## _wp16175

## Source details

**Canonical URL:** [_wp16175](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16175.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16175.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16175.pdf.json)

---

### I. Introduction — context and challenges
- A number of interrelated cyclical and structural challenges bear on the profitability outlook of Italian banks:
  - low potential growth;
  - a large stock of nonperforming loans (NPLs), which account for about one-third of those in the euro area;
  - bank business models exposed to the SME sector and thus highly reliant on the growth outlook;
  - cost challenges, with cost-to-income ratios relatively high and above the EU average.
- Impact on banks:
  - Total (gross) NPLs in 2015 reached about 18 percent of total loans (over €360 billion).
  - Return on equity averaged 3.1 percent in 2015 for Italian banks.
  - Italian banks have taken provisions on up to an average coverage ratio of about 60 percent (implying a carrying value of merely 40 percent of the notional amount).
- Structural capital needs:
  - Ongoing regulatory reform and supervisory actions increase structural needs for more bank capital.
  - Banks in resolution can only receive state funding after 8 percent of liabilities have been “bailed-in.”
  - Banks are currently permitted more lenient risk-weights than under Basel rules, suggesting further capital needs when more restrictive use of internal models is finalized.

### II. Objectives and questions addressed
- Sample and scope:
  - Evaluates current and prospective profitability using granular bank-by-bank data for the 15 largest banks supervised by the Single Supervisory Mechanism (SSM) (these banks account for about 60 percent of system-wide assets).
- Key questions addressed:
  - By how much would profitability of current lending improve if all 15 largest banks achieved a cost structure similar to the EU average or median?
  - What is the likely impact of the ECB’s TLTRO II on funding and lending rates, and how does it affect banks’ prospects for profitability? What is the potential to raise revenues through higher credit growth?
  - Do banks that remain profitable under conservative provisioning have enough capital to lend and support the recovery? What is the scope for NPL resolution to free up capital for lending?
  - How will profitability of new lending evolve under alternative growth projections?

### III. Data and methodology
- Data sources and sample:
  - Publicly available data from the SNL database of S&P Global Market Intelligence for the 15 largest Italian banks supervised by the SSM; end-2015 quarterly data or latest available annual data.
  - System-wide data from the Bank of Italy (2014) used beyond the 15 banks (as of end-2015 there were over 640 banks in the Italian banking system, of which 33 were cooperative banks and 365 were mutual banks).
- Variables and profitability measure:
  - Variables: net interest income/average assets, cost of funds, cost-income ratio, CAR ((Tier 1 capital +Tier 2 capital)/total risk-weighted assets), credit risk-weighted assets, fee and commission income/operating income, total gross loans, loan loss provisions/operating income, net operating income.
  - Net return on equity (RoE) based on net interest margins (NIMs), commissions/fee income, operating expenses, and firm-specific capitalization.
  - Tax rate τ assumed at 35 percent for all banks.
- Provisioning and IFRS 9:
  - Forward-looking LLPs calibrated to default risk of overall loan portfolio using firm-specific credit risk weights from EBA’s Transparency Exercise (cut-off end-Q2 2015).
  - In most cases, forward-looking LLPs are higher than reported LLPs.
  - Under forthcoming IFRS 9, for loans without significant increase in credit risk provisions = expected losses next 12 months; if significant increase, provisions cover lifetime expected losses.
- Operating costs treatment:
  - Profitability changes simulated if each bank’s cost-to-income ratio declined to (i) the EU weighted average or (ii) the EU median, except where banks are already below those benchmarks.
  - Italian system operating costs: 65 percent compared to EU weighted average 63 percent and EU median 53 percent.
  - Out of the 15 sample banks, 5 and 3 banks already below the EU-weighted average and median, respectively.
- TLTRO II scenario:
  - All Italian banks assumed to participate in TLTRO II as of June 2016.
  - All banks cease to remunerate deposits, reducing funding cost to the ECB’s marginal refinancing rate (MRO) of zero percent.
  - Lending rates variable and adjust to decline of marginal policy rate; estimated to lower NIMs of Italian banks by 11 basis points on average.
- Macroeconomic scenarios:
  - (i) Staff’s baseline scenario (corresponds to staff projections in the 2016 Article IV Consultation Staff Report for Italy).
  - (ii) Severe downturn: real GDP growth declines by more than 2 percentage points over the first two years (but recovers above baseline after that).
  - (iii) Stagnation: annual GDP growth is one-half of that in the baseline scenario.
  - PDs from Garrido and others (2016); correlation of nominal growth with corporate PDs estimated at 72 percent; estimated corporate loan PD for 2014 is 1.8 percent.
- Capital and lending capacity:
  - Available capital buffer calculated taking into account Pillar I and II capital requirements under ECB’s SREP.
  - Potential net loan growth calculated assuming unchanged CAR and overall credit quality and a minimum capital buffer of 2 percentage points over the minimum.

