Monday, May 13, 2019

PBS on Uber's economists

Paul Solman interviews Uber economists (John Hall and others) and other economists (Susan Athey and Paul Oyer) on what economists do at Uber:

Sunday, May 12, 2019

UNOS proposal for public comment: Eliminate the use of DSAs and regions from kidney and pancreas distribution

Public comment solicitation is a lengthy process--and this proposal has solicited many lengthy comments (for which you'll have to scroll down at the link...)

Proposal Overview

Status: Public Comment
Sponsoring Committee: Kidney Transplantation Committee & Pancreas Transplantation Committee
Strategic Goal 2: Provide equity in access to transplants
Read the concept paper (PDF; 1/2019)
Contacts: Scott Castro and Abigail Fox
Data request from the OPTN Kidney Transplantation CommitteeProvide simulation data on effect of removing DSA and region from kidney/pancreas/kidney-pancreas organ allocation policy

Executive summary

The Final Rule (hereafter “Final Rule”) sets requirements for allocation polices developed by the Organ Procurement and Transplantation Network (OPTN) and the United Network for Organ Sharing (UNOS), including the use of sound medical judgement, achieving the best use of organs, preserving the ability for centers to decide whether to accept an organ offer, avoiding wasting organs, avoiding futile transplants, promoting patient access to transplantation and promoting efficiency. The Final Rule also includes a requirement that policies “shall not be based on the candidate’s place of residence or place of registration, except to the extent required” by the other requirements of the Final Rule.
In the past year, the United States Secretary of Health and Human Services (HHS) received critical comments regarding the OPTN/UNOS’s compliance with the National Organ Transplant Act (NOTA) and the Final Rule with respect to the geographic units used in lung and liver distribution. The OPTN/UNOS made rapid changes to eliminate using donation service area (DSA) and OPTN/UNOS regions (regions) in lung and liver distribution, respectively. Furthermore, the OPTN/UNOS Executive Committee directed the organ-specific committees to analyze their distribution systems and replace DSAs and regions with more rational units of distribution.
Policy 8: Allocation of Kidney and Policy 11: Allocation of Pancreas, Kidney-Pancreas, and Islets currently use DSA and region as geographic units of distribution. These are poor proxies for geographic distance between donors and transplant candidates because the disparate sizes, shapes, and populations of DSAs and regions result in an inconsistent application for all candidates. As noted in Department of Health and Human Services Administrator Sigounas’s letter to the OPTN/UNOS President, “DSAs and Regions have not and cannot be justified” under the regulatory requirements of the Final Rule.
Members of the OPTN/UNOS Kidney Transplantation Committee, joined by members from the OPTN/UNOS Pancreas Transplantation Committee and the OPTN/UNOS Pediatric Transplantation Committee, created the Kidney/Pancreas Workgroup (hereafter “the Workgroup”) in order to remove DSA and region from kidney and pancreas allocation policies. The Workgroup reviewed OPTN/UNOS data on current distribution practices, engaged Workgroup members on their collective clinical experience, and utilized the OPTN/UNOS Board of Directors’ “Geographic Organ Distribution Principles and Models” to develop five potential allocation options that would eliminate DSA and region from kidney and pancreas allocation policies.
The five variations that the Workgroup chose to model are:
  1. A fixed concentric circle framework with a 150 nautical mile (NM) small circle and a 300 NM large circle
  2. A fixed concentric circle framework with a 250 NM small circle and a 500 NM large circle
  3. A fixed concentric circle framework with a single 500 NM circle
  4. A hybrid framework with a single 500 NM circle that utilizes a small number of proximity points inside and outside of the circle, and
  5. A hybrid framework with a single 500 NM circle that utilizes a large number of proximity points inside and outside of the circle.
These variations will be more comprehensively outlined in this paper’s “What Concepts Are Being Considered?” section. The Workgroup is not limiting itself to consideration of solely these five variations, but rather used these variations as choices to model in the Kidney/Pancreas Simulated Allocation Model (KPSAM) in order to most strategically determine what could be the ideal variation. The Workgroup understands that, given community feedback and additional evidence gathered, it is possible that the framework and variation ultimately selected by the Workgroup may be a combination of these variations, or perhaps a new variation, such as a single-circle hybrid with a smaller concentric circle.
The Workgroup is currently considering these five variations for modifying kidney and pancreas allocation policy to be more consistent with the Final Rule and to provide more equity in access to transplantation regardless of a candidate’s place of residence or registration, except to the extent required by §121.8 (a)(1)-(5) of the Final Rule. The Workgroup requests community feedback in order to better inform the evidence-gathering and decision-making processes.

