Italy: Selected Issues

Italy: Selected Issues


Italy: Selected Issues

Resolving Nonperforming Loans in Italy: a Comprehensive Approach1

A. Background

1. Nonperforming loans (NPLs) in Italy have reached systemic levels. They have tripled since the beginning of the crisis and, according to national definition, now stand at €330 billion (17 percent of total outstanding loans).2 The problem is especially pronounced for bad (impaired, or sofferenze) loans, which amount to more than half of total NPLs (Figure 1). The combination of over-indebted corporates, banks low in capital buffers but high in risks, a legal system complicating corporate restructuring and insolvency, lengthy judicial processes, and a tax system that discourages NPL write-offs, have all contributed to the build-up of NPLs. And while the rise in NPLs has recently decelerated, the stock of NPLs is still growing and the pace of write-offs has not increased significantly.

Figure 1.
Figure 1.

Italy: Nonperforming Loans

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Source: Bank of Italy and IMF Staff EstimatesNotes: In percent of total loans.

2. Loan losses have depressed profitability and hindered Italy’s recovery. Loan losses and high administration costs for NPLs weigh heavily on bank profitability and hamper attempts to build up capital buffers (Figure 2). Banks’ reluctance to write off NPLs can be explained by several factors, including the high risk premia and ROEs targeted by investors. It may also be suggestive of provisioning gaps that would need to be closed. At the same time, corporate debt overhang has grown substantially in the wake of the crisis (see accompanying paper on corporate sector). Hence, despite availability of ample and cheap liquidity, banks have become much more cautious in extending new credit, especially to SMEs. For more risky corporates, bank financing has become largely unavailable or simply too expensive to be economically reasonable. Given the absence of alternative sources of corporate financing, this has constrained investment and undermined the economic recovery.

Figure 2.
Figure 2.

NPLs, Profitability, and Capital: Some Cross-Country Evidence

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

3. The authorities recognize the scale of the problem and have already introduced successive reforms to revamp the Italian insolvency system. Still, more needs to be done and the authorities have set up a commission to make proposals for the reform of the insolvency system by the end of this year. Separately, Bank of Italy (BoI) has issued supervisory guidance on how and when to write-off uncollectable exposures. Recently, NPL resolution has received significant attention as the authorities are considering the establishment of a state-backed asset management company (AMC). At this stage, many details of the proposal are still unknown, however.

4. This paper aims to contribute to the ongoing discussion by proposing a comprehensive strategy to repair private sector balance sheets and support the recovery. For the strategy to be effective, the strong interlinkages between Italian bank and corporate balance sheets need to be acknowledged. Hence, a potential solution needs to involve a broad-based clean up of bank balance sheets, corporate restructuring, and exit of unviable firms from the market. Elements of this strategy include economic, supervisory, and legal measures.

5. The remainder of the paper is structured as follows. Section B presents the distribution of NPLs across banks, sectors, and regions and analyzes the macroeconomic and bank-specific factors that explain the build-up of NPLs, including the deteriorated health of Italy’s corporate sector. Section C outlines the main obstacles to resolving NPLs. Section D discusses some preliminary recommendations for tackling the NPL problem.

B. The Nature of the Problem

Distribution of NPLs by Size, Region, and Sectors

6. The NPL problem in Italy has several features that need to be considered when devising a potential solution (Figure 3).

  • The bulk of bad debt consists of relatively small loans. In terms of the total value of bad debt (impaired loans), more than 75 percent relates to loans above €250,000. However, in terms of the total number of borrowers, about 75 percent of bad debt relates to loans of less than €75,000.

  • About three quarters of bad debt are related to the corporate sector. Since the corporate sector in Italy comprises mostly small and medium sized companies (often with less than 10 employees), this may explain the prevalence of small loans noted above.3 The service sector and less technology-intensive sectors are most affected.

  • The problem has a pronounced regional dimension. Looking at all types of NPLs and all sectors of economic activity, there appears to be a north-south divide, especially in terms of bad loans to the corporate sector. While in early 2009 most regions had bad debt (sofferenze) ratios below 10 percent, by end 2014, most central and southern Italy regions saw their bad debt ratios increase above 20 percent.

Distribution of NPLs Across Banks

7. Large banks hold the lion’s share of NPLs, but NPL ratios are high across all types of banks suggesting a system-wide problem (Figure 4). Specifically, the following patterns emerge from the data:

  • There are a few banks (especially smaller ones) that face extremely high NPL ratios. However, most banks have NPL ratios between 15 and 20 percent. There is no clear correlation between NPL ratios (in percent of total loans) and bank size.

  • As of December 2014, the five largest banks in Italy account for two-thirds of total bad debt and NPLs.

  • NPL ratios are high across different types of banks suggesting a system-wide problem. However, large cooperative banks (banche popolari) faced the highest NPL ratios with the lowest coverage ratio at end-2014.4 This is consistent with the findings from the IMF’s 2013 Financial Sector Assessment Program.

Figure 3.
Figure 3.

Italy: NPLs by Size, Sectors, and Region

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Sources: Bank of Italy; and IMF staff estimates.
Figure 4.
Figure 4.