### IV. Key findings and quantitative results
- System-level profitability and heterogeneity:
  - The banking system is profitable overall but with significant heterogeneity across banks.
  - Larger SSM banks are relatively profitable and could improve through cost reductions and higher credit growth.
  - Some banks generate little or slightly negative profitability from lending under current conditions; monetary easing and cost cutting can help some but not all.
  - Some banks likely to continue struggling to be profitable even under favorable funding conditions and operational efficiency improvements because profitability of new lending may be insufficient to offset declining interest income and high provisioning cost associated with the existing loan book.
- NPLs and provisioning:
  - High stock of impaired assets and cost of risk have dragged down earnings capacity and limited capital buffer rebuild.
  - Impaired balance sheets have weighed down credit growth and the economic recovery.
- Impact of TLTRO II and funding improvement:
  - Assumed elimination of deposit remuneration reduces funding cost to 0 percent (ECB MRO), with an estimated average NIM decline of 11 basis points.
  - TLTRO II can help maintain bank profitability and encourage lending if banks meet minimum net lending growth, but benefits may be uneven across banks (largest banks more able to access capital markets; smaller banks face more challenging liquidity).
- Cost reduction potential:
  - Improving operational efficiency of all 15 SSM banks to the euro area weighted average cost-to-income ratio of 63 percent would result in improving the weighted average net RoE by more than 40 percent.
  - If Italian banks improved operational efficiency to the EU median cost-to-income ratio of 53 percent, the weighted average net RoE would triple.
  - Material profitability gains possible but require sizable consolidation and cost savings.
- Capital constraints on new lending:
  - Even where lending is profitable, existing capital buffers generally constrain the amount of new lending possible without raising additional capital or resolving NPLs to free up capital.
- Profitability of current lending (selected results):
  - Under expected LLP, current lending by about half of the banks in the sample—about 83 percent of the banking sector in terms of total outstanding loans—generate profits amounting to a system-wide weighted-average annual net RoE of 0.7 percent at end-2015.
  - For the 15 largest banks, the weighted average net RoE improves to 2.1 percent.
  - For the system of a total of over 640 banks, the net RoE is –1.6 percent in 2014 (Bank of Italy data) and rises to 1.4 percent if projected to 2015 consistent with the performance of the 15 SSM banks in the sample.
  - The actual RoE of the system amounted to 2.6 percent in 2015 according to recently released data.
- TLTRO II effects on RoE:
  - Under expected loss provisioning and assuming sufficient loan demand, the weighted-average net RoE improves to 2.8 percent with TLTRO II.
  - Using reported provisioning improves overall system profitability to a weighted-average net RoE of 4.0 percent with TLTRO II.
  - Even with TLTRO II, one-third of the banks in the sample would still have current lending be unprofitable.
- Credit growth and required lending growth:
  - To maintain profitability over the amortization period of the current loan book, the estimated required higher lending growth is at least 3.6 percent annually (or about 3.0 percentage points above current credit growth).
  - On average, potential loan growth across all sample banks would amount to 1.4 percent—close to the benchmark lending rate required to access TLTRO II funding at most favorable terms (i.e., at the ECB’s deposit rate of currently –0.4 percent)—but far below the 3.6 percent needed to maintain current profitability given declining NIMs.

### V. Policy considerations highlighted
- Repairing bank balance sheets is a policy priority to facilitate new lending and support the economic recovery.
- Approaches to invigorate the banking system’s “self-healing powers”:
  - Facilitate bank consolidation to enable cost-cutting.
  - Reform insolvency regimes to enable workouts and speed up insolvency processes and enforcement.
  - Set up mechanisms to assist banks in resolving NPLs (examples mentioned in source: GACS and Atlas).
- Importance of continuing recent legislative reforms (August 2015 and May 2016) to speed up insolvency processes and enforcement, especially for new lending.
- Supervisory and structural measures:
  - Supervisors should engage banks to provide credible plans to reduce significantly the NPL overhang over the medium term and closely collaborate with the ECB’s NPL Task Force to incentivize NPL resolution.
  - Complementary measures: enhanced supervision; further insolvency and enforcement reforms; facilitation and development of distressed debt markets; shortening lengthy court procedures and encouraging out-of-court arrangements.
  - Examine viability of banks not subject to the ECB’s Comprehensive Assessment and take follow-up actions in line with regulatory requirements.
  - Facilitate bank consolidation, branch rationalization, and other cost-cutting to improve business models and achieve scale economies.
  - Realistic viability assessments: only banks already profitable (or with a reasonable chance of becoming profitable) can absorb reform costs and build capital buffers.

### VI. Box 1 — Capital Relief and New Lending from NPL Disposal (summary)
- Pricing gap methodology and assumptions:
  - Distressed debt investor expected return / IRR: 10 percent.
  - Servicing/legal fees: 10 percent.
  - Management costs: 2 percent.
  - Unsecured portion of each loan: 15 percent of the principal value (assumed fully provisioned).
  - Rate of collateral decay r_c: 5 percent.
  - Expected country-specific asset recovery time (Italy): 4.5 years (also cited as 4 ½ years).
  - Target CAR: 16 percent.
  - NPLs reduced to historical average: 3.6 percent of gross loan book on average.
- Capital impact and sensitivity:
  - Under conservative assumptions, banks would register losses of up to €22 billion (or 1.1 percent of GDP) at an expected valuation haircut of 31.5 percent should they dispose of their current NPLs.
  - Immediate and large reduction of NPLs would not be expected to result in capital relief at present; under an enhanced debt enforcement and insolvency framework, NPL disposal might free up regulatory capital.
  - Reducing expected foreclosure time to 3 years—or lowering return expectations of distressed debt investors to 6.5 percent—would notably reduce bank losses.
  - Reducing expected foreclosure time to 2.5 years (very optimistic), without changing return expectations, could result in some capital relief and unlock new lending—provided each new loan is capitalized at a CAR of 16 percent and each new loan carries a credit risk weighting of 56 percent.
- Illustrative market evidence:
  - MPS example: move off balance sheet €27.7 billion of gross NPLs, transferring the worst impaired assets at a transfer price of about €9.2 billion (or 33 percent of gross book value).
  - Transaction expected to result in reduction of capital of about €4.8 billion, implying a valuation haircut to net book value of about 34 percent.
- Profitability of new lending under alternative scenarios (supply side, no feedback effects):
  - Baseline scenario: Projected average annual net RoE of 3.2 percent over the next three years.
  - Downside scenario: Projected average annual net RoE of –8.4 percent over the next three years.
  - Stagnation scenario: Projected average annual net RoE of 0.8 percent over the next three years.
  - Improvements in funding cost through TLTRO II improve RoE estimates by about one-third (with heterogeneity caveats).

