Credit scoring has been used for more than half a century by the financial industry all over the world to automise the assessment of risk. In microfinance, credit scoring is still not very common but change is under way. The authors of this chapter believe that credit scoring will become a standard application in microfinance within the next 10 years. This chapter lays out how credit scoring works, how microlenders and their clients can benefit and, finally, what are crucial success factors in the implementation process.
What is behind the increasing interest of microlenders in credit scoring?
In the past years, the demand for credit scoring from microfinance institutions (MFIs) and banks has increased sharply.1 A number of institutions have started implementing credit scoring solutions and their number is likely to grow significantly over the next few years.
The transition of MFIs from non-profit NGOs to commercial entities around the globe, the increasing consciousness of investors and lenders to commercial viability, the growing competition and shrinking yields in saturated microfinance markets and, last but not least, the impact of the financial crisis are key factors driving managers to improve the efficiency of their microlending operations. The 100 largest microlenders reporting to the MIX Market showed average year-on-year portfolio growth of almost 50% between 2005 and 2007.
According to the same source, at least eight MFIs report over a million clients. Several strong global networks have emerged over the last years, increasing the concentration in the industry.
Investors, lenders and supervisors are becoming more concerned about the importance of credit risk management solutions which quantify the risk exposures taken. The growing relevance of Basel II standards in many emerging markets is another important factor.
Exhibit 1: highlights some of the most common reasons why microlenders decide to implement credit scoring.
Beyond the myths: understanding the basics of credit scoring.
Credit scoring is used to predict the probability of a given borrower defaulting on his or her loan. The default probability is calculated based on the influence of defined risk factors on loan repayment performance of a cluster of similar clients in the past. The first commercial use of statistical methods to define the repayment probability of borrowers was made by the mathematicians Bill Fair and Earl Isaac for the American Investment Company in the late 1950s.
To ensure statistical significance, a sample of not less than 500 bad and 500 good loans are collected and analysed. Thirty or more relevant risk factors are reviewed, such as ‘years of experience’, ‘marital status’, ‘type of business activity’, etc., with regard to their contribution to the total risk exposure of one client. In the absence of a sufficient quantity or quality of data, the values can be assigned also based on judgment of experts. However, the predictive power of an ‘expert’ or ‘judgmental’ score card is normally significantly less than for a statistical score card.
The final score for a given applicant is the sum of the scores for each risk factor and corresponds to the likelihood that the client will default on the loan. Loan applications with a low risk and those with a high risk may be approved or rejected automatically without review by a credit committee.
Both credit scoring and ‘manual’ risk assessment approaches use a similar set of risk factors to determine the risk level of a given loan application. The main difference is that under a credit scoring approach risk factors are quantified and weighted based on a statistical analysis, while the traditional risk assessment is made based on the ‘manual’ analysis and weighting of risk factors relying solely on the experience and judgment of involved lending staff.
The discussion on the use of credit scoring in microfinance has been, at times, controversial. Some often used arguments of opponents are discussed below.
Credit scoring works only in countries with an established credit bureau.
While information obtained from a credit bureau can be helpful, a scoring system can work as well without, as do traditional lending schemes.
Credit scoring works only where the credit analysis is based on formal financial statements.
If lending staff are entrusted to reconstruct the financial statements of informally working micro-entrepreneurs under a ‘traditional’ lending scheme, there is no reason to doubt why this should not work with scoring. Both schemes work with the same amount of information.
Credit scoring leads to a loss of personal responsibility of lending staff for the credit decision.
Loan officers remain responsible for the repayment performance of their client portfolio and can be granted the right to disagree with the decision made by the scoring system.
Credit scoring has a long reaction time in taking into account the impact of a changing environment.
The validity of the score card against real repayments is monitored on a regular basis. The frequency of monitoring is normally between one and two times per year, but can be increased upon necessity. Scoring systems can be programmed to ‘learn’ from a changing risk environment in real time.
Impact of credit scoring on microlenders…
The impact of credit scoring in microfinance has not been the subject of scientific research on a larger scale so far. Nevertheless, the existing anecdotic evidence from projects implemented so far allows to make some important conclusions.
Productivity of lending staff, measured by the number of loans disbursed per loan officer shows increases after the implementation of credit scoring between 15% to 30%. This effect is mainly due to the shortening of the loan cycle by abolishing the credit committee and the onsite visit for a significant part of the loan applications.
The enhancement of standardisation of the loan approval process is viewed by microlenders as another significant advantage. As the scoring engine is managed by the head office directly, changes in the risk environment can be fed into the score card centrally, ensuring an immediate effect throughout the entire branch network.
The quantification of risk in the form of a credit score can be used easily to analyse the risk-adjusted return of every single loan and introduce risk-based pricing. A more accurate pricing, taking in account the risk profile of an individual client, provides a significant competitive advantage to microlenders in maturing markets where price becomes as important for clients as the service provided.
Many institutions indicate that the review of their lending processes and MIS solutions, an integral part of a well implemented scoring project, have revealed efficiency gaps which were closed in the course of scoring implementation.
As Exhibit 1 shows, lenders with problems such as poor portfolio quality, high rejection rates of applicants and slow credit approval process can experience dramatic performance improvements as a result of more accurate risk assessments and the application of automatic decision-making processes.
…and their clients
Lending institutions using credit scoring have the possibility to pass a number of potential benefits of scoring to their customers.
With an improvement of the organisational efficiency, credit scoring enables microlenders to offer loans at a lower interest rate—an impact which in particular can be realised in maturing microfinance markets.
Marginal micro credit borrowers can benefit from scoring as it allows microlenders to manage their portfolio at risk ratios more precisely then a traditional lending approach would allow. An institution willing to gradually increase the access of higher risk clients in order to increase their outreach will find in scoring an efficient tool to manage this process. Scoring can help institutions in particular where a high rejection rate is combined with excellent portfolio quality to gradually open their doors for higher risk—but still profitable—clients.
Critical success factors for implementation
The complexity of implementing a credit scoring system is often underestimated. Understanding the concept is not difficult, but making it work at the intersections of the credit division, senior management and IT function has often proven challenging. Three key success factors are described below.
A feasibility assessment before the project start is strongly recommended in order to define if key eligibility criteria are met. One of those is the scale of operations, allowing the microlender to build up a database with the necessary number of good and bad clients within a reasonable period of time. At least 750 loans disbursed per month are considered the treshhold for the implementation of a statistical scoring system.
In most cases, scoring projects aim to improve the efficiency of lending operations. To achieve a satisfactory improvement, other factors impacting efficiency, such as staff qualifications and motivation, design of the lending process and similar factors, should be analysed and improved as part of the implementation.
The MIS solution represents the ‘wheels’ of the scoring engine. An application processing system (APS) is today considered essential for the implementation of a scoring solution in order to ensure the smooth processing of data in real time. Ideally, the MIS platform to carry the scoring engine is in place when the project starts, to avoid implementation risks.
The question is not whether credit scoring will become a standard application in microfinance, but how fast this will happen. Managers of microlenders, investors and donors can play a crucial role in enforcing this process by supporting projects which generate showcases and disseminate information to the industry.
In a recent survey conducted by BFC in three countries in Eastern Europe among 58 banks and microfinance institutions, credit scoring ranked second among nine potential areas of desired technical assistance support.
By Michael Kortenbusch and Peter Hauser, Business & Finance Consulting (BFC).
The Euromoney Environmental Finance Handbook 2009/10
ISBN: 978 184374 593 8
Date Available: September 1st, 2009