论文速递 | Management Science 4月文章合集（上）

编者按：

在本系列文章中，我们梳理了运筹学顶刊Management Science在2024年4月份发布的有关OR/OM以及相关应用的13篇文章的基本信息，旨在帮助读者快速洞察领域新动态。本文为第一部分（1/2）。

推荐文章1

• 题目：Submodular Order Functions and Assortment Optimization 次模排序函数与分类优化
• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2021.04108
• 发表日期：2024/04/01
• 作者：Rajan Udwani
• 摘要：
  • 我们定义了一类新的函数集合，该函数集不仅具有单调性和次加性，还承认在基集的排列上定义的极有限形式的子模性。我们把这种排列称为次模顺序。这类函数包括作为一个子族的单调子模函数。我们给出了在各种约束条件下，具有强逼近保证的次模阶函数最大化的快速算法，并给出了具有多项式查询复杂度的算法所能达到的最高逼近保证的近紧上界。将这一新概念应用于基本选择模型中的约束分类优化问题，我们获得了速度更快且具有更强近似保证的新算法(在某些情况下，第一种算法具有恒定因子保证)。我们还展示了与流模型中单调子模函数最大化的有趣联系，其中我们恢复了最著名的近似保证作为我们推论的必然结果。
  • We define a new class of set functions that, in addition to being monotone and subadditive, also admit a very limited form of submodularity defined over a permutation of the ground set. We refer to this permutation as a submodular order. This class of functions includes monotone submodular functions as a subfamily. We give fast algorithms with strong approximation guarantees for maximizing submodular order functions under a variety of constraints and show a nearly tight upper bound on the highest approximation guarantee achievable by algorithms with polynomial query complexity. Applying this new notion to the problem of constrained assortment optimization in fundamental choice models, we obtain new algorithms that are both faster and have stronger approximation guarantees (in some cases, first algorithm with constant factor guarantee). We also show an intriguing connection to the maximization of monotone submodular functions in the streaming model, where we recover best known approximation guarantees as a corollary of our results.

推荐文章2

题目：Should Bank Stress Tests Be Fair?银行压力测试应该变得公平吗？

• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2022.02060
• 发表日期：2024/04/02
• 作者：Paul Glasserman , Mike Li
• 摘要：
  • 监管压力测试已成为美国大型银行设定资本金要求的主要工具之一。美联储使用保密模型来评估银行特定投资组合在共同压力情景下的特定结果。尽管各机构之间存在相当大的差异，但作为一个政策问题，所有银行都采用相同的模型。个别银行认为有些模型不适合它们的业务。在这场讨论的推动下，我们想知道怎样才能公平地将各自定制的模型汇总到一个通用模型中？我们认为，简单地将各家银行的数据汇集在一起，对各家银行一视同仁，但存在两个缺陷：一是可能会扭曲合理的投资组合特征的影响，二是推测银行身份的过程容易受到合法信息的隐含误导。我们比较了回归公平性的各种概念，以解决这些缺陷，同时考虑到预测准确性和平等处理。在线性模型中，我们认为估计然后剔除中心银行固定效应比简单地忽略银行之间的差异更可取。我们还讨论了非线性模型的扩展。
  • Regulatory stress tests have become one of the main tools for setting capital requirements at the largest U.S. banks. The Federal Reserve uses confidential models to evaluate bank-specific outcomes for bank-specific portfolios in shared stress scenarios. As a matter of policy, the same models are used for all banks, despite considerable heterogeneity across institutions; individual banks have contended that some models are not suited to their businesses. Motivated by this debate, we ask, what is a fair aggregation of individually tailored models into a common model? We argue that simply pooling data across banks treats banks equally but is subject to two deficiencies: it may distort the impact of legitimate portfolio features, and it is vulnerable to implicit misdirection of legitimate information to infer bank identity. We compare various notions of regression fairness to address these deficiencies, considering both forecast accuracy and equal treatment. In the setting of linear models, we argue for estimating and then discarding centered bank fixed effects as preferable to simply ignoring differences across banks. We also discuss extensions to nonlinear models.

