By Pavlina Tcherneva (revised 10/10/14)
Winship has produced a rambling and insulting retort to a very thorough and thoughtful response to his false critique of my original inequality chart.
Here are my brief replies to his last post.
- Let me say it again. His business cycle chart is incorrect. One cannot make any sense of what is happening to the distribution of income growth over time by the way he reports his periods. To use his own words, it does not “convey history correctly”.
You can be the judge of whose method is better (his or mine) in calculating the business cycles, but even if you use his method, his results are wrong. See herewhere I present two business cycle charts (using both methods). Both methods yield completely different and more depressing trends than in his chart. It is unclear how he’s chosen his periods.
In a very disingenuous way, Winship, passes the hot potato and implicates Saez and Piketty in his own error, arguing that they do the same thing in Table 1 here. This is incorrect. Saez and Piketty’s table shows one whole time period and then several discrete periodsof various recessions and expansions and, unlike Winship, they have identified their expansions and recessions correctly. They do not report the history of all business cycles (go ahead, check).
- Winship wants an apology from me about incorrectly claiming that his charts exclude capital gains data. First, I do not claim that his actual data exclude them. Second, Winship used a ton of space to argue how hard it was to measure capitals gains and how, in his words, “these problems argue for excluding taxable realized capital gains”.
So I challenged “why should we exclude them?” and have made a strong argument that the proper correction to Piketty’s data is to include all forms of income from wealth (and provided examples how to do it).
So what exactly is Winship saying? He writes a whole section trying to convince his readers that because of difficulties with capital income measurement, we are better off excluding them, and then want us to believe that he’s actually making the case for including them! Which is it? Do we include them or not?
As a point of logic, excluding them would make the numbers look better, whether he actually does it or not. That’s the essence of my argument. The burden is on Winship to explain to his readers that he is fact including them and why he does it, after making such a long case about the problem with counting them.
- The bottom line is this: if his ‘money chart’ falls so out of line with the global research program on inequality (showing an improvement in the way gains are distributed over the last 25 years), he should have asked himself “what am I doing wrong here?”
- He has neither corrected my original chart, nor has he produced an enlightening alternative.
- He has not addressed any of my essential points about capital
- And as I said in my last post, sweeping the discussion of capital gains under the rug serves the narrative of the inequality deniers—and the failure to engage me on the proper corrections accounting for stock options, carried interest, tax treatments and other government transfers to the wealthy, only reinforces his ideological culpability.
As tantalizing as this debate may have been for our readers, at least in my case, I like to spend my time on research that better illuminates the problems of our time.
We have serious problems to address and we need serious people to address them.
p.s. And by the way, his whole argument that “the rate at which it has risen hasn’t steadily grown” is just bizarre. Just because he can take a derivative, doesn’t mean that he is making any meaningful contribution to the discussion. Whoever talked about the acceleration in the gains captured by the top 10%?
Winship’s approach is symptomatic of the Forbes disease.
They devolved into apologists for a corrupt self-serving system, from a once-respectable publication that actually generated insights.
I don’t understand your point about the Saez-Piketty table 1.
They show the change for the whole period, and then they break it up into five sup-periods that cover the complete period. You can see that if you multiply together the percent changes in the sub-periods, you’ll recover the total change for the whole period (with some small rounding errors). There doesn’t appear to be any double counting going on.
That the years at the end of each period appear twice as a matter of labeling convention. For example, the year 2000 appears in both the sub-period 1993-2000 and 2000-2002. The growth in the 2000-2002 period actually happens in 2001 and 2002, not in 2000 (assuming you are using full-year-end values for income). But to calculate the change you compare income in 2002 to income in 2000, even though the growth itself occurs in 2001 and 2002.
Can you clarify your point or explain if you think I’m not understanding this right?
The point is, whichever way we interpret what Winship was doing, it is clear he was cherry picking his numbers.
I’m just trying to understand the “double reporting” issue, which is different than whether the periods were cherry-picked. I still don’t understand what you mean by “double reporting.” Were Saez and Piketty double reporting under your definition, and if not, how is what they did different than what Winship did? This seems like it should be a straightforward thing to resolve.
Thank you for the clarification question. No, they’re not doing the same “double reporting”. What you said in your first comment was correct about labeling convention and when growth occurs. Let’s use your example: in a 2000-2002 period, the growth rate actually occurs in 2001 and 2002, not in 2000. But what I think he’s doing is that he’s reporting the growth rate that occurred in 2000 in both of his two business cycle calculations. That was the only way I could make sense of his numbers.
I actually tried replicating his chart using his exact same dates and the Piketty spreadsheet we both used. I could not reproduce his results for any of the business cycle periods. All of his numbers looked much better in favor of the bottom 90%, so I concluded he must be double reporting or measuring growth rates in peak years twice.
This is why I strongly objected in my first blog and concluded that something was fishy. I see that my language could have been more precise and I have made a revision to the blogs to clarify this issue.
Also, if you go to my earlier blog, you will now see two business cycle charts – one measures them from post-peak-to-peak year (as I did them originally) and the other measures them from peak-to-peak. I did this in response to his objections on the proper measurement of a business cycle. It’s best if readers have both and decide which one is ‘better’. Both of them however, produce completely different results from his and confirm my original “expansions” chart.