OECD updates country risk assessment model

The export credit guarantees issued by EKN and other Export Credit Agencies (ECA) serve as insurance against payment default on the part of a borrower. Like other insurances, the premium should reflect the level of risk involved in a transaction.

Since different countries represent varying levels of risk (to creditors), OECD has established a Country Risk Expert group (CRE) for the task of rating countries according to eight levels of country risk. Each level is associated with  a minimum premium fee for official export credit support, in order not to be seen as unfair trade support.

The setup forms a key building block in the broad OECD disciplines on Officially Supported Export Credits (see fact box below). To support the CRE in its rating process, a Country Risk Assessment Model (CRAM) was established in 1997. The model is developed and applied by the CRE.

At Credendo, the Belgian ECA commissioned by the OECD to administrate the CRAM model, Pascaline della Faille, Country and Sector Risk Manager, says, “The purpose of CRAM is to classify countries according to their risk.”

In this context, “country risk” refers mainly to the risk of payment default due to transfer and convertibility risk (the risk of a government imposing capital or exchange controls that prevent the conversion of local currency into foreign currency and/or transferring funds to creditors located outside the country). It also includes other measures or decisions of a foreign government which prevents repayments such as a default on government debt. Cases of force majeure (such as war, expropriation, revolution, civil disturbance, floods, earthquakes) may also apply.

Some 140 countries are classified annually by the CRE. High income OECD and high income euro area countries are not classified. Insurance premiums for transactions in these countries are subject to a separate set of market pricing disciplines.

A growing gap

The constantly changing global economic environment produces new types of risk that need to be reflected in country risk assessments. In addition, new data and improved statistical methods offer new opportunities for improving model performance.

Consequently, the methodology is subject to regular revisions with the goal of improving the accuracy of the model. “We also noticed a growing gap between country risk classification determined by the Country Experts and the model outcome, and began looking at ways to improve the model,” adds della Faille.

Credendo teamed up with representatives of the ECAs in Denmark (EIFO), Italy (SACE), and Sweden (EKN) to work on exploring tweaks and upgrades. “The team members brought their expertise in various fields and long practitioner experience to the table in the joint modernisation effort, but the expertise in quantitative modelling at Credendo was really instrumental in further developing the model,” says Johan Fredriksson, Senior Country Analyst at EKN.

The CRE rating process is supported by statistical modelling as well as input from the country experts, who always have the final say on classification. “No mathematical model can completely cover the complexity of a single country,” says Fredriksson. “We need people for that. But a strong predictive tool like CRAM is a valuable tool in the process. In particular, it sets a starting point for the experts’ further deliberations on factors that may not be captured in the model.”

Highlights of the recent revision include a stronger focus on governance and institutional factors, while risks related to the consequences of climate change have been assigned greater weight.

Input to the model comes in the form of data from a host of sources, such as IMF, the World Bank, and universities across the world. “It’s very important to use only well-reputed, non-biased sources,” says della Faille.

Verification against historical data

Running the model results against historical data going back several decades lets Credendo evaluate the predictive capability of the model. “This allows us to review the extent to which our classifications were accurate in predicting actual defaults and ECA’s payment experience,” says della Faille.

At EKN, Fredriksson agrees that adding institutional factors and governance to the model will improve its accuracy: “The single most important set of indicators of the new revision is related to governance and institutions, which adds a structural dimension. All in all, the new model will hopefully be able to flag risks at an even earlier stage than is presently the case.”

Country classifications using the newly approved CRAM model will be applied at the first meeting at the end of January 2024, but della Faille does not foresee a massive upheaval of current CRE classifications. “The overriding objective of the revision is to achieve a stronger predictive capability and a better tool for the human experts that have the final say. A better model reduces the need for analyst adjustments.

Multiple challenges

That aside, the revision did offer multiple challenges to the team. “We added a lot of new criteria and adjusted their respective weight, in order to improve the predictive capability and reduce the need for analyst adjustments. Statistically, we find the model to be very strong and verified according to established practices. We have been able to demonstrate that chosen indicators really improve the model performance.”

Fredriksson adds that, “we have institutionalized a more continuous update process to make sure we capture broad trends in economy, such as the emerging risks of non-payment associated with climate change.”

To ECAs and lenders, the country risk classifications are obviously useful as it set the minimum premium. But potential exporters who are seeking new export markets may use them to get an overview of prevailing levels of risk across geographies. The country classifications can be found on the OECD web page for export credits (see link below).

Finally, what’s the difference between CRAM-based country risk classifications and sovereign ratings issued by commercial rating institutes? The latter mainly refer to the government credit risk related to capital market instruments such as sovereign bonds. The CRE classifications deals in risk of defaults on trade-related payments, not only to the sovereign but to a wider set of buyers in a country.

“Obviously there is a substantial correlation between the two, however, debt default on part of a government need not entail the same for private borrowers, and vice versa,” says della Faille and adds another difference: “The CRAM country risk classifications represent a real commitment on part of the ECAs, since it directly affects the premium we are allowed to charge for our services.”

“Export credits also involve many other dimensions, such as the underlying business and supply relations between the exporters and the buyer,” adds Fredriksson.