1. Applying the core-periphery framework to the New York garment industry

The core-periphery framework can be applied in a variety of ways, as illustrated by the diversity of empirical studies mentioned above. Because core-periphery systems are organized on the basis of tasks and employment differentiation within a given production process, the case study focuses on one industry. This choice should enhance the visibility of core-periphery systems by highlighting the organization of work, within firms and between firms, in the production process of a given product. In addition, a specific industry provides a research setting homogeneous enough to identify key economic and social dynamics supporting the diffusion of new organizational forms. However, the general features of the core-periphery model may materialize quite differently from one industry to the other. We might expect to find the central principles associated with the four employment systems, but not necessarily their "ideal" representations, within a given industry. Industry-specific characteristics might also influence the shape of core-periphery systems, such as the relative importance of the core and the periphery, as well as their scope of diffusion. Thus, the empirical study should allow for a careful weighting of industry-specific versus general characteristics in drawing conclusions about core-periphery systems.

In that perspective, the choice of the New York garment industry provides several advantages. First, this industry is subject to general trends toward increased market diversity and volatility, globalization, as well as the diffusion of new technologies, that are currently transforming a majority of industries. By looking at how New York firms adjust to these new trends, we may draw some insights on more general changes in firms' organizational strategies. Second, this industry has long been organized in networks of specialized firms enhancing its ability to respond to market changes. Because network forms of organization tend to diffuse in a variety of industries, the study of New York firms can help us identify some key dynamics of this new organizational forms as well as their employment implications. Finally, the New York garment industry is subject to strong competitive pressures resulting both from economic factors, such as the ease of entry into the industry, low cost foreign competition, and the growing market power of big retailers, and from social and institutional factors, such as the availability of an abundant immigrant labour force and the declining influence of labour unions. These intense competitive pressures have lead to deteriorating working conditions in the local industry, accompanying the erosion of the New Deal system aimed at regulating competition on a tripartite basis. In that perspective, the New York garment industry appears as an extreme manifestation of broader trends in employment and its regulation. Thus, following Mollenkopf and Castells [1991], we consider New York as a laboratory of the postindustrial economy, a condensed manifestation of broader trends that may be more diffuse and less visible in other industry settings.

The study of the New York garment industry combines two levels of analysis. First, an industry-wide approach allows the identification of characteristics of the economic, social and institutional environment in which local firms operate, as well as their main changes over the last two decades. This approach is based on an extensive collection of secondary data as well as interviews and informal discussions with over 40 representatives of trade unions, employer organizations, joint labour-management associations, training institutions, state officials and industry experts completed in 1994 and 1995. Second, a firm-centered approach reveals the new organizational forms that dominate the local industry. This approach focuses on inter-firm relations and inter-firm differences in employment characteristics, based on interviews in 40 local firms. The methodology used as well as the general characteristics of the firms studied are presented in the appendix. Finally, the general trends supporting the diffusion of core-periphery systems in the New York garment industry are isolated from industry-specific characteristics in order to provide some insights on broader changes in organizational forms and employment conditions in other industries.

2. Economic, social and institutional changes in the New York garment industry

The New York garment industry has undergone substantial changes in its economic, social and institutional characteristics, that have played an important role in the emergence of new competitive and organizational strategies among local firms. These dominant changes will be reviewed with a focus on the women's wear sector, that constitutes the bulk of the garment industry in New York. Adopting an historical perspective, we will see how a form of bureaucratic system was early introduced to regulate competition and employment conditions within the local industry, and how this system has been undermined by the economic and social transformations of the local industry over the past decades.

2.1 Emergence of a bureaucratic system in the New York garment industry: the tripartite regulation of employment and contracting relations

The development of the garment industry in New York between 1880 and 1920 was fuelled by a growing national market for clothing, as well as large scale immigration from Eastern and Southern Europe. The first items to be mass-produced were coats and suits, followed by dresses, shirtwaists and undergarments at the turn of the century. While male Jewish workers dominated in the coat and suit sector, young Jewish and Italian women constituted the majority of the workforce in other sectors. The predominance of immigrant women among production workers has since been an enduring feature of the local industry.

