Egan Research Center, Room 206

Lindsey D. Cameron, University of Michigan
Making out While Driving: Control, Coordination, and its Consequences in Algorithmic Labor
Increasingly, algorithms are changing how work is structured and navigated. Drawing on a 26- month ethnography of the ride hailing service industry, the largest sector of the on-demand economy, this paper describes how the shift from human to algorithmic managers affects the nature of managerial control and worker autonomy. I begin by describing how algorithm-based control systems differ from prior control systems and conceptualize algorithmic work—a set of job-related activities that are structured by algorithms. In this context, algorithms manage by structuring choice at each human-algorithm interaction via nudges, to which drivers respond with a set of work tactics: compliance, deviance, or feigned acquiescence. While these tactics appear to be at odds, drivers frame their actions as evidence of their personal autonomy, in that the actions allow them to build a continuous stream of income and work from a discontinuous set of tasks. This contingent autonomy, or circumscribed choice, demonstrates that though algorithms may be externally viewed as an impersonal and, at times, unforgiving taskmaster, workers perceive otherwise, actively navigating their work environment. Further, by requesting and acquiring consent from workers at each stage of the work process, these algorithmic systems enact control without authority. This paper contributes to the understanding of control in organizations by describing a new type of control system and theorizing how algorithms can be a means of exercising control through enabling autonomy. Lastly, worker autonomy is re- conceptualized as a product of dynamic, human-algorithmic interactions as opposed to a static job feature. In sum, this paper provides insight into how work is structured and performed in the contemporary workplace.

William Attwood-Charles, Boston College
Culture and Collective Identity in Gig Work
Postmates and Favor are on-demand labor platforms that use mobile applications to match delivery requests with couriers, similar to the way Uber and Lyft match drivers with riders. Previous research on couriers has found that they often exist as tight-knit urban subcultures that are comprised of low-income bohemians who define themselves against the middle and professional classes. However, with the advent of the “gig economy,” the ranks of couriers have changed. The ease of registering to become a gig courier, combined with economic hardship and the de-stigmatization of service work via its association with new technologies and start-up culture, has expanded the population of couriers to include middle-class and even upper-class workers from diverse racial and ethnic groups. While the role of culture and collective identity formation is well-theorized in conventional service work, we know very little about the development of identities and culture amongst platform workers.  Some critical accounts of the gig economy argue that it isolates workers, diminishing their ability to develop a collective identity and culture, while more positive accounts emphasize the role of rating systems in eliminating the need for occupational communities in generating trust and deterring malfeasance. In this study, draw on data from 25 in-depth interviews and surveys with Postmates and Favor couriers to explore the question of collective identity formation and culture amongst gig workers. Are gig platforms purely technical enterprises that are empty of culture and collective identity? If not, what processes lead some gig workers and not others to develop a sense of collective identity and culture?

Avijit Sarkar, Mehrdad Koohikamali and James Pick, University of Redlands
Socioeconomic Analysis and Geographic Patterns of Host Participation in the Shared Accommodation Economy – Airbnb in Los Angeles, California
Short-term homesharing platforms, within the broader framework of collaborative consumption are rapidly transforming cities and metropolitan areas worldwide. There is very little existing research analyzing why suppliers and consumers participate in the sharing economy. In fact, factors that motivate suppliers to participate in the sharing economy are neither well understood, nor theoretically grounded. Suppliers or providers in the sharing economy are owners of unused or underused assets. In the context of peer-to-peer short-term homesharing, suppliers are owners who share their key asset –– residential properties, with guests –– usually strangers, with the homesharing platform (for example, Airbnb) as the intermediary. In doing so, suppliers are subject to different levels of control exerted by the sharing platform and engage in rivalry with peers to provide services to customers. Additionally, suppliers possess varying demographic and socioeconomic backgrounds spanning gender, age, race/ethnicity, household income, educational attainment, and attitudes towards trust and sustainable consumption. Additionally, where suppliers reside or where assets shared by them are located can be a key determining factor of supply capacity, which when combined with demand for the assets and the spatial and temporal origin of demand, immediately impacts pricing decisions. Therefore, studying the supply side of the sharing economy is imperative. What underlying factors motivate supply-side providers from diverse backgrounds to share their assets as part of collaborative consumption is as important a question as the motivation of demand-side consumers. This can assist sharing economy platforms better understand the needs, strengths, and challenges faced by providers and accordingly design training programs. Studying the supply side is also imperative to formulating policy and regulation that protect the rights of legitimate hosts engaged in lawful sharing, compared to rogue actors who abuse sharing economy platforms.

