«Working Paper 13-096 May 3, 2013 Copyright © 2013 by Nathan Craig and Ananth Raman Working papers are in draft form. This working paper is ...»
Improving Store Liquidation
May 3, 2013
Copyright © 2013 by Nathan Craig and Ananth Raman
Working papers are in draft form. This working paper is distributed for purposes of comment and
discussion only. It may not be reproduced without permission of the copyright holder. Copies of working
papers are available from the author.
Improving Store Liquidation
Nathan Craig and Ananth Raman
Harvard Business School
May 3, 2013
Abstract Store liquidation is the time-constrained divestment of retail outlets through an in-store sale of inventory. The retail industry depends extensively on store liquidation, not only as a means for investors to recover capital from failed ventures, but also to allow managers of going concerns to divest stores in eﬀorts to enhance performance and to change strategy. Recent examples of entire chains being liquidated include Borders Group in 2012, Circuit City in 2009, and Linens ‘n Things in 2008; the value of inventory sold during these liquidations alone is $3B.
The store liquidation problem is related to but also diﬀers substantially from the markdown optimization problem that has been studied extensively in the literature. This paper introduces the store liquidation problem to the literature and presents a technique for optimizing key decision variables, such as markdown, inventory, and store closing decisions during liquidations.
We show that our approach could improve net recovery on cost (i.e., the proﬁt obtained during liquidations stated as a percentage of the cost value of liquidated assets) by 2 to 7 percentage points in the cases we examined. The paper also identiﬁes ways in which current practice in store liquidation diﬀers from the optimal decisions identiﬁed in the paper and traces the consequences of these diﬀerences.
1 Introduction Store liquidation, deﬁned as the as the time-constrained divestment of retail stores through an in-store sale of inventory, is a critical aspect of the retail industry for both defunct and going concerns. Going concerns use store liquidation to divest sets of stores or even entire concepts.
Struggling and failed retailers liquidate thousands of stores and billions of dollars of inventory each year. For example, as Borders Group entered bankruptcy in early 2011, it held $638.5M dollars of inventory.1 Circuit City and Linens ‘n Things held $1.5B and $795.4M, respectively, of inventory prior to entering bankruptcy. Given the size of these liquidations, even a small improvement in net recovery on cost—i.e., the proﬁt obtained during a liquidation stated as a percentage of the cost value of inventories liquidated—can be substantial. For example, a 1% improvement in Circuit City’s net recovery on cost would have amounted to $15M.
Store liquidation has important implications for ﬁrms and investors. Bankruptcy and, thus, liquidation are common in retailing: for the United States alone, Capital IQ records 2,013 retailer bankruptcy announcements over the decade beginning in 2000. Further, Gaur et al. (2013) ﬁnd that 3.4% of all public retailers during the past 20 years were liquidated in bankruptcy. Store liquidations operated by asset disposition ﬁrms like Gordon Brothers Group (GBG) and Hilco Merchant Resources, such as those conducted for Montgomery Ward (Ordonez et al., 2001) and Syms Corporation, the owner of Filene’s Basement (Mattioli, 2011), help stakeholders recoup funds from failed ﬁrms and allow investors to shunt capital to other ventures. According to managers at GBG, that ﬁrm alone liquidated over $2B of inventory measured at retail value during 2011.
Improving store liquidation in bankruptcy can also directly increase liquidity within the retail industry, since the funds and terms available to retailers using inventory-based lending, a type of asset-based lending in which the collateral is a retailer’s inventory, depend on the estimated NOLV of the inventory (Foley et al., 2012). Inventory-based lending is an important source of capital in the retail sector (Alan and Gaur, 2012). Example inventory-based loans include a $3.28B revolver held by Sears Holdings Corporation and a $1B revolver used by Barnes & Noble.2 Of course, the liquidation value of a retailer can also aﬀect its ability to obtain trade credit as well as the terms of the credit it receives; see, for example, Yang and Birge (2011).
