Reducing Discrimination with More Information

From Cui, Li and Zhang:

We conduct four randomized field experiments among 1,801 hosts on Airbnb by creating fictitious guest accounts and sending accommodation requests to them. We find that requests from guests with African American-sounding names are 19.2 percentage points less likely to be accepted than those with white-sounding names. However, a positive review posted on a guest’s page significantly reduces discrimination: When guest accounts receive a positive review, the acceptance rates of guest accounts with white-sounding and African American-sounding names are statistically indistinguishable.

In other words, taste based discrimination is weak but statistical discrimination is common. Statistical discrimination happens when legitimate demands for trust are frustrated by too little information. Statistical discrimination is a second-best solution to a problem of trust that both owners/sellers/employers and renters/buyers/workers want to solve. Unfortunately, many people try to solve statistical discrimination problems as if they were problems of invidious prejudice.

If you think the problem is invidious prejudice, it’s natural to try to punish and prevent with penalties and bans. Information bans and penalties, however, often have negative and unintended consequences. Airbnb, for example, chose to hide guest photos until after the booking. But this doesn’t address the real demands of owners for trust. As a result, owners may start to discriminate based on other cues such as names. Instead market designers and regulators should approach issues of discrimination by looking for ways to increase mutually profitable exchanges. From this perspective, providing more information is often the better approach. As Cui, Li, and Zhang write in a HBR op-ed:

Our recommendation is for the platform companies to build a credible, easy-to-use online reputation and communication system. Bringing information to light, rather than trying to hide it from users, is more likely to be a successful approach to tackling discrimination in the sharing economy.

Addendum: See also Tyler and I in The End of Asymmetric Information. We need to work with information abundance rather than try to push against the tide.

Comments

Taste-based discrimination exists too though, for example all the orchestras that started taking more women and minorities after they made auditioners play behind a screen.

Ideally you’d both maximize relevant information like other users’ reviews, and minimize irrelevant information that give away racial cues like photos and names.

Bend that dick curve!

It doesn't always work out that way though.
"A measure aimed at boosting female employment in the workforce may actually be making it worse, a major study has found."

https://www.abc.net.au/news/2017-06-30/bilnd-recruitment-trial-to-improve-gender-equality-failing-study/8664888

You are conflating a structural disincentivation toward discrimination as a heuristic for decision making and equality of outcome.

What is your prior?

"A measure aimed at boosting female employment in the workforce"

You do understand that that effort is discriminatory?

"Notice that only the Fred's relatives get promoted?"
"Yeah, but trying to address that would be discrimination against the Fred's relatives."

If there is an individual scoring/rating system that is accurate and earns people's trust, then people want to use that and don't care about ethnicity/religion, and you don't have to hide that info or trick people.

Taste-based discrimination exists too though, for example all the orchestras that started taking more women and minorities after they made auditioners play behind a screen.

There's a Kennedy School study which attributes a portion of the increase in the number of women in such orchestras to blind auditions.

N.B. there are 20,000 musicians employed by 'performing arts companies'. Of the 25 leading performing arts companies, about 3 trade in concert and chamber music. This is a niche occupation and selection is subjective.

I don't know about minorities, but the claim that blind auditions increased the percentage of female musicians hired comes from Goldin & Rouse (2000). They claim that blind auditions explain a quarter of the increase of female representation in American symphony orchestras in the latter half of the last century. However, this estimate is very imprecise and not only its significance but also its sign is dependent on what covariates are included. For example, the raw effect of the introduction of blind auditions was negative on women in their model, with fewer getting selected in the first round of audition. Only after including certain covariates did the positive effect emerge. Also, in auditions with semifinal rounds, blinding had no effect.

There's a large difference between the actual study, with its complex mess of assumptions and fragile effects, and the conventional wisdom that it showed the efficacy of blind auditions in decreasing discrimination.

I suppose this is important information to find out... but I can tell you my wife would be pissed if she learned that she had spent time and effort responding to a fake guest request on her rental. The Laputians should consider the effect of their experiments on those below.

It's totally ok to stick it to business as long as it results in yet-another paper saying that Racism Is Bad.

They needed a control group to separate the effect of perceived class (and known associated anti-social behavior) based on name. Maybe some low-class white names like Cletus and Bubba? Without that control group, the effect of class has not been accounted for.

A reasonable point, but not one that is going to be explored here any time soon.

Middle name "Wayne" might be worth testing as well.

'Statistical discrimination happens when legitimate demands for trust are frustrated by too little information.'

