Private Labels and Exports: Trading Variety for Volume∗ Emily Blanchard†, Tatyana Chesnokova‡, and Gerald Willmann§ March 30, 2017 Abstract This paper explores the role of private label trade intermediation in shaping the range and diversity of exports and imports. Whereas direct sales maintain a firm’s unique prod- uctcharacteristics,or‘brandequity’,tradethroughanintermediaryoftentakestheformof ‘private label’ sales, under which multiple firms’ output is pooled and re-sold under a new private label brand created by the intermediary. This paper shows that these private label arrangements result in greater total export and import volumes and lower average prices for consumers, but fewer independent varieties available to consumers in equilibrium. Nor- mative implications are mixed: consumers trade variety for volume, independent exporters face greater competition from the new private label products, and intermediary firms can capture more of the gains from trade. We explore the implications of competition at the intermediary level and trade costs for the equilibrium pattern of private label and direct exporting and importing activities. Keywords: Private Labels, Intermediaries, Heterogeneous Firms, International Retailers JEL Classifications: F13, F16, D72, E60 ∗Wethanktheeditorandtwoanonymousrefereesforcommentsandsuggestionsthathavegreatlyimproved thepaper. †TuckSchoolofBusinessatDartmouthCollege,[email protected] ‡WasedaUniversity,[email protected] §BielefeldUniversityandIfWKiel,[email protected] 1 1 Introduction Despite the close scrutiny afforded to barriers to trade and economic openness, economics re- search on international market access focuses almost exclusively on physical and political trade costs: tariffs, non-tariff barriers, and the transportation costs associated with physically moving productstomarket. Relativelylittleattentionhasbeengiventothecommercialrealitiesofinter- national market access – the role and potential failure of the market mechanisms through which exportersinonecountryreachconsumersinanothercountry,whetherthroughdirectshipments, wholesalers, product-sourcing arms of international retailing firms or other intermediated trade channels.1 Inthispaper,wejoinasmallbutgrowingliteraturethataddressestheseissuesbyex- aminingtheimportanceofintermediariesintrade. Ourkeycontributionistopointoutthatnot only do intermediaries shape firm’s exporting decisions (which firms export and whether they ship directly or indirectly), but also, crucially, that intermediation can fundamentally change the characteristics of exported products, and thus the variety and nature of imports available to consumers. Our starting point is to recognize that there are two distinct forms of trade intermediation, each with different implications for the equilibrium pattern of trade. The first form is the conventionalnotionadoptedbytheexistingliterature,whereatradeintermediaryservesasago- betweentoreducetheaveragecostoftransportationforpotentialexporterseitherbyresolvingan informationasymmetryorincompletecontractsproblem,2 orbyeconomizingontrade(orsearch) costs.3 Crucially, this existing work implicitly assumes that intermediation does not otherwise change the underlying characteristics of the individual products shipped abroad. In contrast, we consider in this paper the possibility of transformative trade intermediation, under which exported products are fundamentally changed by the process of intermediation. In this paper, wearguethatpooled-producersourcing–thewidespreadpracticeinwhichatradeintermediary sources products from multiple independent producers to re-brand under a new umbrella brand –whatthemarketingliteraturecallsa‘privatelabel’–constitutesanempiricallyimportantand as yet unexplored form of transformative trade intermediation.4 We identify and explore the 1Animportantexceptionisearlyworkattheintersectionbetweenthetradeandindustrial-organizationliter- aturesthatstudiestheeffectoftradeliberalizationonfirms’behaviorandendogenousmarketstructure;see,for example,RaffandSchmitt(2009)orRaffandSchmitt(2012). 2See Rauch and Watson (2004), Feenstra and Hanson (2004), Felbermayr and Jung (2008), and Felbermayr andJung(2011) 3See Blum et al. (2010), Head et al. (2014), Antra`s and Costinot (2011), Ahn et al. (2011), and Akerman (2010). Alongasimilarline,Baietal.(2015)suggestthatintermediationmayreducedynamicmarket learning potentialforexporters. 