How models can help in building risk management culture in a bank

Transcription

1 How models can help in building risk management culture in a bank Djordje Stojanovski, CFA, PRM Banca Intesa Beograd Chief Risk Officer Rome, November 2016 INTERNATIONAL SUBSIDIARY BANKS DIVISION

2 Content About us Business and regulatory environment Risk management culture Using models to build risk management culture 2

3 About us Member of Intesa Sanpaolo Group (ISP) Intesa Sanpaolo Group: largest bank in Italy among 15 largest banks in Europe among 50 largest groups in the World investment grade (by Moodys, S&P, Fitch) Banca Intesa Beograd: largest bank in Serbia market leader in all aspects (assets, capital, loans, deposits, profitability) 1.6 million clients, 200 branches, EUR 4 bln of assets, 3,000 employees member of ISP since 2005 among most important foreign subsidiaries of ISP Group 3

4 Business and regulatory environment Recent history Serbia is not member but is a candidate for the EU membership Serbian banking regulation is converging to the EU but still there are differences in capital standards which require parallel reporting to local and Italina regulator Economy highly dependent on trade and economic growth of EU (primary Germany and Italy) Economy in crisis from 2008 till mid size banks have failed, draining entire Deposit Insurance funds Banking sector suffers from high level of non-performing loans Public Debt to GDP has doubled in the last 8 years 4

5 Business and regulatory environment Current trends Political stability Mild recovery of the economy, industrial production, export, FDIs Slow reduction of non-performing loans: Strenghtening of collection and work-out capacities Improvement of systems for early detection of credit deterioration and proactive credit managemen Higher credit standards and better selection Easier balance sheet cleaning through write offs Development of market for NPL Fierce banking competition due to: Extremely high liquidity of banking sector Limited investment oportunities Negative interest rates on interbank market Higher lending standards reduced number of eligible clients Regulatory storm - cost of regulation constantly increases in last 6 years 5

6 Risk Management Culture Theory Risk Culture is a cornerstone of a sound risk management system of a company A company has good risk culture if it has risk-based governance, i.e. if risk management is part of all processes, services, activities, decisions, performance measures etc. There are numerous books, articles, international standards explaining sound principles and best practices of good risk management culture They explain and elaborate on fundamentals of sound risk culture: Board overview of risk management system Effective reporting and monitoring system Clear definition of roles and responsibilities Segregation of roles between risk decision takers and risk managers Clearly defined risk appetite and risk limits Remunaration system that properly balance between risk and return KPIs Risk parameters and measures used in all business decisions and management areas (underwriting, provisioning, monitoring, pricing, asset management, human resource management, internal controls etc.) Why is it so difficult then to build good risk culture in an organization? 6

7 Risk Management Culture Practice The problem is that books and articles on risk culture are read almost exclusively by risk managers. Even if risk management function participates in the top management, it often does not have enough power to syndicate importance of risk culture to the board and entire organization. In most of the organizations risk management is seen primarily as a segragated function, team of scientists that measure something, set and monitor some limits and thus, make constraint to the business. If we are lucky, we are going to be considered by the rest of the bank as regulatory cost to do business, something as taxes or a pay tool on a highway. Regulatory rules and standards, trainings/courses and audit reports on corporate governance and risk management, may help us to get attention of business departments and top management about the importance of sound controls and risk culture. However, experience tells us that best way of building the risk culture in everybody s minds is to get them feel some benefits from risk management. 7

8 Case study on small business Before 2014 wast majority of loans to small business segment were approved through traditional credit decision making process: Branch Product HQ offer Loan request KO criteria Black list Credit Bureau Client s exposure Client premises Report from site visit Loan contract signing Disburs ement Analysis Approval Legal Final approval Decision making process was centralized. 6 credit analysts were dedicated to SB segment, while Legal Department and Loan Administration capacity was shared with Corporate segment. Process was time consuming and expensive, both from financial and reputational point of view. 8

9 In 2014, the bank has made new credit strategy for next three years that has emphesised two goals: diversification of credit portfolio increase of quality of credit portfolio Implementation of the first goal was planned to be achieved through very ambitious growth rates of small business portfolio: Dynamics of SB loan outstanding volumes plan % +7.48% +5.66% +5.8% % Second goal was meant to be achieved through better and more efficient client sellection. Risk Management has proposed simplification of credit process.the idea was to fully automate credit underwriting for standard products and small tickets. The automation of the credit process would be based on the newly developed credit rating model with some supplementary credit criteria. The estimation was that this approach would boost SB portfolio without compromising credit quality. 9