Well, apart from the fact that when the same information was apparently offered, 'African American-sounding names are 19.2 percentage points less likely to be accepted than those with white-sounding names.'

Almost as if white people are considered more trustworthy/desirable/etc regardless of the lack of information - apart from the apparent color of their skin.

Strange how that works, regardless of how many times someone attempts to avoid the actual mechanism involved - which is making a distinction based on skin color (using a name as a proxy), and only being reassured with additional information that is not required in the case of a white person. As demonstrated by the 19.2% rate.

'Statistical discrimination is a second-best solution to a problem of trust'

Yeah, that must be it, which is why when presented with the same lack of information, white sounding names are favored.

'many people try to solve statistical discrimination problems as if they were problems of invidious prejudice'

Today's new and improved prejudice is only 19.2% invidious, right?

US homicide rates per 100,000:
White (non Hispanic): ~3.5
Black: ~20

I don't know exactly how much information is available in an Airbnb booking but that is a bit of a problematic prior to overcome.

You are missing the point: to take your analogy, yes, some ethnic/racial groups are more violent than others. Some individuals are more violent than average, others less violent. It's more fair to judge individuals by their individual behavior than to judge them by ethnic group averages, even if those are accurate.

the sociology study is getting real people to judge imaginary people
based on a squisky metric (name) in a contrived study that doesn't control for a lotta other important variables.

I don't know that I am the one missing the point. The conclusion here is people are better off with more information individual information so they can decide on factors other than (their gut-feel grasp of) demographic statistics.

Let me see if I have this straight - when a 49 year old woman with an African-American sounding name is looking to book a place to stay overnight through AirBnB, the disparity in murder rates is the reason for her being 19.2% less likely to be accepted.

Right.

It didn't break down effects for gender and age as well, did it?

But for what it is worth, I bet single female guests have a higher take up than males. For what it is worth, I'd rank Female-Middle Age-Black above Male-Young-White for trust in room rental, in the absence of more information.

+1 And this follows my experiences as a landlord. Females and older people have better outcomes for the landlord, outweighing any racial component.

the murder rate in st. louis is 66/100,000
much higher than baltimore or chicago
average u.s. murder rate is about 5-6/100,000

so "white-sounding and African American-sounding names" can be so precisely defined ... that these researchers can thus objectively detect racial discrimination to a tenth-percent accuracy (?)

B.S. "research"

I bet a much-better-than-guess rate can be achieved.

Also, names are class and age as well as race signifiers. That 19% could be confounded with class and youth as well as race effects.

Indeed. In fact, this is just my own rough heuristic, but there seem to be a lot of low class "black names" that are very easy to distinguish as "black". Upper class black names are often different then lower class black names.

Same for whites. As they noted in "Ted" there are certain kings of names that are very "white trash".

Do you know any white people named Lashondra? Does the social security office have any statistics on this?

I wonder what the difference in rental acceptance would be if the same exact african-american woman was called Lashondra versus Tracy? Of course, for that experiment the decision maker must know the woman's appearance. Does a person's name give you any information about probable upbringing?

In Utah, yes.

I met a white woman named LaDawn a few years ago. It was a surprise, for sure.

+1. "Black-sounding" doesn't tell you very much. A lot of these names are typical of what Thomas Sowell has called "ghetto blacks", as opposed to middle and upper class blacks (and represents a point he has made in his books, that activists try to represent ghetto black culture as common to all black people). There aren't quite as many white people with poor-sounding names, but they do exist: Billy Ray, Amber, and the like. They don't tend to get hired or invited either.

Assuming the original study is valid, I wonder why people just don't just change their legal names. They can always keep their original one as a nickname.

Well, the Amber I know is going to be very disappointed to hear her name signifies white trash. Or maybe not, as her family will undoubtedly be able to pay for whatever education she wishes pretty much from petty cash.

Does her sister Megan need to worry about her name?

Hotels work without any easy-to-use online reputation system aimed at guests. Hotels just hold some amount on credit or debit cards in case you damage something.

I think the problem here is poor people being distrustful of other poor people. Who wants to deal with the inconvenience of hosting strange people at home for a couple hundred dollars? AirBnb has the potential to include racists among their hosts because it's the people who has troubles to make ends meet, people living under economic stress.

No. Hotels can handle the hassle of bad guests because of they spread the risk sufficiently over scores of rooms with a well-developed insurance market.

For an individual, a single bad guest is much, much, more of a risk, even if the money is ultimately "recoverable", the hassle is immense. This is why people will pay for trust.

