Risk Data Silos

In an article published in Future Banking the authors have lucidly presented a  case for using Business Intelligence to manage the chaotic world of data silos. They have made specific reference to risk data silos in the context of Basel III. According to Chartis, Banks and financial institutions worldwide will invest $27 billion in their risk management initiatives – that will include tools to dig into these silos - to provide actionable risk intelligence. These initiatives will form the crux of the post crisis response to a new regulatory regime as well as information consumption paradigms that are emerging.  

When we look deeper into these extant data silos, we definitely see more chaos than order even in large financial institutions.  In fact I would argue that the chaos is independent of the size or market share of the bank. 

The center piece of credit risk management is customer intelligence.  From a bank’s perspective, knowing the customer – beyond getting the contact address and drivers license info – is key to managing its default risk. 

Using their customer insights, Banks typically cross sell a variety of loan products   - from home equity lines to credit cards -to existing DDA customers.   

Below are some examples that I have seen in my experience while working at large banks.

a) the credit card SBU would be offering  lower credit line to a high net worth customer resulting in poor customer experience and potential loss of business

b) the loan officer would be happily underwriting a mortgage loan to a customer who  currently has past due payments on Credit card

b) the credit card SBU would  continue to send mail solicitation to its retail banking customer who has just started experience difficulties in making monthly mortgage payments

( Note: Much of this intelligence cannot be garnered from credit bureau data because there is a time lag before the bureau reports it)

All these could have been easily avoided if the bankers had timely and complete vision of truth or appropriate business intelligence.

The recent financial crisis spawned a massive loss mitigation effort by the banks. Many creative strategies have been developed to proactively manage default risk. For example, retail banking relationship – defined as having a checking account or a credit relationship (example – auto loan) provides powerful insights and often early warnings on loan performance.  Intelligence on presence or absence of significant retail relationship is now a key component of a bank’s new account acquiring strategy. However, the existence of data silos is a hindrance to providing a 360^o view of the customer.  

Interestingly, banks have the analytical capability to compute the benefits or loss savings that will accrue if they remedy the situation .  In other words they know in dollar terms how much this intelligence costs them.

Another dimension that forces attention on the chaos is the new regulatory regime such as Basel III. Compliance to the new regulations would automatically mean better data management leading to timely risk intelligence. Banks will need additional investments in time and money for compliance.

  

These insulated data warehouses and silos often weaken the integrity of a comprehensive credit risk management system.

Risk Intelligence from Social Networking – New Dimensions in understanding Consumer behavior

 

I came across a recent news item in Orlando Sentinel about how a Florida judge stopped a debt collections agency from pursuing an auto loan customer who was late on payments. Apparently the collections agency tracked the customer’s Facebook account and contacted her as well as sent messages to all her friends on Facebook. The judge’s ruling in the case may have highlighted privacy concerns, but what is of interest is that social network intelligence are opening up new and creative ways of understanding and managing consumer behavior.

Now, if debt collections agencies can use social networks in recovering payments - well at least until the Florida judge’s ruling – it would be of interest to know how these networks can be used in larger credit risk management initiatives by banks and financial institutions. While the regulatory / compliance / privacy angle is evolving, there is no doubt that social network intelligence provides powerful yet undiscovered dimensions to our understanding of consumer behavior.

Banks and financial services companies have always sought to better understand their customers. Their primary resource has been the credit bureaus (Experian, TransUnion, Equifax) who have a comprehensive record of credit activity of consumers.  From marketing to risk management, the credit bureaus continue to be the single go-to place for insights on customer’s past and also predicting future behavior.  Using their vast store of data, the credit bureaus created credit scores that were designed to predict the probability of a credit outcome such as default on a loan. 

Conventional risk analytical paradigms overlay credit bureau data on set of existing customers whose performance (example payment behavior) is fully tracked internally.  This would provide insights on external behavior - such as payments as agreed to other lenders or shopping for more credit - that may be useful. Matching this insight with internal data would provide risk intelligence that would be used in granting or denying additional loans / lines to the existing customers.