Strong seasonality and fashion changes in the women's wear industry translated into strong variations in employment and earning levels over the year [Waldinger, 1985]. As a result, a variety of mechanisms were used by firms to buffer core functions and employees from unexpected variations in product demand. First, a system known as specification contracting was early established in the industry, that allowed the so-called jobbers or manufacturers to limit their fixed costs and pass on demand fluctuations to contractors. Under this system, jobbers or manufacturers focus on product development, merchandising and sales activities, while the sewing contractors assemble garments according to manufacturers' specifications. The development of contracting firms is historically based on social networks in immigrant communities, and eased by low entry cost into sewing activities [Waldinger, 1986; Bailey and Waldinger, 1991].

The deskilling of entry-level jobs was a second buffering device used by the largest production firms. By resorting to unskilled new hires to meet fluctuations in output demand, these firms could protect their production process from the disruptive effects of high employees turnover and employment fluctuations [Waldinger, 1985]. The employment of home workers also provided a means for rapidly increasing and decreasing the workforce. More generally, the production system was associated with a variety of ways to transfer costs to production workers and increase cost flexibility, such as contractors' practice of subcontracting to individual workers, or the use of a piece rate pay system by which workers are paid by the task. Arbitrary management, unlimited work hours and unsanitary working conditions were common features of the industry's early days [Schlesinger, 1951; Waldinger, 1985].

Elements of a bureaucratic employment system were first introduced in the industry through the establishment of rules aimed at standardizing employment practices and stabilizing competition. Following two major strikes in the shirtwaist and cloak industries, an agreement known as the "Protocol of Peace" was signed in 1910 by the International Ladies Garment Workers' Union (ILGWU) and the employers' organization representing the largest firms in the local industry. It provided for a definite wage scale and a fixed number of work hours per week, standard sanitary conditions, as well as the abolition of subcontracting and homework.

Fashion changes and slower growth during the 1920s undermined the stability embodied in the Protocol, and firms developed a widespread use of contracting during this period [Carpenter, 1972; Waldinger, 1985]. A number of studies reveal the employment and contracting structure that emerged at that time: in 1923-24, 200 jobbers in the coat and suit industry used 18.5 per cent of their contractors for 86.3 per cent of their production; in 1934 in the dress industry, 81 jobbers were found to allocate 71 per cent of production to 32 per cent of their contractors; in the early 1930s, a study of 60,000 workers employed in the coat and suit industry showed that about one third received some income during the entire year, another third received some income during 29 weeks per year, while the remaining third received some income during 17 weeks per year [Schlesinger, 1951]. The emerging picture is that of a minority of contractors and workers accounting for the bulk of local production activities, and complemented by peripheral contractors and workers absorbing seasonal fluctuations in output demand. This dual structure created disparities between the competitive positions of large and small firms, as small peripheral contractors exerted a downward pressure on wages and contracting prices in the industry. These highly competitive conditions were aggravated by the crisis of 1929, which stimulated the introduction of new forms of regulation in the industry. The union and largest firms joined again to stabilize the industry, focusing this time on the regulation of contracting relations. The system of rules introduced by the NRA Codes of Competition in 1933 and maintained in the industry collective agreements had two major objectives: prevent the development of a core-periphery structure in production activities, and recreate a unity of interest between manufacturers and contractors. First, the contractor designation system required manufacturers to select a number of permanent contractors and distribute work equitably among them. Second, manufacturers had to pay contract prices that allowed for the payment of union wages in contracting firms and for contractors' overhead costs and services. As emphasized by Piore [1990, p. 286], this rule system was supported by a strong social consensus based on shared ethnicity in the local industry:

(...) employers welcome union organizations as the key to wage stabilization in the industry, provided that the union can enforce the same wage bargain upon all employers, potential or actual. (...) The immigrant community plays a central role in this process because it is able to bring community pressure to bear upon both workers and employers to adhere to the labour agreement.

This bureaucratic system greatly improved employment conditions in the local industry [BLS, 1914; Best, 1936], but it also considerably rigidified the production system. Although the most constraining aspects of the collective agreements disappeared over the years, their basic principles remain unchanged today. Nevertheless, their regulating impact gradually eroded as local production activities became part of global production systems, while demographic changes in the local workforce weakened the social consensus supportive of bureaucratic rules.