Store liquidation is a valuable tool for going concerns: the ability to redeploy resources by eﬀectively liquidating subsets of stores is key for managers. Store liquidations allow ﬁrms to generate The inventory ﬁgures reported in this paragraph are recorded by Capital IQ.
These ﬁgures are reported in ﬁrm quarterly ﬁlings and are current as of January 1, 2013.
cash from poorly performing stores and chains, as in the cases of Barnes & Noble’s decision to close roughly 200 stores over the coming decade (Trachtenberg, 2013), of Sears Holdings’ liquidation of over 100 Sears and Kmart stores (Lahart, 2011), and of Home Depot’s closing of its EXPO stores (Zimmerman, 2009). Store liquidation also allows managers to free resources to abet a change in strategy, as in Best Buy’s closing of 50 big box stores to fund a new focus on mobile device stores (Bustillo, 2012). Store liquidation is useful in other situations as well: when Pamida, a department store chain, merged with Shopko (LBO Wire, 2012), a similar ﬁrm, managers conducted store liquidations to empty Pamida stores and prepare them for conversion to Shopko stores. Other going concerns that rely on asset disposition ﬁrms to close stores include Dick’s Sporting Goods, Forever 21, J.C. Penney, Rite Aid, and Saks Fifth Avenue.3 The store liquidation problem diﬀers from the markdown optimization problem that has been studied extensively in the literature. First, unlike in the markdown optimization problem, retailers have to close (i.e., stop operating their stores) in the store liquidation problem. The decision of when to close a store and, if needed, move the merchandise to another store is an integral part of the liquidation problem. The decision of which stores to open on a particular day adds a number of binary decision variables to the optimization problem and is a function of demand levels, store operating costs, and inter-store transfer times and costs.
Second, consumers behave diﬀerently during a store liquidation than at a store under normal operation. Hence, there is considerably more demand uncertainty during store liquidation than during normal store operations or even in markdown optimization. It is hard to predict ex-ante how consumers will react to a liquidation event, and the reaction can diﬀer substantially from one store in a liquidation event to another. As the liquidation progresses, the level of demand uncertainty goes down. Consequently, in identifying optimal liquidation approaches, one needs to explicitly incorporate these phenomena—demand uncertainty and forecast updating.
Third, liquidations involve “quirks”—special features and constraints that characterize each event and often individual stores within the same liquidation event. For example, there might be greater ﬂexibility on when some stores in a chain can be shut down because of the lease agreement For more examples, see the client list posted by Hilco Merchant Resources at http://www.
with the store’s landlord. Similarly, as we illustrate in examples later in the paper, there might be limitations on changing markdown levels, inventory transfers, or store closings by deal. Any method to optimize store liquidation should be ﬂexible enough to accommodate these unique features associated with each liquidation.
In this paper, we introduce a method for improving the eﬃciency of store liquidations, i.e., for increasing the net orderly liquidation value (NOLV) of retail stores, with a focus on liquidations conducted by asset disposition ﬁrms. The method comprises a dynamic program that informs markdown, inventory, and store closing decisions as well as a demand forecasting model. We provide techniques for estimating the parameters in our model and a heuristic approach to solve the dynamic program. We compare the performance of our method to practice in selected case studies and show that the net recovery on cost improved by 2 to 7 percentage points. Through these applications, we provide novel insights gleaned from the use of our technique. GBG served as the test site and our collaborator for the research in this paper; we partnered with GBG on liquidating over $3B of inventory.
This paper is organized as follows. The next section provides background on the process of store liquidation. §3 discusses how our work relates to prior literature. §4 presents the full dynamic program. §5 introduces our solution methodology, including the modiﬁed program and the forecasting model. In §6, we discuss the performance of our methods in practice as well as insights garnered while applying the methods. Our concluding remarks are in §7.
