Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Apparel1 PETER DOERINGER AUDREY WATSON Boston University Economists have traditionally argued that technological change and improve- ments in human capital are the key determinants of productivity growth. Busi- ness historians favor broader explanations that include changes in managerial organization and conduct. Alfred Chandler (1977), for example, characterizes the period between 1870 and 1920 as a second industrial revolution because im- provements in technology interacted with new management systems that could tap economies of scale, scope, and organizational learning. Some analysts argue that the world is now experiencing a third industrial revolution, based on a com- bination of new information technologies and modern manufacturing techniques (Greenwood, 1997; Best, 1990; Milgrom and Roberts, 1990~. The apparel industry in the United States participated modestly in the first two revolutions. The sewing machine was the technological innovation that paved the way for factory production of clothing beginning in the 1850s and 1960s. During the second industrial revolution, economies of scale and scope in apparel manufacturing resulted in a rate of productivity growth that actually exceeded the average for all manufacturing between 1860 and 1910 (Kuznets, 1952~. ~ This research is sponsored by the Alfred P. Sloan Foundation, through a grant to the Harvard University Center For Textile and Apparel Research. The paper draws heavily upon materials devel- oped by our colleagues Frederick Abernathy, John T. Dunlop, Janice Hammond, and David Well. We are also grateful to Bruno Courault, Lynn Oxborrow, and Elisabeth Parat, whose work on a counter- part Sloan Foundation study in France and the United Kingdom has added to our understanding of the U.S. experience. David Mowery, members of the New England Economy Study Group, and partici- pants at a conference on small-scale enterprise organized by the French Ministry of Labor and the Center for Employment Studies provided helpful comments on earlier drafts of these materials. Kara Bunting contributed excellent research assistance. 329
330 U.S. INDUSTRYIN2000 The apparel industry, however, became the antithesis of most modern indus- tries in the postwar economy. It is dominated by small and medium-sized firms, technological change has been modest, education requirements are relatively low, and the industry remains labor intensive. The predictable result has been a loss of market share to imports and a substantial decline in employment. With two ex- ceptions commodity products, such as socks and men's underwear that can be mass-produced at low cost using capital-intensive technologies and high-fashion products that are not sensitive to price the prognosis for the apparel industry through the early 1980s was one of continuing decline in market share and jobs. That prediction is now being reassessed (Abernathy et al., 1995; New York Times, 1998~. The apparel industry is adopting modern information technolo- gies, domestic suppliers are serving new just-in-time replenishment markets, and labor productivity has been rising at rates comparable to those in all manufactur- ing. The sustainability of these trends, however, is less certain. Domestic pro- duction may have speed advantages over offshore production, but speed and cost are substitutes in the sourcing decision, and there is no domestic monopoly on production speed. Either increases in domestic production costs or faster produc- tion and delivery speeds among offshore suppliers could threaten the revival of production inside the United States. This chapter examines recent developments in apparel production channels in the United States. It focuses on the growth of new domestic markets for just- in-time apparel supply and on the prospects for U.S. apparel manufacturers to develop the rapid response production capabilities needed to secure these mar- kets against foreign competition. APPAREL'S PLACE IN AMERICAN INDUSTRY In 1995 U.S. apparel manufacturers shipped $62.9 billion in 1992 dollars of product, representing slightly under 2 percent of U.S. manufacturing output (Fig- ure 1; Table 1~. In addition to clothing products, the industry also includes home furnishings and industrial products, such as automobile upholstery, and the share of output accounted for by these non-clothing sectors has been growing in recent years. The apparel industry has long been one of the nation's larger employers. Although employment has fallen by 38 percent since 1970 (Figure 2; Table 2), apparel still accounts for 4.6 percent of manufacturing employment. Earnings in apparel, which were once close to the manufacturing average, are now only 55 percent of average earnings in manufacturing, and real earnings have fallen more than 13 percent since 1970 (Table 3~. Skill and education levels are low in apparel. Ninety percent of production workers are unskilled or semi-skilled (Mittelhauser, 1997), and the percentage of the apparel workforce with less than a ninth-grade education is about double that of the average for manufacturing (Arpan et al., 1982~. The apparel industry is
APPAREL 70,000 60,000 50,000 o In o ._ ._ 40,000 30,000 20,000 1 0,000 o ~ H]~T O Lo O CO ~ ~ 00 ~ O ~ Cal 00 00 00 00 00 00 "I Women's FIGURE 1 Real value of shipments (millions of 1992 dollars), 1970-l99S. Source: Bureau of the Census, Annual Survey of Manufactures (various years). 33 CO 1 ~ Lo Apparel TABLE 1 Real Value of Shipments (millions of 1982 dollars), 1970-1995 All Men's and % of Women's and % of Year apparel Boys' apparel total girls' apparel tote] 1970 41,231.2 14,511.9 35.2 14,351.3 34.8 1975 45,816.6 15,416.7 33.6 16,786.7 36.6 1980 51,614.2 16,008.5 31.0 19,914.7 38.6 1985 54,279.1 16,086.2 29.6 19,904.1 36.7 1986 54,486.0 17,025.6 31.2 20,269.8 37.2 1987 59,264.4 16,854.5 28.4 21,335.0 36.0 1988 58,220.2 16,339.1 28.1 21,235.6 36.5 1989 55,370.0 15,416.6 27.8 19,423.3 35.1 1990 54,866.8 14,555.2 26.5 19,606.4 35.7 1991 54,636.3 15,040.5 27.5 19,606.4 35.9 1992 58,548.8 16,158.7 27.6 21,192.9 36.2 1993 60,062.6 15,471.8 25.8 22,539.8 37.5 1994 62,331.6 16,014.9 25.7 22,538.5 36.2 1995 62,879.9 15,944.7 25.4 22,947.6 36.5 Note: Men's and boys' apparel is the sum of SIC codes 231 and 232; women's and girls' apparel is the sum of SIC codes 233 and 234. Values of shipments are deflated by the producer price indices for all apparel products, men's apparel, and women's apparel, respectively. Source: U.S. Bureau of the Census, Annual Survey of Manufactures (various years).
332 1200 1 000 800 o ._ ~5 o TO 400 600 v) to ° 200 o U.S. INDUSTRYIN2000 I- ~ BiBlele'~ -+ + 1 ~ 1-+ ~ ~ +~ 1950 1960 1970 1975 1980 1985 1990 1991 1992 1993 1994 1995 1996 _ Men's Other FIGURE 2 Employment in the U.S. apparel industry, 1950-1996 (thousands of produc- tion workers). Source: Bureau of Labor Statistics, Employment and Earnings (various years). also a major employer of women and minorities. Female workers make up about 75 percent of the workforce, accounting for almost 11 percent of all females in manufacturing. Around 15 percent of the apparel workforce is African-Ameri- can, 24 percent is of Hispanic origin, and a substantial percentage of the remain- der is Asian (U.S. Department of Labor, 1994~. Apparel is an industry of small firms (Figures 3 and 4; Tables 4 and 5~. Average number of emloyees is 38, and two-thirds of all establishments employ fewer than 20 workers. Average establishment size, however, varies consider- ably across product sectors. With 109 employees, the average men's wear estab- lishment is more than three times the size of the average women' s wear establish- ment (Figure 3; Table 4~. Establishment size had been growing until the early 1980s, when this trend reversed across all product categories. Firms with fewer than 20 employees account for less than 10 percent of the industry's workforce, however, while 37 percent of the workforce is employed in establishments with 250 or more employees (Figure 5; Table 6~. With many small plants, relatively limited economies of scale, and little ver- tical integration, apparel manufacturing is the quintessential example of a com- petitive industry. The four largest dress manufacturers, for example, account for only 6 percent of their market, and the eight largest have only a 10 percent market share.
APPAREL 333 From the beginning of mass production, however, apparel manufacturing has been characterized by elaborate contracting networks in which production is di- vided among "inside" shops (or "manufacturers"), "outside" shops, (or "contrac- tors"), and "jobbers." Clothing jobbers are intermediaries, but they often play a much more extensive role than that of a mere middleman between suppliers and retailers. They design clothing, purchase and often cut material, deliver the fabric parts to contractors for assembly, and market the completed garments. Inside shops resemble manufacturers in other industries. They design their products, buy raw materials, produce their goods in company-owned facilities, and then TABLE 2 Employment in the Apparel Industry, 1899-1996 (thousands of Production Workers) Year Total Men's clothing Women's clothing 1899 338 158 84 1914 548 226 169 1925 515 224 127 1935 631 277 221 1940 819 324 226 1950 1080 380 428 1960 1098 383 439 1970 1196 436 478 1975 1067 382 433 1980 1079 377 436 1985 944 325 367 1986 926 315 348 1987 922 312 344 1988 912 306 336 1989 907 295 333 1990 869 277 317 1991 841 271 309 1992 844 276 304 1993 829 272 289 1994 815 262 278 1995 772 244 259 1996 695 215 227 Note: Data from 1899to 1935 are not directly comparable tolater date. 1899to 1935 dataonmen's clothing represent employment in outerwear, work clothing, shirts, and nightwear. 1940 to 1996 total employment figures are all production workers for SIC 23, apparel and related products. Men's clothing is the sum of production workers in SIC codes 231 and 232; women's clothing is the sum of production workers in SIC codes 233 and 234, except for 1940, for which SIC 234 is not available. Sources: 1914-1935 total production workers, U.S. Department of Commerce, Bureau of the Census, Census of Manufactures; employment in men's and women's clothing, 1914-1935, Drake and Glasser, Trends in the New York City Clothing Industry; 1940-1996, Bureau of Labor Statistics, Employment and Earnings and Supplement to Employment and Earnings (various years).
