CINXE.COM

<meta http-equiv="Content-Type" content="text/html;charset=UTF-8"><base href="https://deliverypdf.ssrn.com/delivery.php?ID=714022009114095066118074092106088126021037057034004075122020004076104025076118127117023061006041103116116084000003115069102126019027055089080083123069084086110072035043045027102065012100013066064091070123087004017113029103067100124022095070105090092&EXT=pdf&INDEX=TRUE"><div style="background:#fff;border:1px solid #999;margin:-1px -1px 0;padding:0;"><div style="background:#ddd;border:1px solid #999;color:#000;font:13px arial,sans-serif;font-weight:normal;margin:12px;padding:8px;text-align:left">This is the html version of the file <a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3018732"><font color=blue>https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3018732</font></a>.<br> Google automatically generates html versions of documents as we crawl the web.</div></div><div style="position:relative"><html> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> <meta name="Author" content="A. Musalem et al."> <meta name="CreationDate" content="D:20170803213914-04'00'"> <meta name="Creator" content="Microsoft® Word 2013"> <meta name="ModDate" content="D:20170804003146-03'00'"> <meta name="Producer" content="Microsoft® Word 2013"> <meta name="Title" content="A Review of Choice Modeling in the Marketing-Operations Management Interface"> <title>A Review of Choice Modeling in the Marketing-Operations Management Interface</title> </head><body bgcolor=#ffffff vlink="blue" link="blue"> <table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=1><b>Page 1</b></a></font></td></tr></table><font size=3 color="#383838" face="Times"><span style="font-size:14px;font-family:Times;color:#383838"> <div style="position:absolute;top:1342;left:235"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:1296;left:455"><nobr>1</nobr></div> </span></font> <font size=5 face="Times"><span style="font-size:34px;font-family:Times"> <div style="position:absolute;top:292;left:192"><nobr>A Review of Choice Modeling in the</nobr></div> <div style="position:absolute;top:333;left:132"><nobr>Marketing-Operations Management Interface</nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:445;left:108"><nobr>Andrés Musalem, University of Chile, <a href="mailto:amusalem@dii.uchile.cl"></a><font color="#0462c1"><a href="mailto:amusalem@dii.uchile.cl">amusalem@dii.uchile.cl</a></font></nobr></div> <div style="position:absolute;top:466;left:108"><nobr>Marcelo Olivares, University of Chile, <a href="mailto:molivares@u.uchile.cl"></a><font color="#0462c1"><a href="mailto:molivares@u.uchile.cl">molivares@u.uchile.cl</a></font></nobr></div> <div style="position:absolute;top:487;left:108"><nobr>Sharad Borle, Rice University, <a href="mailto:sborle@rice.edu"></a><font color="#0462c1"><a href="mailto:sborle@rice.edu">sborle@rice.edu</a></font></nobr></div> <div style="position:absolute;top:507;left:108"><nobr>Hai Che, Indiana University, <a href="mailto:haiche@indiana.edu"></a><font color="#0462c1"><a href="mailto:haiche@indiana.edu">haiche@indiana.edu</a></font></nobr></div> <div style="position:absolute;top:528;left:108"><nobr>Christopher T. Conlon, New York University, <a href="mailto:cconlon@stern.nyu.edu"></a><font color="#0462c1"><a href="mailto:cconlon@stern.nyu.edu">cconlon@stern.nyu.edu</a></font></nobr></div> <div style="position:absolute;top:549;left:108"><nobr>Karan Girotra, INSEAD, <a href="mailto:Karan.GIROTRA@insead.edu"></a><font color="#0462c1"><a href="mailto:Karan.GIROTRA@insead.edu">Karan.GIROTRA@insead.edu</a></font></nobr></div> <div style="position:absolute;top:569;left:108"><nobr>Sachin Gupta, Cornell University, <a href="mailto:sg248@cornell.edu"></a><font color="#0462c1"><a href="mailto:sg248@cornell.edu">sg248@cornell.edu</a></font></nobr></div> <div style="position:absolute;top:590;left:108"><nobr>Kanishka Misra, University of California, San Diego, <a href="mailto:kamisra@ucsd.edu"></a><font color="#0462c1"><a href="mailto:kamisra@ucsd.edu">kamisra@ucsd.edu</a></font></nobr></div> <div style="position:absolute;top:611;left:108"><nobr>Julie Holland Mortimer, Boston College, <a href="mailto:julie.mortimer.2@bc.edu"></a><font color="#0462c1"><a href="mailto:julie.mortimer.2@bc.edu">julie.mortimer.2@bc.edu</a></font></nobr></div> <div style="position:absolute;top:632;left:108"><nobr>Gustavo Vulcano, New York University, <a href="mailto:gvulcano@stern.nyu.edu"></a><font color="#0462c1"><a href="mailto:gvulcano@stern.nyu.edu">gvulcano@stern.nyu.edu</a></font></nobr></div> <div style="position:absolute;top:652;left:108"><nobr>Fanyin Zheng, Columbia University, <a href="mailto:fz2225@gsb.columbia.edu"></a><font color="#0462c1"><a href="mailto:fz2225@gsb.columbia.edu">fz2225@gsb.columbia.edu</a></font></nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:1337;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:1363;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=2><b>Page 2</b></a></font></td></tr></table></div><font size=3 color="#383838" face="Times"><span style="font-size:14px;font-family:Times;color:#383838"> <div style="position:absolute;top:2530;left:235"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:2484;left:455"><nobr>2</nobr></div> </span></font> <font size=5 face="Times"><span style="font-size:34px;font-family:Times"> <div style="position:absolute;top:1480;left:192"><nobr>A Review of Choice Modeling in the</nobr></div> <div style="position:absolute;top:1521;left:132"><nobr>Marketing-Operations Management Interface</nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:1589;left:108"><nobr>The operations management discipline has made substantial progress developing analytical</nobr></div> <div style="position:absolute;top:1609;left:108"><nobr>models to help firms make tactical and strategic decisions (e.g., inventory control, assortment</nobr></div> <div style="position:absolute;top:1630;left:108"><nobr>planning). The application of these models to real contexts requires several inputs, such as an</nobr></div> <div style="position:absolute;top:1651;left:108"><nobr>estimate of the benefits and costs of various levels of the operational decision under study. It is</nobr></div> <div style="position:absolute;top:1671;left:108"><nobr>often the case that the associated benefits depend on how customers react to changes in the</nobr></div> <div style="position:absolute;top:1692;left:108"><nobr>operational decisions being modeled. This in turn requires an empirical validation of the</nobr></div> <div style="position:absolute;top:1713;left:108"><nobr>consumer model that serves as an input when making operation decisions. Bringing together</nobr></div> <div style="position:absolute;top:1734;left:108"><nobr>researchers from economics, marketing and operations management, this paper reviews and</nobr></div> <div style="position:absolute;top:1754;left:108"><nobr>discusses recent findings in terms of modeling and empirically validating consumer choices in</nobr></div> <div style="position:absolute;top:1775;left:108"><nobr>response to operational decisions. We structure this discussion along four major operational</nobr></div> <div style="position:absolute;top:1796;left:108"><nobr>decisions and issues: product variety (including inventory and assortment), service capacity,</nobr></div> <div style="position:absolute;top:1816;left:108"><nobr>pricing and supply chain coordination. For each of these areas, we review recent work and</nobr></div> <div style="position:absolute;top:1837;left:108"><nobr>identify challenges that we believe are important to address in future research aimed at</nobr></div> <div style="position:absolute;top:1858;left:108"><nobr>connecting operations management research to practice.</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:1946;left:108"><nobr><b>1. Introduction</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:1990;left:108"><nobr>Since its origins, the operations management (OM) discipline has had an emphasis on connecting</nobr></div> <div style="position:absolute;top:2032;left:108"><nobr>research to practice. This has led to the development of analytical models to help firms make de-</nobr></div> <div style="position:absolute;top:2073;left:108"><nobr>cisions in a wide range of areas (e.g., inventory, supply chain and capacity management). Earlier</nobr></div> <div style="position:absolute;top:2114;left:108"><nobr>work in operations mangement typically modeled the demand side in these decision models as</nobr></div> <div style="position:absolute;top:2156;left:108"><nobr>exogenous. However, the recent work in OM analyzing, for example, inventory, service levels,</nobr></div> <div style="position:absolute;top:2197;left:108"><nobr>quality and pricing decisions, requires measuring the demand responses to changes in these oper-</nobr></div> <div style="position:absolute;top:2239;left:108"><nobr>ational variables and to incorporate these demand functions into a decision model. This has led to</nobr></div> <div style="position:absolute;top:2280;left:108"><nobr>the development of more sophisticated models of demand that analyze consumers’ choices at a</nobr></div> <div style="position:absolute;top:2321;left:108"><nobr>more granular level.</nobr></div> <div style="position:absolute;top:2363;left:108"><nobr>In contrast to other fields such as marketing and economics, where the main goal is often to un-</nobr></div> <div style="position:absolute;top:2404;left:108"><nobr>derstand in greater detail the underlying mechanisms through which consumers make decisions,</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:2525;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:2551;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=3><b>Page 3</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:3672;left:455"><nobr>3</nobr></div> <div style="position:absolute;top:2663;left:108"><nobr>the focus in OM is on building a demand model that is rich enough to capture the trade-off between</nobr></div> <div style="position:absolute;top:2704;left:108"><nobr>the benefits and costs of improved inventory, service, quality and other operational decisions. De-</nobr></div> <div style="position:absolute;top:2746;left:108"><nobr>spite these differences, there are potential research synergies between these fields and OM. With</nobr></div> <div style="position:absolute;top:2787;left:108"><nobr>the objective of exploring these synergies, we organized a multidisciplinary workshop at the 2016</nobr></div> <div style="position:absolute;top:2829;left:108"><nobr>Triennial Choice Symposium, bringing together researchers from economics, marketing and op-</nobr></div> <div style="position:absolute;top:2870;left:108"><nobr>erations management.<font style="font-size:10px">1 </font>This article originates from the interactions among the authors during this</nobr></div> <div style="position:absolute;top:2911;left:108"><nobr>workshop, and its main objective is to review the progress made so far in terms of empirically</nobr></div> <div style="position:absolute;top:2953;left:108"><nobr>characterizing consumer behavior in response to operational decisions. The focus of the survey is</nobr></div> <div style="position:absolute;top:2994;left:108"><nobr>primarily on published articles, but we also provide some directions for future and emerging trends</nobr></div> <div style="position:absolute;top:3036;left:108"><nobr>in current working papers.</nobr></div> <div style="position:absolute;top:3077;left:108"><nobr>There is a large set of decisions that have been considered in the OM literature. For instance, the</nobr></div> <div style="position:absolute;top:3118;left:108"><nobr>textbook by Cachon and Terwiesch (2013) covers a wide range of issues such as capacity, labor,</nobr></div> <div style="position:absolute;top:3160;left:108"><nobr>product variety, inventory, quality, revenue management and supply chain management. Most of</nobr></div> <div style="position:absolute;top:3201;left:108"><nobr>them directly or indirectly affect the customer experience. For example, capacity decisions that</nobr></div> <div style="position:absolute;top:3243;left:108"><nobr>determine the number of servers in a queuing system directly affect customers’ waiting times,</nobr></div> <div style="position:absolute;top:3284;left:108"><nobr>which may in turn impact customer behavior and choices. Whereas in the operations management</nobr></div> <div style="position:absolute;top:3325;left:108"><nobr>literature there has been extensive work studying analytical models to support these type of deci-</nobr></div> <div style="position:absolute;top:3367;left:108"><nobr>sions, there is much less <i>empirical </i>research that focus on studying the impact of these decisions</nobr></div> <div style="position:absolute;top:3408;left:108"><nobr>on the underlying customer behavior that determines the demand. Hence, this article seeks to par-</nobr></div> <div style="position:absolute;top:3450;left:108"><nobr>tially fill this gap by surveying relevant work in the disciplines of operations, marketing and em-</nobr></div> </span></font> <font size=2 face="Times"><span style="font-size:7px;font-family:Times"> <div style="position:absolute;top:3564;left:108"><nobr>1 <font style="font-size:12px">The 10</font>th <font style="font-size:12px">Triennial Invitational Choice Symposium was held in Lake Louise, Alberta, Canada, May 14-17, 2016.</font></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:12px;font-family:Times"> <div style="position:absolute;top:3583;left:108"><nobr>The symposium is organized in parallel workshops in specific, well-defined themes that are broadly related to hu-</nobr></div> <div style="position:absolute;top:3600;left:108"><nobr>man choice behavior and decision making. Our proposal was on the interface of OM, Marketing and Economics,</nobr></div> <div style="position:absolute;top:3617;left:108"><nobr>which was selected by the committee to form one of the workshops.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:3713;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:3739;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=4><b>Page 4</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:4860;left:455"><nobr>4</nobr></div> <div style="position:absolute;top:3851;left:108"><nobr>pirical industrial organization that has focused on modeling empirically the impact of four opera-</nobr></div> <div style="position:absolute;top:3892;left:108"><nobr>tional variables and issues on demand: (1) product availability and variety; (2) waiting times and</nobr></div> <div style="position:absolute;top:3934;left:108"><nobr>service quality; (3) dynamic pricing / revenue management and (4) supply chain coordination. </nobr></div> <div style="position:absolute;top:3975;left:108"><nobr>The first reason we selected these research areas is because they not only cover a significant por-</nobr></div> <div style="position:absolute;top:4017;left:108"><nobr>tion of the operational decisions studied through analytical modeling in the OM literature, but also</nobr></div> <div style="position:absolute;top:4058;left:108"><nobr>because each of them is likely to affect the customer experience and hence demand. More specif-</nobr></div> <div style="position:absolute;top:4099;left:108"><nobr>ically, the application of assortment planning and inventory management models, which have been</nobr></div> <div style="position:absolute;top:4141;left:108"><nobr>extensively studied in OM, determine the product variety and availability experienced by custom-</nobr></div> <div style="position:absolute;top:4182;left:108"><nobr>ers. Similarly and as previously mentioned, in a queuing system, capacity decisions (e.g., the num-</nobr></div> <div style="position:absolute;top:4224;left:108"><nobr>ber of servers in a queue) have a direct impact on service quality, for example, in terms of the</nobr></div> <div style="position:absolute;top:4265;left:108"><nobr>speed of service provided to customers. In the context of revenue management, dynamic pricing</nobr></div> <div style="position:absolute;top:4306;left:108"><nobr>helps managers to match supply and demand and to implement price discrimination strategies,</nobr></div> <div style="position:absolute;top:4348;left:108"><nobr>which in turn affect the value of offerings purchased by customers. In terms of supply chain coor-</nobr></div> <div style="position:absolute;top:4389;left:108"><nobr>dination, vertical arrangements between manufacturers and distributors may impact the level of</nobr></div> <div style="position:absolute;top:4431;left:108"><nobr>inventory and prices that consumers face at the retail level. The second reason to focus on variety,</nobr></div> <div style="position:absolute;top:4472;left:108"><nobr>waiting times, quality, dynamic pricing and supply chain coordination is that each of these factors</nobr></div> <div style="position:absolute;top:4513;left:108"><nobr>has received attention not only within OM, but also through empirical research in the marketing</nobr></div> <div style="position:absolute;top:4555;left:108"><nobr>and economics literature, hence providing opportunities to identify synergies across areas. </nobr></div> <div style="position:absolute;top:4596;left:108"><nobr>It is important to acknowledge other surveys covering work in the marketing-operations manage-</nobr></div> <div style="position:absolute;top:4638;left:108"><nobr>ment interface (e.g., Chayet et al. 2004; Tang 2010). The focus of this paper is however different,</nobr></div> <div style="position:absolute;top:4679;left:108"><nobr>since we are particularly interested in discussing empirical papers that focus on the demand re-</nobr></div> <div style="position:absolute;top:4720;left:108"><nobr>sponses to factors that are directly related to operational decisions. Understanding these responses</nobr></div> <div style="position:absolute;top:4762;left:108"><nobr>empirically is a critical input for the practical implementation of an operational decision model</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:4901;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:4927;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=5><b>Page 5</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:6048;left:455"><nobr>5</nobr></div> <div style="position:absolute;top:5039;left:108"><nobr>that seeks to manage inventory, assortment, service capacity, pricing and/or supply chains. Readers</nobr></div> <div style="position:absolute;top:5080;left:108"><nobr>interested in other marketing-operations interface issues, such as collaboration and coordination</nobr></div> <div style="position:absolute;top:5122;left:108"><nobr>between marketing and operations, are referred to the comprehensive review by Tang (2010).