Distribution of NPLs Across Banks, End-2014

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Sources: SNL; Bank of Italy; and IMF staff estimates.1/ 2014 data. NPLs are in percent of total gross loans. Full SNL sample of 402 Italian banks.2/Total system wide bad loans and NPLs based on from supervisory returns data from BoI. Data for bad loans from SNL are net of provisions.3/ bad loans and problem loans from SNL based on 2013 data.4/ largest five banking groups by total assets as of December 2014.

Contributing Factors to the Build-up of NPLs

8. Bank specific factors, such as low profitability and past excessive lending, are associated with higher NPLs (Figure 5). Using cross-sectional data from SNL that cover 402 banks, we find a strong negative correlation between profitability (measured as returns on equity) and NPLs.5 This suggests that banks with better risk management practices are, on average, more profitable.

Figure 5.
Figure 5.

NPLs, Profitability, and Lending

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Source: SNL.Notes: NPLs (in percent of total gross loans).

9. Using a dynamic panel regression, we attempt to differentiate between bank-specific and macroeconomic factors. We run fixed effects and GMM regressions of NPLs (problem loans in percent of total gross loans) on various macro variables common to all banks (lagged inflation, NEER percentage change, percentage change in unemployment, lagged real GDP growth, lagged percentage change in house and stock prices), as well as bank-specific variables (lagged equity/assets ratio, lagged ROE, lagged loan growth and lagged Tier 1 ratio). In order to have data for a number of years, we use SNL data for 62 banks for which time series data since 2005 are available. Various robustness checks are performed (see Annex A for details).

10. The analysis suggests that both bank-level and macro-economic factors have affected banks’ asset quality. Higher NPL levels cause lower profitability.6 And higher lending in the past—measured by (lagged) loan growth—is related with higher NPLs, indicating that faster loan book expansion on average results in worse asset quality. A number of macro-economic variables are significant as well. For instance, lower growth, exchange rate appreciations, and falling house prices are significantly associated with higher NPLs.7 Overall, the analysis suggests that the crisis had a profound impact on banks’ asset quality, which was exacerbated by bank-specific factors.

11. The prolonged recession led to higher default risk for firms and banks (Figure 6, left chart). In particular, corporate and bank default probabilities peaked in mid-2012. And while default probabilities have come down substantially since then, NPLs have continued to rise (Figure 6, right chart). At 2 percent, the one-year default probability for corporates is still high, however.8

Figure 6.
Figure 6.

Italy: Default Risk

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Source: Bank of Italy, OECD, Moody’s KMV.Notes: Real GDP growth yoy in percent. Default probabilities for 12 month

Output, Credit Default Risk, and Nonperforming Loans

12. In turn, corporate sector default risk has been driven mostly by the interest rate environment and real activity. A dynamic macro-financial state space model (DSSM) illustrates how the dynamics have changed over the past few years (Figure 7). The estimates suggest that (i) further output losses would fuel credit risk considerably more than in the past; and (ii) higher interest rates would now have a much more substantial impact on corporates than in the past. Similarly, increases in interest rates would negatively impact the market value of banks’ holdings of Italian sovereign assets, as the volume of sovereign bonds in banks’ books has risen substantially.9

Figure 7.
Figure 7.

Italy: Key Drivers of Corporate Sector Default Risk; Sovereign Risk

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Source: IMF, Moody’s KMV, OECD. IMF staff calculations.Notes: The chart on the left shows the true coefficients of corporate default risk’s key drivers resulting from a dynamic State Space model (DSSM). Unemployment is the share of potential workforce that is not employed, STIR is the change in short-term interest rates, GDPreal is the yoy change in real GDP growth, and GovConsumption. Exp is the change in government consumption expenditure. The chart on the right shows the build-up of sovereign exposure and Italian banks’ reliance on ECB liquidity.

C. Obstacles to NPL Resolution

Economic Factors

13. Potentially substantial pricing gaps lead to disincentives for banks to write-off and sell NPLs. Writing off under-provisioned loan exposures requires closing the provisioning gap, eating into banks’ capital. And if investors believe NPLs are not adequately provisioned, transfer prices are discounted. In both cases, the result is that banks would have to realize losses. Structurally weak profitability and, for some banks, low capital buffers give only limited space for balance sheet clean up. A key factor is asset quality, as NPLs weigh heavily on bank profitability (Figure 8). Already thin profit margins make it difficult for banks to digest additional loan loss impairments.

Figure 8.
Figure 8.

Italy: Return on Equity


Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Source: IMF.1/ 2013Q4 data.

14. Italy’s banking system is largely based on relationship lending. Banks have long-standing relationships with their customers, and keeping these relationships is essential for the banks’ reputation. Relationship banking is an important element of many cooperative and regional banks, and has arguably helped many embattled nonfinancial corporation survive the deep recession suffered by the Italian economy. At the same time, related-party lending has likely contributed to the build-up of problem loans in medium-sized and small credit institutions. It also makes it more difficult to resolve NPLs.

15. The previous tax treatment penalized Italian banks that wrote off problem loans more aggressively. Until the 2013 reform, write-offs were only tax deductible in the state of insolvency. Tax deductibility of loan-loss provisions from taxable income was limited to 0.3 percent of outstanding loans, with the remaining part treated as a deferred tax asset (DTA). DTAs were deductable from taxable revenues over a period of 18 years.10 This cap constituted a clear disincentive to provisioning. Since 2013, provisions and write-offs can be deducted in equal installments over five years, and with a higher tax rate. While this approach is still more restrictive than in other countries, incentives for accelerated write-off have increased.11 As of end-2014, DTAs not deducted from capital amounted to €43 bn.