Feedback requested

Saturday, May 11, 2019

How are American transplant centers regulated? How does this influence treatment decisions?

Here's a NYT oped about the fact that transplant centers are regulated based on their one-year graft survival statistics--i.e. on how often the transplant (and the patient) lasts 12 months:

When Is a Transplant Worth It?
A year in a hospital bed is a “success” while dying after 11 months is failure.
By Daniela J. Lamas
Dr. Lamas is a pulmonary and critical care physician.

"The single-minded focus on staying alive for a year begins at the time of a transplant program’s initial certification by the Centers for Medicare and Medicaid Services. If the program’s one-year mortality rate is higher than expected, possibly because surgeons are giving transplants to people who are too sick, that program could be put on probation or lose its certification. That metric is equally important to the United Network for Organ Sharing, which allocates donor organs. A patient looking for information might happen upon the Yelp-style transplant center rankings developed by the Scientific Registry of Transplant Recipients — also based on one-year mortality, particularly for lung transplant programs.
...
"The focus on one-year mortality isn’t necessarily what patients want, and it can have unintended consequences. Dr. Richard Formica, a kidney transplant specialist at Yale-New Haven Hospital, noted that with mortality as the metric of success, surgeons might be apt to discard riskier transplant organs because they worry about their numbers. The concern about program numbers — and the potential repercussions for other patients if a center loses its certification — also might influence the choice of who gets a transplant in the first place. “Do we deny patients who have an increased risk of mortality in the first year?” Dr. Formica asked. “Yes, we do.”

Friday, May 10, 2019

Indian Society of Organ Transplantation meeting in Ahmedabad May 11-12

I'll be in Ahmedabad this weekend, starting with a conference of the Indian Society of Transplantation:

http://isot.co.in/file/ISOT_Mid_term_meeting_11-12_May_Ahmedbaabd.pdf

Here's a draft of the program.

My two talks will be on
History and organization of kidney exchange, and
Taboo transactions and frontiers in ethical kidney exchange

Thursday, May 9, 2019

Shrouded prices for blood tests in the U.S.

One of the features of the American health care system is that prices are heavily shrouded--insurance companies reach negotiated prices with providers, that may be very different with different providers, and very different from list prices, and are not quoted. So prices aren't nearly as informative in health care as in most other markets.

Here's a NYT story that focuses on blood tests:

They Want It to Be Secret: How a Common Blood Test Can Cost $11 or Almost $1,000
Huge price discrepancies like that are unimaginable in other industries. Also unusual: not knowing the fee ahead of time.


Wednesday, May 8, 2019

David Kreps on Behavioral Economics (Nemmers Prize Lecture)

David Kreps won the 2018 NEMMERS PRIZE IN ECONOMICS , and
today, at Northwestern, he is giving his

NEMMERS PRIZE LECTURE

"SOME DIMENSIONS OF BEHAVIOR WITH WHICH ECONOMICS SHOULD CONTEND"


Behavioral economics is generally taken to mean economics in which the behavior of individual agents does not conform to the “standard model” of rational behavior.  However, under this banner, one finds a very large number of specific “nonstandard” models of behavior.  This very large number prompts a standard criticism of behavioral economics:  If any behavior is permissible, any conclusion can be reached.  
Using a small handful of examples, the lecture illustrates and fleshes out a test for the value of work in behavioral economics.  This test is based on three principles:
  1.  Is the behavior in the model systematic, at least in some important contexts?
  2. Does positing this behavior lead to economically significant phenomena?
  3. Either via intuition or, preferably, empirical evidence, does the behavior provide "better" explanations of those significant phenomena, where defining the adjective “better” is the crux of the matter.

Tuesday, May 7, 2019

School choice in San Francisco--update in the NYT

Here's the NY Times story: San Francisco Had an Ambitious Plan to Tackle School Segregation. It Made It Worse.