</nobr></div> <div style="position:absolute;top:5163;left:135"><nobr>The rest of the paper is organized as follows. Section 2 reviews recent work related to product</nobr></div> <div style="position:absolute;top:5205;left:108"><nobr>variety and availability. We describe advances in terms of measuring how consumers react to both</nobr></div> <div style="position:absolute;top:5246;left:108"><nobr>short and long term changes in variety. We also discuss how competitive aspects in terms of both</nobr></div> <div style="position:absolute;top:5284;left:108"><nobr>how firms react to competitors’ variety decisions and how endogenous or strategic changes in</nobr></div> <div style="position:absolute;top:5329;left:108"><nobr>variety should be accounted for when estimating demand. Section 3 discusses research related to</nobr></div> <div style="position:absolute;top:5370;left:108"><nobr>capacity decisions that affect the quality of the service provided to customers. In particular, we</nobr></div> <div style="position:absolute;top:5412;left:108"><nobr>focus primarily on papers describing how consumers behave in waiting environments (e.g., phys-</nobr></div> <div style="position:absolute;top:5453;left:108"><nobr>ical queues and call centers). Section 4 focuses on pricing issues, with a focus on characterizing</nobr></div> <div style="position:absolute;top:5494;left:108"><nobr>demand in response to dynamic pricing and revenue management policies. Section 5 discusses</nobr></div> <div style="position:absolute;top:5536;left:108"><nobr>supply chain coordination research that empirically measures the impact of different supply chain</nobr></div> <div style="position:absolute;top:5577;left:108"><nobr>contractual arrangements. Section 6 provides a discussion related to the selection of choice models</nobr></div> <div style="position:absolute;top:5619;left:108"><nobr>in the context of an operation decision model. Whereas the empirical literature has made progress</nobr></div> <div style="position:absolute;top:5660;left:108"><nobr>in terms of providing more realistic and flexible representations of customer behavior in response</nobr></div> <div style="position:absolute;top:5701;left:108"><nobr>to operational decisions, it is important to contrast this increased flexibility with the tractability</nobr></div> <div style="position:absolute;top:5743;left:108"><nobr>implications for determining optimal operational policies. This flexibility-tractability trade-off is</nobr></div> <div style="position:absolute;top:5784;left:108"><nobr>discussed in this section. Finally, Section 7 concludes the paper by suggesting and identifying</nobr></div> <div style="position:absolute;top:5826;left:108"><nobr>challenges that are important to address in future research aimed at connecting OM research to</nobr></div> <div style="position:absolute;top:5867;left:108"><nobr>practice.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:6089;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:6115;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=6><b>Page 6</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:7236;left:455"><nobr>6</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:6230;left:108"><nobr><b>2. Product variety: Inventory and assortment decisions</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:6274;left:108"><nobr>Product variety is an important dimension of choice that affects customers’ decisions. Firms make</nobr></div> <div style="position:absolute;top:6315;left:108"><nobr>both long term and short term decisions related to how much variety to offer. Long-term decisions</nobr></div> <div style="position:absolute;top:6356;left:108"><nobr>specify the assortment of products to be offered to customers over several periods. However, each</nobr></div> <div style="position:absolute;top:6398;left:108"><nobr>product in a retailer’s assortment may not always be available due to the limited ability of a retailer</nobr></div> <div style="position:absolute;top:6439;left:108"><nobr>to hold enough units, inducing short-run impacts on the variety that customers face.</nobr></div> <div style="position:absolute;top:6481;left:108"><nobr>Inventory and assortment decisions have a long history of active research in the OM literature.</nobr></div> <div style="position:absolute;top:6522;left:108"><nobr>Theoretical models to optimize these decisions need to balance the cost of offering an assortment</nobr></div> <div style="position:absolute;top:6563;left:108"><nobr>or holding inventory against the consequences of not providing enough variety. The latter requires</nobr></div> <div style="position:absolute;top:6605;left:108"><nobr>an estimate of how customers react to variety. In addition, these responses may depend on the</nobr></div> <div style="position:absolute;top:6646;left:108"><nobr>assortment and inventory decisions of competitors. Hickman and Mortimer (2016) survey the de-</nobr></div> <div style="position:absolute;top:6688;left:108"><nobr>mand estimation literature when product availability varies. They primarily focus on changes in</nobr></div> <div style="position:absolute;top:6729;left:108"><nobr>availability that are due to assortment variation and stock-out events. We summarize some of their</nobr></div> <div style="position:absolute;top:6770;left:108"><nobr>key findings in this section. We first discuss empirical research that focuses on estimating customer</nobr></div> <div style="position:absolute;top:6812;left:108"><nobr>responses to inventory changes, particularly due to product stockouts. We then focus on variety</nobr></div> <div style="position:absolute;top:6853;left:108"><nobr>decisions and the challenges for researchers to empirically characterize the demand when firms</nobr></div> <div style="position:absolute;top:6894;left:108"><nobr>strategically select the variety for different markets. Finally we discuss recent research related to</nobr></div> <div style="position:absolute;top:6936;left:108"><nobr>measuring how firms compete with each other in terms of the product variety (assortment and</nobr></div> <div style="position:absolute;top:6977;left:108"><nobr>inventory) offered to consumers.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:7277;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:7303;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=7><b>Page 7</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:8424;left:455"><nobr>7</nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:19px;font-family:Times"> <div style="position:absolute;top:7416;left:108"><nobr><b>2.1 Customer response to stockouts</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:7451;left:108"><nobr>Stockout events are common in retail industries. Gruen, Corsten, and Bharadwaj (2002) report an</nobr></div> <div style="position:absolute;top:7493;left:108"><nobr>average stockout rate of 8.3% in the “fast moving consumer goods” retail category, and that retail-</nobr></div> <div style="position:absolute;top:7534;left:108"><nobr>ers on average lose 4% of annual sales due to items being out of stock. The challenge for retailers</nobr></div> <div style="position:absolute;top:7575;left:108"><nobr>when choosing their inventory policy is that, on the one hand, excess inventory increases holding</nobr></div> <div style="position:absolute;top:7617;left:108"><nobr>and other related costs, while on the other hand, stockout events lead to lost sales.</nobr></div> <div style="position:absolute;top:7658;left:108"><nobr>Stockouts also present an econometric challenge for analyzing demand within a product category. </nobr></div> <div style="position:absolute;top:7700;left:108"><nobr>During periods when a product stocks out, the sales of the product and its available substitutes</nobr></div> <div style="position:absolute;top:7741;left:108"><nobr>may not reflect the true underlying demand because stockouts are not randomly occurring events;</nobr></div> <div style="position:absolute;top:7782;left:108"><nobr>they usually take place during periods of high demand (Kalyanam et al. 2007). For example,</nobr></div> <div style="position:absolute;top:7824;left:108"><nobr>Conlon and Mortimer (2013) find that demand is 250% higher during periods with stockouts than</nobr></div> <div style="position:absolute;top:7865;left:108"><nobr>in periods without stockouts. Although sales are observed, the true demand is unobserved by re-</nobr></div> <div style="position:absolute;top:7907;left:108"><nobr>searchers, and shocks to demand are likely to be correlated with stockout events, leading to a</nobr></div> <div style="position:absolute;top:7948;left:108"><nobr>truncation bias in the estimation of demand which could in turn affect inventory decisions. </nobr></div> <div style="position:absolute;top:7989;left:108"><nobr>Thus, the analysis of sales data which does not account for stockout events may lead to biased</nobr></div> <div style="position:absolute;top:8031;left:108"><nobr>estimates of demand elasticities. Che, Chen, and Chen (2012) attribute the bias in price coefficient</nobr></div> <div style="position:absolute;top:8072;left:108"><nobr>estimates to a high correlation between price promotion activities and stockout events. In other</nobr></div> <div style="position:absolute;top:8114;left:108"><nobr>words, consumers may not be able to purchase a product that is on sale because it is stocked out.</nobr></div> <div style="position:absolute;top:8155;left:108"><nobr>Demand estimates that ignore this effect may conclude that consumers are less price sensitive than</nobr></div> <div style="position:absolute;top:8196;left:108"><nobr>they really are. Likewise, estimates of other demand parameters may also be biased. Musalem et</nobr></div> <div style="position:absolute;top:8238;left:108"><nobr>al. (2010) and Conlon and Mortimer (2013) report that under an assumption that all products are</nobr></div> <div style="position:absolute;top:8279;left:108"><nobr>available, demand estimates for products with frequent stockouts are likely to be understated. Sim-</nobr></div> <div style="position:absolute;top:8321;left:108"><nobr>ilarly, demand estimates for substitute products with infrequent stockouts are likely to be over-</nobr></div> <div style="position:absolute;top:8362;left:108"><nobr>stated.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:8465;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:8491;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=8><b>Page 8</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:9612;left:455"><nobr>8</nobr></div> <div style="position:absolute;top:8603;left:108"><nobr>Broadly speaking, there are three ways in which researchers can handle stockout events in the data.</nobr></div> <div style="position:absolute;top:8644;left:108"><nobr>The first is simply to assume that all products are always available. The second is to ignore data</nobr></div> <div style="position:absolute;top:8686;left:108"><nobr>from stockout periods. Both of these approaches are likely to lead to biased demand estimates. </nobr></div> <div style="position:absolute;top:8727;left:108"><nobr>The third method is to attempt to model the stockout process and adjust demand estimates for</nobr></div> <div style="position:absolute;top:8769;left:108"><nobr>stockout events.</nobr></div> <div style="position:absolute;top:8810;left:108"><nobr>An important challenge when considering stockouts in demand estimation is that the set of prod-</nobr></div> <div style="position:absolute;top:8851;left:108"><nobr>ucts that were actually available for each specific customer is often unobserved. Anupindi, Dada,</nobr></div> <div style="position:absolute;top:8893;left:108"><nobr>and Gupta (1998) were the first to deal with this issue using periodic inventory information. They</nobr></div> <div style="position:absolute;top:8934;left:108"><nobr>assume a Poisson arrival process for each product and develop a missing data approach to infer</nobr></div> <div style="position:absolute;top:8976;left:108"><nobr>item-level demand and substitution patterns (restricted to single-stage substitution). They find sig-</nobr></div> <div style="position:absolute;top:9017;left:108"><nobr>nificant differences between demand rates and the observed sales of products, even for items that</nobr></div> <div style="position:absolute;top:9058;left:108"><nobr>are rarely out of stock. Musalem et al. (2010), Conlon and Mortimer (2013), and Vulcano, van</nobr></div> <div style="position:absolute;top:9100;left:108"><nobr>Ryzin, and Ratliff (2012) extend the missing data approach to allow for flexible substitution pat-</nobr></div> <div style="position:absolute;top:9141;left:108"><nobr>terns with multinomial logit (MNL) and mixed logit demands. </nobr></div> <div style="position:absolute;top:9183;left:108"><nobr>A complementary approach is taken by Che, Chen, and Chen (2012), who exploit daily out-of-</nobr></div> <div style="position:absolute;top:9224;left:108"><nobr>stock data from a single grocery store. With some additional assumptions, they treat changes in</nobr></div> <div style="position:absolute;top:9265;left:108"><nobr>product availability and out-of-stock events as being observed at the individual consumer level.</nobr></div> <div style="position:absolute;top:9307;left:108"><nobr>They are able to analyze whether stockouts have a long term impact on the demand in subsequent</nobr></div> <div style="position:absolute;top:9348;left:108"><nobr>periods. They find that for some product categories there is a positive effect (pent-up demand) and</nobr></div> <div style="position:absolute;top:9390;left:108"><nobr>in other categories there is a negative effect (state dependence and consumer switching).</nobr></div> <div style="position:absolute;top:9431;left:108"><nobr>Kalyanam, Borle, and Boatwright (2007) study how the availability of certain products in the as-</nobr></div> <div style="position:absolute;top:9472;left:108"><nobr>sortment can affect the sales of the entire product category. Their work utilizes stockout-induced</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:9653;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:9679;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=9><b>Page 9</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:10800;left:455"><nobr>9</nobr></div> <div style="position:absolute;top:9791;left:108"><nobr>variation in product assortment and they view stockout events as natural experiments that can re-</nobr></div> <div style="position:absolute;top:9832;left:108"><nobr>sult in lost sales, substitution to other products, and the overall attractiveness of the entire product</nobr></div> <div style="position:absolute;top:9874;left:108"><nobr>category. Likewise, Conlon and Mortimer (2016) show that exogenous stockouts (e.g., generated</nobr></div> <div style="position:absolute;top:9915;left:108"><nobr>from field experiments) can be used to measure the intensity of competition in merger analyses.