Mehmet Cansoy, Fairfield University
The Fault In The Stars: Public Reputation And The Reproduction Of Racial Inequality On Airbnb
Public reputation systems are central to the operation of the “sharing economy.” The platforms that have come to dominate this sector depend on the reputation systems as key features to facilitate exchange among strangers, by providing a measure of trustworthiness. Some existing research even suggests that such systems can be an effective way of reducing statistical discrimination against participants of color, by providing accessible and relevant information about the goods or services they offer (Ayres, Banaji, and Jolls 2015; Cui, Li, and Zhang 2016; Edelman and Luca 2014; Laouenan and Rathelot 2016; Nunley, Owens, and Howard 2011). However, the platforms have been far from egalitarian spaces (Cansoy and Schor 2018; Edelman, Luca, and Svirsky 2017; Ge et al. 2016) and in this paper I show that the public reputation system on Airbnb plays a crucial role in reproducing racial inequality. The analysis in this paper is based on data on Airbnb activity from 2015 and 2016, covering the 10 largest urban Airbnb markets in the USA, with about 276,000 individual units for rent that was available on the platform in this period. I use demographic information from the American Community Survey’s 2016 5-year data release on the census tracts the Airbnb units were located in to understand how public reputation information is affected by the racial composition of the surrounding area, as well as a range of other potential factors like income levels, housing values and educational attainment. I analyze the data in hierarchical models, with a range of controls including for the spatial autocorrelation of the dependent variables.
I identify two critical dynamics in the reputation system that place participants of color at a significant disadvantage on the short-term rental platform. First, it is harder for units located in areas with higher proportions of residents of color to generate reputation information about themselves. Among Airbnb listings that came on the platform during this two-year period, those located in areas with a higher percentage of residents of color were both less likely to receive an initial booking and were available to be booked for longer before receiving their first booking. I present substantial evidence that part of this dynamic can be directly attributed to the racial composition of the area, rather than real-estate characteristics, or socio-economic status. Thus, there is a consequential inequality in having access to reputation information, and whatever the merits of the existence of such information might be for reducing statistical discrimination, the burdens created by its lack fall more heavily on participants of color. On the other hand, even when participants of color are able to generate reputation information on the platform, this information itself suffers from racial bias. I find that among the listings that were able to obtain ratings on Airbnb during this period, being located in neighborhoods with higher concentrations of residents of color significantly increased the odds of a listing having a lower rating. Once again, a significant portion of this effect is directly attributable to the racial composition of the area, controlling for class differences and real estate characteristics.

Egan Research Center, snell 108

Jason Jackson, MIT
The Global Rise of Platform Firms in Urban Mobility Markets
This paper addresses one of the most pressing policy issues in contemporary market society: the controversial rise of the platform or ‘sharing’ economy and the implications for radical changes in industry organization, market structure and the future of work. It does so through analysis of the global emergence of digitally-enabled ridehailing platform firms in urban mobility markets. Ridehailing firms such as Uber, and their global counterparts such as Grab, Ola and Taxify exemplify the rise of the ‘gig economy’ through the use of digital technologies to reorganize work processes and reorder markets. Indeed, ‘Uberization’ has itself become a description of on-demand work, with commentators debating the implications across a range of outcomes from productivity to precarity. However, most of this discussion has focused on market contexts in the Global North, particularly the United States. Yet the structure of labor markets in the Global South presents important challenges to the dominant narrative. For example, many observers understandably lament the implications of the gig economy from the perspective of a Fordist past of high employment, and ‘good’ benefits. Yet in developing countries, high levels of unemployment and underemployment are the labor market norm and informality is pervasive. What are the implications for ‘Uberization’ in the South? This research explores this variation in effects and outcomes on markets and workers. It does so through comparative analysis of competing market structures and organizational forms in urban transportation markets, from ‘traditional’ taxis, motorcycles and rickshaws to ‘modern’ app-enabled mobility providers. The analysis contrasts the effects of the rise of ridehailing firms in industrialized cities such as New York with those on Bangkok, Dar es Salaam and Jakarta. The discourse around the rise of digital technologies suggests that powerful global corporations wielding sophisticated machine learning algorithms and ‘big data’ will radically re-shape urban mobility markets in their image. Our analysis challenges this standard prediction by showing how the effects of digital technology enabled firms varies across different social, political and institutional contexts.