### VII. Box 2 — Italian NPLs: Recent Government Initiatives (summary)
- Atlante Fund:
  - Purpose: backstop ongoing banks’ capital increases; act as a buyer of last resort; could purchase non-investment grade tranches of NPL securitizations; can also invest in real estate assets.
  - Collected €4.25 billion by April 29, 2016.
  - UniCredit SpA and Intesa Sanpaolo Spa each disclosed a €1 billion stake in the fund.
  - Atlante invested €1.5 billion in the capital raising by Banco Popolare di Vicenza, reducing available resources in the fund to €2.7 billion.
  - Minority participation: Cassa Depositi e Prestiti (CDP) with 8 percent.
  - Banks are requested to deduct the amount invested in Atlante from regulatory capital; the impact on capital ratios is estimated to be modest.
- Enhanced debt enforcement measures (adopted April 29, 2016):
  - New loan contract type allowing banks to sell real estate collateral even if borrowers are subject to insolvency proceedings.
  - Creditors and borrowers can renegotiate existing loan agreements so that the new provision applies to outstanding loans.
  - Bankruptcy hearings can be conducted remotely via the internet.
  - Government estimate: it will take less than a year to collect collateral under the new framework.
- Other measures:
  - GACS: government guarantees to securitization of impaired assets; expected to close the pricing gap by around 2–3 percentage points (market participants estimate the pricing gap to be around 20 percent).
  - Decree law to simplify bankruptcy procedures and speed up recovery of collateral.
  - Shortening tax deductibility of write-offs and provisions from five years to one year to increase timely provisioning incentives.
  - ECB-Banking Supervision’s Task Force on NPLs concluding data collection and expected to provide detailed guidance; Bank of Italy launched a new periodic survey on bad debts and recovery procedures.

### VIII. Selected numeric highlights (preserve source precision)
- NPLs in 2015: about 18 percent of total loans (over €360 billion).
- Return on equity (average, 2015): 3.1 percent.
- Provisioning coverage (average): about 60 percent (implying 40 percent carrying value).
- Sample coverage: 15 largest SSM-supervised banks account for about 60 percent of system-wide assets.
- Operating costs: Italian system 65 percent versus EU weighted average 63 percent and EU median 53 percent.
- Individuals per branch (2015 ECB data): Italy 1,979 individuals per branch; EU average 2,111 individuals per branch.
- Assumed tax rate τ: 35 percent for all banks.
- Estimated average NIM reduction under TLTRO II and pass-through: 11 basis points.
- Correlation of nominal growth with corporate PDs: 72 percent.
- Estimated corporate loan PD for 2014: 1.8 percent.
- Minimum capital buffer used in simulations: 2 percentage points over the minimum.
- Under expected LLP, system-wide weighted-average annual net RoE of current lending at end-2015: 0.7 percent.
- Weighted average net RoE for the 15 largest banks: 2.1 percent (end-2015).
- System net RoE in 2014 (Bank of Italy): –1.6 percent; projected to 2015 consistent with sample: 1.4 percent.
- Actual system RoE in 2015 (recent data): 2.6 percent.
- Estimated losses from immediate disposal of current NPLs under conservative assumptions: up to €22 billion (or 1.1 percent of GDP).
- MPS NPLs transfer: €27.7 billion gross NPLs at transfer price of about €9.2 billion (33 percent of gross book value); capital reduction about €4.8 billion; valuation haircut about 34 percent.
- Atlante fund resources collected: €4.25 billion by April 29, 2016; available resources after Banco Popolare di Vicenza transaction: €2.7 billion.
- Distressed debt investor IRR used in sensitivity analysis: 10 percent (alternative sensitivity example: 6.5 percent).
- Expected foreclosure time (Italy) used in pricing gap: 4.5 years (4 ½ years); alternative foreclosure times examined: 3.0 yrs, 3.9 yrs, 2.5 yrs.
- Required higher lending growth to maintain profitability over amortization period of current loan book: at least 3.6 percent annually.
- Average potential loan growth across sample banks: 1.4 percent.
- Historical reference pre-crisis net RoE (2006-09 average): 13.8 percent.
- Projected average annual net RoE for new lending:
  - Baseline (next three years): 3.2 percent.
  - Downside (next three years): –8.4 percent.
  - Stagnation (next three years): 0.8 percent.

*Italic source attribution: IMF Working Paper material as provided in the content unit.*

### References .............................................................................................................

### _wp16175 - References

### I. Introduction — context and challenges
- A number of interrelated cyclical and structural challenges bear on the profitability outlook of Italian banks:
  - low potential growth;
  - a large stock of nonperforming loans (NPLs), which account for about one-third of those in the euro area;
  - bank business models exposed to the SME sector and thus highly reliant on the growth outlook;
  - cost challenges, with cost-to-income ratios relatively high and above the EU average.
- Impact on banks:
  - Total (gross) NPLs in 2015 reached about 18 percent of total loans (over €360 billion).
  - Return on equity averaged 3.1 percent in 2015 for Italian banks.
  - Italian banks have taken provisions on up to an average coverage ratio of about 60 percent (implying a carrying value of merely 40 percent of the notional amount).
- Structural capital needs:
  - Ongoing regulatory reform and supervisory actions increase structural needs for more bank capital.
  - Banks in resolution can only receive state funding after 8 percent of liabilities have been “bailed-in.”
  - Banks are currently permitted more lenient risk-weights than under Basel rules, suggesting further capital needs when more restrictive use of internal models is finalized.