4Hereafter, we use the term ‘private label’ to indicate any pooled product, which need not be a store brand 2 tradeoffs inherent to the two forms of product intermediation, and particularly the potential for horizontal product homogenization and profit-shifting via private label sourcing. Buildingfromrecentempiricalandtheoreticalworkintheheterogeneousfirmsliterature,we develop a model of private label trade intermediation. The theoretical exercise delivers key in- sightsthatbuildonandrefineexistingworkontradeintermediation. Firstandmostconcretely, private label contracts offer an additional form of low-cost market access for exporters, allowing more firms at the lower-end of the firm brand spectrum to reach foreign consumers. The pri- vatelabelchanneltherebyintroducesanotherextensivemarginfortradeadjustmentinaddition to traditional (brand-preserving) trade intermediation and direct exporting. This private label tradechannelhassubtleandimportantdifferencescomparedtoconventional(brand-preserving) wholesale trade intermediation or direct exports.5 Most importantly, when trade intermediaries can pool products under a single private label brand, intermediation leads to product homog- enization in equilibrium — forcing consumers to trade off variety for volume. By identifying a plausible mechanism that would cause the endogenous loss of product differentiation, our model carries immediate welfare implications for the broad class of trade models that leverage ‘love-of-variety’ preferences. Atthesametime, wefindthattheeffectsoftradecostsandmarketpoweramongintermedi- ariesaremorenuancedthanexistingmodelsoftradeintermediationsuggest. When,forinstance, changes in trade costs or market concentration among intermediaries make trade via a private label more attractive, entry by low-end firms causes a ‘brand dilution’ effect, which can drive higher-end firms into brand-preserving exports even in the presence of stronger pro-competitive effects of expanded trade. We show that intermediaries’ (monopsony) market power can have non-monotonic effects on trade: too much concentration, and aggressive private label contracts crowd out direct exporters, limiting overall trade; too little concentration, and private label brandequityisdilutedsomuchthatoverallexportsfall. Theseresultsarenewtotheliterature, and suggest a more complex relationship between trade intermediation and trade patterns than has been found in existing work. Empirical Context. This paper is motivated by recent empirical work that demonstrates boththeimportanceofintermediariesintradeandsystematicvariationinintermediaryinvolve- ment across sectors and trading partners. Using detailed firm-level trade and transactions data for the United States, Bernard et al. (2010) find that intermediaries are disproportionately in- volved in trade with lower wage countries and in consumer goods sectors such as agricultural andisnotnecessarilyassociatedwithaparticularretailer. 5Intheinterestoftractability,thetheorypartofthispaperabstractsfrombrand-preservingtradeintermedi- ation(orinterpreteddifferently,subsumesitasaformofdirectexporting). 3 products, clothing, and footwear. Ahn et al. (2011) and Bai et al. (2015) find similar patterns for China, Akerman (2010) for Sweden, and Blum et al. (2010) for South America.6 At the firmlevel,evidenceisbroadlysuggestivethatthebiggest,mostproductivefirmsexportdirectly, while the majority of (typically much smaller) exporting producers use intermediaries on one or both sides of the border to reach foreign consumers. Taken together, these studies suggest the intermediariesaremostinvolvedintradewithlower-wagecountries,lessdifferentiatedproducts, and smaller exporters. Private label sales are likewise an important feature of the commercial landscape and follow similarpatterns. AccordingtoACNielsen(2005),privatelabelsalescomprisealargeandgrowing shareofretailpurchases,makinguproughly17%ofsalesatsurveyedretailersacross38countries and 80 categories.7 Private label sales are increasing world wide, with the strongest growth in emerging markets (where the share of private label products has been increasing at 11 percent per annum) compared to more modest growth in developed countries (e.g. 4 percent per year in Europe). Like intermediated trade more generally, the importance of private label sales varies markedly across product categories and is greater at the lower-end of the market, where private label products are on average priced 31% lower than their manufacturer-branded counterparts.8 At the same time, private label trade seems particularly important for trade. In interviews reportedbyTimmorandZif(2008),exportmanagerscitedprivatelabelsalesasmoreimportant for success in exporting than for domestic sales. The