Journal Issue
Share
Article

Portugal: Selected Issues

Author(s):
International Monetary Fund. European Dept.
Published Date:
September 2016
Share
  • ShareShare
Show Summary Details

Challenges Confronting Portuguese Banks: Profitability and Asset Quality1

A. Introduction

This paper provides an overview of the main challenges confronting the Portuguese banking system, and identifies two key sources of vulnerability: low profitability and lackluster asset quality. It then studies the drivers of profitability and asset quality in Portuguese banks before, during, and after the global financial crisis and examines policy options for increasing profitability and strengthening asset quality over the medium-term.

We argue that that key characteristics of Portuguese banks contributed to a build-up of vulnerabilities that began in the mid-1990s and continued following euro accession during 1999–2007. The banking sector was profitable throughout this period. However, it also experienced: (i) low levels of capital and increased solvency risk; (ii) an overreliance on wholesale funding, which in turn increased liquidity risk; (iii) concentrated ownership, which acted as a disincentive to capital increases that could dilute ownership shares; and (iv) the misallocation of credit, especially to the construction sector. The global financial crisis in 2008 would begin to expose those vulnerabilities across Europe, and would impact Portuguese banks as well.

Based on empirical analysis of a sample of euro area banks over the period 2010–2015, we find that low profitability was associated with (i) lower levels of capital; (ii) lower liquidity; and (iii) weaker macroeconomic fundamentals (lower GDP growth and higher public debt levels).2 Moreover, the effects of these variables appear to be larger for stressed economies (including Portugal) than for non-stressed economies.

In recent years, Portuguese banks were able to increase liquidity and, to a lesser extent, capital. However, significant challenges remain as low or negative profitability, weighed down by low growth, and poor asset quality, due to legacy issues, continue to impact the outlook for the banking sector.

Given limited policy options to restore profitability and improve asset quality, we argue that Portuguese banks should: (i) further reduce their operating costs; (ii) increase the pace of non-performing loan (NPL) disposal; and (iii) raise capital to absorb losses from restructuring and writeoffs.

B. The Pre-Crisis Period (1995–2007): Build-up of Vulnerabilities

1. During the pre-crisis period, Portuguese banks were profitable compared to European peers (IMF (1998)). On the income side of the balance sheet, net interest income—bolstered by high loan volumes and comparatively high interest rates—was the main driver of profitability. This strong revenue performance offset higher than average operating costs on the expense side and kept even inefficient banks profitable.

Portugal: Income Statement of the Banking System

(Percent of assets)

Sources: Bank of Portugal; and IMF staff calculations.

2. The drivers of bank profitability began to change in the late-1990s. While the banking system remained profitable throughout the pre-crisis period, net interest margins were increasingly compressed by competition following euro area accession. Banks responded by strengthening alternative sources of income and decreasing their costs to preserve profitability (IMF (2005)). Notably, banks further reduced staff costs relative to assets (from 1.3 percent of assets in 1999 to 0.9 percent of assets in 2007). The efforts by Portuguese banks to increase their efficiency relative to European peers were reflected in a lower cost-to-income ratio by 2007. However, regulatory capital to risk-weighted assets was declining (to just 7 percent at end-2007) and lagged behind peers.

Portugal: Efficiency Ratio of Banks, 2007

(Ratio of non-interest expenses to gross income)

Source: IMF Financial Soundness Indicators.

3. Banks also became increasingly reliant on funding from wholesale markets. Portuguese banks migrated from a retail-oriented model, reliant on customers’ deposits, to a wholesale-funded model, characterized by a reliance on interbank lending and bond issuance, which contributed to a stable and positive return on assets (averaging 1 percent between 1999 and 2007). Euro area accession enabled banks to obtain cheaper lending in euro-denominated wholesale markets. By 2007, Portuguese banks had one of the highest loan-to-deposit ratios in the euro area, increasing from 100 percent in 1999 to 150 percent in 2007.

Regulatory Tier 1 Capital, 2007

(Percent of risk-weighted assets)

Source: IMF Financial Soundness Indicators.

4. In an effort to retain profitability, the banking sector rapidly expanded its role in the economy resulting in a misallocation of credit. Banks assets as a share of GDP rose from 184 percent of GDP in 1997 to 250 percent in 2007 (Figure 1). Large capital inflows were intermediated by Portuguese banks and directed mainly towards domestic non-tradable sectors. This led to a misallocation of real capital and the rising leverage of the corporate sector (IMF (2011, 2013)) as reflected in the increase in the share of loans to the real estate and construction sectors from less than 13 percent of non-financial corporation (NFC) loans in 1997 to 38 percent in 2007.

Figure 1.Bank Asset Structure, 1999–2007

Portugal: Structure of Bank Liabilities

(Percent of Total Liabilities)

Sources: Haver Analytics; and IMF staff estimates.