But what type of risk intelligence can be generated using social network intelligence?  While this is unchartered territory, definitely we can identify social network behavior of customers whose payment behavior we already know. We can use the same analytical paradigm of taking a set of existing customers and match it with social networking data – travels (location), payment transactions to glean insights. For example we can track delinquent customers’ travels and see if their delayed payments arise from travel and travel related splurges.  Frequent holiday travel alerts emanating from social network intelligence can lead lenders to proactively minimize / reduce / freeze credit lines. While the legal and regulatory ramifications of such actions are evolving and will be tested in courts or when regulators take a stand, it is well known that Banks have used and will not hesitate to use customer intelligence in creative ways to manage their profitability.

I see a new dimension to credit risk management opening up by using social network intelligence. There is no stopping this genie.

Retail Banking Risk Dashboards

This write-up details KPIs that provide Risk Intelligence on a wide spectrum of consumer lending products. These reports provide actionable intelligence from executive management to operational managers in Banks and financial services institutions. These reports can be used to report on a host of loan products such as

o   Home equity lines of credit (HELOC)

o   Home refinance loans (HRL)

o   Home mortgage

o   Auto loans

o   Credit Cards

o   Unsecured loans

o   Small business loans (SBA) / lines of credit

Further these reports can easily be tailored to report on other unique or exotic consumer loan products offered by individual Banks and financial services providers. Currently most bank use Excel or PowerPoint paper outputs as the delivery media for these reports. Since usually these reports make a big pile of paper reports, they provide a great opportunity for  iPad apps.

Retail Banking Dashboards that support loan products management generally fall under two large groups of reports. The groupings reflect the back-end databases/ source systems from which these dashboards are generated. 

o   Acquisition Related Reports

o   Existing Account Reports

In addition to these reports, Banks also invest resources in building Regulatory / Compliance / Governance / Audit reports.

Acquisition reports are really a point in time view of an ever moving target. For example it provides executives and managers insights on day-to-day account bookings, dollar volume, demographics and risk segmentation of credit qualified prospects and approvals. They provide snapshots on short time spans typically daily or weekly reports. They typically use OLTP / front end underwriting systems as data source to generate the reports.

 On the other hand existing account/ portfolio reports generally source their data from a data warehouse that is refreshed usually every month. In large banks, typically several intelligent cubes or OLAP cubes will be the prime provider of data for such reports.  Although there is a time lag, these reports provide powerful insights to the Chief Credit Officer / Chief Risk Officer and his executive team to efficiently manage their portfolio in accordance with stated corporate objectives - usually financial targets.

Risk Management KPI for Consumer Lending Portfolio - An illustrative list of reports

Acquisition Related Reports

ü  New Account Acquisition Management KPI

ü  Loan applications volume / on line enquiries / pre approved solicitations

ü  Approval Rate  i.e  Approved Applications / Credit Qualified Population

ü  New Customers vs new applications from existing customers – Wallet share?

ü  Customer Segmentation -   Risk Segments  - By Credit Score Bands

ü  Income segments  -  Debt to Income ratio

ü  Current Credit  Bureau Debt Level  - shows overall debt burden

ü  Existing customer – Wallet Share?

ü  Geographic Segments                   

ü  Average  Daily Balance on DDA

ü  Exception Reporting  Low Side Overrides/ High Side Overrides

                        

Existing Account Reports / KPI

ü  Balance and account attrition reports             

ü  Current level of Line Utilization or unpaid balance

ü  New loans sought in last 3 months?  Shopping or Wallet share opportunity?

ü  Credit Bureau Triggers - Any derogs  with any other lender?

ü  Debt to income ratio changes / alerts

ü  Change in income segments – Mass affluence migrations/changes

ü  Credit Score Migration (Change / Deterioration in loan quality)

ü  Adverse public data – possible impact on loan quality

ü  First Payment Defaults

ü  Delinquency Reports / KPI – Past due reports by severity

ü  Charge off reports – Loan default KPIs

ü  Bankruptcy  Loss Reports

ü  Fraud Losses

ü  Forecast vs actual loss reports

As mentioned earlier, this is only an illustrative list to provide guidance in designing Risk Intelligence Dashboards for Retail Banks and Financial Services. Different variations and permutations of these reports can be incrementally added to meet client’s requirements.

Credit Risk Primer

Credit Risk Management has attracted a lot of attention in the recent times. The recent financial crisis has been a trigger for this new interest. The discipline itself, however, is as old as banking and money lending. In this paper, discussion will be focused on the credit risk management function in retail banking and the credit card world.