</nobr></div> <div style="position:absolute;top:9957;left:108"><nobr>Although much of the existing work has relied on observational data, some studies have also used</nobr></div> <div style="position:absolute;top:9998;left:108"><nobr>field experiments to measure the impact of stockouts (e.g., Anderson, Fitzsimons, and Simester</nobr></div> <div style="position:absolute;top:10039;left:108"><nobr>2006) and the effectiveness of inventory planning recommendations that account for stockout-</nobr></div> <div style="position:absolute;top:10081;left:108"><nobr>based substitution (e.g., Lee, Gaur, and Muthulingam 2015).</nobr></div> <div style="position:absolute;top:10122;left:108"><nobr>In terms of promising areas for future research, studying how consumers react to stockouts requires</nobr></div> <div style="position:absolute;top:10164;left:108"><nobr>accurate inventory data and flexible models of consumer behavior. In terms of the former, inven-</nobr></div> <div style="position:absolute;top:10205;left:108"><nobr>tory systems, however, often exhibit inaccuracies or may typically not discriminate between the</nobr></div> <div style="position:absolute;top:10243;left:108"><nobr>inventory on the shelf and in a store’s backroom (DeHoratius and Raman 2008). Hence, the devel-</nobr></div> <div style="position:absolute;top:10288;left:108"><nobr>opment of new methods to correctly identify stockouts should facilitate the advancement of this</nobr></div> <div style="position:absolute;top:10329;left:108"><nobr>research stream. Regarding the models used to characterize consumer choices, most of the extant</nobr></div> <div style="position:absolute;top:10371;left:108"><nobr>literature has relied on variants the multinomial logit model. It would be helpful to consider alter-</nobr></div> <div style="position:absolute;top:10412;left:108"><nobr>native and more flexible models of substitution (e.g., non-parametric choice models) when meas-</nobr></div> <div style="position:absolute;top:10453;left:108"><nobr>uring the impact of stockouts. In addition, the selection of which product and consumer or market</nobr></div> <div style="position:absolute;top:10495;left:108"><nobr>characteristics to use when modeling consumer choices could benefit from the use of machine</nobr></div> <div style="position:absolute;top:10536;left:108"><nobr>learning and optimization techniques for variable selection (e.g., Bertsimas and King 2016). An-</nobr></div> <div style="position:absolute;top:10578;left:108"><nobr>other important issue that deserves attention is the interplay between price and inventory decisions.</nobr></div> <div style="position:absolute;top:10615;left:108"><nobr>In particular, promotional price reductions are likely to more quickly deplete a retailer’s inventory.</nobr></div> <div style="position:absolute;top:10660;left:108"><nobr>It would be important to address whether consumer response to stockouts is affected by the firm’s</nobr></div> <div style="position:absolute;top:10702;left:108"><nobr>promotional efforts. Finally, there are interesting opportunities for expanding our understanding</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:10841;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:10867;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=10><b>Page 10</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:11988;left:450"><nobr>10</nobr></div> <div style="position:absolute;top:10979;left:108"><nobr>of omnichannel settings (Bell et al. 2015; Gallino and Moreno 2014). Since many physical retailers</nobr></div> <div style="position:absolute;top:11020;left:108"><nobr>also have an online presence, another important issue to be addressed in future research is how</nobr></div> <div style="position:absolute;top:11062;left:108"><nobr>online store inventory can mitigate the lost sales from stockouts experienced by customers at brick</nobr></div> <div style="position:absolute;top:11103;left:108"><nobr>and mortar stores. </nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:19px;font-family:Times"> <div style="position:absolute;top:11164;left:108"><nobr><b>2.2 Demand substitution patterns and assortment decisions</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:11199;left:108"><nobr>Choosing the product mix or assortment is another important concern for retailers. In addition to</nobr></div> <div style="position:absolute;top:11240;left:108"><nobr>variation in assortment across retail chains, retailers also customize assortments across stores</nobr></div> <div style="position:absolute;top:11282;left:108"><nobr>within a chain based on market-specific characteristics (Clifford 2010). The variety of an assort-</nobr></div> <div style="position:absolute;top:11323;left:108"><nobr>ment needs to balance potential lift in sales with the increased costs and shelf space constraints of</nobr></div> <div style="position:absolute;top:11364;left:108"><nobr>holding more variety (Kök and Fisher 2007; Kök et al. 2008).</nobr></div> <div style="position:absolute;top:11406;left:108"><nobr>For researchers, variation in product assortment presents a modeling challenge because the deci-</nobr></div> <div style="position:absolute;top:11447;left:108"><nobr>sion of the retailer to carry a product is often positively correlated with consumer preferences for</nobr></div> <div style="position:absolute;top:11489;left:108"><nobr>a particular retailer, location, or season. Bronnenberg, Mahajan, and Vanhonacker (2000) docu-</nobr></div> <div style="position:absolute;top:11526;left:108"><nobr>ment a “positive feedback” relationship showing that retail sales positively impact distribution </nobr></div> <div style="position:absolute;top:11571;left:108"><nobr>(i.e., inclusion of a product in a retailer’s assortment), and distribution positively impacts retail</nobr></div> <div style="position:absolute;top:11609;left:108"><nobr>sales, leading to a “reverse causality” problem that makes the identification of causal effects diffi-</nobr></div> <div style="position:absolute;top:11654;left:108"><nobr>cult.</nobr></div> <div style="position:absolute;top:11696;left:108"><nobr>Tenn and Yun (2008) document that in their data approximately one third of products (or 25% of</nobr></div> <div style="position:absolute;top:11737;left:108"><nobr>sales) have limited availability at the city level. More recently, Quan and Williams (2016) have</nobr></div> <div style="position:absolute;top:11775;left:108"><nobr>documented that Macy’s exhibits a high degree of geographic customization in its shoe assort-</nobr></div> <div style="position:absolute;top:11820;left:108"><nobr>ments, with more than two-thirds of products appearing in less than 10% of locations. In a different</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:12029;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:12055;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=11><b>Page 11</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:13176;left:450"><nobr>11</nobr></div> <div style="position:absolute;top:12167;left:108"><nobr>context, Eizenberg (2014) shows that computer manufacturers such as Dell face substantial oper-</nobr></div> <div style="position:absolute;top:12208;left:108"><nobr>ational costs from maintaining wider product lines, and retire products more quickly than consum-</nobr></div> <div style="position:absolute;top:12250;left:108"><nobr>ers would like.</nobr></div> <div style="position:absolute;top:12291;left:108"><nobr>Complications induced by the fact that assortment decisions are often correlated with local market</nobr></div> <div style="position:absolute;top:12333;left:108"><nobr>preferences are enhanced by the use of sales data aggregated to a zip code, county, or other geo-</nobr></div> <div style="position:absolute;top:12374;left:108"><nobr>graphic area level, as well as the assumption that consumers choose among the products available</nobr></div> <div style="position:absolute;top:12415;left:108"><nobr>anywhere within the region. As Tenn and Yun (2008) point out, this can lead researchers to dras-</nobr></div> <div style="position:absolute;top:12457;left:108"><nobr>tically underestimate the intensity of competition if consumers substitute primarily between prod-</nobr></div> <div style="position:absolute;top:12498;left:108"><nobr>ucts within the same store.</nobr></div> <div style="position:absolute;top:12540;left:108"><nobr>Incorporating information about product assortment at the market level can help to improve de-</nobr></div> <div style="position:absolute;top:12581;left:108"><nobr>mand estimates. We might expect demand to be higher for a product available at 80% of the re-</nobr></div> <div style="position:absolute;top:12622;left:108"><nobr>tailers than for a product available at 20% of the retailers, even if the two products were equally</nobr></div> <div style="position:absolute;top:12664;left:108"><nobr>popular. Bruno and Vilcassim (2008) develop a method to correct demand estimates by scaling</nobr></div> <div style="position:absolute;top:12705;left:108"><nobr>the consumer utilities in a choice model by the distribution intensity of each product in a given</nobr></div> <div style="position:absolute;top:12747;left:108"><nobr>market, which is a proxy for the probability of finding the product available in a store within that</nobr></div> <div style="position:absolute;top:12788;left:108"><nobr>market.</nobr></div> <div style="position:absolute;top:12829;left:108"><nobr>In terms of current and future research, recent work has focused on accounting for the strategic</nobr></div> <div style="position:absolute;top:12871;left:108"><nobr>selection of the product assortment when estimating demand. Intuitively, products with locally</nobr></div> <div style="position:absolute;top:12908;left:108"><nobr>high demand are more likely to be included in local firms’ assortments. Hence, ignoring this se-</nobr></div> <div style="position:absolute;top:12954;left:108"><nobr>lection may lead to biases in demand estimation when inferences need to be made about products</nobr></div> <div style="position:absolute;top:12995;left:108"><nobr>not carried by firms in a specific market. Some recent working papers in marketing and economics</nobr></div> <div style="position:absolute;top:13036;left:108"><nobr>propose methods to account for the endogenous choice of product assortment. These methods rely</nobr></div> <div style="position:absolute;top:13078;left:108"><nobr>on the use of instrumental variables for variety across markets (Iaria 2014) or the formulation of a</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:13217;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:13243;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=12><b>Page 12</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:14364;left:450"><nobr>12</nobr></div> <div style="position:absolute;top:13355;left:108"><nobr>structural model of assortment choice (Musalem 2015; Shah, Kumar, and Zhao 2015). More pro-</nobr></div> <div style="position:absolute;top:13396;left:108"><nobr>gress needs to be made in this area in order to accurately measure how consumers are affected by</nobr></div> <div style="position:absolute;top:13438;left:108"><nobr>long term changes in variety via assortment decisions. </nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:19px;font-family:Times"> <div style="position:absolute;top:13498;left:108"><nobr><b>2.3 Competitive effects and strategic aspects of inventory and assortment</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:13533;left:108"><nobr>In addition to the understanding of how customers react to inventory and variety, it is also im-</nobr></div> <div style="position:absolute;top:13575;left:108"><nobr>portant to consider how firms manage these operational decisions to compete with one another. If</nobr></div> <div style="position:absolute;top:13616;left:108"><nobr>one considered pricing instead of inventory or variety decisions, as the number of competitors</nobr></div> <div style="position:absolute;top:13658;left:108"><nobr>increases, markups charged by firms would typically be predicted to decline. However, the effects</nobr></div> <div style="position:absolute;top:13695;left:108"><nobr>of increased competition on the supply of retailer “effort” on inventory and variety are more am-</nobr></div> <div style="position:absolute;top:13740;left:108"><nobr>biguous For example, one could imagine that the direct effect of increased retail competition is to</nobr></div> <div style="position:absolute;top:13782;left:108"><nobr>reduce the demand at any given retail outlet. Thus, for a fixed inventory or service level, this may</nobr></div> <div style="position:absolute;top:13823;left:108"><nobr>have the appearance of reducing stockouts and improving what customers perceive as the service</nobr></div> <div style="position:absolute;top:13865;left:108"><nobr>level. Under such circumstances, a firm could reduce its effort while maintaining the same per-</nobr></div> <div style="position:absolute;top:13906;left:108"><nobr>ceived service level. Such an accommodating response could even lead a firm to decrease its</nobr></div> <div style="position:absolute;top:13947;left:108"><nobr>perceived service level. Alternatively, a firm could respond to increased competition by raising</nobr></div> <div style="position:absolute;top:13989;left:108"><nobr>its own level of inventory or variety in order to steal business from competitors. In practice, this</nobr></div> <div style="position:absolute;top:14030;left:108"><nobr>effort can take on a number of forms. Firms can: (1) hold additional inventory, (2) expand the</nobr></div> <div style="position:absolute;top:14072;left:108"><nobr>variety of product offerings, (3) increase the frequency of restocking, or (4) invest in technologies</nobr></div> <div style="position:absolute;top:14113;left:108"><nobr>that improve the operations of the firm.</nobr></div> <div style="position:absolute;top:14154;left:108"><nobr>The question of whether competition leads firms to increase or decrease effort in service provision</nobr></div> <div style="position:absolute;top:14196;left:108"><nobr>is theoretically ambiguous and necessitates empirical examination. The main challenges are that</nobr></div> <div style="position:absolute;top:14237;left:108"><nobr>a) the number of competitors is not random and is potentially correlated with service provision,</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:14405;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:14431;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=13><b>Page 13</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:15552;left:450"><nobr>13</nobr></div> <div style="position:absolute;top:14543;left:108"><nobr>and b) unobserved factors may affect both incentives for service provision and the entry of com-</nobr></div> <div style="position:absolute;top:14584;left:108"><nobr>petitors. In many empirical settings, the consequences of retailer effort (e.g., the rate of stockouts)</nobr></div> <div style="position:absolute;top:14626;left:108"><nobr>may be observable, even when the corresponding inputs, such as inventory holdings, are unob-</nobr></div> <div style="position:absolute;top:14667;left:108"><nobr>served.</nobr></div> <div style="position:absolute;top:14709;left:108"><nobr>Along this line of work, Olivares and Cachon (2009) study the relationship between inventory and</nobr></div> <div style="position:absolute;top:14750;left:108"><nobr>competition by examining a cross-section of about 200 U.S. automobile dealers in geographically</nobr></div> <div style="position:absolute;top:14791;left:108"><nobr>isolated markets. They find competition appears to increase the inventory holdings of dealers.