16. In the past, insufficient guidance on accounting under IFRS decelerated NPL write-offs. IAS 39 does not define when and how defaulted loans are to be written off. Since the current accounting regime does not include clear write-off rules, banks (including Italian credit institutions) apply a de-recognition rule (loan cancellation). This practice, however, was supposed to be applied only under certain conditions, like asset transfer, and not as a general practice. BoI recently issued guidance on when and how uncollectable loans are to be removed from the balance sheet.

17. Smaller banks have limited experience and capacity to deal with NPLs. While the largest Italian banks have been able to dispose of NPLs, invest in internal NPL management, and set up decentralized AMCs, the medium and small-sized banks have struggled to bring down NPLs.12 These banks lack risk management capacity, NPL management experience, and access to distressed debt markets.

18. The sectoral composition of NPLs makes the problem more difficult. Unlike in other countries, where NPLs were concentrated in the real estate sector and therefore were relatively easy to value, Italian NPLs are mostly in the corporate sector, and very heterogeneous.

Legal Factors

19. The Italian insolvency system has evolved significantly in the last decade.

  • Successive reforms have resulted in a revamped reorganization procedure (concordato preventivo, reformed in 2005, 2007, 2012, and 2013) and special procedures that only apply to large companies and special enterprises (amministrazione straordinaria, versions “Prodi” and “Marzano” for the reorganization of large companies; liquidazione coatta amministrativa for specially regulated enterprises).

  • The system includes the traditional bankruptcy liquidation (fallimento), which is the procedure most commonly used in practice for corporate distress.

  • The insolvency system also includes some informal or hybrid debt restructuring options (piani di risanamento, accordi di ristrutturazione).

  • Until recently, consumers and small entrepreneurs were outside the scope of the insolvency legislation. But new insolvency procedures were introduced to cover both individuals (consumers and entrepreneurs) and small enterprises (law of January 27, 2012, modified by Decree-Law no. 179 of 2012).

Despite these changes, which make the current framework largely consistent with best international practices, the insolvency system still appears inadequate—due to procedural complexity, coupled with the lack of efficiency of the judiciary—in dealing with the magnitude of the corporate distress problem.

20. The number of insolvency cases has increased substantially since the onset of the crisis. In 2014, enterprise bankruptcies surpassed 15.000 cases (+10.7 percent in comparison with 2013) (Figure 9). It is estimated that more than 82,000 enterprises have been liquidated in insolvency processes since 2008, with an associated loss of about a million jobs (Cerved 2015). By contrast, enterprise reorganizations (concordati preventivi) declined by 20 percent in 2014.

Figure 9.
Figure 9.

Italy: Bankruptcies and Reorganizations

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Sources: Cerved; and IMF staff estimates.

21. The current framework does not seem conducive to the reorganization of viable businesses.

  • Given the number of different techniques available to debtors, by way of informal or hybrid restructuring options (piani di risanamento, accordi di ristrutturazione), and by way of the reformed reorganization procedure (concordato preventivo), one could raise the question of whether successive legislative reforms have been effective.

  • The comprehensive reform of the concordato preventivo in 2005–2007 did not result in an efficient reorganization procedure. Some targeted revisions (e.g., in 2012) intended to make the process more attractive to debtor companies, by allowing them to access the procedure and enjoy bankruptcy protection while preparing a reorganization plan (concordato con riserva, or concordato in bianco). After an initial period in which there was an unprecedented increase in the number of proceedings, including the abuse of the procedure by recalcitrant debtors, corrective measures were introduced (a 2013 reform made it possible for the court to appoint an examiner in these cases) and the number of cases decreased considerably.

  • Overall, most reorganization attempts fail and result in protracted liquidations. The liquidation process is slow, due to frequent and extensive litigation within the insolvency process. In turn, the lengthy process tends to reduce the value of the enterprise even further. The existence of numerous priorities in favor of creditor groups—including public creditors—results in a very low return for unsecured creditors.

22. The lack of legal tools for effective corporate restructuring poses a constraint. There are difficulties in the use of advanced corporate restructuring options, such as recapitalizations and debt/equity swaps. Informal restructuring is not anchored in a set of principles that govern negotiations among financial institutions. The informal restructuring arrangements are limited in their contents and disconnected from the other procedures in the insolvency framework. There are no procedures for the insolvency of groups of companies. There are no specific restructuring options for SMEs.

