Credit Scoring Its Effects And Diffusion Finance Essay - Free Essay Example

Introduction Credit scoring is a statistically derived numeric expression of a persons creditworthiness  that is used by lenders to access the likelihood that a person will repay  his or her  debts.  A credit score is based on, among other things, a persons past credit history. Credit scoring, using standardized formula is a measurement of  credit risk.  Factors  that can reduce  a credit  score  includes   absence of  credit  references, and late payments, and unfavourable  credit card  use. By using a credit score,  lenders determine  whether to grant a  loan, what  rate  to  charge and also the term.   For example, borrowers with a credit score that is under 600 will be unable to receive a prime mortgage and will typically need to go to a subprime lender for a subprime mortgage, in which will typically have a higher interest rate.. The designation for credit scoring is the FICO score is the single best summary score of ones credit worthiness. .A credit score number is often called a FICO score, a California company that developed the system upon which it is based. The score is supposed to distils all the information in your credit report, using a formula to calculate a single number that indicates your credit worthiness. Its designed to give lenders a fast, accurate prediction of the risk involved in giving you a loan. Lenders have attested to the scores value in streamlining the underwriting process and creating more opportunities for consumers to get mortgages. Scores range from the 300s to about 900, with the vast majority of folks falling in the 600s and 700s. The higher the score, the better. Literature Review Credit scoring: its effects and diffusion in the early stage A study was conducted by Jalal Akhavein, Scott Frame, and Lawrence (2005) to illustrate the effects of the introduction of credit scoring on small business credit market and determine the factors that influence the adaptation of this financial innovative among large banking organization during the middle 1990s. Jalal Akhavein, Scott Frame, and Lawrence (2005) stated that there were effects on borrower-lender interactions, loan pricing and credit availability since the introduction of credit scoring.They claimed that borrower-lender interactions might contract since credit scoring allowed lenders to grant or reject loans without physically meeting the borrower. They further stated the credit scoring might influence the price of credit as lenders would offer different price to borrowers according to the score gained. Regarding the availability of credit, they concluded there might be a rise in the number of credit, because cheaper or better information about the repayment prospects allowed lenders to price the credit accordingly, rather than rejecting loan out of fear. Jalal Akhavein, Scott Frame and Lawrence (2005) discussed how the three variables market variables, firms variables and Chief Executive Officer (CEO) variables influence the adaptation of credit scoring. According to the study, banks with market power and in less-concentrated markets relatively tended to adopt new technology credit scoring. The study also suggested that larger banking firms were more likely to adopt new technology because of their economies of scale and ability to pool risk. As suggested by the study, the personal characteristics of CEO had great influences on the adaptation of credit scoring as they had great power over decision making. Construction of credit scoring using data mining and its classification performance A study was conducted by Yap Bee Wah, Ong Seng Huat and Mohamed Husain Nor Huselina (2011) to demonstrate the use of data mining by using credit scoring models to assess credit worthiness and predict default in payment. In the study, credit scoring techniques was applied with the data of payment history of members from a recreational club which had been facing rising subscription fees payment defaults. To identify defaulters, the club used credit scorecard model, logistic regression model and decision tree model, and the performances were compared. The respective error rates are were 27.9%, 28.8% and 28.1%. The study concluded that although no models outperform the other, it is relatively easier to adopt scorecards (Bee Wah , Seng Huat Nor Huselina, 2011). Data mining is the process of extracting useful information from a large database, and this technique has been generally used in many fields, such as banking, finance, telecommunication, manufacturing, healthcare, insurance and others (Bee Wah, Seng Huat Nor Huselina, 2011). As stated by Bee Wah, Seng Huat and Nor Huselina Mohamed Husain, to construct a credit scoring models, data mining techniques is required. They explained useful historical data on payment obtained through data mining could help identify the important the demographic characteristic related to different credit risk and give a score for each borrower. They further suggested some sophisticated data mining software such as ANNs, MARS and SVM . Despite the usefulness of data mining, the study (2011) stated that there are limitations to construct credit scoring models. First, it is subjected to poor quality or unavailability of the data. Secondly, since it is based on historical data, bias occurs when the models are applied to new borrower. Therefore, the study concluded that there is no best model to evaluate credit worthiness, despite of the easy-to-use feature of credit scoring. Credit scoring in developing countries A study is conducted by Thanh Dinh and Stefanie (2007) to identify the borrower characteristics related to default risk and how the use of credit scoring models in the in developing countries such as Vietnam. Thanh Dinh and Stefanie (2007) stated, As the credit market in developing countries such as Vietnam become mature, banks benefit from credit growth, but also face increasing competition of local banks and sophisticated foreign banks. Thus, local banks started to change from relationship banking to transaction banking by adopting credit scoring model (CSM). They further claimed small banks or micro-financiers unable to collect complete and sufficient data of the borrower characteristics and their credit histories to design reliable CSMs. These data include borrowers income, collateral, character, reputation, and standing in the community. Therefore, to overcome the