334 U.S. INDUSTRYIN2000 TABLE 3 Relative and Real Average Weekly Earnings in the Apparel Industry, 1909-1996 Weekly apparel earnings Real weekly apparel as a percentage of earnings in Year weekly manufacturing 1982-1984 dollars 1909 92.8 N/A 1914 86.2 N/A 1925 94.3 N/A 1935 88.6 N/A 1950 76.5 185.1 1960 62.9 190.7 1970 63.3 217.4 1975 58.5 207.4 1980 55.9 195.9 1985 54.0 193.8 1986 54.1 195.6 1987 54.1 193.5 1988 54.1 191.4 1989 54.5 189.0 1990 54.1 183.0 1991 55.0 183.9 1992 55.0 184.3 1993 54.3 182.5 1994 54.3 185.7 1995 54.9 185.5 1996 55.3 187.5 Note: Average weekly earnings for 1909 to 1935 are estimated by dividing total payrolls in men's and women's clothing by total wage earners in both sectors, and dividing the result by 52. Men's clothing includes outerwear, work clothing, shirts, and nightwear. Real earnings are deflated by the consumer price index, 1982-1984 = 100. Sources: 1909 to 1935 apparel payrolls and wage earners, Drake and Glasser, Trends in the New York Clothing Industry; 1940 to 1996 apparel wages and all manufacturing wages, Bureau of Labor Statistics, Employment and Earnings and Supplement to Employment and Earnings (various years). market the output. Outside shops serve as contractors for jobbers and inside shops (Teper, 1937~. The apparel industry is labor intensive. Assets per employee were only 14 percent of the manufacturing average as late as the mid-1980s (Murray, 1995; Rothstein, 1989), and new capital expenditures per worker average less than 15 percent of the manufacturing average (Table 7~. The pace of technological change in apparel has also been relatively modest (Murray, 1995~. The dimensional instability of fabric has made the actual sewing process difficult to automate (Dunlop and Well, 1996~. More manufacturing in- novation has occurred in the preproduction stages. Computer-aided design (CAD) systems have reduced fabric waste and speeded the size-grading of patterns, while
APPAREL 335 marker-making and fabric cutting can now be performed by computer-guided lasers. The high cost of these systems, however, makes their adoption prohibitive for all but the largest firms (Murray, 1995; Rothstein,1989~. More recently, new information technologies are being adopted to link manufacturers to retailers. These systems allow producers to receive electronic point of sale data from stores and to track orders from production through delivery (Abernathy et al., 1995~. The single most important factor affecting the apparel industry has been the globalization of the supply chain. Prior to the 1970s, imports accounted for only about 10 percent of the domestic sales. Although both domestic shipments and domestic value added have continued to grow in real terms, domestic production has steadily lost market share to imports and over half the U.S. market is supplied by foreign producers (AAMA, 1997~. The rate of import penetration has been controlled by the Multi-Fiber Agree- ment (MFA), a complex system of tariffs and quotas which allowed Asian coun- tries to be the dominant source of U.S. imports. Special trade privileges have been granted to Mexico under the North American Free Trade Agreement (NAFTA) and to Caribbean basin countries. These changes in trade policy are shifting production to the western hemisphere as domestic firms are outsourcing assembly to these regions to take advantage of preferential trade arrangements. ~ 140 .m Q to LL Q a) o A: I 40 o Q 80 60 _-~ _ '_ /~ - `_- 2 0 ~: ! i 1 1 1 1 1 1 1 ~ 1 1899 1914 1925 1935 1939 1946 1950 1956 1962 1965 1970 1975 1980 1985 1990 1994 1995 -- Men's ~ Women's FIGURE 3 Average establishment size U.S. apparel industry, 1899-1995. Sources: 1899- 1939, industry averages, Historical Statistics of the United States; men's and women's, Seidman, The Needle Trades; Bureau of the Census, County Business Patterns.
336 1 975 250 to 499 (3.21 %) in. 20 to 249 (43.38%) 20 to 249 (29.83%) ~ \ + (0.80%) - - 1 995 250 to 499 (2.00%) 500 + (0.90%) An, \ / FIGURE 4 Establishments by size class, 1975 and l99S. Source: Bureau of the Census, County Business Patterns (various years). U.S. INDUSTRYIN2000 O to 19 (52.61 %) O to 19 (67.27%) Even greater import pressure can be expected as the MFA is phased out early in the next century. The challenge for the domestic clothing industry is to find ways to shift the basis of global competition from manufacturing cost, where the United States is at a comparative disadvantage in all but the most capital-intensive products, to the speed of supply, where proximity to markets gives the U.S. industry an edge. While employment continues to fall in the industry, these declines are concen- trated in those sectors where foreign competition matters most, and the industry is now reporting employment gains in markets where just-in-time supply is impor- tant (New York Times, 1998~.
APPAREL 337 THE INFLEXIBLE PRODUCTION SYSTEM Since World War II, apparel production has been dominated by a relatively inflexible form of mass production known as the progressive bundle system (PBS). Under PBS, bundles of cut and partly sewn clothing parts are sequentially assembled into complete garments as they pass from work station to work station. Labor is highly specialized, many tasks take only seconds to perform, and the total labor content of a garment is measured in minutes. Workers can become very proficient at these specialized tasks, but learning curves can be as long as six months on the more skilled sewing jobs. PBS assembly lines are difficult and costly to balance because of speed and quality problems inherent in working with soft fabric. Frequent style changes further raise line-balancing costs, making PBS most efficient when there are long production runs of each style. Individual employee differences in proficiency and level of fatigue also raise line-balancing costs. For these reasons, large buffer stocks between work stations are used to pre- vent bottlenecks, resulting in long throughput times for individual garments. For TABLE 4 Average Establishment Size in the Apparel Industry, 1899-1995 Year Industry averageMen's clothingWomen's clothing 1899 272131 1914 303630 1925 284121 1935 335231 1939 375736 1946 38N/AN/A 1950 42N/AN/A 1956 428445 1962 4610248 1965 5111652 1970 5713155 1975 5212353 1980 5813655 1985 4813249 1990 4313041 1994 3811331 1995 3810931 Note: Industry averages 1899-1914 are number of production workers per establishment. 1899 to 1914 men' s are for "men' s, youths' and boys"' and women' s are for "women' s and children' s". For 1946 to 1995, industry figures are for SIC code 23; men's for SIC codes 231 and 232, and women's for SIC codes 233 and 234. Sources: Industry averages, 1914-1939, Historical Statistics of the United States; men's and women's, Seidman, The Needle Trades; 1946-1995, U.S. Bureau of the Census, County Business Patterns (various years).
