16 October 2014

One Life to Live

It was raining cats and dogs this morning here in South Fireplug, CT (although not nearly what my beloved, yet benighted, Bermuda faces tomorrow), so I postponed my daily stroll to get the dead trees NYT and a cuppa joe until it let up. In the meantime, I spent some time with r-bloggers and came across a young pup who talked about predicting life expectancy from historical data. I considered a tongue in cheek rebuttal, but tasks interfered.

The rain let up, and there were still NYTs on the shelf. What do I find? David Leonhardt's, et al take on the issue of quants not looking out the window to see if it's raining. These endeavors have addressed this issue a few times. Time series data of some attribute may be useful in predicting the attribute's future value, IFF the entities which have the attribute are running under God's Laws over which they have no control. On the other hand, when measures are of some aspect of human activity, it's mostly a Wizard of Oz situation: some human(s) are behind the curtain bending, breaking, or re-writing the "laws" of behavior which determine the attribute. It wasn't thermodynamics which propelled house prices, but various Wizards playing various games with the rules of the game. That financial quants were all too happy to ignore the absurdity of their data, well...

Curiously, one might argue that the examples (possibly, save one or two) he gives of odds are of attributes and entities which obey God's Laws. Such odds are accurate, by definition of our understanding of God's Laws. No mention of house prices continuing to rise like smoke from a volcano.

So far as life expectancy goes, if you go and look (I've provided links more than once), you'll see that from 1900 to 2000 life expectancy at birth increased by nearly 70%, but life expectancy at 65 increased about 7% of total lifetime (less if you measure from the start of Social Security). The increase in life expectancy at birth is the number that right wingnuts use to bray, "we have to kill SS because 'we' can't afford it. You all have to buy stocks and bonds from our friends in the financial services industry." Not that they see any conflict of interest. The overhead of SS is about 1%, and protects citizens from oscillations of Mr. Market. Can't say the same thing about that 401(k); all those Wall Street fat cats got fat off "your" retirement nest egg.

So, why is predicting life expectancy from today forward based on the accumulated data over the last X years silly? Because, just as the Wizard of Oz, those added years are the result of human intervention, sporadic and specific. Increase in life expectancy, whether at birth or 65 or some other age, is not a God given gift to humanity for just being on the planet for evermore years. Doesn't work that way. Most of the increase at birth is due to greatly diminished infant and child mortality, this due mostly to public health initiatives; vaccines and anti-biotics being discovered and made widely available. That's not going to happen again. One might even speculate that, with increasing wealth concentration, what had been widely available will become more restricted. With the loonies chirping about autism and vaccines and denying same to their spawn, we are finding increased (although not epidemic levels, yet) incidence of old diseases. And so on. Humanity, despite what some quants believe, isn't Brownian motion or a random walk. No, progress exists because humans change the rules, from time to time. And sometimes those changes benefit most of us rather the the 1% and we all live a bit longer. It isn't God that has provided us with longer lives than we had in 1900, it's us. As we've reached near the limit of our ability to stave off Father Doom, and, perhaps, our willingness to make this ability available to all, we will see a leveling (if not diminishment) of life expectancy at all ages.

No comments: