Vertical Integration and Internet Strategies in the Apparel Industry

by Robert S. Stillman, Robert H. Gertner
Vertical Integration and Internet Strategies in the Apparel Industry
Robert S. Stillman, Robert H. Gertner
The Journal of Industrial Economics
Start Page: 
End Page: 
Select license: 
Select License



We explore the relationship between vertical scope and the ability to respond to a significant economic shock by studying how firms in the apparel industry have adapted to the Internet. We find that vertically integrated specialty retailers, e.g., The Gap, tended to start on-line sales sooner than non-integrated vendors, e.g., Nautica, and depart- ment stores. We also find that the products of vertically integrated retailers are more available on-line than those of non-integrated vendors. These results are consistent with greater contractual barriers, coordination costs and incentive problems that non-integrated brands face relative to integrated companies in responding to the e-commerce opportunity.


THISPAPER EXAMINES the speed with which different firms in the apparel industry began selling on-line and the success they have achieved to date. The apparel industry makes an interesting case study because apparel is one of the leading products sold over the Internet and because there is variance within the apparel industry in organizational form. Some brands are distributed through vertically integrated specialty retailers and catalog companies that handle these brands exclusively. Examples include The Gap, The Limited, Abercrombie & Fitch, Eddie Bauer, Lands' End,

L.L. Bean and J. Crew. Other apparel brands are distributed primarily on a non-exclusive basis through department stores and other non-integrated retailers. Examples include Tommy Hilfiger, Calvin Klein, Polo Ralph Lauren and Nautica. Many of these brands also operate their own retail stores, but sales through these company stores typically are small relative to sales through department stores and other third-party retailers.

*We would like to thank Christine Griffin and Angela Lessuise for their excellent help with the research for this paper. We would also like to thank Raymon Adiletta (Arthur Andersen), Jim Carey (PC Data On-Line), Dan Finkelman (The Limited), Jonathan Hersch (Nautica), Barrett Ladd (Gomez), Richard Last (J.C. Penney) and Shasha Richardson ( for their willingness to share their industry expertise andlor data. We also thank Severin Borenstein, Judy Chevalier, Luis Garicano, David Genesove, Barry Nalebuff, an anonymous referee, and participants in the NBER's E-commerce project for useful comments.

?Author's affiliation: University of Chicago Graduate School of Business and NBER, 1101 East 58th Street, Chicago, IL 60637, USA. email:

1Lexecon, 332 South Michigan Avenue, Chicago, IL 60604, USA.

email: rstillman@ lexecon. com

O Blackwell Publ~rherr Ltd 2001, 108 Cowley Road. Oxford OX4 LIF. UK, and 350 Main Street. Malden, MA 02148, USA.


This variance in distribution models allows us to use the apparel industry as a vehicle for exploring the relationship between organizational form and the ability of firms to adjust to changes in economic conditions. Although the ability to respond to changing economic conditions is a hallmark of economic efficiency, economists know relatively little about the factors that affect it. A common belief is that large bureaucratic organizations suffer from excess inertia and that a significant advantage of small organizations is their flexibility and ability to adapt. Exactly why this should be so, however, is rarely explained, nor have economists developed systematic empirical evidence on the relationship between organizational structure and change. Likewise, despite the large theoretical literature on contracting costs, the flexibility of contractual relationships has not been compared to the flexibility of integrated organizations.

This paper is an attempt to help fill this void. We argue that, depending on the nature of the required adjustments and the nature of transaction costs, vertically integrated firms may be able to adjust to changes in economic circumstances in general, and the Internet in particular, more effectively than non-integrated firms.

Our view is based on knowledge of on-line selling of apparel that we have accumulated through company reports, newspaper and trade press articles, consulting reports, phone calls and interviews with company executives. We know from this research that, when a department store sells apparel on-line, it must coordinate activities among many parties that it does not control. In particular, it typically must obtain permission from its vendors and then negotiate brand guidelines that govern how the brand will be presented on-line. Coordination and other transaction costs are also at the bottom of the so-called channel conflict problems that reportedly constrain the Internet strategies of non-integrated vendors. Our research also identifies an externality that may attenuate the incentives of department stores to invest in developing and promoting on-line sales. When department stores generate incremental sales by investing in their on-line stores, vendors capture a significant share of the gains (unless there are side payments). These factors- coordination costs, channel conflict and externalities-are reasons why vertically integrated specialty retailers such as The Gap may invest sooner and more intensively in on-line sales than department stores and largely non-integrated vendors.

We examine the accuracy of these predictions using data from a sample of 30 firms in the apparel industry. Consistent with our hypothesis, we find that (a) vertically integrated specialty retailers tended to start on-line sales sooner, even after controlling for pre-existing catalog operations and (b) the products of vertically integrated specialty retailers and catalog com- panies are more available on-line than the products of non-integrated vendors.

0 Blackwell Publishers Ltd 2001

We also examine data on web site quality ratings and web site visitors. These results also indicate that the vertically integrated specialty retailers have higher quality sites that are visited more frequently, although the results are of weak statistical significance.


II(i). Projections

Apparel is one of the most common types of product sold over the Internet. Table I presents the results of a monthly survey by the National Retail Federation and Forrester Research of on-line spending by U.S.




Cumulative, January-September 2000 ($ 000)

Small-ticket items:

Health and beauty 
Toys! videogames 
Office supplies 
Linens1 home decor 
Sporting goods 
Tools and hardware 
Small appliances 
Garden supplies 

Total small-ticket items

Big-ticket items:

Air tickets 
Computer hardware 
Hotel reservations 
Consumer electronics 
Car rental 
Food! beverages 

Total big-ticket items

Total online sales $31,313,282

Source; National Retail Federation (

0 Blackwell Publishers Ltd. 2001


Category Sales ($ mm) %I
Store 162,976 88.6
Catalog 17,226 9.4
On-Line I Internet 1,125 0.6
Not Reported 2,535 1.4
Total 183,859 100.0

Note: Total may not sum due to rounding.

