, like the anchor tenant concept suggests, these firms are able to influence suppliers and research laboratories by means of their sophisticated demands. In terms of geographic scope, limits are a difficult issue in aerospace. The works of Beaudry (2001), Smith and Ibrahim (2006) and Cooke and Ehert (2009) are very revealing of this respect. According to the second authors, there are five aerospace clusters in the UK (they cite as a source a 2001 report from the Department of Trade and Industry): North West, South West, East Midlands, North Ireland, and Wales (pp ); yet Cooke and Ehret (2009) argue that there are two aerospace agglomerations in Wales alone, one in the north and one in the south (pp ). To make matters more complex, Beaudry (2001) in her study of cluster effects on aerospace firms growth, proposes counties as the relevant regional dimension.2 This author 2 Beaudry (2001) uses the European NUTS (Nomenclature des Unités Territoriales Statistiques) as the 9

11 identifies 5 counties in Wales Clwyd, Powys, Mid Glamorgan, South Glamorgan, and West Glamorgan- with aerospace activities (p. 434), from which only three Clwyd, Mid Glamorgan, and South Glamorgan- were identified as having a strong fixed effect (pp. 417, 436), thus, presumably considered as clusters. None of the authors mentioned, made completely clear why they chose a determined geographic unit. It seems that the criterion followed by Smith and Ibrahim (2006) is based on broad regional political administrative divisions3 (let s remember that Wales is in itself a country) that ultimately may have the political and economic resources to support clusters. In the case of Cooke and Ehert (2009) it seems that their choice is based on mere geographical proximity4. Since these authors are concerned with learning and local systems, it may be that they consider proximity as a condition to interact. For Beaudry (2001) the approach is different. This author doesn t assume clusters as given. On the contrary, she explored which of the counties do have a positive impact on firms growth. Once that exercise done, just some of the counties were identified as having a fixed effect. In the description about how Montreal came to be an aerospace cluster and later an aerospace regional innovation system, Niosi et al (2005) adopt the metropolitan region as a unit of analysis. At the same time, he recognizes both national and provincial government (outside the metropolitan area) intervention to foster innovation in Montreal. As Niosi et al (2005) stress, one of the main agglomeration forces at play in the case of aerospace is specialized labour. Thus, it seems that the scope of the cluster has to be delimited in some way by the possible mobility (or absence of mobility) of this specialized work force. In this sense, the metropolitan area may provide an adequate geographical scope for clusters in aerospace. geographical unit. According to this author the UK had 65 NUTS at level 3 (the maximum disaggregation level), and for Wales alone, the number is of 8, from which 5 had aerospace firms. 3 Smith and Ibrahim (2006) explain that if [a] region had more people employed in that industry [in this case aerospace] that one would expect given the size of the region in terms of its labor force it can be considered a cluster in that industry (p. 362). Nevertheless, how the regions are pre-defined has an impact on final results. In this case it seems that the whole of Wales was taken as a region, and as such it prevented further divisions within Wales itself. 4 Cooke and Ehret (2009) do not give an extensive explanation for their claim of two agglomerations. Instead, they show a map of Wales in which the agglomerations are visually identified. 10

12 2.3. Innovation and technology policy (ITP), and the institutional environment This subsection 1) sketches the policy measures taken by once newcomer countries to the aerospace sector, and 2) briefly describes the innovation system conditions in Mexico. For the case of Canada, it is interesting to see how military procurement was at the base of its aircraft capacity building. Overhaul and maintenance, and later the licensing of British aircraft military models by British subsidiaries marked the beginning of that industry in Canada (let s not forget that Canada was a part of the British dominion) (Niosi et al, 2005). Other countries with conditions more similar to Mexico, like Brazil and Argentina developed their aircraft industries based on military impulses too, although only Brazil was successful (Hira and De Oliveira, 2007). These countries also started by acquiring licenses to produce foreign planes. Whatever the historical reasons for the absence of a military aircraft industry attempt in Mexico, the fact is that the country starts with only a very limited experience, based in the overhaul and maintenance of military and civil aircraft bought from foreign countries. One may add that experience in the manufacturing of motor vehicles and auto-parts are also important assets. The countries mentioned above could learn from licensing old military and civil aircraft models. However, that possibility is not clear for Mexico. As it was said, Mexico is not interested in developing a domestic military aircraft sector (at least apparently), which left out the possibility of licensing military aircraft production. Thus, the country has to depend on the civil sector alone. These historical examples bear one implication: All countries that in one moment attempted to developed the aerospace sector had to rely at the beginning in the transference of foreign technology by leading firms. The next issue is what kind of technology is more appropriate to seek. According to Goldstein (2002): the traditional trajectory for a developing domestic aerospace sector is a three stage process: first, countries begin with co-production agreements; second, as the industry develops, a viable set of subcontractors develops and finally, the domestic industry is capable of putting all the pieces together and become a final assembler of complete aircraft. 5 Nevertheless, the recent 5 Here quoted from McGuire et al. (2010: 368). 11

