Why do we include a scale weight in our index?

The Corporate Tax Haven Index (CTHI) rests on two components: a haven score, assessing countries' rules, laws and practices that attract corporate profit-shifting, and a scale weighting estimating how much activity multinationals have in each jurisdiction. It mathematically combines the two scores to produce a final index score for each jurisdiction, which is the basis for our index.

The weighting is necessary for several reasons.

Our ranking is designed to identify jurisdictions according to their overall global contribution to the problems of corporate tax avoidance, and in spurring the global race to the bottom that is steadily removing the tax burden from multinationals and shifting it onto everyone else's shoulders. So we seek to identify those jurisdictions where reforms to laws and practices would have the greatest effect.

The top 10 jurisdictions in our index, with an average haven score of 89 percent, account for almost 40 percent of the total reported foreign direct investment (which is our proxy for multinational's activity in a jurisdiction). If we ranked jurisdictions only by their haven score, the top 10 would have an average haven score of 99 percent, but they would account for less than 7 percent of the total reported foreign direct investment. (See the ranking based on haven scores only, here).

Some may argue that by including scale weights, our index "punishes" jurisdictions with large financial sectors. But the mathematical formula we use -- see here for details -- is designed to reduce the relative importance of the scale weighting in the final index scores. So a jurisdiction that improves its haven score is likely to improve its ranking, whether it hosts lots of foreign direct investment or not.

We reduce the scale weighting for two reasons. First, we want to give jurisdictions an incentive to clean up: the easiest and least painful way to do that is to clean up the haven score. That's why we emphasise it. The other reason is that while the haven scores have a relatively narrow range - between 39.05 and 100 percent -- the scale weightings diverge massively, between 0.0000016 and 12.9 percent. So we need to mathematically compress the scale weighting, so that it doesn't dominate the haven score.

More details on the the formula and the scale weight are included in the full methodology.