“Our current system is broken,” said Stevon Cook, president of the district Board of Education, which, late last year, passed a resolution to overhaul the process. “We’ve inadvertently made the schools more segregated.”
...
"About a quarter of the city’s children are enrolled in private school, a higher percentage than in some other major cities, like New York, where it is around 20 percent. The lottery system is thought to be a major reason wealthy parents here opt out of public schools, further worsening segregation."
**********

The San Francisco Unified School District interacted with market designers some years ago, but ultimately turned down their (our) help and decided to deal with the existing problems in-house.  Here are some old blog posts...


Thursday, September 23, 2010

And
Thursday, June 2, 2011

Monday, May 6, 2019

A (first) liver-kidney exchange

Here's a forthcoming paper in the American Journal of Transplantation:

Bi‐organ Paired Exchange – Sentinel Case of a Liver‐Kidney Swap
by Ana‐Marie Torres  Finesse Wong  Janine Sabatte‐Caspillo  Sandy Del Grosso John P Roberts  Nancy L Ascher  Chris E Freise  Brian K Lee
First published: 12 April 2019
https://doi.org/10.1111/ajt.15386

Abstract: "Organ transplantation is the optimal treatment for patients with ESLD and ESRD. However, due to the imbalance in the demand and supply of deceased organs, most transplant centers worldwide have consciously pursued a strategy for living donation. Paired exchanges were introduced as a means to bypass various biologic incompatibilities (blood‐ and tissue‐typing), while expanding the living donor pool. This shift in paradigm has introduced new ethical concerns that have hitherto been unaddressed, especially with non‐directed, altruistic living donors. So far, transplant communities have focused efforts on separate liver‐ and kidney‐paired exchanges, whereas the concept of a trans‐organ paired exchange has been theorized and could potentially facilitate a greater number of transplants. We describe the performance of the first successful liver‐kidney swap."
******
The discussion of the ethical concerns mentioned in the abstract strike me as worth looking at, given that one of the authors, when she was president of the Transplantion Society, argued strenuously that poor patients should not be allowed to participate in American kidney exchange, for ethical reasons.

First, here are the practical steps they took:
"Our team debated the ethical underpinning of this swap. A discussion with the chair of the ethics committee at the time concluded that a full committee review was unnecessary."

Second, they considered the differential risks to the kidney and liver donors, and decided that this did not disallow the donations, since both kidney and liver donation is already accepted.

Finally they discuss the differential benefits to the liver and kidney recipients:
"Another area of contention is that donor-L’s recipient received remarkably less from a “life-enhancing” kidney transplant (rather than a truly “life-saving” liver transplant), despite the fact donor-L took on the substantially greater risk of donor hepatectomy. Our counterpoint is that the kidney recipient was spared from an extended dependency on dialysishad she stayed on the deceased donor waitlist (mortality on the kidney wait-list is 6-8% annually with a significant reduction in quality of life15,16). This does not even account for the superior allograft and patient survival outcomes that comes with a living vs. a deceased donor kidney transplant17. In fact, Merion18made the observation that the risk to patients on the kidney waitlist is not dissimilar to the liver waitlist mortality and reduction in quality of life for those with moderate MELD scores of 12-17. "

Fortunately for the patients involved, they decided that the benefits to the liver and kidney recipients were comparable. The implication is that if they had decided that the 'life-saving' benefits of a liver exceeded the 'life-enhancing' benefits of the kidney, then the exchange would have been unethical, and the ethical course of action would have been not to go ahead with it--which would likely have resulted in a quick death sentence for the liver patient, and perhaps a slower one for the kidney patient.

I'm glad that the dire decree was diverted.

It is this kind of ethical reasoning that led one of the authors to conclude that it would be unethical to go ahead with global kidney exchanges involving patient-donor pairs whose care had to be financed outside of their own countries' insurance coverage, so that, ethically, they could not be offered treatment even when financing was available. (It's lucky that the patients in this liver-kidney exchange apparently had good insurance, too.)
**********
The paper has an interesting back story in some speculative thoughts on market design.

The second reference in the paper is to a paper by computer scientists John Dickerson at Maryland and Tuomos Sandholm at CMU, proposing that multi-organ exchanges might substantially increase transplantation:
 Dickerson J, Sandholm T. Liver and multi-organ exchange. Journal of Artificial Intelligence Research. 2017;60:639

An unusual twist to the story is that it is the liver donor, eager to help her mother get a kidney transplant, who read the Dickerson-Sandholm paper and proposed the idea to the docs at UCSF.