2 The Process of Store Liquidation From the retail asset disposition ﬁrm’s perspective, the ﬁrst step of any liquidation is “getting the deal.” In the case of a bankruptcy liquidation, the liquidator receives information on store characteristics, inventory, and historical performance from the bankrupt retailer. Typical data include store location and square footage, store-level or category-level inventory in terms of cost and retail value, count, and age, as well as current- and last-year store revenues. The liquidator must then ﬁle a bid with the bankruptcy court for the right to liquidate the bankrupt ﬁrm’s inventory within the retailer’s extant retail outlets. The bidding process transpires quickly and is often limited to less than a week. In the case of a going concern liquidation, the asset disposition ﬁrm receives similar information and must engage in a sales process—i.e., earning the right to liquidate from the retailer, often through the estimation of net liquidation proceeds and the negotiation of fees.
During a store liquidation, inventory is sold at an increasing discount in a set of retail stores over a ﬁnite time period. The length of a liquidation is limited by law for both bankrupt ﬁrms and going concerns The majority of U.S. states constrain the length of all liquidation, distressed inventory, and going-out-of-business sales to protect consumers from ﬁrms that might perpetually use liquidation as a marketing tool. See, for example, Ohio Administrative Code Chapter 109:4-3which constrains liquidations to 90 days, and Massachusetts General Laws Part 1, Chapter 93, §28A, which limits going-out-of-business sales to 60 days. Many other jurisdictions impose similar restrictions.
Liquidators may execute a sale for a ﬁxed fee or on an equity basis. In the latter case, the asset disposition ﬁrm pays up front for the right to liquidate the inventory. In the equity case, the liquidator may share some portion of the proceeds with the retailer or the retailer’s estate. For instance, in advance of the ﬁnal liquidation of Borders Group during 2011, liquidators agreed to pay Borders’ estate 72% of the audited cost value of inventory present at the outset of liquidation plus 50% of the net proceeds of the liquidation (Checkler, 2011). Given the speed of store liquidation, even a small improvement in net recovery on cost can translate into a substantial increase in annualized return on investment in the case of an equity liquidation. For instance, suppose an asset disposition ﬁrm improves net recovery from 3%, which is fairly typical, to 5%. If the ﬁrm acquires $100M of inventory at cost and liquidates the inventory over the course of 12 weeks, then the annualized return on investment improves from 13% to 22%.
After securing a deal, the asset disposition ﬁrm quickly begins to execute the sale, often within a few days. Once the liquidation commences, each store is assigned a supervisor. At the store level, the supervisors work to increase the proﬁtability of the liquidation through inventory placement (to provide a pleasant shopping environment throughout a sale, asset disposition ﬁrms collapse a store by moving inventory toward the front of the store and cordoning oﬀ the back of the store) and expense management, including inventory shrinkage and payroll. These supervisors are often recruited from ﬁrms subject to a prior going-out-of-business sale and thus tend to have liquidation experience.
Prior exposure to liquidation is important to managers because stores in liquidation behave very diﬀerently than stores in normal operation. One way to illustrate this is to compare revenues during liquidation to revenues during normal operation. To do this, we construct liquidation multipliers, i.e., the ratio of revenue earned during the liquidation of a store to the revenue generated by that store over the same period during the prior year. Figure 1 presents box plots of store liquidation multipliers from four liquidations in the apparel, book, household furniture, and jewelry segments. As this plot shows, most stores see a signiﬁcant increase in revenue due to liquidation.
The median store in both the book and household furniture liquidations more than doubled its revenue. Moreover, this revenue perspective understates the physical volume of product sold, since liquidation discounts exceed normal operating discounts.
In current practice, asset disposition ﬁrms use central managers to plan a common markdown cadence across all stores. This markdown cadence is chosen based on the managers’ prior experience with a given retail segment. Broadly, the markdowns start relatively low—usually around 25%— and increase to approximately 85% over the course of the sale. Figure 2 plots the realized markdown levels over time across six retail chains in diﬀerent segments. Each store continues to operate until its inventory is sold through or until it turns unproﬁtable, i.e., when revenues exceed operating costs. Remaining inventory is either sold at a large discount to a jobber or, rarely, is transferred to a nearby store in the same chain. As will be discussed in §6, our method represents a substantial departure from this practice.