2.2 From a local to a global industry: the emergence of the large firm and the continuous relocation of production activities

The extreme mobility of garment capital and the development of transportation technologies have favoured a continuous relocation of production activities almost since the inception of the industry. Due to frequent fashion changes and the soft quality of clothing, garment assembly cannot be easily standardized and automated, and labour remains a major part of production costs [Bailey, 1989; Hoffman and Rush, 1988]. On that basis, the search for low labour cost as well as cost flexibility has been an essential part of firms' competitive strategies: a substantial part of production activities moved first to the greater New York area, then to Southern States, and then again to foreign countries.

Starting in the 1920s through the 1950s, the industry expanded in the New York region in response to the higher price of labour and rent in New York City, as well as a labour shortage due to the slowdown of immigration flows [Blumberger and Ong, 1994]. The ILGWU successfully organized the greater New York area and contained labour cost differentials between New York and the rest of the region [Carpenter, 1972; Waldinger, 1986]. However, the regulation of the contracting system proved more difficult to enforce with the geographical dispersal of contractors, and was further undermined when production activities relocated to nonunion settings in Southern states and foreign countries. The post-war trend toward casual clothing increased the possibilities to standardize garment production and reduce production skills requirements, while labour surplus and union-free policies in Southern states provided new location opportunities [Blumberger and Ong, 1994]. Starting in the 1950s, the production of the most standardized items such as undergarments relocated to larger production facilities in North Carolina, followed by less standardized product lines during the 1960s. This trend parallelled the emergence of large sportswear firms developing a variety of products, in opposition to the small specialized manufacturers that typified the New York industry. During the 1950s and 1960s, the share of multiplant firms in the industry's employment and shipments substantially increased [Waldinger, 1986]. The large firms used product diversification and strong brand recognition strategies to smooth seasonality effects on sales. Their large size enabled them to monitor production activities in Southern States, while maintaining design, marketing and sales activities in New York.

Employment growth continued in the South through the following decades, although at a much slower pace. The trend toward larger sized firms and multiplant firms also stabilized, and even reverted during the 1980s. These changes parallelled a significant increase in foreign production. Like the industry relocation to the South, the move to foreign production facilities started with the most standardized products and continued with more fashion-oriented items. However, this more recent trend relies mainly on contracting, and has been largely initiated by large retailers bypassing manufacturers to develop their own products and monitor production contracting. Foreign contracting activities initiated in Japan in the mid-1950s, but quickly moved to Hong Kong due to Japan's economic development and labour cost increases [Bonacich and Waller, 1994]. Large American retailers and buyer groups played a significant role in the growth of the Hong Kong industry during the 1960s, through frequent visits to Hong Kong factories as well as the establishment of local offices to transmit product specifications and monitor local production [Oman, 1989; Steed, 1981]. A similar pattern accounted for the development of the garment industry in Taiwan and South Korea during the 1970s, leading these three countries to produce the bulk of American apparel imports by the early 1980s [Rothstein, 1989]. With the assistance of their Far East contractors, large American firms started developing new production locations in Latin America and Southeast Asia during the 1980s, to compensate for higher contract prices in the Far East [Steele, 1988]. The share of Hong Kong, Taiwan and South Korea in American imports subsequently decreased from 53 per cent to 25 per cent between 1984 and 1993, as their production stabilized and growth took place in other countries [AAMA, 1994].

These locational changes have accompanied a surge in American apparel imports during the 1980s and early 1990s [AAMA, 1994; Bailey, 1993], together with important restructuring in the retailing and merchandising ends of the industry. The importance of customer recognition in maturing markets has led retailers to develop their own private label products, providing both lower prices to customers and higher margins to retailers [KSA, 1992; IFTF, 1994]. Private label products are highly fashion-oriented and price sensitive, increasing retailers' incentive to develop low cost, flexible production networks. They now represent about 25 per cent of women's apparel sales, and import penetration is higher than average for these products [KSA, 1992]. The growth of large retailing chains through consolidation has further improved their ability to develop and manage global contracting networks [Steele, 1988]. Importing initially introduced an important source of rigidity in garment production systems but today, international networks combine cost and flexibility advantages due to substantial reductions in production lead times.

