Estimating tax avoidance.


New findings, new questions

There are now a range of estimates of the global scale of tax avoidance. These include:

* the $600 billion annual tax loss estimated by IMF researchers Crivelli et al. (2015; 2016), which divides roughly into $400 billion of OECD losses and $200 billion elsewhere;

* the $100 billion annual tax losses that UNCTAD's World Investment Report 2015 estimated for developing countries due only to conduit FDI investment through 'tax havens';

* the $100 billion to $240 billion globally that OECD researchers estimate;

* the $130 billion globally that we have estimated as annual losses due to avoidance by US multinationals only; and so on.

With the exception of the latter, which deals only with multinationals responsible for around 1/5 of global FDI, the estimates have not been broken down to country-level to show the underlying pattern. While the OECD research indicates losses in a range from 4% to 10% of corporate income tax revenues, the IMF researchers' estimates suggest that OECD countries may lose 2-3% of their total tax revenues, and lower-income countries much more: to 6-13%.

Petr Jansky and I have now reworked the IMF researchers' estimates to provide the country-level detail, in a paper published by UNU-WIDER today. We chose the IMF paper because it is the only one of the institutional researchers' efforts to date which has been published in a peer-reviewed journal. The researchers were kind enough to share their code, and we were able to replicate quite closely the original findings. We then revised the results by using the ICTD-WIDER Government Revenue Database, which provides significantly better revenue data, and these are our main findings:

* The global tax losses are estimated at around $500 billion (compared to $600 billion in the original);

* The pattern of greater relative intensity in lower-income countries (measured by losses/GDP or losses/tax revenue) is stronger in our results than in the original; but

* The methodology, and country-level findings, raise a number of questions.

Estimated Tax Loss ($bn) United state 188.8 China 66.8 Japan 46.8 India 41.2 Argentina 21.4 France 19.8 Germany 15.0 Pakistan 10.4 Indonesia 6.5 Philippines 6.4 Note: Table made from bar graph. The main questions are three. First, the methodology takes statutory tax rates as the determinant of profit-shifting. The authors do experiment with one set of effective tax rates, and we try with another, but without finding a...

To continue reading