Source. NPD Group, Inc.

library I weeklylaa022200a.htm

consumers. The table shows that, among 'small ticket' items, apparel was the second leading category for the first nine months of 2000. On-line sales of apparel during this period were estimated at $1,568 million, just short of the estimated on-line sales of books ($1,614 million).

Even so, the Internet still accounts for a very small share of total sales of apparel in the United States. Table I1 reports estimates by NPD Group of apparel sales by channel. On-line sales of apparel were approximately $1 billion in 1999. In contrast, total sales of apparel in the United States during 1999 were $184 billion. Thus, on-line sales accounted for only about 0.6% of total sales.

The relative importance of on-line sales in the apparel industry is expected to grow. A study by McKinsey & Co. and Salomon Smith Barney estimates that on-line sales of apparel will grow to $7.8 billion by 2003- an estimate that implies a 67% average annual growth rate between 1999 and 2003.' But no one expects on-line sales to come even close to the volume of in-store sales. According to the consulting firm Mainspring, projections for on-line sales of apparel in 2003 range from two to eight percent of total sales.2 Brick and mortar stores will remain the principal channel for apparel sales for the foreseeable future.

II(ii). Multi-channel Retailing

These estimates of future on-line sales mean that, for retailers, sales per square foot at their brick and mortar stores will continue to be far more important to company profitability than the success of their on-line stores. As James Zimmerman, chairman and CEO of Federated Department

' Cited in 'Pure Play: A Losing Model?' The Industry Standard 192-93 (June 26. 2000) Chu, Standley and Reiss [2000].

0 Blackwell Publishers Ltd. 2001.

Stores puts it, 'We are a department store company. The [Internet] tail will not wag the dog.'3

But, even though on-line sales will remain small relative to in-store sales, every retailer we have studied (including the few who have not yet launched on-line stores) emphasizes the ways in which the Internet can complement their efforts to make shopping easier and to communicate better with their customers. The apparel retailers we have studied take it as given thar the Internet is going to be an important part of their future marketing strategies and the way in which they interact with their customers.

Dan Nordstrom describes the concept of 'multi-channel retailing' as follows:

Consumers . . . want options. They may want to peruse the hottest fashion at their local store, but they may ultimately make a purchase through a catalog or on the Web. And if they don't like what they bought online, they may want to be able to return it to a store, not hassle sending it back through the mail.4

Talbots annual report for 1999 provides additional detail on the relationship between catalog, store and on-line sales:

[Olur research tells us that 70% of store customers who receive our

catalog were prompted to visit Talbots stores because of the catalog.

And a significant number of people who order from

the Company's on-line shopping site-first browse the catalog before

they make their selection^.^

Federated's annual report for 1999 describes how both the department store and direct-to-consumer segments of its business will benefit from multi-channel retail integration. According to Federated, 'clicks can help bricks' by driving traffic to stores, reaching and attracting new customers, deepening relationships with existing customers, expanding the range of merchandise offerings, and enhancing gift registry services. 'Bricks can help clicks' by leveraging trusted store brands, providing merchandising ex- pertise, building on established vendor relationships, providing immediate access to products, and making merchandise returns more ~onvenient.~

3'Federated: Not Planning to Let E-Tail Take Over', WWD (May 24, 1999). In a similar vein, Zirnrnerman is reported to have told business students at Indiana University about the 10 most important issues he faces. Issues one through eight were, 'Grow department store comp-store sales by 3% or more a year. That's the single most critical issue.' Issue number nine was, 'Open new stores when they do make sense.' The Internet ranked only tenth on Zimmerman's list-'Fully exploit our direct-to-consumer side, including e-commerce opportunities.' 'Brick and Mortar Man', Cincinnati Post Courier (Oct. 15, 1999).

'Doom and gloom on the e-tail front', Upside Today (Apr. 28,2000).

"Talbots Catalog Sales', Talbots I999 Annual Report (emphasis in original).

Federated Department Stores, Inc. 1999 Annual Report at 4.

0 Blackwell Publishers Ltd 2001

Non-integrated vendors also recognize the complementarities that exist between on-line sales and the other retail channels. Industry executives suggested to us that the best Internet strategy for a nonintegrated vendor may be to build an information site that controls the 'customer experience,' but which then provides links to the website of one or more 'retail partners' (principally department stores) where the customer can complete his or her purchase. Vendors have a comparative advantage in presenting their products to consumers in the most effective way. Vendors also would like to use the Internet to communicate with and learn more about consumers who purchase their products. But vendors recognize that they do not have the infrastructure to fulfill on- line orders, nor do they have a large enough network of retail stores to make it easy for customers to exercise their option to return unwanted merchandi~e.~

II(iii). Economies of Scope Between Catalog Operations and On-line Sales

The complementarity between on-line sales and the other channels of retail distribution ('multi-channel retailing') suggests that catalog firms and firms with stores may have had an advantage over non-integrated vendors in their ability and incentive to begin selling apparel on-line. Nonintegrated vendors such as Calvin Klein, Polo Ralph Lauren and Tommy Hilfiger have small networks of retail stores. But they do not have catalogs, and most of their sales are through department stores and other third-party retailers.

A related factor that may be even more important is pre-existing catalog operations.* Firms that operate catalogs can (and do) use the same systems to fulfill on-line order^.^The only difference from an order fulfillment perspective is how the orders are received-i.e., over the Internet versus through a call center or by mail. According to Lands' End, 'We've been direct marketing by catalog for more than 25 years. Translating our fulfillment method to the Internet was easy. In fact, once an order arrives at our fulfillment center, there is no immediate way of knowing which sales

7This cooperative model of doing business on-line is still in development in the apparel industry. In our sample of apparel firms (described below), we found only one example of links from a vendor's site to the site of a department store 'partner.' Levi Strauss' web site provides links to J.C. Penney and Macy's, its two largest customers. The other vendors in our sample at this point are merely offering store locator services.

