The Basel II Capital Accord and its expected implications on the Romanian banking system

(1)

The Basel II Capital Accord and its expected

implications on the Romanian banking system

Anca Stefanescu s1752448

Supervisor: Prof. Dr. K.H.W. Knot Co-referent: Drs. E. Wester

MSc. Economics

Faculty of Economics and Business University of Groningen

(2)

Abstract

This paper looks from a quantitative point of view at the changes that occurred in the Romanian banking sector after the implementation of the Basel II Accord. In order to achieve this, it presents the expected impact of the Accord on regulatory capital, based on theoretical and research papers. It also gives an overview of the Romanian banking system and the structure of its balance sheet. Lastly, it describes the capital modifications experienced by the whole sector and by a relevant portfolio of banks.

Overall, at system level there was a 5.7% decrease of its capital ratio, a little sharper than the one expected for Group 2 CEBS banks implementing the same approach. Based on the analysis of the selected portfolio, the considerable reduction was caused by two different trends. On one hand, a number of small and medium banks increased their equity in order to mitigate the effects of the new regulations and saw their capital ratios rise. On the other hand, large banks did not take similar actions, nor did they restructure their assets, and experienced increased risk-weighted assets which caused their capital adequacy ratios to be lower than ever before.

(3)

1. Introduction

The publication of the “International Convergence of Capital Measurement and Capital Standards” (Basel II) brought about the much awaited and needed change to prudential supervisory legislation.

Its key objectives, as presented in the Third Consultative Paper, were to better align economic and regulatory capital, while keeping constant the overall capital level in banking systems. In addition to these objectives, Basel II focused on integrating in the new Framework recent innovations in risk management techniques and financial products, set the foundation for qualitative banking supervision while allowing banks a higher degree of self-regulation and increased public disclosure requirements in order to improve market discipline and counterbalance the increased allowance for self-regulation.

The discussions over the new Accord covered a wide range of topics from expected effects on certain types of banks and lending to the variables in the formulas used to determine regulatory capital and to the improvements and novel elements it introduced compared to the previous version. But the most important issue was the actual expected effect of Basel II on regulatory capital and risk-weighted assets, at national level. This was covered in quantitative impact studies and public consultations carried out by the Basel Committee on Banking Supervision as well as in numerous research papers and Central Banks’ analyses.

For the Romanian banking sector, the Central Bank has taken the necessary steps for the implementation of Basel II, without publishing a report on the quantitative and qualitative changes the system will undergo as a result of the change from Basel I to Basel II. Neither was there any paper covering the quantitative effects of Basel II in Romania written by scholars. The NBR did organize a COREP exercise in September 2007 in which banks were supposed to determine their capital ratios using Basel II, but just eight banks responded to it so no aggregated indicator for the system was made available to the public.

(5)

individual bank’s level. At the end of 2007, 16 out of 29 banks published their capital ratios (Basel I). From these banks, BCR Erste also calculated it according to Basel II. At March 2008, BRD Groupe Société Générale was the only bank out of those 16 to publish its capital adequacy ratio (Basel II) and for another three banks I got a hold of the needed indicators in strict confidentiality.

When looking at the qualitative modifications Basel II facilitated in the Romanian sector, according to the specialist I’ve interviewed as part of data collecting, “the spotlight finally fell on risk management” and the crucial role played by it. In all five banks I’ve analyzed, centralized loans databases were set up, risk management activities were centralized and risk management committees were either put into place or restructured. In addition to changes in managing credit and market risk, the entire infrastructure needed to deal with operational risk was implemented.

The remainder of the paper is organized as follows. Section 2 presents the general expected impact of the Accord on banks’ capital structure and focuses on credit and operational risk because of the insignificant importance of market risk on Romanian banks’ balance sheets. Section 3 discusses the structure of the Romanian banking system. Section 4 details the effects of the Accord on the banking system and in particular on five banks. Lastly, section 5 concludes the study.

2. Expected impact of the Accord on banks’ capital structure

2.1. The Basel II framework

In order to address the weaknesses of the previous capital framework, the new Accord moved away from the simplistic view on risk weights, assets and risk management techniques of Basel I and made efforts to reduce the scope for regulatory arbitrage.

(6)

The Basel II framework, based on three complementary pillars, is much more complex than its predecessor’s because of the need for a more flexible capital adequacy setting and the importance given by the Basel Committee to promoting a healthy and efficient banking sector, not just a well-capitalized one.

Pillar 1, the main focus of this paper, covers the minimum capital requirements and the rules governing the BIS ratio, computed as regulatory capital to risk-weighted assets.

The new Accord leaves unchanged the minimum value for the capital adequacy ratios (4% for tier 1 and 8% for total capital) and the definition of capital. Regulatory capital is divided into tier 1 (equity capital and disclosed reserves) and tier 2, which consists of supplementary capital (e.g. undisclosed reserves and hybrid debt capital instruments), limited to the amount of tier 1.

The considerable changes introduced by Basel II regard mostly the denominator, the risk-weighted assets, and are presented in detail in the remaining of this chapter. The chapter focuses only on the capital charge attracted by credit and operational risk because of the negligible importance market risk has on Romanian banks’ balance sheets.

Pillar 2 outlines the supervisory review process of banks’ internal capital assessment models. While the new Accord encourages the use of internal-ratings based models and increased self-regulation on the banks’ part, it also allows supervisory authorities to evaluate banks’ IRB models, to demand additional capital to be held for risks not covered by Pillar 1 and to intervene at an early stage if there are any threats of capital levels falling below the required amount consistent with the bank’s risk profile.

Enhanced disclosure of quantitative and qualitative risk information is the core of Pillar 3. By demanding from banks greater transparency and relying on market discipline, the Basel Committee encourages a prudent, well-balanced risk strategy and counteracts Pillar 2’s increasing self-regulation.

2.2. Credit risk

Credit risk, as defined as risk of counterparty failure1, is the most important risk faced by financial institutions. Even if all banks already protected themselves against it

(7)

by implementing risk assessment models and holding a capital buffer, the first Capital Accord published by the Basel Committee on Banking Supervision in 1988 was the one to put forward an international standard for setting bank capital requirements.

It divided banks’ on-balance sheet assets into four, general risk buckets2 and it set the minimum total capital requirement at eight percent of total risk-weighted assets. By doing so, it attempted to align the regulatory capital charge to the economic capital held by banks.

Although it was recognized as a very important innovation in banking regulations, Basel I had its flaws, among others the simplistic treatment of credit risk, not taking into account risk-mitigating techniques and newly developed instruments, covering through the required capital charge only credit and market risks. The Basel II Accord has been designed to mend these flaws and allow banks to use their internal risk management models in oreder to better align the regulatory capital requirements with the actual incurred risks.

For credit risk, Basel II encourages banks to improve their internal risk measuring systems by providing a standardized approach and two internal model approaches. The standardized approach is similar to the First Accord, but with an increased risk sensitivity. The internal rating based approaches rely on the banks’ risk modeling techniques, risk parameters and the input from supervisory authority to set aside enough capital to cover unexpected loss “with a probability of 0.999 over one year”3.

