Response to discussion Paper and Call for Evidence on SMEs and the SME Supporting Factor
Go back
The Discussion Paper states that “the application of the SME Supporting Factor has a negligible
impact in terms of reducing the capital ratio on an aggregate level” for the banks included in the EBA’s sample (p. 19). As the Discussion Paper correctly continues to point out, this assessment does not adequately illustrate the impact of the factor on the capital ratio of individual banks. Considering that the EBA sample is based on available data of the largest European banks, the analysis must be considered highly biased.
The factual implementation and the calibration of the factor itself have been based on the concept of neutralizing the increase in capital requirements by the introduction of the capital conservation buffer. The reasoning behind this was to enable banks to provide an adequate flow of credit to SMEs. The necessity to use such measures emphasizes the expectation at that time that the new regulatory requirements, especially the focus on stringent increase in quality and quantity of capi-tal, would potentially further deteriorate the access of SMEs to financing. Although many initia-tives to boost alternative sources of financing for SMEs have been attempted, European SMEs preference for bank financing will continue in ESBG’s view.
Empirical analyses conducted by an ESBG member showed that it is impossible to distinctly iso-late the effect of the SME supporting factor on SME lending due to a vast array of reasons. For example, the temporary nature of the supporting factor calls for decision-makers within banks to conservatively consider its termination and to include this in their calculations as well as, ultimate-ly, in their decision to grant a loan or not. This holds especially true for member states where SME loans are typically long in maturity. Having only recently been introduced with the CRD4 and CRR in January 2014, the supporting factor was part of a major regulatory package with substantial im-plications for banks’ businesses.
The SME scalar has meant that, since its introduction, less capital has been required in relation to these assets. According to one ESBG member, growth has been achieved in part through attractive pricing made possible by the scalar. If it was removed, capital requirements would go up overall which could increase a component of the cost of lending to SMEs. So the SME scalar has indirect-ly helped support the lending to SMEs. Based on this evidence, the supporting factor must be confirmed for a longer period of time which may create further opportunities for good lending growth within this sector at a critical point for growth in Europe.
Finally, the ongoing economic difficulties across the European Union combine with the above factors to make clearly visible effects on SME lending appear highly improbable, let alone any pro-spect of an empirically sound isolation of the supporting factor’s effect.
In some cases, the internal definition of SMEs is somewhat wider than the definition of SME ex-posures subject to the SME supporting factor. In the case of one ESBG member, the internal defi-nition is for example related to both internal risk management practices as well as FINREP report-ing requirements . In accordance with reporting requirements, counterparties that fulfil the EC recommendation 2003/361/EC of 6 May 2003 on definition of micro, small and medium-sized enterprises (SME) are identified to view an enterprise’s ownership connections with other enter-prises. In order to reconcile the internal definition of SMEs with the definition of SMEs subject to the supporting factor, criteria set out in Article 501 of the CRR are used to filter out the subset of SMEs subject to the supporting factor.
As a direct result of the above definition, the amount of Non-Performing Loans does not qualify as an indicator for riskiness as NPLs do not reflect unexpected losses but rather relate to expected losses. The same holds true for accounting data such as the BACH data cited in the Discussion Paper, as these too are merely indicators for expected losses.
Additionally, it is important to note that the interpretation of the proposed risk measure should differ across industries since e.g. the measure of activity (turnover/assets) is naturally low in CRE compared to Retail. Thus, variations in the aggregate measures may only reflect changes in a dom-inant or particular large sector and not variations for a representative SME due to the heterogenei-ty across sectors. In addition, the proposed risk drivers do not have equal weight with regards to their contribution to default probability. Thus, the composite index as depicted in Figure 8 in the discussion paper (EBA/DP/2015/02) is in part misleading and the conveyed message that SMEs in general had a larger decrease in credit quality than larger firms is in ESBG experience largely dependent on regional characteristics and market maturity.
Examining the proposed measures in section 4.3 of the discussion paper, ESBG would like to em-phasise that larger firms compared to SMEs had a larger percentage drop from peek to bottom in default driving aggregate indicators such as coverage and profitability during the global financial crisis. This finding is well aligned with both theoretical empirical studies that conclude that larger firms are more affected by global systemic risk factors since asset correlations increase with the size of the firm. As correctly quoted by the EBA in the discussion paper (pp. 28), lower asset cor-relations for smaller firms should translate into a lower capital requirement. The actual reduction in capital requirements resulting from the SME supporting factor is by far not large enough com-pared to empirical results (for example, see the extensive analysis conducted by Düllmann and Koziol ). This in turn confirms that the SME supporting factor is not only justified but also highly conservative.