13 Japanese example shows that pursuing just one (or some) part of the plane could be a good strategy, instead of trying to build the whole aircraft. The Japanese firm Mitsubishi is now a world player in the design and construction of wings, thanks to the deal made with Boeing. With the high degree of modularity (see section 1.2) and high costs, it is inconceivable for one firm or a new comer country to master all the knowledge and technologies needed to build a new plane. Therefore, concentrating in just one of those modules could be a good strategy. All the countries mentioned above have one characteristic in common; in the early moments of the development of their aircraft industry, there were local-owned firms supported by the state. For instance in the case of Brazil, Embraer was a state enterprise, and was privatized later when it had already achieved a good technological degree. The Canadian firm Bombardier has benefited from government program supports at different moments of its life. Therefore, two implications are drawn in this case; first, government support is an indispensable condition since high costs make new attempts unprofitable; and second, national-owned enterprises seem to be a necessary condition for the further development of capabilities within the industry. Summarizing so far, we have four important historical implications for the development of an aircraft industry in a new comer country: 1) in a first phase, technology have to be licensed or transferred from abroad; 2) in an early moment of the development (perhaps not at the beginning) a national-owned firm should enter the industry; 3) this national-owned firm (or firms) should be specialized in one of the main modules of the plane (or in the assembly of the whole plane, although this option is more difficult); 4) state support has to be present in the different phases of development. For its success, an ITP depends not just in taking the right steps at the right time; it depends also on the well functioning of the innovation system. The national innovation system approach (Freeman, 1987; Lundvall, 1992) and more recently the regional innovation system (Cooke and Morgan, 1998) and the sectoral innovation system (Malerba, 2002) put the emphasis on organizational and institutional patterns needed in order to innovate. This means that firms don t innovate in isolation, on the contrary, there are numerous external factors that contribute either 12

14 directly or indirectly to this innovation effort, and in some cases innovation is originally conceived and realized through collective effort. Although some advances can be seen in innovation activities in Mexico, recent assessments of Mexico s innovation system reveal its still low level of development. In a comprehensive study about Mexico's innovation system, Dutrénit et al (2010) show the low tendency of Mexican firms to do innovative activities in a sustainable fashion (p ). Also, when these efforts occur they are more oriented to adapt foreign technologies than to create new ones (Dutrénit et al, 2010: 87). In the same token, links among firms and other relevant agents are limited. The 2009 OCDE report on Mexico also shows low indicators and lack of adequate measures to tackle innovation problems. For instance, the OCDE report shows that the investment in R&D as a percentage of GDP is at 0.5% (where business R&D plays a particularly small role), versus an OCDE average of over 2% (p. 17). Therefore, we have a scenario in which investment in innovation is weak, relationship between higher education institutions, research centers and industry is poorly developed, and the main focus of government is on regulatory and infrastructure issues, with little attention given to knowledge-related factors (OCDE, 2009: 15). Other important aspects are the regional and sectoral issues of the innovation system in Mexico. While there are some efforts at the state level to foster innovation, the fiscal scheme of Mexico concentrates resources at the federal level; thus, efforts by states are conditioned to a great extent on federal funding (OCDE, 2009: 15). An important issue regarding state support is the duplicity of efforts. Since any public program claiming to support a high technology industry requires important financial resources, it is unlikely that Mexican states, and the federal government itself, could sustain such an industry in several states at the same time. In terms of support for specific economic branches, the Sectoral Economic Program of the Mexican Economic Ministry doesn t contain a specific program for the aerospace sector. It only mentions the desirability for the Mexican economy to upgrade to a high value added sector like aerospace and aeronautics, among others (p. 50). 2.4 Research hypotheses 13