23. These issues should be considered in the context of a less-than-optimal system for the protection of creditor rights. The current system does not afford enough opportunities for enterprises to use efficiently their movable assets as collateral: the system for the creation and registration of security interests is fragmented and rigid. Also, the law does not provide for swift remedies in case of default by borrowers because enforcement necessitates multiple procedural steps in a court system overburdened by the backlog of existing cases. These factors affect both the stock of NPLs and the inflow of NPLs in the system. The judiciary is overburdened and lack specialization in debt enforcement, restructuring and insolvency. The tribunali delle imprese, created in 2012, and whose competences are being revised, represent an attempt at specialization of the judiciary, but insolvency matters are still dealt by the sections of the civil courts, which effectively prevents specialization in the smaller districts. The general problems affecting the civil justice system are especially relevant in the area of insolvency and creditor rights. The institutional framework includes an established profession of insolvency administrators, but their expertise focuses more on liquidation than in the reorganization of enterprises

D. Impact from Balance Sheet Clean-up

24. The removal of bad debt from the banks’ balance sheets has the potential to boost solvency levels (Figure 10, left chart). Under the assumption that sofferenze are adequately provisioned, write-off would cause RWAs to drop, freeing up valuable capital that can be used to extend new loans and to support Italy’s recovery. Estimates show that a full write-off of existing sofferenze would (i) increase the banking system’s Tier 1 capital ratio by 1.1 percentage points, and (ii) free-up capital resources to generate about €150 billion in new loans to the real economy.13

Figure 10.
Figure 10.

Italy: Drivers of Corporate Default Risk and Model Performance

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

25. However, if bad debt was not adequately provisioned, write-off would force banks to close provisioning gaps, hitting bank capital directly (Figure 10, right chart). Sensitivity tests show that provisioning gaps of 3 and 15 percent would result in capital shortfalls of €0.4 to €3.6 billion (in 5 and 8 banks), respectively, against the January 2016 minimum Tier 1 capital requirement plus capital conservation buffer (together 6.625 percent of RWA). Hypothetical provisioning gaps are calculated based on NPL gross value.

26. These results indicate that banks’ solvency levels would on average be sufficient to digest balance sheet clean-up. Banks could strengthen their solvency position through the reduction in RWAs and the generation of new average-risk loans at current market funding and lending rates. Furthermore, potential additional losses at the time of write-off as a result of insufficient loan loss provisioning would be absorbed by banks’ capital while staying well above regulatory minimum capital requirements. In some banks, capital would drop to or below minimum requirements, but at less than 0.2 percent of annual GDP the total shortfall would be relatively small.

E. Recommendations for Bringing Down Nonperforming Loans

27. The size and prevalence of NPLs call for a comprehensive strategy. This section presents some preliminary recommendations to reduce the stock and decelerate the flow of NPLs. In a nutshell, this may involve a “stick-and-carrot” approach, with banks facing strong disincentives/incentives for keeping/removing NPLs from their balance sheets. Separately, to facilitate NPL resolution, the conditions for corporate debt enforcement, debt restructuring, and insolvency would need to be improved systematically.

Reducing the Stock of NPLs

28. Any remaining uncertainty about Italian banks’ asset quality needs to be resolved. Applying a common asset classification system, the ECB’s 2014 Comprehensive Assessment revealed sizeable provisioning gaps among the 15 participating banks (which are expected to be filled by July 2015). However, uncertainty remains about the rest of the Italian banking system. Since January 2015, a new asset classification framework harmonized on the EU level, has become effective and is applicable to all banks operating in Italy. If under the new framework provisioning and capital shortfalls are identified, the coverage would need to be topped up and capital buffers strengthened, including through retention of earnings and capital increases.

29. Regulatory and supervisory action would be required to increase the ability of the banks to deal with NPLs. This would include measures such as the establishment of specialized workout units, the development of codes of conduct for the treatment of arrears (e.g., like in Ireland and Spain), further increase of reporting obligations on NPLs, and guidance on restructuring options available to banks, including on triage approaches (e.g., Iceland).14 This may also include the development of debt restructuring principles for multilateral workouts (Austria, Slovenia).

30. Strong disincentives for keeping bad debt on balance sheets would foster NPL clean up. Supervisory actions that effectively introduce time limits for write-off of vintage NPLs can play an important role in this respect. One option is to raise considerably capital charges for vintage NPLs after a certain period. Separately, a macro-prudential instrument can be introduced that limits going forward the size of the NPL and the impaired (sofferenze) portfolio.

31. In parallel, tax incentives can motivate quicker disposal of NPLs. For instance, consideration could be given to shortening further the five-year period over which banks are allowed to deduct loan-loss provisions from taxable revenues. While there is no best international practice in the recognition of write-offs of NPLs for tax purposes, there is consensus that the regulatory and the tax regime should be as aligned as possible. There is wide variety of regimes, but those that are most widely used are i) the “charge-off” method (USA, Australia), that allows for deductibility of the whole loan once it has been classified as uncollectible for regulatory purposes; and ii) the “reserve method” (France, UK, and Canada), that allows the credit institutions to make tax deductions corresponding to each of the special provisions corresponding to every downgrade in classification of the loans. Both methods present advantages and disadvantages, but both are superior to the system currently in force in Italy.