</nobr></div> <div style="position:absolute;top:14833;left:108"><nobr>Matsa (2011) studies the impact of competition on product stockouts at retail grocers. He estimates</nobr></div> <div style="position:absolute;top:14874;left:108"><nobr>that stores facing local competition have 5% fewer stockouts than comparable stores. This work</nobr></div> <div style="position:absolute;top:14912;left:108"><nobr>addresses the endogeneity problem caused by the retailers’ decisions on product availability by</nobr></div> <div style="position:absolute;top:14957;left:108"><nobr>examining retailers before and after the entry of Wal-Mart into the local market. He reports that</nobr></div> <div style="position:absolute;top:14995;left:108"><nobr>product stockouts decreased by 10% after Walmart’s entry. Work by Ailawadi et al. (2010) and</nobr></div> <div style="position:absolute;top:15040;left:108"><nobr>Huang et al. (2012) suggests that the mechanism behind this effect is that incumbent retailers mod-</nobr></div> <div style="position:absolute;top:15081;left:108"><nobr>ify their assortment after entry from Walmart. </nobr></div> <div style="position:absolute;top:15123;left:108"><nobr>As previously noted, assortment decisions are a key strategic variable that can be adjusted by firms</nobr></div> <div style="position:absolute;top:15164;left:108"><nobr>when facing changes in competition. Watson (2009) studies the impact of competition on the prod-</nobr></div> <div style="position:absolute;top:15205;left:108"><nobr>uct variety offered by firms in the eyeglass industry using a structural model of spatial entry. The</nobr></div> <div style="position:absolute;top:15247;left:108"><nobr>author finds that as the number of rival firms in the market increases, variety at the individual</nobr></div> <div style="position:absolute;top:15288;left:108"><nobr>retailer level increases initially before decreasing. </nobr></div> <div style="position:absolute;top:15326;left:108"><nobr>How firms respond to competition depends on whether the retailers’ required efforts to increase</nobr></div> <div style="position:absolute;top:15371;left:108"><nobr>product availability are strategic substitutes or complements and the vertical relationship between</nobr></div> <div style="position:absolute;top:15412;left:108"><nobr>manufacturers and retailers. Considering both inventory and assortment issues, Conlon and</nobr></div> <div style="position:absolute;top:15454;left:108"><nobr>Mortimer (2017) examine a setting in which a vending machine operator chooses how often to</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:15593;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:15619;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=14><b>Page 14</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:16740;left:450"><nobr>14</nobr></div> <div style="position:absolute;top:15731;left:108"><nobr>restock products under different vertical arrangements with candy manufacturers. They find that a</nobr></div> <div style="position:absolute;top:15772;left:108"><nobr>rebate contracts employed by a dominant manufacturer can be used to incentivize higher levels of</nobr></div> <div style="position:absolute;top:15814;left:108"><nobr>service for the retailer, leading to fewer stockouts. However, they also find that more frequent re-</nobr></div> <div style="position:absolute;top:15855;left:108"><nobr>stocking harms rival manufacturers. Furthermore, in terms of product variety, they also find that</nobr></div> <div style="position:absolute;top:15897;left:108"><nobr>these contracts can be used to restrict the variety and number of rival products that the retailer</nobr></div> <div style="position:absolute;top:15938;left:108"><nobr>carries. </nobr></div> <div style="position:absolute;top:15979;left:108"><nobr>Regarding future work, retailers have expanded their presence by selling their goods in both phys-</nobr></div> <div style="position:absolute;top:16021;left:108"><nobr>ical an online stores. Accordingly, omni-channel is an emerging stream of research that could ben-</nobr></div> <div style="position:absolute;top:16062;left:108"><nobr>efit from a deeper understanding of how firms can compete with each other in terms of their in-</nobr></div> <div style="position:absolute;top:16104;left:108"><nobr>ventory and assortment decisions when each firm may have a presence not only in multiple loca-</nobr></div> <div style="position:absolute;top:16145;left:108"><nobr>tions, but also through multiple formats (e.g., online and brick and mortar).</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:16257;left:108"><nobr><b>3. Capacity and service level decisions in services industries</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:16301;left:108"><nobr>An important area of research in OM deals with the allocation of resources (e.g., capacity) to im-</nobr></div> <div style="position:absolute;top:16343;left:108"><nobr>prove the service level offered to customers. Within this context, customer waiting times, an im-</nobr></div> <div style="position:absolute;top:16384;left:108"><nobr>portant measure of service quality, have been extensively studied through analytical models built</nobr></div> <div style="position:absolute;top:16425;left:108"><nobr>on queuing theory. More recent work in this area has incorporated empirical models that capture</nobr></div> <div style="position:absolute;top:16467;left:108"><nobr>consumers’ reactions to waiting time and delays, including herding behavior (Debo et al. (2010),</nobr></div> <div style="position:absolute;top:16508;left:108"><nobr>Veeraraghavan and Debo (2009)) and customer valuations that depend on the service time (Anand</nobr></div> <div style="position:absolute;top:16550;left:108"><nobr>et al. (2011)) . Bridging these analytical models with real world applications requires empirical</nobr></div> <div style="position:absolute;top:16591;left:108"><nobr>validation of how consumers behave in service systems where waiting times are relevant. This</nobr></div> <div style="position:absolute;top:16632;left:108"><nobr>section describes empirical work studying how consumers react to different service levels, with a</nobr></div> <div style="position:absolute;top:16674;left:108"><nobr>special emphasis on waiting environments. </nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:16781;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:16807;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=15><b>Page 15</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:17928;left:450"><nobr>15</nobr></div> <div style="position:absolute;top:16919;left:108"><nobr>In OM, call centers have been extensively studied because customer waiting time is an important</nobr></div> <div style="position:absolute;top:16960;left:108"><nobr>measure of service that can be measured objectively (see Gans, Koole, and Mandelbaum 2003;</nobr></div> <div style="position:absolute;top:17002;left:108"><nobr>and Aksin, Armony, and Mehrotra 2007 for comprehensive surveys of research in call centers).</nobr></div> <div style="position:absolute;top:17043;left:108"><nobr>An important primitive in queuing models is the abandonment behavior of consumers, which can</nobr></div> <div style="position:absolute;top:17085;left:108"><nobr>be perceived as a proxy for consumer dissatisfaction or lost revenue. Mandelbaum and Zeltyn</nobr></div> <div style="position:absolute;top:17126;left:108"><nobr>(2004) and Brown et al. (2005) incorporate abandonments in queuing systems by modeling cus-</nobr></div> <div style="position:absolute;top:17167;left:108"><nobr>tomers with a patience threshold, abandoning when waiting time exceeds their thresholds. Dura-</nobr></div> <div style="position:absolute;top:17209;left:108"><nobr>tion models are then used to estimate the distribution of this patience threshold across consumers.</nobr></div> <div style="position:absolute;top:17250;left:108"><nobr>A different approach is taken by Akşin et al. (2013), Akşin et al. (2016) and Yu, Allon, and</nobr></div> <div style="position:absolute;top:17292;left:108"><nobr>Bassamboo (2016), who use a structural estimation approach to incorporate customers’ expecta-</nobr></div> <div style="position:absolute;top:17333;left:108"><nobr>tions of waiting time in the decision of abandoning a queue. Using a framework similar to Rust</nobr></div> <div style="position:absolute;top:17374;left:108"><nobr>(1987), these papers model consumer choice as a dynamic programming problem that captures the</nobr></div> <div style="position:absolute;top:17416;left:108"><nobr>trade-off between the value of the service against the current and future expected delay costs. </nobr></div> <div style="position:absolute;top:17457;left:108"><nobr>Customer abandonments are also relevant in services where queues are visible to consumers. Lu</nobr></div> <div style="position:absolute;top:17499;left:108"><nobr>et al. (2013) develop a choice model to estimate the probability of joining a queue as a function of</nobr></div> <div style="position:absolute;top:17540;left:108"><nobr>the number of customers in the queue. Using data from the deli section of a supermarket, they find</nobr></div> <div style="position:absolute;top:17581;left:108"><nobr>that (i) customers tend to react to the number of people in queue but are insensitive to the speed at</nobr></div> <div style="position:absolute;top:17623;left:108"><nobr>which the queue advances; (ii) the impact of queue length is heterogeneous across customers,</nobr></div> <div style="position:absolute;top:17664;left:108"><nobr>where price sensitive customers tend to be less sensitive to service delays. In a different setting,</nobr></div> <div style="position:absolute;top:17706;left:108"><nobr>Batt and Terwiesch (2015) study the abandonment of patients in the waiting room of an emergency</nobr></div> <div style="position:absolute;top:17747;left:108"><nobr>department using patient level data. They show that patient abandonments are sensitive to the</nobr></div> <div style="position:absolute;top:17785;left:108"><nobr>number of patients in the waiting room. Moreover, the flow of patients during a focal patient’s</nobr></div> <div style="position:absolute;top:17830;left:108"><nobr>waiting period also affects their decision to abandon, and this effect depends on the severity level</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:17969;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:17995;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=16><b>Page 16</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:19116;left:450"><nobr>16</nobr></div> <div style="position:absolute;top:18107;left:108"><nobr>of the patients that move through the system. Campbell and Frei (2010) analyze the impact of</nobr></div> <div style="position:absolute;top:18148;left:108"><nobr>waiting time on customer satisfaction and retention in the retail banking industry. Their results</nobr></div> <div style="position:absolute;top:18190;left:108"><nobr>suggest that on average customers are sensitive to waiting time and that there is substantial heter-</nobr></div> <div style="position:absolute;top:18231;left:108"><nobr>ogeneity that can be explained by income and local competition. </nobr></div> <div style="position:absolute;top:18273;left:108"><nobr>All the above studies use consumer-level data to analyze the impact of waiting time on demand.</nobr></div> <div style="position:absolute;top:18314;left:108"><nobr>Allon, Federgruen and Pierson (2011) use instead aggregated market-level data to study the impact</nobr></div> <div style="position:absolute;top:18355;left:108"><nobr>of waiting time on demand in the fast food industry. They develop a structural model similar to</nobr></div> <div style="position:absolute;top:18397;left:108"><nobr>Thomadsen (2007) that is based on a Bertrand Nash competition between differentiated products.</nobr></div> <div style="position:absolute;top:18438;left:108"><nobr>They find that the cost of waiting is relatively high in this industry, with each second of additional</nobr></div> <div style="position:absolute;top:18480;left:108"><nobr>wait being equivalent to a $0.05 increase in price, more than 1% of the cost of a meal. Png and</nobr></div> <div style="position:absolute;top:18521;left:108"><nobr>Reitman (1994) studied the impact of waiting time on demand in retail gasoline markets using</nobr></div> <div style="position:absolute;top:18562;left:108"><nobr>aggregate data on monthly sales by station, average prices and service capacity. They find that</nobr></div> <div style="position:absolute;top:18600;left:108"><nobr>consumers’ willingness-to-pay increases by 1% for a 6% reduction in congestion. Overall, all of</nobr></div> <div style="position:absolute;top:18645;left:108"><nobr>these studies highlight the importance of modeling how customers respond to waiting time; such</nobr></div> <div style="position:absolute;top:18687;left:108"><nobr>empirical models can be used within a decision model to support capacity and staffing decisions</nobr></div> <div style="position:absolute;top:18728;left:108"><nobr>in a wide variety of service industries. </nobr></div> <div style="position:absolute;top:18769;left:108"><nobr>The literature in OM has also looked at service quality measures beyond waiting time. Guajardo,</nobr></div> <div style="position:absolute;top:18811;left:108"><nobr>Cohen, and Netessine (2015) study the effect of product defects and warranty on demand for new</nobr></div> <div style="position:absolute;top:18852;left:108"><nobr>vehicles in the U.S. automobile market. Using aggregate market level data and adapting Berry,</nobr></div> <div style="position:absolute;top:18894;left:108"><nobr>Levinsohn, and Pakes (1995), they incorporate endogenous service quality metrics into consumer</nobr></div> <div style="position:absolute;top:18935;left:108"><nobr>utility. They find that the demand impact of a 1% decrease in price is equivalent in magnitude to</nobr></div> <div style="position:absolute;top:18976;left:108"><nobr>increasing product quality by 2.2%, and is in turn equivalent to increasing the warranty length by</nobr></div> <div style="position:absolute;top:19018;left:108"><nobr>8%. </nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:19157;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:19183;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=17><b>Page 17</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:20304;left:450"><nobr>17</nobr></div> <div style="position:absolute;top:19295;left:108"><nobr>Finally, there are numerous studies in marketing focused on service quality and customer satisfac-</nobr></div> <div style="position:absolute;top:19336;left:108"><nobr>tion using attitudinal data such as service quality perceptions or customer satisfaction ratings (e.g.,</nobr></div> <div style="position:absolute;top:19378;left:108"><nobr>Mittal and Kamakura 2001). Perceptions are valuable metrics, since consumers make decisions</nobr></div> <div style="position:absolute;top:19419;left:108"><nobr>based on how they perceive the quality of a product or service. However, they are less useful to</nobr></div> <div style="position:absolute;top:19461;left:108"><nobr>support operational decisions when perceptions cannot be directly linked with the objective service</nobr></div> <div style="position:absolute;top:19502;left:108"><nobr>level metrics (e.g. waiting time) that are commonly used in queuing and other decision models.