Nobody at a hotel is too upset if a single room is trashed. But if you come back from a weekend to find your AirBnB guest has trashed your primary residence....well....

Well, Axa, if it is racism, why do they stop being racist when they learn their black-named guest is a Systems Engineer from Nairobi and not a crack dealer from South Chicago? Very strange racists.

But hey, go with "poor, distrustful, racist idiots" as your explanation if you want.

'why do they stop being racist when they learn their black-named guest is a Systems Engineer from Nairobi and not a crack dealer from South Chicago'

Did they control for profession and nationality, even if the study did not break down effects for gender and age, apparently?

And considering how certain Americans treated a president they considered to be a Kenyan, it is unreasonable to assume that the sort of birther that believes Obama was a Kenyan is more likely to rent to a Kenyan systems engineer.

It's been 12 years since Hillary started that crap. Leave her alone!

Indeed, hotels can hedge the risk in a easier way.

When the host reads the "black-sounding name" and starts to think about some personal characteristics like "could be a crack dealer".......well, that's racism. As you said, individuals (AirBnB hosts) face so little cases than averages are irrelevant. If the person changes opinion after the guest says "I'm a systems engineer", that only means the host does not support racial segregation.

The "poor distrustful racist idiots" idea is based on the observation of some family members working on hotels/restaurants. As employees they have to follow the company's policy on how to treat guests. As private individuals they're a bit rougher.

The curious thing is that AirBnB drives you to meet people at their most personal space: their home. I'll take the hypocrisy of a professional setting (hotel) any time over honesty (AirBnB).

PS. AirBnB every once in a while publishes some feel-good story on how their business model helps low income people to have extra-income. So, it's AirBnB who sells the story about poor hosts, not me.

Do you have any idea how much Blackman cum I need to clean up after my wife and her boyfriend get done? The fact that he pays me some rent sometimes does not come close to compensating

I'm thrilled to have my own troll. I feel like I've made it.

I feel like a Cuuuuuuuck!

Because Americans passively deferred to corrupt and corrupting Media Establishment accounts over the past decades of Boomer ascendancy, "discrimination" for Americans now means all but exclusively "prejudiced or prejudicial outlook, action, or treatment".

Focusing all but exclusively on this usage in public discourse, Americans have all but lost the "other" sense of "discrimination", to wit: "the process by which two stimuli differing in some aspect are responded to differently; differentiation; the quality or power of finely distinguishing".

We have permitted our corrupt and corrupting Media Establishment to teach us not to observe observable distinctions and to treat "discrimination" across the board not as the power to observe distinctions but as the signal to pretend as if differences and distinctions do not exist, so enamored have we been taught to become with idealistic and illusory "equality".

The reluctance or refusal to observe observable distinctions hardly renders egalitarian conceits "true" in any sense of the word.

When might we surrender our aspirational politics of worthless idealism and the pestilential lie of egalitarianism in order to face the world not as we might prefer it to be but a bit more as it in fact is?

The sign and seal of egalitarianism remains DEATH--hardly the basis for any social or political scheme for the living unless you want to insist on using necrological categories for any and all events transpiring this side of our graves: which might in fact work, given the psychic putrefaction our Cognitive Elites are pleased to treat us to most days.

"The reluctance or refusal to observe observable distinctions hardly renders egalitarian conceits "true" in any sense of the word."

I'm currently reading Life at the Bottom, a set of Dalrymple essays from about 30 year ago, primarily about the white English underclass.

He observes that he can reliably identify by appearance and demeanor men with a high likely hood to engage in domestic partner abuse, but that the nurses in his ward, who may themselves become victims, refuse to admit those markers because they are, in today's terms, politically incorrect.

Its worth a read.

Not surprised. Humans are extraordinarily good at picking up all sort of markers for "violent", "stupid", "defector", "low conscientiousness", "high time discount", "liar" etc. Such...discernment is a hugely selected-for trait.

Of course, like a lot of biology, this turns out to be a hate-fact. Who knew?

Some people don’t want to be like me getting rammed up the ass all day. It’s a sign of how far our society has fallen. I worked and studied hard to get to this position.

Do you have a citation for that? Just how good ARE humans at determining which people are vigilant criminals just by looking at them? I'mcurious. .

It was a type of silent suffering that had become accustomed, rationalized through comedy. The appearance of men, women, and children so tightly packed that you couldn’t help but wonder upon the giant monster, that at least three times out of ten you were either crammed in someone’s armpit, sat next to a homeless person, or got squeezed by the doors, then went back up the platform and took a cab instead.