338 o ~ o lo To ca o so 4= o ~ ° ca o 4= Cal o be a' a' ·_4 VO Cq Cq ·_4 o C) a' U) Em ca o 4= ~ 0 ca 4= 4= o ~ V, so o ~ ~ CO Lo Do A So 8 ;~>, In o o ~ ~ ca o on 4= ~ o o ca on 4= ~ ca on o ~,4 no ~ ca _I O 4= ~ o ~ o o Do 8 ~8 ~ ~n o ~n o ~ ~ ~= - =o o ~ o o o o ~ o ~ ~ ca ~ ca 4 ° 4 ° CM ~ ~ ~ CM CM ~ CM ~ 4 ° 4 ° 8 ~ -~= 8 ~ oo oo o o ~ oo 4= ~ o ~ ~ ~ ° o~ ca ca ~o ~ ~ ~ ~ ~ ~o ;^ o ~ ~ ~ ~ ~ o 4= ° 4- ~ ~ ~ o o o o oo ~ CM CM CM CM CM ~ ~ ca ca o. m= o ~ ~ ~ ~ ~ ~ o ~ o ~ ~ ~ o oo ~ o oo ~ ~ . . . . . . o ~ ~ o o o ~ ~ ~ ~ ~ o . . . . . . ~ o oo ~ ~ ~ ~ o ~ ~ ~ o cM ~ o ~ oo oo . . . . . . cM cM o ~ ~ ~ cM . . . . . . ~ ~ ~ ~ o o . . . . . . ~ oo ~ ~ oo ~ . . . . . . ~ o ~ o ~ ~ ~ oo oo ~ ~ ~ ~o m 0 V ~o V, .^ ca V ~o m~ ca ~ 0 V ·s .~ · L#~ . ^ ~ . . ~ ca o v
APPAREL 1 975 250 to 499 (20.58%) \ 20 to 249 (59 20 to 249 (53.60%) 500 + (12.85%) 1 995 250 to 499 (18.40%) 500 + (18.40%) FIGURE 5 Employment by establishment size class, 1975 and l99S. Source: Bureau of the Census, County Business Patterns (various years). 339 O to 19 (6.71 %) O to 19 (9.60%) example, standard industry practice is to have one day's buffer stock between operations. This implies that a given pair of pants requiring 40 operations will take 40 days to move through the line, even though the average direct labor time for a pair of pants is only about 24 minutes (Dunlop and Well, 1996~. This inflexible system had its origins in the development of mass markets in the 1920s and 1930s, but it became widespread in the manufacturing and retailing environment of World War II. The mass demand for military garments, War Production Board regulations limiting the variety of civilian styles that could be offered, and the preferences of consumers for "quality over variety" created op- portunities for long production runs of identical products (Disher, 1947~. Long
340 ca 8 ~ In o o ~ O ca o ~ 4.. ° o ~ ca 4.. ° o o ca o o o ~ 1 a' a' ·_4 VO a' o o C) a' En ca o o ~,4 o ~ ca ;~. o o o ca ;^ o 4= ~ ca ;^ to 4= ~ o ~ ¢ . . . . . . CM ~ ~ ~ Do Do ~ ~ ~ ~ o ~ . . . . . . o ~ o o ~ Do Do ~ CM . . . . . . Do ~ ~ ~ ~ ~ Do ~ ~ Do o o ~ CM Do ~ Do ~ . . . . . . . . M cM cM ~ oo ~ ~ ~ ~ . . . . . . o o o ~ ~ ~ ~ o ~ o ~ ~ ~ oo oo ~ ~ ~ · - ~ ca oo ~ ~ ~ oo ~ . . . . . . ~ o ~ ~ ~ ~ ~ oo ~ o ~ ~ . . . . . . ~ ~ ~ o ~ ~ cM . . . . . . ~ o ~ ~ o ~ o ~ ~ ~ o ~ . . . . . . cM cM ~ ~ ~ cM . . . . . . ~ ~ o o ~ ~ o o o o o o . . . . . . o o o o o o ~ o ~ o ~ ~ ~ oo oo ~ ~ ~ 4= o ca ca .~ ^~ o ~ o ~ ~ ~ . . . . . ~ ~ ~ ~ cM . . . . . ~ ~ ~ o ~ o o ~ ~ cM . . . . . oo ~ ~ ~ ~ ~ ~ oo o ~ ~ ~ ~ o o cM ~ ~ ~ ~ . . . . . ~ ~ o ~ ~ ~ 00 ~ ~ ~ . . . . . 00 00 (~N 00 . . . . . O ~ 0 0 ~ ~ 00 ca ~ 0 C) ·s V V, ca .g ~ ca ca ^~ ~ o V .^ ~ ~o ~o ~ C) ~ v m V, g .ca ~ ·~ 0 .ca V ~ ~o =~ t ^~ V, 4o~ o Z ~Q
APPAREL TABLE 7 New Capital Expenditures per Production Worker (1982 dollars), 1960-1995 341 Year All industries All apparel Men's end boys' apparel Women's and girls' apparel 1960 2476.7 240.6 191.9 195.5 1970 4085.8 637.9 539.3 429.2 1975 5093.9 617.8 410.6 613.0 1980 5878.8 627.2 638.0 495.2 1985 6346.4 717.3 583.5 534.1 1986 5915.9 722.6 627.5 502.1 1987 5751.2 723.3 663.2 549.5 1988 5682.8 657.3 618.1 578.1 1989 6628.4 806.7 669.3 677.0 1990 6839.8 769.2 540.4 535.4 1991 6780.9 700.5 585.3 391.5 1992 6863.3 885.4 832.6 628.4 1993 6629.8 886.7 759.2 490.0 1994 7039.7 1018.2 612.0 562.4 1995 7655.4 1100.2 635.8 559.3 Note: Men's and boys' apparel is the sum of SIC codes 231 and 232; women's and girls' apparel is the sum of SIC codes 233 and 234. New capital expenditures are deflated by the producer price index for capital equipment. Source: U.S. Bureau of the Census, Annual Survey of Manufacturers (various years). runs offered substantial scale economies, particularly through the adoption of Tayloristic PBS manufacturing systems and the substantial division and special- ization of labor. The result was the growth of relatively large, highly efficient firms that specialized in the mass production of garments. The war also changed the apparel industry from a "buyer's market" to a "seller's market." Retailers, "anxious for all available supplies," could no longer demand rapid response production as they had in earlier decades. As a result, they were forced to accept the longer order times and production smoothing long sought by manufacturers (Disher, 1947~. After the war, clothing styles were no longer regulated, and style change again became important (Frank, 1953~. Instead of returning to the prewar system of relatively flexible production, however, the demand for mass fashion at rea- sonable prices and the low unit costs made possible by PBS allowed the large manufacturing sector to continue to dominate the production channel. Manufac- turers defined the styles of garments to be produced at different price points, set production and delivery schedules, and consequently had considerable influence over the level of inventories held by retailers (Disher, 1947~. Many large manu- facturers also developed "branded" products that gave them a marketing advan- tage over "store" brands.
342 U.S. INDUSTRYIN2000 The efficiency of this system kept clothing costs low at both the wholesale and retail levels. During the 1950s, before imports became significant, retail apparel prices rose less than 7.5 percent (less than one-third the rate of all con- sumer prices) and wholesale apparel prices were comparably stable (Tables 8 and 9~. Low prices, coupled with the control of style and marketing by manufactur- ers, presumably lowered the resistance of mass retailers to the large inventories and long supply times inherent in the inflexible mass production system. The accommodation that retailers made to the PBS system, however, laid the foundation for the subsequent shift from domestic to offshore supply chains. When lower-cost foreign clothing became available in the 1960s and 1970s, the adaptations that retailers had already made to inflexible domestic mass produc- tion predisposed them to view foreign suppliers as relatively easy substitutes for domestic suppliers. The even longer lead times and greater inflexibility of for- eign supply channels were only an extension of the inflexible characteristics of the domestic supply chain, and the costs of added inflexibility were more than compensated for by the labor cost advantages of imports. THE FLEXIBLE PRODUCTION CHANNEL Currently, new demands from downstream retailers and inflexibilities in fabric supply, as well as pressures from foreign competition, are transforming the inflexible manufacturing-driven domestic production system. Unlike apparel, the retail and textile sectors are characterized by large firms and rising levels of concentration. For example, the four largest apparel retailers held 17.9 percent of the market in 1992, compared with 6.4 percent in 1972. The corresponding fig- ures are 27.6 percent and 11.2 percent for women's specialty shops and 53.1 percent and 38.8 percent for department stores (Bureau of the Census, Census of Retail Trade, various years). Increased concentration, along with the availability TABLE 8 Percentage Change in Consumer Price Indices, 1950-1996 Apparel less All items footwear Men's and boys' Women's and girls' apparel apparel 1950-1960 22.827.4211.065.39 1960-1970 31.0827.2231.7826.63 1970-1980 181.6744.1943.7333.70 1980-1990 58.6232.0434.6827.60 1990-1996 20.054.486.061.80 1950-1996 551.04171.82200.47131.78 Source: Bureau of Labor Statistics.
