Mitsubishi Research Institute, Inc. Launches Joint Initiative for Mortgage Fraud Detection

Mitsubishi Research Institute, Inc. (MRI) (President and CEO: Kenji Yabuta) announced that on July 1, it has commenced a collaborative initiative to enhance mortgage fraud detection with financial institutions through the Mortgage Loan Data Consortium (the Consortium *), operated by MRI. The Consortium will pool data on fraudulent cases held by its member financial institutions to promote measures for the deterrence and early detection of mortgage fraud.

1. Background

In recent years, the importance of preventing fraudulent use in mortgage operations has been increasing. Fraudulent use refers to acts such as using loans intended for owner-occupied housing to purchase investment properties, or attempting to illegally pass the screening process by inflating property prices or falsifying income documents.

These fraudulent activities are limited in number for any single financial institution, making it difficult to grasp the overall picture or detect early signs. In light of this situation, there is a growing need for initiatives where multiple financial institutions share knowledge and data to identify the characteristics and trends of fraud.

Leveraging the expertise gained from building and operating the Consortium, MRI has called upon participating regional banks, online banks, and credit guarantee companies to launch a joint effort for mortgage fraud detection. By combining MRI's expertise in analyzing and operating a joint database for financial institutions nationwide with the practical knowledge of participating financial institutions, the aim is to enhance fraud countermeasures in mortgage operations.

2. Overview and Features of This Initiative

This initiative aims to enhance analysis by aggregating data on fraudulent cases, which are limited in handling by individual financial institutions, to extract characteristics highly correlated with fraud, under appropriate management based on various laws and contracts.

In a trial conducted with some financial institutions from July 2025, we identified case attributes (age group, purpose of funds, etc.) highly correlated with fraud and built an AI-powered fraud detection model. The model has confirmed high accuracy, with a majority of fraudulent cases concentrated among those with high scores, and its effectiveness for practical use, such as pre-screening cases with a high probability of inaccuracy, has been verified.

Based on these results, we will begin providing analysis reports and risk scores to Consortium member financial institutions that have provided fraud data.

Specifically, as shown in the figure:

1 Provision of visualized fraud trends and analysis results of regional characteristics

2 Presentation of fraud risk scores for each case

3 Sharing of the latest fraud tactics and cases (key countermeasures)

4 Sharing of information to raise awareness regarding fraudulent use (e.g., information on vendors)

This will enable participants to grasp fraud trends and tactics that were difficult to detect on their own, and also allows for post-mortem reviews and revision of countermeasures using fraud risk scores for individual cases. Furthermore, by sharing information to raise awareness about fraudulent use, it complements the knowledge of other banks, leading to fraud prevention across regions and business types.

Implementation Details of Joint Fraud Detection

Note: Service details are subject to change.

Created by Mitsubishi Research Institute, Inc.

3. Future Plans

Through this initiative, MRI will provide an environment for participating financial institutions to enhance their fraud countermeasures and support the strengthening of risk management in mortgage operations.

Following the rollout of the above initiative, we plan to offer a fraud detection API service that can determine the inaccuracy rate in real-time upon receiving a case, and a service to streamline the screening of cases determined to have a relatively high fraud risk.

We will continue to expand our fraud countermeasure services tailored to practical needs, incorporating feedback and requests from participating financial institutions.

* Mortgage Loan Data Consortium <https://www.mri.co.jp/service/mortgage-data-consortium.html>

This is a joint database operated by MRI since 2010, accumulating various information such as application details, credit information, default records, prepayment records, and interest/guarantee fees for mortgage applicants. Approximately 40 financial institutions are members, and based on over 5 million data points, it provides high-value information such as benchmark analysis based on nationwide data.