A critical factor having a significant influence on the variance in earnings of lending institutions is the underlying risk profile of the credit portfolio. A proportion of this variance can be attributed to the prevailing general economic climate. However, a significant proportion of the remaining variance can be directly attributed to incorrect pricing for the given level of payment default risk presented by credit applicants.
If the true level of profitability by risk category can be quantified, then risk managers together with representatives from finance, operations and sales can better manage:
- take-on volume by risk category and associated default rates;
- absolute sales volume;
- cost of sales; and
- collection department efficiency
This greater degree of management control is provided by the following influencing levers:
- Credit policy
- Operational efficiencies
- Early and late stage collection strategies
Where to Start?
Profitability modelling should enable an institution to accurately quantify the impact on overall profitability of changes to one, or all of these levers. This, in turn, will assist the institution in determining and executing both credit policy and pricing decisions in a more scientific manner.
The best methodology is to identify key specific states within your data (e.g. up to date, 1 payment in arrears) and the possible events (e.g. a missed payment) that cause accounts to migrate between these identified states.
These movements are quantified by calculating transition matrices that measure the probability of a specific event occurring for each account within a given state and at a specific exposure point since its inception. This approach is beneficial as it does not solely focus on predicting a single ultimate event (e.g. 3 payments in arrears) as is commonly done within traditional profit tools. As all of the states and events are interdependent, possible errors that might occur when trying to scale independent curves for default, attrition, pre-payment and recoveries, are eliminated.
A further advantage of this methodology is that it can also be used to predict the number of accounts and the value of expected payments within various arrears states by exposure periods, thus allowing for better collections management.
The transition matrices predict the anticipated volume of accounts within specific states that trigger income and expense events. These income and expense events are defined using the product rules and operational processes followed by the institution, such that a model is developed that quantifies expected income and expenses associated for each individual account throughout its lifetime. Profitability models should use a statement view depending upon the term of the credit product and should typically be split into 5 main sections:
- Population movements
- Capital Flows
- Profitability metrics
Profitability models must calculate the following:
- NPV (the Rand value generated by selling a contract at the required discount rate)
- IRR (the rate of return generated by the contract)
- Perfect Price (the price that must be charged in order to achieve the required rate of return)
- Break-even Default (the default rate that can be supported in achieving the required return)
- Loss Given Default (LGD)