After the recent financial crisis, banks are investing heavily in strengthening their risk infrastructure by bringing in new technologies and tools like Business Intelligence (BI) Tools. These new tools will enhance timely delivery of key risk metrics that will help risk managers mitigate portfolio losses. 

A commonly found definition for Credit Risk is that it is the risk of loss due to non payment on loans as agreed. When the borrower is unwilling or unable to adhere to the repayment schedule, the loan account becomes delinquent and is eventually charged off if no further payments are received. This is a real risk that directly impacts the lender’s bottom line. Identifying, quantifying and mitigating this risk of loss of capital and interest are the principal focus of Credit Risk (CR) Management , the subject of this primer. 

CR is a key function in all banking and lending institutions - whether they are small, medium, or global financial institutions. It plays a pivotal role by designing the Bank’s loss management and regulatory compliance policies. The degree of sophistication, however, of the risk function varies across these organizations.

Organizationally, individual banks may have different structures for CR. Two common arrangements are generally found: In one, multiple responsibilities making up the CR role report to a single function or strategic unit.; In the second, CR reports to the Chief Credit Officer and the functions are often named differently.

Credit Risk Management Function

CRM functions in a bank typically include the following responsibilities:

Ø Credit Risk Policy

· New account acquiring system

· Portfolio Risk management (existing loan portfolio)

Ø Loss Forecasts

Ø Decision Sciences and Predictive analytics

Ø Collections

Ø Risk IT

Ø Risk Governance , Compliance and Audit

Credit Risk Policy: This group is responsible for design, testing and approval of the bank’s credit risk policies. Policies include new account acquiring policy, portfolio risk management policy, fraud risk management policy, etc. For example, new account risk policy will determine the cut-off credit score above which new accounts / loans will be solicited. These decisions are supported by detailed and in-depth analytics on customer data that is often supplemented with additional data purchased from credit bureau or other external vendors. On the other hand, for loss mitigation on existing portfolio, customer retention strategies would constitute portfolio risk management policies.

Loss Forecasts: Banks compute Allowance for Loan and Lease Losses (ALLL) forecasts to provide for loss reserves. While several methodologies are available for computing loss estimates, the portfolio delinquency and flow rate methodology is widely used. This method uses loan past due information and rate of worsening delinquency to forecast losses. Many banks also use macroeconomic indicators / metric overlays like unemployment rate to fine tune their forecasts. Adequate loss reserves safeguard the bank as well as the depositors against expected losses in loan portfolio. Federal banking regulators such as Federal Reserve, Office of the Comptroller of the Currency (OCC), Federal Deposit Insurance Corporation (FDIC) and other Bank supervisory /regulatory agencies have issued detailed guidelines on loss reserves for banks. 

Decision Sciences and Predictive Analytics: This group uses statistical and data mining techniques to review historical customer data to identify behavior patterns and predict future customer behavior. This group custom builds scoring models that are targeted to predict given outcomes. Smaller banks buy scoring models from credit bureaus or other vendors. 

Collections: Often known as the Customer Support group, this group designs collection policies and strategies to recover receivables and past due payments from delinquent customers. 

Risk IT is outside of the Information Systems (variously known as IS or IT in many financial institutions) and generally reports to the risk business leadership. They are charged with implementing / executing risk policies on the different platforms and IT infrastructure in use. 

Risk Governance, Compliance and Audit group provides management oversight to CRM and ensures that all regulatory policies are in compliance.

Role of Data Analytics

Data analytics provides an important role in CR management. These can be regular or ad hoc analytics designed to provide specific information e.g., loan portfolio segments or customer responses to solicitations. Typically, large volumes of data are processed to segment the portfolio and compute key risk metrics. Banks usually use SAS, Cognos, Business Objects or MicroStrategy as their analytical toolset. 

Risk Reports, Dashboards & KPIs play an important role in Credit Risk Management. Reports such as those which show 30-Day, 60-Day and 90-Day delinquent accounts provide quick insights into the state of the portfolio. Segmentation of delinquency by customer profiles like risk-score, household-income and credit-utilization bands quickly deepen Management’s understanding of portfolio performance. Many banks have separate group within Credit Risk Department dedicated to producing risk reports.