Essay

Share and Share Alike?

Considering Racial Discrimination in the Nascent Room-Sharing Economy

Michael Todisco *

Introduction

The motel’s neon vacancy sign lures the weary traveler from the freeway. She enters and inquires after a room. The receptionist, without looking up from his crossword, slides a clipboard across the front desk. An application? When did motels become prep schools?

The traveler scribbles in the requested information. Name: Monique Jackson. Reason you chose our motel: Sleep. She notices but ignores the advice that reads, “Giving the motel more information will make us more likely to accept your booking request.” Monique hands the application back to the receptionist, who takes out a worn Polaroid camera, snaps a quick picture of her, and tapes it to the application. He disappears behind the desk and quickly reemerges.

“No can do,” he says. “Owner tells me not to rent you a room.”

*    *    *

The above is a near-paradigmatic violation of the Civil Rights Act of 1964, which guarantees full and equal access to any place of public accommodation. 1 42 U.S.C. § 2000a (2013). Does it make any sense, then, that the substance of the above transaction is likely occurring hundreds of times each day online, with almost total impunity?

Airbnb and other housing-focused companies of the new “sharing economy” facilitate virtually unregulated discrimination—both implicit and intentional—in housing and accommodations. Before the host accepts or declines a guest’s request, Airbnb provides them with information about the prospective guest that serves as a heuristic for race. This layer of choice separates Airbnb from the traditional hotel model: armed with a guest’s racial information, hosts have a nearly unfettered ability to decline potential guests. With Airbnb’s exponential growth, this issue jeopardizes the integrity and efficacy of our civil rights laws.

I. The Discrimination Problem

Airbnb is an online marketplace for peer-to-peer rentals. 2 About Us, Airbnb, https://www.airbnb.com/about/about-us (last visited Mar. 10, 2015). Individuals can list any accommodation, whether a private room in a shared house, an entire dwelling, or something in between. 3 Id. Listed accommodations may be available on a night-by-night or more long-term basis. 4 Id. At the end of 2014, Airbnb had 925,000 listings and over 25 million customers. 5 PrivCo, Private Company Financial Report: Airbnb, Inc. 8 (2015).

Before accepting or denying any request, Airbnb hosts are furnished with the guest’s first name, often a picture, and other personal information. (Guests are encouraged by Airbnb to share information with their prospective hosts to increase their odds of acceptance.) While the racial salience of profile pictures is obvious, social science shows that race also can be inferred from a person’s first name 6 Marianne Bertrand & Sendhil Mullainathan, Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination, 94 Am. Econ. Rev. 991, 997 (2004). or her unique background and interests. 7 See Michal Kosinski et al., Private Traits and Attributes Are Predictable from Digital Records of Human Behavior, 110 PNAS 5802 (2013). To illustrate: “Emily from Marin visiting for a spa weekend” presents as racially different from “Imani from Oakland visiting her cousin.”

Quantitatively speaking, there is pervasive racial bias among Airbnb users. A recent Harvard study found significant bias against minority Airbnb hosts, but there is no reason why that bias would not also function against minority guests. 8 Benjamin Edelman & Michael Luca, Digital Discrimination: The Case of Airbnb.com (Harvard Bus. Sch., Working Paper No. 14-054, 2014), available at http://www.hbs.edu/faculty/Publication%20Files/14-054_e3c04a43-c0cf-4ed8-91bf-cb0ea4ba59c6.pdf. The study coded public profile pictures of all Airbnb hosts into racial categories and found that black hosts are paid twelve percent less for equivalent rentals than their nonblack counterparts. 9 Id. While the data revealed a bias among Airbnb users, raw numbers cannot demonstrate why the bias exists. The two principal explanations are intentional discrimination and implicit bias.

