The recent sovereign debt crisis and the subsequent economic recession highlighted the importance of having sound banks that are able to withstand adverse shocks and, more generally, to support the economy even during downturns.
The cost borne by taxpayers in the countries that experienced a banking crisis created the need to implement new regulation to reduce the banking sector risk and increase its resilience to macroeconomic shocks. The creation within the Banking Union of the Single Supervisory Mechanism (SSM) required therefore a thorough evaluation of all the risks on the balance sheets of banks whose supervision was going to become a direct responsibility of the European Central Bank.
Therefore, in 2014 as a preliminary activity to the launch of the SSM, on the initiative of the European Banking Authority (EBA) and in collaboration with the ECB an extensive and accurate asset quality review (AQR) of the 130 leading European banks to be supervised by the Frankfurt authorities was conducted, followed by a stress test exercise. Since then, stress tests have been performed every two years.
On January 31 the EBA, in cooperation with the European Systemic Risk Board (ESRB), released the macroeconomic scenarios for the EU-wide stress test. Before then, the EBA published the new methodology, which however shares many similarities with that of the exercise conducted in 2016 together with some novelties. For the first time, the exercise will incorporate IFRS 9 accounting standards and will cover all relevant risk areas.  Moreover, the exercise will allow banks to use their internal models to assess the impact of the adverse scenario on net fee and commission income (NFCI), which represents between 22% and 30% of euro area banks' revenues.  In particular, the banks will be able to use models that embed both macroeconomic relationships and managerial policies.
In this article we discuss the treatment of NFCI in the EBA stress test. The rest of the article is structured as follows. A first section briefly describes the features of the 2018 EU-wide stress test scenarios, as well as the assumptions and constraints on the NFCI items. The second section discusses the empirical literature on NFCI models. The third section presents some analysis of the NFCI for the Italian banking sector over the last decades and some evidence on the NFCI in the adverse scenario. Conclusions follow.
The 2018 EU-wide stress test involves 119 EU banks under the ECB supervision.
To evaluate the resilience to adverse macroeconomic and financial scenarios, banks simulate their balance sheets over a three-year horizon. They are allowed to use both FINREP and internal data to project the NFCI conditional on the scenarios provided by EBA.
The two scenarios are a baseline, defined by National Central Banks and in line with the December forecasts published by the ECB, and an adverse scenario, which assumes a slowdown of the economy mainly due to global shocks.
Whatever the outcome of the test, it will not be interpreted as a "pass-or-fail". Nevertheless, it will be used to assess capital needs in the context of the prudential review and evaluation process, the so-called SREP (Supervisory Review and Evaluation Process).
The adverse scenario starts in the first quarter of 2018 and reflects four systemic risks that cause a sizeable repricing of risk premia in global financial markets and a negative impact on the economic growth and bank profitability. Moreover, they have an impact on the sustainability of both public and private debt increasing the liquidity risk of the banking sector. 
Compared to the baseline, at the end of the horizon in 2020, the adverse scenario for the Italian economy implies: 
As in previous exercises the EU-wide stress test is performed on the assumption of static balance sheets, or rather, zero growth in assets and liabilities and a stable business mix. In particular, assets and liabilities that mature within the three-year horizon should be replaced with similar financial instruments. Moreover, loans remain stable although their composition between performing and non-performing exposures is expected to change.
Focusing on NFCI, each bank provides projections on both revenues and costs, to the first level of FINREP granularity (Table 4.1).
For each of the selected FINREP items, banks indicate:
NFCI projections are subject to specific constraints.
In the baseline scenario the projection for each year cannot exceed the value at the end of 2017, if this value is positive; it must be set to zero if the 2017 value is zero or negative.
In the adverse scenario banks are instead required to comply with one of the following two approaches:
- the cumulative value from 2018 to 2020 under the adverse scenario;
- the initial state multiplied by three and reduced by 10%.
If the cumulative value obtained with the models is lower than the initial value (multiplied by three and reduced by 10%), then banks are allowed to use these projections for the next three years; alternatively, they need to use the initial state reduced and equally distributed over the three years.
So far the empirical banking literature has mainly focused on net interest income and other banking and insurance income while there are only a few studies on NFCI. These can be classified in two main strands.
The first strand analyses the differences between traditional income and market-related businesses, in particular their response to cyclical variations (See among others Smith, Staikouras and Wood (2003), Busch and Kick (2009)).
A second strand identifies the most relevant drivers of NFCI and measures the sensitivities to macro variables and stress test scenarios.