Euro area: Selected Loan-to-Deposit Ratios

(Percent)

Sources: Bank of Portugal; OECD; and ECB.

5. The misallocation of credit to the construction sector was especially pronounced. In Portugal, loans to the construction sector surged after the country joined the euro area, increasing from €8.4 billion in 1999 to €26.1 billion in 2008. However, over the same period the share of the construction sector in employment and value added declined along with productivity (Reis (2013). Moreover, the value added in the construction sector stopped growing in the early 2000’s and remained stable throughout the decade, while bank credit to the construction sector ballooned (Figure 1). The decoupling in the dynamics of value added and bank credit in Portugal are in sharp contrast to what would be expected from an efficient banking sector. For example, in Spain growth in lending to the construction sector was largely a reflection of increases in the value added of that sector.

6. The ownership structure of Portuguese banks exacerbated the risk of credit misallocation. The banking sector was dominated by five banking groups with highly concentrated ownership: Caixa Geral de Depósitos (CGD), Banco Comercial Português (BCP), Banco Espírito Santo (BES), Banco Português de Investimento (BPI) and Santander-Totta. Large economic groups played a major role as shareholders in the banking sector (Valerio et al. (2010)), which likely facilitated relationship-based lending with large groups in the construction and real estate sectors. This would be consistent with Taboada’s (2011) finding that increases in domestic ownership of banks adversely affect the allocation of capital through increased lending activity to less productive industries, especially in the non-tradable sector.

7. Portuguese banks headed into the crisis with pronounced vulnerabilities. Banks’ low capital levels and reliance on potentially volatile funding sources limited their ability to absorb shocks. On the asset side, credit misallocation to the real estate and construction sectors sowed the seeds for poor asset quality and the subsequent rise of non-performing loans once the economic downturn began.

C. The Crisis Hits (2008–2013): Crystallization of Risks

8. The global financial crisis resulted in a sudden stop of capital flows to Portugal that tested the resilience of its banking system. The sudden stop led to an economic slowdown, which was further amplified by an increase in interest rates reflecting higher funding costs for banks. The reliance of Portuguese banks on wholesale funding markets resulted in liquidity issues which forced banks to increase ECB funding (IMF (2013)). Excessive leverage in the corporate sector made firms unable to repay their loans, resulting in a rise in non-performing loans, especially in the construction sector. Banks reacted by deleveraging, through a decline in outstanding loans on the asset side, and by attracting deposits to change their funding structure and move towards a retail-based model.

9. The quality of the assets on banks’ balance sheets continued to deteriorate throughout the crisis period. The real estate and construction sector were especially hard hit by the economic slowdown, and banks’ disproportionate exposure to these sectors took a mounting toll. The ratio of NPLs to total loans nearly tripled from less than 4 percent in late-2008 to 11 percent by 2013, resulting in large impairment charges. NPLs increased at such a steep rate that banks were unable to fully absorb the losses and saw their provisioning coverage ratios decline.

10. Bank profitability deteriorated sharply during the crisis. The lack of profitability observed over this period is consistent with the findings of our empirical analysis of the drivers of bank profitability in stressed euro area economies (Box 1). We find that in stressed economies (such as Portugal’s during 2010–15), the level of bank capital and the weight of securities in the asset portfolio tend to have a significant impact on bank profitability. Portuguese banks went into the crisis with the second-lowest levels of capital in the euro area, and strong sovereign bank linkages. Net interest margins were compressed to very low levels (from 1.9 percent in 2008 to just 1.1 percent in 2013), especially relative to euro area peers, because higher bank lending rates were unable to offset higher funding costs. Overall, the banking system experienced losses each year between 2011 and 2014.

Portugal: Non-Performing Loans and Provisioning

Source: IMF Financial Soundness Indicators.

Net Interest Margin

(Share of Gross Income)

Sources: European Central Bank; and OECD.

Box 1.Portugal: An Empirical Analysis of Drivers of Bank Profitability, 2010–15

Data. We collect data on 183 euro area banks with at least €20 billion in assets over the period 2010–15 from SNL Financial and Bankscope. Table 1 shows that Portuguese banks have: (i) lower capital ratios than most other euro area banks (except Greece); (ii) a higher share of deposits, (iii) a higher RWA density due to lower asset quality and a more retail-oriented business (as customers’ loans have higher risk weights than debt securities); (iv) lower net income than most other euro area banks (except Greece, Spain and Ireland); and (v) have the highest operating costs to assets ratio (with Italy and Greece).