There is the very real possibility that hosts intentionally decline guests because of race, and Airbnb users have voiced such complaints. YouTube personality Tommy Sotomayor explained, “I got declined twice by the same person trying to rent a place, and if you look over their history they only rent to white people . . . . All they knew is [that] I was black.” 10 TommysCallToAction, @Airbnb Ask Tommy Sotomayor to Put His Video on Private Until They Can Make a Deal!, YouTube (Oct. 13, 2013), https://www.youtube.com/watch?v=0V5clzFs0Vo; see also Jorge Rivas, Airbnb Racism: Neither Unique nor Surprising, Fusion (Jan. 23, 2014), http://fusion.net/story/4689/airbnb-racism-neither-unique-nor-surprising. In support of his claim, Sotomayor showed that the properties remained vacant over the dates for which he was declined. 11 TommysCallToAction, supra note 10; see also Rivas, supra note 10.

Even if Airbnb users are not intentionally discriminating, the data suggest that they are falling prey to implicit biases. Again, social science provides an explanation: “White Americans, on average, show strong implicit preference for their own group and relative bias against African Americans.” 12 Nilanjana Dasgupta, Implicit Ingroup Favoritism, Outgroup Favoritism, and Their Behavioral Manifestations , 17 Soc. Just. Res. 143, 147 (2004) (citations omitted). This bias can be deeply subliminal: studies demonstrate different levels of brain activation when viewing different races. 13 Elizabeth A. Phelps & Laura A. Thomas, Race, Behavior, and the Brain: The Role of Neuroimaging in Understanding Complex Social Behaviors , 24 Pol. Psychol. 747, 750-52 (2003) (showing that white subjects experienced greater amygdala activation to black faces than to white faces).

II. Neither Commercial nor Intimate

Airbnb users find themselves in a soft spot of the law: somewhere between the commercial sphere, where discrimination is strictly prohibited, and the intimate-relationship sphere, where discrimination, even if socially reviled, is beyond governmental reach.

If Airbnb users are in the intimate-relationship sphere, their choices to discriminate are protected. The Supreme Court holds sacred the First Amendment right to free association. In Roberts v. U.S. Jaycees, the Court confirmed that “highly personal relationships [are entitled to] a substantial measure of sanctuary from unjustified interference by the State.” 14 468 U.S. 609, 618 (1984). The Court explained that it would protect relationships that mirror those between family members: relationships defined by relative smallness, a high degree of selectivity, and seclusion from others. 15 Id. at 619-20.

Airbnb users fall short of this standard. First, the guest-host relationship is neither selective nor small. Although Airbnb jealously guards its data, a subpoena from the New York State Attorney General has divulged some trends. 16 Research Dep’t & Internet Bureau, N.Y. State Office of the Att’y Gen., Airbnb in the City 4 (2014), available at http://www.ag.ny.gov/pdfs/Airbnb%20report.pdf. Hosts average around fourteen bookings a year—the equivalent of each host welcoming a new stranger into his home every single month. Moreover, this average is likely driven down by a glut of inactive users. High-volume hosts regularly host for more than 100 nights a year, with some booking for more than 365 nights a year (meaning those hosts inevitably hosted more than one guest on certain nights). 17 Id. at 8, 14.

Nor is the Airbnb relationship highly secluded. Guests often rent the entire dwelling, meaning that guests and hosts will never share a common space (and may never even meet). Even when the rental is for a single room in a shared home, the level of interaction—probably capped at a few hours—does not approach that of a family-type relationship.

More holistically, the Airbnb relationship fails the sniff test set forth by the Supreme Court. The Court explained that intimate relationships involve “deep attachments and commitments” and share “a special community of thoughts, experiences, and beliefs.” 18 Roberts, 468 U.S. at 619-20. Renting an apartment to a stranger from the Internet seems categorically different from the relationships envisioned in Jaycees.

III. The Legal Landscape

If Airbnb users are not in the intimate sphere, then the laws that govern the commercial housing sphere—the Civil Rights Act of 1964 and the Fair Housing Act (FHA)—will apply. 19 Section 1982 also probably applies. See 42 U.S.C. § 1982 (2013). Section 1982 has a somewhat narrower scope than the FHA but is largely duplicative with regard to racial discrimination. 93 Am. Jur. 3d Proof of Facts 415, § 5 (West 2015).