</nobr></div> <div style="position:absolute;top:19543;left:108"><nobr>An important goal for future research is to fill this gap by linking consumer perceptions of service</nobr></div> <div style="position:absolute;top:19585;left:108"><nobr>quality with objective measures of service that are affected by operational decisions. Another op-</nobr></div> <div style="position:absolute;top:19626;left:108"><nobr>portunity identified in our survey is to consider consumer choice heuristics in response to waiting</nobr></div> <div style="position:absolute;top:19668;left:108"><nobr>time and other service metrics. An example is the working paper by Yu et al. (2016), extending</nobr></div> <div style="position:absolute;top:19709;left:108"><nobr>the structural model developed in Akşin et al. (2013) by incorporating loss aversion, where cus-</nobr></div> <div style="position:absolute;top:19750;left:108"><nobr>tomers’ reaction to waiting times depends on reference points formed through previous service</nobr></div> <div style="position:absolute;top:19792;left:108"><nobr>experience. Another alternative is to model in more detail the process through which consumers</nobr></div> <div style="position:absolute;top:19833;left:108"><nobr>form expectations about service times, for example, by incorporating a learning process in settings</nobr></div> <div style="position:absolute;top:19875;left:108"><nobr>with repeated service interactions (Emadi and Swaminathan 2017; Craig et al. 2016; Musalem et</nobr></div> <div style="position:absolute;top:19916;left:108"><nobr>al. 2016). </nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:20028;left:108"><nobr><b>4. Dynamic pricing decisions and revenue management</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:20072;left:108"><nobr>In OM, dynamic pricing can be useful to control the utilization of capacity and the consumption</nobr></div> <div style="position:absolute;top:20114;left:108"><nobr>of inventory when these cannot be adjusted in the short-term. The more recent literature in this</nobr></div> <div style="position:absolute;top:20155;left:108"><nobr>area has identified the importance of capturing forward-looking behavior of customers (i.e., stra-</nobr></div> <div style="position:absolute;top:20193;left:108"><nobr>tegic customers), where demand responses to price changes depend on customers’ expectations</nobr></div> <div style="position:absolute;top:20238;left:108"><nobr>about future prices (e.g., Cachon and Swinney 2009; Mersereau and Zhang 2012; Jerath and</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:20345;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:20371;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=18><b>Page 18</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:21492;left:450"><nobr>18</nobr></div> <div style="position:absolute;top:20483;left:108"><nobr>Netessine 2010). The discussion herein is focused on work that has made progress in terms of</nobr></div> <div style="position:absolute;top:20524;left:108"><nobr>incorporating forward-looking behavior in empirical models of demand. </nobr></div> <div style="position:absolute;top:20566;left:108"><nobr>In the airline industry, Li, Granados, and Netessine (2014) study strategic customers by analyzing</nobr></div> <div style="position:absolute;top:20607;left:108"><nobr>detailed bookings data. They consider routes where prices follow different price trajectories and</nobr></div> <div style="position:absolute;top:20649;left:108"><nobr>therefore create varying expectations about future prices, which helps to identify the fraction of</nobr></div> <div style="position:absolute;top:20690;left:108"><nobr>consumers that are forward-looking. There are also several studies that consider customer forward-</nobr></div> <div style="position:absolute;top:20731;left:108"><nobr>looking behavior in the retail sector. Hendel and Nevo (2006) analyze customer stock-piling be-</nobr></div> <div style="position:absolute;top:20773;left:108"><nobr>havior, where promotions induce customers to increase current purchases to build-up household</nobr></div> <div style="position:absolute;top:20814;left:108"><nobr>inventories for future consumption, thereby reducing sales in the following periods at the regular</nobr></div> <div style="position:absolute;top:20856;left:108"><nobr>price. This structural model identifies the underlying mechanism of the “post-promotion dip” ef-</nobr></div> <div style="position:absolute;top:20897;left:108"><nobr>fect that has been reported in the marketing literature (Van Heerde and Neslin 2008). In the context</nobr></div> <div style="position:absolute;top:20938;left:108"><nobr>of durable goods like consumer electronics, customers not only have expectations about future</nobr></div> <div style="position:absolute;top:20980;left:108"><nobr>prices but also about the introduction of new substitutable products. To account for this effect,</nobr></div> <div style="position:absolute;top:21021;left:108"><nobr>Gordon (2009) develops a structural model that incorporates heterogeneity in customers’ replace-</nobr></div> <div style="position:absolute;top:21063;left:108"><nobr>ment behavior, which can be useful to design new product introduction policies that improve prof-</nobr></div> <div style="position:absolute;top:21104;left:108"><nobr>its by segmenting the market.</nobr></div> <div style="position:absolute;top:21145;left:108"><nobr>A different type of intertemporal price discrimination is relevant in the context of products with a</nobr></div> <div style="position:absolute;top:21187;left:108"><nobr>short lifecycle, such as video-games and fashion apparel. Markdown pricing is commonly used in</nobr></div> <div style="position:absolute;top:21224;left:108"><nobr>these industries to “skim” customers with a higher valuation to purchase early in the season at</nobr></div> <div style="position:absolute;top:21270;left:108"><nobr>higher prices. This price segmentation strategy become less effective when customers can antici-</nobr></div> <div style="position:absolute;top:21311;left:108"><nobr>pate future price reductions. Nair (2007) studies this problem in the video-game industry, devel-</nobr></div> <div style="position:absolute;top:21352;left:108"><nobr>oping a structural model where customers solve an optimal-stopping problem of when to buy the</nobr></div> <div style="position:absolute;top:21394;left:108"><nobr>product during its lifecycle. In fashion apparel, Soysal and Krishnamurthi (2012) study a similar</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:21533;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:21559;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=19><b>Page 19</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:22680;left:450"><nobr>19</nobr></div> <div style="position:absolute;top:21671;left:108"><nobr>problem, but where products may become unavailable towards the end of the season due to stock-</nobr></div> <div style="position:absolute;top:21712;left:108"><nobr>outs. In this case, customers face a trade-off between buying early at higher prices and waiting for</nobr></div> <div style="position:absolute;top:21754;left:108"><nobr>prices to go down, but facing the risk of a stockout. They find that scarcity strategies that increase</nobr></div> <div style="position:absolute;top:21795;left:108"><nobr>the risk of stockouts can be effective to induce forward-looking customers to buy at higher prices.</nobr></div> <div style="position:absolute;top:21837;left:108"><nobr>A related problem is considered by Moon, Bimpikis, and Mendelson (2017), where customers</nobr></div> <div style="position:absolute;top:21878;left:108"><nobr>incur a cost to monitor prices at an online retailer when searching for price discounts. They find</nobr></div> <div style="position:absolute;top:21919;left:108"><nobr>substantial heterogeneity in customers’ monitoring costs, which provides opportunities to segment</nobr></div> <div style="position:absolute;top:21961;left:108"><nobr>the market by introducing randomized markdowns. </nobr></div> <div style="position:absolute;top:22002;left:108"><nobr>In terms of future research, most of the models that address strategic consumer behavior and the</nobr></div> <div style="position:absolute;top:22044;left:108"><nobr>associated pricing strategies assume that customers are fully rational agents. However, it is well</nobr></div> <div style="position:absolute;top:22085;left:108"><nobr>acknowledged that customers exhibit behavioral biases when making decisions (e.g., Chen et al.</nobr></div> <div style="position:absolute;top:22126;left:108"><nobr>2012). Enriching pricing models with these behavioral biases following the prolific line of research</nobr></div> <div style="position:absolute;top:22168;left:108"><nobr>on behavioral economics (e.g., Ariely and Wallsten 1995) could lead to different pricing strategies</nobr></div> <div style="position:absolute;top:22209;left:108"><nobr>and business insights.</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:22280;left:108"><nobr><b>5. Supply chain coordination</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:22324;left:108"><nobr>The coordination of supply chains has received substantial attention in the OM literature, particu-</nobr></div> <div style="position:absolute;top:22365;left:108"><nobr>larly through the development of analytical models. Although supply chain arrangements dictate</nobr></div> <div style="position:absolute;top:22407;left:108"><nobr>the terms under which manufacturers and distributors interact with each other, they also have an</nobr></div> <div style="position:absolute;top:22448;left:108"><nobr>impact on the variety, service levels and prices faced by consumers. In this section we discuss</nobr></div> <div style="position:absolute;top:22490;left:108"><nobr>empirical research about channel coordination issues, particularly in terms of price setting ar-</nobr></div> <div style="position:absolute;top:22531;left:108"><nobr>rangements between manufacturers and retailers, product line and exclusivity contracts, product</nobr></div> <div style="position:absolute;top:22572;left:108"><nobr>quality and performance incentives. </nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:22721;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:22747;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=20><b>Page 20</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:23868;left:450"><nobr>20</nobr></div> <div style="position:absolute;top:22859;left:108"><nobr>Vertical arrangements in supply chains considering the role of upstream and downstream pricing</nobr></div> <div style="position:absolute;top:22900;left:108"><nobr>decisions have been empirically studied by Sudhir (2001) and Villas-Boas (2007). Both of these</nobr></div> <div style="position:absolute;top:22942;left:108"><nobr>papers study the US grocery industry, and model the vertical interaction between retailers and</nobr></div> <div style="position:absolute;top:22983;left:108"><nobr>manufacturers as a static pricing game. They consider different supply-side models of the indus-</nobr></div> <div style="position:absolute;top:23025;left:108"><nobr>try to infer the nature of the pricing game played between manufacturers and retailers (Nash ver-</nobr></div> <div style="position:absolute;top:23066;left:108"><nobr>sus manufacturer or retailer price leadership). In their empirical analyses, both authors estimate</nobr></div> <div style="position:absolute;top:23107;left:108"><nobr>an aggregate discrete-choice demand model, similar to Berry et al. (1995). Cost-based instru-</nobr></div> <div style="position:absolute;top:23149;left:108"><nobr>ments are used to address concerns about price endogeneity, but estimation of the demand model</nobr></div> <div style="position:absolute;top:23190;left:108"><nobr>does not rely on information from supply-side relationships for identification. Using their de-</nobr></div> <div style="position:absolute;top:23232;left:108"><nobr>mand-side estimates, the authors infer the supply-side relationships between agents. For exam-</nobr></div> <div style="position:absolute;top:23273;left:108"><nobr>ple, different supply-side models imply different marginal costs and retail pricing patterns, given</nobr></div> <div style="position:absolute;top:23314;left:108"><nobr>the estimated demand elasticities. This allows the authors to evaluate how well different supply-</nobr></div> <div style="position:absolute;top:23356;left:108"><nobr>side models explain variation in observable outcomes. </nobr></div> <div style="position:absolute;top:23397;left:108"><nobr>One can also consider the nature of the pricing contract between upstream and downstream firms</nobr></div> <div style="position:absolute;top:23439;left:108"><nobr>(e.g., the use of linear pricing, two-part tariffs, or revenue sharing). Mortimer (2008) studies the</nobr></div> <div style="position:absolute;top:23480;left:108"><nobr>impact of revenue-sharing between movie distributors (i.e., studios) and video rental stores. Prior</nobr></div> <div style="position:absolute;top:23521;left:108"><nobr>to the availability of monitoring technology for rental transactions, a retailer paid a fixed fee to a</nobr></div> <div style="position:absolute;top:23563;left:108"><nobr>distributor for each unit of rental inventory, and kept all rental revenues. This is referred to as a</nobr></div> <div style="position:absolute;top:23604;left:108"><nobr>linear pricing contract. After distributors acquired the ability to monitor rental transactions, video</nobr></div> <div style="position:absolute;top:23646;left:108"><nobr>rental stores could choose between the linear pricing contract and a revenue-sharing contract, in</nobr></div> <div style="position:absolute;top:23687;left:108"><nobr>which the rental store purchased inventory at a low upfront cost and split the rental revenue with</nobr></div> <div style="position:absolute;top:23728;left:108"><nobr>the distributor. Mortimer analyzes the welfare effects of the availability of revenue sharing, ac-</nobr></div> <div style="position:absolute;top:23766;left:108"><nobr>counting for the fact that a retailer’s choice of contract will be determined endogenously. She</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:23909;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:23935;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=21><b>Page 21</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:25056;left:450"><nobr>21</nobr></div> <div style="position:absolute;top:24047;left:108"><nobr>finds the revenue-sharing improves welfare overall, with the greatest gains going to smaller re-</nobr></div> <div style="position:absolute;top:24088;left:108"><nobr>tailers.</nobr></div> <div style="position:absolute;top:24130;left:108"><nobr>Vertical arrangements can take a wide variety of forms beyond different pricing games, includ-</nobr></div> <div style="position:absolute;top:24171;left:108"><nobr>ing quantity requirements, full-line forcing, exclusive dealing, and vendor allowances. Many of</nobr></div> <div style="position:absolute;top:24213;left:108"><nobr>these arrangements require downstream firms to meet conditions to qualify for payments from</nobr></div> <div style="position:absolute;top:24254;left:108"><nobr>producers, and anti-trust agencies are especially interested in these conditional pricing practices.</nobr></div> <div style="position:absolute;top:24295;left:108"><nobr>Genchev and Mortimer (2016) review the empirical work in this area. In many cases, papers</nobr></div> <div style="position:absolute;top:24337;left:108"><nobr>studying these contractual arrangements consider additional data that permit researchers to iden-</nobr></div> <div style="position:absolute;top:24378;left:108"><nobr>tify their effects. </nobr></div> <div style="position:absolute;top:24420;left:108"><nobr>Asker (2016) studies exclusive dealing in the market for beer in Chicago. Anheuser-Busch and</nobr></div> <div style="position:absolute;top:24461;left:108"><nobr>other upstream firms employed exclusive dealing arrangements with beer distributors, in which a</nobr></div> <div style="position:absolute;top:24502;left:108"><nobr>distributor agrees not to distribute products of rival producers. There are two potential effects of</nobr></div> <div style="position:absolute;top:24544;left:108"><nobr>these types of contracts. On one hand, they may encourage efficient investment and thus produce</nobr></div> <div style="position:absolute;top:24585;left:108"><nobr>efficiency gains. On the other hand, they may be used to foreclose a rival producer from the mar-</nobr></div> <div style="position:absolute;top:24627;left:108"><nobr>ket. Using data on the exclusive dealing arrangements as well as distribution networks used in</nobr></div> <div style="position:absolute;top:24668;left:108"><nobr>the Chicago market, he estimates retailer costs and develops a test for foreclosure. He finds in-</nobr></div> <div style="position:absolute;top:24709;left:108"><nobr>significant evidence that foreclosure occurs in this market, and concludes that intervention to ban</nobr></div> <div style="position:absolute;top:24751;left:108"><nobr>these contractual forms would likely reduce social welfare.</nobr></div> <div style="position:absolute;top:24792;left:108"><nobr>A few years after the introduction of revenue sharing in the video rental market, movie distribu-</nobr></div> <div style="position:absolute;top:24834;left:108"><nobr>tors introduced a third contract for acquiring rental inventory. The contract bundled a distribu-</nobr></div> <div style="position:absolute;top:24871;left:108"><nobr>tor’s titles, which partly addressed distributor’s adverse selection concerns stemming from retail-</nobr></div> <div style="position:absolute;top:24913;left:108"><nobr>ers’ endogenous selection of contract. Ho et al. (2012) consider the impact of this bundling con-</nobr></div> <div style="position:absolute;top:24958;left:108"><nobr>tract, referred to as full-line forcing (FLF). Under a FLF contract, a retailer agrees to accept all</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:25097;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:25123;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=22><b>Page 22</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:26244;left:450"><nobr>22</nobr></div> <div style="position:absolute;top:25235;left:108"><nobr>titles from a distributor in exchange for paying a lower upfront cost per unit of inventory and</nobr></div> <div style="position:absolute;top:25276;left:108"><nobr>keeping a more generous share of the rental revenue. The authors report that 80% of all retail</nobr></div> <div style="position:absolute;top:25318;left:108"><nobr>stores have at least one FLF with a distributor. After estimating a nested-logit demand model, the</nobr></div> <div style="position:absolute;top:25359;left:108"><nobr>authors estimate supply-side parameters using moment inequalities. The inequalities are based on</nobr></div> <div style="position:absolute;top:25401;left:108"><nobr>the assumption that, on average, profits from the set of movies selected by a retail store are</nobr></div> <div style="position:absolute;top:25442;left:108"><nobr>higher than profits from any other potential set of movies chosen for that store. Given these esti-</nobr></div> <div style="position:absolute;top:25483;left:108"><nobr>mates, the authors analyze what profit would have been if retailers were not offered FLF. Build-</nobr></div> <div style="position:absolute;top:25525;left:108"><nobr>ing on this empirical structure, (Ho et al. 2012) measure the effects of FLF on market coverage,</nobr></div> <div style="position:absolute;top:25566;left:108"><nobr>leverage, and efficiency. They report that FLF contracts increased market coverage and effi-</nobr></div> <div style="position:absolute;top:25608;left:108"><nobr>ciency in this setting, but had little impact on leverage across distributors. As a result, they esti-</nobr></div> <div style="position:absolute;top:25649;left:108"><nobr>mate that FLF contracts increased consumer surplus in this market.