### II. Objectives and questions addressed
- The paper evaluates current and prospective profitability of Italian banks using granular bank-by-bank data for the 15 largest banks supervised by the Single Supervisory Mechanism (SSM) (these banks account for about 60 percent of system-wide assets).
- Key questions:
  - By how much would profitability of current lending improve if all 15 largest banks achieved a cost structure similar to the EU average or median?
  - What is the likely impact of the ECB’s TLTRO II on funding and lending rates, and how does it affect banks’ prospects for profitability? What is the potential to raise revenues through higher credit growth?
  - Do banks that remain profitable under conservative provisioning have enough capital to lend and support the recovery? What is the scope for NPL resolution to free up capital for lending?
  - How will profitability of new lending evolve under alternative growth projections?

### III. Data and methodology
- Data sources and sample:
  - Publicly available data from the SNL database of S&P Global Market Intelligence for the 15 largest Italian banks supervised by the SSM; end-2015 quarterly data or latest available annual data.
  - System-wide data from the Bank of Italy (2014) used beyond the 15 banks (as of end-2015 there were over 640 banks in the Italian banking system, of which 33 were cooperative banks and 365 were mutual banks).
- Variables used (from SNL or constructed): net interest income/average assets, cost of funds, cost-income ratio, CAR ((Tier 1 capital +Tier 2 capital)/total risk-weighted assets), credit risk-weighted assets, fee and commission income/operating income, total gross loans, loan loss provisions/operating income, and net operating income.
- Profitability measure:
  - Net return on equity (RoE) based on net interest margins (NIMs), commissions/fee income, operating expenses, and firm-specific capitalization.
  - Tax rate τ assumed at 35 percent for all banks.
  - Net RoE formula uses CARt, average RWA, net interest income, fees and commissions, operating cost, LLPt−1*, where LLP* denotes forward-looking provisioning standard.
- Provisioning and IFRS 9:
  - Forward-looking LLPs calibrated to default risk of overall loan portfolio using firm-specific credit risk weights from EBA’s Transparency Exercise (cut-off end-Q2 2015).
  - In most cases, forward-looking LLPs are higher than reported LLPs.
  - Under forthcoming IFRS 9, for loans without significant increase in credit risk provisions = expected losses next 12 months; if significant increase, provisions cover lifetime expected losses.
- Treatment of operating costs:
  - Investigate profitability changes if each bank’s cost-to-income ratio declined to (i) the EU weighted average or (ii) the EU median, except where banks are already below those benchmarks.
  - Italian system operating costs marginally higher than weighted average of EU banks: 65 percent compared to 63 percent; EU median is 53 percent.
  - Out of the 15 sample banks, 5 and 3 banks already below the EU-weighted average and median, respectively.
- ECB’s TLTRO II scenario:
  - All Italian banks assumed to participate in TLTRO II as of June 2016.
  - All banks cease to remunerate deposits, reducing funding cost to the ECB’s marginal refinancing rate (MRO) of zero percent.
  - Lending rates variable and adjust to decline of marginal policy rate; estimated to lower NIMs of Italian banks by 11 basis points on average.
- Macroeconomic scenarios for forward-looking analysis:
  - (i) Staff’s baseline scenario (corresponds to staff projections in the 2016 Article IV Consultation Staff Report for Italy).
  - (ii) Severe downturn: real GDP growth declines by more than 2 percentage points over the first two years (but recovers above baseline after that).
  - (iii) Stagnation: annual GDP growth is one-half of that in the baseline scenario.
  - Probabilities of default (PDs) from Garrido and others (2016); correlation of nominal growth with corporate PDs estimated at 72 percent; estimated corporate loan PD for 2014 is 1.8 percent.
- Capital and lending capacity:
  - Available capital buffer calculated taking into account Pillar I and II capital requirements under ECB’s SREP.
  - Potential net loan growth calculated assuming unchanged CAR and overall credit quality and a minimum capital buffer of 2 percentage points over the minimum.

### IV. Key findings and quantitative results
- System-level profitability and heterogeneity:
  - The banking system is profitable overall but with significant heterogeneity across banks.
  - Larger SSM banks are relatively profitable and could improve through cost reductions and higher credit growth.
  - Some banks generate little or slightly negative profitability from lending under current conditions; monetary easing and cost cutting can help some but not all.
  - Some banks likely to continue struggling to be profitable even under favorable funding conditions and operational efficiency improvements because profitability of new lending may be insufficient to offset declining interest income and high provisioning cost associated with the existing loan book.
- NPLs and provisioning:
  - High stock of impaired assets and cost of risk have dragged down earnings capacity and limited capital buffer rebuild.
  - Impaired balance sheets have weighed down credit growth and the economic recovery.
- Impact of TLTRO II and funding improvement:
  - Assumed elimination of deposit remuneration reduces funding cost to 0 percent (ECB MRO), with an estimated average NIM decline of 11 basis points.
  - TLTRO II can help maintain bank profitability and encourage lending if banks meet minimum net lending growth, but benefits may be uneven across banks (largest banks more able to access capital markets; smaller banks face more challenging liquidity).
- Cost reduction potential:
  - Material profitability gains possible if cost-to-income ratios converge toward the EU weighted average or median, but sizable consolidation and cost savings would be required.
- Capital constraints on new lending:
  - Even where lending is profitable, existing capital buffers generally constrain the amount of new lending possible without raising additional capital or resolving NPLs to free up capital.