By “giant monster” I assume you are referring to the average black penis right?

a lobster !

Apparently you can tell if someone is a murderer just by looking at them!
I must be wierd, because when I look at a black person, I see little which is objectively observable beyond skin color.

Delightful Freudian slip from the resident SAHM. “Look at black person” not meets not talks to-looks at. Very little listening or interacting with black folks eh. Just a few peeks now and then when you are especially bored with the husband.

I guess because whenever YOU look at a black man your eyes are covered by cum.

Only if I’ve just raw dogged his girlfriend. One of the draw backs of an over-sized dick is I can’t always control the spray.

Were I to follow Media Establishment guidance, I guess I'd follow the recent Netflix example and walk through life blindfolded, probably arming myself along the way with the consolations that all people are persons of good will and humane courtesy, that each of us is so wonderfully and purely empathetic with all of humanity that we're willing to think for successive moments that we have transcended our own respective subjective conditions all for the charm of entertaining idealistic bliss! What rapture!

Empiricist that sometimes I am, though, I watched the Netflix offering blindfolded with the volume off: I can tell you I derived no idealistic infatuations as a consequence.

Or you could walk around reserving judgement about things and holding as certain only the things you can actually observe with some measure of objectivity.

--which, foregoing Media Establishment guidance, I do after my fashion: I discriminate (observe observable distinctions) with discrimination (whatever measure of fine judgment I might claim legitimately).

I don't trouble myself with "objectivity" much since no one else seems to approach it too closely (it is said that the original Potemkin villages--those that antedated our beloved Internet--enjoyed objective status also, insofar as they were subject to observation: yet just how any Potemkin village capable of being observed corresponds to "objective reality" launches a new topic which I might prefer to address in some tale of speculative fiction).

¿por qué leer la nueva york times?
para los titulares
"Smollett Timeline: Mystery Deepens as Police Review the case"
doesn't it actually look like the "mystery"
has gotta lotta less "deep"

Fake hate crimes are valid. They advance the agenda.

¿por qué amamos a CNN?
na los titulares
"new Orleans police shootout leaves five bystanders shot "

which effectively leaves out the part about the armed robbery suspect who started the whole mess

it is starting to look like what the sociology dept. calls
assymetrical information is actually deliberate misinformation

got it
socioloigists
interpret fake hate crimes
symbolically not literally?

I must admit, I had a wobbly moment with this one. Now, I thought it was weird on first hearing it (MAGA country? Hats? How did the attackers possibly recognise him? Why no notable injuries or medical report?) but none of the media remarked on it or seemed to show any scepticism whatsoever. Was it just me?

"Maybe," I thought "Maybe it was a real attack. There must be some, after all. These incongruous details may be incomplete or incorrectly reported".

Hahahahaha. With this and Covington, its starting to look like the MSM can't print the fake news fast enough.

... is it OK for AirBnB 'customers' to racially discriminate in their choices of AirBnB hosts/owners ?

Why the double standard ?

Asymmetric information? The AirBnB owner knows very little and the applicant. The applicant know something salient about the AirBnB that’s hard to lie about: its location. From location a lot can be gleaned.

"See ... I": I hope your students protest at this sort of barbarism.

I repeat a tale. We have several times had a house to let in a college town. After canvassing opinions among people with experience we decided we would not let to any student or teacher of Law, Economics, or Business Studies.

The tenants we ended up with were archaeologists, engineers, and vets. As tenants they varied from good to excellent.

I see no compelling reason to abandon this discrimination.

I should add - we discriminated on one other ground: no smokers.

Does anyone here object to that vile act of discrimination?

The state of Virginia, and the DC, send out mystery tenants in a sting operation it is rumored to see if landlords discriminate based on color. There's a hefty fine if they do. We don't discriminate since our properties are largely slums that are easy to rent, mostly to people who have marginal jobs and can barely afford the security deposit. And they're cash cows.

Bonus trivia: amazing how the "right" laissez faire economists now doesn't even mention the traditional argument that discrimination is impossible in a free market: the people who are discriminated against, if they are truly not a problem, will be catered to by savvy business people who want to make a nice, risk-free profit. After all, if Regina Jones is truly not a burden on your rental, why not advertise "we don't discriminate based on color, all applicants welcome"? Because Regina, though a nice woman, is likely to bring in 20 of her closest friends and relatives after you rent the place "just for her". And they'll throw a party. With rappers. With party favors. With complaining neighbors. With calls to the police. But it's all good!

If "archaeologists, engineers, and vet [students]" are whiter than "Law, Economics, or Business Studies," you might get into trouble. Disparate impact and all that.