</nobr></div> <div style="position:absolute;top:25690;left:108"><nobr>Hristakeva (2016) studies the impact of financial incentives provided by manufacturers to retail-</nobr></div> <div style="position:absolute;top:25732;left:108"><nobr>ers in order to gain distribution for their products. She considers incentives that are made in the</nobr></div> <div style="position:absolute;top:25773;left:108"><nobr>form of lump-sum payments known as vendor allowance (e.g., slotting fees), rather than as a</nobr></div> <div style="position:absolute;top:25815;left:108"><nobr>function of the amount sold by the retailer. One of the research challenges in this area is that ven-</nobr></div> <div style="position:absolute;top:25856;left:108"><nobr>dor allowances are not observed. To overcome this data limitation Hristakeva (2016) uses varia-</nobr></div> <div style="position:absolute;top:25897;left:108"><nobr>tion in retail assortment to construct a lower bound of the payments made to retailers. After esti-</nobr></div> <div style="position:absolute;top:25939;left:108"><nobr>mating demand for yogurt in the grocery market, she considers the counterfactual profits from all</nobr></div> <div style="position:absolute;top:25980;left:108"><nobr>possible one-product deviations from the observed assortment. That is, she considers the counter-</nobr></div> <div style="position:absolute;top:26022;left:108"><nobr>factual expected profit if the retailer replaced an existing product with a product not offered in</nobr></div> <div style="position:absolute;top:26063;left:108"><nobr>the observed retailer assortment. Using these estimates, she suggests that in this market, slotting</nobr></div> <div style="position:absolute;top:26101;left:108"><nobr>fees are at least 5% of the retailer’s revenue, and that removing them can increase consumer sur-</nobr></div> <div style="position:absolute;top:26146;left:108"><nobr>plus by 2.5%.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:26285;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:26311;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=23><b>Page 23</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:27432;left:450"><nobr>23</nobr></div> <div style="position:absolute;top:26423;left:108"><nobr>Beyond assortments and prices, vertical contacts have also been used to manage product quality.</nobr></div> <div style="position:absolute;top:26464;left:108"><nobr>Guajardo et al. (2012) study the aircraft engine industry. In this market manufacturers offer two</nobr></div> <div style="position:absolute;top:26506;left:108"><nobr>types of contract of after-sales support: a) time and material contracts, where the supplier is paid</nobr></div> <div style="position:absolute;top:26547;left:108"><nobr>based on the resources consumed when maintenance is required; and b) performance based con-</nobr></div> <div style="position:absolute;top:26589;left:108"><nobr>tracts (PBC), where manufacturers are paid based on value delivered to the consumer (e.g., an</nobr></div> <div style="position:absolute;top:26630;left:108"><nobr>airline consumer paying as a function of the number of hours the engine was in use). The authors</nobr></div> <div style="position:absolute;top:26671;left:108"><nobr>use a unique dataset where they observe the contract choice and performance for each consumer.</nobr></div> <div style="position:absolute;top:26713;left:108"><nobr>They estimate the impact of contract choice, using a control function to account for endogeneity,</nobr></div> <div style="position:absolute;top:26754;left:108"><nobr>on performance. They estimate that PBC result in 25%-40% better performance. </nobr></div> <div style="position:absolute;top:26796;left:108"><nobr>Overall these empirical studies on vertical contracts show the importance of modeling the endog-</nobr></div> <div style="position:absolute;top:26837;left:108"><nobr>enous choice of observed contracts. This can be done using exogenous variation in the institu-</nobr></div> <div style="position:absolute;top:26878;left:108"><nobr>tional setting (Asker 2016, Mortimer 2008), moment inequalities (Ho et al. 2012a; Ho et al.</nobr></div> <div style="position:absolute;top:26920;left:108"><nobr>2012b) and Hristakeva 2016), or using instruments or control functions (Guajardo et al. 2012). </nobr></div> <div style="position:absolute;top:26961;left:108"><nobr>In terms of future research, we see two main areas to contribute to this literature. The first is re-</nobr></div> <div style="position:absolute;top:27003;left:108"><nobr>lated to the bargaining process between vertical channel members. Current research mostly con-</nobr></div> <div style="position:absolute;top:27044;left:108"><nobr>siders the game between vertical channel members as Nash or one of the members as the leader.</nobr></div> <div style="position:absolute;top:27085;left:108"><nobr>This assumption does not consider the role of bargaining, say with repeated offers, which likely</nobr></div> <div style="position:absolute;top:27127;left:108"><nobr>exists in vertical relationships. New dataset and models with channel member bargaining is a</nobr></div> <div style="position:absolute;top:27168;left:108"><nobr>promising area of research. The second area for future research should focus on the expansion of</nobr></div> <div style="position:absolute;top:27210;left:108"><nobr>the set of interactions between vertical channel members. Current research has considered static</nobr></div> <div style="position:absolute;top:27251;left:108"><nobr>models of pricing, assortment and quality in vertical channels. Future research could extend this</nobr></div> <div style="position:absolute;top:27292;left:108"><nobr>literature by considering dynamic interactions between members. Another potential area is to</nobr></div> <div style="position:absolute;top:27334;left:108"><nobr>consider other interactions between channel partners. This could include role of information and</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:27473;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:27499;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=24><b>Page 24</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:28620;left:450"><nobr>24</nobr></div> <div style="position:absolute;top:27611;left:108"><nobr>data sharing. Some vertical channel relationships (e.g. category captain arrangements in retail</nobr></div> <div style="position:absolute;top:27652;left:108"><nobr>settings), are based on sharing category information and consumer insights between retailers and</nobr></div> <div style="position:absolute;top:27694;left:108"><nobr>manufacturers.</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:27765;left:108"><nobr><b>6. Balancing flexibility and tractability</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:27809;left:108"><nobr>When formulating a decision model that requires characterizing demand, there is typically a trade-</nobr></div> <div style="position:absolute;top:27850;left:108"><nobr>off between the level of detail at which the demand model is specified and the tractability of the</nobr></div> <div style="position:absolute;top:27892;left:108"><nobr>resulting decision model. Although there have been important advances in the estimation of econ-</nobr></div> <div style="position:absolute;top:27933;left:108"><nobr>ometric models that provide flexibility in the demand specification through choice models, the</nobr></div> <div style="position:absolute;top:27974;left:108"><nobr>application of some of these models has been limited because their use may yield decision models</nobr></div> <div style="position:absolute;top:28016;left:108"><nobr>that are difficult to solve using analytical/numerical methods. This section discusses some of these</nobr></div> <div style="position:absolute;top:28057;left:108"><nobr>flexibility-tractability challenges focusing (for brevity) on specific examples related to assortment</nobr></div> <div style="position:absolute;top:28099;left:108"><nobr>decisions.</nobr></div> <div style="position:absolute;top:28140;left:108"><nobr>Assortment optimization focuses on choosing a subset of products to offer in order to maximize</nobr></div> <div style="position:absolute;top:28181;left:108"><nobr>an objective, such as revenue, contribution margin or margin minus inventory costs. When demand</nobr></div> <div style="position:absolute;top:28223;left:108"><nobr>is modeled using a multinomial logit model (MNL), Talluri and van Ryzin (2004) show that the</nobr></div> <div style="position:absolute;top:28264;left:108"><nobr>optimization problem is computationally tractable since the MNL satisfies the general conditions</nobr></div> <div style="position:absolute;top:28305;left:108"><nobr>of revenue-ordered assortments; i.e., given a set of n products, there are n possible complete reve-</nobr></div> <div style="position:absolute;top:28347;left:108"><nobr>nue-ordered sets, and one of them is the assortment that maximizes expected revenues. Mahajan</nobr></div> <div style="position:absolute;top:28388;left:108"><nobr>and van Ryzin (2001) show that the optimal assortment also has a simple structure when the ob-</nobr></div> <div style="position:absolute;top:28430;left:108"><nobr>jective function accounts for inventory costs captured through a newsvendor model. </nobr></div> <div style="position:absolute;top:28471;left:108"><nobr>The MNL has a number of limitations in terms of its ability to characterize consumer choices, the</nobr></div> <div style="position:absolute;top:28512;left:108"><nobr>most salient being the independence from irrelevant alternatives (IIA) that imposes severe re-</nobr></div> <div style="position:absolute;top:28554;left:108"><nobr>strictions on the substitution patterns. Going beyond the MNL, however, increases the complexity</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:28661;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:28687;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=25><b>Page 25</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:29808;left:450"><nobr>25</nobr></div> <div style="position:absolute;top:28799;left:108"><nobr>of the assortment problem substantially. When demand is specified through a Nested logit, Davis</nobr></div> <div style="position:absolute;top:28840;left:108"><nobr>et al. (2013) showed that the problem is polynomially solvable when the nest dissimilarity param-</nobr></div> <div style="position:absolute;top:28882;left:108"><nobr>eters of the choice model (i.e., the nested logit parameters that measure the dissimilarity among</nobr></div> <div style="position:absolute;top:28923;left:108"><nobr>alternatives within a nest) are less than 1 and the customers always make a purchase within the</nobr></div> <div style="position:absolute;top:28965;left:108"><nobr>selected nest. Gallego and Topaloglu (2014) studied the same problem under cardinality and space</nobr></div> <div style="position:absolute;top:29006;left:108"><nobr>constraints on the offered assortment. They show that the optimal assortment under cardinality</nobr></div> <div style="position:absolute;top:29047;left:108"><nobr>constraints can be obtained efficiently by solving a linear program, but the assortment optimization</nobr></div> <div style="position:absolute;top:29089;left:108"><nobr>problem under space constraints remains NP-hard. Li and Rusmevichientong (2015) study a spe-</nobr></div> <div style="position:absolute;top:29130;left:108"><nobr>cific case of the Nested MNL where nests are organized as trees with specified depth, and develop</nobr></div> <div style="position:absolute;top:29172;left:108"><nobr>a polynomial algorithm to find the optimal assortment.</nobr></div> <div style="position:absolute;top:29213;left:108"><nobr>Another natural extension of the basic MNL is the Mixed MNL, where the market is described by</nobr></div> <div style="position:absolute;top:29254;left:108"><nobr>a heterogeneous mix of consumers, each of them described by an MNL. Rusmevichientong and</nobr></div> <div style="position:absolute;top:29296;left:108"><nobr>Shmoys (2014) show that the assortment optimization problem under this demand model is NP-</nobr></div> <div style="position:absolute;top:29337;left:108"><nobr>hard but identify special cases where revenue-ordered sets are optimal. For the general case, nu-</nobr></div> <div style="position:absolute;top:29379;left:108"><nobr>merical experiments suggest that heuristics based on revenue-ordered sets can perform well. </nobr></div> <div style="position:absolute;top:29420;left:108"><nobr>Feldman and Topaloglu (2015) provide methods to compute upper bounds on the optimal revenue</nobr></div> <div style="position:absolute;top:29461;left:108"><nobr>for the assortment problem with mixed MNL demand.</nobr></div> <div style="position:absolute;top:29503;left:108"><nobr>A more recent line of work has focused on non-parametric models of consumer choice. Rather</nobr></div> <div style="position:absolute;top:29544;left:108"><nobr>than characterizing choices through the utility of each alternative, the class of rank-based models</nobr></div> <div style="position:absolute;top:29586;left:108"><nobr>assumes each consumer ranks alternatives based on her preferences. Consumers are segmented</nobr></div> <div style="position:absolute;top:29627;left:108"><nobr>into a finite number of types defined by their corresponding rank orders, which determines de-</nobr></div> <div style="position:absolute;top:29668;left:108"><nobr>mand. These nonparametric models have recently been used in the revenue management and retail</nobr></div> <div style="position:absolute;top:29710;left:108"><nobr>operations literature, such as Mahajan and van Ryzin (2001a) who use this class of models in</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:29849;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:29875;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=26><b>Page 26</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:30996;left:450"><nobr>26</nobr></div> <div style="position:absolute;top:29987;left:108"><nobr>assortment decisions. Farias et al. (2013) develop a robust approach to estimate the distribution</nobr></div> <div style="position:absolute;top:30028;left:108"><nobr>over customer types that produces the worst-case revenue compatible with the observed data for a</nobr></div> <div style="position:absolute;top:30070;left:108"><nobr>given fixed assortment. Alternatively, van Ryzin and Vulcano (2015) propose a column generation</nobr></div> <div style="position:absolute;top:30111;left:108"><nobr>procedure to obtain maximum likelihood estimates for the proportions of the customer types.</nobr></div> <div style="position:absolute;top:30153;left:108"><nobr>The models and associated assortment optimizations described above are based on preferences</nobr></div> <div style="position:absolute;top:30194;left:108"><nobr>inferred at the aggregate market level. Going one step further in the granularity of the preferences,</nobr></div> <div style="position:absolute;top:30235;left:108"><nobr>the latter nonparametric models of demand can also be applied to individual consumers.</nobr></div> <div style="position:absolute;top:30277;left:108"><nobr>Jagabathula and Vulcano (2017) model individual customer preferences through directed acyclic</nobr></div> <div style="position:absolute;top:30318;left:108"><nobr>graphs that represent partial orders of the revealed preferences between some of the offered prod-</nobr></div> <div style="position:absolute;top:30360;left:108"><nobr>ucts, and report promising accuracy results for the prediction of next purchases.