10 Main issues of the previous credit process: TIME TO CASH A LOT OF ROLES IN DECISION- MAKING Process length was too long (bottle necks in Credit Management and Legal Department). Completely centralized model of decision making Too much manual work loan requests go through 4-5 people regardless of (how small) the exposure TREATMENT? Identical loan approval process for SB segment and Corporate LACK OF CLEAR RULES High discretion level in decision making. Nontransparent decision making criteria. CUSTOMER (DIS)SATISFACTION Low level of customer satisfaction with service provided Average time to cash period for SB loans was 15 days. Only the credit analysis phase took 6 days in average. Almost 20% of approved loans were not realized either because customers did not want to accept additional conditions required by Credit Department or because they found other sources of financing. This was a recipe for customer dissatisfaction. 10

11 Characteristics of the new credit process: AUTOMATION Automation of decision making process for low risk clients/ low exposures, based on scoring and pre-defined criteria. Fully standardized approach. DECENTRALIZATION RATING MODEL Decentralization of decision making authority to regional level in accordance with an acceptable level of risk and workload. Client s rating is key element of credit decision making proces. TRANSPARENCY Clear guidelines for credit analysis and decision making for loans that are processed by Credit Department SIMPLIFICATION Simplifying product catalogue - reducing the number of commercial products New process would be applied to all standard products and exposures at client level of up to EUR 25k. This covers 89% of all SB clients and apx. 37% of volumes. 11

12 Developing new rating model Using previous modelling experience (scoring schemes for consumer lending to individuals and SME rating models), available information (analytical data basis) and SAS Enterprise Miner, modelling team in Risk Management Department developed SB rating model for managerial purposes. Variables included in the model cover financial and behavioural data, the clients delinquency/blockade history information as well as some of the owner s individual data Sources of information are financial statements, credit bureau data, payment operations data Cluster analysis has shown that we cannot have more than 7 meaningful grades, thus we developed an S&P like scale. The intention was not to copy or mimic S&P data on PDs but just to have something convenient and userfriendly to the analyst and relationship managers. At first, the model was used for limited pre-approved campaign and shown good performance statistics 12

13 In the process, model was accompanied with additional criteria (knock out rules): Knock-out criteria residency current account blockade max 5 consecutive days in previous12 months or 10 days in previous 6 months no delays in repayment max EUR 40 in delay less than 30 days positive net results no start up frim accepted minimum EU 15 k of turnover client or owner not on the fraud list or the black list clean Credit Bureau report Additional criteria annual debt service max 30% of annual income no delays in repayment in last 12 months no outstanding restructured or rescheduled loans maximum loan amount is 20% of annual income maximum overdraft amount is 5% of annual income 13

14 New credit decision making scheme Rating Maturity Cumulative PD 12 months 18 months 24 months 36 months 60 months AAA 0.37% 0.55% 0.74% 1.11% 1.85% AA 0.50% 0.75% 1.00% 1.50% 2.49% A 1.23% 1.84% 2.45% 3.68% 6.13% BBB 2.09% 3.13% 4.17% 6.26% 10.43% BB 3.88% 5.82% 7.76% 11.65% 19.41% B 8.22% 12.33% 16.45% 24.67% 41.12% CCC 35.33% 52.99% 70.65% % % Automated Machine Decentralized Regions/Branches Centralized Credit Department unsecured secured 14

15 Performance Observed date: Number of loan % requests Total requests ,0% Applications in progress 627 1,6% Rejected applications ,0% Approved amount (EUR) Delinquency >90d, > 100 % Delinquency amount (EUR) Disbursed loans ,4% ,65% ,28% FINAL SCENARIO RATING CLASS DISBURSED LOANS ANALYSIS Scenario A (BM) ,9% ,46% ,28% Scenario B1 (SM) ,9% ,16% ,83% Scenario B2 (RD) ,0% ,71% ,86% Scenario C (CMD) ,2% ,77% ,22% AAA ,8% ,13% ,06% AA ,8% ,23% ,08% A ,9% ,49% ,35% BBB ,7% ,53% ,23% BB ,2% ,31% ,63% B ,4% ,93% ,02% CCC 382 1,0% ,62% ,34% N/A ,2% ,96% ,23% % 15

