Thread by @wwwojtekk

Following up on the major discrepancy between NYT visualization and the published QJE paper (see the quoted thread). Below is the comparison of what is in the two sources. They generally don't line up and the discrepancy goes in the direction of the trend visible in NYT graphics

https://twitter.com/wwwojtekk/status/1182695438077562881

To illustrate - compare 1969, 1984, 1999, 2014

Just to be clear: PSZ (2018) is a peer-reviewed paper published in QJE, a top economics journal, gabriel-zucman.eu/files/PSZ2018Q…
The NYT visualization is based on data from the forthcoming book that's supposed to follow the same methodology with just some adjustments. Details unknown.

A few comments:
-PSZ data: gabriel-zucman.eu/files/PSZ2017A…, Table TG1
-NYT (h/t @davidawaldron): static01.nyt.com/newsgraphics/2…
- I had to merge categories in P0-50, P50-90 (NYT data) and P99-99.9 (PSZ) so that they match
- Top 400 and P99.999-P100 are not the same - 400 vs 2434 people in 2014

I don't know why the series do not match. DINA paper is supposed to be the conceptual source of all of it, it goes until 2014. Separating top 400 (speculative, but still) shouldn't have much of an effect elsewhere in the distribution. I'm puzzled. These are big differences.

Particular years surprise me. Here is 1998. NYT average tax rates are lower everywhere except tiny difference at the bottom. How that can be if we are just reallocating the same taxes and income?

There is too little income at the bottom and it does not seem that my rough construction of average tax rate can explain it (it varies from 0.24.2 to 0.28.5 by decile). This one is just mathematically puzzling.

Better ordering:

https://twitter.com/wwwojtekk/status/1183121913830346752

Updated 1998

https://mobile.twitter.com/wwwojtekk/status/1183121913830346752

Updated

https://mobile.twitter.com/wwwojtekk/status/1183121913830346752