Research notes · Finance

Why do careful estimates of profit shifting differ by an order of magnitude?

What the authors found

Dharmapala reviews the measurement of profit shifting and finds the divergence between microeconomic estimates from firm data and macroeconomic estimates from national accounts is not a matter of better or worse technique. The two approaches draw the boundary between tax avoidance and ordinary behavioural response to taxes in different places, so they are measuring different things. He also documents that OECD countries substantially strengthened anti-avoidance rules between 2000 and 2014.

What our data adds

This database sits firmly on the macro side of the divide Dharmapala describes, and the note on the royalty ledger shows what that view looks like: Luxembourg paying 13.21 per cent of GDP for the use of intellectual property, Cyprus receiving 8.12 per cent. Those figures are dramatic and, by this paper's argument, systematically ambiguous — they include real licensing alongside transfer pricing, and no macro series can separate them. The microdata studies that can produce much smaller numbers for the same phenomenon.

What this cannot show

The most useful thing this note can do is warn against a figure derived from our own data. It would be easy to total the royalty flows through low-tax jurisdictions and present the result as the scale of profit shifting. That number would be a macro estimate with all the boundary problems Dharmapala identifies, and quoting it without them would be exactly the error the paper is about.

The data behind it

Related notes

• If a firm books a sale in a tax haven while the goods are sold somewhere else, what does that look like in the statistics?
• What happens to a country's creditworthiness when some of its debts are secret?
• If profits are booked in the wrong country, which national statistics are wrong?

Was this useful?

It answered my question Partly No

No account, no comment box. One click, counted and nothing else.