In the empirical analysis below, we treat 1995 as the start of the 'e-commerce age.' We selected this date because it was the year after Netscape introduced the browser that arguably spawned public access to the World Wide Web.

'Order fulfillment covers a range of activities including receiving product from vendors; managing warehouse inventories; answering customer questions through call centers; processing on-line orders; picking product from inventory, packaging the product and then arranging for shipping; and handling mail returns.

0 Blackwell Publishers Ltd 2001

channel the customer used because the fulfillment process for each is identical.'1° According to J.C. Penney, 'There's essentially little difference between shipping out Web orders and shipping out catalog orders. The fulfillment skills we learned with the catalog business translate well to the Web, so when we decided to go on line, we were able to focus the better part of our energies on building a strong Web infrastructure-things like site design and server capacity-rather than on figuring out how to fulfill orders on the web.'"

The evidence of economies of scope in order fulfillment between catalog operations and on-line sales seems compelling. This theme appears in company statements and various consulting reports. It also was a recurring theme in our interviews with apparel industry executives. It is somewhat less clear, however, whether these economies can be realized effectively by a non-catalog firm through out-sourcing. There are firms, such as Fingerhut (before its acquisition in 1999 by Federated) and Hanover Direct (a catalog company), that provide fulfillment services to third-party e-retailers.12 J.C. Penney recently entered the third-party fulfillment business by agreeing to provide fulfillment services for Ann Taylor's new on-line shopping site.13

The decision by an e-commerce firm whether to make or buy fulfillment services turns in part on scale. Given the prices charged by third-party fulfillment firms (about 10% of sales price14), it is estimated that a firm must ship 8,000 to 10,000 packages per day to justify the investment required to build and operate a distribution center itself.''

The make or buy decision turns on concerns about customer service. 'Fulfilling customer orders and expectations is no longer an afterthought that happens after the all-important sale. It is a critical component of serving the customer and gaining repeat business and a key business function for any online retailer.'16 And, because order fulfillment is so important to customer satisfaction, there are concerns among e-retailers that out-sourcing risks loss of control., which makes educational toys, recently made the investment required to bring ful- fillment in-house and explained, 'Our [fulfillment] partners were very responsive to our needs, but we wanted to bring things under our own roof

'Paging Rudolph', Chain Store Age (Oct. 1, 2000).

'''Penney's Net Advantage', Chain Store Age (Sep. 1, 2000).

l2 'The Many Channels of Fingerhut', Chain Store Age (June 1,2000).

13'J.C. Penney Co.: Agreement is Set to Handle Orders for anntaylor.com3, Wall Street Journal (June 14,2000).

l4 Chu, Standley and Reiss [2000].

l5 Bhise, Farrell, Miller, Vanier and Zainulbhai [2000].

16'Fulfilling Relationships', Part 1 (

librarylweekly laa0007 11a.htm) (downloaded Nov. 1, 2000).

O Blackwell Publishers Ltd. 2001

so we're better able to react to changes in demand.'17 Our interviews with apparel industry executives uncovered similar thinking.''

Thus there appear to be both strong economies of scope between catalog operations and on-line orders in the area of order fulfillment, and a general (though not universal) belief that third-party order fulfillment specialists is risky, together suggesting that apparel firms that had catalog operations in place at the start of the e-commerce age had an advantage with respect to starting up on-line sales.


In this section, we argue that vertically integrated firms in the apparel industry such as The Gap had a greater ability and incentive to adapt to the Internet than non-integrated department stores and vendors. Our reasons for this hypothesis are all based ultimately on transaction costs. For expositional convenience, we divide the reasons into three groups- coordination costs, externality problems and channel conflict.

III(i). Coordination Costs

The concept of coordination costs has already surfaced implicitly in our discussion of the alleged disadvantages faced by non-integrated vendors that rely primarily on department stores and other third-party stores for retail distribution. It is widely accepted that there are important comple- mentarities associated with a multi-channel approach to retailing. In a multi-channel, 'click and mortar' approach, on-line marketing is coordinated with in-store promotions; on-line purchases can be returned by mail or at the store; stores have kiosks so that customers can make on-line purchases of products that the store does not have in its inventory, etc. Because non-integrated vendors lack a large network of brick and mortar stores (and because they also lack catalog operations), their ability to take advantage of multi-channel complementarities may be constrained.

But note the role of coordination costs in this discussion. If coordination costs were zero, a non-integrated vendor such as Tommy Hilfiger could sell on-line directly from its own web site and negotiate arrangements with its retail distribution partners that addressed the disadvantages to on-line

l7 'Paging Rudolph', Chain Store Age (Oct. 1,2000).

l8 There are several 'pure play' companies that sell apparel on-line including, , and As the names ofa number of them suggest, these companies specialize in discount merchandise, which they obtain from manufacturers or resellers. The way in which these pure play companies handle order ful- fillment lends further support to the view that third-party fulfillment is an imperfect substitute for in-house fulfillment. All of these companies have their own warehouses, arrange shipping themselves, and handle billing internally.

0 Blackwell Publishers Ltd. 2001

selling that the Tommy brand allegedly faces. If coordination costs were zero, Tommy Hilfiger could negotiate side-deals so that on-line marketing was coordinated with in-store promotions, returns of goods purchased on the Tommy web site could be returned at any department store that carries Tommy merchandise, etc. Alternatively, Tommy Hilfiger could negotiate cooperative arrangements of the type described above in which the Tommy site would provide customer information and control the 'customer experience,' but would have links to department store sites where the customer would make his or her on-line purchase. Either way, if coordination costs were zero, Tommy Hilfiger could realize all of the benefits of multi-channel retailing. Thus, the fact that the lack of brick and mortar stores is seen as an obstacle to on-line success for non-integrated vendors is evidence that coordination costs matter.

