21. Net expected credit losses

The allowance for expected credit losses is recognized in the financial statements in the following manner:

• Financial assets measured at amortized cost: the allowance reduces the gross carrying amount of the financial asset; changes in the allowances amount are recognized in the income statement;
• Off-balance sheet liabilities of a financial nature and financial guarantees: the allowance is presented as a provision under liabilities; changes in the provisions amount are recognized in the income statement;
• Financial instruments measured at fair value through other comprehensive income: the carrying amount of assets recognized at fair value is not additionally decreased by the allowances; however, each change in the measurement is divided into the impairment component, which is recognized in the income statement, and the component relating to other changes in the fair value measurement, which is recognized in other comprehensive income;
• Financial assets measured at fair value through profit and loss: no allowances for expected credit losses are recognized.

With regard to impairment, the Group applies the concept of expected losses.

The impairment model is applicable to financial assets that are not measured at fair value through profit or loss, comprising:

• debt financial instruments comprising credit exposures and securities;
• lease receivables;
• other financial assets;
• off-balance sheet financial and guarantee liabilities.

Expected credit losses are not recognized for equity instruments.

Impairment allowances for exposure reflect 12-month or lifetime expected credit losses on such exposures for a given financial asset.

The time horizon of an expected loss depends on whether a significant increase in credit risk occurred since the moment of initial recognition. Based on this criterion, financial assets are allocated to 3 stages:

• Stage 1 – exposures in which the credit risk is not significantly higher than upon initial recognition and no evidence of impairment is found;
• Stage 2 – exposures in which the credit risk is significantly higher than upon initial recognition, but no evidence of impairment is found;
• Stage 3 –assets in respect of which evidence of impairment is recognized, including assets granted or purchased with evidence of impairment recognized (upon being granted or purchased).

A material increase in credit risk is verified according to the likeliness of default and its changes with respect to the date of originating the loan.

The Group uses a model based on a marginal PD calculation, i.e. the probability of default in a given month, to assess a material increase in credit risk for mortgage exposures and other retail exposures. This probability depends on the time that has passed from originating the exposure. This enables reflecting the differences in credit quality that are typical of exposures to individuals over the lifetime of the exposure. The marginal PD curves were determined on the basis of historic data at the level of homogeneous portfolios, which are separated according to the type of product, the year of their origination, the loan currency and the credit quality at the time of origination. The marginal PD is attributed to individual exposures by scaling the curve at the level of the portfolio to the individual assessment of the exposure / Customer using application models (using data from loan applications) and behavioral models. The Group identifies the premise of a material increase in risk for a given exposure by comparing individual PD curves over the exposure horizon as at the date of initial recognition and as at the reporting date. Only the parts of the original and current PD curves which correspond to the period from the reporting date to the date of maturity of the exposure are compared as at each reporting date. The comparison is based on the average probability of default over the life of the loan in the period under review adjusted for current and forecast macroeconomic indicators.

The result of this comparison, referred to as α  statistics, is referred to the threshold value above which an increase in credit risk is considered material. The threshold value is determined on the basis of the historical relationship between the values of the α statistics and the default arising. In this process the following probabilities are minimized:

• classification into a set of credit exposures with a significant increase in the level of credit risk (based on the α statistic), for which no event of default took place during the audited period (type I error)
• non-classification into the set of credit exposures with a significant increase in the level of credit risk (based on the statistics) for which an event of default occurred during the audited period (type II error).

According to data that is applicable at the end of 2020, an increase in the PD parameter of at least 2.6 compared to the value at the time of its recognition in the Group’s accounting records in respect of mortgage exposures and an increase of at least 2.5 in respect of other retail exposures constitutes a premise of a significant deterioration in credit quality (unchanged compared to end of 2019).

With respect to credit exposures for which the current risk of default does not exceed the level provided for in the price of the loan, the results of the comparison of the probability of default curves as at the date of initial recognition and as at the reporting date do not signify a material increase in credit risk.

The Group uses a model based on Markov chains to assess material increases in credit risk for institutional customers. Historical data is used to build matrices of probabilities of Customers migrating between individual classes of risk that are determined on the basis of the Group’s rating and scoring models. These migrations are determined within homogeneous portfolios, classified using, among other things, customer and customer segment assessment methodologies.