APPAREL TABLE 9 Percentage Change in Producer Price Indices, 1950-1996 343 All finished All Men's and boys' Women's and girls' Children's and goods apparel apparel apparel infants/apparel 1950-1960 18.444.727.031.4316.10 1960-1970 17.6616.8024.5710.3723.53 1970-1980 123.9255.6178.3238.3848.13 1980-1990 35.4532.4731.6533.6032.38 1990-1996 10.156.479.903.276.16 1950-1996 365.60168.45244.01113.73198.54 Source: Bureau of Labor Statistics. of offshore suppliers, has shifted decision-making power within the production channel from clothing manufacturers to mass retailers. Increased cost pressures following the wave of leveraged buyouts and merg- ers in retailing in the 1980s encouraged retailers to reduce the costs of inventories by adopting new information technologies, such as electronic point of sale (EPOS) data and computerized ordering and stock management programs. These cost- cutting practices are known as "lean retailing" (Berg et al., 1996~. The prolifera- tion of clothing styles, colors, and sizes as well as the shortening of product life cycles in the 1980s further intensified the incentives for adopting lean retailing practices. In one industry sample, the number of stock-keeping units (SKUs)- the codes that define individual products by style, color, and size increased by 63 percent between 1988 and 1992. At the same time, the year-to-year turnover in styles rose. Between 1984 and 1992, the number of selling seasons in a year grew from 2.8 to 3.2 for basic fashion products and from 2.9 to 3.7 for fashion products (Abernathy et al., 1995~. Within selling seasons, the importance of "short-cycle" products those with a planned sales period of only 5 to 10 weeks, compared with 12 to 20 weeks for traditional "seasonal" products has also risen, accounting for more than one-third of apparel consumption in 1988 (Rothstein, 1989~. More products and more rapid style change tend to raise inventory and mark- down costs and to increase the possibility of lost sales. They also raise uncer- tainty about consumer demand because there are fewer products with a market history and less time in a season to adapt to demand fluctuations. In the face of these pressures, mass retailers sought further reductions in inventories and markdowns through the extension of lean retailing to include just-in-time supplies. Leadership for this change came across a wide spectrum of retailing mass merchandisers such as Wal-Mart, large department store chains such as Dillards, national variety chains such as J.C. Penney, and national spe- cialty chains such as The Limited (Abernathy et al., 1995~. These extended lean retailing efforts have focused on "basic fashion" products positioned between commodity products and more fashionable women's wear products.
344 U.S. INDUSTRYIN2000 In principle, just-in-time supply allows retailers to place smaller initial or- ders because replenishment supplies can be obtained throughout the selling sea- son in response to actual sales. As a result, inventories, stockouts, and mark- downs would be reduced. Just-in-time delivery is not consistent with the supply capabilities of the inflexible domestic PBS system, however, and is beyond the reach of distant offshore supply chains. Introducing a quick supply capability into domestic PBS supply channels has involved both a willingness among retail- ers to pay a cost premium for quick and accurate fulfillment of replenishment orders and to provide the information systems needed to link domestic clothing manufacturing to retail sales data. The main instrument for building just-in-time supply chains has been the transfer of new information technologies from lean retailers to apparel manufac- turers. Examples include electronic data interchange (EDI) of point-of-sale data between retailers and clothing manufactures and the use of EPOS computer pro- grams to trigger quick-response shipments and initiate new production. The spread of new computer and information technologies within the domes- tic supply chain has been well documented by the Harvard Center for Textile and Apparel Research (Abernathy et al., 1995~. Between 1988 and 1992 the use of bar coding at the detailed product level grew by 2.75 times; EDI links between clothing retailers and manufacturers grew by more than 7 times; the receipt of EPOS data on either an individual store or a company-wide basis rose from 13 percent of sales volume to 32 percent; and the use of programmed ordering mod- els also increased (Abernathy et al., 1995~. The effective use of EPOS and EDI by clothing manufacturers, however, requires considerable knowledge about the information systems and lean retail- ing practices of specific retailers. Much of this knowledge is proprietary and cannot be readily acquired through arms-length market relationships. Lean re- tailers, therefore, have reinforced information linkages by forming privileged business relationships, or "partnerships," with their just-in-time suppliers to fa- cilitate the transfer of "match-specific" knowledge. These partnerships often in- volve technical assistance in improving operations planning, management, and logistics between retailers and manufacturers. INDICATORS OF PERFORMANCE UNDER JUST-IN-TIME PRODUCTION Rapid replenishment capabilities have spread through the supply chain. For example, 73 percent of sales volume for national chains and 66 percent of sales volume for mass merchants in 1992 were replenished on a daily or weekly basis (Abernathy et al., 1995~. These data understate the growing importance of the practice, however, because the retail sectors that use rapid replenishment most intensively have also been increasing their overall share of domestic apparel pro
APPAREL 345 auction. Between 1988 and 1992 the percentage of domestic shipments going to national chains rose by 64 percent and that going to mass merchants rose by 12 percent, compared with a 7 percent decline for department stores and a constant share for specialty stores (Abernathy et al., 1995~. Adopting information technologies, forming retailer-supplier partnerships, and achieving higher replenishment speeds has brought corresponding improve- ments in clothing manufacturing practices. For example, work-in-process inven- tories fell by one-third, from three weeks of supply to two weeks, between 1988 and 1992, while manufacturing throughput times have been reduced (Abernathy et al., 1995; Dunlop and Well, 1996; Berg et al., 1996~. Firms that adopt the new information technologies and other innovations associated with rapid replenishment capabilities perform better than those that do not. Comparisons between "high innovation" and "traditional" apparel firms show large differences in the time from fabric purchase through completion of manufacturing (79.2 days compared with 128.9 days), replenishment supply speeds (9.2 days compared with 26.3 days), and operating profits (10.5 percent compared with 5 percent) (Abernathy et al., 1995~. Even firms that adopt the minimal level of innovation consistent with supplying lean retailers perform sub- stantially better than traditional firms. Aggregate Productivity Growth Given these sweeping changes in the apparel supply channel, it is appropri- ate to look for wider evidence of their impact on business performance. One aggregate indicator of performance is the growth in productivity. To establish a baseline from which to compare productivity growth during the lean retailing period, we calculated the trend rate of growth in real value-added per worker hour between 1950 and 1995. This trend rate for the entire industry is 3.3 percent a year, a record that is slightly better than that for all manufacturing (Table 10; Figure 6~. We also looked at a second baseline period, 1950-1970, when the PBS manufacturing system dominated the apparel industry and when imports were relatively unimportant. Trend annual productivity growth in this period was very close to that for the entire postwar period 3.4 percent between 1950 and 1960 and 3.2 percent between 1960 and 1970 (Table 10~. Trend productivity growth was somewhat higher for women's and girls apparel and somewhat lower for men's and boy's apparel (Figures 7 and 8~. These aggregate trends, however, are the net result of a far more complicated set of forces than the organization of production in the industry. These forces include growth of imports since the early 1970s and shifts in the importance of mass manufacturers as well as technological change. The magnitudes of some of these changes, such as those associated with imports and increased offshore con- tracting, are presumably large relative to those associated with manufacturing technology. Although productivity growth cannot be partitioned into its compo
346 U.S. INDUSTRYIN2000 TABLE 10 Value-Added Per Worker Hour (in 1982 dollars), 1950-1995 All All Men's and boys' Women's and girls' manufacturing apparel apparel apparel 1950 1960 1970 1975 1980 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 13.4 20.3 28.6 31.6 32.9 40.3 43.3 45.5 47.2 46.7 45.8 46.5 49.1 50.0 52.1 53.3 4.9 7.0 9.6 11.4 13.2 16.7 17.4 18.6 18.4 18.3 18.3 18.7 19.7 19.9 21.2 21.8 5.1 7.3 9.8 11.0 12.8 16.6 17.1 17.9 17.0 16.9 16.3 17.5 18.2 17.1 19.4 20.8 4.4 6.4 8.9 10.6 12.6 15.7 17.1 18.3 18.9 17.7 17.9 18.5 20.2 20.0 20.6 20.5 Annual Growth Rates: 1950-1995 (%) 3.1 3.3 3.1 3.4 1950-1960 (%) 4.1 3.4 3.5 3.8 1960-1970 (%) 3.4 3.2 2.9 3.4 1970-1980 (%) 1.4 3.2 2.7 3.4 1980-1990 (%) 3.3 3.3 2.4 3.5 1970-1987 (%) 2.7 3.9 3.5 4.2 1987-1995 (%) 2.0 2.0 1.9 1.4 1990-1995 (%) 3.1 3.4 4.9 2.7 Note: Men's and boys' apparel is the sum of SIC codes 231 and 232; women's and girls' apparel is the sum of SIC codes 233 and 234. Manufacturing value-added is deflated by the producer price index for all finished goods; apparel value-added figures are deflated by the producer price indices for all apparel products, men's apparel, and women's apparel respectively. Source: U.S. Bureau of the Census, Annual Survey of Manufactures (various years). nent parts with any precision, the direction and timing of these different influ- ences can be examined. Import Penetration The period 1970-1987 largely predates lean retailing and approximates a time when import growth was the dominant influence on the industry. Productivity growth during this period was 3.9 percent a year, substantially above the long- term trend, and it was particularly high in women' s and girls' apparel. We sus- pect that the conventional wisdom of imports displacing low value-added do- mestic production is the primary explanation for this superior productivity performance.
APPAREL 22 20 ~_ _ so 18 16 ~ 14- 2 1 2 a' 10 > 8 6 347 a..." Of 1960 1965 1970 1975 1980 1985 1990 1995 FIGURE 6 Real value added per worker hour, all apparel, 1960-l99S. Source: U.S. Bureau of the Census, Annual Survey of Manufactures (various years). 20 ~_ Cal oo 18 I~ 1 6 4 - ~ 10 > 12 8 6 _~ 1960 1965 1970 1975 1980 1985 1 9gO 1995 FIGURE 7 Real value added per worker hour, men's apparel, 1960-l99S. Source: U.S. Bureau of the Census, Annual Survey of Manufactures (various years).