The concept of coordination costs also surfaced implicitly in our discussion of the risks to customer satisfaction allegedly associated with out-sourcing of fulfillment services. If coordination and other transaction costs were zero, on-line sellers could use third-party fulfillment specialists without worrying about principal-agent problems and the incentive of the specialist (agent) to deliver fulfillment services at the levels desired by the on-line seller (principal). Thus, the fact that concerns exist about the reliability of third-party fulfillment specialists is additional evidence that coordination costs matter.

Several recent papers on vertical integration have focused on coordination issues.19 Gertner [I9991 analyzes how vertical integration affects coordination through different dispute resolution mechanisms. In an integrated firm, a dispute between two units can be resolved by a manager who has control rights over both units' assets. In contrast, a dispute between units that are independently owned will be resolved according to contractual rules that allocate the control right to one or the other unit. The different dispute resolution mechanisms affect the incen- tives to share information and thereby impacts efficient coordination. Although beyond the scope of the model, an implication may be that a vertically integrated firm could more easily share the information needed to coordinate an efficient response to a changing environment.

Knez and Simester [I9991 study the make-or-buy decisions of a producer of complex, technological products requiring many specialized parts. They demonstrate that the decision to produce internally depends on the com- plexity of coordination required in the design and manufacturing of a particular part. This suggests that some coordination problems can be more easily resolved in a vertically integrated firm.

19~lthough issues related to coordination were raised by Coase [I9371 most of the subsequent literature on vertical integration has focused on its effect on the incentives to make relation-specific investments.

O Blackwell Publishers Ltd. 2001

With this as background, we now discuss other coordination costs that may put non-integrated department stores at a disadvantage relative to vertically integrated specialty retailers with respect to their ability and incentive to adapt to the Internet. We understand from our interviews that, before a department store can offer a vendor's products for sale on its web site, the creative departments of the store and vendor typically must negotiate a set of brand guidelines that will govern the on-line presentation of the vendor's brand. The creative departments must also negotiate who will shoot the pictures that appear on the web site and who will have responsibility for the layout of the on-line product description. Once these brand guidelines have been established, it becomes easier for a department store to add additional products from the same vendor; the on-line presentation of the additional products will be governed by the previously negotiated brand guidelines, and job responsibilities will be allocated as before. But, every time the department store wants to add a new vendor to its web site, a new set of negotiations typically is required.

There may be similar discussions from time to time between depart- ments of a vertically integrated firm such as The Gap regarding on-line marketing strategies and product presentation.20 Our hypothesis, however. is that the frequency and costs of these inside-the-firm discussions are consider'ably lower than the costs of the negotiations required between non-integrated brand-owning vendors and non-integrated, web site-operating department stores.

Department stores face other vendor-related problems that a vertically integrated specialty retailer is unlikely to face. For various reasons, certain vendors reportedly do not allow certain department stores to offer their products for sale on-line:

Vendors are another problem. Two years after its launch, is still trying to convince designers to let the store

sell their merchandise online. Kate Spade, for example, won't allow

her handbags to be sold on . . .

Neiman Marcus has hit the same wall. Karen Katz, president and

CEO of Neiman Marcus Direct, which manages the company's

catalog and e-commerce divisions, says is not

permitted to sell Giorgio Armani, Chanel and Estee Lauder because

these companies are in the midst of sorting out their own e-commerce


*'One of the insights of the modern literature on vertical integration is that integration does not magically eliminate the sources of disputes that arise when firms deal with one another on an arm's length basis. As a first approximation, integration converts inter-firm problems into intra-firm, inter-divisional problems. See, e.g., Katz [1989].

'Saks Shops for Customers Online', The Industry Standard 58, 59 (Aug. 28, 2000).

0 Blackwell Publishers Ltd. ZOO1

Macy's and J.C. Penney ran into similar problems in 1999 when Levi Strauss experimented with direct sales from its own web site and requested that its retailers cease offering Levi products on-line. Macy's and J.C. Penney are Levi Strauss' two largest customers.22

III(ii). Externality Problems

Neither the department store industry nor the apparel manufacturing industry is perfectly competitive. This means that when a department store's on-line operations generate incremental sales, the profits from the sale are shared with the vendorlbrand owner-a form of externality. The magnitude of this externality is suggested by a recent study by McKinsey & Co. and Salomon Smith Barney of profitability per order in the on-line retail category in the fourth quarter of 1999.~~

According to this study, the average gross margin on on-line sales for 'apparel manufacturers' was 46%. In contrast, the gross margin for 'multilabel apparel retailers' was 9%. These figures suggest that, when a department store's web site generates an incremental sale of $100, the sale generates $9 in gross profit for the department store and approximately $37 (= $46 -$9) in gross profit for the vendor1 brand owner.

This externality will limit the incentive of retailers to invest in de- veloping and promoting their web sites unless there is some form of co-op funding or restructured pricing. But this requires negotiation, monitoring, enforcement and other transaction costs. We understand from our inter- views that co-op funding does take place and that vendors make payments to department store web site operators in return for premium placement on the web site. Even so, we conjecture that these solutions are imperfect and that an externality problem remains.

The fact that the relationship between vendors and department stores is non-exclusive makes the externality problem even harder to solve. First, because vendors sell through multiple department stores, they would need to negotiate with multiple department stores if they wanted aggregate on- line investment in their brands to be at the optimal levels. Second, because department stores handle multiple brands, the willingness of brand owners to finance department store on-line investments may be dampened by free rider concerns. For example, if Polo Ralph Lauren hypothetically sub- sidized the development of the web site, the vendor might be concerned that other vendors would take a free ride on the improvements in the site.


discussed in more detail below, in October 1999, Levi Strauss reversed its decision to sell on-line direct from its own web site, at which point Macy's and J.C. Penney resumed on- line sales of Levi products. Levi's web site now has links that allow on-line shoppers to make purchases from the web site of either department store.