32. A further development of the Italian market for restructuring NPLs would help banks improve asset quality quickly and efficiently. Involving outside investors that work together with the banks on corporate restructuring, or directly purchase NPLs, can be seen as a regular tool in the management of NPLs. The transfer of NPLs would relieve banks from the burden of debt collection and foreclosure, and quicker resolution can help conserve recovery values, facilitate debt restructuring and debt/equity conversions, inject capital into firms, and clean-up Italy’s corporate sector.15

33. Intra-segmental cooperation can help smaller banks improve their NPL management infrastructure. As discussed in Section C, smaller banks appear to have insufficient risk and NPL management capacity and limited (or no access) to distressed debt markets.16 In order to reduce the stock of NPLs, banks in the same segment of the banking system (e.g., banche popolari) could cooperate in cleaning up balance sheets. One possibility is to establish a corporate restructuring vehicle (CRV) within a bank segment, aimed at managing NPLs and orderly corporate restructuring. Such a vehicle could be owned by the banks but managed independently, in order to facilitate the design of appropriate incentive schemes, as well as the build-up of NPL management capacity within the segment. Separately, sofferenze need to be resolved decisively (see below).

34. AMCs can also play a useful role in developing a distressed debt market. This can be achieved through setting uniform market standards for pricing distressed assets, conducting regular auctions to facilitate price discovery, and publishing data that facilitates the price discovery, including, for instance, payment history, default rates, and collateral values.

35. A centralized, state-backed AMC could also help by kick-starting a distressed debt market. A centralized solution could be a useful complement, not a substitute, to the economic incentives, supervisory disincentives, and legal changes outlined above. Unlike the AMCs in Spain or Ireland, however, this vehicle could be of smaller scale—concentrating on SMEs, only temporary, and with limited government involvement. Furthermore, transfer prices would need to be conservative enough not to invoke EU State Aid rules.

36. Any such vehicle would need to be properly designed and in line with best international practice (see Box 1). Specifically, the AMC should (i) purchase assets at market prices; (ii) use transparent and uniform valuation criteria; and (iii) have strong governance, including operational independence and accountability. Given the small fiscal space and EU State Aid rules, the government should have minimum involvement (minority stake, limited guarantees) and receive adequate remuneration (conservative pricing, equity warrants).

Addressing Corporate Debt Overhang and Bank Asset Quality

The AMC could evaluate SME sofferenze and incagli, and restructure or resolve those that meet selective eligibility criteria. The centralized AMC could be managed by an asset valuation and coordination entity that decides on restructuring (corporate restructuring vehicle; CRV) or liquidation (corporate liquidation vehicle; CLV). The AMC could implement a triage approach, in which the decision to restructure rests on a conservative estimate of a firm’s ability to service its debt using standard criteria or assessing firms, including interest coverage and future cash-flows. Purchase of NPLs should be selective and based on clear eligibility criteria. Also, the strategy could prioritize the removal of those NPLs that are considered critical from a public policy point of view, and are important for the rehabilitation of financial institutions and the corporate sector.1 Nonviable firms should not be subject to management by the CRV, but be liquidated and removed from the market. The CLV on the other hand should be concerned only with the quick disposal of assets to conserve collateral values, and a maximization of asset values. The concentration of collateral provides the AMC with substantial power and could speed up the insolvency process and help preserve collateral values. Hence, this approach is very different from an indiscriminate purchase of defaulted loans.

Efficient transfer pricing would allow for the implementation of a state-backed AMC without distortion of competition. Troubled loans need to be classified and provisioned for correctly. This is essential for the strategy to be effective, and to avoid triggering State Aid rules. Assets subject to transfer need to be priced at or below the current market value (as otherwise this would constitute a public subsidy). Any loss from the revaluation would need to be taken by the bank before or at the time of transfer. Tax incentives for disposal and supervisory disincentives for holding on to NPLs will foster the transfer of NPLs to the AMC.

The properties of the AMC:

  • Operational independence and limited public ownership. The government should have no means to intervene in the operational work of the AMC. The centralized, state-backed AMC would need to be predominantly owned by private investors, with the sovereign holding a minority stake. The entity should be insulated from political interference in both the disposal and management of NPLs. CRV and CLV should be established as independent entities, with different goals and targets.

  • Temporary character. Banks could be given a period of six months to transfer existing sofferenze and incagli to the AMC. This would help jump-start the market, while preserving bank-lending standards going forward. Furthermore, the vehicle should be self-liquidating in order to limit moral hazard and warehousing of assets. Incentive structures should be designed in a way that profit maximization within a predefined period of time is warranted.

  • Limited scope. The AMC would concentrate on SME sofferenze and incagli as the market failure is most prominent in that part of the corporate sector. About two thirds of sofferenze and incagli are related to the corporate sector, 30 percent of which is SME exposure. Gross (net) of provisions, the potential volume of the vehicle is estimated at €55–65bn (€25–30bn). Exposure size restrictions would reduce the volume further.

  • Limited role of the Government. The involvement of the state is limited to three areas: (i) the design of the AMC; (ii) the provision of guarantees on the mezzanine tranches of the emitted assets (in order to improve investor sentiment and enhance marketability of the assets emitted by the AMC); and (iii) the provision of equity in form of a minority stake in the AMC.

  • Sale and resolution of NPLs. The AMC aims at both the sale and the securitization of exposures. Recent NPL sales by large Italian banks clearly underline the market appetite for higher yield investments. But the identification problem needs to be solved in order to increase volumes. As regards securitization, a guarantee for the mezzanine tranches of Asset Backed Securities issued by the CRV will help attract institutional investors facing minimum credit quality requirements. Such bulk sales allow for quick resolution and the generation of immediate cash-flows. Individual sales, on the other hand, can be used for larger loans that profit from specific market valuation. The possibility of granting special enforcement powers to the AMC could raise constitutionality issues, and cannot be a substitute for a general improvement of the legal enforcement and insolvency regime, which should benefit all economic actors.