348 22 20 ~_ cat 0 18 = 16 ~ 14 o G. 12 <' 10 - > 8 6 U.S. INDUSTRYIN2000 ;if-1 1960 1965 1970 1975 1980 1985 1990 1995 FIGURE 8 Real value added per worker hour, women's apparel, 1960-l99S. Source: U.S. Bureau of the Census, Annual Survey of Manufactures (various years). The Effects of Lean Retailing Productivity growth after 1987 and extending to the present is subject to a more complex set of influences than earlier periods. This was the time when lean retailing became important in creating new markets for rapid replenishment do- mestic production in certain product lines. But it was also a period that witnessed a shift in the origin of imports from Asia to the Caribbean basin and Mexico as well as continued growth in imports. Lean retailing can affect productivity in several different ways, some of which operate in contradictory directions, and its net effect is difficult to predict. For example, it can raise productivity growth by promoting lean manufacturing practices that make labor more efficient. Conversely, if production batches are smaller or permit less specialization of labor as would be the case with modular production, the growth in labor productivity will be reduced. A second effect occurs through the mix of products produced domestically. Lean retailing has focused on basic fashion products. To the extent that the value- added of basic fashion products is below the average for domestic production, an increase in replenishment production will also slow the observed rate of produc- tivity growth. A third consequence of lean retailing has been to favor large suppliers over smaller ones. If there are differences in value-added per worker by size of firm, then the changing size composition of firms will affect productivity growth.
APPAREL 349 Somewhat surprisingly, there is roughly a U-shaped distribution of real value- added per worker hour by size of establishment in the apparel industry with the smallest firms (fewer than 20 employees) often having the highest productivity. On balance, the growth in importance of large firms under the influence of lean retailing has probably had a relatively minor effect on aggregate productivity in the industry. A final aspect of lean retailing is to shift responsibility for various "value- added services" required to make garments floor-ready from retailing to manu- facturing. For example, the volume of floor-ready merchandise shipped rose from 40.6 percent to 50.3 percent between 1988 and 1992, and increases were also registered in the labeling and pricing of garments by clothing manufacturers (Abernathy et al., 1995~. Business units with the highest replenishment pressures were the most likely to provide these services. Although data are lacking on the actual value-added per worker hour of these services, tasks like price tagging and floor-ready packing of garments presumably contribute less to value-added than the average assembly job and probably have had a negative effect on the growth in labor productivity. Changing Trade Regimes The shift in the national origin of U.S. apparel imports toward the Caribbean basin and Mexico began in 1985 as the result of preferential trading arrangements that eased restrictions on imports from these regions. The Caribbean basin trade preferences are conditional upon imports being assembled from U.S.-made parts, while NAFTA requires only that the fabric originates in North America. This requirement, along with geographic proximity, has encouraged domestic clothing manufacturers to retain high value-added design, cutting of U.S.-made fabric, and marketing functions while transferring lower value-added assembly work to foreign countries (Mittelhauser, 1997~. This "contracting" effect contributes to raising average labor productivity in the U.S. apparel industry. Net Effects on Productivity Change Although the precise contribution of each of these factors to changes in pro- ductivity cannot be determined, their net effect can be observed. Productivity growth has decreased for the industry as a whole during the period 1987-1995, with the decline being sharpest for women's garments (Table 10~. This picture, however, is misleading because the timing and pattern of productivity change vary by product. For example, basic fashion has been the product sector where lean retailing should have the greatest effect on domestic productivity. We would expect to see the lean retailing directly raising productivity through its effects on lean manu- facturing and its bias toward large firms Conversely, product lines such as fash- ionable dresses or blouses are least likely to be affected by either lean retailing or
350 U.S. INDUSTRYIN2000 TABLE 11 Annual Growth Rates of Real Value Added per Worker Hour, Selected Products Men's arid boys' Men's arid boys' Women's and misses' Women's and misses' suits arid coats shirts dresses suits arid coats (SIC 231) (SIC 2321) (SIC 2335) (SIC 2337) 1960to 1970 2.6 3.0 3.5 3.2 1970 to 1980 2.0 2.1 1.9 4.7 1980to 1990 2.2 4.2 5.2 4.6 1970 to 1987 3.1 3.7 4.4 4.9 1987to 1995 0.8 0.7 0.5 0.4 1990 to 1995 3.5 1.1 1.7 -1.3 Note: Men' s and boys' shirts includes nightwear before 1987. Source: U.S. Bureau of the Census, Annual Survey of Manufactures (various years). import penetration and fashion products are also more likely to be produced by small firms where productivity growth has been relatively slow in recent years. These effects are likely to slow productivity growth in product lines such as fash- ionable dresses and blouses. Examining productivity change between 1987 and 1995 in four of the largest product lines for which consistent data are available reveals a slowdown in pro- ductivity growth below the long-term trend. This slowdown, however, was con- centrated in the late 1980s. After 1990, as lean retailing gained strength, there was a widespread rebound in most of the product lines (Table 11~. Moreover, productivity growth in men's and boys' clothing, where style tends to be more "basic" than "fashion," is at an historical high, while the lowest productivity growth was recorded for product lines such as women's suits and coats and women' s dresses, where we would expect to see the weakest influences of trade and lean retailing (Tables 10 and 11~. These patterns lend support to the thesis that the effects of lean retailing on productivity are positive. The Reinvigoration of the Large Firm Sector A second way to assess the effects of lean retailing on the apparel supply chain is to examine growth patterns by size of firm. Between the 1950s and 1970, when the inflexible PBS production channel was expanding and before foreign imports and lean retailing were significant, the fraction of large establishments (250 or more employees) in the apparel industry more than doubled (Table 5~. Thereafter, imports led to widespread employment declines, particularly among large PBS manufacturers whose markets were particularly vulnerable to import penetration. One would expect, however, that the growth of rapid replenishment markets would limit or reverse this decline because of the preferred position of large firms in lean retailing partnerships. This effect is confirmed by the data. The employ
APPAREL 351 ment share of the very largest firms (500+ employees) grew from 12.9 percent in 1975 to 15.4 percent in 1990 and to 18.4 percent in 1995 (Table 6~. This growth in large firms also appears to be associated with an increase in the importance of "inside" manufacturers. The share of total value-added held by inside manufacturers grew across a broad range of product lines throughout the period 1972 to 1992 (Table 12~. A similar trend throughout this period can be seen in the share of employment accounted for by manufacturers in all of the product lines except women's dresses (Table 13~. This growth in the share of value-added by manufacturers does not necessarily imply a reduction in contract- ing, since manufacturers also took on more value added services under lean re- tailing and had the option of foreign contracting as well. Comparable data for domestic contractors, however, confirm a reduction in the amount of domestic contracting activity by manufacturers. The overall trend away from domestic contracting and toward consolidating production within manufacturing began when import penetration was the main influence on the apparel industry. During 1987-1992, the part of the "lean retail- ing" period for which there are data, this trend is limited to a more narrow range of products. The share of both value added and employment held by manufactur- ers continued to increase for the men's wear products examined (except for the value-added in men' s and boys' shirts) and for women' s blouses. Contrary to the long-term trend, however, contractors gained a larger share of value-added in all of the women' s wear products and in men' s and boys' shirts. The share of em TABLE 12 Percent of Value Added, by Type of Firm: Selected Products, 1972, 1987, and 1992 1972 1987 1992 Manufacturers Men's and Boys' Suits and Coats (SIC 231) Men's and Boys' Trousers and Slacks (SIC 2325) Men's and Boys' Shirts (SIC 2321) Women's and Misses' Blouses (SIC 2331) Women's and Misses' Dresses (SIC 2335) Women's and Misses' Suits and Coats (SIC 2337) 55.5 45.9 46.1 25.2 30.8 28.5 69.0 54.0 75.5 38.8 48.4 45.1 77.4 60.0 71.5 44.9 40.8 44.4 Contractors Men's and Boys' Suits and Coats (SIC 231) Men's and Boys' Trousers and Slacks (SIC 2325) Men's and Boys' Shirts (SIC 2321) Women's and Misses' Blouses (SIC 2331) Women's and Misses' Dresses (SIC 2335) Women's and Misses' Suits and Coats (SIC 2337) Note: Men's and boys' shirts include nightwear in 1972. Source: U.S. Bureau of the Census, Census of Manufactures (various years). 31.1 34.5 29.9 40.5 38.1 34.0 22.2 20.8 9.9 31.1 28.5 24.0 18.5 13.8 17.1 32.6 34.3 25.1
352 U.S. INDUSTRYIN2000 TABLE 13 Percent of Total Production Worker Employment, by Type of Firm: Selected Products, 1972,1987, and 1992 1972 1987 1992 Manufacturers Men's arid Boys' Suits and Coats (SIC 231) Men's arid Boys' Trousers arid Slacks (SIC 2325) Men's arid Boys' Shirts (SIC 2321) Women's and Misses' Blouses (SIC 2331) Women's and Misses' Dresses (SIC 2335) Women's and Misses' Suits and Coats (SIC 2337) 46.9 37.6 40.9 25.7 32.4 27.9 5.7 51.5 56.6 26.6 27.2 32.7 67.2 61.4 71.1 32.4 24.4 30.0 Contractors Men's arid Boys' Suits and Coats (SIC 231) Men's arid Boys' Trousers arid Slacks (SIC 2325) Men's arid Boys' Shirts (SIC 2321) Women's and Misses' Blouses (SIC 2331) Women's and Misses' Dresses (SIC 2335) Women's and Misses' Suits and Coats (SIC 2337) 48.5 54.8 51.8 69.0 61.1 63.8 39.8 45.6 41.9 68.5 69.5 60.5 31.2 34.1 27.9 63.6 72.4 62.4 Note: Men's and boys' shirts include nightwear in 1972. Source: U.S. Bureau of the Census, Census of Manufactures (venous years). ployment held by contractors also increased for women's dresses and suits and coats (Tables 12 and 13~. By and large, we read these data as confirming the conclusion that lean retailing has shifted domestic production toward manufac- turers in product lines, such as men's wear, that have a relatively lower fashion content, while at least some higher fashion products, such as dresses, are becom- ing more contractor-intensive. THE LIMITS TO GROWTH OF REPLENISHMENT MARKETS By 1992, almost half of all sales by domestic producers were being shipped on a weekly or shorter replenishment basis. The substantial progress that has been made in serving rapid replenishment markets and in improving the perfor- mance of the U.S. apparel industry is cause for optimism. If past trends can be extended, the apparel industry faces a brighter future than would have been pre- dicted a decade ago. Some signs, however, point to limits on future improve- ments in rapid replenishment speeds and business performance. One set of limiting factors may be the difficulty of managing rapid response production. Despite new information technologies and various other changes in supply chain management, rapid replenishment appears to be accomplished, in part, by manufacturers' holding larger inventories of finished goods from which they can provide frequent shipments. This result can be seen in data on "innova- tive" clothing firms the types of suppliers that are most likely to be serving