推荐文章3

题目：Offline Returns for Online Retailers via Partnership通过合作关系为在线零售商提供离线退货

• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2023.01291
• 发表日期：2024/04/03
• 作者：Leela Nageswaran , Elina H. Hwang , Soo-Haeng Cho
• 摘要：
  • 网上购物者通常更愿意将商品退还给商店（即 “线下退货”），而不是邮寄回去。我们研究了一种新的商业行为——退货合作，即网络零售商与商店零售商合作提供线下退货。我们试图找出退货合作何时对商店和在线零售商都有利。顾客会在网上购物和实体店购物渠道之间做出选择，并决定是否退货，以及在需要时通过哪个渠道退货。我们发现，提供差异化产品的商店和网络零售商有合作的动机，例如 Everlane 和 Cost Plus World Market 之间的退货合作（通过 Happy Returns——一家连接网络零售商和商店零售商的服务提供商）。在这种情况下，网络零售商通过整合从商店退回商品的运输而节省的成本所带来的收益超过了高退货率所带来的损失。我们的研究表明，与最初的直觉相反，退货合作关系也可能以提供类似产品的零售商为特征，如亚马逊-科尔公司的案例。在这种情况下，在线顾客向线下退货的迁移确保了商店零售商的合作动机。然而，当线下退货的便利性促使更多的在线购物者退货时，就会限制在线零售商的积极性。因此，我们提醒大家，在产品差异化有限的情况下，零售商到店购物不应过于方便，以至于让退货合作关系难以形成。
  • Online shoppers often prefer to return items to stores (i.e., “offline return”) rather than mail them back. We study a new business practice, return partnership, wherein online retailers partner with store retailers to offer offline returns. We seek to identify when return partnerships benefit both the store and online retailers. Customers choose between the online and store channels for their purchase and decide whether, and through which channel, to return an online purchase if needed. We find that store and online retailers who offer differentiated products have incentives to partner, as in the case of the return partnership between Everlane and Cost Plus World Market (formed via Happy Returns—a service provider that connects online retailers with store retailers). In this case, the online retailer’s benefit from the cost savings achieved by consolidating shipments of returned items from stores outweighs the loss associated with a high return rate. We show that, contrary to initial intuition, return partnerships may also feature retailers who offer similar products, as in the case of Amazon-Kohl’s. In this case, online customers’ migration to offline returns ensures a store retailer’s incentive to partner. However, it limits the incentive of an online retailer when the added convenience of an offline return induces more returns of online purchases. Thus, we caution that with limited differentiation in product offerings, store visits should not be too convenient for a partnership to form.