15 Based on the discussion of the three previous sections, we propose a series of general hypotheses and then some empirical hypotheses regarding the questions posed in the introduction section. General hypothesis 1: The relatively high codification of the technological knowledge base of the aircraft is reflected in part in the high degree of modularity that exhibits the production of this artefact. This is true also for the different sub-systems that made up the aircraft. Given this modularity, a modified version of the PLC for aerospace, predicts that system and sub-system integrators have the possibility to transfer some of their simpler activities to low-cost locations while keeping integration activities in their home locations. General hypothesis 2: Initial conditions in countries with underdeveloped innovation systems (like Mexico) cannot procure all relevant capabilities, knowledge and infrastructure needed to sustain the production of complete aircraft systems, sub-systems or R&D activities. General hypothesis 3: Given the low technological profile of aerospace activities (at least in an initial state) more likely to be transferred to a low-cost country like Mexico, the main decision factors for firms looking for a place to establish there, are related to traditional manufacturing advantages rather than to Marshallian agglomeration forces. General hypothesis 4: Given the stringent standards of the aerospace sector, the potential transfer of activities (as simple as they may be) by foreign firms to their subsidiaries requires certain degree of capabilities and infrastructure in the host locality. Question 1: Do Mexican aerospace cluster exhibit similar centripetal forces found in other well known aerospace clusters? Hypothesis 1: Aerospace firms are attracted to Mexican clusters because of reasons related with low cost operations, and the manufacturing capability of the country. Question 2: Are there firms that might be considered anchor tenants? 14

16 Hypothesis 2: New to the world innovation in aerospace firms located in Mexico will be minimal in the short and medium term. However, new to the country processes and products will certainly be introduced. Hypothesis 2a: When innovation does occur, complementary ideas and capabilities may be related to foreign sources of knowledge. Hypothesis 3: Being part of a corporate company, subsidiaries in Mexico have more supply and demand links with their corporate location than with local agents. Question 3: Are there differences between Mexican localities regarding their attraction factors? Hypothesis 4: When some local input links do occur, firms located in inland cities may exhibit more input links when compared to firms located in border cities. This is because the ampler industrial base observed in inland cities. Question 4: Do Mexican government policy towards the sector exhibit similar measures taken by other successful countries? Hypothesis 5: The lack of a sectoral program dedicated to the sector and the underdeveloped innovation system of Mexico, makes unlikely the appearance in the near future of: The development of a specialized infrastructure that encourages foreign firms to take the risks to transfer more complex activities to their subsidiaries. The targeting of selected aeronautic technologies on which firms seem to be more capable of learning. The concentration of support in just certain clusters. The creation of a national-owned firm in the top Tiers of the pyramid. 15

17 3. Empirical research 3.1 Data Prior to the field application of the Questionnaire that was carried on from May 25 th to June 28 th of 2009 at 30 aerospace firms in five cities located in four states; the Ministry of Economy had identified 161 companies distributed in 15 Mexican states giving employment to approximately 20,000 people. Attempting to cover all the localities and companies was neither necessary nor practical. Since the focus of this research is on the agglomeration forces and characteristics of aerospace clusters, the approach taken here was to concentrate on localities that exhibit some features that qualified them as a potential cluster. As was explained in section 1.3, the analytical dimensions for this study are: 1) firms technological characteristics 2) agglomeration forces and 3) ITP and the institutional environment. Based on those dimensions, three criteria were used to select the locations for the field research. The first step in analyzing the structure of a cluster is the mere existence of such a structure. Given that firms are the most important organizations in aerospace clusters, the 1) total number of aerospace firms that a location hosts is the first criterion. The aerospace industry is organized in layers of system integrators, sub-system integrators, and numerous specialized suppliers. As we climb the layers, the firms are supposed to be more knowledge intensive. That knowledge generation serves as an attractor to other firms, and as such it becomes an agglomeration force. A system integrator firm is characterized for being a knowledge intensive organization. This system integrator firm can have several facilities scattered around the world. It is important to note that the technological content of the activities carried on in those different facilities can vary broadly. Nonetheless, a priori, the reputation of the firm is taken as a signal of its technological level. Therefore 2) the reputation of the aerospace firms in the location is the second criterion. In other parts of this text it was explained that the aerospace sector requires a strong commitment from government. Since it is a high technology sector it requires support at the financial, infrastructural and educational levels. Thus, 3) the institutional support in terms of policy, infrastructure, and education, that characterize the location in general and the aerospace industry in particular, is the third criterion. 16