By using data on French banks from 1993 to 2007, Coffinet et al. (2009) confirm the role of GDP growth rate and stock market returns in explaining fees and commissions. Exploiting annual data from 1981 to 2003 for ten advanced economies, Albertazzi and Gambacorta (2009) show that non-interest income is positively related to its own lag, stock market volatility and inflation rate, and negatively to long-term interest rates.
In a recent paper, Kok et al. (2017) run a scenario analysis for a sample of banks present in all 19-euro area countries and subject to the direct supervision of the Single Supervisory Mechanism. In particular, using annual data from 1995 to 2015, they project the net fee and commission with the 2016 EU-wide Stress Test scenarios and show a high sensitivity of NFCI to its own lag, GDP growth rate, stock market returns and short-term interest rates. Focusing on the Italian stress test results, the authors show that in the baseline scenario net fee and commission income has an average growth rate of about 2.3% while in the adverse scenario it grows close to 1.9% on average.
Casolaro and Gambacorta (2004) study the relation existing between bank profitability and main financial and economic indicators, using data on the Italian banking sector from 1984 to 2002. They show that stock market volatility and GDP growth rate have the highest explanatory power for NFCI.
Over the last 20 years, fee and commission income for the Italian banking system has been on an upward trend. In particular, from 1998 to 2008 NFCI experienced a higher growth rate (9.6%) than the average over the entire 20-year period (6.4%), and with limited variability (Figure 4.1).
Moreover, the share of NFCI on total assets is more stable (showing an average value over the whole sample of about 0.67%) than the net interest income on total assets, which instead is on a decreasing trend that became more pronounced since the financial crisis (Figure 4.2).
These trends are mainly explained by the changes in the intermediation activity and in business models due to the financial crisis. Indeed, banks experienced a noticeable reduction in revenues from traditional activities, which are only partially compensated by the increase in fees and commission income.
By using an econometric model, we forecast net fee and commission income consistent with the EBA scenarios. The model uses quarterly data from 1998 to 2016, and includes macro and financial variables that have been extensively used in the economic literature. Consistent with previous literature, NFCI estimates exhibit a positive and significant relationship with the business cycle (Gambacorta and Albertazzi (2009), Kok et al. (2017)), with the inflation rate (Demirgüç-Kunt and Huizinga (1999)) and with both stock and bond market performance.
Applying the EBA constraints to the projections obtained from the model, the cumulative value for the three-year horizon in the adverse scenario is higher than the cumulative value of the initial state reduced by 10%. Therefore, the use of internal models would benefit the sector (because net commissions would suffer a penalty of 10%, which is worth about 7 billion, rather than 20% compared to the values of 2017). It is worth mentioning that this exercise is performed at an aggregate banking sector level, therefore we cannot infer clear conclusions for each individual bank, which may use internal data and project NFCI conditional on their business policies.
Over the last years, NFCI has increasingly contributed to bank profitability. This is true for the banking sector in Italy as well as in the rest of the euro area countries: in response to squeezing margins on lending and some other basic services, banks have tried to boost their revenues through NFCI.
Results shows that banks’ decision to adopt internal models is quite important and drives the projections. In fact, NFCI projections will not only depend on the scenarios outlined by the EBA, but also on the choices made by the banks themselves and that can be taken into account when building internal models. The exercise conducted for the Italian banking sector shows that banks that use internal models for NFCI penalize their projected revenues less than banks that choose not to use models.
Albertazzi, U. and L. Gambacorta, Bank Profitability and the Business Cycle, Journal of Financial Stability 5, 2009.
Busch, R. and T. Kick, Income Diversification in the German Banking Industry, Deutsche Bundesbank Discussion Paper 09, 2009.
Casolaro, L. and L. Gambacorta, A model of the profit and loss accounts for the Italian banking system, Temi di discussione Banca d’Italia 519, October 2004.
Coffinet, J., S. Lin and C. Martin, Stress Testing French Banks' Income Subcomponents, Banque de France Working Paper 242, 2009.
Demirgüç-Kunt, A. and H. Huizinga, Determinants of Commercial Bank Interest Margins and Profitability: Some International Evidence, The World Bank Economic Review, Vol. 13, pp. 379-408.2, 1999.
Kok, C., H. Mirza and C. Pancaro, Macro stress testing euro area banks' fees and commissions, ECB Working Paper series, No 2029, 2017.
Smith, R., C. Staikouras and G. Wood, Non-Interest Income and Total Income Stability, Bank of England Working Paper 198, 2003.
EBA - Adverse macro-financial scenario for the 2018 EU-wide banking sector stress test.
EBA - Methodological Note, 2018 EU-wide stress test.
EBA - Final draft ITS amending ITS on Supervisory reporting of FINREP - Annex III.
Prometeia - The 2018 EBA/ECB stress test Prometeia Insights – White Paper.