2.2.1. Sovereign bond debt Basel I

Regarding sovereign claims, Basel I divided countries into OECD and non-OECD countries. The countries in the first category were given a 0% risk weight for their sovereign debt, with the ones in second category received 100% risk weight.

Basel II

The standardized approach

2_{Basel Committee (1988), Annex 2}

(8)

In the standardized approach for measuring credit risk, risk weights for sovereign debt are more detailed, based on credit ratings provided by external credit rating agencies like Standard & Poor’s, Fitch or Moody’s, as shown in table 1.

Table 1. Sovereign creditworthiness risk weights Credit

assessment AAA to AA- A+ to A- BBB+ to BBB- BB+ to BB- B+ to B- Below B- Unrated

Risk weight 0% 20% 50% 100% 100% 150% 100%

Source: BIS, International Convergence of Capital Measurement and Capital Standards, June 2004

If a bank uses two rating agencies that supply different credit ratings for a sovereign claim, the estimation that translates into a higher risk weight is used. In case a bank uses three rating agencies with different assessments, the higher risk weight will be chosen from the two lowest ones (see table 4).

Another alternative for determining sovereign bonds’ risk weight would be the approval of Export Credit Agencies’ ratings by the national supervisory authority. To be recognized, these agencies have to use the methodology imposed by the OECD and classify countries’ credit risk into 8 categories with corresponding minimum premium rates as presented below.

Table 2. Corresponding risk weights to the OECD determined risk scores

According to the quantitative impact study 5 (QIS 5), the abolition of the distinction between OECD and non-OECD countries causes a slight change in the capital charge for sovereign bond debt across most banks as shown in the table below.

Table 3. Variation of capital charge using the standardized approach

G10 CEBS Other non-G10

Banks

Group 1 Group 2 Group 1 Group 2 Group 1 Group 2 Capital charge variation (%) 0.2 -0.1 0.4 0.1 0.1 14.3

Source: CEBS (2006)

For most groups of banks, the variation from 0% risk weight for all OECD sovereign bonds to 20%, 50% or even 100% for the national debt of OECD countries like Greece, Mexico or Turkey, as shown in table 4, will translate in a small variation of the

ECA risk scores 0 or 1 2 3 4 to 6 7

(9)

capital charge, because of the reduced weight of sovereign debt in banks’ assets – less than 1% for the analyzed banks. The only notable exception is the increase of almost 15% of the minimum capital charge for other non-G10 Group 2 banks, caused most probably by the banks’ portfolio structure.

Table 4. Sovereign debt capital charge under Basel I and Basel II Country

Basel I capital charge (%)

Standard

& Poor’s Fitch Moody’s

Overall rating

Basel II capital charge (%)

Australia 0 AAA AA+ Aaa AAA 0

Austria 0 AAA AAA Aaa AAA 0

Belgium 0 AA+ AA+ Aa1 AA+ 0

Canada 0 AAA AAA Aaa AAA 0

The Czech Republic 0 A A+ A1 A+ 20

Denmark 0 AAA AAA Aaa AAA 0

Finland 0 AAA AAA Aaa AAA 0

France 0 AAA AAA Aaa AAA 0

Germany 0 AAA AAA Aaa AAA 0

Greece 0 A- A A1 A 20

Hungary 0 BBB BBB A3 BBB 50

Iceland 0 BBB- BBB- Baa1 BBB- 50

Ireland 0 AAA AAA Aaa AAA 0

Italy 0 A+ AA- Aa2 AA- 0

Japan 0 AA AA Aaa AA 0

Korea 0 A A+ A2 A 20

Luxembourg 0 AAA AAA Aaa AAA 0

Mexico 0 BBB+ BBB+ Baa1 BBB+ 50

The Netherlands 0 AAA AAA Aaa AAA 0

New Zeeland 0 AA+ AA+ Aaa AA+ 0

Norway 0 AAA AAA Aaa AAA 0

Poland 0 A- A- A2 A- 20

Portugal 0 AA- AA Aa2 AA 0

Slovakia 0 A+ A+ A1 A+ 20

Spain 0 AA+ AAA Aaa AAA 0

Sweden 0 AAA AAA Aaa AAA 0

Switzerland 0 AAA AAA Aaa AAA 0

Turkey 0 BB- BB- Ba3 BB- 100

Source: the rating agencies’ websites

According to a report4 based on 2004 credit ratings, the main winners will be countries with lower risk weights (i.e. China, Chile, South Africa, Hong Kong, Thailand

(10)

and Malaysia) and the following countries will face an increased capital charge and more difficult access to credit markets: Iceland, Turkey, Poland, Greece, Hungary, Slovakia, the Czech Republic, South Korea and Mexico. Another report5 claims that Singapore, Cyprus, Malta and South Africa will benefit from lower risk weights and confirms the findings of the Morgan Stanley report regarding Poland, Hungary and Turkey. Table 5 presents the findings.

In these analyses, the external ratings for foreign currency sovereign claims were used because it was necessary to level the evaluation field due to a clause in the new Accord that states: “a lower risk weight may be applied to banks’ exposures to their sovereign of incorporation denominated in domestic currency and funded in that currency. Where this discretion is exercised, other national supervisory authorities may also permit their banks to apply the same risk weight to domestic currency exposures to this sovereign funded in that currency.”6

Table 5. Basel II’s impact on sovereign risk weighting

Basel I Basel II

Standard approach External

rating Zone A Zone B

Option 1 Option 2

AAA USA Singapore 0%

AA Belgium Cyprus 0%

A Poland Malta 20%

BBB Hungary South Africa 50%

BB Turkey Romania 100%

B N/A Argentina 100%

Under B N/A Ecuador 150%

Unrated N/A

0%

Tanzania

100%

Source: Nomura, Basel II and banks, September 2005; updated values

Internal ratings-based approach

Using the FIRB approach, banks determine the PDs for sovereign exposures by using their internal risk-scoring models. The rest of the variables needed to determine the risk weight for sovereign claims - LGD, EAD and maturity - are given by the supervisory authority. Senior unprotected sovereign claims are allocated a 45% LGD, subordinate claims on sovereign have a 75% LGD and maturity is set at 2.5 years.

5_{Nomura (2005)}

(11)

In the AIRB approach, banks use their internally estimated variables (PD, LGD and EAD) and maturity. The value used for maturity cannot be greater than 5 years. It is expected7 that banks’ internally determined LGDs will be lower than the 45% value assigned in the FIRB approach, causing a bigger reduction in capital charge required for the same claim.

As most investor-grade sovereign debt has a low default history, the capital charge assigned to it is expected to be very low. But as mentioned before, the QIS 5 report shows that switching from a 0% (100%) risk weight for (non-)OECD countries to internally determined ones causes a percentage increase in the capital charge across every group of banks, as shown in table 6. Compared to the standardized approach, the effects of the shift are more noticeable, although in real terms, the fluctuation is still of relative small impact.

Table 6. Variation of capital charge using the most likely internal ratings based approach

Banks

Group 1 Group 2 Group 1 Group 2 Group 1 Group 2 Capital charge variation (%) 1.3 0.6 0.9 0.5 0.1 0.1

Source: CEBS (2006)

Concerns

A number of researchers8 and central banks9 expressed concerns regarding the omission from the new Accord of the risk reduction effects of international diversification of banks’ portfolios.