23 'Pure Play: A Losing Model?' The Industry Standard 192-93 (June 26, 2000).

0 Blackwell Publishers Ltd. 2001

The externality problem may also affect the timing of on-line invest- ments. Initial investments in e-commerce are motivated in part by a desire to learn about the on-line channel and to experiment with different con- cepts of multi-channel retailing. As Dillard7s explained when it launched its on-line store in 1999, 'We're going to put our toe in the water. We're not going to do much business the first year, but we're going to learn.'24 Thus, initial investments in e-commerce have a significant real option com- ponent. The incentive to invest in this option to learn will be attenuated for department stores because of the externality problem that limits their ability to capture the full benefits of their investments in e-commerce.

III(iii). Channel Conflict

On October 29, 1999, Levi Strauss announced a reversal of its e-commerce strategy. Rather than continuing to invest in on-line sales of Levi products direct from its own web site, Levi would stop on-line sales to consumers after Christmas, 1999, and it would permit on-line sales by retailers, beginning with J.C. Penney and Macy's, its two largest customers.25 At the time, many articles appeared that suggested that Levi's decision was driven by concerns about channel conflict. According to these articles, Levi reversed its e-commerce strategy because of complaints from its retailers who were upset that, by selling direct to consumers from its own web site, Levi was competing against them for sales.26

There is considerable evidence that many firms in a variety of industries regard channel conflict as a significant issue. In June 1999, Home Depot wrote to its vendors that, 'We recognize that a vendor has the right to sell through whatever channel it desires. However, we too have the right to be selective in regard to vendors we select, and we trust that you can understand that a company may be hesitant to do business with its competitors.'27 Liz Claiborne explicitly cites channel conflict as a reason for not selling on-line, and stated in late 1999 that, 'We are currently not selling Liz Claiborne products on-line because we don't want to compete with our retail partners. Instead, we are looking for our retailers to help us tap the potential of this emerging channel of distrib~tion.'~~

Channel conflict is also reflected in the manner in which certain manu- facturers have begun on-line sales. For example, it appears to be common

24'Dillard's Plans to Begin Internet Sales This Summer', Dolt Jones News Service (May. 16, 1999). 25 See, e.g., '501 Blues', Business 2.0 53-56 (Jan. 2000) and 'Levi Strauss Will Halt Sales On Its Web Site', Sun Francisco Chronicle (Oct. 29, 1999). 261t should be noted that Levi Strauss denied that retailer opposition contributed to its

decision. 'Levi's to Quit Selling Online', Sun Francisco Examiner (Oct. 29, 1999).

"'Internet Channel Conflicts', Stores (Dec. 1999).

28 Id.

0 Blackwell Publishers Ltd. 2001

for manufacturers to assure their retailers that prices on the manufacturer's site will be at or above the suggested retail prices.29 Other manufacturers have responded to channel conflict by limiting the scope of their on-line product offerings, in some cases only offering entirely new products that are not available ~ff-line.~'

During our interviews, we heard various views on the significance of channel conflict in the apparel industry. Industry executives agree that channel conflict is a major issue for non-integrated vendors contemplating selling apparel directly from their own web site. But executives of department stores and specialty retailers state, in contrast, that they are not worried at all about the analogous channel conflict that could arise between store managers and the managers of these companies' on-line operations. The executives of the department stores and specialty retailers point to the low volume of on-line sales to explain their lack of concern about internal channel conflict. They say they are not worried primarily because the volume of on- line sales to date has been so small that, even if there has been some diversion of sales from the stores, the effects have been nearly imperceptible.

These different views about the significance of channel conflict are consistent with our hypothesis about the effects of vertical integration. On- line sales by department stores and specialty retailers enhance the attractiveness of their total retail operations (multi-channel retailing). This is a benefit to the overall corporation. Eventually there may be some concerns by store managers within these firms about diversion of sales and the effect on their compensation. But our hypothesis is that the costs of resolving such conflicts inside the firm will be relatively low.31 Hence, it is not surprising that executives of department stores and specialty retailers are not especially concerned about internal channel conflict.

The situation is different with respect to on-line sales by non-integrated vendors. The ability of customers to buy Polo Ralph Lauren apparel on- line at enhances the overall attractiveness of Polo Ralph Lauren's retail operations, which is a long-run competitive threat to de-

29'E-Commerce Report', New York Times C-7 (Jan. 3, 2000) (describing Nike's pricing policies); Thunder House, 'Strategies for Addressing Channel Conflict on the Web',Bifocals (Mar. 1999) (discussing the pricing policies of Estee Lauder and Rubbermaid); 'Internet Channel Conflicts', Stores (Dec. 1999) (discussing the pricing policies of VF Corporation, an apparel manufacturer). There is also empirical evidence showing that manufacturers' actual on-line pricing practices are consistent with these assurances. See Carlton and Chevalier [2001].

30'New Covenants Ease Online Channel War', Computerworld (May 17, 1999); 'Doom and Gloom on the E-tail Front', Upside Today (Apr. 28, 2000) (discussing Procter& Gamble's creation of a new line of beauty products for its web site).

31 An example of how internal channel conflict problems might be addressed is suggested by the way in which J.C. Penney measures the performance of its store managers, who are evaluated in part based on a store-level profit and loss statement that includes an attribution of income for catalog and on-line sales to customers who pick-up merchandise at the manager's store or whose zip code is in the store's 'trading area'.