Hongyao Ma, Fei Fang and David C. Parkes, Harvard University
Spatio-Temporal Pricing for Ridesharing Platforms
Many ridesharing platforms emphasize the importance of providing reliable transportation. For example, Uber’s mission is “to connect riders to reliable transportation, everywhere for everyone.”1 Whereas taxi systems have reliable pricing but often unreliable service, these platforms use dynamic “surge” pricing to guarantee rider wait times do not exceed a few minutes [5]. The platforms also provide the flexibility for drivers to drive on their own schedule, which significantly increases both driver surplus and driver supply [2]. As an example, Uber advertises itself as “work that put you first— drive when you want, earn what you need.” Despite having radically changed the way people get around in urban areas, there remain a number of problems with the pricing and dispatching rules governing the ridesharing platforms, leading in turn to various kinds of market failure, and undercutting the mission of providing reliable yet flexible transportation. A particular concern, is that trips may be mis-priced relative to each other, incentivizing drivers to strategize [3, 1]. For example, many platforms hide trip destinations from drivers before the pick-up. However, experienced drivers will call riders to ask about trip details, and cancel those that are not worthwhile [3]. Drivers also strategize in the following scenarios, where there is spatial imbalance and temporal variation of rider trip flows:
• (Spatial mis-pricing) When the price is substantially higher for trips that start in location A than an adjacent location B, drivers in location B that are close to the boundary will decline trips. This spatial mis-pricing leads to drivers’ “chasing the surge”— turning off a ridesharing app while relocating to another location.
• (Temporal mis-pricing) When large events such a sports game will soon end, drivers can anticipate that prices will increase in order to balance supply and demand. In this case, drivers will decline trips and even go off-line in order to wait in place.
• (Network externalities) The origin-based “surge pricing” used in practice does not correctly factor market conditions at the destination of a trip. As a result, drivers decline trips to destinations where the continuation payoffs are low, e.g. quiet suburbs with low prices and long wait times.

Kwong Chan, Northeastern University
Why Sharing Firms Don’t Always Share
Platforms are naturally embedded in the process of trust formation amongst buyers and sellers and in fact trust signaling by intermediaries is essential to overcome noisy trust signals that prevent efficient market function (Boerlijst, Nowak & Sigmund, 1997). Yet there are clear reasons for platforms to actively intervene in the process of trust between buyers and sellers.

1. The malfeasant rebels: Even if all buyers and sellers are nice to one another, new entrants to the system can take advantage of the pervading niceness, leading to previously nice actors altering their behavior such that the whole system stops assuming exchanges partners will act honestly. The result is that the market yields minimally overall utility (Imhof, Fudenber & Nowak 2007).

2. Group behavior: If multiple participants build trust, they may coordinate to extract benefits from other participants (Press Release: “University of Southampton team wins Prisoner’s Dilemma competition”, 07 October 2004), and consequently produce negative experiences for individual participants who are harmed by coordinated actions of groups.

3. Market bypass: Over repeated interactions buyers and sellers are able to build relationships that help them directly build trust under uncertainty (Williamson 1985) and motivate the actors to bypass the intermediary, cutting off a source of revenue for the platform. Even if the platform retains an intermediary position in the exchange, it will lose the capacity to match buyers and sellers in a manner that optimized overall market utility. Thus if buyers and sellers constantly demand specific exchange partners to increase their individual benefit, the overall efficiency of the system may be reduced as a result of lower overall resource availability to other potential partners.

Major Findings: An incentive therefore exists for firms to obscure credibility signals provided by buyers and sellers in order to prevent malfeasant behavior that harms nice participants and reduce the chance of the platform being cut out of future transactions. Yet obscuring credibility signals and creating subsequent social frictions are antithetical to concepts of sharing, social cohesion and community and may define a limit to truly socially-motivated exchanges when economic return is prioritized by an intermediary. The more successful firms in the sharing economy may in fact simultaneously encourage users to behave socially through word-of-mouth marketing and peer ratings, yet use this information to optimize business operations at the expense of participant choice in transaction partner, and channel of consumption. Participants may no longer be able to choose to connect with another participant without the platform, or access detailed information provided by other users of the platform. The role of trust in the sharing economy is thus double-edged. Users signal trust to improve the overall ability of the system to determine credibility, but the intermediary that collects the trust may hoard the information in a manner that reduces the utility of individual transactions for a many users. This may be implemented to stop participants bypassing the platform or to protect other users of the platform.

Social media companies have long wrestled with the role of moderating communications between users for safety reasons. The messaging platform WhatsApp for example actively restricts coordinated behavior in an attempt to protect some users (The Guardian, WhatsApp to restrict message forwarding after India mob lynchings, July 20, 2018) while Facebook regularly blocks user groups that it deems promote opinions that are harmful to others (NBC News, Civil rights groups call for tech firms to crack down on hate groups, Oct 25, 2018). It is clear platforms would be more prepared if they deploy active strategies for moderating signals between participants.