In normal times, internationally active banks and banks with a high percentage of below investment grade sovereign claims would benefit form diversification because portfolios that include both developed and developing countries’ sovereign claims would have a lower risk exposure than an undiversified one.

The failure to take into account the portfolio and diversification effects might result in inaccurate measuring of risk, a higher capital charge for sovereign debt of emerging market countries and, indirectly, a higher concentration of lending to less risky

7_{Illing, M. and Paulin, G. (2004)}

(12)

sovereign debt of economically advanced countries leading to potentially higher systemic risk.

In contrast, during turbulent periods, the issues of co-movement10 and contagion arise between national economies, making the risks the public sector is exposed to much more difficult to diversify. This also would mean that the expected benefits of international diversification of portfolios will not be reaped when most needed and that the overall relevance of this effect is justly neglected by the Accord.

2.2.2. Bank claims Basel I

The first Accord set a 20% risk weight to the following claims11: - Claims on banks located in an OECD country;

- Claims guaranteed by OECD incorporated banks;

- Claims on banks incorporated in countries outside the OECD with a residual maturity of up to one year;

- Claims with a residual maturity of up to one year guaranteed by banks incorporated in countries outside the OECD.

For all other bank claims on banks not incorporated in an OECD country, with a maturity longer than one year, the designated risk weight was 100%.

Basel II

Standardized approach

The national supervisors can choose one of the two options available under the standardized approach for all claims to banks incorporated in that country. The first option defines the risk weight for bank debt as being one category less than that of the sovereign bond debt of the country the bank is incorporated in. Also, a 100% risk weight is assigned to claims on banks located in countries with sovereign debt rated between BB+ and B- and to claims on banks in countries with unrated sovereign claims.

The other option, expected to be chosen by most national supervisor agencies, sets the risk weights according to banks’ external ratings. Unrated bank claims have a 50%

(13)

risk weight. In case the claim is short-term (three months or less), either rated or unrated, a risk weight one category more favorable than the one for the bank debt can be applied, but no less than 20%. The only exception consists of short-term claims on banks rated below B- who maintain the 150% risk weight. The two options are presented in the tables below.

Table 7. Standardized approach - option 1 Credit assessment of Sovereign AAA to AA- A+ to A- BBB+ to BBB- BB+ to BB- Below B- Unrated Risk weight 20% 50% 100% 100% 150% 100%

Table 8. Standardized approach - option 2 Credit assessment of Banks AAA to AA- A+ to A- BBB+ to BBB- BB+ to BB- Below B- Unrated Risk weight 20% 50% 50% 100% 150% 50% Risk weight for short-term claims 20% 20% 20% 50% 150% 20%

If the national supervisor chooses to apply a preferential risk weight to banks’ exposure to domestic currency sovereign debt, it can also apply a favorable treatment to short-term bank debt issued in domestic currency. The risk weight assigned to it will be one category less favorable than the sovereign debt, but no less than 20%.

Similar to the treatment of sovereign debt, if a bank has three ratings, it will be assigned the lower risk weight between the two highest ratings.

Under this approach, for either option as shown above in the tables, OECD-incorporated banks will see their risk weights increase noticeably or, in the best case, remain constant. Comparing Basel I with option 1 of Basel II standardized approach, banks from Western Europe keep the 20% risk weight and the following banks show an increase of:

- 50% for banks in the Czech Republic, Greece, Island, Korea, Poland and Slovak Republic;

(14)

If option 2 is compared to Basel I, all OECD-located banks rated below A+ will see the risk charge for their long-term claims increase. Only for above BB+ claims with maturity under three months, option 2 resembles Basel I.

All banks situated in non-OECD countries will have lower or equal risk weights compared to those set by the first Accord except banks in very low rated countries (below B-) like Argentina, Bolivia, Granada, Malawi, Moldova or Paraguay.

As stated before, option 2 seems to be the likely choice of regulators, at least in Europe because of the statements issued by the British regulating agency, FSA, one of the trendsetters in European regulation. In its 2003 consultative paper12 regarding the implementation of the Basel II Accord, FSA decided to use option 2 for its QIS 3 and further consultations because of its higher risk sensitivity, more in line with the ideas of the new Accord. Among other countries that used option 2 in QIS 3 are Spain, the Netherlands and Sweden13.

This option will favor a 20% risk weight for well-rated (above BB+) and unrated short-term debt, a preferential treatment unavailable under the first option, and it will be more advantageous for banks with good rating located in lower-graded countries. On the other hand, option 2 will bring a higher capital charge for smaller banks unrated or rated above A- (e.g. Spanish cajas, German sparkassen, Italian savings banks), with headquarters in the EU (except for those located in Greece, Czech Republic, Hungary, Poland and Slovakia).

Regardless of the option adopted by the supervisory agency, under the standardized approach, exposure to bank claims will attract higher capital requirements for most banks. Generally, the effects in the total capital charge will be reduced due to the small percentage of bank debt in overall bank actives, around 5%. As shown in table 9, the QIS 5 found a small increase for all Group 1 banks and for Group 2 banks except Group 2 CEBS banks. The small, European banks show reduced variations due to their portfolio’s structure being less oriented towards interbank lending.

12_{FSA (2003)}

(15)

Table 9. Variation of capital charge using the standardized approach

Banks

Group 1 Group 2 Group 1 Group 2 Group 1 Group 2 Capital charge variation (%) 1.5 0.2 1.8 -0.7 1.2 2.3

Source: CEBS (2006)

Internal based ratings

The capital requirement for bank claims is determined similarly to that for sovereign debt. Under the FIRB approach, all inputs except PD are given by the supervision authority: LGD is set at 45% for senior unsecured claims and at 75% for subordinated unsecured claims and maturity is given as 2.5 years Banks will use their internal models to determine the debt’s one year PD subject to a 0.03% minimum. Under the AIRB approach, all variables used and the maturity are determined internally.

Because many banks implementing the Basel II Accord are expected to choose one of the two approaches based on internal ratings as they will attract a lower capital charge than the standardized alternative (as seen in table 10), the expected impact of the new Accord is diminished and the risk weight attracted by bank claims will be reduced. Table 10. Variation of capital charge using the most likely IRB approach

Banks

Group 1 Group 2 Group 1 Group 2 Group 1 Group 2 Capital charge variation (%) 0.4 0.1 -0.2 0 0.2 -0.9

Source: CEBS (2006)

A report14 using QIS 5 data for European banks finds that, under AIRB, average LGD is 37.7% for Group 1 banks and 39.4% for Group 2 banks, both considerably lower than the assigned LGD under FIRB.

Another report15 done five years ago employs for its analyses “the industry standard LGD of 60% for senior debt” and 75% for subordinated debt. It expects that, in both cases, the banks will use much lower figures and suggests that a more accurate LGD for western European banks would be around 10% although, when taking into account the recent financial crisis and how defaults appear to cluster, this figure might seem optimistic. It also anticipates a decrease in the risk weights corresponding to subordinated

14_{CEBS (2006)}

(16)

debt from 100% under Basel I to around 24% under FIRB although for non-OECD banks with lower ratings subordinated debt will attract an higher capital charge.