Furthermore, the individual creditworthiness assessment on the ESBG member. In some cases, a wide range of detailed balance sheet information is included while some banks apply e.g. conserva-tive stressed interest rates in the assessment of SME borrowers’ creditworthiness. In addition, it can happen that a wide range of supervisory approved internal models is used together with de-tailed balance sheet information and econometric models in the corporate credit risk assessment processes.
Several academic studies have analysed asset correlations of SME exposures, notably the research paper by Düllmann and Koziol (2014) , who found that the risk weights of SMEs are empirically much lower in relation to those of large corporates and that the capital requirements suggested by the Basel framework are significantly overstated.
One ESBG member conducted an extensive analysis based on 8 years of historical credit data covering 1 million SME exposures held by its member banks, the results of which have been shared with the EBA. This analysis confirms the findings of Düllmann and Koziol (2014), suggest-ing an overstatement of SME risks within the capital regime.
ESBG considers it to be difficult to conclude if the SME lending volume has increased or de-creased due to one single factor since there are many factors that are relevant in this equation. From the figures provided by the EBA using data from the “ECB Monetary and Financial Institu-tions Interest Rate Statistics” ESBG can draw the conclusion that the supporting factor may re-cently have had a positive effect in terms of the reduction of the spread between the interest of large companies and small companies. This reduction in interest rates, both in general and, in par-ticular, for SMEs, should have a positive impact on the provision of loans to the real economy in the medium term as soon as the economy starts having a relevant growth rate.
After several years of credit contraction, evidence is supporting a gradual recovery of new loans granted to SMEs in a few ESBG member countries. According to the National Central Bank data of one country (Figure 1), in the first half of 2015 new loans to SMEs rose by 13% compared with the same period of last year, accelerating its pace of growth from 2014. The recovery of SMEs credit started earlier and is being stronger than the recovery of total credit to households and non-financial corporations. In fact, the total flows of new credit decreased in 2014 and did not achieve positive growth rates until 2015, as a result of the more positive performance of credit to larger companies in recent months.
Moreover, Table 1 below illustrates the effect on exposures since the introduction of the SME supporting factor in one other ESBG member. As can be seen, exposures to firms eligible for the supporting factor increased more than other SME lending directly after the introduction in 2014. However, during the first half of 2015, exposures to SMEs not eligible for the supporting factor increased by almost 9 percent while exposures to SMEs subject to the supporting factor increased by a modest 0.7 percent. This reversal in trends can possibly be attributed (i) the roll out of an A-IRB model during the second half of 2014 and (ii) the fact that smaller firms are more dependent on local economic conditions, thus being more severely affected by the resurrection of the Euro-pean sovereign debt crisis during 2015.
As indicated by the growth in exposures presented in Table 1 above, the demand stimulating effect of the supporting factor does not necessarily show up in exposure data. This since demand and supply are affected by e.g. macroeconomic conditions and monetary policy measures making it impossible to disentangle the specific demand and supply side shifts through econometric model-ling, given the short time series currently available. However, the effect on demand due to reduced lending rates can be assessed by the means of estimated demand elasticities together with indica-tive estimates of the price effect attributed the supporting factor. As discussed in the empirical literature, the credit demand elasticity in Europe differs substantially between countries and re-gions, varying from -0.01 to -10.81 . To not exaggerate the effect of the supporting factor, an esti-mate of Eurozone semi-price elasticity of -1.1 is applied. Using a decrease in lending rates by 25 bps due to the introduction of the supporting factor, the supporting factor can be said to have in-creased SME credit by at least 0.28 percent. This result is highly dependent on the assumed semi-elasticity. Using the lower end estimates (ceteris paribus), we find that SMEs subject to the sup-porting factor credit demand may have increased by as much as 2.7 percent in some European countries.
From the analysis of the portfolio of one member, it is shown that the figures from net new busi-ness in this segment (March’14 to June’15) has risen more than 60%, while the segment of SME’s without the supporting factor has shrunk almost a 50%. This evolution, as said before, is the result of several inputs, one of them being the supporting factor.
Q1: Do you have systems in place to track the reduction in capital due to the application of the SME Supporting Factor (capital relief)? Yes/No. Please explain and provide evidence.