推荐文章4

• 题目：Offline Planning and Online Learning Under Recovering Rewards回收奖励下的离线规划与在线学习
• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2021.04202
• 发表日期：2024/04/03
• 作者：David Simchi-Levi , Zeyu Zheng , Feng Zhu
• 摘要：
  • 受直播电子商务、促销和推荐等新兴应用的启发，我们引入并解决了一类具有以下两个特征的非稳态多臂老虎机问题： (i) 决策者可以在每个时间段内从 N 个不同的拉杆中拉出并收集最多 $K(\ge 1)$ 的奖励；(ii) 一个拉杆的预期奖励在拔出后立即下降，然后随着拉杆闲置时间的增加而非参数地恢复。我们以在 T 个时间段内最大化预期累积奖励为目标，设计了一类纯周期性策略，这些策略共同设定了拉动每个拉杆的时间段。对于所提出的策略，我们证明了离线问题和在线问题的性能保证。对于离线问题，当所有模型参数都已知时，所提出的周期性策略获得的长期近似率为 $1-O(1/\sqrt{K})$ ，当 K 增长到无穷大时，该近似率是渐进最优的。对于模型参数未知且需要动态学习的在线问题，我们将离线周期性策略与置信上界步骤相结合，构建了在线策略。事实证明，与离线基准相比，所提出的在线策略近似具有 $\tilde{O}(N\sqrt{T})$的遗憾。我们的框架和策略设计可以为更广泛的离线规划和在线学习应用提供启示，这些应用具有非稳态和奖励恢复的特点。
  • Motivated by emerging applications, such as live-streaming e-commerce, promotions, and recommendations, we introduce and solve a general class of nonstationary multi-armed bandit problems that have the following two features: (i) the decision maker can pull and collect rewards from up to $K(\ge 1)$out of N different arms in each time period and (ii) the expected reward of an arm immediately drops after it is pulled and then nonparametrically recovers as the arm’s idle time increases. With the objective of maximizing the expected cumulative reward over T time periods, we design a class of purely periodic policies that jointly set a period to pull each arm. For the proposed policies, we prove performance guarantees for both the offline and the online problems. For the offline problem when all model parameters are known, the proposed periodic policy obtains a long-run approximation ratio that is at the order of $1-O(1/\sqrt{K})$ , which is asymptotically optimal when K grows to infinity. For the online problem when the model parameters are unknown and need to be dynamically learned, we integrate the offline periodic policy with the upper confidence bound procedure to construct on online policy. The proposed online policy is proved to approximately have $\tilde{O}(N\sqrt{T})$ regret against the offline benchmark. Our framework and policy design may shed light on broader offline planning and online learning applications with nonstationary and recovering rewards.

推荐文章5

题目：Exercising Market Power Without Using Prices: Service Time in Online Grocery在不使用价格的情况下行使市场力量:在线杂货的服务时间

• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2022.01820
• 发表日期：2024/04/04
• 作者：Itai Ater , Adi Shany
• 摘要：
  • 本文研究了在全国统一定价的情况下，在线杂货商如何利用服务时间来应对当地的竞争和需求状况。利用三年来每周两次从 172 个以色列本地市场收集的综合数据，我们表明，在线杂货商在所有市场都会设定相同的价格。相比之下，在竞争更激烈的市场和低需求日，服务时间更短。接下来，我们利用进入决策的地区和时间差异，研究当新的在线杂货商进入市场时，原有企业如何调整其服务时间。在低需求日和垄断市场中，现有企业的服务时间明显缩短。这种下降始于进入市场前不久，当新进入者构成较大竞争威胁时，下降幅度更大。在高需求日和竞争性市场中，我们没有发现服务时间在进入市场前几个月有明显变化。我们的研究结果表明，当价格反应迟钝时，企业会利用服务时间来行使其本地市场支配力，而运营方面的考虑因素会影响企业的反应程度。
  • This paper studies how online grocers use service time to respond to local competition and demand conditions when prices are uniformly set at the national level. Using comprehensive data collected twice a week over three years from 172 Israeli local markets, we show that an online grocer sets identical prices in all markets. By contrast, service time is shorter in more competitive markets and on low-demand days. Next, we exploit regional and temporal variation in entry decisions to examine how the incumbent adjusts its service time when new online grocers enter the market. The incumbent’s service time falls significantly on low-demand days and in monopolistic markets. This decrease begins shortly before entry and is greater when the entrant poses a larger competitive threat. On high-demand days and in competitive markets we do not find a significant change in service time in the months surrounding entry. Our findings suggest that firms use service time to exercise their local market power when prices are unresponsive and that operational considerations affect the extent to which they respond.