Retailers' involvement in manufacturing activities has been echoed by manufacturers' move into retail activities aimed at strengthening their own brand strategies. Large manufacturers have increasingly invested in store-into-the-store strategies, where they are responsible for managing their own selling space within department stores. They have opened their own stores and factory outlets, and developed direct selling via catalogues and other means [KSA, 1993]. They also tend to increase product variety and reduce production times, relying on global sourcing to combine cost advantage and flexibility. Today, virtually all major women's wear manufacturers have developed international production networks [Bonacich and Waller, 1994]. Thus, the expansion of global production networks over the last two decades is associated with aggressive marketing strategies by large American firms, aimed at promoting strong brand recognition in matured consumer markets.

Parallel to the growth of imports, production jobs in the United States have declined since the early 1970s, but this trend has not been homogenous across the country: women's wear employment has been booming in California since 1970, and the decline of employment in New York has stabilized since the mid-1970s.

2.3 Changing labour market characteristics and the renewal of the New York production base

As shown in Table 2, New York's share of national employment has remained stable since the mid-1970s, after a decline from 33.5 per cent to 17.2 per cent between 1958 and 1972 (see appendix). This stabilization has been accompanied by a trend toward smaller firms' size, as well as declining real wages for production workers. Industry experts have also identified a resurgence of sweatshop conditions in local firms during the 1980s [GAO, 1988, 1989], defined by the violation of labour laws concerning health and safety conditions, record keeping, overtime pay, minimum wage and homework. These important changes in employment characteristics are closely linked to a substantial increase in immigration from Asia and Latin America since the 1970s, and the resulting transformation of the industry's ethnic base. Table 3 indicates that the share of immigrants among New York garment workers increased from 48 per cent in 1970 to 88 per cent in 1990, with an increase from 4 per cent to 38 per cent in the share of foreign-born Asians, and from 13 per cent to 29 per cent in the share of foreign-born Hispanics (see appendix). The Asian penetration is significant not only among production workers but also in management positions, with a shift from 1 per cent to 13 per cent between 1970 and 1990, while the share of Hispanic increased from 3 per cent to 8 per cent over the same period.

The entry of the new immigrants in the New York garment industry has led to the emergence of a major production centre in Chinatown [ILGWU; 1983], as well as a recent stabilization of production activities in the traditional midtown Garment District. While traditional ethnic groups concentrated in highly skilled production activities, the new immigrants specialized in the production of low-priced sportswear [Waldinger and Lapp, 1988]. Chinese firms progressively upgraded their production capacities to serve higher-priced markets [Brookman, 1994; Furman, 1993], while Korean firms employing an Hispanic workforce have more recently entered low-priced segments in the Garment District. Relying on ethnic networks, the immigrant firms are able to offer low cost, highly flexible production services in order to meet the increased requirements of retailers and manufacturers. While the New York production base had declined during the postwar period, due to labour shortage as well as the development of mass production in Southern states, the recent inflow of immigrants combined with accelerated fashion changes to provide a new competitive advantage for local production: within the global production networks of American retailers and manufacturers, New York immigrant firms find their niche in the production of short runs [Telesis, 1989].

The flexibility provided by New York contractors is obtained through a substantial weakening of the tripartite regulation system. The ILGWU has successfully organized the Chinatown industry and plays a major role in the ethnic community by providing health benefits to union workers and their families [ILGWU, 1983]. However, the institutional regulation of contracting and employment relations is loosely enforced in the local factories [Chin, 1992; Kwong, 1987]. Production activities also tend to develop in the Chinatown satellites of Brooklyn and Queens, where union coverage is less important [Zhou, 1992], while Korean factories in the Garment District remain largely unorganized.

These trends seem to point to a trade off between employment creation and the quality of jobs in the New York garment industry: the entry of the new immigrants has renewed the local production base, but employment conditions have strongly deteriorated over the last decade, while the bureaucratic rules for employment regulation became increasingly difficult to enforce. Because a substantial part of garments designed and marketed in New York are produced in foreign countries, the union controls only a portion of manufacturers' production networks and has limited power in negotiating and enforcing collective agreements. New York contractors also have to compete with contractors in Asia and Latin America who operate in loosely regulated labour markets [Bureau of International Affairs, 1990; International Labour Organization, 1995; World Bank, 1993], and are under continuous pressure to reduce costs and increase flexibility in order to avoid further relocation of production activities. Parallel to the globalization of production networks, Asian and Hispanic immigrants replaced the Russian Jews and Italians who had fuelled the initial development of the local industry. The massive entry of new immigrants, a typically exploitable labour force [Piore, 1979], reinforced cost competition incentives in the local industry. It combined with a generational change to weaken the social foundations of the bureaucratic system.