Italy: Structure and Debt of Firms

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Source: ORBIS. IMF staff calculations.
1 He (2004), p.11.

Decelerating the Flow of Nonperforming Loans

37. Banks would need to demonstrate their efforts to reduce NPLs. BoI and the Single Supervisory Mechanism (SSM) are recommended to take an even more proactive approach to supervision, including the issuance of further guidance on and regulation of NPL management in banks. For instance, banks should be required to report regularly to the supervisory authority detailed loan portfolio information and develop decisive NPL workout plans if needed. In order to complement these efforts, supervision should strictly enforce banks’ compliance with existing and future rules and guidance.17

38. Applying the same provisioning approach for all banks would alleviate the problem going forward. Forthcoming IFRS 9 will include a clear definition of write-off.18 When the new accounting rules become applicable, scheduled for 2018, banks can no longer apply loan de-recognition and keep loans on balance sheets until all legal means are exhausted. They would be forced to write off earlier, opening the way for corporate restructuring or liquidation. Common definitions can help improve cross-country comparability, increase transparency, and reduce forbearance going forward.

39. Better cooperation among banks could become a catalyst for decisive clean-up. Common asset disposal and management platforms would also pave the way for further consolidation. The recent reform that requires the largest ten banche popolare to transform into joint-stock companies has already had a positive impact on bank valuations, and going forward will result in a second wave of consolidation in the sector.

40. To prevent a further increase in NPLs, banks may need to improve their credit risk and NPL management capacity. Specifically, banks may need to build up internal expertise in collateral valuation, insolvency procedures, and financial restructuring. Alternatively, these activities could be outsourced to specialized third parties; or banks could cooperate and use synergies as discussed above.

Legal Recommendations

41. Further revisions to the insolvency framework would be necessary to allow the swift liquidation of non-viable businesses and a better reorganization of viable enterprises.

  • A rationalization and better integration of the different options available to enterprises in distress (in informal, hybrid and formal insolvency procedures) would ensure that these procedures work efficiently and seamlessly (e.g., possibility of “pre-packed” arrangements).

  • Insolvency procedures could be initiated at an earlier stage, increasing the recovery possibilities, with effective provisions establishing directors’ liability (wrongful trading).

  • The insolvency regime should also offer wider restructuring possibilities, including recapitalizations and debt/equity swaps without allowing shareholders to interfere with these solutions.

  • Creditor priorities may need to be revisited to increase the rate of return for unsecured creditors. Public creditors may have to take a more constructive role in enterprise restructuring, possibly assisted by guidelines on public sector participation in insolvency processes.

  • Insolvency processes should be further streamlined, reducing appeals and opportunities for delay, while, at the same time, providing for adequate protection for all participants.

  • The increased flexibility in the sale of assets should be effectively implemented, and supported by the use of mechanisms such as Internet platforms.

  • The reform could also include a simplified debt restructuring mechanism for SMEs, preferably a hybrid framework (see Bergthaler and others, 2015), based on a triage approach establishing the viability of SMEs according to basic objective indicators. Principles for debt restructuring could be useful for the restructuring of the debt of large companies. A more specific set of principles could address the situation of over-indebted SMEs.

42. The legal environment of credit would greatly benefit from reforms of the law of secured transactions. Changes in this area should improve the possibility of creating security interests over movable assets, with corresponding reforms of the registration systems (Box 2), and coordination among existing registry systems. Fiduciary contracts and other mechanisms to increase out-of-court enforcement options are necessary for the take up of new security interests and the improvement of the existing ones. These reforms would reduce the inflow of NPLs by improving the position of creditors and the realization of collateral.

43. The institutional framework for insolvency and creditor rights remains a substantial challenge. The program of judicial reforms, especially focused on civil justice, should continue. Further special measures to address the backlog at the court seem necessary. These could include more specialization of judges in commercial matters, especially in debt enforcement, restructuring and insolvency, and the reinforcement of insolvency administrators.

The Importance of Efficient Secured Transaction Laws

Security rights provide certainty for creditors and increase access to finance. The importance of the secured transactions system has been highlighted by legal and economic experts, and has given rise to detailed best practice recommendations (UNCITRAL, 2007). The possibility of using different types of collateral creates opportunities for more efficient financing arrangements. This is the case with the creation of security interests over movable collateral: there are important differences in the flexibility and efficiency of secured transactions over movable assets. These affect the availability of credit for enterprises, particularly SMEs.

The use of movable collateral in the context of entrepreneurial activities requires a high degree of flexibility in the creation of security interests. These should cover not only specific assets, such as machinery, but also categories of assets, such as inventory, equipment and accounts receivable, without the need to specify each and every asset covered by the security interest. The main reason is that these security interests should allow the debtor to remain in possession of the assets. This implies that the debtor will be able to use the assets, and even, with respect to inventory, to sell the assets in the ordinary course of business. Dispossession of the debtor is not necessary because these security interests are recorded in public registries. Registries are “notice-based”: they merely provide notice to the public of the creation of the security interest. Registries are computerized and allow for interaction with other registries and databases.

