{"id":619407,"date":"2019-11-03T09:40:02","date_gmt":"2019-11-03T17:40:02","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-project&p=619407"},"modified":"2022-10-26T10:00:51","modified_gmt":"2022-10-26T17:00:51","slug":"market-design-center","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/market-design-center\/","title":{"rendered":"Market Design Center"},"content":{"rendered":"
Markets match individuals to scarce resources or to each other.\u00a0 Online platforms are modern marketplaces, with users who interact to accomplish their own personal goals. We must therefore take care when designing the rules and infrastructure that govern online platforms in order to guide participants to beneficial outcomes.\u00a0 In the process, we circumvent a wide range of market failures.<\/p>\n
The academic literature uses economic theory to build a market design playbook.\u00a0 This playbook recommends rules that induce good incentives for participants, build market thickness, and keep congestion low.\u00a0 Following this playbook, economists and their peers have improved a wide array of public markets, including spectrum allocation, centralized labor markets, organ transplantation, school choice systems, affordable housing programs, and more.<\/p>\n
Our group helps apply the market design playbook to practical markets at Microsoft.\u00a0 By understanding user preferences and modeling their behavior, we can design a platform’s rules so that the designer’s goal is reached even when customers act strategically. Sometimes this means making it impossible to benefit from gaming the system. Sometimes this means embracing strategic behavior to learn about customer preferences. Using a combination of game theory, algorithm design, and microeconomic analysis, we design markets and platforms that are stable in the face of manipulation and improve economic outcomes for everyone involved.<\/p>\n
If you are interested in building a market for one of Microsoft’s many platforms, or optimizing an existing one, please get in touch by emailing one of the people below.<\/p>\n","protected":false},"excerpt":{"rendered":"
Markets match individuals to scarce resources or to each other.\u00a0 By designing the rules and infrastructure that govern markets, we can guide participants to optimal outcomes.\u00a0 In the process, we circumvent a wide range of market failures. If you are interested in building a market for one of Microsoft's many platforms, or optimizing an existing one, please get in touch.<\/p>\n","protected":false},"featured_media":619422,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"research-area":[13561,13556,13548],"msr-locale":[268875],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-619407","msr-project","type-msr-project","status-publish","has-post-thumbnail","hentry","msr-research-area-algorithms","msr-research-area-artificial-intelligence","msr-research-area-economics","msr-locale-en_us","msr-archive-status-active"],"msr_project_start":"","related-publications":[],"related-downloads":[],"related-videos":[],"related-groups":[],"related-events":[],"related-opportunities":[],"related-posts":[],"related-articles":[],"tab-content":[{"id":0,"name":"Resources","content":"Online talks<\/strong>.\r\n\r\nIn Fall of 2019, the Simons Institute for the Theory of Computing at UC Berkeley hosted a 6-week program on Online and Matching-Based Market Design<\/a>, co-organized by Microsoft researcher and Market Design Center member Nicole Immorlica.\u00a0 The program included 3 workshops.\u00a0 The talks are all recorded and can be accessed on website of the corresponding workshops: Matching and Objectives<\/a>, Platform Markets<\/a>, and Information Design and Data Science<\/a>.\r\n\r\nCourse syllabus.<\/strong>\r\n\r\nMicrosoft researchers and Market Design Center members Nicole Immorlica and Brendan Lucier co-teach a course on Markets for Networks and Crowds<\/a> at Harvard University.\u00a0 The course syllabus including lecture notes, readings, and assignments, can be found on the course website.\r\n\r\nTextbook.<\/strong>\r\n\r\nMicrosoft researcher and Market Design Center member Nicole Immorlica is co-editing a textbook on Market Design, to be released in 2022-2023.\u00a0 The introductory chapters on one-sided allocation and two-sided matching will be linked to here when they are available."},{"id":1,"name":"Projects","content":"Markets for Sustainability.<\/strong>\r\n\r\nCarbon removal is a crucial part of global sustainability efforts. Unfortunately, the current market for high-quality carbon removal is extremely thin. Better incentives for production and investment are sorely needed.\u00a0 Our group is actively investigating market mechanisms that improve the supply of carbon removal credits reliably and efficiently.\r\n\r\nOne of our projects in this space relates to the development of afforestation contracts.\u00a0 Afforestation (growth of net new trees) pulls carbon from the atmosphere. Trees on farms can also improve agricultural productivity and provide additional income to farmers through the sale of carbon reduction credits. Buyers of such credits, however, want to ensure that the farmer is properly incentivized to safeguard tree growth over time. Using the lens of contract theory, this research shows how to calculate the optimal stream of payments over time to ensure incentive alignment. This research can be applied to enlarge the supply of carbon reduction credits, especially outside the developed world. For more information, see our published paper Contract Design for Afforestation Programs<\/a>.\r\n\r\nTerritory Management and Salesforce Matching.<\/strong>\r\n\r\nMicrosoft enterprise customers are assigned teams of sales and support professionals to help manage and maintain their contracts and products.\u00a0 Each year, Microsoft must form a matching between its salesforce and customers, spanning many salesperson roles, geographic regions, and account types. This is a complex multi-objective matching problem, which must satisfy multiple constraints and balance different criteria of what makes a successful match.\r\n\r\nThe Market Design Center has worked with the Behavioral Analytics team to design and implement an algorithm to evaluate and propose assignments of salespeople to sales accounts. Our solution leveraged the theory of simulated annealing and many-to-many matchings to balance exploration and convergence to a high-valued feasible assignment. In addition to direct optimization, this algorithmic framework is used to evaluate counter-factual adjustments to the market, such as estimating the marginal benefit of increasing the number of salespeople in a given role.\r\n\r\nOnline Advertising in Bing.<\/strong>\r\n\r\nThe market design center has a long-running engagement with the Bing Advertising team, and we work closely with them to analyze pricing and allocation rules for ads on Bing.\u00a0 One sample engagement centers around pricing rules for so-called Rich ads.\u00a0 Rich ads have varying sizes, and as such, in a poorly-designed system, large ads can poach space from small ones.\u00a0 Our solution to selling Rich ads leverages the theory of core auctions to maximize the value of displayed ads while maintaining good revenue for the platform.\u00a0 For more information, see our published paper Fast Core Pricing for Rich Advertising Auctions<\/a>.\r\n\r\nAnother theme of engagement centers around the design of autobidders.\u00a0 An autobidder employs machine learning techniques to optimizing bidding strategies on behalf of advertisers.\u00a0 To understand the efficacy of an autobidder design, it is not enough to consider the individual guarantees obtained for an advertiser in isolation: one must also consider emergent behavior that arises when autobidders interact with each other.\u00a0 Our work in this space develops guidelines for the development of auction rules and autobidder designs that jointly achieve high-welfare outcomes in the aggregate, avoiding market failures due to unintended algorithmic collusion.\u00a0 For more information, see the following papers:\r\n\r\n \t