1 January 2019 in Manufacturing & Service Operations Management
We present an approach to forecast customer orders of ready-to-launch new products that are similar to past products. The approach fits product life cycle (PLC) curves to historical customer order data, clusters the curves of similar products, and uses the representative curve of the new product’s cluster to generate its forecast. We propose three families of curves to fit the PLC: bass diffusion curves, polynomial curves, and simple piecewise-linear curves (triangles and trapezoids). Using a large data set of customer orders for 4,037,826 units of 170 Dell computer products sold over three and a half years, we compare goodness of fit and complexity for these families of curves. Fourth-order polynomial curves provide the best in-sample fit with piecewise-linear curves a close second. Using a trapezoidal fit, we find that the PLCs in our data have very short maturity stages; more than 20% have no maturity stage and are best fit by a triangle. The fitted PLC curves of similar products are clustered either by known product characteristics or by data-driven clustering. Our key empirical finding is that, for our large data set, data-driven clustering of simple triangles and trapezoids, which are simple to estimate and explain, perform best for forecasting. Our conservative out-of-sample forecast evaluation, using data-driven clustering of triangles and trapezoids, results in mean absolute errors approximately 2%–3% below Dell’s forecasts. We also apply our method to a second data set of a smaller company and find consistent results. The online appendix is available at https://doi.org/10.1287/msom.2017.0691 .
1 July 2017 in Manufacturing & Service Operations Management
Online purchases constitute about one-tenth of U.S. retail sales. The supply chains that support online retailing are fundamentally different from those that support traditional brick-and-mortar stores. Traditional solutions are not always appropriate to solve online retailing’s operations problems; thus, there is an opportunity to understand and improve these novel supply chains. One key characteristic of the inventory systems for online retailing is demand spillover, whereby a stockout at a fulfillment center (FC) results in demand spilling over to another FC. For this context we examine how to allocate inventory to the FCs under a periodic-review joint-replenishment policy, with an objective to minimize outbound shipping costs for a fixed amount of inventory. We first show how traditional decentralized allocation policies may perform suboptimally and induce dynamics (whiplash) that result in costly spillover. We find that this phenomenon increases with the prevalence of local stockouts and with the level of inventory imbalance. We then describe that inventory imbalance occurs in online retailing because of operational realities and provide evidence based on real data. Finally, we propose a heuristic to allocate inventory accounting for possible spillover during the lead time. We test the heuristic by a simulation and show that it performs better than the status quo policy, is robust to operational realities, and captures over 90% of the possible improvement as compared to a pseudo-optimal policy. The online appendix is available at https://doi.org/10.1287/msom.2016.0614 .
1 January 2022 in Manufacturing & Service Operations Management
Problem definition: Firms providing products and services to low-income base of the pyramid (BOP) customers are increasingly utilizing independent contractor agents rather than employees in their distribution models. We empirically investigate the best way to help agents perform better. Academic/practical relevance: BOP customers represent one-third of the world’s economy but make five United States dollars or less daily. Providing goods and services to these customers is difficult for traditional firms because most retail activity occurs at small-scale independent outlets. Improving agent performance can help firms reach customers in this environment. We enhance the literature on agent-based models in BOP settings, decision making, technology in developing economies, and field experiments. Methodology: In partnership with a Tanzanian mobile money operator, we perform a randomized, controlled trial with 4,771 agents to examine how differing types of guidance, and whether in-person training is offered, impact agents’ inventory management. Mobile money is a platform whereby firms in developing economies provide financial services to customers via cell phones. Mobile money agents service customer withdrawals and deposits as branchless banking outlets. Every day, they decide how much money to stock to service customers’ transactions, from which they earn commissions. Results: We find that those agents given only explicit recommendations (as opposed to summary statistics or both) who were invited to in-person training (as opposed to simply received an automated notification) improve their performance. Agents in other treatments showed no statistically significant change. The effect is concentrated in agents who never replenished their money at a bank and whose money inventory levels were low in the pretreatment period. Managerial implications: We show empirically how firms can better manage agents, thereby improving the value proposition of serving BOP customers. We show the utility of segmentation based on agent heterogeneity. This can improve firm performance, agent profits, and customer service.
1 January 2019 in Manufacturing & Service Operations Management
We provide and describe a data set of N = 8,935 weekly, normalized customer orders over the entire product life cycle for 170 Dell computer products sold in North America over a three and a half year period, from 2013–2016. Total orders for these products exceeded four million units and well over one billion dollars in revenue. While Dell is historically known for fulfilling customer demand with a build-to-order approach, the products in this data set were designated as build-to-stock products. There are three elements in the data that, depending on the research application, researchers may want to identify or mitigate. First, some products have seemingly anomalous orders representing one-time purchases from large customers. Second, there are negative values for some products representing order cancellations. Third, end-of-life sales may be significantly influenced by management action. We present approaches for cleaning the data to address these issues.The supplemental material is available at https://doi.org/10.1287/msom.2017.0692 .
1 May 2023 in Manufacturing & Service Operations Management
Problem definition: Mobile money systems—platforms built and managed by mobile network platform operators (MMPOs) to allow money to be stored as digital currency—connect millions of poor and “unbanked” people to the formal financial system. Unfortunately, low service levels because of the suboptimal management of cash and digital currency (e-float) inventory impede the development of these ecosystems. Accordingly, we seek to answer the question of how agents should manage inventories of cash and e-float. Academic/practical relevance: This paper extends inventory theory to the mobile money context, unique in that sales of cash generate inventory of e-float and vice versa. In doing so, we address a key pain point for an emerging sector that improves lives at the base of the pyramid. Methodology: We develop an analytical heuristic to determine initial stocking levels for cash and e-float and analyze its performance on simulated and actual data. Results: By partnering with an MMPO, we tested the performance of the heuristic inventory policy with data from more than 35 million transactions. The heuristic captured 99.9998% of the optimal profit on simulated data and, on actual data, we found that following the recommendations could increase agents’ profits by an average of 15.4%. Managerial implications: We develop a pragmatic inventory policy that performs nearly optimally. We also analyze under which conditions the performance deteriorates and examine heterogeneity among agents with respect to the heuristic’s impact on their performance. Thus, we equip MMPOs with guidance as to whom to target and how. By contributing to service level and profit improvements, this work can make mobile money a more effective financial inclusion tool in the developing world as well as improve the livelihoods of agents. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.1175 .
22 September 2017 in SSRN Electronic Journal
We provide and describe a data set of N = 8935 weekly, normalized customer orders over the entire product life cycle for 170 Dell computer products sold in North America over a three and a half year period, from 2013-2016. Total orders for these products exceeded 4 million units and well over a billion dollars in revenue. While Dell is historically known for fulfilling customer demand with a buildto-order approach, the products in this data set were designated as build-to-stock products. There are three elements in the data that, depending on the research application, researchers may want to identify or mitigate. First, some products have seemingly anomalous orders representing one-time purchases from large customers. Second, there are negative values for some products representing order cancellations. Third, end-of-life sales may be significantly influenced by management action. We present approaches for cleaning the data to address these issues.