An individual highest acceptable value of the probability of default is set for each class of risk and portfolio on the date of the initial recognition of the credit exposure, which, if exceeded, is identified as a material increase in credit risk. This value is set on the basis of the average probability of default for classes of risk worse than that at initial recognition of the exposure, weighted by the probability of transition to those classes of risk in the given time horizon.

In accordance with the data as at the end of 2020 and 2019, the minimum deterioration in the class of risk which constitutes a premise of a material improvement of the credit presented compared to the current class of risk were as follows:

The Group uses all available qualitative and quantitative information to identify the remaining premises of a material increase in credit risk, including:

• restructuring measures introducing forbearance for a debtor in financial difficulties;
• extending the period for the repayment of a significant amount of principal or interest by more than 30 days;
• identified early warning signals as part of the monitoring process, suggesting a material increase in credit risk;
• a significant increase in the LTV ratio;
• an analyst’s assessment according to an individual approach;
• quarantine for Stage 2 exposures, which have not shown premises for impairment in the previous 3 months.
• filing for consumer bankruptcy by any of the joint borrowers;
• transferring the credit exposure to be managed by the Group’s restructuring and debt collection units.

The premise for the impairment of a credit exposure is, in particular:

• extending the period for the repayment of a significant amount of principal or interest by more than 90 days;
• a deterioration in the debtor’s economic and financial position during the lending period, expressed by the classification into a rating class or class of risk suggesting a material risk of default (Rating H);
• the conclusion of a restructuring agreement or the application of relief in debt repayment, which is forced by economic or legal reasons arising from the customer’s financial difficulties (until the claim is recognized as remedied);
• filing a motion for the debtor’s bankruptcy, placing the debtor into liquidation or the opening of enforcement proceedings with respect to the debtor.
• declaration of consumer bankruptcy by any of the joint borrowers to the list of premises of impairment.

In accordance with the Regulation (EU) No 575/2013 of the European Parliament and of the Council of 26 June 2013 on prudential requirements for credit institutions and investment firms (“CRR”), the Group defines a state of default if it assesses that the debtor is unable to repay the loan liability without resorting to exercising the collateral or if the exposure is overdue more than 90 days. The premises of default are identical to the premises for impairment of the exposure.

The model for the calculation of the expected credit loss is based on applying detailed segmentation to the credit portfolio, taking into account the following characteristics at product and customer level:

• type of credit product;
• currency of the product;
• year of granting;
• assessment of risk of the customer’s default;
• the customer’s business segment;
• method of assessing the customer risk.

The Group calculates expected credit losses on an individual and on a portfolio basis.

The individual basis is used in respect of individually significant exposures. The expected credit loss from the exposure is determined as the difference between its gross carrying amount (in the case of an off-balance sheet credit exposure – the value of its balance sheet equivalent) and the present value of the expected future cash flows, established by taking into account the possible scenarios regarding the performance of the contract and the management of credit exposure, weighted by the probability of their realization.

The portfolio method is applied to exposures that are not individually significant and in the event of a failure to identify premises of impairment.

In the portfolio method, the expected loss is calculated as the product of the credit risk parameters: the probability of default (PD), the loss given default (LGD) and the value of the exposure at default (EAD); each of these parameters assumes the form of a vector representing the number of months covering the horizon of estimation of the credit loss.

The Group sets this horizon for retail exposures without a repayment schedule on the basis of behavioral data from historical observations. The loss expected both in the entire duration of the exposure and in a period of 12 months is the sum of expected losses in the individual periods discounted using the effective interest rate. The Group adjusts the parameter specifying the level of exposure at the time of default by the future repayments arising from the schedule and potential overpayments and underpayments to specify the value of the asset at the time of default in a given period.