0 Blackwell Publ~shers Ltd. 2001

partment stores and other third party retailers. In deciding whether to offer on-line sales capability, non-integrated vendors need to assess how their department store partners are likely to respond to this threat and whether there should be changes in the terms of the contractual relation- ship between vendors and department stores. Finding an equilibrium solution to these inter-firm issues is likely to require time, negotiation and experimentation. Taken together, it is not surprising that executives in the apparel industry regard channel conflict between vendors and depart- ment stores as a major issue.


Apparel is a very large product category. In selecting a sample of firms in this large industry to study, we began with The Gap because it is widely regarded as one of the leaders in on-line sales of apparel. Thus, our first objective in constructing a sample was to identify vendors and retailers offering merchandise similar to that offered in The Gap. To narrow the field further, we focused on men's wear and women's wear. We excluded children's wear, athletic wear, footwear, intimate apparel and discount stores.

Our first step was to identify apparel firms that were listed as com- petitors of The Gap in The Gap's profile in Hoover's On-Line that also met our product category criteria. Hoover's On-Line ( is a business information 'portal' that offers profiles of thousands of public and privately held companies. This generated 24competitors of The Gap-eight specialty retailers, eight department stores and eight vendors.

We then identified designer, jeans and sportswear brands from the Fairchild 100 list of most familiar brands that were not already in the sample, but whose Hoover's On-Line profile listed The Gap as a com- petitor. This added two vendors to our sample.

We then identified apparel retailers from the Stores List of Top 100 Specialty Stores that were not already in the sample, but for whom Hoover's On-Line listed The Gap as a competitor. This added three specialty retailers to the sample. A further search for department stores on the Stores List of Top 100 Retailers produced no additional department stores that Hoover's On-Line considered competitors of The Gap.

Table I11 shows the 30 firms in our sample and supplies background information that was used in the empirical analysis described below. The table indicates whether the company is selling apparel direct from its own web site and, if so, when on-line sales began.32 The table also shows sales

32 Identifying the year in which on-line sales began required judgment in some cases. J.C. Penney, for example, had a web site during the Christmas season of 1994. But, to place orders. a customer was instructed to call an 800 number. Our rule in constructing Table IV was as follows: We did not consider on-line sales to have begun unless the site displayed merchandise and it was possible for customers to purchase the merchandise on-line.

0 Blackwell Publishers Ltd. 2001

0 Blackwell Publishers Ltd. 2001 

for the most recently available fiscal year, whether the company had catalog operations in 1994 (prior to the start of the e-commerce age), whether the company has catalog operations today and (in many cases) when catalog operations began. The information on when on-line sales and catalog operations began was obtained from web sites, company reports, newspaper articles and phone calls to the companies.


This section reports the results of our empirical analysis of whether there is a relationship between organizational form and the speed and success with which firms in the apparel industry have adapted to the Internet.

V(i). Speed

On-line sales are viewed as a complement to in-store and catalog sales, and all of the apparel retailers we have researched either have begun on- line sales or plan to do so. This, however, does not mean that faster is necessarily better and, indeed, the firms that have gone slower have their explanations. Saks Fifth Avenue did not launch its on-line store until the autumn of 2000. According to an article in the trade press:

Saks says it waited to launch its site until it was certain that the technology-enlarge and zoom functions, along with easy checkout and payment-would work. 'We spent a tremendous amount of time making sure that no one would ever have to use the Back button,' says Denise Incandela, the COO of Saks Direct, which oversees the store's e-commerce offerings.33 , which did not launch until October 1998, suggested to us in an interview that one of its concerns was modem speed. Selling apparel on-line requires displaying high quality pictures of the merchandise. Downloading these pictures can become a source of consumer frustration if the consumer has a slow modem.

The Limited is the only specialty retailer in our sample that has not begun on-line sales, but is engaged in e-commerce through its controlling interest in Intimate Brands, Victoria's Secret owner. However, The Limited does not offer on-line shopping for its Limited, Express, Lane Bryant or Lerner New York brands, although it expects to offer on-line shopping for these brands in the relatively near future. The executive we interviewed emphasized that because brand names and product differ- entiation are so important in the sale of apparel, there are no first-mover

33 'Saks Shops for Customers Online', The Industry Standard 58, 59 (Aug. 28. 2000)

0 Blackwell Publishers Ltd 2001

advantages associated with on-line sales, so they are able to wait and learn from others about what works on the Internet. In his view, apparel brands are sufficiently differentiated that first-mover advantages are a non-issue.

Despite these idiosyncratic differences in e-commerce strategies, the timing of on-line sales in the apparel industry is also the product of systematic factors related to whether a firm is vertically integrated. The evidence of this relationship is summarized in Tables 111, IV and the first column of Table V. Table I11 is a time line that shows, by year, when different companies began selling apparel on-line. The companies that had catalog operations prior to 1995 are marked with an asterisk. Corporate sales for the most recent fiscal year on are in italics.

Table I11 by itself shows some clear patterns. There are seven specialty retailers in our sample that did not have pre-existing catalog operations. Six of these seven firms are now selling apparel on-line. In contrast, there are two department stores in our sample that did not have catalog operations prior to 1995. Neither of these chains has yet begun on-line sales of apparel. Similarly, there are nine non-integrated vendors in our sample that did not have pre-existing catalog operations. Only three of these nine firms have begun selling apparel on-line. Table IV reports cumu- lative frequency distributions and is self-explanatory.