A.  The Civil Rights Act of 1964

Title II of the Civil Rights Act of 1964 prohibits racial discrimination in “any inn, hotel, motel, or other establishment which provides lodging to transient guests.” 20 42 U.S.C. § 2000a(b)(1). Airbnb properties would, at a minimum, fall under the broadly interpreted category of “other establishment” providing lodging. 21 For example, “other establishment” has been interpreted to cover YMCAs, United States v. YMCA, 310 F. Supp. 79, 82 (D.S.C. 1970), and trailer parks, Dean v. Ashling, 409 F.2d 754, 755-56 (5th Cir. 1969). Moreover, the legislative history of Title II supports a reading that would encompass Airbnb users. Congressional hearings recognized “the fact that a number of people open their homes to transient guests, often not as a regular business, but as a supplement to their income.” 22 Statutory History of the United States: Civil Rights 1194 (Bernard Schwartz ed., 1970) [hereinafter Statutory History].

Not every lodging establishment, however, is subject to the reach of Title II. Section 2000a(b) exempts from compliance any “establishment located within a building which contains not more than five rooms for rent or hire and which is actually occupied by the proprietor of such establishment as his residence” 23 42 U.S.C. § 2000a(b). A nearly identical exception exists in the FHA, discussed in Part III.B below. See 42 U.S.C. § 3603(b)(2). —the so-called “Mrs. Murphy exception.” 24 2 Statutory History, supra note 22, at 1194. In other words, Mrs. Murphy may refuse to rent to minorities so long as she actually lives in the building she rents out and there are fewer than five units in that building.

The application of the Mrs. Murphy exception to Airbnb users demonstrates the difficulty of jamming square-pegged new technologies into round-holed existing laws. Do Airbnb hosts “actually occupy” their residence? Hosts who rent out just a single room (and are staying in their units contemporaneously with their guests) would almost certainly qualify. But would a host who rents out her entire apartment still actually occupy it? While these hosts do not actually stay in their apartment for the night it is rented to their guest, many still generally reside in that apartment and call it home. Since Congress in 1964 did not consider the possibility that homeowners might rent out their bed one night and sleep in it the next, we are left to wonder.

Nevertheless, the purpose of § 2000a(b) is perhaps instructive. The animating rationale of the Mrs. Murphy exception is tied to the unsavory history of racial politics in the United States. Opposition to the Act painted a portrait of “the ancient widow operating a three or four room tourist home who would, by force of the bill, be required to accommodate transients without regard to race.” 25 Harry T. Quick, Note, Public Accommodations: A Justification of Title II of the Civil Rights Act of 1964, 16 W. Res. L. Rev. 660, 672 (1965). If § 2000a(b) protects Mrs. Murphy from the noninvasive tasks of a landlord (i.e., occasionally receiving rent payments from her tenant), then the exception would likely protect Airbnb hosts from the more intimate consequences of an Airbnb rental (i.e., sharing their bed, bathroom, cutlery, etc.).

But consider the policy implications of allowing Airbnb hosts to use this exception. Given the proliferation of mega-hotels and the relative decline of the mom-and-pop landlord, § 2000a(b) is now a provision of “limited practical significance.” 26 James D. Walsh, Note, Reaching Mrs. Murphy: A Call for Repeal of the Mrs. Murphy Exemption to the Fair Housing Act, 34 Harv. C.R.-C.L. L. Rev. 605, 606 (1999). This extension would breathe new life into an unfortunate provision that looked moribund. Airbnb has over 900,000 units—a number that more than doubled in 2014. With just one more year of growth at the 2014 rate, this extension would result in millions of new Mrs. Murphy rooms—nearly the equivalent impact of allowing the three largest hotel chains in the world to discriminate at will. 27 See Zainab Mudallal, Airbnb Will Soon Be Booking More Rooms than the World’s Largest Hotel Chains, Quartz (Jan. 21, 2015), http://qz.com/329735.

B.  The Fair Housing Act

The FHA 28 42 U.S.C. §§ 3601-3619. prohibits discrimination in the rental of “dwellings”—generally any place of residence meant to last more than a few weeks. 29 See, e.g., Lakeside Resort Enters., LP v. Bd. of Supervisors, 455 F.3d 154, 160 (3d Cir. 2006) (holding that a facility whose residents would generally stay for about two weeks was a “dwelling”). Because Airbnb also serves as a significant marketplace for long-term sublets, 30 See Monthly Stays Made Easy with Airbnb Sublets, Airbnb, https://www.airbnb.com/sublets (last visited Mar. 10, 2015). the FHA applies.

