2020 Universal registration document and annual financial report - BNP PARIBAS 195

4Consolidated finanCial statements for the year ended 31 deCemBer 2020

4

Notes to the financial statements

As a backstop, when an asset becomes 30 days past due, the credit risk is deemed to have increased significantly since initial recognition and the asset is therefore placed into stage 2.

Forward Looking Information

The Group considers forward-looking information both when assessing significant increase in credit risk and when measuring Expected Credit Losses (ECL).

Regarding the assessment of significant increase in credit risk, beyond the rules based on the comparison of risk parameters between initial recognition and reporting date (see Significant increase in credit risk section), the determination of significant increase in credit risk is supplemented by the consideration of more systemic forward looking factors (such as macro-economic, sectorial and geographical risk drivers) that could increase the credit risk of some exposures. These factors can lead to tighten the transfer criteria into stage 2, resulting in an increase of ECL amounts for exposures deemed vulnerable to these forward looking drivers.

Regarding the measurement of expected credit losses, the Group has made the choice to use 3 macroeconomic scenarios by geographic area covering a wide range of potential future economic conditions:

■ a baseline scenario, consistent with the scenario used for budgeting;

■ an adverse scenario, corresponding to the scenario used quarterly in Group stress tests;

■ a favourable scenario, allowing to capture situations where the economy performs better than anticipated.

The link between the macroeconomic scenarios and the ECL measurement is mainly achieved through a modelling of the probabilities of default and deformation of migration matrices based on internal rating (or risk parameter). The probabilities of default determined according to these scenarios are used to measure expected credit losses in each of these situations.

The weighting of the expected credit losses under each scenario is performed as follows:

■ 50% for the baseline scenario;

■ the weighting of the two alternative scenarios is computed using a relationship with the position in the credit cycle. In this approach, the adverse scenario receives a higher weight when the economy is in strong expansion than in lower growth period in anticipation of a potential downturn of the economy.

In addition, when appropriate, the ECL measurement can take into account scenarios of sale of the assets.

Macroeconomic scenarios The three macroeconomic scenarios are defined over a three-year projection horizon. They correspond to:

■ a baseline scenario which describes the most likely path of the economy over the projection horizon. This scenario is updated on a quarterly basis and is prepared by the Group Economic Research Department

in collaboration with various experts within the Group. Projections are designed for each key market of the Group (France, Belgium, Italy, the United States and the eurozone) using key macro-economic variables (Gross Domestic Product GDP and its components, unemployment rate, consumer prices, interest rates, foreign exchange rates, oil prices, real estate prices, etc.) which are key drivers for modeling risk parameters used in the stress test process.

As from 31 December 2020, in addition to the geographic breakdown, prospective parameters are detailed at sector level to better reflect the heterogeneity of economic trajectories in conjunction with lockdown measures and the partial interruption in activity;

■ an adverse scenario which describes the impact of the materialisation of some of the risks weighing on the baseline scenario, resulting in a much less favourable economic path than in the baseline scenario. The starting point consists of a shock on GDP. This shock is applied with variable magnitudes, but at the same time to economies as the crisis is considered to be a global crisis. Generally, these assumptions are broadly consistent with those proposed by the regulators. The calibration of shocks on other variables (unemployment, consumer prices, interest rates) is based on models and expert judgment;

■ a favourable scenario which reflects the impact of the materialisation of some of the upside risks for the economy, resulting in a much more favourable economic path. To achieve an unbiased estimation of provisions, the favourable scenario is designed in such a way that the probability of the shock on GDP growth (on average over the cycle) is equal to the probability of the corresponding shock in the adverse scenario. The magnitude of favourable GDP shocks generally corresponds to 80%-95% of the magnitude of adverse GDP shocks. Other variables (unemployment, inflation, interest rates) are defined in the same way as in the adverse scenario.

Factoring the specific features of the health crisis in the baseline scenario After a historical drop in the first half of 2020, followed by a marked rebound in the third quarter, activity fell again in the fourth quarter, reflecting stricter measures implemented by governments (including lockdowns in some European countries) due to the increase in Covid-19 cases. However, this fall is less pronounced than the shock in the second quarter thanks to government and central bank measures taken to limit the impact of the crisis on the economic activity. In 2021, a favourable trend in growth rates is expected, reflecting: a) a mechanical catch-up; b) government and central bank measures; c) positive developments in terms of vaccines. The pace of growth is expected to normalise in 2022 and 2023, and activity is expected to return to pre-crisis levels in 2022 in most mature economies.