Yes, systems have been created to track the reduction in capital due to the application of the SME Supporting Factor. For example, in the case of one ESBG member, the stipulated requisites are implemented in the capital adequacy compilation routines and exposures relevant for the preferen-tial treatment are flagged. Thus, relevant exposures can be singled out and the different outcomes of the calculations, whether applying the SME Supporting Factor or not, can be compared. Risk Exposure is reported both including and excluding the impact of the supporting factor, in accord-ance with current COREP templates.The Discussion Paper states that “the application of the SME Supporting Factor has a negligible
impact in terms of reducing the capital ratio on an aggregate level” for the banks included in the EBA’s sample (p. 19). As the Discussion Paper correctly continues to point out, this assessment does not adequately illustrate the impact of the factor on the capital ratio of individual banks. Considering that the EBA sample is based on available data of the largest European banks, the analysis must be considered highly biased.
Q2: In your experience, is the reduction in capital requirements due to the application of the SME Supporting Factor (capital relief) being used to support lending to SMEs? Yes/No. Please explain and provide evidence.
Yes. Based on ESBG’s experience, the reduction of capital requirements due to the application of the SME Supporting Factor has been used to support lending to SMEs.The factual implementation and the calibration of the factor itself have been based on the concept of neutralizing the increase in capital requirements by the introduction of the capital conservation buffer. The reasoning behind this was to enable banks to provide an adequate flow of credit to SMEs. The necessity to use such measures emphasizes the expectation at that time that the new regulatory requirements, especially the focus on stringent increase in quality and quantity of capi-tal, would potentially further deteriorate the access of SMEs to financing. Although many initia-tives to boost alternative sources of financing for SMEs have been attempted, European SMEs preference for bank financing will continue in ESBG’s view.
Empirical analyses conducted by an ESBG member showed that it is impossible to distinctly iso-late the effect of the SME supporting factor on SME lending due to a vast array of reasons. For example, the temporary nature of the supporting factor calls for decision-makers within banks to conservatively consider its termination and to include this in their calculations as well as, ultimate-ly, in their decision to grant a loan or not. This holds especially true for member states where SME loans are typically long in maturity. Having only recently been introduced with the CRD4 and CRR in January 2014, the supporting factor was part of a major regulatory package with substantial im-plications for banks’ businesses.
The SME scalar has meant that, since its introduction, less capital has been required in relation to these assets. According to one ESBG member, growth has been achieved in part through attractive pricing made possible by the scalar. If it was removed, capital requirements would go up overall which could increase a component of the cost of lending to SMEs. So the SME scalar has indirect-ly helped support the lending to SMEs. Based on this evidence, the supporting factor must be confirmed for a longer period of time which may create further opportunities for good lending growth within this sector at a critical point for growth in Europe.
Finally, the ongoing economic difficulties across the European Union combine with the above factors to make clearly visible effects on SME lending appear highly improbable, let alone any pro-spect of an empirically sound isolation of the supporting factor’s effect.
Q3: Is your internal definition of SMEs in line with the definition of SME exposures subject to the SME Supporting Factor? Yes/No. If no, how are you reconciling the internal definition of SMEs with the definition of SMEs subject to Supporting Factor? Please explain and provide specific examples.
It depends on the ESBG member.In some cases, the internal definition of SMEs is somewhat wider than the definition of SME ex-posures subject to the SME supporting factor. In the case of one ESBG member, the internal defi-nition is for example related to both internal risk management practices as well as FINREP report-ing requirements . In accordance with reporting requirements, counterparties that fulfil the EC recommendation 2003/361/EC of 6 May 2003 on definition of micro, small and medium-sized enterprises (SME) are identified to view an enterprise’s ownership connections with other enter-prises. In order to reconcile the internal definition of SMEs with the definition of SMEs subject to the supporting factor, criteria set out in Article 501 of the CRR are used to filter out the subset of SMEs subject to the supporting factor.
Q4: In monitoring the total amount owed to you, your parent and subsidiary undertakings, including exposures in default, by the borrower and its group of connected clients (as defined in CRR Article 4(1)(39)), what reasonable steps do you take to ensure that amount does not exceed EUR 1.5 million in accordance with Article 501(2)(c)?
ESBG members employ different individual methods to achieve a consolidated view in order to monitor the EUR 1.5 million exposure limit. For example, some banks follow an approach similar to the methodology applied within the large exposures regime.Q5: Do you see merits in having a harmonised definition of SMEs for reporting purposes? Yes/No. Please explain and provide specific examples.