### V. Policy considerations highlighted
- Repairing bank balance sheets is a policy priority to facilitate new lending and support the economic recovery.
- Approaches to invigorate the banking system’s “self-healing powers” given limited scope for inflation or public bailouts within the euro area:
  - Facilitate bank consolidation to enable cost-cutting;
  - Reform insolvency regimes to enable workouts and speed up insolvency processes and enforcement;
  - Set up mechanisms to assist banks in resolving NPLs (examples mentioned in source: GACS and Atlas; see Box 2).
- The paper underscores the importance of continuing recent legislative reforms (August 2015 and May 2016) to speed up insolvency processes and enforcement, especially for new lending.

### VI. Selected numeric highlights (preserve source precision)
- NPLs in 2015: about 18 percent of total loans (over €360 billion).
- Return on equity (average, 2015): 3.1 percent.
- Provisioning coverage (average): about 60 percent (implying 40 percent carrying value).
- Sample coverage: 15 largest SSM-supervised banks account for about 60 percent of system-wide assets.
- Operating costs: Italian system 65 percent versus EU weighted average 63 percent and EU median 53 percent.
- Individuals per branch (2015 ECB data): Italy 1,979 individuals per branch; EU average 2,111 individuals per branch.
- Assumed tax rate τ: 35 percent for all banks.
- Estimated average NIM reduction under TLTRO II and pass-through: 11 basis points.
- Correlation of nominal growth with corporate PDs: 72 percent.
- Estimated corporate loan PD for 2014: 1.8 percent.
- Minimum capital buffer used in simulations: 2 percentage points over the minimum.

*Source: _wp16175 - References (IMF working paper content provided).*

### 12.7 percent.

### _wp16175 - 12.7 percent.

### Profitability of current lending and provisioning levels
- Under expected LLP, current lending by about half of the banks in the sample—about 83 percent of the banking sector in terms of total outstanding loans—generate profits amounting to a system-wide weighted-average annual net return on equity (RoE) of 0.7 percent at end-2015.
- A number of smaller banks (representing about one-eighths of total loan volume of all banks in the sample) are likely to experience losses.
- For the 15 largest banks, the weighted average net RoE improves to 2.1 percent, but three smaller banks (accounting for about 5 percent of the outstanding stock of loans in Italy) still generate losses from current lending.
- For the system of a total of over 640 banks, the net RoE is –1.6 percent in 2014 according to the latest available data published by the Bank of Italy (and rises to 1.4 percent if projected to 2015 consistent with the performance of the 15 SSM banks in the sample).
- The actual RoE of the system amounted to 2.6 percent in 2015 according to recently released data.
- Historical analysis (assuming expected loss provisioning) suggests that since 2012 lending rates on average were far below what would have been required for banks to fund sufficient loan loss reserves ex post; the rise of NPLs would have implied a higher minimum lending rate to maintain profitability.
- The high level of LLPs relative to net income and high operating costs reveal a fundamental lack of profitability in core banking business for many smaller banks.

### Potential impact of reducing operating costs
- Improving operational efficiency of all 15 SSM banks to the euro area weighted average cost-to-income ratio of 63 percent would result in a significant improvement of banks’ earnings capacity from current (and future) lending, improving the weighted average net RoE by more than 40 percent.
- If Italian banks improved operational efficiency to the EU median cost-to-income ratio of 53 percent, the weighted average net RoE would triple.
- The analysis emphasizes that lowering costs is a preferable structural solution to raising lending rates or tightening credit standards, which could otherwise further weaken economic activity and asset quality.

### Potential impact of monetary easing (including TLTRO II)
- The ECB’s TLTRO II improves the pass-through of lower bank funding costs to credit supply and mitigates adverse impacts of negative rates on profitability.
- Under expected loss provisioning and assuming sufficient loan demand, the weighted-average net RoE improves to 2.8 percent with TLTRO II.
- Using reported provisioning improves overall system profitability to a weighted-average net RoE of 4.0 percent with TLTRO II.
- Even with TLTRO II, one-third of the banks in the sample would still have current lending be unprofitable.
- There remain some banks that would experience very negative profitability even under optimal funding conditions.
- Since Q3 2015 the equity risk premium of Italian banks has risen and price-to-book ratios have declined, with the average cost of equity now exceeding the return on equity—reflecting market expectations of deteriorating future profitability and raising the cost of capital raising for impaired banks.
- The ECB’s expanded asset purchase program and negative marginal policy rate are estimated to lower the NIM of Italian banks by 11 basis points on average.

### Credit growth, NIM dynamics, and capital buffers
- Given deposit stickiness and a high share of variable rate loans, lower rates benefit mostly new lending and do not offset negative impacts on existing loans if credit growth is insufficient.
- To maintain profitability over the amortization period of the current loan book, the estimated required higher lending growth is at least 3.6 percent annually (or about 3.0 percentage points above current credit growth).
- Banks generally exceed regulatory capital adequacy requirements, so prudentially there is no immediate need for further capital, but capital buffers may only support a limited amount of new lending.
- On average, potential loan growth across all sample banks would amount to 1.4 percent—close to the benchmark lending rate required to access TLTRO II funding at most favorable terms (i.e., at the ECB’s deposit rate of currently –0.4 percent)—but far below the 3.6 percent needed to maintain current profitability given declining NIMs.
- Continued weak credit demand and high impaired assets may further delay improvement of banks’ earnings capacity; reductions of NPLs combined with structural reforms to reduce foreclosure times and strengthen debt enforcement and insolvency frameworks could free up regulatory capital to support new lending.