Concerns

Representatives of the banking industry16 and theoreticians17 have voiced their worries over the possible negative effects the new Accord might have on interbank lending for banks located in non-OECD and emerging countries.

The Accord states that the capital charge for banks claims issued by unrated banks located in developing countries cannot be lower than “the weighting of the sovereign in which they are incorporated”18. Most of the concerned sovereign are low-rated and will attract a 100% or even a 150% risk weight. Although all internationally active banks will use one of the two IRB approaches, local banks in developing countries will probably adopt the standardized approach, at least at the beginning. They will face an increased capital charge for their claims and will have low ratings which might discourage borrowers to get rated and lenders to finance interbank activity.

There are still discussions over the effects of the increased capital charge for loans to lower rated banks. On one hand, it has been argued by Griffith-Jones and Pratt (2001) among others that G10 banks are likely to reduce the quantity of lending to banks located in emerging countries and cause a supply shortage. On the other hand, Saporta et al (2002) consider that there will not be any negative effects on interbank lending because economic capital currently exceeds and will exceed the amount of required regulatory capital determined by the new standards.

2.2.3. Corporate exposure - wholesale and small and medium enterprises Basel I

Under the first Accord, all claims on small and medium enterprises and corporations were assigned a flat risk weight of 100%.

Basel II

16_{Reserve Bank of India (2003), China Banking Regulatory Commission (2003), Asian Bankers}

Association (2003)

(17)

Corporate exposures are defined as “debt obligations of a corporation, partnership, or proprietorship”19. These exposures are divided into three categories:

- Claims on large corporations with sales in excess of €50 million;

- Claims larger than €1 million on small and medium enterprises (SME) identified as corporate debt of a firm with sales between €5 million and €50 million;

- Specialized lending composed of 5 sub-classes: project finance (PF), object finance (OF), commodities finance (CF), income-producing real estate (IPRE) and high-volatility commercial real estate (HVCRE).

IPRE refers to a method of funding for real estate20 like multifamily buildings where the repayment of the loan depends on the rental payments made by the tenants or on the sale of the building. HVCRE lending is the type of lending that could experience higher loss rate volatility due to certain factors, compared to other kinds of specialized lending. Because HVCRE loans are thought to default simultaneously, they will require a higher capital charge, computed using a substitute asset correlation formula.

According to this approach, corporate exposures are given a risk weight that varies from 0% to 150% depending on their external rating or lack of, as seen in the table below. At national discretion, banks are allowed to assign a 100% risk weight to all their corporate claims, disregarding external ratings. To prevent abuse of this option, banks have to ask for approval for their choice from the supervisory authority before using it. Table 11. Sovereign creditworthiness risk weights

Credit assessment AAA to AA- A+ to A- BBB+ to BB- Below BB- Unrated

Risk weight 0% 20% 100% 150% 100%

As it can be seen, for unrated corporations, corporations rated between BBB+ and BB- and SME which are expected to be mostly in the unrated bucket, there will not be

(18)

any change from the first Accord. Only well-rated firms will benefit from lower capital where as those rated below BB- will see their risk weight rise to 150%.

The QIS 5 notes that exposures to corporations average at around 20% of bank assets for both groups of banks, with a notable increase to more than 35% for other non-G10, Group 2 banks. SME corporate lending covers around 10% of banks’ portfolios across the analyzed countries. Despite the high proportion of loans to corporations and SME with sales between €5million and €50million, the contribution to the overall capital requirements is modest as confirmed by table 12, due to minor reductions across Group 2 banks and small increases for Group 1 banks, except for those located in CEBS countries.

Specialized lending accounts for around 5% of the portfolios of Group 1 banks located in G10 and CEBS countries and has a small percentage in the portfolios of the other analyzed banks, resulting in a general negligible effect on the capital requirements. Table 12. Variation of capital charge for banks using the standardized approach

G10 banks CEBS banks Other non-G10 banks Portfolio

Group 1 Group 2 Group 1 Group 2 Group 1 Group 2

Corporate loans 0.9 -1 -0.3 -0.6 0.4 -0.2

SME loans (to firms with sales

between €5m and €50m ) -0.2 -0.1 -0.4 0.2 0 -0.1

Specialized lending -0.3 0.1 -0.4 0 0 0.1

Total effect 0.4 -1 -1.1 -0.4 0.4 -0.2

Source: CEBS (2006)

Under the FIRB approach, banks determine internally the PD for corporate and SME loans, subject to a minimum of 0.03%, and retrieve the rest of the variables from the supervisory authority: LGD (set at 45% for senior debt), EAD and maturity (usually set at 2.5 years). Banks using the AIRB approach will calculate their own variables and effective maturity based on their credit risk measuring models.

(19)

IRB approaches except for high-volatility commercial real estate lending (HVCRE), which can only be calculated using a special risk-weight function.

For the loans to SME, “a firm-size adjustment (i.e.

(

)

45 5 1 * 04 . 0 − S− ) is made to the corporate risk weight formula”21, where S is a proxy for annual sales expressed in million Euros, ranging from €5 million to €50 million (if sales are lower than €5 million, they are rounded at €5 million).

Compared to the single risk weight applied by Basel I to all corporate and SME loans, Basel II reflects better the riskiness of the claims, with the IRB approach being the most risk sensitive, as seen in the figure below. This increased risk sensitivity will alleviate one of the main weaknesses of the first Accord, regulatory capital arbitrage.

Figure 1. Capital requirements for a corporate loan under Basel I and Basel II

Source: Reserve Bank of New Zeeland: Bulletin, vol. 68, no. 3, 2005

This will probably cause low-rated firms to either look for financing from banks using the standardized approach or to issue bonds. The flip coin might be that highly rated corporations will be better off getting funds from sophisticated banks that have implemented either of the IRB systems.

In the before mentioned report22, analysts have estimated the capital requirements of foundation IRB using a model portfolio and Moody’s PDs and found that they will drop considerably as shown in table 13. AA corporate claims’ risk weights fall to just

(20)

14% from Basel I’s 100%, with A and BBB rated debt capital charge dropping to 40%, making bank borrowing more attractive. The main difference from the standardized approach is the lack of an upper limit for the capital required for low-rated company loans.

For the AIRB approach, the roughly estimated capital charge was higher than in the FIRB case, due to the use of a very conservative LGD of 60%. Another study, done by the FDIC23, focused on this approach exclusively and used Standard & Poor’s PDs for a portfolio of commercial and industrial loans. It showed that the capital requirement for this type of portfolio will be directly proportional with the loan’s estimated LGD. It also determined that, for all commercial and industrial exposures rated BBB or better, as well as for exposures rated BB or better with LGD under 50%, capital requirements will decrease.

Similar calculations determined that the preferential treatment of SME exposures advantages the most SME with sales of €5 million and below, as they require a capital charge over 20% lower than that of corporate claims under the IRB approach, as table 13 also illustrates.

