Launch of the Credit Risk Database Philippines Web-based Scoring System (CRDPh System)

To
BANGKO SENTRAL NG PILIPINAS
Subject
MEMORANDUM No. M-2025- 027
The Bangko Sentral rig Pilipinas IBSP) is pleased to announce the launch of the CRDPh System. All BSP-Supervised Financial Institutions Launch of the Credit Risk Database Philippines Web-based Scoring System ICRDPh System) OFFICE OF THE DEPUTY GOVERNOR REGIONAL OPERATIONS AND ADVOCACY SECTOR The Project to establish a credit risk database (CRD)in the Philippines is ajoint initiative of the BSP and the Japan International Cooperation Agency. It aims to enhance access to credit of small and medium enterprises (SMEs) by helping financial institutions (F1s) assess the creditworthiness of their SME borrowers, creating a truly inclusive economy for the country. The Project is participated by 33 pioneer F1s. CRD is a large-scale database of anonymized financial. nori-financial and borrower performance-related (due dates and past due status) data of small and medium enterprises (SME) provided by pioneer F1s from which a statistical credit scoring model was developed. The CRD scoring model can generate the probability of default and credit score toutputs) of an SME borrower in a particular group of similar attributes, As data is anonymized. CRD does riot provide borrower level credit history unlike existing credit registries/bureaus.
3. Phasel of the CRD Project, which started in 2020. involved the construction of database
of SME data. and development and validation of CRD scoring model and its deployment to the pioneer F1s through a standalone CRD scoring tool. The ongoing Phase 2 of the CRD project will support the development of new CRD services and a transition plan toward a sustainable, permanent operation beyond the project-based phase,
4. The CRDPh System is an automated platform designed to streamline the secured
submission and processing of data and retrieval of CRD outputs. The system was developed to shift the standalone CRD scoring tool into a web-based platform,
5. By complementing existing models and tools. the CRDPh System expects to enhance
the F1s' credit riskbased evaluation of SME borrowers. address information asymmetry. improve loan pricing. reduce dependence on collateral. and ultimately support access to finance of SMEs. 6, The use of CRD outputs is subject to the relevant regulatory expectations on credit risk management. To participate in the CRDPh System and for queries and concerns, please contact the CRD Project Implementation Unit at or at (632) 87082186. Attached as Annex A is the list of Frequently Asked Questions on CRD and CRDPh System. For information and guidance. .2^_ July 2025 BERNA ^ I ' MULO PUYAT vernor

Frequently Asked Questions IFAqsj on the
Credit Risk Database (CRDj and Credit Risk Database Philippines Web-based Scoring System ICRDPh System) CFO Prqject
I. What is CRD Project?
The CRD Project is a joint initiative of the Bangko Sentral rig Pilipinas and Japan International Cooperation Agency (JICA) under the Technical Cooperation Program to establish CRD in the Philippines. The Philippines is the first country to adopt CRD outside Japan. The Project expects to enhance the financial institutions' (F1s) risk-based evaluation of creditworthiness of SME borrowers. improve loan pricing, reduce dependence on collateral. and ultimately support access to finance of SMEs. To effective Iy implement the Project, JICA appointed CRD Japan to provide technical assistance to Bangko Sentral,
2. What is Japan's experience in implementing CRD?
CRD Japan was founded in March 2001. From 52. its membership as of 01 April2025. grew to 165, primarily comprising of private F1s, credit guarantee corporations, credit-rating agencies, and government-affiliated F1s. ' The positive impact in Japan's SME financing landscape and guarantee system after the introduction of CRD include growth in the average annual loans granted to SMEs. increase in guaranteed loans without collateral. and charging of guarantee fee became risk-based (from a fixed rate)
3. How will the CRD apply in the Philippines if it has been developed based on the
technology and experience of. apan?
CRD Japan developed its credit scoring model using statistical methodologies and metrics Ie. g. . logistic regression and accuracy ratio IARl) that are widely used and thus. can be adopted in developing the credit scoring model of CRD Philippines. CRD Philippines' model is also built using the database containing data of Philippine SMEs submitted by F1s
4. What is the governance framework for CRD Project?
As part of the Project's governance framework. Joint Coordinating Committee PCC) and Third-party Evaluation Committee are constituted. The JCC is composed of officials from relevant government agencies/institutions. JICA, Embassy of Japan and industry associations. JCC facilitates interQrganizational coordination toward the accomplishment of the Project's key deliverables and targets. The TPEC is an independent body that objective Iy evaluates the validity and appropriateness of the construction and validation of the CRD scoring model. For Phase I. TPEC is chaired by key personnel from Financial Services Agency - Japan, and members include officers/personnel from the Bangko Sentral. selected financial institutions, and University of the Philippines School of Statistics Upon project completion. the JCC and TPEC compositions and guidelines will be reviewed to consider the establishment of permanent bodies for CRD Philippines, to ensure the sustainability of CRD, up hold data integrity. and promote the responsible and consistent use of the scoring model.
Annex A
accessed on 9 May 2025.12:39PM
Page I of 4

