Payday loans are short-term unsecured cash loans based on the borrower’s personal check held for future deposit or on electronic access to the borrower’s bank account. Borrowers write a personal post-dated check for the amount borrowed plus the fee and receive cash. In some cases, borrowers sign over electronic access to their bank accounts to receive and repay payday loans.
Payday lenders hold the checks until the borrower’s next payday when loans and the loan fee must be paid in lump sum. Payday loans are made by payday loan stores, or at stores that sell other financial services. Loans are made online or on mobile devices. Payday loan companies serve customers who need money quickly and cannot get the money from banks or from credit cards.
The repayment cycle for such loans is very small averaging about two weeks. The loan is usually linked with the employment status of the customers hence most of its applicants are employed. The payment is usually done through checks or debit cards.
The payday market is a niche compared with mainstream consumer and credit-card loans, two markets where start-ups are now applying data science to lending. The cost factor maybe the only hindrance in the use of predictive models in payday lending. This barrier has been removed with the advent of open sources like R, MySql etc.; that have made it possible to develop predictive models and forecasting solutions for payday lending in a cost-effective manner.
We assisted one of our clients who runs a Pay Day Loan company operating multiple stores across the country. The client also provides a wide range of other related financial services for customer in revenue.
The client’s existing system was used to estimate Cash order quantity and End Daily Cash balance, which was primarily based on analysis of limited store wise variables. This resulted in sub optimal availability of cash and at times surplus cash resulting in money lying unutilized and cash deficit preventing lending money to the customers causing loss in revenue.
Design and development of structured solution framework comprise
i. The various components in the Cash Flow were evaluated & analysed to determine the Volume and percentage of change with peak and trough periods in their time series based on our experience and understanding on similar business.
ii. Various Time Series forecasting algorithms were tested to determine the most accurate and viable option. The Weighted Moving Averages Algorithm with an additional parameter to capture the seasonality in the time-series, was implemented.
iii. A system integrating the Client’s web interface to this new model was formulated. This enabled the Cash order recommendations to be displayed on the Client’s web interface.
The cash ordering & forecasting model resulted in reduction on the residual cash by 20% at stores. This resulted in decreasing cost of idle money for the client and in turn adding to the top line growth for the business. An accurate forecasting of the cash balance reduced risk of holding surplus cash overnight at the stores. The new flexible system allowed change in the cash ordering schedule for unforeseen circumstances.
The ordering process is rendered easily through the interface with the display of baseline recommendations. Simplified reporting and monitoring of cash flow was instrumental in addressing changing cash flow trends in timely manner. We offered affordable analytical solutions to optimize the costs and maximize revenues for the client’s business.
The future of payday lending looks brighter because of data analytics. Processes will be easier, faster, and more cost-effective for both the lenders and the borrowers. There isn’t anything that big data has not touched. It is nearly impossible to find some aspect of business that remains unaffected by the rise of big data and the technologies that come along with it.