same study also indicates that exporting underaprivatelabelismorefrequentlyobservedinconsumergoodssectorslikefood,beverages, and textiles. Oneexampleofanindustrywhereprivatelabelsareparticularlywidespreadisinappareland textiles, which is also among the most highly traded sectors globally. In the U.S., the expansion of private labels in apparel began in the 1980s, and within a decade private labels constituted about 25% of the total US apparel market.9 One of the first retailers that pursued private label 6In related work, Head et al. (2014) test the impact of international retailers’ local Chinese operations and subsequentexport activity fromChina,whileBaskerandVan(2010a)andBaskerandVan(2010b)considerthe impactofWal-Mart(amajortradeintermediary)onU.S. imports fromChina 7Surveyedretailersincludedsupermarkets,hypermarkets,massmerchandizersandsomedrug-andconvenience stores. Aseparate2011studyin‘PrivateLabelMagazine’reportedsimilarfiguresforindividualretailers: in2010, private label sales made up 18% of revenue at Wal-Mart, 24% at Costco, and 30% at Target Corp. stores (cf. Private.Label.Magazine(2011)). 8According to the same ACNielsen (2005) study, the highest private label market shares are in refrigerated food (32 percent) and paper, plastic & wraps (31 percent), and lowest in cosmetics (2 percent). There is also substantial variation in the price differential between private label and manufacturer-branded products ranging frompersonalcareproducts(whereprivatelabelssellfor46percentlessonaverage)torefrigeratedfood(witha pricedifferentialof16percent). (ibid) 9SeeGereffi(1999). 4 strategyisJ.C.Penney,whoseprivatelabellinesaccountforupto60%ofwomen’sapparelsales. As J.C. Penney is a pure retailing company and not a manufacturer, it imports apparel for its private labels from lower-wage countries, such as Mexico. In 1994, J.C. Penney established a buyingofficeinMexicoCityanditsprocurementofapparelwentfrom$7millionin1994to$100 million in 1999. J.C. Penney buys apparel products, such as tee-shirts, underwear, and jeans, from twenty two independent Mexican manufacturers.10 Here, a major international retailer – J.C. Penny – is serving as an intermediary that links Mexican exporters with U.S. consumers via private label arrangements. To fix ideas further, and to highlight another connection between trade intermediation and pooled-producer sourcing, consider a second example - the wine industry in New Zealand. Over the period between 2001 and 2011, New Zealand’s wine market has seen a dramatic expansion of private label exports.11 Bulk wine is exported not in bottles, but in massive wine “bladders” of 20,000 liters or more. This bulk wine is shipped to retailers who blend and bottle it under private labels in major markets like the U.S., U.K., and Germany. In 2007, bulk exports were only 5 percent of total exports of New Zealand wine, but by 2011, this share had increased to 35 percent.12 Many have argued that the dramatic increase in bulk and private label exports has amplified competitive pressure on independent New Zealand producers: a business survey conductedinearly201213 citedbulkwineexportsasakeyreasonthat56percentofNewZealand wineries suffered negative profits in 2011. Thus, at the same time that bulk wine homogenizes private label products via physical pooling, it can also reduce the market for independent vari- eties through a pro-competitive effect. Both of these effects contribute to a loss of variety for consumers, even as overall exports rise. We view this potential tradeoff between variety and volume as an important and as yet un- explored feature of trade intermediation. Accordingly, we build a model to highlight the tension betweenprivatelabelanddirectexporting,andshowthatthenatureoftradeintermediationcan have important implications for the firms and varieties that succeed in the global marketplace. The reader may note that while we present and frame our analysis from the perspective of the exporting country, exports turn into imports the moment they reach their destination, so our results apply more generally to imports as well. 10Forexample,J.C.PenneyalsoimportsdirectlyfromaMexican-ownedmanufacturerLibrainTorreon,Mexico which claims the title of ‘Blue Jeans Capital of the World’. See Bair and Gereffi (2003), and Bair (2002) for a reviewoftheapparelindustryinMexico. 11From2001to2011,theshareofwineexportsinbulkincreasedfrom20percenttoalmost50percentinnew worldcountries. SeeRabobank(2012). 12SeeANZ(2012)report. 13AscitedinPPB(2012)report. 