CROPhsystem
5. What is CRDPh System?
CRDPh System is a web-based platform that was developed primarily to facilitate transition of CRD from project-based to its sustainable operation in the Philippines. A series of pilot runs were conducted prior to the launch of the CROPh System. Apart from the credit scoring service, it will be enhanced to provide additional services such as comparison of CRD outputs with the F1's internal rating/score. consultancy service for/analysis of F1's SME client, and statistical information. The FAQs. Terms and Conditions, Participant Registration Form. and Data Privacy Notice for the CRDPh System are available in Partic^, ation of F1!s 6, Who are the pioneer F1s and participating F1s? The pioneer F1s refer to F1s which voluntarily shared data for the construction of CRD database and development of the CRD scoring model. On the other hand. participating F1s are F1s which will utilize and access the CRDPh System to generate CRD outputs and periodically submit data, Thus. participating F1s are considered as both data contributors and users of the CRD scoring model. While no information of individual borrower is obtained but the statistical data. participating F1s are responsible in ensuring that all data submissions conform to applicable anonymization standards and data protection regulations. utilizing the prescribed templates set by System Administrators. The CRDPh System. for its part. employs encryption methodology and secure transmission channels to ensure compliance with relevant laws and Bangko Sentral regulations on data governance.
7. Are F1s required to use the CRD scoring model?
No. While the Bangko Sentral may encourage its utilization. the F1's use of the CRD scoring model is on a voluntary basis. as such, the F1's management is responsible for determining the appropriate application and use of the CRD outputs. These are intended as additional information to support credit evaluation and approval process. The use of the CRD outputs is subject to the relevant regulatory expectations on credit risk management under the Bangko Sentral's manual of regulations and other applicable issuances or regulations issued by other relevant supervisory authorities.
Annex A
CRDBe"e, its a. What insight does the CRD provide?
The CRD outputs may be used by F1s to:
a. Assess the capacity of SMEs to repay theirloan; b. Validate or develop their internal rating systems by grouping customers into categories in accordance with the degree of estimated PD and other parameters set by the F1. such as quantitative and/or qualitative factors; and
c. Align pricing decisions/strategies with the borrower's credit risk profile
Once additional services are developed. the CRDPh System will also provide F1s with statistical information on SMEs which will be useful in understanding trends and patterns in the SME sector and formulation of business strategies. Furthermore. the CRDPh System is envisioned to encourage F1s to efficiently manage SME data as they are required to periodically submit SME data to continuously enhance the CRD database.