The first column of results in Table V reports the results of a Tobit regression where the dependent variable is the first year of on-line sales. If the company did not initiate on-line sales by November, 2000, we treat the observation as censored. There are four explanatory variables: a dummy variable to indicate if the firm is a non-integrated vendor (VENDOR), a dummy variable to indicate if the firm is a department store (DEPT), a dummy variable (PRE95CAT) to indicate if the firm had catalog operations prior to 1995 and the sales of the company in its most recent fiscal year SALES).^^ Given these are reduced-form regressions, one could question whether we should include PRE95CAT in the regression analysis. Since there are economies of scope between catalog and e-commerce operations, and some of the same organizational factors may cause both catalog sales and early Internet sales, including PRE95CAT may weaken our results. We choose to control for catalog operations, none the less, to confirm that economies of scope between catalog and e-commerce operations are not the sole explanation for the relation between organizational form and timing of on-line sales.35

The coefficient on both VENDOR and DEPT are positive and statistic-

34The omitted category of apparel firms in our sample is vertically integrated specialty retailers.

350mitting PRE95CAT from the regression leads to a larger coefficient on VENDOR and a smaller coefficient on DEPT but does not change the qualitative interpretation of the results.

O Blackwell Publishers Ltd. 2001


Cumulative Frequency of On-Line Sales Starting i=

1996 1998 2000 Total


Frequency Specialty Dept Specialty Dept Specialty Dept Specialty Dept 2 Distribution Retailers Stores Vendors Retailers Stores Vendors Retailers Stores Vendors Retailers Stores Vendors

Catalog before 3 1 0 4 4 1 5 6 1 5 1995 60.0'Xr 16.7'%1 0.0%) 80.0% 66.7%) 100.0'%, 100.0% 100.0%) 100.0%r

No catalog 0 0 0 2 0 1 6 0 3 7 2 9 $ before 1995 0.0% O.Onh 0.0% 28.6'%, 0.0% 11.1% 85.7'%1 0.0% 33.3% rn








SPEED RATINGS VISITORS Tobit Regression of OLS Regression OLS Regression Time of On-Line of Gomez of Unique Sales Scores Visitors

First year of on- Gomez Ratings for PC Data Estimate

line sales, censored Fall 1999, Spring of Unique Visitors

if not on-line by 2000 and Summer 12/99, 5/00, and Dependent Variable 1 1 12000 2000. 9/00 (in 1000s)

Dummy for non-integrated vendor

Dummy for department store

Dummy if catalog < 1995

Sales (in $ millions)


Gap Dummy

J.C. Penney Dummy



ally significant. This confirms that the non-integrated apparel companies were slower at introducing on-line sales, even controlling for pre-existing catalog operations. The size of the coefficients are economically significant, implying that the non-integrated firms are approximately two years behind integrated firms.

These results are consistent with our argument that vertically integrated firms in the apparel industry have begun on-line sales earlier because of advantages with respect to coordination costs and investment incentives. They are inconsistent with timing of on-line sales in the industry's being due to firm-specific factors that are unrelated to organizational form.

V(ii). Product Coverage

The timing of initial on-line sales may not fully capture the differences in the e-commerce strategies across organizational forms. Another important feature that may differ across the companies in our sample is the breadth and depth of product coverage available on-line.

0 Blackwell Publishers Ltd. 2001

There are reasons to think that specialty retailers have advantages in extending the scope of their on-line product offerings. Specialty retailers do not have to negotiate with vendors over whether a brand can be sold on-line and, if so, the guidelines for the brand's on-line presentation. In addition, the incentive of specialty retailers to invest in content manage- ment systems is not attenuated by the externality problem discussed above-the difficulty that department stores have in capturing the full gains from incremental sales.

A consumer can buy from the Gap or J. Crew website nearly everything that he or she can buy in their stores. Table VI shows that the same is generally true for all of the specialty retailers and catalog companies in our sample. Their on-line product selection typically is the same or greater than the product selection available in their stores or through their catalogs.

A customer who is used to buying Nautica merchandise at a department store may have more difficulty buying Nautica merchandise on-line. First, his or her favorite department store may not have an on-line store. May Department Stores is one of Nautica's biggest customers. None of the May department store chains is selling on-line. Second, even if his or her favorite department store is selling on-line, the department store's on-line

When Did On-Line Description of On-Line Product
Specialty Retailers: Sales Begin? Offering
Abercrombie &Fitch 1999 118 items. Complete product line
    available on-line.
American Eagle 1998 188 items. Everything in catalog
Outfitters   is available on-line. On-line store
    is meant to mimic physical store.
Eddie Bauer Sep-96 806 items. Entire catalog
    available on-line. More available
    on-line than in stores.
  Nov-97 270 items. More items on-line
    than in the stores.
J. Crew Jul-97 353 items. The stores contain
    60% of the merchandise sold
    on-line and in the catalog.
L.L. Bean Nov-96 The catalog and on-line offerings
    are the same.
Lands' End Jul-95 Catalog and on-line selection are
    the same. Outlet stores may have
    some items not in catalog1
Talbots Nov-99 Almost everything in the catalog
    is on-line. More in the catalog
    than in store.

0Blackwell Publishers Ltd 2001




Department Calvin Levi   Polo Ralph Tommy
Store Klein Strauss Nautica Lauren Hilfiger
Dillards Yes Yes Yes Yes Yes Yes, 2 No Yes, 17 Yes, 15 Yes, 28
Bloomingdales Yes Yes Yes Yes Yes Yes, 61 No Yes, 7 Yes, 73 Yes, 9
Macys Yes Yes Yes Yes Yes Yes, 57 Yes, 42 Yes, 8 Yes, 41 No
Nordstrom Yes Yes Yes Yes No Yes, 18 No Yes, 4 Yes, 59 No
JC Penney Yes Yes No    
jcpenney .corn Yes, 115 Yes, 300+ No    
Saks Yes No No Yes No Yes, 8 No No Yes, 19 No

Notes Based on research conducted August-November 2000

store may offer only a limited selection of Nautica apparel. As shown in Table VII, this is the situation (as of November, 2000) for Nordstrom and . Table VII also identifies situations where a brand's merchandise is available in the store, but not available at all for on-line purchases as well as the number of distinct products it offers from that vendor. Third, the customer trying to buy Nautica apparel on-line will not receive much assistance from Nautica's web site. Nautica's web site does not offer on-line sales. It is an informational site that describes its product line. And it currently has no links that would help the customer find a place to buy Nautica products on-line. Instead, the site merely offers a store locator function.