5 Findings. We build a model of private label sourcing and trade intermediation that ties the prevalence of private label exports to fundamentals (market size, preferences, and costs of exporting). We start with a tractable heterogeneous firms model of intermediated trade, to which we introduce private label contracts. The model incorporates micro-founded building blocks from earlier work to identify an intuitive, plausible, and general sorting mechanism by whichfirmsofdifferingex-anteproductcharacteristicsself-selectintoexportmodes. Thelargest exporters ship products directly (perhaps by establishing a foreign wholesale subsidiary, as in Felbermayr and Jung (2008)), while smaller exporters ship indirectly through intermediaries, and the smallest and least productive firms do not export at all. The model highlights the key differences between two distinct business models for trade intermediation: brand-preserving exports, which would be preferred by larger exporters in more differentiatedproducts(whereexportingfirmscompeteonbrand-equityandcost),versusprivate labelcontracts(whereexporterscompeteoncostalone). Weshowthattheavailabilityofprivate label trade intermediation increases total exports, reduces profits of the direct exporters, and induces some former direct-exporters to switch to private label exporting. The net effect on the totalnumberofexportersisambiguous,however: privatelabeltechnologyprovidesanadditional mode of accessing the export market, leading to entry, but it simultaneously introduces a stark pro-competitive effect, pushing firms to exit. We find that the second effect dominates the first (i.e. there is net exit) when the intermediary’s cost advantage over direct exporters is large, products are less differentiated, or exports from the rest of the world are large. Using our model to study intensive and extensive margin adjustments to changes in trade costs, we find that an increase in variable trade costs reduces the range of direct exporters and shifts the range of exporters who use the intermediary toward higher brand equity. This strengthens the private label brand, but reduces the total range of exporting firms. At the same time, on the intensive margin, the quantity exported by an individual exporter via the intermediary remains unchanged, whereas a (surviving) individual direct exporter exports more following the reduction in competition from fewer differentiated products. Net, the extensive margin dominates, so that total export volume falls with an increase in variable trade costs, even as the most successful (direct) exporters become larger. Turningtofixed costs,weagainfindasymmetricresultsfordirectvs. intermediatedexporters. Because intermediaries allow private label exporters to share the burden of fixed trade costs (whereasadirectexportermustbearfixedcostsalone),anincreaseinthefixedcostofexporting would cause the range of firms who use the intermediary to increase and the range of direct exporters to shrink; the effect on the total number of exporting firms is generally ambiguous. Ontheintensivemargin, anincreaseinfixedcostswouldleadafirmthatusesintermediationto 6 export more, while the net effect on total export volume and individual direct export volumes are, in general, ambiguous. In a final step, we use the model to explore the implications of market power exercised by tradeintermediaries. Weconsiderfirstacaseinwhichmanyretailerseachoffer(exclusive)access to a subset of destination market consumers, and second a scenario where a single retailer faces the threat of entry by a potential competitor and thus has to reduce the fee it sets to extract profit from exporting firms. While the insights from the baseline version of the model prove robust to the variation in market structure, a reduction in market power due to potential entry has significant implications. Small reductions in the market power of the intermediary render exportingunderitsprivatelabelmoreattractive: whiledirectexportsfall,privatelabelandtotal exports increase. As the pricing power of the retailer vanishes further, however, the negative aspects of the private label start to dominate as lower brand equity exporters are absorbed into the private label pool, leading to an increase in direct exports and a decline in overall export volumes. Related Literature. Our paper is most closely tied to the growing trade literature on the importance of intermediaries in trade. As noted earlier, the existing body of work treats trade intermediaries as a go-between that reduces the average cost of transportation for potential exporters. Theliteratureofferstwodifferentwayshowtheintermediariesfacilitatethisreduction