The consultative document can be found at http://www.bis.org/publ/bcbs201.htm
At the heart of the proposed method to identify the G-SIBs (Globally Systemically Important Banks) are the list of indicators that will be used to measure the importance of each bank.
The way the score is then calculated is explained in paragraph 17:
17. For each bank, the score for a particular indicator is calculated by dividing the individual bank amount by the aggregate amount summed across all banks in the sample for a given indicator.6 The score is then weighted by the indicator weighting within each category. Then, all the weighted scores are added. For example, the size indicator for a bank that accounts for 10% of the sample aggregate size variable will contribute 0.10 to the total score for the bank (as each of the five categories is normalised to a score of one). Similarly, a bank that accounts for 10% of aggregate cross-jurisdictional claims would receive a score of 0.05. Summing the scores for the 12 indicators gives the total score for the bank. The maximum possible total score (ie if there were only one bank in the world) is 5.
First, the sample banks must be chosen beforehand. This is necessary, as calculation of interconnectedness in a system requires the system to first be defined. The initial 73 banks were chosenbased on size and supervisory judgement. The total score of these 73 banks will be computed, then the 27 or 28 with the highest scores will be deemed G-SIBs. The cutoff is based on observed clustering of the scores. (27 or 28 banks have considerably higher scores than the rest, with apparent gap) Lastly, the G-SIBs are then divided into 4 equal sized score buckets, which then determine the additional common equity that a bank must hold in addition to Basel III requirements.
This method must causes the world's largest banks some concern, as not only they will have to hold more capital, it is also not easy to estimate their future positions in the G-SIBs queue and thus how much the additional buffer will be. There are parameters that will change over time even when the guideline had been finalized. An example of this is the 27/28 cutoff. I think this is a necessary complexity, for the type of estimation the committee is trying to produce.
Is indicator-based method the way to go, or should BCBS holds on for model-based method to get more mature? Will the additional capital requirements do their job in (sufficiently) strengthening the world's largest banks and dis-incentivise them from getting even bigger? I'm sure there will interesting discussions to follow.