18 Regarding the number of firms, TABLE 1 shows the distribution of the aerospace firms in 15 Mexican states. TABLE 1. Aerospace firms by state State # firms State # firms Baja California 49 Distrito Federal 6 Sonora 26 Jalisco 5 Nuevo León 19 Estado de México 5 Chihuahua 14 Aguascalientes 2 Tamaulipas 10 Puebla 2 Querétaro 8 Yucatán 2 Coahuila 6 Guerrero 1 San Luis Potosí 6 Total 161 Source: Promexico 6 (2009) PowerPoint presentation with data from the Ministry of Economy There was no unique criterion to set apart the states which (by their number of firms) could eventually be identified as an aerospace cluster. However, it seemed intuitive to set the lower threshold by including either Tamaulipas or Querétaro (10 and 8 firms respectively). Given the attention Querétaro was receiving at that moment we decided to set the lower limit to include Querétaro. Therefore, initially we selected Baja California, Sonora, Nuevo León, Chihuahua, Tamaulipas and Querétaro. In terms of the reputation of the aerospace firms the issue was somehow clear. Based on the data of Promexico (2009), Baja California stood first with the presence of companies such as Honeywell Aerospace, Gulfstream, Lockheed Martin Aeronautics, Rockwell Collins, and GKN Aerospace. In a second group, the states of Querétaro and Chihuahua counted with firms like Bombardier, Messier Services and ITR in the former, and Honeywell Aerospace, Labinal, and Cessna in the latter. Following the list, the states of Sonora and Nuevo León had within their borders companies like ESCO turbines and Goodrich in the former, and MD Helicopters and Frisa Wayman Gordon in the latter. At the time of this evaluation there was little information about the state of Tamaulipas, but recent accounts reveals only one medium profile company like Chromalloy. This information leads us to explore the institutional environment prevailing in the states of Baja California, Querétaro, Chihuahua, Sonora, and Nuevo León. 6 Promexico is an office dependent of the Ministry of Economy, which is in charge of both, promoting Mexico as a recipient of foreign investment, and helping Mexican firms that want to go international. 17

19 By means of their web pages, regional development offices in Baja California, Querétaro, and Nuevo León were the ones that stood up with more information about the firms, institutions and activities related to the aerospace sector. The three regional development offices of these states, were promoting three specific aspects of their respective clusters: 1) existence of a business chamber grouping aerospace firms, 2) the setup of education and training programs targeted to the sector in universities and technical schools, and 3) the support of state government in attending and promoting the sector. Information about Chihuahua and Sonora was scanty. With just the preliminary information mentioned so far, Baja California and Querétaro were the states that have the highest potential for developing an aerospace cluster. Accordingly, those two states were automatically selected to be included in the field work. Even if there was scanty information regarding the institutional support and promotion of the regional development office in Chihuahua, we decided to keep it for the research because of the important firms established there, and because of the importance of the state in the national economy in general. If the contribution of Chihuahua to the national economy is important, the Nuevo León contribution is even greater; also, the educational institutions in that state are considered among the best in the country. Therefore we included Nuevo León in spite of the lack of big names located there. The few high profile firms in the state of Sonora, plus the scanty evidence about the infrastructure targeted to support the sector were factors for not including it in the research. Once the four states were selected, the second step was to select the cities. In Baja California two cities accounted for the majority of firms, Mexicali (the state capital) and Tijuana (which, together with Ciudad Juárez is the most important border city in terms of manufacturing). In the state of Querétaro, some firms were scattered around and others were within the capital city of Querétaro. Choosing the city was not a problem. The metropolitan area of Monterrey that is comprised of several counties is where the aerospace activity takes place in Nuevo León. Although no firm is located in the actual city of Monterrey, all of them are accessible by car from Monterrey. The state of Chihuahua has two main cities and industrial centers, Chihuahua (the state capital) and Ciudad Juarez. Although preliminary information showed that Ciudad Juarez had some aerospace activity, it turned out that some firms were mislabelled as aerospace and the others were only partially involved in aerospace. The city of Chihuahua is both the host 18