in transportation costs for exporters. In the first group of papers: Rauch and Watson (2004), FeenstraandHanson(2004),FelbermayrandJung(2008),andFelbermayrandJung(2011),the intermediary resolves an information asymmetry or incomplete contracts problem. The second explanation is suggested by Blum et al. (2010), Head et al. (2014), Antr`as and Costinot (2011), Ahn et al. (2011), and Akerman (2010), where the role of the intermediaries is in economizing on trade (or search) costs. Along a somewhat different line, Bai et al. (2015) develop and findempiricalsupportforastructuralmodelinwhichtradeintermediationmayreducedynamic marketlearningpotentialforexporters. Incontrasttothesepapers,ourcontributionistoanalyze transformative trade intermediation via product pooling and homogenization. A pair of papers, orthogonal to our own but similar in spirit, look at potential implications for vertical quality adjustmentsinresponsetointermediation;seeDasguptaandMondria(2011)andIacovoneetal. (2016)fortheoreticalmodelofendogenousqualityupgradinginaheterogenousfirmsframework. Our paper also contributes to existing work on private labels. In the (small) industrial organization literature on private labels, Mills (1995) and Gabrielsen et al. (2007) focus on how private label introduction affects the division of profits between manufacturer and retailer. For a broader review of the literature see Berg`es-Sennou et al. (2004). In the marketing literature, 7 study of private labels center on consumer behavior, largely through case studies. Unlike these papers, we do not aim to explain the existence of private label sourcing, but rather to analyze theeffectsofavailabilityofthisformoftradeintermediationonthevarietyandselectionoffirm exports. Our paper is the first to analyze the effects of private label intermediation in a market setting where firms produce differentiated products. The remainder of the paper proceeds in the usual sequence. The next section presents a simple model of private label trade intermediation, and analyzes the effects of a private label option for firm behavior and the equilibrium pattern of trade. Sections 3 and 4 present a set of policy-relevant comparative statics, evaluating the effects of changes in (fixed and variable) trade costs and variations in market power at the intermediary level. Section 5 concludes. 2 The Model Inwhatfollows,wedevelopasimplepartialequilibriummodelofdirectandintermediatedtrade in which horizontally differentiated firms in a small open economy – ‘Home’ – compete to serve consumersinaforeigntradingpartner,‘Foreign’. Exporterschoosebetweentwodifferentmarket access channels to reach consumers abroad. The first channel is direct exports, which preserve exporters’ unique brand-equity, but entail a higher fixed cost. The second option is private label trade intermediation via an international firm-retailer. Under this option, exporters are required to pool their products under a single umbrella private label brand that is controlled by the intermediary. The fixed and variable costs of exporting via the private label contract are set endogenously by a profit-maximizing international firm-retailer. In the baseline version of the model, this intermediary acts as a (single price) monopsonist; Section 3 later relaxes this assumption to consider variation in the degree of market power. We present the model in stages, beginning with the basic set-up and then introducing direct exporting and private label trade intermediation in turn. Sections 3 and 4 then use the model to explore a series of comparative statics exercises. 2.1 Basic Model Consumers. The Foreign target market consists of a mass of L consumers. These consumers areservedbybothHomefirmsandtherestoftheworld. Inkeepingwithasmall-countrysetting, we treat Home exporters as atomistic profit maximizers that take as given the aggregate sales fromtherestoftheworld. Tokeepthemodelassimpleaspossible, wefocusononlytheforeign targetmarket, omittingadomesticmarketatHome. Thissimplificationisoflittleconsequence, 8 and does not change the key results for firm selection or trade patterns.14 Consumer preferences are identical and given by the following quadratic utility function, which mirrors that in Ottaviano et al. (2002): (cid:90) 1 (cid:90) 1 (cid:18)(cid:90) (cid:19)2 U =qc+α λ qcdi− γ (qc)2di− η qcdi . 