3. Firms' strategies and core-periphery forms of organization in the New York garment industry

In order to better understand the new dynamics that have replaced the bureaucratic form of regulation in the New York garment industry, we can use the core-periphery framework to study how manufacturers and contractors achieve product quality and production cost flexibility, two organizational objectives that place conflicting demands on their employment and contracting systems.

The detailed analysis of the economic, employment and contracting characteristics of 16 manufacturers and 21 contractors reveals a complex combination of cost, quality and flexibility objectives in these firms' strategies, as well as a variety of skills and stability levels in their employment and contracting systems ^{(Endnote 1)} . It points to the emergence of a new form of dualism in the local industry, based on a core-periphery structure among manufacturers as well as contractors. Table 4 presents the various firms' profiles identified in the local industry (see appendix).

3.1 Core and periphery among New York manufacturers

The core-periphery structure emerging among New York manufacturers is defined by the nature of manufacturers' relations with retailers and final consumers, the primary orientation of manufacturers' competitive strategies, as well as specific organizational and contractual characteristics, that mainly differentiate core manufacturers from second-tier and third-tier manufacturers.

Core manufacturers develop close partnerships with retailers and experienced growing sales and profits during the early 1990s, as well as limited sales seasonality. These large firms develop a wide range of products in the growing sportswear segment, and pursue strong, long-term image building strategies based on their own brand names, a situation that dominates in designers' segments, or supporting retailers' brand names in lower price segments. Designers gain access to final customers by selling in their own stores and/or participating in retailers' advertising and merchandising activities. Conversely, retailers are closely involved in product development and production activities for private label products. The main competencies of core manufacturers lie in product development and production management activities, supported by a consistent use of new technologies such as computer-aided design systems, as well as human resource development policies that resemble the commitment model. Skills development processes are fully integrated to the work process on the basis of a flexible and decentralized work organization. Core firms rely on experienced managers to informally train workers, and implement pay-for-education policies to support employees' formal training in relation to work activities. They maintain close links with training institutions, such as the Fashion Institute of Technology and the Parson School of Design, through regular internships and hiring. They also pay higher than average wages in order to attract and retain talented employees, and provide employment stability based on employee performance. These employment practices apply to marketers, designers and production managers, who embody the core competencies of the firm, but not to production-oriented activities such as sample sewing.

Core manufacturers develop a large variety of styles over the year, and yet derive most of their sales from a limited number of styles: designers develop cheaper derivatives of their collection lines that make up most of their annual sales, while private label manufacturers market-test a variety of styles before producing the most successful. As a result, these firms have to combine creativity and adaptability to market trends with large production volumes. To meet these requirements, core manufacturers resort to a dual contracting network. They develop commitment systems based on close trading relations with a few core contractors accounting for the bulk of their production activities, and resort to peripheral contractors under the lines of a traditional system. Core contractors acquire an idiosyncratic knowledge of manufacturers' production requirements, and receive a consistent flow of work over the year. Friendship and loyalty are an essential part of these contracting relations, which fits Uzzi's [1994] conceptualization of manufacturer-contractor linkages in the New York garment industry as based on trust, information exchange and joint-problem solving. Core manufacturers also resort to short-term relations with peripheral contractors to meet sudden changes in output demand and lower their production costs. In these production settings, they achieve product quality through direct control and exert strong price and lead time pressures in contract negotiations.

The core and periphery components of these contracting networks are connected through contractors' selection and development processes. Manufacturers constantly test new contractors and progressively increase their volume of activity with those who exhibit both motivation and potential for developing specific skills. The training process occurs through manufacturers' intensive involvement in contractors' activities and may last one to three years, during which the factories produce almost exclusively for one manufacturer. Once trained, contractors tend to demand higher prices and enlarge their clientele, so that manufacturers select new contractors to enter the training process. As a result, sophisticated manufacturers achieve a balance between quality, cost, and flexibility requirements by maintaining a three-tiered contracting network, including well-trained factories, in-training factories, and factories that have just entered their production network. This core-periphery strategy applies to contracting networks in New York as well as in the Far East, Central America and Southeast Asia, where core manufacturers have established production management and quality control staff to engage in factory search and development activities. The New York production base retains a specific advantage for these global manufacturers, based on its proximity as well as its skilled and low cost labour force. Short production runs are thus contracted out locally, while large production volumes are contracted out overseas.