推荐文章6

题目：Salient Cues and Complexity显著线索和复杂性

• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2022.00652
• 发表日期：2024/04/05
• 作者：Markus Dertwinkel-Kalt , Mats Köster
• 摘要：
  • 重要决策往往很复杂，现有证据表明，复杂性会影响经济行为。然而，复杂性究竟何时以及如何产生影响还是一个未决问题。我们假设，突出的线索——在选择情境中脱颖而出——会减轻复杂性对选择的影响。我们以投资组合选择为背景，从理论上提出并通过实验验证了这一假设。我们发现，无论是在简单问题还是复杂问题中，被试都会寻求具有极端显著上行趋势的高右倾投资组合，而回避具有极端显著下行趋势的高左倾投资组合。然而，复杂性确实会影响对称投资组合的选择，因为对称投资组合既没有突出的上风，也没有突出的下风。在没有显著提示的情况下，受试者会天真地分散投资。反应时间和记忆方面的证据支持我们基于显著性的解释。
  • Important decisions are often complex, and existing evidence suggests that complexity can affect economic behavior. It is an open question, however, exactly when and how complexity matters. We hypothesize that salient cues—standing out in the choice context—mitigate the effect of complexity on choices. We theoretically develop and experimentally test this hypothesis in the context of portfolio selection. We find that, in both simple and complex problems, subjects seek highly right-skewed portfolios, which have an extreme and salient upside, and avoid highly left-skewed portfolios with an extreme and salient downside. Complexity does affect, however, choices among symmetric portfolios, which have neither a salient upside nor downside. Absent a salient cue, subjects diversify naively. Evidence on response times and memory supports our salience-based explanation.

推荐文章7

• 题目：Reducing Interference Bias in Online Marketplace Experiments Using Cluster Randomization: Evidence from a Pricing Meta-experiment on Airbnb利用聚类随机化减少在线市场实验中的干扰偏差:来自Airbnb定价元实验的证据
• 期刊：Management Science
• 原文链接：https://doi.org/10.1287/mnsc.2020.01157
• 发表日期：2024/04/05
• 作者：David Holtz , Felipe Lobel , Ruben Lobel , Inessa Liskovich, Sinan Aral
• 摘要：
  • 在线市场设计师经常进行随机实验，以衡量拟议的产品变化的影响。然而，鉴于市场是内在联系的，通过个体水平随机实验获得的总平均处理效果(TATE)估计可能会因为违反稳定单位治疗值假设而产生偏差，我们将这种现象称为“干扰偏差”。聚类随机化(即在相似个体的“聚类”水平上随机分配处理任务的做法)是一种成熟的实验设计技术，用于消除社会网络中的干扰偏见，但目前尚不清楚它是否会在市场环境中有效。在本文中，我们利用在Airbnb上进行的元实验或“实验中的实验”来提供在线市场环境中干扰偏差的经验证据，并评估聚类随机化作为减少市场TATE估计中干扰偏差的工具的可行性。我们的元实验结果表明，对我们所研究的平台费用增加的个人层面随机评估产生的TATE估计中，至少有20%可归因于干扰偏差，而使用聚类随机化则可消除干扰偏差。我们还发现，在需求受限的地区，卖方实验中的干扰偏倚更为严重，而聚类随机化在减少干扰偏倚方面的效果随着聚类质量的提高而提高。
  • Online marketplace designers frequently run randomized experiments to measure the impact of proposed product changes. However, given that marketplaces are inherently connected, total average treatment effect (TATE) estimates obtained through individual-level randomized experiments may be biased because of violations of the stable unit treatment value assumption, a phenomenon we refer to as “interference bias.” Cluster randomization (i.e., the practice of randomizing treatment assignment at the level of “clusters” of similar individuals) is an established experiment design technique for countering interference bias in social networks, but it is unclear ex ante if it will be effective in marketplace settings. In this paper, we use a meta-experiment or “experiment over experiments” conducted on Airbnb to both provide empirical evidence of interference bias in online marketplace settings and assess the viability of cluster randomization as a tool for reducing interference bias in marketplace TATE estimates. Results from our meta-experiment indicate that at least 20% of the TATE estimate produced by an individual-level randomized evaluation of the platform fee increase we study is attributable to interference bias and eliminated through the use of cluster randomization. We also find suggestive, nonstatistically significant evidence that interference bias in seller-side experiments is more severe in demand-constrained geographies and that the efficacy of cluster randomization at reducing interference bias increases with cluster quality.