Table 1.Descriptive statistics 2010–15
CountryATBEFRDEGRIEITLUNLPTESEA
Assets (€bn)571222621127653922321952154127
CET1 ratio10.915.112.914.810.216.012.116.613.311.012.013.2
Capital ratio15.1181519.111.922.614.819.916.512.113.516.5
Deposits/Assets39.758.644.551.152.624.951.24845.760.152.848.5
RWA/Assets54.331.135.336.463.444.251.235.131.858.84943.2
Income/Assets0.130.360.330.19−2.30−0.630.150.520.05−0.28−0.330.06
Operating costs/Assets (%)1.441.181.371.151.760.692.091.110.781.651.331.39
Number of banks178355751038115733183
Number of observations984820133727532205727331391240
Sources: SNL Financial; Bankscope; and IMF staff calculations.
Sources: SNL Financial; Bankscope; and IMF staff calculations.

Econometric framework. We estimate a panel with cross section and time fixed effects:

Where Πi,t is the profitability of bank i on year t (measured by net income in percentage of assets, return on assets or return on equity), δt and Tt are respectively cross section and time fixed effects, Xi,t are bank-specific variable related to solvency (equity capital ratio, Tier 1 ratio, total capital ratio), funding (deposit to assets, deposits to wholesale funding), risk appetite (risk weighted assets to total assets), size (log of total assets), Zi,t are macroeconomic variables (GDP growth, public debt to GDP ratio, EONIA et VIX).

Results. For bank-specific variables, the size of the bank (capturing scale efficiency) and the current level of capital have the most (positive) impact on profitability, with liquidity (measured by deposits to assets) having a more limited contribution (Table 2). For macroeconomic variables, real GDP growth has a significant and large positive while public debt has a negative (and smaller) effect on profitability. Since the sample includes banks in stressed countries (Greece, Ireland, Italy, Portugal and Spain) and non-stressed (Austria, Belgium, Germany, France, Luxembourg and Netherlands) we run separate estimations for each type of country.

Table 2.Drivers of profitability
Net income/Assets All banksNet income/Assets Stressed CountriesNet income/Assets Other countries
Total assetsO.948***0.7120.718***
Equity/Assets0292***0.316***0.229***
Securities/Assets−0,010−0,045**0.020***
RWA density−0.004−0.017−0,002
Deposits/Assets0.028***0.031*0,018***
GDP growth0.241***0.250***0.036
Public debt−0.023***−0.040***0.006
EONIA1.1391.559***−0.199
VTX.−0.088−0.174***0.032***
Adjusted R20.420.420.34
Observations1115445670

In stressed economies, the impact of capital on profitability is slightly larger and trading operations tend to reduce profitability (due to the high exposure to domestic sovereign bonds), while the size of the bank is not significant. Regarding macroeconomic variables, domestic public debt has a larger negative effect on profitability and the level of interest rates (measured by the EONIA) has a large positive impact on profitability (possibly due to the widespread use of variable rate loans in those countries and a higher reliance on net interest income).

In other economies, the size of the bank is the main factor along with capital and trading has a positive contribution to profitability. Macroeconomic variables are not significant (as banks are larger and hence more diversified internationally) except the VIX which is positive, implying that higher volatility results in increase in profits (possibly due to income from derivatives trading and/or trading gains on sovereign bonds portfolio).

The results are broadly similar with recent studies on banks’ profitability in Europe (ECB (2015), Cheng and Mevis (2015)), which also find that the level of capital is the main factor within bank specific variables.

11. The weakening of the banking sector resulted in the need for significant public support. During the Economic Assistance Program (EAP), public support was used to prop up bank capital, through the set-up of the Bank Solvency Support Facility (BSSF) and private issuance (IMF (2011, 2012)). The public sector injected €7.2 billion into the banking system, mainly through the issuance of contingent convertible debt instruments (€4.3 billion purchased by the BSSF). Injections of private capital significantly altered the ownership structure of two banks, BCP and BPI, and further concentrated ownership and voting rights in both banks. As of end-2012, BPI was effectively controlled by CaixaBank and Angolan investors (19.5 percent for Santoro Finance, as the Brazilian bank Itaú sold its stake), and BCP by Sonangol (19.4 percent for the Angolan state-owned oil company). At the same time, exposures of Portuguese banks to Angola increased significantly, and exposures to Brazil declined markedly.

Portugal: Bank Exposure to Selected Countries

(Billions of U.S. dollars)

Sources: BIS; and IMF staff calculations.

Portugal: Sources of Capital for Large Banks

(Billions of euros)

Sources: Bank of Portugal; and IMF staff calculations.

D. The Post-Crisis (2014–16): More Constraints, Less Room for Maneuver

12. Despite a moderate economic recovery, the operating environment for banks remains challenging. The banking system was unable to restore profitability even as the economy rebounded. Banks took some steps to reduce costs in the context of their restructuring plans, but costs did not go down enough to offset the sharp declines in net interest income from the low interest rate environment. While a few banks were able to aggressively reduce their costs, the remaining large banks were unable to do so. Overall, the banking system has been unable to return to durable profitability.