Non-core New York manufacturers are typically small and specialized by product category, so that the coordination of the line and image-building strategy are left to retailers. They operate in the dress and sportswear segments and develop products under their own labels, although most of them also entered the private label business in recent years to meet retailers' growing demand. Their private label products are either derivatives of their own lines or fully specified by retailers, in which case manufacturers only arrange for the production process. Second-tier manufacturers limit sales seasonality based on stable relations with retailers. They compete mainly on service and value and provide an important flexibility in the products offered. They stress the importance of employee skills but implement training activities only on an occasional basis. Similarly, they adopt a core-periphery contracting structure but do little to further contractor skills development. They also lack the sophistication necessary to manage global production networks and focus on local production and its "quick turn" advantage. Third-tier manufacturers experienced declining sales and profits in the early 1990s and their sales level strongly fluctuates during the year. They compete mainly on cost and are highly dependent on fluctuations in product demand. CEOs exhibit no awareness of the role of human resource in developing a long term competitive advantage. Likewise, their contractors' selection process as well as quality control systems are largely under-developed. Third-tier manufacturers also give primacy to cost considerations in production location decisions and typically resort to foreign production through import agents. Their employment and contracting characteristics tend to resemble a traditional system, as do, although to a lesser extent, those of second-tier manufacturers.

These various manufacturers' profiles reveal important organizational differences between global, multiproduct manufacturers that invest in their human resources and develop close links with retailers and core contractors, and local, specialized manufacturers that service retailers with unequal success. While the traditional New York industry was based on specialization and complementarity between small firms [Waldinger, 1986], current trends place small, specialized manufacturers in an increasingly marginal position. Indeed, only large firms generate sufficient sales volume to compensate for the cost of brand building and global sourcing strategies. These transformations among manufacturers have directly impacted the New York contracting base, generating a growing division between core and periphery contractors at the local level.

3.2 Core and periphery among New York contractors

The study of contractors' profiles shows how manufacturers' practices translate in terms of contracting structure in the local industry. Core-periphery systems appear as the central organizing principle of production activities in Chinese and Korean firms, as revealed by the dominant characteristics of core and periphery contractors, as well as contractors in intermediary positions. By contrast, the bureaucratic system is more prevalent in the organizational characteristics of contractors belonging to the traditional Jewish and Italian ethnic groups.

The sales and profits of core contractors have been stable or growing during the early 1990s, with moderate sales seasonality. Most of these large firms are organized in family groups of several factories, and operate in the sportswear segment. They maintain close and stable links with their main clients, for whom they work exclusively during the industry's seasonal peaks while enlarging their clientele during slow periods. As one owner explains:

Our main challenge is work continuity. Over the year, the activity is profitable for three months, even during six months, and unprofitable for three months. During these last three months, you can lose a lot of money, all the profits accumulated over the year. So the key is to pick up the right relationship during the busy months, when there is a shortage of capacity in the industry. You'd better make sure that you help the right manufacturer then, the one that will help you during the slow times.

Core contractors compete mainly on product quality, which involves not only conforming to manufacturers' technical specifications, but also ensuring that products meet the client's objectives by providing extra-efforts to deal with unspecified aspects of the production process. They handle unexpected problems, quickly learn new styles and meet short deadlines, maintaining a complex balance between quality, flexibility and pricing requirements. Core contractors take advantage of technological changes and develop incremental innovations on specific parts of the production process, although their major source of differentiation is human capital. Owners build on more than ten years of contracting experience in the local industry to maintain their core positions. They also operate a continuous investment in core workers' skills, in line with the principles of a commitment employment system. A high level of quality is instilled in all production operations through continuous reinforcement of needed work behaviour. Technical skills are developed informally and through job rotation, with occasional resort to formal training. Continuity in employment and earnings over the year is the central return provided to regular employees. As a result, core contractors are able to attract and retain skilled and motivated workers and experience low employee turnover. Current employees serve as the main recruitment channel to select workers with adequate skills and appropriate personality traits. Workers' steadiness and responsibility allow these contractors to control production quality and timing, while multiskilling is essential to handle changing production requirements. Core contractors adjust to fluctuations in product demand through highly flexible work schedules based on work sharing. They also shelter core employees from seasonal fluctuations in output demand by hiring temporary workers to perform less skilled tasks during peak periods, and subcontracting to low cost contractors. This last buffering device puts core contractors in direct relation with the second type of contracting firm identified in the local industry.

