What is Early Warning? Loxon’s early warning system is a risk monitoring application providing efficient support to your credit monitoring activity. The main goal of EWS is to identify possible threats in retail and/or non-retail credit portfolio (in early stage) and to support the Bank to take the necessary measures on time.
Early Warning for prudent financial professionals
- Collect all relevant data from your source systems for Early Warning calculation
- Support the controlled manual data (e.g. qualitative information) capture
- Schedule recurring events (customer visit and phone calls, manual control of watch-listed cases etc.)
- Provide you full-scale parameterization so that you can change the models and the qualitative questionnaires later on
- Calculate different types of signal values and determine signal flags
- Calculate Probability of Default with statistical prediction methods (scorecards)
- Send different levels of signals (warnings, alerts)
- Execute Root cause analysis, identify and display the possible root causes
- Assign customers to the appropriate risk classification buckets or customer segments
- Provide you the opportunity to overwrite the classification result manually in a controlled way
- Support monitoring review from customer
- Support the escalation process and the pre-workout phase Generate documents for decision making (Breach report, Classification proposal etc.)
Our valued customers
How does Early Warning work?
The early recognition of potentially problematic exposures or clients could be based on the up-to-date prediction of the probability of default (PD), using a behavioural approach. The prediction model could be expert-based or statistical.
Obviously, any statistical model requires a thorough backtesting (incl. root cause analysis, EWS signal significance tests) and an expert based validation (plausibility checking, reject inference considerations). The major advantage of a behavioural concept is that the discriminatory power of it (i.e. how well the model can differentiate between good and bad customers in advance) significantly outperforms the strength of a purely application data based rating or scoring system.
The simple reason for this is that a behavioural model can incorporate all the precious information generated after the origination (e.g., account balance history, changes in availability information, latest financial data, industrial outlook, rating path).
Based on the forecasted PD, you can identify the most profitable way of customer relationship management. The business goal is to select the action type which would cause the largest increase in the expected profit or in similar customer value measures, calculated on the basis of the expected reduction in the PD, lengthened customer lifetime and the cost of the treatment.
In many cases, you will experience an improvement in the creditworthiness of the client (a decreasing PD predicted by the behavioural model). In these cases a X-sell or up-sell action could generate the maximum profit. The expert-based prediction system evaluates several signals concerning every client based on the available quantitative and qualitative data and on the parameterized signal rules.
The statistical system evaluates the applied scoring function (linear, logit or probit) at the given input parameters and suggests the optimal action.
The Loxon Early Warning System can support the escalation process and the pre-workout or any other optimized (segment-based) customer management workflow, customized to your specific needs. The System will historically register all the actions taken by the Bank and the reactions of the customer.
The list of problematic clients (and information about the clients) can be produced quickly and efficiently with the help of the reporting module. The reports can easily be customized and new reports can be added with a minimum requirement of human resources.