</nobr></div> <div style="position:absolute;top:30401;left:108"><nobr>Regarding future directions, an important departure from the choice-based demand paradigm is</nobr></div> <div style="position:absolute;top:30442;left:108"><nobr>the flexibility of the consideration set definition. The standard practice so far has been to assume</nobr></div> <div style="position:absolute;top:30484;left:108"><nobr>that in each store visit, customers evaluate all products in the assortment; i.e., their consideration</nobr></div> <div style="position:absolute;top:30525;left:108"><nobr>set was essentially the offer set exhibited by the seller. But of course, customers are boundedly</nobr></div> <div style="position:absolute;top:30567;left:108"><nobr>rational and pay limited attention to assortments. Ignoring these behavioral biases may introduce</nobr></div> <div style="position:absolute;top:30608;left:108"><nobr>significant errors when modeling choices, as reported in the recent econometrics related literature</nobr></div> <div style="position:absolute;top:30649;left:108"><nobr>(e.g., see Manzini and Mariotti 2014). Bringing this to the operational level through tractable as-</nobr></div> <div style="position:absolute;top:30691;left:108"><nobr>sortment and price optimization models is another promising venue to explore.</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:30762;left:108"><nobr><b>7. Concluding remarks</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:30806;left:108"><nobr>In contrast with other disciplines that study consumer choices, the OM field has focused on using</nobr></div> <div style="position:absolute;top:30847;left:108"><nobr>choice models that are sufficiently rich to capture the effect of important operational variables</nobr></div> <div style="position:absolute;top:30888;left:108"><nobr>(e.g., inventory and waiting times) on demand, but at the same time can keep the underlying deci-</nobr></div> <div style="position:absolute;top:30930;left:108"><nobr>sion model tractable for optimization purposes. Simple and parsimonious demand models (e.g.,</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:31037;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:31063;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=27><b>Page 27</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:32184;left:450"><nobr>27</nobr></div> <div style="position:absolute;top:31175;left:108"><nobr>MNL), while potentially less accurate in terms of representing the underlying choice behavior,</nobr></div> <div style="position:absolute;top:31216;left:108"><nobr>lead to optimization problems that are easier to solve and may yield reasonable solutions in prac-</nobr></div> <div style="position:absolute;top:31258;left:108"><nobr>tice. A more complex demand model with many attributes, nests, or latent classes may provide a</nobr></div> <div style="position:absolute;top:31299;left:108"><nobr>better approximation to a wide range of choice behavior and reduce the specification error, but the</nobr></div> <div style="position:absolute;top:31341;left:108"><nobr>associated optimization of operational decisions becomes less tractable, except possibly for par-</nobr></div> <div style="position:absolute;top:31382;left:108"><nobr>ticular cases. </nobr></div> <div style="position:absolute;top:31423;left:108"><nobr>Progress has been made in the estimation of demand systems that account for the effect of product</nobr></div> <div style="position:absolute;top:31465;left:108"><nobr>availability, service quality and other operational variables. Endogeneity is an important challenge:</nobr></div> <div style="position:absolute;top:31506;left:108"><nobr>inventories and assortments are chosen by managers, and ignoring this endogeneity could lead to</nobr></div> <div style="position:absolute;top:31548;left:108"><nobr>biases in the estimation. This review discusses several approaches to account for endogeneity,</nobr></div> <div style="position:absolute;top:31589;left:108"><nobr>including structural models, instrumental variables and field experiments. Nevertheless, the OM</nobr></div> <div style="position:absolute;top:31630;left:108"><nobr>field, which is focused on prescribing how these decisions should be made in practice, is in a</nobr></div> <div style="position:absolute;top:31672;left:108"><nobr>privileged position to help our understanding on how to handle these endogeneity problems.</nobr></div> <div style="position:absolute;top:31713;left:108"><nobr>Hence, we foresee two lines of research where OM can interact with other fields in the context of</nobr></div> <div style="position:absolute;top:31755;left:108"><nobr>choice modeling: (1) studying how more sophisticated demand models can be analyzed in the</nobr></div> <div style="position:absolute;top:31796;left:108"><nobr>context of a decision model; and (2) using knowledge about how operational decisions are made</nobr></div> <div style="position:absolute;top:31837;left:108"><nobr>in practice to inform empirical researchers about the effect of operational variables on demand,</nobr></div> <div style="position:absolute;top:31879;left:108"><nobr>hence, addressing endogeneity issues. We hope that this article helps to foster more interdiscipli-</nobr></div> <div style="position:absolute;top:31920;left:108"><nobr>nary research along this line between OM, marketing, economics and other related fields.</nobr></div> </span></font> <font size=4 face="Times"><span style="font-size:25px;font-family:Times"> <div style="position:absolute;top:31991;left:108"><nobr><b>References</b></nobr></div> </span></font> <font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:32035;left:108"><nobr>Ailawadi KL, Zhang J, Krishna A, Kruger MW (2010) When Wal-Mart Enters: How Incumbent</nobr></div> <div style="position:absolute;top:32056;left:144"><nobr>Retailers React and How This Affects Their Sales Outcomes. <i>J. Mark. Res. </i>47(4):577–593.</nobr></div> <div style="position:absolute;top:32073;left:108"><nobr>Akşin Z, Armony M, Mehrotra V (2007) The modern call center: A multi‐disciplinary</nobr></div> <div style="position:absolute;top:32097;left:144"><nobr>perspective on operations management research. <i>Prod. Oper. </i>16(6):665–688.</nobr></div> <div style="position:absolute;top:32114;left:108"><nobr>Akşin Z, Ata B, Emadi S, Su C (2016) Impact of delay announcements in call centers: An</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:32225;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:32251;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=28><b>Page 28</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:33372;left:450"><nobr>28</nobr></div> <div style="position:absolute;top:32363;left:144"><nobr>empirical approach. <i>Oper. Res. </i>65(1):242–265.</nobr></div> <div style="position:absolute;top:32380;left:108"><nobr>Akşin Z, Ata B, Emadi SM, Su CL (2013) Structural estimation of callers delay sensitivity in call</nobr></div> <div style="position:absolute;top:32404;left:144"><nobr>centers. <i>Manage. Sci. </i>59(12):2727–2746.</nobr></div> <div style="position:absolute;top:32421;left:108"><nobr>Allon G, Federgruen A, Pierson M (2011) How much is a reduction of your customers’ wait</nobr></div> <div style="position:absolute;top:32446;left:144"><nobr>worth? an empirical study of the fast-food drive-thru industry based on structural estimation</nobr></div> <div style="position:absolute;top:32466;left:144"><nobr>methods. <i>Manuf. Serv. Oper. Manag. </i>13(4):489–507.</nobr></div> <div style="position:absolute;top:32487;left:108"><nobr>Anand K, Paç M, Veeraraghavan S (2011) Quality–speed conundrum: Trade-offs in customer-</nobr></div> <div style="position:absolute;top:32508;left:144"><nobr>intensive services. <i>Manage. Sci. </i>57(1):40–56.</nobr></div> <div style="position:absolute;top:32529;left:108"><nobr>Anderson E, Fitzsimons G, Simester D (2006) Measuring and mitigating the cost of stockouts.</nobr></div> <div style="position:absolute;top:32549;left:144"><nobr><i>Manage. Sci. </i>52(11):1751–1763.</nobr></div> <div style="position:absolute;top:32570;left:108"><nobr>Anupindi R, Dada M, Gupta S (1998) Estimation of consumer demand with stock-out based</nobr></div> <div style="position:absolute;top:32591;left:144"><nobr>substitution: An application to vending machine products. <i>Mark. Sci. </i>17(4):406–423.</nobr></div> <div style="position:absolute;top:32611;left:108"><nobr>Ariely D, Wallsten TS (1995) Seeking Subjective Dominance in Multidimensional Space: An</nobr></div> <div style="position:absolute;top:32632;left:144"><nobr>Explanation of the Asymmetric Dominance Effect. <i>Organ. Behav. Hum. Decis. Process.</i></nobr></div> <div style="position:absolute;top:32653;left:144"><nobr>63(3):223–232.</nobr></div> <div style="position:absolute;top:32673;left:108"><nobr>Asker J (2016) Diagnosing Foreclosure due to Exclusive Dealing. <i>J. Ind. Econ. </i>64(3):375–410.</nobr></div> <div style="position:absolute;top:32694;left:108"><nobr>Batt R, Terwiesch C (2015) Waiting patiently: An empirical study of queue abandonment in an</nobr></div> <div style="position:absolute;top:32715;left:144"><nobr>emergency department. <i>Manage. Sci. </i>61(1):39–59.</nobr></div> <div style="position:absolute;top:32736;left:108"><nobr>Bell D, Gallino S, Moreno A (2015) Showrooms and Information Provision in Omni-channel</nobr></div> <div style="position:absolute;top:32756;left:144"><nobr>Retail. <i>Prod. Oper. Manag. </i>24(3):360–362.</nobr></div> <div style="position:absolute;top:32777;left:108"><nobr>Berry S, Levinsohn J, Pakes A (1995) Automobile prices in market equilibrium. <i>Econom. J.</i></nobr></div> <div style="position:absolute;top:32798;left:144"><nobr><i>Econom. Soc.</i>:841–890.</nobr></div> <div style="position:absolute;top:32818;left:108"><nobr>Bertsimas D, King A (2016) OR Forum—An Algorithmic Approach to Linear Regression. <i>Oper.</i></nobr></div> <div style="position:absolute;top:32839;left:144"><nobr><i>Res. </i>64(1):2–16.</nobr></div> <div style="position:absolute;top:32860;left:108"><nobr>Bronnenberg BJ, Mahajan V, Vanhonacker WR (2000) The Emergence of Market Structure in</nobr></div> <div style="position:absolute;top:32881;left:144"><nobr>New Repeat-Purchase Categories: The Interplay of Market Share and Retailer Distribution.</nobr></div> <div style="position:absolute;top:32901;left:144"><nobr><i>J. Mark. Res. </i>37(1):16–31.</nobr></div> <div style="position:absolute;top:32922;left:108"><nobr>Brown L, Gans N, Mandelbaum A, Sakov A, Shen H, Zeltyn S, Zhao L (2005) Statistical</nobr></div> <div style="position:absolute;top:32943;left:144"><nobr>analysis of a telephone call center: A queueing-science perspective. <i>J. Am. Stat. Assoc.</i></nobr></div> <div style="position:absolute;top:32963;left:144"><nobr>100(469):36–50.</nobr></div> <div style="position:absolute;top:32984;left:108"><nobr>Bruno H, Vilcassim N (2008) Structural demand estimation with varying product availability.</nobr></div> <div style="position:absolute;top:33005;left:144"><nobr><i>Mark. Sci. </i>27(6):1126–1131.</nobr></div> <div style="position:absolute;top:33025;left:108"><nobr>Cachon G, Terwiesch C (2013) <i>Matching supply with demand : an introduction to operations</i></nobr></div> <div style="position:absolute;top:33046;left:144"><nobr><i>management </i>3rd ed. (McGraw-Hill, New York, NY).</nobr></div> <div style="position:absolute;top:33067;left:108"><nobr>Cachon GP, Swinney R (2009) Purchasing, pricing, and quick response in the presence of</nobr></div> <div style="position:absolute;top:33088;left:144"><nobr>strategic consumers. <i>Manage. Sci. </i>55(3):497–511.</nobr></div> <div style="position:absolute;top:33108;left:108"><nobr>Campbell D, Frei F (2010) Market Heterogeneity and Local Capacity Decisions in Services.</nobr></div> <div style="position:absolute;top:33129;left:144"><nobr><i>Manuf. Serv. Oper. Manag. </i>13(1):2–19.</nobr></div> <div style="position:absolute;top:33150;left:108"><nobr>Chayet S, Hopp WJ, Xu X (2004) The Marketing-Operations Interface. Simchi-Levi, D.; Wu, S.</nobr></div> <div style="position:absolute;top:33170;left:144"><nobr>D. & Shen ZJ, ed. <i>Handb. Quant. Supply Chain Anal. Model. E-bus. Era</i>. (Springer US,</nobr></div> <div style="position:absolute;top:33191;left:144"><nobr>Boston, MA), 295–334.</nobr></div> <div style="position:absolute;top:33212;left:108"><nobr>Che H, Chen X (Jack), Chen Y (2012) Investigating Effects of Out-of-Stock on Consumer</nobr></div> <div style="position:absolute;top:33232;left:144"><nobr>Stockkeeping Unit Choice. <i>J. Mark. Res. </i>49(4):502–513.</nobr></div> <div style="position:absolute;top:33253;left:108"><nobr>Chen Y, Su X, Zhao X (2012) Modeling Bounded Rationality in Capacity Allocation Games</nobr></div> <div style="position:absolute;top:33274;left:144"><nobr>with the Quantal Response Equilibrium. <i>Manage. Sci. </i>58(10):1952–1962.</nobr></div> <div style="position:absolute;top:33291;left:108"><nobr>Clifford S (2010) Atlanta Hats? Seattle Socks? Macy’s goes local. <i>The New York Times</i></nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:33413;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:33439;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=29><b>Page 29</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:34560;left:450"><nobr>29</nobr></div> <div style="position:absolute;top:33551;left:144"><nobr>(October).</nobr></div> <div style="position:absolute;top:33572;left:108"><nobr>Conlon CT, Mortimer JH (2013) Demand estimation under incomplete product availability. <i>Am.</i></nobr></div> <div style="position:absolute;top:33592;left:144"><nobr><i>Econ. J. Microeconomics </i>5(4):1–30.</nobr></div> <div style="position:absolute;top:33613;left:108"><nobr>Conlon CT, Mortimer JH (2016) An Experimental Approach to Merger Evaluation, working</nobr></div> <div style="position:absolute;top:33634;left:144"><nobr>paper.</nobr></div> <div style="position:absolute;top:33654;left:108"><nobr>Conlon CT, Mortimer JH (2017) Efficiency and Foreclosure Effects of Vertical Rebates:</nobr></div> <div style="position:absolute;top:33675;left:144"><nobr>Empirical Evidence, working paper.</nobr></div> <div style="position:absolute;top:33696;left:108"><nobr>Craig N, DeHoratius N, Raman A (2016) The Impact of Supplier Inventory Service Level on</nobr></div> <div style="position:absolute;top:33717;left:144"><nobr>Retailer Demand. <i>Manuf. Serv. Oper. Manag. </i>18(4):461–474.</nobr></div> <div style="position:absolute;top:33737;left:108"><nobr>Davis J, Gallego G, Topaloglu H (2013) Assortment optimization under variants of the Nested</nobr></div> <div style="position:absolute;top:33758;left:144"><nobr>Logit Model. <i>Oper. Res. </i>62(2):250–273.</nobr></div> <div style="position:absolute;top:33779;left:108"><nobr>Debo L, Parlour C, Rajan U (2012) Signaling Quality via Queues. <i>Manage. Sci. </i>58(5):876–891.</nobr></div> <div style="position:absolute;top:33799;left:108"><nobr>DeHoratius N, Raman A (2008) Inventory record inaccuracy: an empirical analysis. <i>Manage.</i></nobr></div> <div style="position:absolute;top:33820;left:144"><nobr><i>Sci. </i>54(4):627–641.</nobr></div> <div style="position:absolute;top:33841;left:108"><nobr>Eizenberg A (2014) Upstream Innovation and Product Variety in the U.S. Home PC Market.</nobr></div> <div style="position:absolute;top:33861;left:144"><nobr><i>Rev. Econ. Stud. </i>81(3):1003–1045.</nobr></div> <div style="position:absolute;top:33878;left:108"><nobr>Emadi SM, Swaminathan JM (2017) Impact of Callers’ History on Abandonment: Model and</nobr></div> <div style="position:absolute;top:33903;left:144"><nobr>Implications, working paper.</nobr></div> <div style="position:absolute;top:33924;left:108"><nobr>Farias V, Jagabathula S, Shah D (2013) A non-parametric approach to modeling choice with</nobr></div> <div style="position:absolute;top:33944;left:144"><nobr>limited data. <i>Manage. Sci. </i>59(2):305–322.</nobr></div> <div style="position:absolute;top:33965;left:108"><nobr>Feldman J, Topaloglu H (2015) Bounding optimal expected revenues for assortment</nobr></div> <div style="position:absolute;top:33986;left:144"><nobr>optimization under Mixtures of Multinomial Logits. <i>Prod. Oper. Manag. </i>24(10):1598–</nobr></div> <div style="position:absolute;top:34006;left:144"><nobr>1620.</nobr></div> <div style="position:absolute;top:34027;left:108"><nobr>Gallego G, Topaloglu H (2014) Constrained assortment optimization for the Nested Logit</nobr></div> <div style="position:absolute;top:34048;left:144"><nobr>Model. <i>Manage. Sci. </i>60(10):2583–2601.