Peripheral contractors experienced important seasonal fluctuations as well as declining sales in the early 1990s. They do not develop close relations with manufacturers and typically get new orders through intermediaries, such as core contractors or trucking companies that carry garments from manufacturers to contractors and back. Their clientele increases during peak periods, when manufacturers and core contractors look for additional production capacities in the local industry. Peripheral contractors are typically small and ill-equipped and compete mainly on cost, although facing strict quality requirements that translate into non-payment for defective products. Owners have little industry experience and a limited knowledge of production and quality control systems. High variations in employment and earning levels over the year, frequently combined with an unhealthy work environment, prevent them from attracting skilled and motivated workers. Employees have narrowly defined skills and owners report poor motivation levels as well as high employee turnover. Peripheral contractors recruit new hires through signs posted in the street, sharing a pool of low skilled workers who transit between the densely clustered factories in Chinatown and the midtown Garment Center. The size of their workforce fluctuates with activity levels and no positions are protected from work instability. Factories also close and reopen, often with different ownership, according to business fluctuations. Limited management skills combine with intense price competition to produce a downward pressure on wages and employment conditions in peripheral factories. Indeed, these firms typify the traditional employment systems described in the literature.

The dynamic link between core and periphery positions is illustrated by intermediary contractors who are engaged in a training process with one manufacturer, and combine core and periphery characteristics in their economic and employment profile. These contractors may eventually reach core status, although the transition process is not systematic. Some contractors fail to develop the versatility of their workforce and diversify their clientele, and remain "stuck" in intermediary positions. Others fail to maintain a core status due to tight quality-flexibility-pricing requirements in the local industry. More importantly, the seasonality of output demand in the local industry generates an excess of production capacity during peak periods that cannot be absorbed during slow times, so that peripheral factories structurally outnumber core factories. Periphery firms also maintain a competitive pressure on core contractors: because trust links can easily form and dissolve, the frontier between core and periphery positions is blurred and core contractors continuously strive to maintain their quality-flexibility-pricing advantage.

While core and periphery contractors belong to the new ethnic groups, owners being predominantly Asian among the firms studied, contracting firms owned by European immigrants have distinct organizational characteristics. They operate in high-price, declining market segments such as evening couture, and owners as well as employees are over 50 years old on average. Work schedules have limited flexibility, and employees alternate work and employment by resorting to the unemployment insurance system during idle times. Owners also rely on state and union mechanisms to recruit workers. They are rather supportive of union rules and resent the unbridled competition of Asian factories. These contrasts between Asian- and European-owned firms highlight the social rupture associated with the shift from a bureaucratic to a core-periphery system, together with market changes and the new contracting strategies of manufacturers.

The various firm profiles identified among manufacturers and contractors reveal a complex core-periphery structure in the New York industry. The functional or vertical division of work between manufacturers and contractors defines a first, easily predictable dimension of core-periphery systems. However, horizontal segmentation patterns have emerged within each of these groups: the global sportswear manufacturers and their core contractors have gained dominant positions, while small manufacturers and contractors operate in marginal positions. Firms' positions along this vertical and horizontal segmentation are associated with different competitive strategies, performance levels and human resource management systems. In addition, firms' market price segment modifies the shape of core-periphery systems. Both the relative importance of the periphery and the competitive pressures exerted on the core appear to be stronger in lower price segments: manufacturers are more vulnerable to retailers' pressure and have less control on production planning, so that they tend to exert more price pressure on contractors, provide their core contractors with a less consistent flow of work, and resort more heavily to occasional contractors than manufacturers in higher price segments. Core contractors also pay lower wages and make greater use of temporary workers in lower price segments. Finally, one could add an ethnic and institutional dimension to the new core-periphery system. While manufacturers belong to the traditional ethnic groups, new immigrants predominate among contractors. The Chinese factories are unionized and dominate the high price sportswear segment, while Korean firms are nonunion and specialize in low price sportswear.

