Feasibility is a backwards looking concept. It seeks to answer the question: has anyone else achieved a target like this? To answer this question, we require historical data. Using this data the feasibility dimension can be quantified by looking at how peers have performed in a similar situation. If many peers were able to achieve a similar goal within the same time span, then that goal has a high feasibility. If few of the peers have ever come close to the target, then the feasibility would be low.

Typically, we look at three to five years. This depends however on the availability of data, whether we can find peers that have a long enough history (for feasibility), and whether the uncertainty grows too fast the further we look (for ambitiousness).

No, the feasibility assessment is only based on historical progress by other countries. For some indicators, recent developments such as technological improvements might have led to more possibilities to reach targets within smaller time spans and the assessment may therefore be considered a conservative estimate of what is feasible.

The average feasibility path characterizes the typical historical development of peer countries when they were at a similar level as the country of interest is now. It is calculated by taking the median of the peer country values in each year after the base year until the horizon.

To assess the ambitiousness of possible targets, we model a business-as-usual scenario that describes how the indicator may develop over the chosen time horizon. If the target is significantly better than the business-as-usual scenario, we can be confident that it would not be achieved without additional policy effort. We would classify such a target as high ambitiousness. If the target is within the estimated range of the business-as-usual scenario it would be achieved anyways, and we would classify the target as low ambitiousness.

Ambitiousness is concerned about whether a target falls within the range of a business-as-usual scenario or exceeds such expectations. An ambitious target would be significantly better than the business-as-usual scenario and require additional policy effort. Feasibility on the other hand is concerned with whether there is any historical precedence that a target like that can be reached. It looks for historical cases of progress similar to the progress needed for the country to reach its target. If many peer countries have made similar progress in the past as is necessary to achieve the target, then it would be highly feasible.

In principle any model could be used, as long as it is suitable to make short term predictions based on statistically sound and transparent methodology and provides uncertainty intervals around the estimate. We employ four models. The country model only relies on the country's own historical trends. The income and regional models leverage the common trends of other countries in the same income classification and countries from the same geographic region, respectively. We also offer the global model, which uses common trends of all countries available for the target indicator.

Comparable peers are countries that are similar to the country of interest based on their income classification and/or geographic region. The analysis needs at least five countries in the comparable peer group to proceed.

Benchmarkable peers a subset of the comparable peers. They are particularly relevant for the target country, since they are countries that at some point in the past that had a similar value as the most recent value of the indicator of the target country. In order to qualify as a benchmarkable peer, the country must have had at least one value within the tolerance (see next).

We aim to identify peer countries that once were at a level similar to the base year value of the country of interest. ‘Similar’ is context-dependent and therefore adjustable with the tolerance bandwidth. Tolerance specifies the acceptable deviation in historical data points of peer countries from the base year value of the country of interest. For example, if the country is at value X in the chosen base year, the model selects peer countries that have had values within X +/- tolerance level in any prior year(s). The larger the tolerance, the more peer countries will be selected.

The number of years prior to the base year doesn't affect the feasibility assessment and serves only for visualization purposes. The years after the base year do affect the feasibility assessment, by changing the interval. If your target lies five years ahead, it makes sense to set the year range to +5 to evaluate the feasibility of the final target and the intermediate trajectory to reach the target.

How far you can look into the future depends on the indicator. It is important to note that the feasibility concept only assesses historical developments. For some indicators, technological developments and economic conditions make it plausible to reach targets faster than historically possible. In that case, there is no historical precedence but that doesn't mean the target is infeasible to reach.

Let's say we are looking at a target that is set for Y years from the base year. If the peer country value that is similar to the base year value is less than Y years ago from the most recent data value, we cannot project the peer path for a Y years range. This means that there is not enough data. We can choose to include or exclude these kinds of peers. If we include them, we can have more possible peer countries and see more accurate feasibility intervals for intermediate targets. If we exclude them and only consider peers with full date range, we might end up with fewer peers and we use the same amount of data for the entire time interval.

If a peer country has had a similar level of development as the country of interest in multiple years, we select the occurrence that had the most similar value as the most recent value of the target country. This assures that each peer country only enters the feasibility assessment once and that the most relevant information is used.

The FAB dashboard is available for all indicators present on the ESG Data Portal. Some indicators are not available for all countries and all years. When data is missing for one or more years within a country, the data is interpolated to fill the gaps. We are currently working on a more sophisticated way to estimate missing data points using advanced machine learning methods.

All the data used on this dashboard is publicly available. You can find all data on the Sovereign ESG Data Portal, which the FAB Dashboard is part of.

• Blog post: Ambitious, yet feasible: Setting FAB targets for sustainable financing instruments
• Policy Research Working Paper: Could Sustainability-Linked Bonds Incentivize Lower Deforestation in Brazil’s Legal Amazon?

The FAB dashboard has been designed and developed by Dieter Wang with support from Philipp Kollenda, Veerle de Smit and Calvin Elikem under the supervision of Gianleo Frisari. The dashboard is part of the Sovereign ESG Data Portal and has been supported by the Global Platform for Sustainability