Egan Research Center, Room 406

Dafna Goor, Harvard Business School, and Amir Grinstein, Northeastern University
Social Interactions in the Sharing Economy: A Double Edge Sword?
The sharing economy – peer-to-peer platforms that drive collaborative consumption (Milanova and Maas 2017; Zervas, Proserpio, and Byers 2017) – represents a prevalent societal trend. This trend has led to the exponential growth of numerous businesses, like Airbnb and Uber, which attract significant attention from consumers, investors, regulators, and the popular press.
There are multiple motivations for consumers to participate in the sharing economy. A preliminary study (N = 148, Mturk) suggests that consumers’ economic and social motivations are highly associated with entering the sharing economy, while environmental and cultural motivations are only marginally important. Indeed, prior work mostly highlights the economic value as a key factor for the success of the sharing economy (e.g., Eckhardt and Bardhi 2015; Neoh, Chipulu, and Marshall 2017). Nevertheless, the sharing economy is embedded with social motivations and interactions (Bucher, Fieseler, Fleck, and Lutz 2018; Milanova and Maas 2017), and some product categories, such as shared dining, are associated more with social motivations than economic motivations (see preliminary study reported below). Companies also try to capitalize on these social drivers. An analysis of 79 advertisements of 10 leading sharing economy brands broadcasted between 2016-2018, found that the vast majority of advertisements build on social-related messages (68.4% emphasized people and 89.9% emphasized community).

The current research investigated the social dimension of the sharing economy, specifically the role of social interactions. Research on social interactions as well as very limited work on social interactions in the sharing economy context, offer two conflicting views. One view highlights benefits from meeting new people and feeling part of a community (e.g., Albinsson and Perera 2012; Möhlman 2015). A different view argues that a sharing setting may elicit social closeness that could be experienced as unpleasant and inauthentic (e.g., Bialski 2012; Bucher et al. 2018). It may also increase concerns regarding lack of privacy (e.g., Bucher et al. 2018; Milanova and Maas 2017). In this research, we study when and why social interactions in the sharing economy may backfire and have a negative influence on consumers’ well-being (e.g., enjoyment during the consumption experience), satisfaction (e.g., with the service provider), and engagement (e.g., recommending the experience to others). Specifically, we posit and demonstrate the following:
H1: Sharing (vs. traditional) economy experiences would decrease consumer well-being, satisfaction, and engagement.
H2: The negative influence of sharing (vs. a traditional) economy experiences on consumer well-being, satisfaction, and engagement is due to social awkwardness.
H3: Highlighting the social benefits in sharing economy experiences would attenuate the negative influence of sharing (vs. traditional) economy on consumer well-being, satisfaction, and engagement.

David R. Keith and Sergey Naumov, MIT
Is Sharing More Sustainable? New Product Sales During the Transition from Low- to High- Product Utilization
The sharing economy has grown rapidly in recent years, allowing consumers to use expensive and durable products without the burden of ownership. Prominent examples of sharing businesses include Zipcar, which rents cars parked in local neighborhoods to members by the hour, Rent the Runway, a fashion startup that allows women to rent designer clothing on a one- off or subscription basis, and Spinlister, a peer-to-peer marketplace facilitating the lending of sporting equipment such as snowboards and surfboards. Sharing businesses seek to create value by increasing the utilization of durable products, mutualizing access to these products for non- owners. Sharing has been viewed by some as a more democratized and resource-efficient alternative to the increasing trend of ‘hyper-consumption’ (Botsman & Rogers, 2010). Others are more skeptical, suggesting that these sharing businesses are merely a different form of capitalism enabled by lower transaction costs (Martin, 2016). Is the sharing economy more sustainable than conventional business models? Does the sharing of durable products reduce the rate at which new products are purchased, and resources consumed?

The growth of the sharing economy over the past decade can be attributed in no small part to the emergence of digital peer-to-peer platforms, enabled by the ubiquity of smart hand- held devices (Taylor, 2018).. The term ‘sharing economy’ has been the source of some controversy (Yglesias, 2013), however, because not all peer-to-peer platforms are pure sharing businesses, including many businesses specifically identified as being emblematic of the sharing economy. For example, Zipcar is a sharing business in that members forego owning their own private vehicle, using a smaller fleet of shared vehicles more intensively (the business model that is the focus of our paper). However, Zipcar also competes with conventional car rental companies to some extent, where it is less obvious that there is a net increase in vehicle utilization, even if Zipcar does have advantages, such as the ability to rent by the hour and the siting of cars in local neighborhoods.