Priority of the security interest is another crucial element in the design of the legal framework. The protection of the borrowers’ interests and the corresponding positive effects for borrowers require that the position of the secured creditor be granted a priority status. Registration establishes the point in time when the lender acquires his priority rights versus other creditors and third parties.

Finally, an efficient secured transactions system requires swift enforcement of the secured claim. In the case of movable collateral, the speed in enforcement of the secured claim is critical, because the depreciation of assets is fast. Enforcement mechanisms may include summary judicial procedures, or out-of-court enforcement. The possibility of using fiduciary contracts can be understood in this context as one way of improving the enforcement of security interests over movable collateral without having to resort to the court system.

The economic importance of efficient secured transactions laws has been emphasized in the economic literature. According to recent research, loan-to-values of loans collateralized with movable assets are on average 21 percentage points higher in countries with strong-collateral laws relative to immovable assets. Further, stronger collateral laws tilt collateral composition away from immovable to movable assets (Calomiris and others, 2015).


  • Bank of Italy, 2015, Financial Stability Report, No. 1 (April).

  • Bergthaler, W., Kang, K., Liu, Y., and Monaghan, D., 2015, “Tackling Small and Medium-Sized Enterprise Problem Loans in Europe,” SDN/15/04 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Calomiris, C.W., Larrain, M., Liberti, J., and Sturgess, J., 2015, “How Collateral Laws Shape Lending and Sectoral Activity,” Columbia Business School Research Paper No. 14–58.

    • Search Google Scholar
    • Export Citation
  • Cerved, 2015, Osservatorio su fallimenti, procedure e chiusure di imprese, 4q 2014.

  • Garrido, J. M., 2012, Out-of-Court Debt Restructuring (Washington: World Bank).

  • International Monetary Fund, 2013, “Italy: Financial System Stability Assessment,” IMF Country Report 13/300 (Washington).

  • Jassaud, N., 2014, “Reforming the Corporate Governance of Italian Banks,” IMF WP/14/181 (Washington: International Monetary Fund).

  • Jassaud, N., and Kang, K., 2015, “A Strategy for Developing a Market for Nonperforming Loans in Italy,” IMF WP/15/24 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Klein, N., 2013, “Nonperforming Loans in CESEE: Determinants and Impact on Macroeconomic Performance,” IMF WP/13/72 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Kopp, E., 2015, “Macro-Financial Dynamics in OECD Countries,” forthcoming.

  • Lanau, S., Esposito, G., and Pompe, S. 2014, “Judicial System Reform in Italy—A Key to Growth,” IMF WP/14/32 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Rath, C., 2014, “Effects of Collateral Law Reform on Firms’ Access to Finance,” Essays on the Political Economy of Institutions, Theses and Dissertations, UWM Paper 483.

    • Search Google Scholar
    • Export Citation
  • Savafian, M., Fleisig, H., and Steinbuks, J., 2006, “Unblocking Dead Capital: How Reforming Collateral Laws Improves Access to Finance,” Private Sector Development Viewpoint, No. 307 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • UNCITRAL, 2007, Legislative Guide on Secured Transactions (New York: United Nations, published 2010).

  • World Bank, 2003, Analyzing and Managing Banking Risk, A Framework for Assessing Corporate Governance and Financial Risk, 2nd ed., 2003, (Washington).

    • Search Google Scholar
    • Export Citation

Annex I. Dynamic Bank-by-Bank Panel Regressions

Fixed Effects Panel Regression 2005–2014

article image
Sources: SNL, World Economic Outlook.

an increase in REER indicates appreciation.

System GMM Estimation, 2005–2014

article image
article image
Sources: SNL, World Economic Outlook.

an increase in REER indicates appreciation.

Annex II. State Space Modeling and Filtering

Dynamic State Space models (DSSM) have a number of important advantages:

  • First, nonlinearities in the functional relationship between variables can be modeled, and structural changes over time can be included explicitly. This is an important property as the influence of different macrofinancial factors on credit risk typically changes during periods of financial distress (“correlation breakdown”).

  • Second, DSSMs make it easier to handle measurement errors and missing values as they allow describing the observed data as a function of the underlying state process. This is an important feature when dealing with snapshots of asset quality data.

  • Third, DSSM are not as myopic as classical time-series models, which tend to wash out older historical realizations while concentrating mainly on recent performance. However, there is always the possibility for past patterns to reemerge. SSMs have a long memory, and model each observation as a function of its entire history.

  • Fourth, the performance of a DSSM can be measured using true out-of sample criteria, and the forecast error is not invariably an in-sample error (as in classical regression models), because in any estimation at point in time t < T, only the data up to time t is known.

This allows for much more realistic description of the dynamic relationship between variables over time. Second, the measurement of credit risk variables tends to be fairly blurred, and it is unlikely that the default or default probability time-series at hand can give the true risk. The DSSMs helps push imprecise measurements towards the true values, thereby increasing precision.