Fake edit: minorities probably smoke at higher rates, too. So that one also iffy.

Sorry, OC, but this humourless prick occasionally pretends to be me.

>"Unfortunately, many people try to solve statistical discrimination problems...."

... with the other facts that they have on hand, even if those facts are not necessarily 100% relevant.

Gee. What a shocker!

I feel like people don't understand the basics of discrimination. Above we see Edward Burke defend discrimination as merely the observation of difference. Clamence cites racial disparities in murder rates. Note also that artificial intelligence readily engages in racial discrimination...because it works!

But here is the deal. As a society we have decided that racial discrimination is not acceptable because it is not fair. White folks might be statically less likely to cause trouble with an AirBnb rental, but this has absolutely nothing to do with whether a particular person, say Lashondra Davis will be a good customer. It is simply not fair to Lashondra to assume she matches the statistical profile of all black Americans. She could easily be a great customer, and furthermore, she didn't choose a skin color and there is no way for her to change it.

Racial discrimination is unfair even if it is logical, effective and profitable. Ethics demands that AirBnb hosts do not discriminate based on features over which individuals have no control. Same goes for artificial intelligence. The effectiveness of AI needs to be actively reduced in order to ensure fairness to individuals.

Re: the non-acceptability of making decisions based on race - Ivy League admissions offices, and I believe every single Democrat running for President, don’t seem to agree.

It is ethically unfair to cause harm to individuals by disallowing discrimination.

You rate the harm done to those that most accommodate those they wish not to as inferior to the harm caused to those that have to deal with discrimination. I reject this moral judgement.

I was taught growing up that discrimination was based on ignorance and that it made the world a worse place. Now we learn that it's based on FACTs and that it makes the world a better place. And we are told that only ethical thing to do is to make our lives and the world overall a worse place. That's not how this was sold.

People should not have to put up with their house being trashed because Lashondra wants to rent a room. Respectable people that want to rent a room should not be locked out of renting a room because a landlord has withdrawn from the market since he can't use discrimination to effectively protect himself from negative outcomes.

>You rate the harm done to those that most accommodate those they wish not to as inferior to the harm caused to those that have to deal with discrimination. I reject this moral judgement.

It has nothing to do with me or how I rate relative harm. We, as a society, decided that racial discrimination was wrong--decades ago. This was codified into the law and it is (and has been) simply illegal for a hotelier to deny rentals based on race. I was just providing the theoretical background for those that have forgotten or never knew.

You are free to disagree of course, but I don't think you are going to have much luck moving society back to 1950. It is widely accepted that racial discrimination is wrong, across the political spectrum.

Those laws were passed based on certain assumptions about WHY discrimination exists. As those assumptions have proven false, even those that once supported them have turned against them. My father marched for civil rights, but he's against affirmative action and school busing.

School busing raises another point. That was supposed to be an unmovable blank of discrimination, and yet it was immensely unpopular and overturned basically everywhere. So what is with this "we decided racial discrimination is wrong" stuff. Progressives claim "soft segregation" in public school districting is discrimination, and yet everyone is massively opposed to redistricting if it involves their district.

The entire "disparate impact" notion of discrimination is still part of "law" but wildly disagreed with.

The attempt to extend civil rights law to "make them bake the cake" is also widely unpopular. It's almost like people who had some sympathy for the trials and tribulation of blacks in the Jim Crow south don't actually think that gays that want to chase down and ruin christian bakers face the same kind fo troubles they did.

I don't think "we decided" much of anything. I think some people shoved something down other peoples throats, and in fact they actually try not to live by their own rules whoever they can get away with it.

I didn't say anything about school busing, affirmative action, disparate impact, or gay wedding cakes.

The AI should be improved to seek out and incorporate additional information so as to distinguish between "Lashonda" and "average statistical profile".

Well, sure, but you still have to explicitly and actively exclude those factors which we, as a society, have decided are not fair...AI doesn't know anything about fairness, data is data and all it cares about is using the data to get the best score. So if you give it racial data, it will use it, unfairly.

That's reasonable. Just don't include race along with all of the other data available, including credit history, etc. Make the AI colorblind. Is some statistical disparity falls out of that because the AI happens to (on average) rate black people less trustworthy based on objective factors like credit history and criminal records, despite not using race an an input, then that should be totally ok. The reasons why black people end up with worse credit history and worse criminal records are a separate issue. As long as the AI isn't actually judging people individually, that's a fair outcome.