Table 13. Risk weights for SME and Corporate loans: Basel II IRB Approach

Ratings RW non-SME RW for €5m turnover % reduction

A- and above 14% 11% 22%

BBB category 40% 32% 21%

BB category 98% 77% 22%

B category 163% 122% 25%

C category 246% 198% 20%

Source: Morgan Stanley, Basel II A-Z, November 2004

On the whole, under either IRB approach, table 14 proves that the capital requirements for corporate, SME and specialized lending will decrease considerably across all bank groups compared to the standardized approach. Except for other non-G10, Group 2 banks, all banks will face lower capital charges for SME loans and a negative overall effect on the required capital of around 7%. This exception can be explained by the structure of those banks’ portfolios, which might include a large share of low-rated or unrated corporate and SME claims. SL clearly has a small, negative effect on the capital

(21)

charge of big, international banks and a reduced, positive one on the smaller, specialized banks.

Table 14. Variation of capital charge using the most likely IRB approach

Corporate loans -5 -4.5 -4 -3.6 -4.3 -4.1

SME loans (to firms with sales between €5m and €50m ) -1.3 -2.2 -1.3 -2.4 -3.3 2.5 Specialized lending -0.4 0.2 -0.7 0.2 -0.4 0.3 Total effect -6.7 -6.5 -6 -6.8 -8 -1.3 Source: CEBS (2006) Concerns

The main concern voiced24 regards the possible negative effects and the overall risk reallocation in the banking systems that might arise from the competition between large, internationally active banks and smaller banks.

Research25 has shown that large banks will adopt one of the IRB approaches and decrease risk-taking by moving towards a portfolio of well-rated corporate loans. Similarly, small, unsophisticated banks will tend to use the standardized approach and lend to high-risk borrowers. This will cause them to take more risks, especially in a highly competitive environment and possibly increase the overall risk of the banking system, given certain conditions.

Another problem might be the rating of debtors by local rating agencies, which might be biased towards them, rating them higher than an international rating agency would. If domestic banks will use these ratings, they might underestimate the riskiness of their portfolio of corporate debt.

Comments made by Kredittilsyne, the Norges Bank and the European Association of Craft, Small and Medium-sized Enterprises on Consultative Paper 3 point out the generic description of SL and its sub-groups, asking for a more detailed presentation, especially for income-producing real estate and high-volatility commercial real estate, and a clearer distinction between corporate lending and specialized lending.

24_{Song Shin, H. (2003); Fisher, S. (2002); Claessens, S. et al. (2008)}

(22)

2.2.4. Retail exposure Basel I

Basel I treated retail exposure very generally, assigning a 100% risk weight to all claims, except for those secured by a property rented or owned by the client which had received a 50% risk weight.

Basel II

According to the new Accord, an exposure qualifies for retail treatment if one of the following conditions applies to it:

- The exposure is towards individuals or small businesses, with a value of maximum €1 million on a consolidated basis or not more than 0.2% of the overall retail portfolio (securities are not treated as retail exposure);

- Residential mortgage loans irrespective of the size go under the retail umbrella only if the loan is given to the owner of a single or a small number of apartments in a building or complex;

The retail treatment is applied to exposures only if they are part of a pool of exposures managed by the bank on a characteristic manner. At the supervisory discretion, a minimum amount of loans within a pool can be set so that the loans in that pool are treated as retail exposures.

Retail claims are grouped into four categories: residential mortgage credits, commercial mortgage claims, other retail claims and past due loans.

For commercial mortgages, there will not be a change in their treatment, which means the risk weight remains 50%. Residential mortgage claims will be assigned a 35% risk weight, decreased from the one under Basel I.

Other retail claims will be risk-weighted at 75%, also lower than under the first Accord. Nevertheless, for all retail exposures except commercial mortgages, the national authority can use supervisory discretion to increase the risk weight, if considered necessary.

(23)

more than 20% of the outstanding debt and 50%, at national discretion, if the specific provisions are more than 50% of the outstanding debt.

Residential mortgage non-performing loans are risked weighted at 100%, net of specific provisions. The supervisory authority can set a 50% risk weight if the specific provisions are more than 20% of the outstanding debt. Commercial past due mortgages will be weighted at 100%.

The two significant reductions in risk weights coupled with considerably high importance of retail portfolios for all banks, especially for G10 and European Group 2 banks (over 40%) and for their Group 1 counterparts (around 30%), are the main drivers for the capital requirements decrease under the standardized approach. Other non-G10 banks have comparably smaller SME and retail portfolios, which account for a lesser drop in the capital charge, of around 4%, as shown in table 15. Because Group 2 banks located in Europe and G10 countries are smaller and not internationally active, they are more retail-oriented and benefit from bigger declines in capital charges of over 10% compared to Group 1 banks. In all banks except those located in other non-G10 countries, the most dominant reduction comes from retail mortgage portfolios, followed by other retail exposures and SME retail loans.

Table 15. Variation of capital charge for banks using the standardized approach

Retail mortgage -6.3 -6.2 -7.8 -7.2 -4.1 -0.1 QRRE -0.1 -0.3 -0.2 -0.3 0 -0.6 Other retail -0.7 -2.5 -1.0 -3.1 0 -2.5 SME retail -0.4 -1.2 -0.9 -1.7 0 -0.8 Total effect -7.5 -10.2 -9.9 -12.3 -4.1 -4.0 Source: CEBS (2006)

Under the IRB approach, banks are required to segment retail exposures into residential mortgage exposures, qualifying revolving retail exposures (QRRE) and other retail claims.

(24)

borrow and repay the loan in the limits set by the bank, a maximum value for each exposure of €100.000, banks have to prove to the supervisory authority the low volatility of sub-portfolios’ loss rates and save for analysis the data on loss rates so that the supervisor can approve the treatment of the exposures as QRRE.

For retail claims, unlike for most other claims, only the advanced internal rating-based approach is available. Banks separate the claims into the subdivisions previously presented with similar risk characteristics and determine internally for each of them the variables needed to calculate the capital charge: PD (not lower than 0.03%), LGD (minimum 10% for all residential mortgages) and EAD. The risk functions in which the variables are applied have an additional factor, the asset value correlation (AVC), which is used to show “the correlation of losses among the assets within a given asset class”26. There are three distinct risk-weight functions, based on the generic one used for sovereign, bank and corporate claims, for each of the sub-group of retail claims. For residential mortgages, AVC is set at 0.15, for QRRE, AVC is 0.04 and for the other retail loans, this indicator is determined using a formula dependent on PD.

A report27 from 2006 estimates wide variations for risk weights under the IRB approach and presented in the table below. For almost all groups the capital charge is lower under the IRB approach than under Basel I. Comparing the standardized approach with the IRB one, it is interesting to point out that for all sub-groups except prime residential mortgages there can be a decrease just as well as an increase in capital requirements, making it clear that the only sure winners are banks with a high percentage of well rated residential mortgages in their portfolios who implement the IRB approach. Table 16. The comparison between Basel I and Basel II – risk weights for retail loans

Asset Basel I _(%) Basel II Standardized _{approach (%)} _{approach (%)}Basel II IRB

Residential mortgage (prime) 50 35 8 to 25

Residential mortgage (non-prime) 50 35 23 to 49

Revolving retail 100 75 51 to 110

Other retail 100 75 61 to 90

Source: Deloitte Touche Tohmatsu, Basel II – The Securitisation Framework

26_{Lang, W. et al. (2006)}

(25)

Another analysis28 looks at risk weights for retail portfolios and finds that for residential mortgages which have a historically low default rate, a PD of 1% and a LGD of 10% give a 13% risk weight while a PD of 2% and a LGD of 10% translate into a 20% risk weight. The conclusion is similar to that of the Deloitte Touche Tohmatsu report regarding banks with high-quality mortgage lending. QRRE include credit card loans, which have a high LGD so the experts suggest a 4% PD and 80-90% LGD for the credit card portfolio, which gives a pretty high risk weight of around 89% but still lower than under Basel I. For other classes of retail exposures, the risk weights vary dramatically depending on the banks’ portfolio structure and what PDs and LGDs are used.