problem, they suggested the banking markets in developing countries must be matured enough to have a large database. Based on the case study in Vietnam, Thanh Dinh and Stefanie (2007) concluded significant advantages can be observed. First and foremost, CSM help to reduce default rate. Also, it helps banks to run risk-based pricing to manage its loan portfolio. Finally, can reduce the time and cost spent by the loan officer on loan assessment. Overall, it increases the competitiveness of the banks. Proposed Credit Scoring Model: Reassign Credit Scoring Model (RCSM) Due to the wave of financial innovation, more and more companies were forming better strategies with the help of credit scoring models. Hence, during the past decades, various credit scoring techniques models have been developed to keep on improving for a better credit approval scheme. (Rong-Ho Chun-Ling, 2009). These models include linear discriminant analysis (LDA), logistic regression (LR), multivariate adaptive regression splines (MARS), classi ¬Ãƒâ€šÃ‚ cation and regression tree (CART), case based reasoning (CBR), and arti ¬Ãƒâ€šÃ‚ cial neural networks (ANNs). A model is proposed by Rong-Ho and Chun-Ling (2009) called reassign credit scoring model (RCSM). As claimed by them, the hybrid model combined ANNs, MARS and CBR approaches to eliminate the Type 1 errors rejecting good credit applicants. They further explained, RCSM model is divided to 2 phase, MARS first obtains input variable, ANNs then classify credit applicants into good or bad credit group ÃÆ' ¢Ãƒ ¢Ã¢â‚¬Å¡Ã‚ ¬Ãƒâ€šÃ‚ ¦Consequently, CBR revaluate the rejected applicants by ANNs by comparing the similarities between rejected cases and CBR database which contains good and bad casesÃÆ' ¢Ãƒ ¢Ã¢â‚¬Å¡Ã‚ ¬Ãƒâ€šÃ‚ ¦ lastly, by looking at the amount of SG(similarities with good database case) and SB(similarities with bad database case), it classifies the rejected applicants into conditional accepted and rejected classes. Figure 1 show the process of RCSM. Rong-Ho and Chun-Ling (2009) stated that the proposed RCSM demonstrated the advantages of MARS, ANNs and CBR, the model eliminates Type I error and increases the approval rate of credit, potentially, increasing banks revenue. Even though the misclassification cost of Type II error (reassigning a rejected customer as good when they are actually bad) was higher than Type I error, the cost could be decreased while increasing Type I errors were eliminated (Rong-Ho Chun-Ling, 2009). Figure 1: The process of RCSM. Source: Rong-Ho, L., Chun-Ling, C. (2009). Constructing a reassigning credit scoring model. Expert Systems with Applications, 36(2), 1685-1694 Recommendation and Conclusion Improve Data Mining Quality To improve data quality, we suggest that the organization need to ensure the imprecise data to be cleaned in order to get more useful and efficient mining. We suggest that they should clean the unnecessary data frequently. They can remove the duplicate records, normalizing the values or number in representing information in the database. For example, no is represented as a 0, cross or sometimes even as a N, X and so on throughout the database. Besides, they also can removing unuseful data fields by identifying anomalous data points. For example, they can remove individuals whose age between 120-130. Moreover, they also can improve the data quality by standardizing data formats to avoid bias concept. For example, all the date is using DD/MM/YYYY format. This can let all the data look more consistence so that they can avoid some error that may face by user. Lastly, they also should carefully verify the information before database entry, make sure the credibility, authenticity and fidel ity of data. In my opinion, they should have some expertises for database checkups regularly to ensure that old data is refreshed and continues to perform wisely for the coming year. Sensitive to the changes of economic As our suggestions, Credit scoring should take consideration to the changes in economic such as inflation or deflation. High inflation could slow down the economic growth and hence credit growth. This is because economic fluctuation might effect the consumer behavior , amount of loan as well as the capacity of individual. Therefore, we suggest that credit scoring should come out with different steps and evaluation based on different economic situation . For example, during recession , the credit score should be more tight and should charge higher interest rate due to the default risk is higher. In short, economic circumstances can be an extra predictive information in credit scoring, in order to forecast the future credit performance of the individual efficiency, bankers should consider both individual and economic situation before making credit evaluation. Improve in diffusion of small business credit scoring in early stage To improve in small business in credit scoring in early stage, we suggest the banking organization to restate the score for each new applicants of credit. This might clear the old data and restructure again scores for accurate determination on eligibility for loan. For example, scores for last month might differ from scores for this month. Besides that, organization should hire a special agent to check the information given is valid or not as the information of applicant might differs in other banks. Other than that, we suggest the CEO of banking organization to have a briefing given by for example Federal Bank, this will ensure any new technology implementation on credit scoring is reliable and this also gives confident and limitation for manager to make decision on credit scoring pertaining issues. The diffusion of small business credit scoring is more likely to develop in future as technology advance makes things easier with less risk. Reassign credit scoring model (RCSM) As reassign credit scoring model (RCSM) might only provide credit manager with numerical scores which is insufficient for the credit manager for explain negative credit decision. Thus, we suggest RCSM to have some explanations in neither accepting nor rejecting the loans. Furthermore, the professionally designed, tested and validated credit scoring model can be expensive for the banking organization to purchase them. Thus, we suggest the reassign credit scoring model to customize the needs of each bank so that mimic the decision making style and the risk tolerance of the companies can be reduced.