Use Collection Analytics

Analytics

Analytics provides organizations with better visibility into the factors that drive revenues, costs, and shareholder value. Today’s business challenges demand timely financial information that enables executives, managers, and front-line employees to make better decisions, take action, and correct problems before they affect the company’s financial performance.

Analytics provide insight to the people who can impact business performance.

Collections + Analytics = Improved Recovery.

Collection Analytic is nothing but an intelligent analysis, using combination of mathematical and statistical tools, to arrive at a “behavioral scoring”, depicting accurate picture of the customer’s propensity and ability to pay, which is used to segment defaulters, prioritize collections activities to maximize recoveries and reduce collections costs.

Effective collection analytics empower the collection staff to focus on the right debtors, maximizing the payments collected by a combination of segmentation, scorecards, and strategies to help manage delinquent accounts better.

Collections Analytics segments and identifies accounts representing those that self-cure, those that cure with engagement, and those that will improve, remain stable, or grow more delinquent. Using Collections Analytics you can assign treatments and protocols to each of your segments depending on the account’s recovery score, outstanding balance, and balance age. This ensures collection teams are distinguishing well-intended, “accidental” debtors from those under economic hardship or potential cases of actual fraud. Strategies can be interfaces with client interaction further enhancing productivity by leveraging and optimizing investments in existing systems and providing a seamless integration to tools already being used to help collection teams.

Why collection analytic?

Because it increases amount of collection and more importantly reduces the losses by 1-2%. This translates into almost the existing profit of many well-run companies; will company not like to double its profit?

Predictive models

Predictive models analyze past performance to assess how likely a customer is to exhibit a specific behavior in the future in order to improve collection effectiveness. Many types of models are available and uniformly, all of them deduce the analysis to a scoring. A good scoring model requires varied data accurately and on timely manner.

Data is at the heart of everything.  With expertise in the interpretation and use of credit bureau, clients’ customer data and customer contact history, Analytics turns this data into information, which enables organizations to predict how defaulters will behave in the debt collections process.

Do Indian companies require them? Yes, banks and big finance companies certainly require. But, are they ready? One is not sure.

Very few banks and lenders have applications data, credit bureau data, credit scores, and collateral details electronically. Some do not have a dedicated debt management system. Migration or pulling data from one to another system creates its own set of problems.

This apart, hardly any bank or lender has captured and recorded customer contact/ interaction history systematically and digitally. Without this crucial information, no predictive model implemented will churn out any meaningful results. It would take a few years more, provided companies realize the value and importance of recording customer contact/ interaction history.

Predictive models are of no use to Indian banks and lenders, in near term.

Descriptive models

This may be appropriate for India. Descriptive models quantify relationships in data in a way that is often used to classify default customers into groups. In other pages, we have classified and grouped them under four groups/ quadrants:

I           High Delinquent        -           High Risk

II         High Delinquent         -           Low Risk

III        Low Delinquent         -           High Risk

IV        Low Delinquent         -           Low Risk

Additionally, we may use six sigma concept to further identify high risk accounts within the groups based on payment pattern falling outside six sigma (standard deviation). This would help rank-order the cases within the groups or in the collection lists.

We will discuss the level of effort, type of effort, timing of effort and the right strategy for each the above groups in future blogs.

Classify To Bring Focus

Delinquencies need to be classified appropriately to bring the right focus. Different institutions use different type of delinquency/ collection reports. Each has its advantage and disadvantages.

Some foreign banks use collection reports based on ‘exposure given default’ – EGD. This gives a list in descending order of exposure. Here, the focus is on accounts with high exposure. Obviously, they do not want loss per account to be huge. Here, classification does not reflect neither level of delinquency or level of risk.

Some NBFCs use collection reports based on ‘past due amount’. This gives a list in descending order of default amount or past due amount. There is no clarity on delinquency and risk levels.

Most MNC lenders use collection reports based ‘days past due’ – DPD. This gives a list in descending order of number of days accounts in default. One can get ‘months past due’ by dividing DPD by 30. Months past due gives number of installments overdue.  Similarly, many credit card lenders use ‘buckets’ generally representing different months past due, say first bucket to mean account with past due from 1-30 days. This certainly indicates the level of delinquency and not fully the level of risk. Yes, default accounts with high DPD tend to be high risk; it gets increasingly difficult to collect with an increase in DPD.

Then, how do we measure risk, delinquency level and classify them broadly based on delinquency and risk, to bring the right focus by allocating and assigning resources to mitigate losses?

It is important to have accounts classified in the following matrix.

I High Delinquency - High Risk	III Low Delinquency - High Risk
II High Delinquency - Low Risk	IV Low Delinquency - Low Risk

Collection reports rak ordered by delinquency rate would help classify, based on risk.