A comparison of the product mix, brand image, and size of J. Crew and Nautica make it difficult to explain why J. Crew khakis should be easily available on-line while Nautica khakis are not. We believe that at least part of the reason that department store selection is limited is higher coordination costs and externality effects and that channel conflict con- cerns have limited the on-line sales of the non-integrated vendors. Why have most vendors not created links to sites where their merchandise can be purchased on-line? We suspect that this is another manifestation of coordination costs--coordination costs avoided by specialty retailers whose sites combine both product information (displayed in a way that promotes the brand) and the opportunity to make on-line purchases.

O Blackwell Publishers Ltd. 2001

V(iii). Web Site Ratings

We also attempt to compare the quality of web sites across the companies in our sample by analyzing data on the e-commerce ratings assigned to firms in our sample by, a company that specializes in reviewing e-commerce operations. Gomez publishes reviews quarterly and has a special section on apparel firms. The Gomez reviews are prepared by company employees based on examination of web sites, performance monitoring of secure and non-secure pages, and customer service inte- raction over the telephone and through the Internet. The Gomez analysts rate e-commerce businesses based on up to 150 subjective criteria. The specific criteria used in rating apparel sites include whether the site provides an on-line fit guide, a virtual changing room, and an on-line gift registry. The general criteria relate to ease of use, customer confidence, on site resources, and relationship services.36

Gomez provided us with historical data on its ratings of apparel sites for autumn 1999, spring 2000 and summer 2000. Gomez 'overall scores' have a theoretical range between 0 (the worst) and 10 (the best). Gomez reported ratings for fourteen firms in our sample, which does not include all firms that had on-line businesses. The average Gomez rating for the firms in our sample is 5.5 with a standard deviation of 1.1.

To examine these data, we pooled the data and estimated a simple regression model. The dependent variable was the Gomez rating. The explanatory variables were dummy variables indicating if the firm was a vendor or department store; dummy variables for the autumn 1999 and spring 2000 seasons (to control for time-related fixed effects in the Gomez ratings); a dummy variable if the company had catalog operations prior to 1995; and overall company sales. We calculate standard errors assuming that the multiple observations for each firm are not independent.

The results are reported in the middle column of Table V. The coefficients on the vendor and department store dummy variables are negative. The vendor coefficient is statistically insignificant while the department store coefficient is significant. The point estimate for the department store co- efficient implies that the department store web sites are approximately one standard deviation below the mean for our sample.

V(iv). WebSite Visitors

We also analyzed data on web site visitors as another possible method of examining web site quality and investment. The data for this analysis were provided by PC Data On-Line ('PC Data'). PC Data provided the estimated number of unique web visitors per month for the months of



0 Blackwell Publishers Ltd. 2001.

December, 1999; May, 2000 and September, 2000. If a consumer visits a web site once or 10 times during a month, he or she is counted as one unique visitor for that month. We were told by PC Data that unique web visitors is the most commonly used measure of web site popularity in the e-commerce community. PC Data's estimates of web site visitors are based on a panel of 120,000 households in the United States, each of whom has loaded software that allows PC Data to track their Internet activity.

PC Data collected data on eighteen of the companies in our sample. The average number of unique visitors for our sample firms is 680,000 with a standard deviation of 895,000. There are two clear outliers-J.C. Penney and The Gap; the number of unique visitors to these two sites is many times the number of visitors to any other site. The average and standard deviation of unique visitors drops to 414,000 and 355,000, respectively without J.C. Penney and The Gap.

We have run a regression incorporating dummy variables for J.C. Penney and The Gap in addition to the set of regressors in our previous analyses. We pool the three sample periods and include dummies for each sample and adjust the standard errors for clustering. The results are summarized in the last column of Table V. The signs on the VENDOR and DEPT coefficient estimates are negative, implying less traffic for the sites of the non-integrated firms. The coefficient for VENDOR is statistically and economically significant while the estimate for DEPT is only marginally significant.


This study provides a striking example that the extent of vertical inte- gration can have an economically significant impact on the costs of adapting to important changes in the economic environment. The coordination, externality, and channel conflict issues appear to be greater and more difficult to overcome in non-integrated apparel firms. It does not follow that these costs can not and will not be overcome, merely that it has been more difficult and more costly for these firms to respond to the e-commerce opportunity.


Bhise, H., Farrell, D., Miller, H., Vanier, A. and Zainulbhai, A., 2000, 'The Duel for the Doorstep', The McKinsey Quarterly, No. 2, pp. 32-41. Carlton, D. and Chevalier,J., 2001, 'Free Riding and Sales Strategies for the Internet', Journal of Industrial Economics (this issue). Chu, J., Standley, A. and Reiss, A., 2000, 'Dressed for Success: Apparel Competitors Move Online', Mainspring, August 14, 2000. Coase, R. H., 1937, 'The Nature of the Firm', Economica 4, pp. 386405.

0 Blackwell Publishers Ltd. 2001

Gertner, R., 1999, 'Coordination, Dispute Resolution, and the Scope of the Firm', unpublished manuscript, Graduate School of Business, University of Chicago.

Katz, M., 1989, 'Vertical Contractual Relations', in R. Schmalensee and R. Willig (eds), Handbook of Industrial Organization, Vol. 1 (Publisher here, city).

Knez, M. and Simester, D., 1999, 'Direct and Indirect Bargaining Cost and the Scope of the Firm', unpublished manuscript, Sloan School of Management. M.I.T., Cambridge, Massachusetts.

0Blackwell Publishers Ltd. 2001

  • Recommend Us