Resources
Resources
Abstract
A minimal set of data fields for evaluating overdue account receivables in personal loans business is proposed to include risk factors, loan characteristics and behavior parameters. Important practical problem is loans portfolio evaluation by imperfect and/or incomplete data. In this case it is proposed to evaluate risk factors mainly by finding hidden links between accounts (i.e. relatives by family names or co-workers by work phone numbers). Such links may indicate fraud cases or possible material problems in households. Analyzing loan parameters such as interest to loan ratio allows to roughly estimate number of due payments before arrears.
The article proposes minimal set of data fields for loans portfolio evaluation and describes methods of analysis and possible conclusions that can be derived from such analysis.
Risks evaluation by imperfect data
Abstract
Optimization approach is proposed for efficient management of debt collection for personal loans. Traditional rule-based technique offers to make decisions of debt collection strategy upon rules such as ‘Stop dunning if cost of collection exceeded probable amount of recovery.’
Linear optimizing model has been proposed by the author:
(target function)
(boundary condition)
where:
xij - binary variable equal to 1 if we select strategy i for account j,
ci - planned cost of strategy i,
Cj - dawn cost of debt collection for account j,
pij - recovery prognosis for account j in case strategy i is selected.
Linear optimizing lets us to apply business limitations. For example, resources limitation can be expressed as:
where:
li - resources required for strategy i per one account,
L - total available resources.
Dunning strategy optimization approach