Thomas Piketty has had a bad few days. The Financial Times has launched a critique of the data behind the French economist's bestseller "Capital in the twenty-first century." In such a sweeping study, errors are almost inevitable. But he has also fallen prey to sloppy spreadsheets.
Piketty deserves credit for making his background data publicly available. Even so, the Frenchman's spreadsheets would, in places, embarrass the greenest investment-banking analyst. Exhibit A is his habit of hard-coding adjustments into Excel formulas.
For instance, Piketty's tables include a column calculating the share of wealth owned by the top one per cent in France. For the period from 1810 to 1960, he pulls figures from various raw data tables and multiplies them by a hard-coded factor. This is 1.05 in all cases except for 1910, where, without explanation, it is 1.08. For 1970, he simply writes in a percentage for the wealth share, rather than referencing one of the sets of source data. More recent years involve both approaches.
More From This Section
Showing all inputs individually makes them easier to identify, better separates the logical steps involved in arriving at the final number, reduces the scope for errors, and makes mistakes easier to spot.
Though it's tricky to deal with information covering many different periods, Piketty's layout for dates in his spreadsheets is also a bit haphazard. This may have made it harder for him to catch data transcription errors - of which the FT seems to have found at least one.
It's not clear these problems are big enough to undermine Piketty's conclusions about inequality, wealth accumulation and economic growth. In an initial response, the French academic noted the diversity of the sources in his long-term and wide-ranging study, and said adjustments were required.