APPAREL 353 rapid replenishment markets that hold larger inventories of finished goods and have fewer turns per year of finished goods inventories than do less innovative firms (Abernathy et al., 1995~. Our surveys in the United Kingdom show that reliance on inventories of finished goods is an even more common device for meeting rapid replenishment pressures than in the United States (Doeringer et al., 1998~. What appears to be happening, in part, is that lean retailing is shifting inventories of finished goods from retailers to clothing manufacturers. Moreover, the flexible production channel being constructed to serve rapid replenishment markets is biased toward large firms (Abernathy et al., 1995~. A cluster of factors that link large suppliers to lean retailing explain this bias. First, large firms are better able than small firms to finance and manage the costly information technologies required by lean retailers. Second, there are likely to be substantial coordination efficiencies within partnerships between mass retailers and large suppliers compared with those with a fragmented supply chain of nu- merous smaller firms. The link between lean retailing and large suppliers is fur- ther reinforced by the focus of rapid replenishment on basic fashion products. Domestic suppliers of basic fashion products tend to be larger than suppliers of fashion products (Abernathy et al., 1995~. Mass retailers in the United Kingdom show a similar tendency to form part- nerships with large suppliers. Some large retailers in the United Kingdom are even providing performance incentives to their partners by promising to increase the volume of orders to suppliers who meet quick response targets. These retail- ers eventually plan to phase out their smaller suppliers. The rationale of building just-in-time supply chains around large firms and basic fashion products may prove short-sighted. Although large firm production channels have made substantial strides in speeding production and delivery to replenishment markets, they may never be able to achieve the response speeds and small order sizes that are reported historically for small firms (Magee, 1930) or that we routinely find in our interviews in the small firm sector. The tendency of large firms to accomplish rapid replenishment by increasing finished goods inventory may also indicate that diminishing returns are occurring in the gains to lean manufacturing in large firms. If so, reliance on large firms for rapid replenishment may restrict the future growth of lean retailing markets for basic fashion products and is likely to inhibit the extension of lean retailing to fashion products. PROSPECTS FOR THE SMALL FIRM SECTOR Not only have the largest firms been growing in importance in the apparel industry, but there has also been substantial growth in the relative importance of firms with fewer than 20 employees (Tables 5 and 6~. One explanation of small firm growth is that rising imports forced firms in the medium-size range (between 20 and 249 employees) to downsize. This downsizing thesis is consistent with
354 U.S. INDUSTRYIN2000 the timing of the decline of medium-size firms and the growth of small firms beginning after 1970 and accelerating since the late 1980s. According to this interpretation, the industry is "hollowing out" and becom- ing more dualistic (Palpacuer, 1996~. In the future the industry will consist of a core of large, highly efficient, information-intensive manufacturers that will serve the relatively secure markets for commodity apparel products and basic fashion replenishment items. A periphery of small, marginal firms will be left with highly uncertain residual markets. A second hypothesis is that the hollowing-out of the middle of the industry and the relative growth of small firms is related to changing patterns of out- sourcing. For example, the downsizing of medium-sized firms to become small firms can be seen as a response to new opportunities for international specializa- tion within the production channel as well as to declining demand for domestic production. Such "internationally specialized" small firms will continue to buy fabric, make patterns, cut fabric, and market the final product, but their garments will be assembled in the Caribbean basin and Mexico. This specialization will allow the United States to retain the more highly skilled workers and the higher value-added tasks in which it holds a comparative advantage. Indirect evidence of such specialization is found in the changing occupa- tional mix in the apparel industry. Between 1983 and 1994, employment in ap- parel manufacturing fell by 16 percent, but this decline was concentrated in sew- ing and unskilled jobs. Employment in other occupations such as designers, technicians, and marketing that are involved in onshore pro- and post-assembly activities either grew or remained unchanged during this period (Mittelhauser, 1997~. A third possibility is that the growth in small firms reflects a set of produc- tion advantages that are important in fashion markets. Compared with basic fash- ion, fashion products are likely to have smaller initial orders, more uncertain markets, and shorter product seasons. Small firms have a proven capacity to produce small lots of fashion garments with short throughput times, whereas large firms are not sufficiently flexible to serve such markets. These advantages of speed and flexibility may also position small firms for a role in serving rapid replenishment markets and may be a foundation for extending rapid replenish- ment to fashion. History provides ample precedent for this possibility. Today's rapid replen- ishment pressures from lean retailers echo those of the 1920s and 1930s. This was a period when the new and rapidly growing demand for mass fashion led to product proliferation. "In the manufacturing industry there developed a period of the wildest experimentation in design. This in turn led to great confusion, not only among manufacturers as to what to make, but also among retailers as to what to buy," according to the National Retail Dry Goods Associations (1936~. Because of this "great confusion," retailers were generally reluctant to order goods ahead of actual demand, preferring instead to order a few items at the start
APPAREL 355 of the season followed by rapid replacement of the styles that sold (Teper,1937~. Uncertainty about fashion trends was further aggravated by the "piracy" of cloth- ing designs. Lower-priced copies of garments appeared with startling rapidity, sometimes before the originals reached the stores, lessening the value of the origi- nal designs. This problem of piracy increased the rate at which the better-priced manufacturers came up with new styles and created further incentives for retail- ers to bring new styles to market as quickly as possible (Teper, 1937~. Problems of fashion uncertainty and piracy of styles were particularly preva- lent in women's garment production and aggravated the seasonal volatility of that sector (Grieg, 1949~. Women's wear production was traditionally scheduled around two selling seasons, spring and fall, with each having a mid-season "fill- in" period. Styles were set and orders placed relatively close to the selling season, and production and delivery followed orders with a relatively short lag. Spring styles for women, for example, were shown in early December, with the first orders delivered in January and the fill-in orders occurring in March (Carpenter, 1972~. By 1939 an estimated 125,000 different dress styles were produced in New York; slightly fewer than half of these were moderately priced garments with an average production run of 997. The remaining styles represented better- priced dresses with an average production run of only 267 (Hochman, 1941~. Unlike current lean retailing partnerships, lean retailing efforts in this period relied on the rapid response capability of small firms. Pressures from retailers for ever more rapid supply response and lower costs led to the development of quick response mass production systems (Bryner, 1916~. A "frantic insistence upon immediate deliveries when orders are finally placed" (Teper, 1937) increased the pressures for rapid response manufacturing. Most orders placed after the begin- ning of the season were "for immediate delivery, that is, a week or ten days" (Magee, 1930~. Retailers also frequently returned merchandise that had been ordered but went unsold (Teper, 1937~. The quickest of these rapid response systems was the "piece" or "complete garment" method, which was widely used in small and medium-sized shops (Bryner, 1916~. Under this system a single operator performed the basic assem- bly of the garment, with an additional worker completing finishing operations such as felling linings and making buttonholes. Similar systems and similar production speeds and flexibility still appear to be common, but today's small firms are often characterized as inefficient because they lack advanced technologies and sophisticated management practices. Other factors may offset these inefficiencies, however; the data on productivity by size of firm shows that firms with fewer than 20 employees often outperform all but the largest firms. Although these various pieces of evidence point to the potential for positive efficiency contributions from rapid replenishment suppliers in the small-scale sector, it is premature to conclude that this sector can be integrated into lean retailing production channels in the foreseeable future. The weaknesses in infor