Airbnb users could fairly easily establish a prima facie FHA claim. To do so, a plaintiff must show: (1) she is a member of a protected class; (2) the defendant knew of her class membership; (3) she was prepared to rent the property; and (4) the defendant refused to deal with her. 31 Hamilton v. Svatik, 779 F.2d 383, 387 (7th Cir. 1985). Tommy Sotomayor’s story discussed above, for example, would seemingly meet this standard.

Plaintiffs prove an FHA violation by showing either disparate treatment or disparate impact. Proving disparate treatment requires some proof of discriminatory intent, 32 Id. which often comes from a landlord’s racially infused explanation for denial; proof of disparate treatment thus inherently varies on a case-by-case basis. Because Airbnb hosts can write explanatory notes to rejected guests, a racially charged justification might meet this standard.

Establishing a prima facie FHA violation, however, does not necessarily require proof of intent if disparate impact can be shown. Even a facially race-neutral policy can violate the FHA if it “actually or predictably results in discrimination.” 33 Gashi v. Grubb & Ellis Prop. Mgmt. Servs., Inc., 801 F. Supp. 2d 12, 16 (D. Conn. 2011). This disparate impact is generally established by using statistical evidence to contrast the affected and unaffected populations. 34 Id.

Disparate impact on Airbnb could be statistically proved. First, code guests into racial categories using profile pictures and first names. Next, examine the relative response and acceptance rates for racially diverse and nondiverse guests. This method could be used to establish prejudice at any level—whether by individual host, city, or country.

In addition to disparate treatment and disparate impact, the FHA provides a cause of action for advertisements that indicate racial preference. 35 42 U.S.C. § 3604(c) (2013). This provision applies irrespective of the advertiser’s intent. 36 Ragin v. N.Y. Times Co., 923 F.2d 995, 999-1000 (2d Cir. 1991). In Ragin v. New York Times Co., the Second Circuit held that the FHA prohibits any advertisement that “would discourage an ordinary reader of a particular race from answering it.” 37 Id. The Ragin court found that a company’s repeated use of all-white models in its advertisements established a prima facie case. 38 Id.

The parallels between Ragin and Airbnb host profiles are tantalizing. Host profiles display pictures of and reviews from each of the host’s past guests. If a host successfully rents her unit to only white users, her profile will be a de facto Ragin advertisement—subtly but powerfully communicating a hostile message to minority guests.

IV. The Solutions

Under current law, Airbnb is not liable for FHA or Title II violations that occur over its website. While vicarious liability is the norm for housing claims, 39 Homeowners, landlords, and real estate companies, for example, are liable under the FHA for the discriminatory actions of their employees or agents “in accordance with traditional agency principles,” meaning that the FHA “normally imposes vicarious liability upon the corporation but not upon its officers or owners.” Meyer v. Holley, 537 U.S. 280, 282 (2003). the Communications Decency Act (CDA) exempts websites from liability for user-generated content. 40 47 U.S.C. § 230(c) (2013). For example, the Seventh Circuit held that the CDA exempted Craigslist from any FHA claims arising from its users’ discriminatory posts. 41 Chi. Lawyers’ Comm. for Civil Rights Under Law, Inc. v. Craigslist, Inc., 519 F.3d 666, 670-72 (7th Cir. 2008); cf. Fair Hous. Council v. Roommates.com, LLC, 521 F.3d 1157, 1175 (9th Cir. 2008) (en banc) (declining to extend CDA immunity because the website required users to choose from potentially discriminatory options in creating user profiles). With a suit against Airbnb out of the question, at least three alternative solutions exist.