For consumers, the sharing of expensive and durable assets can hold considerable appeal, allowing users to avoid buying products that they may only use occasionally, paying only for what they use (and when). The effect that sharing will have on societal resource consumption, and the firms that manufacturing these products, however, is less clear. When the founders of Rent the Runway approached designers to purchase dresses for the purpose of rental, most were wary, believing that it would cannibalize their existing sales (Schwartz, 2018). A reduction in sales is of course necessary if shared use is to result in the reduced consumption of materials. However, products that are used more intensively may wear out faster, and need to be replaced sooner, offsetting the effect of increased utilization to some extent. In the automotive industry, where the emergence of mobility-as-a-service is expected to lead to increased vehicle utilization in coming years, analysts’ estimate of the effect that shared use will have on the rate of new vehicle sales vary wildly, from a dramatic reduction in sales (Arbib & Seba, 2017), to no significant change (Bert, Collie, Gerrits, & Xu, 2016), and even a visible increase (PwC, 2017). Resolving this uncertainty is critical if shared use is to be effective sustainability tool.

In this paper, we develop an analytical model of the adoption of shared product use, showing that increasing the rate of product utilization has significant and unexpected consequences for the rate of new product sales. With higher utilization, the same number of customers can be served by a smaller inventory of products in use, by definition. However, if product lifetime is defined solely by use (for example, a device with a battery that can be recharged a finite number of times), the rate of new product sales is counter-intuitively independent of utilization in steady state, because the more intensively a product is used, the faster it wears out, increasing the rate of installed-base turnover proportionally. To the extent that this assumption holds, the belief that the ability to serve customers with a smaller product inventory will result in fewer sales is a correlation fallacy, where people fail to appreciate that products used more intensively will need to be replaced more quickly. If the product’s life is defined by age as well as use (for example, because a particular dress goes out of style after a period of time), increasing utilization does lead to fewer new product sales in steady state, because more uses are able to occur in fixed period. Important transitional dynamics exist also, because the total rate of new product sales is the sum of the declining rate of sales of the low utilization product, plus the increasing rate of sales of the high utilization product. A transitional dip or surge in sales can be observed, depending on relative strength of two factors: the rate at which used inventories of the low utilization product are repurchased, offsetting new sales; and the utilization multiplier of the new high-utilization product, which dictates how many units of the new high-utilization product are needed to satisfy adopters of shared-use. Finally, we demonstrate the application of this model to the future of new vehicle sales in the United States, estimating the rate of new vehicle sales under a range of plausible scenarios about the future. Taking factors such as population growth, vehicle utilization, and vehicle durability into account. we show that vehicle sales are likely to remain steady or increase in coming decades, with the potential for a temporary surge in sales if the build-up of a fleet of autonomous vehicles is required.

Our paper makes both theoretical and practical contributions. For the literatures on sustainability and sustainable operations, our analysis provides important boundary conditions on when sharing business models are likely to lead to a meaningful reduction of resources. For firms operating in these durable goods markets, our model provides a framework for making effective decisions about product portfolios and production capacity during periods of potentially turbulent change.

Tamar Makov, Yale University
Who wants my half eaten sandwich?- Food waste in the sharing economy
It is often claimed that the “sharing economy,” as implemented via networks of mobile apps and users, yields environmental benefits through the efficient redistribution of already-existing assets and resources. Yet, little is known about how these networks actually function and, indeed, whether they deliver on their promises. In this research, we aim to reveal insights into the nature and dynamics of the sharing economy through a deep dive into a real-world food sharing network that aims to reduce household food waste.

Each year, approximately 1.3 billion tons of food are not consumed, an extraordinary waste of embodied resources and energy (HLPE, 2014), not to mention the ethical travesty of wasting a full third of the global food harvest while food insecurity plagues both developing and developed courtiers (Coleman-Jensen, Rabbitt, Gregory, & Singh, 2016; Loopstra et al., 2015). Reducing food waste (defined as surplus food at the post retail or consumer stage) is therefore widely recognized as critical for improving resource efficiency and meeting the nutritional demand of a growing human population (Godfray et al., 2010; Poore & Nemecek, 2018; Shepon, Eshel, Noor, & Milo, 2018; Springmann et al., 2018; Tilman & Clark, 2014).

While food loss accrues throughout food system, in developed economies it incurs disproportionally at the post-retail and consumer levels (FAO, 2011, 2018). Since most of this food is edible when disposed of, redistributing edible yet unwanted food from primary to secondary consumers could yield substantial environmental and social benefits (Buzby, Hodan, & Hyman, 2014; Buzby & Hyman, 2012; Conrad et al., 2018; Vanham, Bouraoui, Leip, Grizzetti, & Bidoglio, 2015).