All banks using the IRB approach except other non-G10, Group 2 banks will see a drastic decrease in their capital requirements, with retail mortgages being the main driver downwards, possibly due to low PD levels in this sector. For Group 2 banks this will be double than that under the standardized approach. For other non-G10, Group 1 banks, the capital requirements are lower with more than 16%, four times bigger than under the standardized approach. For the other Group 1 banks, as table 17 shows, reductions of around 10% are significant, although there is a slight positive effect from revolving credit exposures and SME retail credits which have been assigned relatively high PD and LGD values.

The only banks with a relatively high increase in minimum required capital of 6.6% are small, nationally active and located in non-CEBS or G10 countries. The likely explanations are a high percentage of low-grade retail and commercial mortgages in the retail portfolio and the use of high values for the input variables.

Table 17. Variation of capital charge using the most likely IRB approach

G10 countries CEBS countries Other non-G10 banks Portfolio

(26)

Concerns

While national authorities29 and professional associations30 endorse the separate treatment of retail exposures according to their risk profiles, they have complained about the severe discrepancy between the risk weights set by the standardized approach and those that are determined by using the IRB approach. The difference is seen as big enough to distort pricing of mortgages although the risk of the underlying asset would be the same. Some have asked for close monitoring because of the negative effects the low capital requirements for residential and commercial mortgages might have in case of economic downfall.

Researchers31 and China Banking Regulatory Commission32 have analyzed whether the preferential treatment for SME and retail portfolios is indeed justified. The general opinion is that although retail and SME claims are less sensitive to systemic risk and benefit from the law of large numbers and the high degree of diversification, SME and retail loans are generally riskier than unrated corporate claims and shouldn’t require less regulatory capital. Also a point was made about the fact that even though most of SME and retail loans are secured, the collection of default debt from SME and individuals is more difficult than from large corporations.

2.3. Operational risk

The recent changes in the banking activity due to increasing globalization of business through mergers, consolidated activity and outsourcing, financial innovations like derivatives and asset-backed securities, and technological developments have made operational risk an ever more significant issue to be dealt with. This is why, in its attempt to better align economic and regulatory capital, the Basel II Accord introduced an explicit capital charge for banks’ operational risk. This is defined33 as “the risk of loss resulting from inadequate or failed internal processes, people and systems or from external events”

29_{Kredittilsynet and Norges Bank (2003)}

30_{The World Council of Credit Unions (2003); European Association Of Craft, Small and Medium - sized}

Enterprises (2003); Asian Bankers Association (2003)

(27)

and includes legal risks like fines applied by supervisory authorities or arising from lawsuits but excludes strategic and reputational risks.

Although banks have been addressing operational risk through internal controls and auditing, workforce training, business process reengineering and managerial actions, the new Accord presents a detailed classification of loss events and provides a menu of three approaches for determining the capital required, with a varied level of sophistication and risk sensitivity. Banks are encouraged to implement one of the more advanced approaches as soon as they put into operation the necessary infrastructure, gather the required data and meet the pre-set qualifying criteria.

Basic indicator approach

This is the least sophisticated approach, based on the underlying principle that operational risk is directly proportional to the size of the bank, average gross income acting as a proxy for it. There are no specific requirements for banks who want to implement the BIA, but it seems clear the Basel Committee thought this simple and easy to execute approach for smaller, nationally active banks with less developed internal operational risk assessment procedures.

The capital charge is determined by using the following formula, where α is set at 15% and gross income (GI) is defined as net interest income plus net non-interest income. GI is averaged over the previous three years and any negative or null values for GI are excluded from the denominator.

KBIA = 3 * 3 1

∑

GI α

The standardized approach

This is a more refined approach than the BIA that takes into account the fact that different business lines attract different levels of operational risk but still uses a regulatory formula to determine the operational risk capital.

(28)

risk, with a supervisory factor (β) as presented in table 18. The full capital charge results from adding up the average over three years for these individual charges as the formula shows: KTSA = 3 0 ; * max 3 1 8 1

∑

⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ _β GI

A certain function in the formula restricts the effect negative gross income might have on the capital charge because it replaces it with zero.

Table 18. Business units, business lines, indicators and values for beta

Business units Business lines Indicator of operational risk Beta factors

Corporate finance Gross income 18%

Investment

banking Trading and sales Gross income or VaR 18% Retail banking Annual average assets 12% Commercial banking Annual average assets 15% Banking

Payment and settlement Annual settlement throughput 18%

Agency services Gross income 15%

Asset management Total funds under management 12% Others

Retail brokerage Gross income 12%

Source: BIS, International Convergence of Capital Measurement and Capital Standards, June 2004; Deloitte Touche Tohmatshu, Basel Capital Check 4, 2002

Because of the varied beta factors, it seems apparent that banks oriented toward retail activities and asset management will clearly be advantaged when using TSA instead of the BIA. Investment banks will be more inclined towards adopting the BIA, if not the most advanced approach, whereas non-internationally active, traditional banks will have to decide based on the composition of their asset portfolios.

All banks applying for TSA have to meet a series of minimum requirements regarding its board of directors, operational risk management team and system, and amount of resource available, with internationally active banks having even tougher standards to meet.

(29)

these lines, the indicator used is total outstanding loans and advances and the capital charge is determined by multiplying beta with the indicator and 0.035.

Advanced measurement approach

Under the AMA, the capital requirement for operational risk will be determined based on the banks’ internal operational risk measurement systems, which have to be approved by the supervisory authority. Also, similar to applying TSA, banks have to comply with a long list of general, quantitative and qualitative standards. They will have to compute both expected and unexpected losses, based on internal data and supervisory inputs, do scenario analyses and evaluate factors related to their business environment and internal controls. Data on internal loss has to be collected for at least five years but a three-year period is accepted when a bank implements the AMA for the first time.

The capital charge will be determined on the basis of business lines, similar with TSA. For each business line, operational risk will be divided into seven loss types: internal fraud; external fraud; employment practices and workplace safety; clients, products and business practices; damage to physical assets; business disruption and system failures; and execution, delivery and process management.

A supervisory assigned value for the exposure indicator will be given to all business line/loss type combinations but banks can provide their own data on the exposure indicator. The same applies for the probability of loss event and loss given event. Expected loss is the product of these three variables. To determine the capital requirement, a scaling factor γ, determined by the supervisor for each business line/loss type combination, is applied to the expected loss and the overall capital charge is the sum of all adjusted expected losses.

Banks will have to calculate the capital requirement for both expected and unexpected loss unless they can prove to the supervisory authority they measured and took into account expected loss with a 99.9% confidence interval over a one-year time horizon, in which case the capital charge will apply only to unexpected loss.