*Source: IMF staff calculations and analysis in _wp16175 - 12.7 percent.*

### Box 1 updates already published material with the latest data (Aiyar and others, 2015).

### _wp16175 - Box 1 updates already published material with the latest data (Aiyar and others, 2015).

### Context and key findings
- Considering the high loan restructuring costs and the track record of low effectiveness of debt enforcement in Italy, distressed debt investors demand a high rate of return.
- Under conservative assumptions, banks would register significant losses if they were forced to dispose of their current NPLs at an expected foreclosure time of 4 ½ years.
- Current insolvency reforms, when fully implemented, are expected to significantly shorten the recovery time of new loans. Reforms that dramatically lower the time to resolve the existing stock of NPLs would go a long way toward addressing concerns related to losses.
- Higher loan growth will not solve the profitability challenge of a number of smaller banks in Italy: high expected provisions, low interest earnings, and high operating costs imply that new lending is unlikely to ameliorate losses or cost pressures under the given conservative provisioning standards.
- Market pressures since early 2016 reflect, along with global factors, investor discomfort with prospects of some banks to get ahead of their profitability challenge absent strong action (for instance, accelerating disposal of high stocks of NPLs).

### Pricing gap methodology and assumptions
- The sale price of NPLs reflects expected recovery time and required return by distressed debt investors; a shortfall below net book value is the “pricing gap”.
- Assumptions used in calculations:
  - Distressed debt investor expected return / IRR: 10 percent.
  - Servicing/legal fees: 10 percent.
  - Management costs: 2 percent.
  - Unsecured portion of each loan: 15 percent of the principal value (assumed fully provisioned).
  - Rate of collateral decay r_c: 5 percent.
  - Expected country-specific asset recovery time (Italy): 4.5 years (also cited as 4 ½ years).
  - Banks reduce current stock of NPLs (end-Q3 2015) by selling distressed loans; sale price = reported net loan value less country-specific haircut, calculated as net present value after fees and costs.
  - Target capital adequacy ratio (CAR): 16 percent.
  - NPLs reduced to historical average: 3.6 percent of gross loan book on average.
- Pricing gap formula (expressed in the source) defines implied coverage ratio and pricing gap as:
  - implied coverage ratio = 0.15 + (1 − ((0.85 (1 − r_c) e^(−rt) (1 − L)) − M)^0.85)
  - pricing gap = implied coverage ratio − actual coverage ratio
  - where L and M are servicing/legal fees and management costs (in percent), r is discount rate / IRR (10 percent), r_c is collateral decay (5 percent), and t is years (Italy: 4.5 years).

### Capital relief calculations and sensitivities
- Immediate and large reduction of NPLs would not be expected to result in capital relief at present; under an enhanced debt enforcement and insolvency framework, NPL disposal might free up regulatory capital.
- The capital relief estimations are highly dependent on the assumption on IRR, which significantly impacts results.
- Numerical outcomes and chart annotations from the exercise (as reported in the source):
  - Italy: Foreclosure Time 4.5 yrs. → Capital Relief -0.4% (percent of total assets).
  - An alternative foreclosure time cited: 3.0 yrs.
  - Another foreclosure time shown: 3.9 yrs.
  - Distressed Debt Investor Return (Percent) sensitivity illustrated (chart labels): example value 6.5%.
  - Euro Area regression reported: y = 1.0895x − 0.0138; R² = 0.7849.
- Country labels and chart context (reported): SVN, ITA, PRT, IRL, AUT, SVK, ESP, NLD, DEU, FRA, BEL; Net NPL (lhs) and Capital relief metrics plotted; Euro Area: Net NPLs and Foreclosure Time (percent of total assets/years), 2015.

### Policy implications and recommendations (from the source)
- Accelerating disposal of the high stocks of NPLs would help address loss concerns and investor discomfort about bank profitability prospects.
- Effective improvements to debt enforcement and insolvency frameworks are crucial to shorten recovery times and reduce pricing gaps, thereby making NPL disposal more likely to generate capital relief.
- Without significant reductions in expected provisions or material improvements in earnings and operating efficiency, new lending alone is unlikely to resolve profitability pressures for smaller banks.

*Source: Box 1, _wp16175 - Box 1 updates already published material with the latest data (Aiyar and others, 2015).*

### Box 1. Capital Relief and New Lending from NPL Disposal (concluded)

### Box 1. Capital Relief and New Lending from NPL Disposal (concluded)

### Capital impact of immediate NPL disposal
- Under conservative assumptions, banks would register losses of up to €22 billion (or 1.1 percent of GDP) at an expected valuation haircut of 31.5 percent should they dispose of their current NPLs.
- These losses would outweigh any potential reduction in capital requirements because the removal of NPLs reduces the risk density of the existing loan portfolio.
- The extent of capital relief would vary significantly across banks owing to the uneven distribution of NPLs and their capital intensity (depending on the relevant credit risk measurement methodology).
- Capital relief and attendant lending capacity could be lower if:
  - banks face larger haircuts on NPL sales;
  - banks do not use all of their freed-up capital for new lending (which also depends on demand for loans);
  - banks do not keep their capitalization unchanged.
- Banks using the internal ratings-based approach (IRB) might experience much smaller capital relief for disposal of NPLs that are sufficiently provisioned under current accounting standards (IAS 39).
- Assumption note: fully provisioned loans are simply written-off, so capital relief is only generated for partially provisioned NPLs.

### Sensitivity to foreclosure times and investor return expectations
- Reducing expected foreclosure time to 3 years—on both legacy and new NPLs—or lowering return expectations of distressed debt investors to 6.5 percent would notably reduce bank losses.
- Reducing expected foreclosure time to 2.5 years (described as very optimistic), without changing return expectations, could result in some capital relief and unlock new lending—provided there is corresponding demand for new loans—assuming:
  - each new loan is capitalized at a CAR of 16 percent; and
  - each new loan carries a credit risk weighting of 56 percent.
- A two-year reduction in recovery times (per Bank of Italy (2016) study using IRR of 15-25 percent) would reduce the pricing gap by approximately 10 percentage points.