I think this is a perfect statement of the situation, except that 'society' has agreed that this constitutes discrimination nor is it clear to me why we are ethically bound to do so. Ugly people don't choose to be ugly. Dumb people don't choose to be dumb (more or less). Yet it really sucks to be ugly or dumb. I'm not sure why having a harder time getting AirBnB rentals calls for government interference. When African-Americans couldn't rent most hotel rooms, that was an injustice. This is not the same thing. It's worth debating what an appropriate ethics would be, not just jumping to the conclusion the left has reached on the matter. Because that is not working out right now. It is a very real cause of Trump.

I meant, 'society' has not agreed...

it may not call for government intervention, but that doesn't mean there shouldn't be some sort of ethical norm that says it's wrong to discriminate on the basis of certain factors people can't control, in certain situations like renting hotel rooms. I think society would (and should) frown upon refusing to rent to ugly people in the same way it rejects refusing to rent to blacks. And Airbnb is totally within it's rights to enact policies which will discourage people from doing so.

in other words african americans have to prove themselves worthy others don't...

White people have to prove themselves not racist.

Men have to prove themselves not rapists.

Middle easterners have to prove themselves not terrorists

Etc.

Etc.

Everyone is affected by stereotypes.

I'm a white male and I've never had to prove myself not racist or not a rapist.

Not that you know of. When you're walking at night and a woman sees you coming the other way, you can be sure she's thinking of you differently than she would had you been a woman.

"See. . . I" ??

Need to study a declined language so you don't make there mistakes.

There are two kinds of "discrimination"

One is a trustworthy black person being seen as untrustworthy due to a lack of credible information.

The other is the fact that, on average, there are fewer trustworthy black people. So that if we did have perfect information, you would still see a "disparate impact" on blacks.

A lot of people would say "accurately evaluating trustworthiness isn't discrimination." But its disparate impact and certainly something progressives would consider disrcimination. So long as they do, they will forever be looking for ways to hide reality.

If the black individual in question is I fact trustworthy, then judging him untrustworthy based on a group association is not "accurate".

And? Is his trustworthiness effected by someone’s faulty assumption? No it’s not so what exactly is the problem unless you think his honest is contingent on having it recognized and affirmed by everyone he meets. But more broadly what is your point. Unless you arguing for a world with no heuristics or built it assumptions.
Which of course you aren’t as you are extremely prejudicial to myriad fashionable bogeymen.

Is his trustworthiness affected [sic] by someone’s faulty assumption?

No, but he's not able to benefit from it if everyone assumes he's not trustyworthy, is he? The problem is that he's being treated as being untrustworthy anyway.

I'd like to convince all of my potential employers that I would be a great employee. I get degrees in relevant subjects. I try to accumulate verifiable work experience at companies in the industry. I try to interview well, etc.

Yet, anyone who has hired knows that even these metrics, far more accurate than mere race, are horribly incomplete. They lament how hard it is to hire well because even these metrics, the best we have, don't tell us that much about potential hires.

Incomplete information is just something you are going to need to live with. All of us put work into tryin to signal trustworthiness. It's impossible to communicate our real trustworthiness in verifiable and accurate fashion. That's just the world we live in. People have to make decisions based on incomplete information based on the heuristics they have.

How would you feel if, in spite of your degrees, you work experience, and your interviews, you found out that companies were using your ethnic heritage to decide whether to hire you or not?

I'm white. I already know companies don't hire me because of my ethnic heritage. It's their STATED COMPANY POLICY.

I don't think any major company turns down brown talent. All companies are absolutely desperate to hire brown people, and give them massive legs up. Whites are obviously discriminated against. And aren't Asians suing Harvard because everyone knows they have an Asian quota.

I also think that non-whites do show high degrees of ethnic nepotism in hiring decisions. Whites probably are suckers for universally disarming in that field.

Can you name these companies that refuse to hire white people? A stated company policy should make it easier for you to prove your case and win a discrimination lawsuit in court.

So basically your answer is: "I would be pretty pissed off and carry a giant chip on my shoulder because of it."

Yes, that would be the first line.

My point is that if every single person could be judged accurately, it may still be the case that say 10% of whites are "untrustworthy" and 30% of blacks are "untrustworthy". Progressives would say that this, despite being 100% accurate, is racist.

BAD PROGRESSIVES!

There. Can we get back to talking about statistical discrimination?