356 U.S. INDUSTRYIN2000 mation technology and modern management practices are a barrier to small firms supplying lean retailers. Bringing small firms into lean retailing production chan- nels and extending rapid replenishment to fashion products will require new insti- tutions that can link mass retailers to the small firm sector. INSTITUTIONAL OBSTACLES TO RESTRUCTURING APPAREL PRODUCTION CHANNELS Lean retailing is currently providing the motivation and the leadership neces- sary for efficient restructuring of domestic apparel production channels based upon rapid supply chain response. Progress toward this goal, however, is im- peded by incompatibilities in relationships between sectors, missing institutional links, and institutional biases in channel reforms. One obstacle to production channel reform is the self-reinforcing character of efficient relationships among the different sectors of the channel. Once pro- duction and distribution arrangements become compatible with channel-wide ef- ficiency, any one sector in the channel will find it difficult to respond to changing markets or technology unless the other sectors also adapt accordingly. Efficient transformation of the production channel is also slowed by the ad- aptation biases inherent in the established institutions of the channel, or what economists call "path dependency." Established production arrangements mean that production channels evolve in directions that extend, rather than radically depart from, existing systems. This "systemic" aspect of production channels biases the restructuring process toward incremental, rather than radical, changes in channels. One example, as noted earlier, is the successful adaptation of mass retailers to the efficiencies of the inflexible domestic supply system of the post- war period. Once this adaptation was made, it became more likely that retailers would next choose to lower costs further by turning to even slower and lower- cost supplies from Asia, instead of by developing rapid replenishment capabili- ties within the domestic supply chain. A related obstacle is the trend toward growth of the large-firm sector. Present reforms in production channels revolve around the relationship between mass retailers and large clothing suppliers. Mass retailers that are promoting restruc- turing through organizational relationships with their suppliers find it easier to coordinate production and logistics with a relatively small number of partners. As a result, retailers are deliberately reducing the number and increasing the size of their suppliers. As large firms gain a bigger market share, they become less dependent upon contractors and jobbers. The decline in contractors was documented earlier; the decline in jobbers has been even faster. In 1967 there was one jobber for every 3.2 contractors in the women' s clothing sector, but that number fell to one jobber for every 6.9 contractors by 1992 (Bureau of the Census, Census of Manufactur- ers, various years). Similar changes can be seen in specific product lines (Table
APPAREL TABLE 14 Number of Contractors per Jobber, by Sector, 1972, 1987, and 1992 357 1972 1987 1992 Manufacturers Men's arid Boys' Suits and Coats (SIC 231) Men's arid Boys' Trousers arid Slacks (SIC 2325) Men's arid Boys' Shirts (SIC 2321) Women's and Misses' Blouses (SIC 2331) Women's and Misses' Dresses (SIC 2335) Women's and Misses' Suits and Coats (SIC 2337) 2.66 2.86 2.53 3.47 3.50 2.62 5.28 6.39 5.00 6.64 12.47 4.06 5.47 5.41 5.19 8.65 6.60 5.23 Note: Men's and boys' shirts include nightwear in 1972. Source: U.S. Bureau of the Census, Census of Manufactures (venous years). 14~. Much of this decline in jobbers occurred before lean retailing was important, but it has been sustained or extended during the lean retailing period in four of the six product lines we examined. As production networks between large and small suppliers and between small suppliers and large retailers become more tenuous, the potential for lean retailing and just-in-time production to create new jobs for the domestic supply chain may be threatened. Domestic job growth depends on continually increasing replenish- ment speeds. Filling Institutional Gaps in Apparel Production Channels Reversing the exclusion of flexible, small-firm production from lean retail- ing systems requires a reconception of the role of intermediaries in apparel pro- duction channels. Historically, the problems in the small-firm sector lack of scale economies, limited managerial capacity, and fragmentation have been offset by various intermediaries between small producers and retailers. The tra- ditional intermediaries have been jobbers, and sometimes "inside" manufactur- ers. Lean retailers, however, have been eliminating such intermediaries from their production channels by choosing to develop direct relationships with large manufacturers and by performing more of the design function once controlled by manufacturers. At the same time, manufacturers are relying less on domes- tic contracting. One alternative to jobbers as intermediaries would be for small firms to form multifirm partnerships along the model of trade associations. Historical experi- ence has shown, however, that competitive market pressures have limited the success of trade associations in coordinating business decisions among their mem- bers. Furthermore, many small firms lack the managerial sophistication and fore- sight needed to participate effectively in such collective efforts.
358 U.S. INDUSTRYIN2000 Another alternative is for trade unions to serve as intermediaries, either on their own or in conjunction with trade associations or jobbers. Unions once pro- vided managerial assistance to improve the efficiency of the fragmented supply chain, and the most successful examples of coordination through trade associa- tions occurred within a framework of collective bargaining with unions. For example, the Amalgamated Clothing Workers contributed to improved industry performance during the 1920s by conducting efficiency studies aimed at lower- ing assembly costs, providing management assistance to troubled companies, and eliminating inefficient work rules (Fink, 1977~. The International Ladies Gar- ment Workers Union (ILGWU) played a similar role through its industrial engi- neering department, established in 1930 (Disher, 1947~. The ILGWU also en- couraged "larger production units, closer relationships between jobbers and their contractors, training for management, planning, cost accounting, and fair trade practices to govern dealing with retailers" (Seidman, 1942~. It is unlikely that unions, however, could play the same intermediary role today, largely because union membership in the industry has declined substantially. A third possibility is for governmental or not-for-profit organizations to serve as intermediaries. Examples of such not-for-profit organizations are (TC)2, an organization sponsored by apparel and textile companies, unions, and the U.S. government which provides technology-based R&D and technical assistance to the U.S. apparel industry, and the Garment Industry Development Corporation which offers both technical assistance and marketing services to the small-scale fashion sector in New York City. Similar organizations have been set up in France and the United Kingdom (Doeringer et al., 1998~. None of these organizations, however, is providing the full range of intermediation services needed to link small-scale flexible producers with lean retailers. Another institutional obstacle is the lack of quick response relationships be- tween apparel firms and their textile suppliers. Despite the enormous changes that have occurred in clothing production channels in recent years, minimum orders for fashion fabric averaged about 3500 square yards in 1992, and mini- mum delivery speeds averaged more than two and one-half months (Abernathy et al., 1995~. These numbers are almost unchanged from 1988. Clothing manufacturers have been unable to change textile supply practices through market forces, while mass retailers lack the direct economic relationships with the textile sector through which they might initiate reforms. Increasing capi- tal intensity in textile production may be one explanation for these rigidities, but examples can be found in the United Kingdom of large retailers coordinating fabric supplies through direct partnerships with textile manufacturers. The Possibility of Global Rapid Response Replenishment markets have also been protected because delivery lags and throughput times average about one-third longer for foreign than for domestic
APPAREL 359 production (Abernathy et al., 1995). Recent developments in trade policy, how- ever, may be undermining the long-term future of rapid replenishment production in the United States by laying the foundation for quick response production chan- nels in Mexico and the Caribbean basin. Special trade legislation exempts imports from the Caribbean from quantity restrictions, and tariffs are levied on only the value-added from assembly, pro- vided the garments are sewn from U.S.-made parts. Imports from Mexico face a preferential tariff regime, with tariffs to be phased out over ten years, as long as the garments are of NAFTA origin, defined as being made of NAFTA-produced fabric (Trebilcock and Howse, 1995~. These provisions encourage geographic specialization within the production channel whereby capital- and skill-intensive stages of production remain in the United States while labor-intensive clothing assembly is performed offshore. A by-product of this arrangement is that it encourages new contracting net- works in which nearby offshore contractors could become part of a rapid re- plenishment supply chain managed by U.S. manufacturers and jobbers. If such "partnerships" can somewhat narrow the disadvantages of longer throughput and slower delivery times, the cost advantages of offshore products might be suffi- cient for them to enter rapid replenishment markets in the United States. This possibility may explain a part of the increase in imports from the Caribbean ba- sin and Mexico from 7 percent of total imports in 1984 to 29 percent in 1995 (AAMA, 1996; Mittelhauser, 1997~. The potential for such global rapid re- sponse production channels makes it all the more imperative that domestic re- sponse times be shortened. CONCLUSION New information technologies, new forms of coordination between retailers and suppliers, and the emergence of quick replenishment clothing markets are improving the economic prospects of production channels in the United States. Sustaining this scenario in the future, however, depends upon retailers' being able to continue offsetting the higher cost of clothing produced in the United States by the cost savings of lean retailing. The faster, the more reliable, and the more accommodating the domestic supply chain, the larger the proportion of goods that can be produced in the United States. The sustainability of the advantages of lean retailing production channels remains somewhat in doubt. Lean retailing has favored large firms because of their advantages in acquiring new technologies and developing more sophisti- cated planning, production, and logistics practices. In addition, lean retailers find it easier to form partnerships with large firms, rather than trying to coordinate supplies among a number of small firms. The large-firm sector, however, may encounter internal obstacles to continuing to raise delivery speeds.
