Slide 1

GGFI Methodology

GGFI will assess ratings for financial centres calculated by a ‘factor assessment model’ that uses two distinct sets of input:

  • Instrumental factors: objective evidence of both environmental credentials and green finance sought from a wide variety of comparable sources. Click here for details.

    For example,; information on  how  a given financial centre’s activity contributes to lowering carbon dioxide/GHG emissions  evidence of a financial centre’s commitments and achievements on ESG disclosure, and about the trading and regulatory environment of green finance, eg  for instance volume trades in carbon or green bonds, as well as more global data such as the Ease of Doing Business Index (supplied by the World Bank), the Government Effectiveness rating (supplied by the World Bank) and the Corruption Perceptions Index (supplied by Transparency International).


Not all financial centres will be represented in all the external sources, and the statistical model will take account of these gaps.

  • Green financial centre assessments: evidence of how a centre’s environmental market is viewed will be captured through an online questionnaire.

For the first edition of GGFI, analysis will be carried out on a range of financial centres across the world. Responses will be collected via an online questionnaire which will run continuously. A link to this questionnaire will be emailed to our growing list of respondents, which includes green finance professionals and experts, at regular intervals and other interested parties can fill this in by accessing these web pages;

In order to avoid home centre bias, the centre that a respondent is based in will be excluded from the assessment.

The financial centre assessments and instrumental factors will be used to build a predictive model of green financial centres using a support vector machine (SVM).

SVMs are based upon statistical techniques that classify and model complex historic data in order to make predictions on new data. SVMs work well on discrete, categorical data but also handle continuous numerical or time series data. The SVM that will be used for GGFI will provide information about the confidence with which each specific classification is made and the likelihood of other possible classifications.

A factor assessment model will be built using the centre assessments from responses to the online questionnaire.  The model will then predict how respondents would have assessed centres with which they are unfamiliar by answering questions such as:

If a respondent gives Singapore and Sydney certain assessments then, based on the instrumental factors for Singapore, Sydney and Paris, how would that person assess Paris? 

Predictions from the SVM are re-combined with actual financial centre assessments (except those from the respondents’ home centres) to produce the GGFI – a set of green financial centre ratings.

Over time, GGFI will be dynamically updated, either by updating and adding to the instrumental factors or through new green financial centre assessments. These updates will permit, for instance, a recently changed low carbon index to affect the rating of the centres.

The process of creating the GGFI is outlined diagrammatically below:

The strengths of this approach are:

A wide range of indices can be used for each instrumental factor;

  • A strong international community of respondents can be developed as the GGFI progresses;
  • Sector-specific ratings are available using the business sectors represented by questionnaire respondents. This makes it possible to rate a centre as highly influential in carbon trading (for example) whilst less advanced in enhanced analytics (for instance).
  • The factor assessment model can be queried in a ‘what if’ mode - “how much would London carbon emissions costs need to increase in order to enhance London’s ranking against Singapore?”

Part of the process of building the GGFI will be sensitivity testing to changes in instrumental factors, the refining of the questionnaire, the development of the rating community and testing of the accuracy of predictions given by the SVM against actual assessments.