(30)

significant percentage of the overall activity and the bank presented a practical timetable for adopting the AMA for all relevant lines of activity.

In line with the practices of the industry, the new Accord recognizes the importance of insurance and its effect on risk mitigating. Banks using the AMA can reduce up to 20% their capital requirement for operational risk, if the insurance issuer and the insurance policy meet conditions stipulated by the Accord concerning their quality. The operational risk charge

As the QIS 5 data collected in table 19 shows, operational risk will bring about the highest increase in required capital across all analyzed banks. In addition, there was considerable dispersion among the estimated capital charges because of different portfolio structures among analyzed banks.

Comparing the capital charges the BIA and TSA might attract, there is a clear decrease for banks using TSA, due to the 12% beta factor assigned to retail banking and asset-management operations.

Another notable point is that the AMA does not attract a lower capital charge than the other two approaches, for any Group 1 bank. Although most Group 1 banks are expected to be implementing this approach, it seems that the required investments into more sophisticated risk management systems do not translate into a lower capital charge. But it might also be possible that the calculations made by banks implementing the AMA are still incomplete and that diversification and insurance effects haven’t been integrated in the actual capital charge.

CEBS located, Group 2 banks are only nationally active, small banks who choose to implement the AMA, which resulted in a 5.4% increase in capital requirements, lower than that following the application of the BIA or TSA.

Table 19. Variation of operational risk capital charge for banks (%)

Source: CEBS (2006)

G10 banks CEBS banks Other non-G10 banks Approach

Basic indicator approach 6.3 8.3 - 8.9 - 13.5

The standardized approach 5.7 7.6 5.5 7.9 4.0 5.2

(31)

Fitch Ratings published a report34 in 2004, after surveying the progress made by 50 of the world’s largest banks and financial institutions in implementing operational risk management. Its findings confirm those of the QIS 5: the capital charge calculated under the AMA (without reflecting the diversification effects and home-host principles) is not necessarily lower than under the BIA or TSA. The survey stated that worries still remain about the recognition of diversification benefits and insurance and the scarcity of internal data needed for a proper loss distribution function.

Concerns

The BIA and TSA, presented above, seem to be enforcing a capital charge that looks more like a tax on banks’ profit rather than a charge on actually computed operational risk. Many35 have argued that gross income is not a good indicator of the level of operational risk a bank is facing. In making this correlation, the Basel II Accord disadvantages small banks and those with high profitability, could give way to regulatory arbitrage and discourages the implementation of better risk-assessing and mitigating techniques. To offset part of these effects, the Basel Committee approved the use, at national discretion, of the ASA, decision received well by some critics36.

Most of the comments made in response to the Committee’s CP 337 with respect to operational risk asked for recalibrating the values for α and β. Other commentators38 requested a subsequent re-calibration of the variables, to reflect changes in risk profiles due to actual loss experience from the newly collected data on expected and unexpected loss.

3. The structure of the Romanian banking system

3.1. General presentation

The evolution of this banking structure in the last twenty years has been marked by privatizations and reforms which transformed the ex-communist single-bank system

34_{Fitch Ratings (2004b)}

35_{Credit Suisse (2003); Fitch Ratings (2004a); Morgan Stanley (2004); China Banking Regulatory}

Commission (2003); America’s Community Bankers (2003); HM Treasury (2004)

36_{Brazilian Federation of Banks (2003)}

37_{The World Council of Credit Unions (2003); Asian Bankers Association (2003); The Reserve Bank of}

India (2003); Bundesverband Investment and Asset Management (2003)

(32)

into a competitive one. The most important reform measures were taken after 1999 and dealt with the problem of high percentage of non-performing loans in banks’ portfolios, very low level of intermediation and high degree of concentration. Although it was a delayed process, it was similar to that in other Balkan states39: liquidation of insolvent banks, privatization of profitable ones, entry of foreign banks, write-offs of toxic loans and effective supervisory actions from the Central Bank.

The lengthily process resulted in a healthy bank sector with assets amounting last year to 61.5% of GDP, holding a dominant position in the Romanian financial system, as shown in the table below.

Table 20. Evolution of financial assets as percentage in GDP

Financial intermediaries 2003 2004 2005 2006 2007

Banks 30.8 36.6 44.6 50.6 61.5

Insurance firms 1.8 1.9 2.2 2.5 3.0

Investment funds 0.l 0.2 0.2 0.3 0.3

Financial investment firms (SIFs) 1.4 1.3 1.8 2.3 2.8

Leasing firms 1.8 3.0 3.6 3.4 5.0

Other non-banking financial institutions 0.4 0.6 0.9 1.3 1.5

Total 36.3 43.6 53.3 60.4 74.1

Source: NBR (2008b)

On the Romanian financial market there are five financial investment firms (SIFs) which represent a distinct type of closed-end investment funds. Their precursors40 (Private Ownership Funds) were started in 1992 to handle a series of management and employees buy outs of state-owned businesses and, later on, a mass privatization program. Since 1999, the SIFs are listed on the stock exchange and are, in essence, collective placement organisms.

Although intermediation is still underdeveloped41 compared to the European Union, it has constantly strengthened over the last years. This trend is also reflected by the significant expansion of local branches which translated into an almost 30% increase for 2007 alone. Additional to the expanding network of branches, the number of bank employees has also increased to over 60.000 and so did their efficiency determined as the ratio of bank assets per employee (see table 21).

39_{Stubos, G. and Tsikripis, I. (2007)}

(33)

Table 21. Evolution of branches, employees and efficiency for Romanian banks

2003 2004 2005 2006 2007

Assets (€ millions) 15,000 23,200 35,400 51,911 72,095

Local branches 3,387 3,031 3,533 4,470 6,340

Employees 46,567 49,702 52,452 58,536 66,039

Employee efficiency (assets/employee in € thousands) 322 466 675 887 1,092

Source: data from ECB (2008)

The Romanian banking system is two-tiered42, with an independent central bank - the National Bank of Romania (NBR) - and 29 registered banks in the second tier (see Appendix 1). In addition, a state-owned central cooperative bank (CREDITCOOP) and two savings and house credit banks43 are active, plus eleven branches of financial institutions from other EU member states. The NBR has also approved 126 foreign banks like Commerzbank, Barclays or Hypo Public Finance Bank to operate in Romania on the basis of the “single European passport”44.

The degree of concentration in the system is moderate and in a slight annual decrease, as reflected by the Herfindahl-Hirschmann index (HHI)45 presented below.

The total value of assets held of the top five largest banks has decreased over time due to the entry of new banks, aggressive strategies used by the other players and mergers and acquisitions between small and medium banks.

Last year was an important year for the banking sector, echoed by the 12% decrease of the HHI: Millennium Bank, a subsidiary of the Portuguese group Millennium BCP, entered the market in a strong manner, opening 39 offices in 8 cities in the first month; Fortis Bank, Finicredito, Depha Bank and La Caixa opened local branches; Blom Bank closed the Egyptian subsidiary’s local branch and moved the activity to the French subsidiary’s branch, taking advantage of the fact that the subsidiary will belong to a credit institution from another member state; Banca Transilvania branched out, opening a subsidiary in Cyprus; and UniCredit România and HVB Ţiriac Bank merged and formed UniCredit Ţiriac Bank, the sixth largest bank.

42_{Lapteacru, I. (2006)}

43_{These banks offer bauspar contracts: a client first saves at least 40% of the needed sum at that bank, after}

which the bank offers a loan for the remaining amount

44_{NBR (2008b)}

45_{According to the Federal Trade Commission, “markets may be “unconcentrated” (HHI below 1000),}

(34)

In 2008, all but two banks (CEC Bank and Eximbank) are privately owned, reflecting the ongoing privatization process. For CEC Bank, a medium term strategy46 was developed to boost its market share, upgrade its network of braches and ATMs, improve its risk management framework and privatize it in 2011.

From the privately-owned banks, three are domestically owned (Banca Transilvania, Banca Comercială Carpatica and Libra Bank) and 37 have foreign capital, out of which 11 are branches of financial institutions located in the EU (as presented in table 3). Analyzing at the end of last year the countries from where the bank capital originated, the top three are Austria with 22% of total capital, Greece with 21,7% and the Netherlands with 7,7%47.

Table 22. Structural indicators for the Romanian banking system

2003 2004 2005 2006 2007 2008

Number banks (including CREDITCOOP) 39 40 40 39 42 43

Privately owned banks 36 38 38 37 40 41

Foreign owned banks 29 30 30 33 36 37

- of which branches 8 7 6 7 10 11

% in total assets of privately owned banks 62.5 93.1 94.0 94.5 94.5 n/a % in total assets of foreign owned banks 58.2 62.1 62.2 88.6 87.8 n/a % in total assets of top 5 banks 63.9 59.2 58.8 60 56.3 n/a

HHI 1264 1120 1124 1171 1040 n/a

Source: NBR (2008b)

3.2. Assets

As shown before, the level of assets held in the banking sector has grown steadily, at a great pace, with annual growth rates exceeding 40%. After Romania joined the EU in 2007, reports from the World Bank48 and the European Central Bank49 confirm that competition has intensified. This process was similar to that experienced in the other new member states50 and put downward pressure on the concentration indicator (the HHI), as shown before. 46_{Fitch Ratings (2007)} 47_{NBR (2007)} 48_{IMF (2008a)} 49_{ECB (2008)}

50_{Bulgaria, Cyprus, the Czech Republic, Estonia, Hungary, Lithuania, Latvia, Malta, Poland, Slovenia and}

(35)

Some small and medium size banks, through expansionary and somewhat unorthodox crediting policies, have doubled or tripled their market share and, in the process, reduced the assets of the top three banks (Banca Comercială Română, BRD Groupe Société Générale and Raiffeisen Bank) with around 5% and limited their growth.

In 2008, the effects of the credit crunch are still relatively unfelt in the Romanian banking sector as a 100% increase in total banks’ net profit is expected. After the second quarter, total assets in the system were evaluated at €76,5 billion, a 7.5% increase in real terms from the same period last year. Banca Comercială Română and BRD Groupe Société Générale, the two biggest banks, account for 36% of total assets, again in a downwards slope of 3.3% since last year due to aggressive strategies of small banks.

The structure of bank assets, presented in table 23, shows that internal assets still hold the dominant share, over 90%, of total assets, mostly due to the high percentage of claims on the non-banking sector. Claims on the governmental sector have increased in importance last year, reflecting the growing need of funding of local administrative branches. Treasury and interbank operations, after peaking in 2005, account for a small segment of the total assets, with the interbank market covering roughly 4% of total domestic assets.

Table 23. Structure of assets held by local banks as percentage in total assets

2001 2002 2003 2004 2005 2006 2007 Internal assets, of which: 85.5 91.7 94.3 94.3 96.5 97.4 98.3 - Claims on NBR and credit

institutions of which:

27.2 32.0 29.3 36.5 40.0 34.9 28.8 + Claims on BNR 23.4 28.6 26.3 28.5 37.5 31.3 24.9 - Claims on the non-banking

(36)

From 2001 onwards, the growth of claims on non-governmental institutions mimicked that of bank assets and was the engine behind their expansion as figure 2 shows.

In 2007, non-governmental credit has shown the same accelerated growth, similar to the other new member states51, with a 50.5% increase in real terms, up to € 41.5 billions, as seen in figure 3. This was triggered by factors in the demand and the supply of credit like high GDP and income per capita growth, relaxed crediting conditions, easy access to external financing and excess of liquidity on the interbank market.

By 2008, credit growth cooled off but remained positive as a direct effect of the decrease in supply and demand. Demand from non-financial companies and the population shrunk as a result of economic conditions and tighter crediting rules imposed by the Central Bank52. The supply contracted on account of more expensive financing available for local banks after Standard & Poor’s and Fitch lowered Romania’s rating to BB+; the lack of liquidity at the level of foreign parent banks for local subsidiaries and branches; and the pessimistic forecast for economic growth. Another important factor in the tightening of credit supply ensued from the complex process of reevaluating their risk profiles banks have undergone as they implemented and complied with the Basel II Accord.

Figure 2. Growth of non-governmental Figure 3. Evolution of non-governmental credit and bank assets as percentage in GDP credit and bank assets

Source: NBR (2008b) Source: NBR (2008b)

51_{ECB (2008)}

52_{Regulation 11/2008, an update to regulation 3/2007, translates into a lower maximum amount that can be}

(37)

Analyzing the structure of these claims from the currency point of view, it is noticeable that, from 2007 onwards, credits in foreign currency have increased faster than credits in Romanian Lei (RON), managing in December 2007 to account for more than 50% of total non-governmental claims. The main reasons for this evolution are: the change in the exchange rate which resulted in a stronger RON; branches and subsidiaries of foreign banks getting relatively easy funding from their parent banks and the NBR abandoning at the end of 2006 the restrictions on the amount of credit in foreign currency banks can lend out53.

As it is expected, the most attractive foreign currency is the Euro, with the Swiss franc making an unexpected entry in 2005 on the mortgage loans market. These loans were tempting both for banks and clients. Foreign-owned banks received the funds relatively cheap from the central headquarters. Clients were attracted because the loans had the lowest interest rate on the market and more relaxed granting criteria. This meant a person could loan more money at a lower interest rate than if he applied for a similar loan in Euros, USD or RON.

In 2008, the spectacular growth of claims in foreign currency was tempered by the international financial situation and by the measures the NBR had taken to curb credit demand and supply. One of these measures was a recommendation followed by seven out of ten banks to stop offering loans in Swiss francs, a currency considered “exotic” and dangerous in times of financial unrest. This slow-down was responsible for the decelerated growth of overall non-governmental credit.

There was also a shift in the loans structure as loans in RON have comprised 50.4% of all loans in July 2008. The reorientation of firms and the public towards credit in the national currency happened despite the increase in interest rates due to high fluctuations for the 1 month ROBOR, especially at the end of 200854.

53_{BNR’s rule 11/2005 stated that banks can extend loans in foreign currency up to three times the amount}

of their bank capital

54_{In the last 12 months, ROBOR hit a minimum of 7.9% on 20.11.2007, a maximum of 51.19% on}

The Basel II Capital Accord and its expected implications on the Romanian banking system