Piketty’s Backtrack

[Dreadnaught]

Over-under that Piketty gets a Nobel prize in the next 20 years?

This is like all the socialist-tourism kids who flocked to Venezuela after 2003 to tour the triumph of Chavez. I don't hear them talk about Venezuela these days.

Why Thomas Piketty’s Revisions Don’t Fix His Book

http://blogs.wsj.com/washwire/2015/0...-fix-his-book/

By SALIM FURTH

A year after the English-language edition of “Capital in the Twenty-First Century” dominated American bestseller lists, Thomas Piketty is trying to cover his retreat with an article aimed at academic economists who have largely rejected the book. In the new article, Mr. Piketty tries to guide people toward his tome’s stronger points and away from its weaknesses, which include his cavalier use and abuse of data.


Mr. Piketty reiterates, for example, that the rise of labor inequality in the U.S. has very little to do with his central thesis, which is that when the interest rate (“r”) is greater than the growth rate (“g”), wealth inequality rises.

Sections II and IV of Mr. Piketty’s article say that simple versions of economic models do not support the claims he made in “Capital.” He argues–without using enough math to be falsifiable–that the addition of some tweaks could make the models give the results he desires. He does not say that in his book he used simple versions of the models.

One important failure in Mr. Piketty’s book was that it relied on substitutability between capital and labor that is far beyond the range supported by data. In a 2014 paper, Brent Neiman and Loukas Karabarbounis offered a defense of high substitutability (though not necessarily as high as Mr. Piketty suggests it is), based on an economy with multiple sectors. In his new article, Mr. Piketty borrows their idea, which was not available when he wrote his book. Of the approach used throughout “Capital,” Mr. Piketty now writes that it is “not [his] favored interpretation of the evidence.”

In other ways, Mr. Piketty remains consistent. His argument for why the inequality “r > g” should increase wealth inequality is tweaked in Section II but still has a central flaw: He does not deal with the fact that r and probably g are outputs of the same economic processes that create wealth.

But even the predictive value of that famous inequality goes under the revisionist’s knife:

Mr. Piketty in 2015: “I do not view r > g as the only or even the primary tool for considering changes in income and wealth in the twentieth century, or for forecasting the path of inequality in the twenty-first century.” (Page 1)

Mr. Piketty in 2014: “[T]here is every reason to believe that r will be much greater than g in the decades ahead . . . clearly all the ingredients are in place for the top centile and thousandth of the global wealth distribution to pull farther and farther ahead of the rest.” (Page 463)

“Capital in the Twenty-First Century” contained a lot of 19th- and 20th-century history, which Mr. Piketty now says was his primary topic. But the book also made strong claims about the future. If Mr. Piketty had wanted readers to take his predictions and policy recommendations less seriously, perhaps his book title should not have included the words “Twenty-First Century.”

Salim Furth is senior policy analyst in macroeconomics at the Heritage Foundation’s Center for Data Analysis. He is on Twitter: @salimfurth.

[wiggin]

Hmm. On the one hand, I agree that Piketty is guilty of overreach when it comes to his conclusions. On the other hand, I think that to completely dismiss his work is premature. He had the unenviable task of trying to make sense of some extremely dirty data sets in his analysis. I have no doubt that there are serious concerns with some of the assumptions he made in his models, which is inevitable in this sort of long-term analysis of complex trends. This means that his approach and conclusions are certainly up for debate, but it doesn't mean he's wrong.

There's a reasonable debate to be had about the return on capital vs. that on labor, and understanding those trends will indeed provide some useful insight into a variety of factors, including inequality. Piketty added some useful thoughts and rigor to the discussion, even if they were flawed. In fact, there are few people who would even take on such a challenging proposition (getting and normalizing those data over two centuries is somewhere between challenging and impossible). I applaud him for his effort, and view his book more as a 'working paper' than as a definitive treatise.

That being said, he certainly has gotten a bit of an ego trip from his book's unexpected success, and has done a poor job of handling thoughtful criticism of his analysis and conclusions. I don't think we should throw out the intellectual baby with the bathwater, though.

[Loki]

When an author makes significant mistakes with his data, the burden of proof is on him to show that the rest of the data is solid. In fact, I'd want someone else to confirm his hypotheses using similar data and statistical techniques.

[wiggin]

When an author makes significant mistakes with his data, the burden of proof is on him to show that the rest of the data is solid. In fact, I'd want someone else to confirm his hypotheses using similar data and statistical techniques.

I'm not sure it's 'mistakes' with his data (though there were some) so much as assumptions in his normalization and extrapolation from incomplete and mismatched data sets.

That being said, Piketty at least follows typical economics procedure and made reams of data available for others to peruse - that's primarily how all of this criticism has been leveled, by people taking a careful look at his data and methods. It's a lot more verifiable (and disprovable) than most books purporting to show some big trends that affect the world (a la Gladwell).

It may be that given more time, data, and research Piketty's underlying theory about the returns on capital and labor will be definitively disproven. More likely, they'll be incorporated into a broader and more complex story that accounts for rising inequality without completely eliminating his basic idea. That's how science (even a social science) works. Just because some of your analysis or data is questionable doesn't mean all of it is.

[Loki]

A theory needs to be proven, not disproven. Given his shoddy evidence, we have no reason to take his theory seriously until he (or someone else) provides better evidence for it. I'm not saying that we won't find that parts of his theory have some validity to it, but we have no reason for believing that at this time. It's not enough for a theory to sound good or plausible.

The fact that he's not a pseudo-scientist is not a sufficient reason to take his results seriously; it's a reason to not dismiss them out of hand.

[wiggin]

Fair enough. I just don't like demonizing an academic who tried to work with some extraordinarily challenging data sets. He may have been biased, or sloppy, or wrong... but he tried to put together a very thoughtful approach to understanding income inequality, and tried to base it on data. I have no idea if he's right or wrong - I suspect given the chorus of criticism from some economists, he's almost certainly not 100% right - but I think he's made a worthwhile theoretical and analytical contribution to the discussion.

Long term trends like this are notoriously difficult to tease out from shoddy and incomplete data, but they are also the only way to get a better theoretical underpinning to economics (rather than bouncing from cycle to cycle and coming up with bespoke explanations for each anomaly). I applaud the attempt, and I'm betting that all sorts of research - critical and supportive - will be performed because of the conversations he has sparked. That can't be a bad thing.