Here's a press release from CMU, celebrating the event, and Sandholm's contributions:
Computer Science Idea Triggers First Kidney-Liver Transplant Swap
Sandholm says multi-organ exchanges could boost number of transplants
***********

Update: here's a May 11 story in the Washington Post


Sunday, May 5, 2019

Do child labor laws apply to social media?

The Guardian asks the question:
'It's not play if you're making money': how Instagram and YouTube disrupted child labor laws

"while today’s child stars can achieve incredible fame and fortune without ever setting foot in a Hollywood studio, they may be missing out on one of the less glitzy features of working in the southern California-based entertainment industry: the strongest child labor laws for performers in the country.

"Those laws, which were designed to protect child stars from exploitation by both their parents and their employers, are not being regularly applied to today’s pint-sized celebrities, despite the fact that the major platforms, YouTube and Instagram, are based in California. The situation is a bit like “Uber but for … child labor”, with a disruptive technology upending markets by, among other things, side-stepping regulation."

Saturday, May 4, 2019

Global kidney health atlas from the International Society of Nephrology

Here's the 2019 Global Health Atlas

Kidney transplantation is nowhere readily available to everyone who needs it,  but the wealthy countries of the West provide it most widely:

Annual costs of dialysis are roughly the same as costs of transplantation in the year the transplant is performed (and of course dialysis costs stay roughly constant until the patient dies or is transplanted, while the costs of a successful transplant become largely the costs of immunosuppressive drugs).


Friday, May 3, 2019

Donating eggs for fertility: an update

Slate brings us up to date:

INSIDE THE QUIETLY LUCRATIVE BUSINESS OF DONATING HUMAN EGGS

"The first US child conceived from a donated egg was born in 1984. Since then, the procedure has grown into a thriving industry. Demand from aspiring parents, along with a dearth of regulations, have spawned matchmaking agencies that offer to help parents find the perfect young woman whose eggs will result in the equally perfect child.

"Donating eggs can be lucrative, with agencies paying as much as $50,000 per cycle in some cases.
...
"Ads or marketing materials targeting potential donors rarely mention the risks or common complaints. Liz Scheier donated eggs three times between 2005 and 2007, and says she was told there were no known risks associated with egg donation. Today, Scheier is a media liaison for We Are Egg Donors, a women’s health organization that works with more than 1,500 donors to promote transparency and advocate for their concerns. She says that donors nowadays hear the same line she did, delivered almost verbatim. But it’s missing one key detail. “There are no known risks because no one has looked,” she adds.
...
"Egg donation has thrived in the US in part because there are few laws regarding the transfer of unfertilized eggs for reproductive purposes, according to industry experts. They say a handful of states have policies that touch on some aspect of egg donation, generally from the perspective of the recipient.

With few regulations, the US has become a magnet for well-off wannabe parents in other countries where egg donation is regulated, or outright illegal. Egg donation is barred in China, Germany, Italy, and Norway; paying women to donate eggs is prohibited in most of Europe, as well as in Canada and other nations.
...
"In 2011, an egg donor sued ASRM [American Society for Reproductive Medicine] over the organization’s compensation guidelines, which the donor claimed were a form of illegal price-fixing; other donors later joined the case. In a 2016 settlement, ASRM agreed to eliminate its payment suggestions, pay $1.5 million in legal fees, and give the plaintiffs $5,000 each. Agencies were freer to offer donors more money.
...
"Since then, donor pay has soared, particularly for attractive, well-educated donors."


HT: Stephanie Lo

Thursday, May 2, 2019

Legal marijuana in California struggling to compete with the well established black market

Yesterday I posted about the competition the legal market for marijuana in Canada is facing from the pre-existing illegal market. Today we turn to California. It's still to early to draw conclusions, but one market design lesson is that even if the attraction of a legal market is regulation and tax revenue, regulators and tax authorities may have to ease up at least in the beginning if they want the legal market to out-compete the legacy black market.

In the NY Times, Thomas Fuller has been following the market(s):

‘Getting Worse, Not Better’: Illegal Pot Market Booming in California Despite Legalization  By Thomas Fuller

"It’s been a little more than a year since California legalized marijuana — the largest such experiment in the United States — but law enforcement officials say the unlicensed, illegal market is still thriving and in some areas has even expanded.
...
"The struggles of the licensed pot market in California are distinct from the experience of other states that have legalized cannabis in recent years. Sales in Colorado, Oregon and Washington grew well above 50 percent for each of the first three years of legalization, although Oregon now also has a large glut of pot.

"But no other state has an illegal market on the scale of California’s, and those illicit sales are cannibalizing the revenue of licensed businesses and in some cases, experts say, forcing them out of business.
...
"California gives cities wide latitude to regulate cannabis, resulting in a confusing patchwork of regulation. Los Angeles, San Francisco, San Jose and San Diego have laws allowing cannabis businesses, but most smaller cities and towns in the state do not — 80 percent of California’s nearly 500 municipalities do not allow retail marijuana businesses. The ballot measure legalizing recreational marijuana passed in 2016 with 57 percent approval, but that relatively broad support has not translated to the local level. Cities like Compton or Laguna Beach decisively rejected allowing pot shops.

"Regulators cite this tepid embrace by California municipalities as one of many reasons for the state’s persistent and pervasive illegal market. Only 620 cannabis shops have been licensed in California so far. Colorado, with a population one-sixth the size of California, has 562 licensed recreational marijuana stores.
...
"And the monetary incentives of trafficking also remain powerful: The price of cannabis products in places like Illinois, New York or Connecticut are typically many times higher than in California."

Wednesday, May 1, 2019

Competition between legal and illegal cannabis in Canada

Canadian cannabis competition:

Canada's legal weed struggles to light up as smokers stick to black market
Six months after legalisation, licensed producers are unable to keep up with the demand or quality of neighborhood dealers

"Six months after Canada became the first G7 country to legalise marijuana, the bold experiment is still struggling to get off the ground.

"Legal producers were unable to meet the sudden surge in demand, and struggled for weeks to fill orders, leaving marijuana stores with empty shelves.

"As a result, the vast majority of cannabis sales in the country – roughly $5bn – are made on the illegal markets, compared to $2bn in legal sales, according to government figures from January 2019.
...
"The Tobins are also competing against illegal “grey market” stores, which alongside marijuana sell edibles and hashish – items that licensed stores cannot yet offer.
...
"Canadians who purchase their cannabis from illegal sources also save a significant amount of money: the average price for a gram of illegal cannabis is 36% cheaper than its legal counterpart, Statistics Canada has found."
********

Here's a price survey and other stats from Statistics Canada: Cannabis Stats Hub

Tuesday, April 30, 2019

Drone delivery of a kidney for transplant

Here's the press release from the University of Maryland,
Pioneering Breakthrough: University of Maryland's Schools of Medicine and Engineering First to Use Unmanned Aircraft to Successfully Deliver Kidney for Transplant at University of Maryland Medical Center

"BALTIMORE and COLLEGE PARK, MD— In a first-ever advancement in human medicine and aviation technology, a University of Maryland unmanned aircraft has delivered a donor kidney to surgeons at the University of Maryland Medical Center for successful transplantation into a patient with kidney failure.
...
"Among the many technological firsts of this effort include: a specially designed, high-tech apparatus for maintaining and monitoring a viable human organ; a custom-built UAS with eight rotors and multiple powertrains to ensure consistently reliable performance, even in the case of a possible component failure; the use of a wireless mesh network to control the UAS, monitor aircraft status, and provide communications for the ground crew at multiple locations; and aircraft operating systems that combined best practices from both UAS and organ transport standards.

“We had to create a new system that was still within the regulatory structure of the FAA, but also capable of carrying the additional weight of the organ, cameras, and organ tracking, communications, and safety systems over an urban, densely populated area—for a longer distance and with more endurance,” said Matthew Scassero, MPA, director of UMD’s UAS Test Site, part of A. James Clark School of Engineering. "

HT: Alex Chan

Monday, April 29, 2019

Refugee resettlement in the U.S., by HIAS, using matching technology

The Atlantic brings us up to date:
How Technology Could Revolutionize Refugee Resettlement
A software program called “Annie” uses machine learning to place refugees in cities where they are most likely to be welcomed and find success.

"Monken, an associate director at HIAS, a migrant-assistance charity, tells me Njabu and his family were specifically placed in Pittsburgh “because of the high employment probability forecasted by Annie.”