The first is a legal solution: operate within the current paradigm and bring enforcement actions against individual Airbnb users. The deterrence effect from successful suits could discourage future acts of discrimination. 42 The Seventh Circuit suggested this method in Craigslist. 519 F.3d at 672. However, this method seems inefficient, labor intensive, and costly. The National Fair Housing Alliance (NFHA), a consortium of 220 nonprofits dedicated to ending housing discrimination, pursued this route, reporting in May 2009 that it had filed over 1000 complaints against discriminatory online postings in the past year alone. 43 Nat’l Fair Hous. Alliance, Fair Housing Enforcement: Time for a Change 32 (2009), available at http://www.nationalfairhousing.org/LinkClick.aspx?fileticket=dsT4nl HikhQ%3D. Given the persistence of online discrimination in the face of this deluge of suits, the NFHA concluded “that pursuing complaints against the thousands of discriminatory advertisers who use the internet” was not feasible. 44 Id.

The second is a political solution: lobby Congress to change the CDA to exclude housing violations. This carve-out would not be as radical as it may seem, as the CDA already excludes certain types of activity from its protection. 45 47 U.S.C. § 230(e)(1), (2), (4). For example, websites are liable for user-generated intellectual property violations. 46 Id. § 230(e)(2). Anyone who has used YouTube has felt the effectiveness of this provision: the site fastidiously removes pirated content. 47 See Google, How Google Fights Piracy 9-12 (2013), available at https://docs.google.com/file/d/0BwxyRPFduTN2dVFqYml5UENUeUE. Google is YouTube’s parent company. Without liability, Airbnb currently has little incentive to fix its discrimination problem. Were the CDA amended, Airbnb would be forced to find internal solutions.

The third is a social solution: pressure Airbnb to evolve. Airbnb is an image-conscious company and might respond to public momentum. 48 For example, Airbnb swiftly made a public comment denying the findings of the Harvard study discussed above in note 8. See Liz Gannes, Airbnb Design May Lead Black Hosts to Charge Less, Say Researchers, Re/code (Jan. 17, 2014, 3:45 PM PST), http://recode.net/2014/01/17/airbnb-design-may-lead-black-hosts-to-charge-less-say-researchers. First, Airbnb could wipe the guest’s name and picture from any booking requests (and only reveal that content after the booking is made). While profile pictures and names can establish trust in an online community, it is hard to imagine that such things are strictly necessary, particularly if other information—such as the guest’s reviews by previous hosts—were made available. Alternatively (and perhaps in a manner that better preserves host autonomy), Airbnb could run statistical analyses of its hosts’ behavior toward racially diverse guests. For hosts that appear to have a statistically significant aversion to minorities, Airbnb could use “testers” to further investigate. Testers are pairs of individuals of different races—often one white and one black—who are virtually identical in all other significant aspects (i.e., age, credit score, income). 49 This method is commonly used by the Department of Housing and Urban Development. See, e.g., Office of Policy Research & Dev., U.S. Dep’t of Hous. & Urban Dev., Housing Discrimination Against Racial and Ethnic Minorities 2012, at xii, 5-6 (2012), available at http://www.huduser.org/portal/Publications/pdf/HUD-514_HDS2012.pdf. If a host denies the nonwhite tester and then accepts the white counterpart, Airbnb could demand a racially neutral explanation from that host and, absent a satisfactory explanation, ban or suspend him from the site.

Conclusion

The sharing economy is here to stay. Companies such as Airbnb and Uber are massive and only getting bigger. Airbnb grew 69% in 2014; 50 PrivCo, supra note 5, at 6. Uber, 155%. 51 PrivCo, Private Company Financial Report: Uber Technologies, Inc. 6 (2015). Moreover, the scope of the sharing economy is limited only by the human imagination. New companies are born every day—ranging in kind from peer-to-peer lending to shared Wi-Fi access to crowd-sourced dog-sitting services. 52 See How Peer Lending Works, Lending Club, https://www.lendingclub.com/public/how-peer-lending-works.action (last visited Mar. 10, 2015); What Is Fon?, Fon, https://corp.fon.com/en (last visited Mar. 10, 2015); How It Works, DogVacay, http://dogvacay.com/how-it-works (last visited Mar. 10, 2015).

This new economy has the potential to exponentially increase access to once unaffordable or unattainable goods and services. Or it can perpetuate the ugliest of human tendencies. Let’s proceed thoughtfully.

Michael Todisco is a J.D. Candidate at Stanford Law School, Class of 2016. He thanks the editors at the Stanford Law Review.