The static multivariate linear regression model with a constant, i.e.,


can be specified to include time-varying coefficients:


A dynamic multivariate regression model can be expressed in State Space form as


where θt denotes the regression coefficients in the state vector (β0, βi,…,βn), Ht′ = (1, x1,t,…,xn,t) the n regressors, and Ft is the state transition matrix from time t to t+1. The variance of the multidimensional noise vectors vt and wt are driven by the time-variant matrices Vt and Wt. If the variances in Vt are larger than zero, the coefficients become time-varying.1 Hence, for the dynamic multivariate state space regression model, both Vt and Wt are nonzero.2

The dynamic coefficients can be estimated using the Kalman Filter, a recursive linear estimation algorithm. The Kalman filter is a set of recursion equations determining, conditional on the information available at time t, optimal estimates of the state vector. The estimation is performed recursively, i.e.,


with the updating equations:


The variance of the state vector θt forecast is denoted by Pt|t. And the variance of the forecast of θt+1 is described by Pt+1|t. As new data becomes available, the Kalman Filter updates both forecasts based on the two variance-covariance matrices Pt+t and Pt+1|t. The updating equations (10) and (11) show how the variance of the state vector and the variance of its forecast are updated. As more and more new observations come in, variance decreases as uncertainty reduces:


The Kalman Gain, Pt|t−1Ht (HtPt|t−1Ht + Rt)−1, assigns the optimal weight to the unexpected forecast error εt, which depends on the new observation yt. The update of θt|t due to the new observation yt is a combination of the old estimate forecast and the forecast error.

The DSSM model outperforms other models in describing Italian corporates’ default probabilities (Figure A1). The chart gives in-sample estimates for (i) multivariate contemporaneous static regression without lags, (ii) with lags,3 and (iii) out-of-sample fitted values for the DSSM. Even the out-of-sample fit of the dynamic regression is better than the in-sample fit of the static models.

Figure A1.
Figure A1.

Italy: Key Drivers of Corporate Sector Default Risk

Citation: IMF Staff Country Reports 2015, 167; 10.5089/9781513559681.002.A003

Sources: IMF WEO, OECD; IMF Staff Calculations.

Prepared by Jose Garrido (LEG), Emanuel Kopp (MCM), and Anke Weber (EUR).


NPLs in Italy consist of four categories: impaired/bad debt or sofferenze (loans in a state of insolvency), substandard (incagli), overdue, and restructured. In 2014, the ECB performed an AQR of Italy’s largest 15 banks, based on harmonized definitions of loan quality, asset classification, and provisioning. This resulted in an aggregate nonperforming exposures ratio for the 15 banks of almost 22 percent.


Of course, if most loans had been to households, the prevalence of small loans would likely have been even higher.


This is consistent with Jassaud (2014), who concludes that due to their corporate governance structure, large cooperatives tend to display lower buffers and weaker asset quality metrics than the system average.


Data for the 402 banks do not typically span beyond 2012 for most variables.


The causality between RoE and NPLs could be two ways as higher NPLs also worsen banks’ equity position. This is why all independent variables are lagged.


House prices are only significant in the fixed effects regression, not when performing system GMM.


This means that one in 50 corporates which have not defaulted already are expected to do so within 12 months.


However, Italian banks do have significant unrealized capital gains from sovereign bond holdings. See BoI Financial Stability Report April 2015.


This has also contributed to the build-up of Deferred Tax Assets (DTAs) in Italian banks, as remaining provisions could be deducted in equal installments over a period of 18 years (and at a lower net present value).


In fall 2014, UCG sold €2 bn in NPLs in the market. And in February 2015, the bank announced the sale of its entire participation in UniCredit Credit Management Spa, including a €2.4 bn NPL portfolio, to a private AMC (Fortress/Prelios) aimed at both liquidation (bad debt) and NPL management (other NPLs), in particular for SME NPLs. Net of provisions, bad debts at UCG stood at about €36 billion in 2013.


These results reflect the impact from (i) write-off of sofferenze, (ii) generation of new, average-risk loans, (iii) reduced loan administration costs for bad debt, and (iv) the profit margin for newly generated loans, based on current (TLTRO) funding rate and average effective SME lending rates observable from the market.


See Act 107/2009, “Act on measures to assist individuals, households and businesses due to extraordinary circumstances in the financial market,” Ministry of Social Affairs and Social Security, 2009.


See also Jassaud and Kang (2015); IMF Global Financial Stability Report April 2015.


The 2013 FSAP demonstrated that medium-sized banks are less profitable, more vulnerable to macrofinancial shocks, experience relatively more phase-out of capital due to the implementation of Basel III, and were not able to strengthen capital like other banks did (in particular due to their ownership structure).


The BoI’s earlier guidance on writing off defaulted exposures would need to be enforced strictly. As mentioned, IAS 39 does not define when and how uncollectable loans are to be written off. Therefore, until the new international accounting rules become effective, it is critical that banks adhere to existing BoI guidelines.


Loan de-recognition requires the exhaustion of all legal means or to waive contractual rights on the loan before the exposure can be removed from the balance sheet. IFRS 9 will include a definition of write-off different from loan cancellation.


Since Wt > 0, the state space model also includes the correlation patterns between the variables in Ht.


The signal-to-noise ratio V/W > 0 indicates how adaptive the coefficients are.


The optimal multivariate regression models were determined using the branch-and-bounds algorithm for best subset selection. Expected sign tests were included.

Italy: Selected Issues
Author: International Monetary Fund. European Dept.