0 i i 2 i 2 i Intheexpression,qc isindividualconsumptionofatradablenum`erairegood15 andqc isindivid- 0 i ual consumption of each given differentiated product i. The parameter α expresses the intensity of preferences for the differentiated product relative to the num`eraire, while parameters γ and η are both strictly positive, which ensures that consumers prefer dispersed consumption of va- rieties (love of variety). Based on this starting point, then we introduce a new ‘brand equity’ parameterλ foreachproducti, whichactsasaverticaldemandshiftparametertoindicatethe i (heterogeneous) strength of demand for each horizontally differentiated product. The key advantage of this utility function is that the resulting inverse market demand for product i is linear in (own) quantity: Q γ p =λ α−η − q , i i L L i where q is aggregate consumption of product i and Q is the aggregate consumption of all non- i num`eraireproductsavailableinthemarketplace.16 UsingQH andQW todenoteaggregatesales of differentiated goods to Foreign from Home and the rest of the world, respectively, aggregate consumption in Foreign is then equal to Q=QW +QH. Exporting Firms. We assume a single factor of production – labor – and categorize firms into two sectors: a basic num`eraire sector, 0 and the remaining differentiated goods sector. The num`erairegoodisproducedunderconstantreturnstoscalewithaunitcost,whichimpliesaunit wage to labor. In the differentiated goods sector, all firms have the same constant marginal cost of production, denoted by c, and differ only in the exogenous firm ‘brand equity’ parameter, λ . i Brand equity can be interpreted as the inherent popularity of the product, (exogenous) quality, oranyotherfirm-specificdemandshifter, forinstanceasinDemidovaetal.(2012).17 Hereafter, 14Givenoursetupwithquasi-linearpreferencesandanum`erairegood,itwouldbearelativelysimplematter to close the model, but the extra modeling apparatus required to add a domestic market and impose balanced tradedoesnotyieldenoughadditionalinsighttowarranttheadditionalcomplexity. 15Whilewefocusattentionontheexportsideinourpartialequilibriumapproach,thenum`eraireconceptually allowsforbalancedtrade. 16We assume that every consumers’ income is sufficient to ensure positive consumption of each differentiated product,i. 17Demidovaetal.(2012)offeranempiricalbasisforusingfirmspecificdemandshocks,basedonevidencefrom Bangladeshiapparelexporters. 9 werefertoafirmwithadrawofλasaλ-typefirm. Finally, fortractability, letλbedistributed uniformly over the unit interval, [0,1]. Each firm randomly draws its parameter λ and then makes a decision whether to export or not. Exporters compete in quantities and there is free entry. Home firms can serve Foreign consumersthrougheitherdirectexports(DE),whichrequiresbothasignificantfixedcost,FDE, to set up a direct marketing link or a store front in the export market, and a per unit trade cost ofcDE.18 Oncesetup,thedirectexportersellsunderitsowndistinctlabel,preservingitsbrand equity, λ . Alternatively, a Home exporter can access the Foreign market via the distribution i network of an international firm-retailer (IR), which we discuss in detail shortly. 2.2 Direct Exports To fix ideas, suppose for a moment that firms can export directly or not at all. Given our assumptions, it is immediately clear that only the firms with sufficiently high brand equity will choose to export. This self-selection mechanism parallels earlier work on intermediated trade (and exporting by heterogeneous firms more generally), and is sufficiently straightforward that we relegate the formal derivation to Appendix A1. Theintuitionisasfollows. Underourassumptions,allHomeexportershaveidenticalcostsof productionandmarketaccess(FDE andcDE),butfirm-leveldemandismonotonicallyincreasing in brand-equity, λ, which acts as a vertical demand shifter for each firm. Thus, the firm-level profit from exporting is (strictly) increasing with λ: πDE(cid:48)(λ)>0. Given our assumption of free entry, each firm’s profit from the outside option – not exporting – is zero. Thus, a λ-type firm optimally exports if and only if πDE(λ) ≥ 0. Hereafter we define λDE to be the threshold zero profit exporter under direct exports, given implicitly by: π(λDE)≡0. 2.3 Private-Label Trade Intermediation We now introduce the possibility of private label trade intermediation via an international firm- retailer(IR).ThisIRhasanestablisheddistributionnetworkintheForeigntargetmarketanda ‘privatelabel’technologythatallowsittosellallofitssourcedproductsunderasingleumbrella private label brand, denoted by k. The brand-equity of the private label is determined by the set of exporters that sell through the IR. Specifically, let: (cid:82) λg(λ)dλ λ ≡ Γ , k K 18Weassumethatneitherofthesecostsisprohibitive. 10