Saturday, May 3, 2014
Piketty Review from Nassim Taleb: The Pikettistas' Reasoning Error
This is a great review from Taleb regarding Piketty's "Capitlal:
1) I am appalled to see how economists are making the mistake of interpretation in the Piketty craze (which we showed contains questionable measurement --note: the paper is about measurement & we've been working on it for a long time; it is not about Piketty but we believe that this applies to his work). His measure of inequality could be inconsistent across samples, so under fat tails, with his measure,Average(Inequality(A)+Inequality(B)) < Inequality(A+B). It can rise over time as populations get larger (under fat tails), or rise simply from model error or breadth of the measurement.
2) But the problem is more severe than his sloppy empiricism, it is at the purely logical level: if the process is fat tailed, then wealth is generated at the top, which means increases in wealth lead to increases of measured inequality. Within population wealth creation is a series of small probability bets. So it is natural that the pool of wealth (measured in years of spending, as Piketty does) increases with wealth. Consider 100 people, for a tail α that corresponds to our measurement, the additional wealth should come come 1 person, which the remaining bottom 50 contributing nothing. It is not a zero sum gain: eliminate that person, and there will be almost no wealth increases. In fact the rest are already benefiting from the contribution of the minority. (I do not consider overpaid corporate managers and bankers with no skin-in-the-game and such empty suits as wealth contributors, but rather the product of warped incentives; I am talking about real people, risk takers like entrepreneurs).
A wealth tax meaning to punish the wealth generator is absurd: since the payoff is severely clipped on the upside, it would be a lunacy to be a risk taker with small probability bets, with wins of 20 (after tax) rather than 100, then disburse all savings progressively in wealth tax. The optimal strategy is to go become an academic or a French-style civil servant, the anti-wealth generators. To see the cross-sectional problem temporally: Compare someone with lumpy payoffs say an entrepreneur who makes $4.5 million every 20 years to a professor like Krugman who earns the same total over the period ($225K in taxpayer-funded income). The entrepreneur over the VERY SAME income ends up paying 75% in taxes, plus wealth tax on the rest while the rent-seeking tenured academic who doesn't contribute to wealth formation pays, say 30%.)
The problem with economists is that they are not (with very few exceptions) familiar with fat tails and make general statements that violate the true probabilistic payoff. In Mediocristan changes over time are the result of the collective contributions of the center, the middle. In Extremistan these changes come from the tails. Sorry, if you don't like it but that is purely mathematical
3) Worry about skin in the game, not inequality. Worry about equality in opportunity not outcome. Worry about the powerful corporations taking over the system via lobbyists and blocking artisans. Worry about the class of privileged mandarins-WNSITG (with no skin in the game) taking over the system via "grandes ecoles"...
PS- The top is unstable. In The Black Swan I explain that the "fairness" of the fat tailed world is that, in the absence of government intervention to support the powerful, the wealthy and powerful becomes more fragile and likely to be displaced. The top is the US is not the same in 2014 as it was in 1984, and it is a healthy system. In places with cronism (Europe), the top is sticky.
PPS- Asset Price Inflation. It would be interesting to figure out how much of the wealth increase (in the past couple of decades) came from asset price inflation, mainly real estate, rather than accumulation. 400% in addition to consumer price inflation? And this is largely the aim and result of low-interest rate policies by central banks since 1987, in the aftermath of Black Monday. The ratio Kapital/NonKapital exploded right there... And had Piketty looked at his "inequality" and Debt-to-GDP he would have perhaps seen the real evil: asset inflation with-stop-from-Central-banks.
PPPS- Rething about Piketty's book, and his "data", one wonders whether pages and pages of correlation-implies-causation can sway economists when the message fits their agenda (dispossess the risk-taking money by nonrisktakers is something that dissgusts me) but more rigorous work can leave them indifferent.
PPPPS- One reason I left data out of The Black Swan (except for illustrative purposes) is that it seems to me that people flood their story with data in the absence of logical argument. Just a little bit of data is needed when one is right (disconfirmatory empiricism).