She was referring not to a person, but to a software program. Named for Annie Moore, the Irishwoman who was the first person to pass through Ellis Island, the New York outpost that served as the gateway for millions of immigrants to America, Annie is at the core of an ambitious experiment, one that, were it deployed more widely, could transform how refugees are allocated and treated around the world.
...
"Developed at Worcester Polytechnic Institute in Massachusetts, Lund University in Sweden, and the University of Oxford in Britain, the software uses what’s known as a matching algorithm to allocate refugees with no ties to the United States to their new homes. (Refugees with ties to the United States are resettled in places where they have family or community support; software isn’t involved in the process.)
...
"The software itself is in its infancy right now. For one, it is lacking in data: HIAS has been using Annie since last summer and has placed about 250 people via the software so far. There’s no exact number on how many refugees Annie must place in order to measure the program’s success. Instead, the software’s efficacy will be measured over several years and through the economic outcomes of the cases that go through the algorithm. Back-testing using data from previous years has yielded promising results, but the real outcomes will take a long time to discover. (Acquiring more data will be its own challenge: The Trump administration’s policy of reducing the number of refugees resettled in the United States means that last year the country accepted fewer refugees, just 22,491, than at any other point since President Jimmy Carter signed the Refugee Act of 1980.)"
************
Here's a website related to the algorithm and related academic work:

Refugees.AI
Resettlement that empowers refugees and communities

"Contact Us
If you would like to work with us on designing matching systems for refugee resettlement, please drop us a line."

************
Here's an earlier post on this work:
Thursday, October 4, 2018

Sunday, April 28, 2019

Hayek at auction at Sothebys

It closed last month, but you can still take a look at the outcome of Sotheby's sale called
Friedrich von Hayek: His Nobel Prize and Family Collection
ONLINE BIDDING CLOSED19 Mar 2019 |
Sale L19409

"Sotheby’s is honoured to offer the Nobel Prize and Family Collection of Friedrich August von Hayek (1899-1992). Hayek was a towering intellectual figure of the twentieth century and his writings have had a profound impact in shaping the modern world. He was awarded the Nobel Prize in 1974 for “pioneering work in the theory of money and economic fluctuations”. This month sees the 75th anniversary of his seminal work, The Road to Serfdom."

Despite the title of the sale, Hayek's Nobel medal doesn't seem to have been on offer, although his U.S. Presidential Medal of Freedom , awarded by President George H.W. Bush in 1991, went for 112,500 GBP.

The most expensive item was his annotated copy of Adam Smith's Wealth of Nations, for which the hammer came down at 150,000 GBP.

The auction notes state:
"Hayek’s underlining includes phrases from the first paragraph of Chapter II which reads “This division of labour, from which so many advantages are derived, is not originally the effect of any human wisdom, which foresees and intends that general opulence to which it gives occasion. It is the necessary, through very slow and gradual consequence of a certain propensity in human nature which has in view no such extensive utility; the propensity to truck, barter, and exchange one thing for another.” (p.12) and later “it is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own interest.” (p.13)."
***********

I've blogged about previous Nobel auctions:

Thursday, October 27, 2016 Re-auction of Kenneth Wilson's, 1982 Nobel medal for Physics

Thursday, February 26, 2015 Kuznets' Nobel Prize medal auction

Saturday, April 27, 2019

Sniping in soft-close auctions (right before the time that triggers an extension)

From Marketing Letters:

Marketing Letters
pp 1–13Cite as
  • Wen Cao, 
  • Qinyang Sha, 
  • Zhiyong Yao, 
  • Dingwei Gu, 
  • Xiang Shao
Abstract: The existing studies suggest that sniping is an equilibrium strategy in hard-close online auctions, but not in soft-close ones. In this paper, we use a unique, large-scale data set from soft-close Overstock and hard-close eBay to document sniping phenomena under the two different closing rules. Estimation results show that sniping is prominent on both websites, but they are prevalent at different times. On eBay, sniping occurs right before the auction close, while on Overstock sniping happens predominantly in a short window of time before the triggering period, during which any additional high bid automatically extends the online auction. Furthermore, the revenue effect of sniping is significantly stronger on Overstock than on eBay.