360 U.S. INDUSTRYIN2000 Small firms have the potential for very fast throughput and are a potential source of high-speed replenishment production. Small firms have been excluded from the rapid response production channel in the United States, however, and the current institutional structure of that channel may preclude their integration with lean retailing. Meanwhile, international sourcing of rapid response produc- tion may become possible by using low-cost suppliers that are "closer to home" (Murray, 1995~. Further reforms to speed production and delivery are critical if domestic clothing suppliers are to retain their position in replenishment markets for basic fashion products and if they are to expand into replenishment markets for more fashionable products. Such efforts, however, must be differentiated by type of product, size of firm, and position in the apparel production channel. The current focus of policies to achieve these reforms often neglects these important distinc tions. For example, few innovations are equally applicable to all products and sizes of firm. Technologies that can be successfully adopted by large firms are likely to be different from those that small firms can use, and technologies for serving mass markets for basic fashion may differ from those appropriate to lower-vol- ume fashion markets. The results of current R&D initiatives, such as those car- ried out by (TC)2, are available to all firms in the apparel industry, but they are often tailored to the larger firms that have traditionally led in the adoption of innovation. Policies for training and management assistance, because they are often or- ganized on a local or regional level, tend to be differentiated by size of firm and product. These programs address issues of work organization, quality control, production skills, and commercial knowledge that are often a barrier to integrat- ing small firms into mass retailing supply channels, but they do not tackle the larger problem of how to establish such linkages. The challenge for policy is to introduce reforms that focus on the institu- tional processes that govern the relationships among the different sectors of the production channel. These may also involve developing new intermediaries that can provide access to information technologies, develop new production channel relationships, and stabilize production in fragmented markets. Lean retailers are currently positioned to lead these reforms, but other candidates include clothing unions, jobbers, employer associations, and government agencies. REFERENCES Abernathy, F., J.T. Dunlop, J.H. Hammond, and D. Well. (1995). "The information-integrated chan- nel: a study of the U.S. apparel industry in transition." Brookings Papers on Economic Activity: Microeconomics 1995. Washington, DC: Brookings Institution. American Apparel Manufacturers Association. (Various years). "Focus: An Economic Profile of the Apparel Industry." Arlington, VA.
APPAREL 36 Arpan, J.S., J. de la Torre, and B. Toyne. (1982). The U.S. Apparel Industry: International Challenge, Domestic Response. College of Business Administration, Georgia State University, Atlanta. Berg, P., E. Applebaum, T. Bailey, and A. Kalleberg. (1996). "The Performance Effects of Modular Production in the U.S. Apparel Industry." Industrial Relations 35(3, July):356-373. Best, M. (1990). The New Competition. Cambridge, MA: Harvard University Press. Braun, K. (1947). Union-Management Co-Operation. Washington, DC: The Brookings Institution. Bryner, E. (1916). The Garment Trades. Cleveland: The Survey Committee of the Cleveland Founda- tion. Carpenter, J.T. (1972). Competition and Collective Bargaining in the Needle Trades, 1910-1967. Ithaca: New York State School of Industrial and Labor Relations, Cornell University. Chandler, A.D. Jr. (1977). The Visible Hand: The Managerial Revolution In American Business. Cambridge, MA: Harvard University Press. Disher, M.L. (1947). American Factory Production of Women's Clothing. London: Devereaux Publi- cations, Ltd. Doeringer, P., B. Courault, L. Oxborrow, E. Parat, and A. Watson. (1998). "Apparel Production Channels: Recent Experience and Lessons for Policy from the U.S., UK, and France," Geneva: International Institute for Labour Studies, Business and Society Programme. Discussion Paper DP/97/1998. Drake, L., and C. Glasser. (1942). Trends in the New York Clothing Industry. New York: Institute of Public Administration. Dunlop, J.T., and D. Weil. (1996). "Diffusion and Performance of Modular Production in the U.S. Apparel Industry." Industrial Relations 35(3, July):334-55. Dunlop, J.T., and D. Weil. (1993). "The Diffusion of Human Resource Innovations: Lessons from the Apparel Industry." Working Paper, Harvard Center for Textile and Apparel Research. Dunlop, J.T., and D. Weil. (1995). "Human Resource Innovations in the U.S. Apparel Industry." Working Paper, Harvard Center for Textile and Apparel Research. Fink, G., ed. (1977). Labor Unions. Westport, Conn.: Greenwood Press. Frank, B. (1953). Progressive Apparel Production. New York: Fairchild Publications. Galenson, W. (1960). The CIO Challenge to the AFL. Cambridge, MA: Harvard University Press. Greenwood, J. (1997). The Third Industrial Revolution. Washington, DC: AEI Press. Grieg, G.B. (1949). Seasonal Fluctuations in Employment in the Women's Clothing Industry in New York. New York: Columbia University Press. Hochman, J. (1941). Industry Planning Through Collective Bargaining: A Program for Modernizing the Dress Industry. New York: International Ladies Garment Workers Union. Hufbauer, G., and K.A. Elliott. (1994). Measuring the Costs of Protection in the United States. Wash- ington, DC: Institute for International Economics. Kuznets, S. (1952). Changes in the National Incomes of the United States of America Since 1870. London: Bowes and Bowes. Magee, M. (1930). Trends in Location of the Women's Clothing Industry. University of Chicago Press. McNair, M., and E. May. (1960). "The American Department Store, 1920-1960: A Performance Analysis Based on the Harvard Reports." Bureau of Business Research Bulletin 166. Harvard University Graduate School of Business, Cambridge, MA. Milgrom, P., and J. Roberts. (1990). "The Economics of Modern Manufacturing: Technology, Strat- egy, and Organization. " American Economic Review 80(June):511-528. Mittelhauser, M. (1997). "Employment Trends In Textiles and Apparel, 1973-2005." Monthly Labor Review 121(8, August):24-35 Murray, L.A. (1995). "Unraveling Employment Trends in Textiles and Apparel." Monthly Labor Review 118(8, August):62-72. National Retail Dry Goods Association. (1936). Twenty-five Years of Retailing, 1911-1936. New York.
362 U.S. INDUSTRYIN2000 New York Times. (1998). "Apparel Making Regains Lost Ground" (January 13, 1998): Al9. Palpacuer, F. (1996). "Strategies Competitives, Gestion des Competences et Organisation en Re- seaux: Etude de cas de l'Industrie New Yorkaise de l'Habillement." Unpublished doctoral the- sis, Universite Montpelier I. Rothstein, R. (1989). Keeping Jobs in Fashion: Alternatives to the Euthanasia of the U.S. Apparel Industry. Washington, DC: Economic Policy Institute. Seidman, J. (1942). The Needle Trades. New York: Farrar & Rinehart. Teper, L. (1937). The Women's Garment Industry. New York: International Ladies' Garment Work- ers' Union. Trebilcock, M.J., and R. Howse. (1995). The Regulation of International Trade. London: Routledge. U.S. Department of Commerce. Bureau of the Census. (Various years). Annual Survey of Manufac- tures. Washington, DC: U.S. Government Printing Office. U.S. Department of Commerce. Bureau of the Census. (Various years). Census of Manufactures. Washington, DC: U.S. Government Printing Office. U.S. Department of Commerce. Bureau of the Census. (Various years). Census of Retail Trade. Wash- ington, DC: U.S. Government Printing Office. U.S. Department of Commerce. Bureau of the Census. (Various years). County Business Patterns. Washington, DC: U.S. Government Printing Office. U.S. Department of Commerce. Bureau of the Census. (1975). Historical Statistics of the United States, Colonial Times to 1970. Washington, DC: U.S. Government Printing Office. U.S. Department of Labor. Bureau of Labor Statistics. (Various years). Employment and Earnings. Washington, DC: U.S. Government Printing Office. U.S. Department of Labor. Bureau of Labor Statistics. (1994). Employment, Hours, and Earnings: United States 1909-1994. Washington, DC: U.S. Government Printing Office.