16 Portfolio distribution BB 14,69% B 4,86% CCC 0,64% AAA 23,34% AAA AA A Product maturity analysis BBB 19,41% A 15,64% AA 21,42% BBB BB B CCC Product maturity Number of products maturity <= 6M M < maturity <= 12M M < maturity <= 18M M < maturity <= 24M maturity > 24M 541 * M - months Rating distribution 30,00% 25,00% 20,00% 15,00% 10,00% 5,00% 0,00% AAA AA A BBB BB B CCC Disbursed Rejected 16

17 Performance Vintage analysis Percentage of clients in deliquency MONTH COUNT 1m 2m 3m 4m 5m 6m 7m 8m 9m 10m 11m 12m 9, ,00% 0,00% 0,00% 0,00% 0,00% 0,36% 0,36% 0,54% 0,54% 0,54% 0,72% 0,72% 10, ,00% 0,00% 0,00% 0,00% 0,00% 0,00% 0,16% 0,31% 0,31% 0,47% 0,63% 0,63% 11, ,00% 0,00% 0,00% 0,00% 0,00% 0,00% 0,00% 0,00% 0,42% 0,63% 0,74% 0,84% 12, ,00% 0,00% 0,00% 0,00% 0,00% 0,00% 0,24% 0,24% 0,40% 0,48% 0,55% 0,79% 1, ,00% 0,00% 0,00% 0,00% 0,00% 0,33% 0,33% 0,66% 0,83% 0,83% 0,66% 0,66% 2, ,00% 0,00% 0,00% 0,00% 0,12% 0,23% 0,46% 0,46% 0,58% 0,92% 0,92% 1,04% 3, ,00% 0,00% 0,00% 0,00% 0,09% 0,09% 0,18% 0,27% 0,27% 0,53% 0,44% 0,53% 4, ,00% 0,00% 0,00% 0,00% 0,00% 0,27% 0,48% 0,64% 0,86% 0,86% 1,02% 1,12% 5, ,00% 0,00% 0,00% 0,13% 0,33% 0,40% 0,60% 0,67% 0,73% 0,93% 1,00% 1,13% 6, ,00% 0,00% 0,00% 0,00% 0,00% 0,04% 0,12% 0,19% 0,35% 0,46% 0,50% 0,65% 7, ,00% 0,00% 0,00% 0,00% 0,06% 0,23% 0,40% 0,40% 0,51% 0,80% 0,86% 0,86% 8, ,00% 0,00% 0,00% 0,00% 0,13% 0,13% 0,26% 0,40% 0,66% 0,86% 0,86% 9, ,00% 0,00% 0,00% 0,00% 0,11% 0,16% 0,32% 0,42% 0,74% 0,95% 10, ,00% 0,00% 0,00% 0,04% 0,18% 0,27% 0,44% 0,62% 0,80% 11, ,00% 0,00% 0,00% 0,10% 0,10% 0,24% 0,39% 0,43% 12, ,00% 0,00% 0,00% 0,00% 0,00% 0,04% 0,13% 1, ,00% 0,00% 0,00% 0,07% 0,14% 0,14% 2, ,00% 0,00% 0,00% 0,00% 0,05% 3, ,00% 0,00% 0,00% 0,05% 4, ,00% 0,00% 0,00% 5, ,00% 0,00% 6, ,00% Low and stable default rate Annual default rate 0.83% by count and 0.3% by amount Average interest rate on SB loans is 8.9%, with 3.5% funding costs it is clear that the CoR is acceptable. 17

18 Achieved results Approximately 65% of loan applications are approved through automated process Time to cash reduced from 15 days to 6 days Time to yes reduced from 6 days to 2 days Higher standardization and transparency of credit decision making process Good performance of approved loans credit quality not compromised Significant growth rates increased volumes and profits (over 30% growth in less than 2 years) Significant increase in market share (market share in new lending goes up to 30% compared to 15% market share) 18

19 Remember Credit Strategy goals from the beginning of the presentation? Credit quality Difersification 19

20 Consequences for the RISK CULTURE Top management and business have significantly improved their risk awareness Risk models are now finally perceved as advantage and not a burden to the business Risk management IS NOT considered anymore as pure regulatory cost Risk parameters (rating, PD, expected loss) have become common language of top and senior management, as well as of all decision makers/risk takers Risk measures are accepted as reliable parameters for measuring business performance (RAPM as KPIs) Front office (busines divisions) has started to raise initiatives for further use of rating model as managerial tool for other aspects of credit management process (for example pricing policy) Communication and trust between risk takers and risk managers have been upgraded to a higher level. We finally play as a team. 20

21 21

22 Discussion 22

23 23