Yes. A more harmonised definition of SMEs for reporting purposes would be welcomed since it would increase comparability across banks and jurisdictions and support supervisors. This in turn could contribute to more aligned supervisory practices across European competent authorities.Q6: Do you agree with the proposed measures of SME riskiness? Yes/No. Are some of these measures more relevant than others? Yes/No.
In order to ensure a meaningful and productive discussion, there needs to be a mutually agreed upon definition of what is meant by SME riskiness. As the SME factor is embedded in the EU’s implementation of the Basel regulatory capital regime, ESBG maintains that SME riskiness should be defined in line with these rules. The capital regime’s purpose is to determine capital require-ments in order to adequately cover unexpected losses. Therefore, SME riskiness should be defined accordingly. In particular, expected losses should be disregarded in this context as they are not within the scope of the regulatory capital regime. Expected losses are not considered a risk within the internal risk management of a bank, but they are rather priced into lending conditions and ade-quately covered by accounting provisions.As a direct result of the above definition, the amount of Non-Performing Loans does not qualify as an indicator for riskiness as NPLs do not reflect unexpected losses but rather relate to expected losses. The same holds true for accounting data such as the BACH data cited in the Discussion Paper, as these too are merely indicators for expected losses.
Additionally, it is important to note that the interpretation of the proposed risk measure should differ across industries since e.g. the measure of activity (turnover/assets) is naturally low in CRE compared to Retail. Thus, variations in the aggregate measures may only reflect changes in a dom-inant or particular large sector and not variations for a representative SME due to the heterogenei-ty across sectors. In addition, the proposed risk drivers do not have equal weight with regards to their contribution to default probability. Thus, the composite index as depicted in Figure 8 in the discussion paper (EBA/DP/2015/02) is in part misleading and the conveyed message that SMEs in general had a larger decrease in credit quality than larger firms is in ESBG experience largely dependent on regional characteristics and market maturity.
Examining the proposed measures in section 4.3 of the discussion paper, ESBG would like to em-phasise that larger firms compared to SMEs had a larger percentage drop from peek to bottom in default driving aggregate indicators such as coverage and profitability during the global financial crisis. This finding is well aligned with both theoretical empirical studies that conclude that larger firms are more affected by global systemic risk factors since asset correlations increase with the size of the firm. As correctly quoted by the EBA in the discussion paper (pp. 28), lower asset cor-relations for smaller firms should translate into a lower capital requirement. The actual reduction in capital requirements resulting from the SME supporting factor is by far not large enough com-pared to empirical results (for example, see the extensive analysis conducted by Düllmann and Koziol ). This in turn confirms that the SME supporting factor is not only justified but also highly conservative.
Q7: Are other aspects relevant in your assessment of the creditworthiness/riskiness of potential SME borrowers? Yes/No. If yes, please provide a list of those aspects and explain how you measure SME riskiness.
First of all, ESBG would like to refer to its answer to question 6 and again stress the importance of focusing on a correct definition of SME riskiness that relates to unexpected losses.Furthermore, the individual creditworthiness assessment on the ESBG member. In some cases, a wide range of detailed balance sheet information is included while some banks apply e.g. conserva-tive stressed interest rates in the assessment of SME borrowers’ creditworthiness. In addition, it can happen that a wide range of supervisory approved internal models is used together with de-tailed balance sheet information and econometric models in the corporate credit risk assessment processes.
Q8: In your experience, are SMEs as cyclical or more/less cyclical than large enterprises?
ESBG experience is that large enterprises are much more affected by cyclical global systemic risk factors than SMEs. One ESBG member points out that SME cyclicality is even considerably lower than currently suggested by the Basel framework. In order to assess the cyclical fluctuation of de-fault rates, asset correlations are being used as a measure in accordance with the framework.Several academic studies have analysed asset correlations of SME exposures, notably the research paper by Düllmann and Koziol (2014) , who found that the risk weights of SMEs are empirically much lower in relation to those of large corporates and that the capital requirements suggested by the Basel framework are significantly overstated.
One ESBG member conducted an extensive analysis based on 8 years of historical credit data covering 1 million SME exposures held by its member banks, the results of which have been shared with the EBA. This analysis confirms the findings of Düllmann and Koziol (2014), suggest-ing an overstatement of SME risks within the capital regime.