Friday, April 26, 2019

Repugnant blood samples (for gender testing in pregnancy) in China

The South China Morning Post has the story: blood samples for gender tests apparently are a leading indicator of abortion of female fetuses:

Chinese blood mule, 12, caught trying to smuggle 142 samples into Hong Kong for sex testing
"Youngster apprehended at Shenzhen port with more than 1.4 litres of blood from expectant mothers in her backpack
Samples had papers requesting DNA tests to show if fetuses were male or female"

Thursday, April 25, 2019

Should NYC school choice diversify school assignments to match applicant demographics?

Some commentators are concerned that features closely correlated with race, for example, can be used in computerized algorithms that don't explicitly use race (see previous two posts here and here). But below is a proposal that sees using features correlated with race as an advantage for achieving diversity in NYC schools, with a view towards making admissions look as diverse as applications.

Here's a report from FastCompany:

How to fix segregation in NYC schools? Let students hack the algorithm
A Nobel Prize winner’s algorithm helps decide which students are placed in which New York schools. A team of students is trying to upgrade it.

"Many of the most desirable, highest-performing schools have a gross disparity between the racial breakdown of who applies versus who eventually attends those schools.

"Data acquired from the Department of Education by IntegrateNYC through a freedom of information request and provided to Fast Company bleakly demonstrates this point. For instance, while white students accounted for one quarter of students who applied in 2017 to Beacon High School, 42% of that Fall’s freshman class was white. At Eleanor Roosevelt High School, 16% of applicants that year were black, yet less than 1% of admitted students were black.
"Part of the problem is that the education children receive from kindergarten to eighth grade is not equal. Students who live in more affluent, largely white neighborhoods have better middle schools, which better prepare students for high school entrance exams. Students from wealthier families are also more likely to be able to afford private test prep for their students. But the city’s current admissions process does nothing to correct this.
...
"The solution students came up with was to create a new matchmaking algorithm that prioritizes factors highly correlated with race such as a student’s census tract, whether they receive free or reduced-price lunch, and whether English is their second language. Such an algorithm would boost disadvantaged students higher up in the matchmaking process, provided they have already passed a school’s screening process."
***********

In NYC, school principals have a lot of agency in determining the input of the school matching algorithm, in the form of preference lists for their schools. The city (i.e. the NYCDOE) provides guidelines for schools. So another approach to achieving more and different diversity would be to provide different guidelines and requirements for schools, that would change the inputs to the matching algorithm (the schools' rank order lists of students), rather than trying to modify the algorithm. My guess is that this would be a more effective, nuanced, and flexible approach.

Wednesday, April 24, 2019

Insurance, privacy, surveillance, algorithms, and repugnance

The NY Times is on the case:

Insurers Want to Know How Many Steps You Took Today
The cutting edge of the insurance industry involves adjusting premiums and policies based on new forms of surveillance.
By Sarah Jeong

"Last year, the life insurance company John Hancock began to offer its customers the option to wear a fitness tracker — a wearable device that can collect information about how active you are, how many calories you burn, and how much you sleep. The idea is that your Fitbit or Apple Watch can tell whether or not you’re living the good, healthy life — and if you are, your insurance premium will go down.
...
"artificial intelligence is known to reproduce biases that aren’t explicitly coded into it. In the field of insurance, this turns into “proxy discrimination.” For example, an algorithm might (correctly) conclude that joining a Facebook group for a BRCA1 mutation is an indicator of high risk for a health insurance company. Even though actual genetic information — which is illegal to use — is never put into the system, the algorithmic black box ends up reproducing genetic discrimination.

"A ZIP code might become a proxy for race; a choice of wording in a résumé might become a proxy for gender; a credit card purchase history can become a proxy for pregnancy status. Legal oversight of insurance companies, which are typically regulated by states, mostly looks at discrimination deemed to be irrational: bias based on race, sex, poverty or genetics. It’s not so clear what can be done about rational indicators that are little but proxies for factors that would be illegal to consider.
...
"A. I. research should march on. But when it comes to insurance in particular, there are unanswered questions about the kind of biases that are acceptable. Discrimination based on genetics has already been deemed repugnant, even if it’s perfectly rational. Poverty might be a rational indicator of risk, but should society allow companies to penalize the poor? Perhaps for now, A.I.’s more dubious consumer applications are better left in a laboratory."

HT: Julio Elias