IBM Research: Cognitive Credit Risk


We are at the beginning of a revolution in financial services fueled by new technologies in blockchain, cognitive computing, mobile and cloud—collectively referred to simply as FinTech. In this domain, IBM ResearchChina is particularly active in the areas of cognitive credit risk analysis, cognitive compliance as well as blockchain services and solutions.

Cognitive financial services help the financial services industry to address today’s multiple challenges, including declining return on equity, expanding regulatory requirements, relentless fraud and security threats, quickly-shifting customer segments, and growing non-traditional competition from new financial players.

Blockchain is a technology for a new generation of transactional applications that establishes trust, accountability and transparency while streamlining business processes. It is a design pattern made famous by Bitcoin, but its enterprise applications go far beyond.

IBM InnovationCognitive CreditIncreasingly, traditional banking does not adequately provide financial services for people and businesses that lack an initial financial basis (e.g., income statement, loan repayment history) or a proof of identity in the form of a credit bureau entry, national ID, address, etc. Therefore, static scoring approaches based on such hard financial data lead to increasing numbers of non-performing loans and lower numbers of disbursed loans.

IBM research in this area focuses on building credit risk schemes that integrate alternative, or soft, financial data such as social media data, mobile phone usage data (calls, top-ups, mobile payments) as well as online and in-store purchase records. Assessing risk based on alternative financial data is changing the financial problem of scoring into an informational problem. Very large and multi-dimensional social, telco and retail data sets are required to be processed, analyzed and integrated. In order to scale, big data and machine learning technique are applied.

BigCredit is a system for advanced credit risk analysis developed by IBM Research – China. The system uses traditional and non-traditional sources of data for assessing credit worthiness and predicting loan defaults. A mix of machine-learning techniques is applied so that the most predictive variables and according weights are automatically determined as the basis for a dynamic scoring model. The solution is implemented on Apache Spark, a high-performance open-source cluster computing framework that helps to scale the system to fast scorecard generation for millions of consumers.

Blockchain Services and Solutions

With blockchain, we can re-imagine the world’s most fundamental business interactions and open the door to invent new styles of digital interactions. It has the potential to vastly reduce the cost and complexity of cross-enterprise business processes. The distributed ledger makes it easier to create cost-efficient business networks where virtually anything of value can be tracked and traded without requiring a central point of control.

IBM Research – China is developing blockchain solutions for financial services and the Internet of Things. This work extends IBM’s Open Blockchain and the Linux Foundation’s Hyperledger Project. They are tailoring IBM’s blockchain technology offerings to the particular requirements of the local market in Greater China.