The calculation of expected credit losses encompasses estimates of future macroeconomic conditions. In terms of portfolio analysis, the impact of macroeconomic scenarios is taken into account in the amount of the individual risk parameters. The methodology for calculating the risk parameters includes the study of the dependencies of these parameters on the macroeconomic conditions based on historical data. Three macroeconomic scenarios based on the Bank’s own forecasts are used for calculating the expected loss – a baseline forecast with a probability of 75% and two alternative scenarios, with a probability of 5%(optimistic) and 20% (pessimistic). The scope of the forecast indicators includes the GDP growth index, the rate of unemployment, the WIBOR 3M rate, the LIBOR CHF 3M rate, the CHF/PLN exchange rate, the property price index and the NBP reference rate. The final expected loss is the weighted average probability of scenarios from expected losses corresponding to individual scenarios. The Group ensures the compliance of the macroeconomic scenarios used for the calculation of the risk parameters with macroeconomic scenarios used for the credit risk budgeting processes. The baseline scenario uses the base macroeconomic forecasts. The forecasts are prepared on the basis of the quantitative models, taking into account adjustments for the presence of one-off events.

The extreme scenarios apply to cases of so-called internal shock, as a result of which the so-called external variables (foreign interest rates) do not change with respect to the baseline scenario. The extreme scenarios are developed on the basis of a statistical and econometric analysis, i.e. they do not reflect the events described, but the forecast path. Two scenarios are identified, optimistic and pessimistic. The share of the scenarios for the GDP path that falls between the optimistic and the pessimistic scenario is referred to as the probability of the baseline scenario. Such an assumption is used to forecast GDP growth, using a potential rate of growth of the Polish economy that varies over time, calculated with the use of quarterly data provided by the Central Statistical Office. The values of other macroeconomic variables used in the scenarios (rate of unemployment, property price index) are estimated after the extreme paths of GDP growth are defined.

The rate of unemployment is calculated on the basis of the quantified dependence on the difference between GDP growth and the potential rate of economic growth. The result is adjusted for significant structural changes taking place in the Polish economy, which are not encompassed by the quantitative model, in particular:

• the ageing of the Polish population (and the appearance of unsatisfied demand for labour, which will limit the scale of increase in the rate of unemployment in a situation in an economic downturn);
• the Polish labour market is nearing full employment (restrictions of supply mean that there is increasingly less space for a further decline in the rate of unemployment);
• the inflow of immigrants (only partly included in the official statistics).

The level of the property price index is set on the basis of changes in GDP, taking into account the conditions of supply and demand on the market based on the data and trends presented by the NBP in the publication “Information on housing prices and the situation on the residential and commercial property market in Poland” and the Group’s own analyses. The forecasts of WIBOR and LIBOR deposit rates are mainly prepared on the basis of assumptions regarding central bank interest rates. The CHF/PLN exchange rate is a cross rate of the EUR/PLN and EUR/CHF exchange rates. Its forecasts are a combination of the forecasts for these two rates. The EUR/PLN and EUR/CHF forecasts are prepared on the basis of a macroeconomic analysis (current and historical) based on econometric methods, as well as on a technical analysis of the financial markets.

Both the process of assessing a material increase in credit risk and the process of calculating the expected loss are conducted monthly at the level of individual exposures. They use a dedicated computing environment that allows for the distribution of the results to the Group’s internal units.

The Group has separated the portfolio of financial assets with low credit risk by classifying financial instruments for which the average long-term default rate does not exceed the probability of default specified by the rating agency for the worst class investment rating. This portfolio includes, in particular, exposures to banks, governments, local government entities and housing cooperatives and communities.

The impact of an increase/decrease in estimated cash flows for the Group’s loans and advances portfolio assessed for impairment on the basis of an individual analysis of future cash flows arising both from repayments and foreclosure of collaterals, i.e. the exposures for which an individual method is applied and the impact of an increase/decrease in the portfolio parameters for the Group’s loans and advances portfolio assessed on a portfolio basis is presented in the table below:

The table below presents the estimated sensitivity of the level of allowances for expected credit losses to macroeconomic conditions, calculated as the change in the level of allowances for expected credit losses related to not-impaired exposures resulting from realisation of particular macroeconomic scenarios as at December 31, 2020 and December 31, 2019. Before the start of the COVID-19 pandemic, the Bank applied a model assuming that the dependence of the allowance level changes on the interest rate changes was the strongest statistically – thus the 2019 optimistic scenario had a negative impact on the allowance level.

The tables below present the adopted forecasts of the main macroeconomic indicators together with the assumed probabilities of their implementation.