20 of important aerospace firms, and the target of state government support. All firms in the city of Chihuahua were located within the city. Once the cities of Mexicali, Tijuana, Querétaro, Monterrey, and Chihuahua were chosen, the next step was to choose the firms. For Mexicali and Tijuana, the information the regional development office provided regarding the firms was very complete. The interest of the research is centered on big firms with a substantial activity in aerospace. For that reason we targeted firms with more than 100 employees and with a major involvement in aerospace activity. All firms with those characteristics contained in a report of the Baja California government were asked to participate in the research. In the end, 15 firms agreed to respond to the questionnaire; 8 from Mexicali and 7 from Tijuana. The process was similar for Monterrey, in which 4 firms took part in the study. It turned out that most of the firms in Monterrey were suppliers that were involved in other sectors as well, thus not a segment in our priorities. For Querétaro, the process was similar and 5 firms agreed to respond the questionnaire. Finally, thanks to the guidance of the regional development office of the state of Chihuahua we contacted 6 important firms that were not initially contemplated. After my arrival, the regional development offices of the four states were cooperative and provided guidance about which firms were worth of asking to participate. Thanks to that information, some smaller firms with a very high involvement in the industry were also included. In the end, 30 firms distributed in five cities answered the questionnaire in a face-to-face interview (see map below). 19

21 MAP 1. Five selected cities in four Mexican states Method The Global outsourcing and R&D best practices in the aerospace industry Questionnaire 7 is the main information source of this research. This tool was designed to gather information directly from aerospace firms in Mexico. Professor Jorge Niosi (research director), Professor Majlinda Zhegu and I collaborated in the elaboration of the questionnaire. The ample experience in the aerospace sector of Professors Niosi and Zhegu was of great help in identifying the cluster issues which firms in this sector face. The questionnaire and the research itself is part of an ampler research project that seeks to compare the development of the aerospace industry in different developing countries. Part I of the Questionnaire asks for general information about the size, ownership, markets and products of the firms. Section II.1 is about the firms innovation and the sources of ideas for those innovations. Additionally, Section II.2 asks for the incentives that firms have had to locate in the places where they are currently located. Part III is about the role of the firm in the industry, and aspects of communication. 7 The Spanish version that was actually used is called Cuestionario acerca de las prácticas de subcontratación e investigación y desarrollo de la industria aeroespacial en México. 20

22 Other primary data sources include four semi-structured interviews with the four regional development offices in charge of promoting the aerospace industry in the four selected Mexican states Baja California, Querétaro, Nuevo León, and Chihuahua-. Secondary data sources include aerospace publications of the Mexican Ministry of Economy. Since most of the questions included in the Questionnaire represent categorical statistical variables, descriptive statistical tools are used to illustrate the clusters characteristics. When using cross tabulation, the chi-square test for independence of two variables 8 is presented. Given the size of the sample, in some cases the expected count in some cells is lower than the recommended (usually 5 counts). For this reason, the Yates correction for continuity statistic 9, a non parametric test, is also shown. This non-parametric test returns a p-value using a Monte Carlo simulation (Verzani, 2005) Results There are two pieces of information in the survey that are relevant to answer question 1 of this paper. First, there is a question that lists possible local advantages, and asks firms if they benefit from those advantages or not. Second, there is an open question about why the firm chose that specific cluster to establish in the first place; the diverse reasons given by the respondents were compared and grouped. TABLE 2. Local advantages Advantages Answer Counts Column % Labour No % Yes % Industrial Areas No % Yes % Suppliers No % Yes % Clients No % Yes % 8 The statistical software used to obtain this statistic was SPSS. 9 The statistical software used to obtain this statistic was PSPP and verified with the R program (both free programs). 21