4. Some lessons from the New York garment industry

While the rationale for developing core-periphery systems applies to all industries, the extent to which firms can organize their operations into core and periphery components, the mechanisms used to operate this separation, as well as the intensity of the competitive pressures driving these changes may vary quite importantly from one industry to the other. Inter-industry comparisons of firms' economic and employment characteristics show that firm's size and capital intensiveness, as well as the gender, educational and occupational distribution of the workforce are major determinants of their employment practices [Bartel, 1991; Delattre and Eymard-Duvernay, 1983; Ichniowski, 1990]. Thus, it is not surprising that the low capital intensiveness, low educational level and high proportion of women and production workers in the workforce of the garment industry translate into a strong core-periphery division, an important periphery, as well as strong competitive pressures exerted on the workforce.

Nevertheless, the trends toward tighter competitive conditions, globalization, and increased workforce diversity, that favoured a resurgence of core-periphery systems among garment contractors and increasingly divide manufacturers, are not specific to the garment industry, nor is the long term decline in real average wages for production workers. Consequently, by conceptualizing the mechanisms that produced core-periphery systems in the garment industry, as well as the outcome of their development, we can build a framework of broader applicability to evaluate current trends in employment characteristics.

Figure B builds both on the literature explored in this paper and on trends identified in the New York garment industry to emphasize the key regulating mechanisms of the core-periphery system in comparison with those of the bureaucratic system (see appendix). Although this framework is of broad applicability, the nature and scope of its individual components may vary, quite importantly, depending on the industry considered. It highlights major differences between the dominant characteristics of the post-World War II period, associated with a bureaucratic system, and those of the current period in which core-periphery systems are developing.

After World War II, the concentration of production and consumption activities in national economies, the prevalence of mass production systems in large production settings, as well as the preponderance of white males in the workforce favoured the development of bureaucratic employment systems supported by the New Deal system of labour-management relations. These developments generated a trend toward increasing average wages and declining inequalities in the United States. Their social outcome was the growth of the middle class, which in turn sustained growing consumer markets, the economic foundation of mass production systems. For example, Boyer [1993] analyses how new rules of wage formation based on collective bargaining, as well as the growing role of the state based on Keynesian principles, contributed to a smoothing of business cycles after World War II, so that downward employment adjustments became less frequent and less steep than during the 1920s and 1930s.

In the current period, the spatial dispersion of production and consumption activities, the development of "flexibly specialized" production systems, as well as increased race and gender diversity in the workforce are associated with a declining influence of institutional rules and the development of core-periphery employment and contracting systems. Their depressing effect on wage levels, combined with increased wage inequalities, erodes both the size and the confidence of the middle class, which in turn reinforces the sluggishness and unpredictability of consumer markets. Changes in the composition of the workforce and the organization of production are eroding the social consensus supportive of bureaucratic rules, based on the homogeneity of the workforce and the cohesion of production and consumption activities within national economies. From the perspective of the labour force, Loveridge [1987] and Piore [1975] explain how segmentation processes are linked to social processes of stigmatization based on race, ethnic and gender differences. From the perspective of industrial organization, the trend toward smaller production settings and increased diversity in the forms of production also fragments the workforce and weakens its social cohesion. The globalization of production activities contributes to this evolution by integrating a variety of national cultures within a given production process.

Such a core-periphery perspective contributes to renew dual labour market theory. On one hand, it relies on the same rationale concerning firms' economic and employment characteristics. In dual labour market theory as conceptualized by Piore [1975, 1980], the segmentation of the labour market is linked to uneven standardization, stability and uncertainty in output demand. On the other hand, important characteristics distinguish the two frameworks. In the traditional theory, the core is defined along the lines of a bureaucratic system, and the periphery is associated with a traditional system. The core also appears as dominant, and the periphery as a residue of pre-industrial practices. Indeed, this theory describes the regulating mechanisms of the post-World War II period, and highlights the fact that the bureaucratic system never completely dominated the organizational landscape. However, the dynamics supporting labour market segmentation appear to be reversed today: it is the periphery, and not the core, that is developing and influences the average evolution of wages and employment conditions. Core positions are also more exposed to competitive pressures in the absence of formal rules to shelter them from periphery positions.

Endnote 1:
See methodological appendix for a general description of the sample studied and methods of analysis.