03 May 2017

Watch Your Head [update]

Not for the first time, I suppose, but a question: isn't 99.44% of quant work just overhead, not output? And for the .56% that is sold output, well that's just sold to MegaCorp to do overhead tasks that MegaCorp chooses not to do in-house.

The recent United fiasco and Southwest mea culpa got me to cogitating, not for the first time, about whether we're actually making much "progress" in both the technical and cultural meanings; have the last couple of generations of invention produced anything as meaningful as the steam engine or washing machine? Gordon's book gets it right.

My current bete noir is dealing with some tracking of mandatory healthcare consumption, specifically my annual physical. The insurer decided I have to have one, despite the evidence (and more each day) that there's really not much value to an annual physical, as commonly executed; certainly the one I get.
The unequivocal conclusion: the appointments are unlikely to be beneficial. Regardless of which screenings and tests were administered, studies of annual health exams dating from 1963 to 1999 show that the annual physicals did not reduce mortality overall or for specific causes of death from cancer or heart disease. And the checkups consume billions, although no one is sure exactly how many billions because of the challenge of measuring the additional screenings and follow-up tests.

Of course, the fact that I did have said physical, all with Windows based software tracking each and every move, didn't get from my doctor's office to the insurer and/or from the insurer to the third-party nanny nagger. Many a slip twixt the cup and the lip. And, of course, the insurer threatens to charge multiple hundreds of dollars per month for failure to have the physical. Sounds like an opportunity much like Catch-22. Losing notifications along the way is to the insurer's advantage. I wonder whether how many quants in Hartford are calculating how many notifications to lose and stay under the regulatory radar? Do the quants get paid on a commission-only basis? Said commission on the penalty charges?

With cheap diagnostics of mortal conditions (pan can, for example), then annual (or so) screenings make sense. As it is, the resting EKG I got is utterly worthless as a diagnostic of my underlying cardiac health.
The test is not useful in routine checkups for people who do not have risk factors for heart disease such as high blood pressure or symptoms of heart disease, like chest pain.

So, the insurer pays this third-party to track whether I've done my Good DooBees each year. The doctor's office tells the insurer, who tells the third-part, who then tells the insurer. I think that's a decent definition of a circle jerk. Which they've managed to screw up, so to speak.

The whole point of quant, in the profit making world at least, is that the cost of a quant exercise brings greater value than the cost of the exercise. We see, may be, this with the overbooking scandal. On the one hand, airlines claim they overbook, and do some calcs to set the percent of such overbooking:
Ms. Owens, along with her main job of setting various fares on a single flight, tweaks the overbooking numbers. Then, each week a report comes out that lists all US Airways flights that bumped 10 or more people. The analyst with the most flights on the report is stuck with a stuffed toy crow for the week. And occasionally they hear from angry airport workers who handled the bumping.
On the other hand, the airlines sell "unbooked" seats through discount brokers, which pretty much guarantees more butts than seats. That text is from 2007, if you noticed, so perhaps the quants have gotten more stats since then, but I doubt it. There's really no penalty for not doing the job smart. Well, a toy crow isn't much punishment.

Not the best use of all the data available as mentioned in an earlier missive. The airlines know that the consumer traveller, with those early and last minute bought tickets, are in no refund land. The discretionary no shows are nearly always business junket guys (and, I suppose, gals) on expensive tickets, so they'll lose far more than just the revenue of the high priced ticket: the cost of the bump has to be included. Not to mention that those steeply discounted broker tickets don't come close to making up the difference.

Could the airlines be bumping out Ma and Pa Kettle who just got excess, cheap tickets from a broker to accommodate Mr. Master of the Universe who just found out he has to get to Orlando for a sales meeting and is willing to pay full freight? What do you think?

So, back to the healthcare situation. In my callow youth, I worked on a Progress application called Optimed (it was bought a few times since I left and before it was sold by the company that bought it, so I don't know if it's still around) which was (is?) a pre-qualification screener. The clients would add rules for certain procedures, based on some, sort of, established criteria shipped with the application. The point, of course, was an ever more elaborate ring-around-a-rosy to justify denying services. Welcome to health as profit. So, the value of quant in this sort of case is cost avoidance. I remain convinced that, modulo fraud avoidance, it's likely cheaper to allow physicians to be doctors, and fire all those $100K/annum quants. YMMV.

The idle educated keep taking more of GDP for nothing gained. What a way to run a country?

Turns out, the subtitle of one version of these endeavors (penned nearly a decade ago, BTW), "It's the Distribution, Stupid" is gaining some traction:
"The 'jobs of the future' are likely to be performed by robots," said Nathaniel Borenstein, chief scientist at Mimecast, an email company. "The question isn't how to train people for nonexistent jobs. It's how to share the wealth in a world where we don't need most people to work."

No comments: