In the past few days Thomas Piketty and a number of his less scrupulous defenders have taken to declaring the data dispute around Capital in the 21st Century a settled matter since the publication of Piketty’s retort to Chris Giles of the Financial Times.
I submit that it is anything but settled, and would draw your attention to five specific areas of his work that remain unresolved.
1. Piketty has yet to provide a clear, transparent, and accurately sourced description of how he calculates his wealth inequality measurements for the United States.
As I’ve documented here and here, Piketty migrates quite freely between at least four different data sources when he assembles his widely acclaimed estimate for the United States. He also does so in ways that are evident in his raw excel sheets, but obscured or entirely missing from his annotation – particularly his switches between estate tax and survey-based estimates of inequality around the 1970s. Neither Piketty’s technical appendix nor his addendum in response to Giles address this discrepancy.
2. Piketty’s methods for estimating the top 10% of wealth distributions are simplistic and non-transparent.
Although imperfectly constructed, most of Piketty’s wealth data for the top 1% is obtained from other sources such as surveys or tax data. He often speaks of parallel estimates for the top 10% though, in some cases suggesting very pronounced trends. Examination of Piketty’s excel files suggests that most of his top 10% figures are directly derived from his top 1% calculations, though in generally opaque and amateurish ways that simply add in very rough estimates or that attempt to calibrate the upper figure to additional sources that are not clearly cited and explained.
3. Piketty is being methodologically inconsistent in his tolerance for divergent techniques for the estimation of wealth inequality.
Wealth inequality is notoriously difficult to measure, though the two most common approaches attempt to estimate it from (a) tax data or (b) survey sampling. Piketty severely chastised the Financial Times for using a survey-based estimate of wealth distributions in the UK, which suggested less pronounced inequality trends than he claims in his book. He also strongly endorsed the methodological superiority of a tax-based estimate…for the UK.
Yet in constructing his own time series for the United States, Piketty quite freely migrated between tax-based and survey-based estimates by other scholars. He also displayed a preference-of-convenience for US survey data over US tax based estimates when the latter showed a flat trend during a period where his argument anticipated an uptick. He has yet to address this seemingly glaring methodological inconsistency.
4. Piketty’s historical data is still full of basic factual errors, unexplained gaps, and suspect methodologies.
Errors of this sort are numerous and ubiquitous in Piketty’s book. They range from small errors and oversights in dealing with historical data to major errors of interpretive significance to his thesis. A partial list of these problems include:
- Erroneous representations of US income tax rates
- Missing historical data for US estate tax rates
- Erroneous representations of historical US minimum wage rates
- Made-up figures for 19th century US tax revenue receipts
- Faulty averages of UK wealth inheritance transfers that confuse (or possibly manipulate) estimates for the wrong decades
- ADDENDUM (6/10/2014) – Use of questionable and possibly made-up data for the Soviet Union and other communist states, artificially distorting his global index for Capital/Income ratios.
None has been addressed or corrected by Piketty as of this writing.
5. Piketty is severely overstating the support for his conclusions found in the work of other scholars.
Piketty has largely attempted to sidestep the aforementioned problems in his US wealth data by claiming vindication in the results of a more recent study-in-progress by his frequent collaborators Gabriel Zucman and Emmanuel Saez. Although Piketty and many of his supporters have already accepted the validity of this study, it has yet to be published or subjected to academic peer review. It presently only exists in the form of a PowerPoint slideshow. A visual examination of its claimed results suggests their model might be skewed by one or more exogenous factors relating to their interest rate data. And while it has gone virtually unnoticed by Piketty’s defenders, Wojtek Kopczuk – another leading scholar of inequality measurement (and sometimes Saez collaborator) who is more familiar with the unpublished Saez-Zucman technique than the public viewers of its PowerPoint – has recently suggested that it has an underlying methodological fault.