It's an inevitable result of incomplete information. There will always be incomplete information. Information gathering is a cost and the potential profit to be gained doesn't always exceed the cost. In insurance we call information gathering "underwriting cost". Underwriting costs increase prices. If underwriting costs are high relative to prices than it will be a very large increase in prices, and may make the value prop to the best risks so bad they leave the risk pool causing a pricing death spiral.

If you want to not be discriminated against, try to provide the most accurate verifiable information you can. Realize that it will never be economical for any buyer or seller to achieve 100% accuracy.

"it's too expensive to call your references, so I'm just going to use your black sounding name to decide to trash your resume."

Do companies do full reference checks on every resume they receive?

My experience has been that I send out lots fo resumes and most get trashed without callback. I've only ever had references checked when I've already got the job.

How much time and effort are you expecting someone trying to get $100 out of their extra bedroom one weekend to expend? Should they be calling references?

So you're totally cool with companies trashing your resume because of your white-sounding name? That would totally not bother you at all?

And if it actually turns out that 30% of whites are "untrustworthy" and 10% of blacks are "untrustworthy," do you think that would change anything all, even it were 100% accurate?

Including your own overtly clear assumptions, obviously.

I don't know. Will the whites pass laws that force people to rent to 90% of whites, even if 20% cause serious damage, and then enforce it no matter who it hurts and use maximum social and legal sanction to anyone that gets in their way.

Lord knows that's what blacks do.

Useful results to know though I believe airbnb already requires renters to have a valid credit card on file before you can even bid. One would think the credit card requirement would function as additional information in lieu of a positive review. I'm wondering what else might serve to overcome statistical discrimination. A credit card requirement is a fairly high bar.

How shitty do you think black people are? Contrary to your assumptions the majority of blacks qualify for credit of some kind. Have you in your entire life known anyone who couldn’t qualify for a credit card? I have high school friends who OD on Vicodin before their 30th birth day that had multiple credit cards. And these people would absolutely have wrecked an AirBnB if they felt like it.

Actually, I have known quite a few people who could not get a credit card. People who have declared bankruptcy or anyone whose failed to repay a significant debt (including back taxes) would probably have problems obtaining one.

But it's a fair point that lack of credit isn't the only potential pitfall when renting to someone.

> Our recommendation is for the platform companies to build a credible, easy-to-use online reputation and communication system.
ROFL. Other useful recommendations:
> Our recommendation is for the Treasury to build an easy-to-use system to transmute base metals into gold.
> Our recommendation is for the health service providers to build a robust production process for the elixir of life.
> Our recommendation is for the utilities to build a cheap, easy-to-use source of free energy.
> Our recommendation is for the nuclear power operators to build a cheap, easy-to-use solution that makes radioactive waste vanish into thin air.
> Our recommendation is for software outsourcing companies to build a cheap, easy-to-use artificial intelligence system that converts vague and contradictory problem statements in natural English into tested and validated software products.

Your scepticism of our credible, easy-to-use reputation and communication system is unwarranted; look at all the 5-star reviews we have!

Here's another good one that's relevant to blog, news, Youtube etc. comment sections:
> Our recommendation is for the platform companies to build a credible, easy-to-use online reputation and communication system.
Jesus. This has been the holy grail of all user-produced content platform companies and their predecessors for 25 fucking years, ever since Eternal September happened. One might almost suspect that it's an extremely difficult if not insoluble problem, but of course economists know better -- all one needs to do is publish a paper with recommendations and stuff just happens.

Wake me up....when September ends...

The recommendation in this post may work for African Americans who receive positive reviews, but would probably intensify discrimination against those who receive no reviews, or negative ones. The key issue is that racism is alive and well decades after the enactment of the Civil Rights Act, and while the poor choice of language here is probably a regrettable error, rather than an expression of prejudice, the words as they are written are not good - the idea that knowledge of a person’s race is a “legitimate demand for trust,” is foundational to racist thinking.

I like transparency, and I like markets, but they aren’t going to solve the problem. I don’t know how well it works in practice, but the philosophical case for affirmative action is increasingly ironclad.

What is the philosophical case for affirmative action?

What problem are you trying to solve?

I'm not saying it's possible to fix, I'm not saying everyone thinks it's worth fixing even (as it's not their problem), but you literally don't know what racism is?

Look, set side your grievances for a second. Let's say whites are now unfairly treated in some realms, etc. Are you saying there's not something difficult about being black, in this country? Is there nothing there that makes you say "I know that group has it rough, even if I don't like how they go about handling it"? Really, just no problem of race there at all?