Portugal: Non-Performing Loans and Provisioning

Source: IMF Financial Soundness Indicators.

13. Legacy issues related to the misallocation of credit continue to hamper banks’ performance following the crisis. The debt of non-financial corporations rose to over 120 percent of GDP during the crisis, and remained at 109 percent of GDP at end-2015 (see IMF (2015)). NPLs continued to rise even as the economy began to recover due to the large number of non-financial corporations that were overly indebted (IMF 2013). Bank capital levels, still among the lowest in the EU, left insufficient buffers for banks to fully absorb losses from mounting provisions and impairments.

Common Equity Tier 1 Capital, end-March 2016

(Percent of Risk-Weighted Assets)

Source: European Banking Authority.

14. High profile bank resolutions during this period also impacted the broader banking system. Two banks were resolved: BES in August 2014 due to large intra-group lending and Banif in December 2015 as the bank had been unable to implement an effective restructuring plan. Both resolutions had a significant fiscal cost of more than 4 percent of GDP (€4.9 billion for BES and €2.2 billion for Banif).3 Moreover contingent liabilities stemming from litigation regarding BES/Novo Banco and the transfer of some of Banif assets to the Resolution Fund could entail further costs in the future for the government but also for the banks (due to their recurring contribution to the Portuguese resolution fund on top of their contributions to the European Resolution Fund).

Return on Equity, end-March 2016

(Percent)

Source: European Banking Authority.

E. The Medium Term: Remaining Challenges

15. In the near future, economic growth, inflation and interest rates are expected to remain low, which will increase the importance of strengthening the banking system to ensure it can support the recovery. Deteriorating asset quality and low capital levels make the Portuguese banking system vulnerable. Therefore, strengthening the banking system will be an important element in ensuring it can support the recovery. While liquidity has been restored, legacy issues related to the misallocation of credit have not been dealt with, resulting in low profitability which makes it difficult to increase capital. Low capital levels in turn hamper the pace of balance sheet clean-up, creating a negative feedback loop. Therefore, banks will need to continue to adjust to the difficult operating environment and make tough decisions to restore profitability by increasing income and reducing expenses.

16. Banks will need to further bolster new and stable income streams. Alternative Sources of income such as fees and commission could be strengthened, building on the universal banking model used by Portuguese banking groups. While the share of income stemming from commission and fees is in line with the euro area, commission and fees as percentage of banking assets have declined since 2012. Existing proposals at the European level, such as the Capital Market Union could be an opportunity for banks to develop further a fee-based business model, which could be supported by securitization markets. This would help reduce capital requirements and improve risk-sharing.

Bank Compensation

(Share of total revenue, percent)

Sources: SNL Financial; and IMF staff estimates.

17. Further reduction in operating costs is also warranted, as the decline in costs has not been able to compensate for a higher level of impairments and dwindling net interest income. Compared to other euro area countries, compensation of bank employees accounts for a large share of revenue, which puts further downward pressure on banks’ profitability. Further restructuring will also be needed, including a rationalization of the branch network and continued divestment of non-core assets and lines of business that are not profitable.

18. Banks will also need additional capital to absorb the losses stemming from changing their business models. Streamlining bank operations around core, profitable income streams and dealing with legacy assets will require significant loss absorption capacity by some banks. In the current environment, banks’ efforts to meaningfully bolster their capital buffers are hampered by the high cost of capital – as both new and existing shareholders are weary of providing additional capital to an unprofitable banking system with unresolved balance sheet weaknesses.

19. The regulatory environment imposes additional constraints on banks. The entry into force of the Bank Recovery and Resolution Directive, the Single Resolution Mechanism and future Minimum Requirements for own funds and Eligible Liabilities (MREL) impose additional requirements on banks. While regulatory requirements are ultimately aimed at strengthening bank balance sheets, in the short-run they will impose costs on banks as they seek to adjust their business models and shore up their capital base. Over the medium-term, however, regulatory requirements to further impair NPLs and raise additional capital would improve asset quality and free resources to lend to more productive firms, especially in the tradable sector.

20. The challenges currently confronted by Portuguese banks arose over the course of decades, and will require decisive action to overcome. Banks will need to move aggressively to restore profitability by identifying alternative stable income streams and lowering costs, while seeking a solution to long-standing legacy issues stemming from the misallocation of credit. Capital levels will need to be strengthened, likely even surpassing regulatory requirements, to absorb losses.

References

Prepared by Antoine Bouveret and Irene Yackovlev.

The positive association between capital and profitability during 2010–15 appears to be different from the pre-crisis period when high profitability was associated with low capital, as elaborated below.

See IMF (2015a) and IMF (2016) for further details regarding the resolution BES and Banif.

Other Resources Citing This Publication