</nobr></div> <div style="position:absolute;top:34069;left:108"><nobr>Gallino S, Moreno A (2014) Integration of online and offline channels in retail: The impact of</nobr></div> <div style="position:absolute;top:34089;left:144"><nobr>sharing reliable inventory availability information. <i>Manage. Sci.</i></nobr></div> <div style="position:absolute;top:34110;left:108"><nobr>Gans N, Koole G, Mandelbaum A (2003) Telephone call centers: Tutorial, review, and research</nobr></div> <div style="position:absolute;top:34131;left:144"><nobr>prospects. <i>Manuf. Serv. Oper. Manag. </i>5(2):79–141.</nobr></div> <div style="position:absolute;top:34151;left:108"><nobr>Genchev B, Mortimer JH (2016) Empirical Evidence on Conditional Pricing Practices *. <i>Anti-</i></nobr></div> <div style="position:absolute;top:34172;left:144"><nobr><i>trust Law J. </i>forthcoming.</nobr></div> <div style="position:absolute;top:34193;left:108"><nobr>Gordon BR (2009) A dynamic model of consumer replacement cycles in the PC processor</nobr></div> <div style="position:absolute;top:34213;left:144"><nobr>industry. <i>Mark. Sci. </i>28(5):846–867.</nobr></div> <div style="position:absolute;top:34234;left:108"><nobr>Gruen T, Corsten D, Bharadwaj S (2002) Retail out-of-stocks: A worldwide examination of</nobr></div> <div style="position:absolute;top:34255;left:144"><nobr>causes, rates, and consumer responses.</nobr></div> <div style="position:absolute;top:34276;left:108"><nobr>Guajardo JA, Cohen MA, Kim SH, Netessine S (2012) Impact of Performance-Based</nobr></div> <div style="position:absolute;top:34296;left:144"><nobr>Contracting on Product Reliability: An Empirical Analysis. <i>Manage. Sci. </i>58(5):961–979.</nobr></div> <div style="position:absolute;top:34317;left:108"><nobr>Guajardo JA, Cohen MA, Netessine S (2015) Service competition and product quality in the US</nobr></div> <div style="position:absolute;top:34338;left:144"><nobr>automobile industry. <i>Manage. Sci. </i>62(7):1860–1877.</nobr></div> <div style="position:absolute;top:34358;left:108"><nobr>Van Heerde HJ, Neslin SA (2008) Sales promotion models. <i>Handb. Mark. Decis. Model.</i></nobr></div> <div style="position:absolute;top:34379;left:144"><nobr>(Springer), 107–162.</nobr></div> <div style="position:absolute;top:34400;left:108"><nobr>Hendel I, Nevo A (2006) Measuring the implications of sales and consumer inventory behavior.</nobr></div> <div style="position:absolute;top:34420;left:144"><nobr><i>Econometrica </i>74(6):1637–1673.</nobr></div> <div style="position:absolute;top:34441;left:108"><nobr>Hickman W, Mortimer JH (2016) Demand estimation with availability variation. Basker E, ed.</nobr></div> <div style="position:absolute;top:34462;left:144"><nobr><i>Handb. Econ. Retail. Distrib. </i>(Edward Elgar, Cheltenham, UK), 512.</nobr></div> <div style="position:absolute;top:34483;left:108"><nobr>Ho J, Ho K, Holland Mortimer J (2012) Analyzing the Welfare Impacts of Full-line Forcing</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:34601;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:34627;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=30><b>Page 30</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:35748;left:450"><nobr>30</nobr></div> <div style="position:absolute;top:34739;left:144"><nobr>Contracts. <i>J. Ind. Econ. </i>60(3):468–498.</nobr></div> <div style="position:absolute;top:34760;left:108"><nobr>Ho K, Ho J, Mortimer JH (2012) The Use of Full-Line Forcing Contracts in the Video Rental</nobr></div> <div style="position:absolute;top:34780;left:144"><nobr>Industry. <i>Am. Econ. Rev. </i>102(2):686–719.</nobr></div> <div style="position:absolute;top:34801;left:108"><nobr>Hristakeva S (2016) How Do Vertical Contracts Affect Product Availability? An Empirical</nobr></div> <div style="position:absolute;top:34822;left:144"><nobr>Study of the Grocery Industry, working paper.</nobr></div> <div style="position:absolute;top:34842;left:108"><nobr>Huang Q, Nijs VR, Hansen K, Anderson ET (2012) Wal-Mart’s Impact on Supplier Profits. <i>J.</i></nobr></div> <div style="position:absolute;top:34863;left:144"><nobr><i>Mark. Res. </i>49(2):131–143.</nobr></div> <div style="position:absolute;top:34884;left:108"><nobr>Iaria A Consumer benefit of big-box supermarket: The importance of controlling for endogenous</nobr></div> <div style="position:absolute;top:34905;left:144"><nobr>entry, working paper.</nobr></div> <div style="position:absolute;top:34925;left:108"><nobr>Jagabathula S, Vulcano G (2017) A Partial-Order-Based Model to Estimate Individual</nobr></div> <div style="position:absolute;top:34946;left:144"><nobr>Preferences Using Panel Data. <i>Manage. Sci. </i>(forthcoming).</nobr></div> <div style="position:absolute;top:34967;left:108"><nobr>Jerath K, Netessine S (2010) Revenue management with strategic customers: Last-minute selling</nobr></div> <div style="position:absolute;top:34987;left:144"><nobr>and opaque selling. <i>Management </i>56(3):430–448.</nobr></div> <div style="position:absolute;top:35004;left:108"><nobr>Kalyanam K, Borle S, Boatwright P (2007) Deconstructing each item’s category contribution.</nobr></div> <div style="position:absolute;top:35029;left:144"><nobr><i>Mark. Sci. </i>26(3):327–341.</nobr></div> <div style="position:absolute;top:35049;left:108"><nobr>Kök A, Fisher M, Vaidyanathan R (2008) Assortment planning: Review of literature and</nobr></div> <div style="position:absolute;top:35070;left:144"><nobr>industry practice. <i>Retail supply Chain Manag.</i></nobr></div> <div style="position:absolute;top:35091;left:108"><nobr>Kök AG, Fisher ML (2007) Demand Estimation and Assortment Optimization Under</nobr></div> <div style="position:absolute;top:35112;left:144"><nobr>Substitution: Methodology and Application. <i>Oper. Res. </i>55(6):1001–1021.</nobr></div> <div style="position:absolute;top:35132;left:108"><nobr>Lee J, Gaur V, Muthulingam S (2016) Stockout-based substitution and inventory planning in</nobr></div> <div style="position:absolute;top:35153;left:144"><nobr>textbook retailing. <i>Manuf. Serv. Oper. Manag. </i>18(1):104–121.</nobr></div> <div style="position:absolute;top:35174;left:108"><nobr>Li G, Rusmevichientong P, Topaloglu H (2015) The d-Level Nested Logit Model: Assortment</nobr></div> <div style="position:absolute;top:35194;left:144"><nobr>and price optimization problems. <i>Oper. Res. </i>63(2):325–342.</nobr></div> <div style="position:absolute;top:35215;left:108"><nobr>Li J, Granados N, Netessine S (2014) Are consumers strategic? Structural estimation from the</nobr></div> <div style="position:absolute;top:35236;left:144"><nobr>air-travel industry. <i>Manage. Sci. </i>60(9):2114–2137.</nobr></div> <div style="position:absolute;top:35257;left:108"><nobr>Lu Y, Musalem A, Olivares M, Schilkrut A (2013) Measuring the effect of queues on customer</nobr></div> <div style="position:absolute;top:35277;left:144"><nobr>purchases. <i>Manage. Sci. </i>59(8):1743–1763.</nobr></div> <div style="position:absolute;top:35298;left:108"><nobr>Mahajan S, van Ryzin G (2001a) Inventory competition under Inventory competition under</nobr></div> <div style="position:absolute;top:35319;left:144"><nobr>dynamic consumer choice. <i>Oper. Res. </i>49(5):646–657.</nobr></div> <div style="position:absolute;top:35339;left:108"><nobr>Mahajan S, van Ryzin G (2001b) Stocking retail assortments under dynamic consumer</nobr></div> <div style="position:absolute;top:35360;left:144"><nobr>substitution. <i>Oper. Res. </i>49(3):334–351.</nobr></div> <div style="position:absolute;top:35381;left:108"><nobr>Mandelbaum A, Zeltyn S (2004) The impact of customers patience on delay and abandonment:</nobr></div> <div style="position:absolute;top:35401;left:144"><nobr>Some empirically-driven experiments with the MMnG queue. <i>OR Spectr. </i>26(3):377–411.</nobr></div> <div style="position:absolute;top:35422;left:108"><nobr>Manzini P, Mariotti M (2014) Stochastic Choice and Consideration Sets. <i>Econometrica</i></nobr></div> <div style="position:absolute;top:35443;left:144"><nobr>82(3):1153–1176.</nobr></div> <div style="position:absolute;top:35464;left:108"><nobr>Matsa DA (2011) Competition and Product Quality in the Supermarket Industry. <i>Q. J. Econ.</i></nobr></div> <div style="position:absolute;top:35484;left:144"><nobr>126(3):1539–1591.</nobr></div> <div style="position:absolute;top:35505;left:108"><nobr>Mersereau A, Zhang D (2012) Markdown pricing with unknown fraction of strategic customers.</nobr></div> <div style="position:absolute;top:35526;left:144"><nobr><i>Manuf. Serv. Oper. Manag. </i>14(3):355–370.</nobr></div> <div style="position:absolute;top:35546;left:108"><nobr>Mittal V, Kamakura WA (2001) Satisfaction, Repurchase Intent, and Repurchase Behavior:</nobr></div> <div style="position:absolute;top:35567;left:144"><nobr>Investigating the Moderating Effect of Customer Characteristics. <i>J. Mark. Res. </i>38(1):131–</nobr></div> <div style="position:absolute;top:35588;left:144"><nobr>142.</nobr></div> <div style="position:absolute;top:35608;left:108"><nobr>Moon K, Bimpikis K, Mendelson H (2017) Randomized markdowns and online monitoring.</nobr></div> <div style="position:absolute;top:35629;left:144"><nobr><i>Manage. Sci. </i>(forthcoming).</nobr></div> <div style="position:absolute;top:35650;left:108"><nobr>Mortimer JH (2008) Vertical Contracts in the Video Rental Industry Vertical Contracts in the</nobr></div> <div style="position:absolute;top:35671;left:144"><nobr>Video Rental Industry1. <i>Source Rev. Econ. Stud. </i>75(1):165–199.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:35789;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> <div style="position:absolute;top:35815;left:0"><hr><table border=0 width=100%><tr><td bgcolor=eeeeee align=right><font face=arial,sans-serif><a name=31><b>Page 31</b></a></font></td></tr></table></div><font size=3 face="Times"><span style="font-size:16px;font-family:Times"> <div style="position:absolute;top:36936;left:450"><nobr>31</nobr></div> <div style="position:absolute;top:35927;left:108"><nobr>Musalem A (2015) When Demand Projections are Too Optimistic: A Structural Model of</nobr></div> <div style="position:absolute;top:35948;left:144"><nobr>Product Line and Pricing Decisions. <i>SSRN Electron. J.</i></nobr></div> <div style="position:absolute;top:35968;left:108"><nobr>Musalem A, Olivares M, Bradlow ET, Terwiesch C, Corsten D (2010) Structural estimation of</nobr></div> <div style="position:absolute;top:35989;left:144"><nobr>the effect of out-of-stocks. <i>Manage. Sci. </i>56(7):1180–1197.</nobr></div> <div style="position:absolute;top:36010;left:108"><nobr>Musalem A, Shang Y, Song JSJ (2016) An Empirical Study of Customer Spillover Learning</nobr></div> <div style="position:absolute;top:36030;left:144"><nobr>About Service Quality. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2811774.</nobr></div> <div style="position:absolute;top:36051;left:108"><nobr>Nair H (2007) Intertemporal price discrimination with forward-looking consumers: Application</nobr></div> <div style="position:absolute;top:36072;left:144"><nobr>to the US market for console video-games. <i>Quant. Mark. Econ. </i>5(3):239–292.</nobr></div> <div style="position:absolute;top:36093;left:108"><nobr>Olivares M, Cachon GP (2009) Competing Retailers and Inventory: An Empirical Investigation</nobr></div> <div style="position:absolute;top:36109;left:144"><nobr>of General Motors’ Dealerships in Isolated U.S. Markets. <i>Manage. Sci. </i>55(9):1586–1604.</nobr></div> <div style="position:absolute;top:36134;left:108"><nobr>Png IPL, Reitman D (1994) Service time competition. <i>Rand J. Econ.</i>:619–634.</nobr></div> <div style="position:absolute;top:36155;left:108"><nobr>Quan TW, Williams KR (2016) Product Variety, Across-Market Demand Heterogeneity, and the</nobr></div> <div style="position:absolute;top:36175;left:144"><nobr>Value of Online Retail. <i>SSRN Electron. J.</i></nobr></div> <div style="position:absolute;top:36196;left:108"><nobr>Rusmevichientong P, Shmoys D, Tong C, Topaloglu H (2014) Assortment optimization under</nobr></div> <div style="position:absolute;top:36217;left:144"><nobr>the Multinomial Logit Model with random choice parameters. <i>Prod. Oper. Manag.</i></nobr></div> <div style="position:absolute;top:36237;left:144"><nobr>23(11):2023–2039.</nobr></div> <div style="position:absolute;top:36258;left:108"><nobr>Rust J (1987) Optimal replacement of GMC bus engines: An empirical model of Harold Zurcher.</nobr></div> <div style="position:absolute;top:36279;left:144"><nobr><i>Econom. J. Econom. Soc.</i>:999–1033.</nobr></div> <div style="position:absolute;top:36300;left:108"><nobr>van Ryzin G, Vulcano G (2015) A market discovery algorithm to estimate a general class of non-</nobr></div> <div style="position:absolute;top:36320;left:144"><nobr>parametric choice model. <i>Manage. Sci. </i>61(2):281–300.</nobr></div> <div style="position:absolute;top:36341;left:108"><nobr>Shah D, Kumar V, Zhao Y (2015) Diagnosing Brand Performance: Accounting for the Dynamic</nobr></div> <div style="position:absolute;top:36362;left:144"><nobr>Impact of Product Availability with Aggregate Data. <i>J. Mark. Res. </i>52(2):147–165.</nobr></div> <div style="position:absolute;top:36382;left:108"><nobr>Soysal GP, Krishnamurthi L (2012) Demand dynamics in the seasonal goods industry: An</nobr></div> <div style="position:absolute;top:36403;left:144"><nobr>empirical analysis. <i>Mark. Sci. </i>31(2):293–316.</nobr></div> <div style="position:absolute;top:36424;left:108"><nobr>Sudhir K (2001) Structural Analysis of Manufacturer Pricing in the Presence of a Strategic</nobr></div> <div style="position:absolute;top:36445;left:144"><nobr>Retailer. <i>Mark. Sci. </i>20(3):244–264.</nobr></div> <div style="position:absolute;top:36465;left:108"><nobr>Talluri K, van Ryzin G (2004) Revenue Management under a General Discrete Choice Model of</nobr></div> <div style="position:absolute;top:36486;left:144"><nobr>Consumer Behavior. <i>Manage. Sci. </i>50(1):15–33.</nobr></div> <div style="position:absolute;top:36507;left:108"><nobr>Tang CS (2010a) A review of marketing-operations interface models: From co-existence to</nobr></div> <div style="position:absolute;top:36527;left:144"><nobr>coordination and collaboration. <i>Int. J. Prod. Econ. </i>125(1):22–40.</nobr></div> <div style="position:absolute;top:36548;left:108"><nobr>Tang CS (2010b) A review of marketing-operations interface models: From co-existence to</nobr></div> <div style="position:absolute;top:36569;left:144"><nobr>coordination and collaboration. <i>Int. J. Prod. Econ. </i>125(1):22–40.</nobr></div> <div style="position:absolute;top:36589;left:108"><nobr>Tenn S, Yun JM (2008) Biases in demand analysis due to variation in retail distribution. <i>Int. J.</i></nobr></div> <div style="position:absolute;top:36610;left:144"><nobr><i>Ind. Organ. </i>26(4):984–997.</nobr></div> <div style="position:absolute;top:36631;left:108"><nobr>Thomadsen R (2007) Product positioning and competition: The role of location in the fast food</nobr></div> <div style="position:absolute;top:36652;left:144"><nobr>industry. <i>Mark. Sci. </i>26(6):792–804.</nobr></div> <div style="position:absolute;top:36672;left:108"><nobr>Veeraraghavan S, Debo L (2009) Joining Longer Queues: Information Externalities in Queue</nobr></div> <div style="position:absolute;top:36693;left:144"><nobr>Choice. <i>Manuf. Serv. Oper. Manag. </i>11(4):543–562.</nobr></div> <div style="position:absolute;top:36714;left:108"><nobr>Villas-Boas SB (2007) Using retail data for upstream merger analysis. <i>J. Compet. Law Econ.</i></nobr></div> <div style="position:absolute;top:36734;left:144"><nobr>3(4):689–715.</nobr></div> <div style="position:absolute;top:36755;left:108"><nobr>Vulcano G, van Ryzin G, Ratliff R (2012) Estimating primary demand for substitutable products</nobr></div> <div style="position:absolute;top:36776;left:144"><nobr>from sales transaction data. <i>Oper. Res. </i>60(2):313–334.</nobr></div> <div style="position:absolute;top:36796;left:108"><nobr>Watson R (2009) Product variety and competition in the retail market for eyeglasses. <i>J. Ind.</i></nobr></div> <div style="position:absolute;top:36817;left:144"><nobr><i>Econ. </i>57(2):217–251.</nobr></div> <div style="position:absolute;top:36838;left:108"><nobr>Yu Q, Allon G, Bassamboo A (2016) How do delay announcements shape customer behavior?</nobr></div> <div style="position:absolute;top:36859;left:144"><nobr>An empirical study. <i>Manage. Sci. </i>63(1):1–20.</nobr></div> </span></font> <font size=3 color="#c3c3c3" face="Times"><span style="font-size:13px;font-family:Times;color:#c3c3c3"> <div style="position:absolute;top:36977;left:249"><nobr>Electronic copy available at: https://ssrn.com/abstract=3018732</nobr></div> </span></font> </body> </html>