24 May 2019


May be industry has had enough of R's silly syntax? The news.
After having been in the top 20 for about 3 years, statistical language R dropped out this month. This is quite surprising because the field of statistical programming is still booming, especially thanks to the popularity of data mining and artificial intelligence. It seems that there is a consolidation going on in the statistical programming market. Python has become the big winner. A possible reason for this is that statistical programming is finding its way from university to industry nowadays and Python is more accepted by the industry.

Not the least is R's memory constrained execution engine. In these days of Big Data and Machine Learning using same, R (sans extensions) doesn't add up. Add the fact that much of what goes on these days isn't statistics, just number crunching.

22 May 2019

Dee Feat is in Dee Flation - part the thirty sixth

Yet another reportage about the bewilderment of the Professionals and Professional Pundits regarding the lack of inflation. You only get inflation if the supply of moolah outstrips the supply of output. You don't need to understand all of Samuelson's book to know that; just any Econ 101 at your local community college.

All that GDP growth and earlier TARP and QE money went to the rich folks. It's a tautology: as income/wealth get infinitely large, marginal propensity to consume gets infinitely small. All the excess moolah never makes it to Macy's. There's also the reason the asset markets have gone up into the stratosphere starting with TARP and all the rest. The rich folks take all this new, excess, moolah and chase assets. Mostly, they chase Treasuries since they're scared little bunnies.

20 May 2019

Carnac Times Six

Yes, there were a few analysts who saw the Great Recession coming. And, no, so far as I can tell, they didn't acquire this insight aided by DSGE models. They looked at the vulnerable structures and pulled the fire alarm. They didn't, so far as I can tell, run any fancy stat pack routines, Bayesian or otherwise. Here's one post mortem for your reading pleasure.

Of particular interest ('He' is Raghuram Rajan):
He said the rollout of complicated instruments such as credit-default swaps and mortgage-backed securities made the global financial system a riskier place.

If you follow the link in his section, you find this bit:
He then focused on the "credit default swaps" which promised to repay delinquent loans in exchange for moderate insurance premiums. Noting that nobody really knew how realistically these swaps were priced, Rajan said that the banks were probably taking excessive risks because they trusted that the insurer would repay them. In these circumstances, a sudden increase in defaulting loans could exceed the reserves of the insurer, leading to a financial crisis. This is exactly what happened two years later, leading to the 2008 financial crisis.

The "It's all Blythe Masters fault!" quote. Not mentioned in either reporting is the key fact of CDS: it's exactly the same as betting on someone else's throw of the dice at a casino. The CDS is a pure bet by an entity not attached to the asset being bet on. And, crucially, there was no limit to the betting. Now you know why AIG crashed. Or, as reported
Or you could just sell GE credit default swaps. You get money from other banks, and all you have to give is the promise to pay if something bad happens. That's zero money down and a profit limited only by how many you can sell.

That was from 2005, long before the cracks began to appear. But he noted that the CDS, et al, had changed the rules of the game in a negative way. Alas, such insight comes not from running fancy quant models, but from reading the newspaper; and having eyes and ears inside financial institutions. What data can tell you: how are the gross money flows changing? But you don't need regression to see such. As the Great Recession proved, time series analysis, of any variety, only works when the rules of the game are stable. Change the rules...?

17 May 2019

Let's Make Some Smores

Not just Brownies make smores around the campfire. But PG&E now admits that it keeps on burning down NoCal.

The strongest argument for allowing the Damn Gummint to interfere in the Free Market is PG&E's behavior. We should be e(x)ternally grateful to them. Without strong regulation of large entities (too big to fail ring a bell?), compensating the injured for gross externalities is a joke. Corporations aren't constrained to hold sufficient resources to cover damages they may inflect on the public. The whole point of the corporation is to wall off assets from seizure. An ounce of prevention and all that. Next time you fly in a 737Max, if you ever do, count your blessings if you land in one piece.

16 May 2019

Simple Arithmetic

It's something 'everybody knows': pharma spends more on sales & marketing than the piddling amounts of its revenue on research, and what it does spend is largely on tweeking existing product in order to maintain a patent (thus, pricing) position. Well, this bit doesn't fully support all of the 'knows', but it does go from commando to briefs.
Biotech's famed four horsemen -- $BIIB, $AMGN, $CELG, $GILD - spent $163B in cash over past 5 years. 68% or $87B went to stock buybacks, while $29B went to major M&A... from JPM analysts

It's also worth remembering that the $29B spend is zero-sum from a research point of view; some compound(s) have been robbed from Peter to benefit Paul. Could be less than zero if the purpose is to squelch a competitive compound. From the macro point of view, nothing good has happened.

There have been, at least annually, such remarks. But it's laborious to get the numbers, since one has to parse through myriad 10-K. These only looked at the Gorillas.

15 May 2019

Dee Feat Is In Dee Flation - part the thirty fifth

About the only benefit to doing house/dog/cat/roach sitting for a brother-in-law in DC, is that, oft times, the NYT doesn't show up, thus compelling me to do the Post puzzle. The horror! Made more mysterious, given that the front-page section of the Times is, at minimum, 50% Washington reporting. I suppose it's parochial envy. So, one of those days meant getting Robert Samuelson's column. Again, not for the first time, an econ/business reporter tells only a bit of the story of 'flation. (You'll have to deal with their stupid ad policy!)

Even if you choose not to kneel at their feet, here's the point: Samuelson, as many others, writes about the failure of economists to predict episodes of 'flation, both in- and de-. And, he tosses up his hands without offering an explanation. Which is, simply, that most in the econ biz curry to the idea that there is only wage-push inflation, all the while ignoring the fact that the growth in GDP since the Great Recession has been driven by most of said growth going to the 1% (and nearby). No mo money, no inflation. IOW, income inequality has the beneficial (for some) side-effect of stable prices. No mo money, no price hikes. Well, drugs excepted. And tariffs.

Events drive the data, not the other way round. At the macro level, anyway.