Q9: Do you agree with the proposed methodology to assess the own funds requirements in relation to SME riskiness? Yes/No. If no, please provide alternative methodologies or indicators, if available.
The Basel formula for calculating risk weights is highly dependent on the single risk factor which is to be interpreted as the systemic component of credit risk. Thus, if asset correlations are lower for SMEs, their own funds requirement should be lower.Q11: Do you agree with the above interpretation of statistical data on lending trends and conditions? Yes/No. If no, please explain.
ESBG only partially agrees with the interpretation of statistical data on lending trends and condi-tions. More specifically, ESBG supports the idea that demand side factors have mainly contributed to the general lending trends and conditions.Q13: Have changes to your SME credit lending and assessment policies and procedures been driven by other factors (e.g. competition from alternative sources of SME financing as described in section 4.1)? Yes/No. Please explain and provide specific examples.
Yes. The uncertain macroeconomic environment in combination with exceptionally low interest rates has increased the focus on a prudent application processes with conservative stressed interest rates as a major input to the credit granting process.Q14: In your experience, is there an impact of the SME supporting factor on the volume of SME lending compared to other loans? Yes/No. Please explain and provide evidence.
ESBG believes that it is quite early to tell whether the supporting factor is fulfilling its objective. In any case, it seems to be too early to say that the supporting factor may have had a neutral or a negative effect since institutions have had little time to apply it.ESBG considers it to be difficult to conclude if the SME lending volume has increased or de-creased due to one single factor since there are many factors that are relevant in this equation. From the figures provided by the EBA using data from the “ECB Monetary and Financial Institu-tions Interest Rate Statistics” ESBG can draw the conclusion that the supporting factor may re-cently have had a positive effect in terms of the reduction of the spread between the interest of large companies and small companies. This reduction in interest rates, both in general and, in par-ticular, for SMEs, should have a positive impact on the provision of loans to the real economy in the medium term as soon as the economy starts having a relevant growth rate.
After several years of credit contraction, evidence is supporting a gradual recovery of new loans granted to SMEs in a few ESBG member countries. According to the National Central Bank data of one country (Figure 1), in the first half of 2015 new loans to SMEs rose by 13% compared with the same period of last year, accelerating its pace of growth from 2014. The recovery of SMEs credit started earlier and is being stronger than the recovery of total credit to households and non-financial corporations. In fact, the total flows of new credit decreased in 2014 and did not achieve positive growth rates until 2015, as a result of the more positive performance of credit to larger companies in recent months.
Moreover, Table 1 below illustrates the effect on exposures since the introduction of the SME supporting factor in one other ESBG member. As can be seen, exposures to firms eligible for the supporting factor increased more than other SME lending directly after the introduction in 2014. However, during the first half of 2015, exposures to SMEs not eligible for the supporting factor increased by almost 9 percent while exposures to SMEs subject to the supporting factor increased by a modest 0.7 percent. This reversal in trends can possibly be attributed (i) the roll out of an A-IRB model during the second half of 2014 and (ii) the fact that smaller firms are more dependent on local economic conditions, thus being more severely affected by the resurrection of the Euro-pean sovereign debt crisis during 2015.
As indicated by the growth in exposures presented in Table 1 above, the demand stimulating effect of the supporting factor does not necessarily show up in exposure data. This since demand and supply are affected by e.g. macroeconomic conditions and monetary policy measures making it impossible to disentangle the specific demand and supply side shifts through econometric model-ling, given the short time series currently available. However, the effect on demand due to reduced lending rates can be assessed by the means of estimated demand elasticities together with indica-tive estimates of the price effect attributed the supporting factor. As discussed in the empirical literature, the credit demand elasticity in Europe differs substantially between countries and re-gions, varying from -0.01 to -10.81 . To not exaggerate the effect of the supporting factor, an esti-mate of Eurozone semi-price elasticity of -1.1 is applied. Using a decrease in lending rates by 25 bps due to the introduction of the supporting factor, the supporting factor can be said to have in-creased SME credit by at least 0.28 percent. This result is highly dependent on the assumed semi-elasticity. Using the lower end estimates (ceteris paribus), we find that SMEs subject to the sup-porting factor credit demand may have increased by as much as 2.7 percent in some European countries.
From the analysis of the portfolio of one member, it is shown that the figures from net new busi-ness in this segment (March’14 to June’15) has risen more than 60%, while the segment of SME’s without the supporting factor has shrunk almost a 50%. This evolution, as said before, is the result of several inputs, one of them being the supporting factor.