Delinquency rate is, as defined elsewhere in my blogs, the rate of amount past due as compared to amount billed so far for the contract.

Top 20% of accounts on the list are certainly high risk accounts; balance can be classified as low risk accounts. These can be further classified based on DPD. Accounts with DPD higher than 90 may be classified as high delinquent and remaining low delinquent.

Now, we have 4 quadrants which constitute Debt Lifecycle as follows:

I        High delinquent – High risk

II       High delinquent – Low risk

III      Low delinquent -  High risk

IV      Low delinquent – Low risk

It is clear that the focus needs to be on high risk accounts and more on Quadrant I accounts which are High delinquent – High risk. Different quadrants require different levels of curing like calling, field visits, repossession, agency collection, legal actions, etc.

Try them; you will see results. Losses will come down.

Collection analytic will help further classify to bring razor sharp focus; I will write on collection analytics on different pages in near future.

Reduce Losses

Losses are a function of mindless orientation towards sales numbers and lack of credit management.

It is true that sales are important in any business, so is in finance business. Sales bring profits.

In many a bank, more so in MNC banks, I have heard that the top guy with orientation of marketing or credit is named alternatively for any branch. The first guy would show result in terms of sales and credit guy would clean up. It may be true because a country head or a product head in MNCs are normally for a three years term and they are under stress to show results to get a ‘better position’. Lest, they would be shown door.

In a competitive business like finance business, every thing gets measured by the boss, except the credit quality which shows up after an average period of 1-2 years. Losses occur after the term is over and the horse is bolted. It is important that banks and NBFCs have adequate credit analytics to show and even predict the quality of credit and portfolio regularly, at an interval not less than 3 months.

Probability of loss is 100% when there is lack of credit management in finance or banking business. They have a very high correlation with losses.

Credit losses arise because of:

1. External fraud by crime groups
2. Internal fraud by sales guys jointly with or without Direct Marketing Agents
3. Lack of credit history of customers
4. Bad credit policy
5. Too many subjective and qualitative assessment of credit
6. Too many exceptions because of interference of the top guy/ seniors.

Strong collection mechanism and strong credit management would deter crime groups and internal employees attempting to defraud by falsifying the application and documents. There are agencies like Experian is available, whose services will help avoid frauds in application stage to a large extent, at least in retail finance business, where number of applications is huge.

CIBIL

Credit history is made available by CIBIL in India; I believe they also offer credit score/ rating. They have done a commendable job of getting about 90 % of lenders to share the credit history of their customers. My experience, at least about 5 years ago, was that not many banks could share the right data in full, and from across India. More computerization of banks had happened since then and I am sure the data provided currently would be highly reliable. My suggestion would be to avoid all customers whose credit history is not traceable in CIBIL.

Should one use exclusively CIBIL credit scoring in approval process?

Rather not. CIBIL score is based on assessment of the ability of the customer to pay on the existing loans. It has no recourse to any other vital information that would be required to fully assess “ability and intention” of the customers and to quantify the credit risk of the customer more accurately.

Most of credit policies are made by "cut & paste" of competitor policies. It must be simple, clear and purposeful, supported by an internal credit scoring model, giving weight to credit history, credit score offered by CIBIL. Credit policy must be company specific and must be internal. There could be enough discussion on formulating the policy among heads of all departments including collection, sales & marketing. But, the interference must not be allowed with the policy till the next review meeting; feedbacks are welcome. Collection analytics like early default analysis, standard deviation of payments, ratio of high-risk accounts sourced in the last one year may be the bases for tinkering the policy and the norms like LTV, etc.

Strong collection department, and preferably internal recovery & legal vertical under collection department will go a long way in achieving better recovery ratio. Of course, the collection analytics will  help to direct efforts appropriately to achieve better collection and the right focus to mitigate losses.

Repossession of underlying asset is the most important activity to reduce or even mitigate losses. Why do you have an underlying asset as security, if you cannot have recourse to?

It is normal and easy to blame that collections are poor and losses are higher, because of inefficiency of collection department. May be. I recall here the famous statement of one my vice-president Carlos who once said, “Best of collections will not solve collections problem, but better credit will”.

Currently, the profit margins are 5-6% in retail finance business. Net margins are as low as 3-4%. It is clear that no bank or company can afford losses beyond this. What is important is that for, every 1% loss, the company's sales will have to be 33% more,  to have the same level of risk adjusted return.

Is it easy to do sales? Even if you can, is capital easily available? Even if capital is available, can you assure “return”? O boy! losses are too costly.

Is there an option? Reduce Losses.