How will F1s with and without own credit scoring model for SME loans benefit from
CRD?
F1s without credit scoring model for SME loans will primarily benefit from CRD as it will provide a robust credit evaluation tool for risk-based lending based on a database larger than the F1's database. On the other hand. F1s with existing credit scoring model for SME loans can also use the CRD to validate its internal credit rating system. 10, What will be the cost to F1s to participate in the Project or to become members in the future? F1s participating in the CRDPh System will incur cost attributable to their data preparation (e. g. . cost of human resources and/or system in preparing/anonymizing financial. nori-financial and borrower performance data), The use of CRDPh System is currently free-of-charge. In the future, however, guidelines may be issued on the applicable fee structure to sustain the operation of CRDPh System. CPDSc@ring
How does PD work?
PD has a corresponding easy-to-understand credit score which ranges from I to 100, and they are inversely related, A high PD Iconsidered a low score) signifies high credit risk or a high probability that the borrower will fail to repay its loan obligation while a low PD (considered a high score) signifies low credit risk. To generate PD using the CRDPh System. F1s will have to input anonymized financial statement (FS) and nori-financial data of existing or potential SME borrowers.
What is meant by PD and credit score toutputs) of an SME borrower in a particular
group of similarattributes?
SME borrowers with similar attributes in terms of the credit riskindicators in the scoring model. such as geographical area. business type, industry, and certain financial ratio. will have similar PD and credit score. Th's highlights the importance of providing complete information by the F1s to generate more accurate results/CRD outputs, 13, What are the financial indicators used for the CRD scoring model?
Annex A
The financial indicators used as explanatory variables in the scoring model are turnover/turnover period, solvency ratio. profitability ratio. stability ratio. asset/debt terms, solvency - debt service capacity, growth rate (rate of change) of financial indicators, and balance check items (cash flow and cost).
Is it possible to build a robust credit scoring model based on FS of SMEs in the
Philippines?
Prior to the project implementation, JICA conducted a feasibility study for the introduction of CRD in the Philippines. The results of the study disclosed that although the quality of FS is riot necessarily high in the Philippines, the result of the "Data Quality Examination" confirmed that the development of a scoring model using FS data is fairly realistic. Additionally. the use of the data analytical outputs of CRD will be acceptable as they are calculated based on statistical theory of the Law of Large Numbers. The said theory states that. the results of a test on a sample get closer to the average of the whole population as the sample size grows. That is. it becomes more representative of the population as a whole. 2 The Study for the Introduction of Credit Risk Database (CRD) in the Philippines (available in

In this regard, participating F1s need to periodically submit accurate. complete, and truthful financial, non-financial, and performance-related information of SME borrowers to the CRDPh System, utilizing the prescribed templates or any other acceptable means for data encoding in line with the standards set by System Administrators. The submitted data will be utilized to update the CRD database, refine the CRD scoring model, and for other purposes related to the CRD Philippines operations.
15. What are the reasons the FS-based CRD model is considered reliable?
The CRD scoring model is considered reliable in view of the following:
a. The AR of the credit scoring model is at 0.6 which means that it has an acceptable predictive power. The validation also shows that the model maintains an acceptable AR and is stable, b. The financial indicators used in the model. which were calculated based on the available FS. are correlated to the default variable and are statistical Iy significant in predicting default.
c. The test on the discriminatory power of the model indicates that the model can
discriminate/differentiate between defaulted and non-defaulted SME borrowers. ,6. How will the CRD Philippines ensure that its credit scoring model is robust and reliable over time? The data submitted by participating F1s undergo quality checking and cleansing. and independent model validation is undertaken on a periodic basis. to ensure the scoring model's continued relevance. accuracy, and reliability in generating credit scores over time.
17. What are the differences among CRD. Credit Information Commission ICIC). and credit
bureaus?
The table below shows how CRD differs from CIC and credit bureaus:
Annex A
Target
S merit
Data
Collection
CRD
SME borrowers
. Anonymized SME data
(financial. nori-financial. performance-related ) . To continuously enhance the database. data submission is required upon voluntary participation in CRDPh System. Output Unlike CIC and credit bureaus, CRD does not obtain information of individual borrower and does riot provide borrower-level credit history. CRD obtains and generates statistical data, . SME's PD and credit score . Benchmark statistics and other valuable information of the SME sector C, C All borrowers tindividuals and firms Submitting entities must submit all credit data of their borrowers Credit bureaus All borrowers (individuals and firms) Credit information from the credit registry (as special accessing entity I, public records. and lenders through their voluntary submission Specific borrower's credit record. e. g. . loan payment history. and status of credit accounts Specific borrower's credit report and credit score