After reading the op-Ed, it’s notable that the authors tested negative reviews as well as positive ones and found that both types reduced discrimination. So now the question is, do most platform users have such reviews, or do most have either neutral reviews, or none at all? By the authors’ logic, it appears that African Americans in the neutral/no-review category would be better off with a system that makes it harder to identify their race.

Or they could just accumulate reviews.

When I first stated selling on EBAY I had to put in more work and sell for a slightly lower price when I didn't have any feedback. As I built feedback I got better deals. Feedback is an asset you have to invest in. There is no way around that.

Right, so if you're black it makes it significantly harder to get your first review, and you probably have to be a better renter than a white person to get a good one. One negative review and you are toast.

The key issue is that racism is alive and well decades after the enactment of the Civil Rights Act,

No, the key point is that no system run by human beings is altogether fair.

Typical conservative “nothing is perfect, so let’s accept slavery” argument. We’re working on progress, conservatives are terrified of using people power to make POSITIVE change in people’s lives. Audacity to hope, not HATE.

There’s blatant racism going on at Airbnb. A half dozen federal discrimination lawsuits backed by Warren/Harris Department of Racial Justice and Social Equality Czar will help guide them to a better path. Here’s a hint, don’t allow names to populate until the reservation has been accepted.

Prior to transaction, neither party can see name or race or picture or email. Only ratings, adjusted for race.

>Typical conservative “nothing is perfect, so let’s accept slavery” argument.

Yes. That's exactly the typical argument.

How can you even feed yourself every day?

I for one can spot the spoofs.

That reads like a typical Polar Bear poster comment. Of course Nanook of the North famously hunted polar bears.

You can certainly hide race, but Tabarrocks point is that you need some additional information to replace it. Making it so that people can't see other people's names and faces before renting will likely reduce trust not increase it, and cause fewer rentals to be available on the market.
What you want is to ADD information that more accurately identifies black renters who are trustworthy. Maybe instantaneous criminal background checks, or cross-linking of reviews from other services like Uber.

Reducing Discrimination with More Information
-------------

"Increasing discrimination with more information" is the correct, classical form. The new information adds a dimensional axis orthogonal to race, make us able to discriminate good tenants. It also requires more transaction costs which the digital systems can reduce. So, the really correct title is:
"Increasing information with bigger data providing another axis of symmetry about which we can measure queue size", if we included everything in the Theory of Everything. The idea being, an axis of symmetry has a detectable queue. When that queue is stable, is is indistinquishable from a mathematical 'force field', that is, elements of the system can be treated as if they are subject to some magical pull or pull of 'nature', and thus do Newton's calculus.

The new proof of the Born rule says this is true if all agents are "packing a sphere" as maximum entropy and uniqueness apply. Nothing to do with anything, actually, and could be called the Theory of Nothing, it just predicts, mathematically, the basic condition counting stuff.

There you have it, the complete theory of everything, from trees to quarks.

"African American sounding names" may be taken as a proxy for the social class of the parents which may correlate with the riskiness as an airbnb guess of offspring.

I would like to see a study comparing AA sounding names with southern poor white sounding names like Billy Bob etc. Then compare blacks with non-AA sounding names with whites with similar names when pictures exist to identify race.

Or compare blacks with AA sounding names and blacks without AA sounding names with picture profiles.

If you are interested in embedded discrimination in the sharing economy, you might want to read this article which examines Uber and TaskRabbit patterns in low SES and minority neighborhoods.

Some conclusions:

"We present evidence that the interaction between common sharing economy platform design decisions and these four geographic principles lead to structural geographic biases in the sharing economy, biases that reinforce existing advantages. Specifically, our results suggest that high- population density, high-income neighborhoods receive the largest benefits from the sharing economy and poor urban neighborhoods and outer-ring suburbs receive fewer benefits.
(3) We find evidence that, due to the pervasive correlation between poverty and race/ethnicity in the United States and many other parts of the world, in many cases, black and Latino neighborhoods tend to be less well-served by the sharing economy."

A very good paper which also deals with market definition in the mobile economy.

http://www.brenthecht.com/publications/tochi_preprint_sharingeconomygeography.pdf

I predict that this kind of online reputation mechanism, which on its face seems an unmixed blessing, will become dominated by Silicon Valley's largest data aggregators (Google, Facebook, and so on), and that they will do their best to bias it against the same persons and groups who are currently targets of exclusion by the "deplatforming" movement. In fact, Big Data will do their best to become just as able to ruin dissidents' lives as the Chinese Social Credit system -- and to regulate the net so that conservatives don't get any chance to return the favor.

Accountability is only a good thing when it works in all directions.

Comments for this post are closed