23 Policy Co-location Infrastructure Universities and Research Centers Financial advantages Incentives No % Yes % No % Yes % No % Yes % No % Yes % No % Yes % No % Yes % TABLE 2 captures the local advantages that firms consider they have by being located in a Mexican aerospace cluster. From the ten local advantages listed, six were identified as such by more than 50% of the firms. The local advantage most cited with 93.3% of positive answers is the Infrastructure. This means that almost all surveyed firms considered that the infrastructure of the locality provided an advantage. By Infrastructure the questionnaire makes reference to the transportation and telecommunications infrastructure. The existence and quality of Industrial Areas was considered by 83.3% of the firms as an advantage. The third most cited advantage was the Labour Force with 86.7%. Respondents usually complemented their answer by adding that the workers were skilled and accustomed to the factory environment. The fact that the locality host universities and research centers was the fourth most cited advantage with 76.7%. It is important to mention that almost all the affirmative answers to this question did so because of the university and not because of the research centers. The fifth advantage was the existence of policy measures with 63.3%. Virtually all the respondents were thinking about the maquiladora program 10 when answering positively to this question. This policy measure can be considered as a horizontal policy because is not specifically targeted to any sector in particular. The sixth most cited advantage was to be Co-located with other aerospace firms. Half of the firms answered yes while the other half said they did not find any advantage of being close to other aerospace firms. The items that received less positive answers were Incentives, Clients, Financial Aid, and 10 The maquiladora is a fiscal regime established in Mexico back in the 1960s that allows duty-free inputs under the condition of subsequent export. 22

24 Suppliers. Particularly the last three items have a negligible percentage of positive answers. TABLE 3. Former attraction factors Attraction factors Answer Counts Column % US proximity No % Yes % Low labour costs No % Yes % Low operation costs No % Yes % Owner's region origin No % Yes % Experience in industrial sectors No % Yes % Another important piece of information is TABLE 3, which shows the original attraction factors that firms took in account before establishing in Mexico. This was an open question in which similar answers were grouped (This may explain the lower percentages compared to TABLE 2). From the several answers given to this question we are going to present the five most prevalent. Of all surveyed firms, 50% declared that Proximity to the United States was among the reasons for establishing in Mexico. This is the attraction factor with major prevalence in aerospace firms in Mexico. The previous Manufacturing Experience of the Mexican localities was the second most cited reason with 26.67%. It is important to note that this experience is not necessarily related with aerospace. The Low Cost of the Labour Force was the third most cited attraction factor with 23.33%. Low Operation Costs and Owners original location follow with 20% each. As hypothesis 1 suggests, the centripetal forces that Mexico s aerospace clusters exhibit, are related to manufacturing advantages. Specifically, the advantages more valued by aerospace firms are related to the 1) industrial infrastructure, 2) the skilled labour force, and 3) the low operation costs. Regarding the first aspect, it is clear that the transport and telecommunication infrastructure [TABLE 2, 93.3%], AND the presence of industrial areas [TABLE 2, 83.3%], procures firms with the facilities needed to carry on manufacturing activities, and gives the overall impression that Mexican localities have a relevant experience in manufacturing [TABLE 23

25 3, 26.7%]. On the second aspect, the presence of a labour pool that in addition to be cheap [TABLE 3, 23.3%] is it considered skilled [TABLE 2, 86.7%], makes possible the undertaking of manufacturing processes that sometimes demand technological sophistication. The presence of universities in these localities adds to this capacity [TABLE 2, 76.7%]. In addition to the two mentioned aspects, the presence of the maquiladora policy [TABLE 2, 63.3%] keeps operation costs on low levels. In order to answer question 2 of this paper, the survey has information related with the new products' degree of novelty (world, country, firm), as well as the knowledge links that in some way helped to developed that novelty. There is also information about the inputs and outputs flows of the firms, which will be very useful to explore their degree of connectedness to the local environment. TABLE 4. Novelty degree of new products Degree of the novelty introduced Counts % None % Firm % Country % World % TABLE 4 shows that almost all firms produced at least one new product in the last three years. Only 20% of the firms declared that the new product they manufactured was a world novelty. This does not mean that these firms conceived and design the product. The participation in the world novelty of these firms was due to either creating a novel process to manufacture a product designed (and even manufactured) elsewhere, or by manufacturing for the first time a prototype, in other words a product that had not been manufactured before. According to some personal communications with interviewees, this is the result of optimization strategies followed by foreign firms, in which the subsidiary is in charge of not just manufacturing, but also of the design of the process, while the parent company gets more concentrated in design, testing and prototyping. We have to wait to know if this trend consolidates or not. It is important to note that the design of the manufacturing process requires certain advanced skills. 24