### Illustrative market evidence (Banca Monte dei Paschi di Siena example)
- MPS announced a plan to move off balance sheet €27.7 billion of gross NPLs, transferring the worst impaired assets at a transfer price of about €9.2 billion (or 33 percent of gross book value).
- Transaction structure and amounts:
  - riskiest equity tranche worth €1.6 billion retained by MPS;
  - €1.6 billion in mezzanine notes purchased by the Atlante Fund;
  - €6 billion of senior notes to be sold to private investors (conditional on a government guarantee via GACS).
- The transaction is expected to result in reduction of capital of about €4.8 billion, implying a valuation haircut to net book value of about 34 percent.

### Profitability of new lending under alternative scenarios
- Structural context: Italian banks’ lending-focused business model, heavy exposure to households and SMEs, results in rigid cost structures and highly cyclical profitability.
- Scenario-based assessment (supply side, no feedback effects) for new lending over next five years:
  - Baseline scenario:
    - Projected average annual net RoE of 3.2 percent over the next three years (even under conservative expected loss provisioning).
    - This remains far below the pre-crisis average of 13.8 percent.
    - Estimates are based on macroeconomic projections before the U.K. referendum; spillovers could lower profitability.
  - Downside scenario:
    - Projected average annual net RoE of –8.4 percent over the next three years.
    - Scenario comprises a severe negative shock to growth resulting in a cumulative output loss of more than 5 percent over the medium term followed by a dynamic recovery.
  - Stagnation scenario:
    - Projected average annual net RoE of 0.8 percent over the next three years.
    - Scenario halves current growth over a five year forecast horizon.
- Default risk is the key determinant in profitability projections under adverse scenarios.
- Improvements in funding cost through the impact of the ECB’s TLTRO II improve RoE estimates by about one-third (with caveats on heterogeneity and rising asset encumbrance).

### Consequences and interaction with NPL overhang
- High provisioning expenses impede sustainable profitability and slow the repair of balance sheets through retained earnings.
- Even with improved demand, banks’ capacity and willingness to lend likely remain modest if provisioning needs persist, weighing on the recovery.
- A persistent pricing gap between net book value and market price of NPLs is driven by:
  - depressed housing market and structural deficiencies slowing recovery of collateral;
  - lengthy foreclosure processes increasing required haircuts;
  - absence of a developed distressed debt market providing pricing benchmarks.
- Reducing NPLs significantly over the medium term and improving operating efficiency can raise profitability, stimulate lending, and improve resilience.

### Policy recommendations and reforms
- Supervisors should engage banks to provide credible plans to reduce significantly the NPL overhang over the medium term and closely collaborate with the ECB’s NPL Task Force to incentivize NPL resolution.
- Complementary measures to support NPL resolution and banking resilience:
  - Enhanced supervision.
  - Further insolvency and enforcement reforms (beyond recent measures), including improvements to insolvency frameworks for corporates and households.
  - Facilitation and development of distressed debt markets.
  - Shortening lengthy court procedures and encouraging out-of-court arrangements to accelerate asset recovery and loan restructuring.
- Structural and sectoral measures:
  - Examine viability of banks not subject to the ECB’s Comprehensive Assessment and take follow-up actions in line with regulatory requirements.
  - Facilitate bank consolidation, branch rationalization, and other cost-cutting to improve business models and achieve scale economies.
  - Realistic viability assessments: only banks already profitable (or with a reasonable chance of becoming profitable) can absorb reform costs and build capital buffers.

### Recent government initiatives (overview)
- GACS: Garanzia Cartolarizzazione Sofferenze
  - Mechanism for government guarantees to securitization of impaired assets; provides guarantees for senior tranches rated investment grade.
  - Guarantees priced at market terms based on expected losses to comply with EU State aid rules.
  - Market participants expect a positive though modest impact: GACS may close the pricing gap by around 2–3 percentage points, while market participants estimate the pricing gap to be around 20 percent.
- Atlante fund:
  - Industry-sponsored backstop fund to recapitalize troubled banks and possibly buy junior tranches of NPL securitizations.
- Other measures:
  - Decree law to simplify bankruptcy procedures and speed up recovery of collateral (expected to impact new NPLs gradually).
  - Shortening tax deductibility of write-offs and provisions from five years to one year to increase timely provisioning incentives.
  - Reforms in insolvency (including out-of-court resolution) and bank corporate governance encouraged.
  - ECB-Banking Supervision’s Task Force on NPLs concluding data collection and expected to provide detailed guidance; Bank of Italy launched a new periodic survey on bad debts and recovery procedures.

*Source: _wp16175 - Box 1. Capital Relief and New Lending from NPL Disposal (concluded).*

### Box 2. Italian NPLs: Recent Government Initiatives (continued)

### Box 2. Italian NPLs: Recent Government Initiatives (continued)

### Atlante Fund: design, resources, and actions
- Purpose: backstop ongoing banks’ capital increases; act as a buyer of last resort; could purchase non-investment grade tranches of NPL securitizations; senior tranches targeted to other institutional investors; can also invest in real estate assets.
- Ownership and participants:
  - Majority: largest Italian banks, nonbank financial institutions and banking foundations.
  - Minority participation: Cassa Depositi e Prestiti (CDP) with 8 percent.
- Resources and commitments:
  - Collected €4.25 billion by April 29, 2016.
  - UniCredit SpA and Intesa Sanpaolo Spa each disclosed a €1 billion stake in the fund.
- Investment activity and capital implications:
  - Atlante invested €1.5 billion in the capital raising by Banco Popolare di Vicenza, taking over 99 percent stake in May 2016.
  - As a result, available resources in the Atlante fund dropped to €2.7 billion.
  - Banks are requested to deduct the amount invested in Atlante from regulatory capital; the impact on capital ratios is estimated to be modest.
- Note: capital impact of contributions scales to the available capital buffer after application of SREP requirements.

### Enhanced debt enforcement measures (adopted April 29, 2016)
- Objective: expedite foreclosures on NPLs to corporate and smaller and medium-sized enterprises (SMEs).
- Main changes to the foreclosure process:
  - (i) A new type of loan contract will allow banks to sell real estate collateral even if borrowers are subject to insolvency proceedings, removing the need to wait for completion of lengthy insolvency processes before repossessing collateral.
  - (ii) Creditors and borrowers can renegotiate existing loan agreements so that the new provision applies to outstanding loans.
  - (iii) Bankruptcy hearings can be conducted remotely via the internet.
- Government estimate: it will take less than a year to collect collateral under the new framework.

### Forward-looking provisioning methodology (Box A1: Calculating Forward-Looking Provisions Based on Risk-Weighted Assets)
- Estimation objective: forward-looking (expected loss) provisioning LLP* by aligning loan loss provisions (relative to operating income) to the average risk density of the current loan portfolio.
- Core identity (as presented):
  - loan loss provisions / net operating income = (0.00092×RWA^2 − 0.06×RWA + 1.662) × loss given default / 100.
- RWA calculation for forward-looking scenarios: use IRB approach under Basel III (BCBS, 2005):
  - RWA = K × 12.5 × EAD
  - where
    - K = LGD × [N(√(1/(1−R) × G(PD) + √(R/(1−R)) × G(0.999)) − PD] × (1 + (M−2.5)b / (1−1.5b))
    - b = (0.11852 − 0.05478 × ln(PD))^2
    - R = AVC × (0.12 × (1 − e^(−50*PD))/(1 − e^(−50)) + 0.24 × (1 − (1 − e^(−50*PD))/(1 − e^(−50))))
  - N(•) and G(•) denote the cumulative distribution function and the quantile function of the standard normal distribution, respectively.
- Parameter choices and simplifications for analysis:
  - AVC = 1 (AVC = 1.25 only if company is a large regulated financial institution or unregulated financial institution as defined; otherwise AVC = 1).
  - LGD = 45 percent.
  - Maturity adjustment term [1 + (M−2.5)b / (1−1.5b)] is ignored for simplicity due to absence of granular data on weighted-average maturity (transforms formula to that used for residential mortgage exposures while retaining AVC term for R).
- Data sources and scope:
  - RWA of performing credit exposures as of end-June 2015 from the EBA Transparency Exercise, except Banca Monte dei Paschi di Siena SpA and Banco Popolare Società Cooperativa (data from SNL).
  - RWAs of aggregate loan portfolio for a given PD calculated via IRB specification above for forward-looking macro scenarios.
- Footnote: Owing to the absence of granular data on the maturity of the loan portfolio, this simplified approach was chosen (without loss of generality).

### Empirical findings and scenario analyses (figures and notes)
- Breakeven and actual lending rates (dispersion, 2006–2014 and sample banks):
  - Figures present percentiles (90th-95th, 75th-90th, IQR, 10th-25th, 5th-10th), Median, and Average for breakeven and actual lending rates; expected loss provisions derived from RWAs per Annex, Box A1.
- Difference between actual and breakeven lending rates:
  - Time series (median and weighted average) 2006–2015 show the gap; negative gap implies loss-making.
- Estimated net return on equity (RoE) of current lending under expected loss provisioning (end-2015):
  - Sample split into three tiers (Top Tier*, Medium Tier, Bottom Tier; */ five banks each, ordered by RoE, weighted by total loans).
  - TLTRO-II scenario assumptions: funding rate at MRO (0%) via TLTRO II (and full rollover of existing TLTRO); any new deposits at 0%; lending rates adjust according to marginal policy rate (since end-2015: -20 bps) and expected pass-through from term spread compression at historical elasticity of NIMs; banks maintain capital ratio as of end-2015.
- Aggregate profitability from new lending with expected loss provisioning under macro scenarios (return on equity, percent) for 2015–2021:
  - Baseline, Downside (adverse), and Stagnation scenarios show projected RoE series 2015–2021 with and without TLTRO II benefit.
  - Note: TLTRO II is assumed to lower funding costs to 0% and NIMs are reduced by 11 bps due to recent ECB easing measures.
  - Historical reference: 2006-09(avg.) = 13.8%.
- Macro scenario assumptions (percent, end-2015 baseline and projected paths 2016–2021):
  - Shown for Default Risk (Loans), Lending Rate, Funding Cost, Real Growth, Headline CPI, Short and Long-Term Interest Rates across Baseline, Stagnation, Severe downturn scenarios (figures present relative paths).
- Aggregate profitability results indicate sensitivity of RoE to funding costs (including TLTRO II), lending rates, and provisioning calibrated to each scenario.

### Bank capitalization and profitability (cross-sectional findings)
- Capitalization metrics (percent), end-2015:
  - Sample split into three tiers (Top Tier*, Medium Tier, Bottom Tier) ordered by CAR and weighted by total loans.
  - Reported metrics include Tier 1 capital ratio, Capital adequacy ratio (CAR), and weighted-average CAR and Tier 1 ratios.
- Profitability snapshots (figures A2–A4):
  - End-2015 estimated net RoE under expected loss provisioning and reported provisioning, with and without TLTRO-II funding benefits, showing variation across tiers and cumulative loan shares.
  - Aggregate projections under macro scenarios indicate pressures on RoE over 2016–2021.

*Italic source attribution: IMF Working Paper material as provided in the content unit.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16175.pdf_