Pain points in traditional risk management

Traditional risk management in the financial institutions had largely relied on borrowers' cash flow, collateral, guarantors, existing credit history and offline investigations. However, the practice has proven to be problematic because

Based on the PBOC's public data, less than a quarter of the 1.4 billion population in China has a documented credit history in September 2016 (by contrast, over 90% of the adult population in the U.S. have credit scores).

SME with limited years of operating history (i.e. mom-and-pop stores) usually do not have complete audited financial statements or strong assets to serve as collateral. For example, according to the World Bank, around 70% of assets in small business consist of accounts receivable, inventory and equipment, which are not the traditional collateral accepted by lenders.

The risk profiles of retail clients and institutional clients are ever-changing and thus, financial institutions usually lack timely indicators to predict the likelihood of future defaults.

Due to these reasons, those undocumented individuals and SMEs have long been dismissed by Chinese banking systems and have had trouble accessing personal and business loans. In recent years, however, financial institutions have been much more open to these underpenetrated communities, so as to fuel their loan growth and to compete with rising Internet players. As financial firms still lack the data and technical expertise to make credit decisions for "long-tail" clients, fintech firms have come forth to fix this issue.

AI-driven risk management is powered by multidimensional data. For example, to measure risk profiles for SME, fintech companies collect data -- transactional, operational, tax and procurement -- from central banks, tax authority, customs, and online channels. Channels are more diverse when it comes to assessing consumers, including telecom companies, mobile operators, e-commerce websites, social media platforms and other Internet companies. Even though alternative data appears to be less relevant to borrowers’ creditworthiness than traditional financial data, a large and dynamic data pool that keeps evolving with new data would eventually lead to strong correlations. As companies serve more partners and end users, the platform generates strong network effects for all parties involved. As one of the most mature fintech applications in our view, the market size of intelligent risk management is expected to grow to CNY 7.6 billion (USD 1.1 billion) in China by 2020, with Bairong and Tongdun counting among the primary representatives.

Investing in intelligent, data-driven risk management

By studying the firms that have realized the highest revenue figures and raised the most capital, we think the key factors when picking investment targets as follows.

There are thousands of firms in China that call themselves big data players, whether they have amassed millions or billions of data. The bigger the data pools, the better results the credit assessment models are likely to generate. We believe it is ideal for fintech companies to accumulate in-house data and maintain long-term, stable data partnerships. That way, they can consistently source data from diversified parties and will not experience high volatility if one or two vendors terminate partnerships. Investors can tell this from the contracts and conversations with companies' data suppliers.

Accurate risk prediction is crucial to clients. While open-source ML framework and AI algorithms are widely used, the credit assessment models and credit scores generated vary from company to company. It is inevitable that models sometimes contain errors or flaws that lead to incorrect approvals/denials of loan applications. Fintech companies should frequently refine algorithms so as to increase accuracy rates.

Financial firms usually choose the pre-lending credit assessment as a starting point and are likely to try other products from the same vendor to meet new needs. Top fintech service providers offer a wide array of products to capture cross-selling opportunities. Launching an ecosystem is stronger than selling a standalone product, as financial organizations will find it hard to integrate if they use anti-fraud services, identity verification and overdue debt monitoring supplied by different vendors.

Tongdun Differentiates it as a trusted third-party and a global visionary

Based on the above mentioned investment criteria, we selected Tongdun Technology, a top AI-driven risk management and banking customer acquisition provider headquartered in Hangzhou, Zhejiang. Serving more than 10,000 client companies including around 5,000 financial institutions and around 5,000 Internet companies, it maintains retention rates of around 95%. In our view, the company differentiates itself the rest of the players in the following aspects,

Multiple AI laboratories to focus on fundamental research: Competition in the "AI+Fintech" sector highly depends on the ability to attract key talent. Impressively, out of its 1,300 total employees, Tongdun has recruited more than 100 scientists in its AI laboratories to explore applications of NLP, computer vision, federated learning, etc., which likely leads to product coverage in the hundreds and catalyze even more aggressive innovation.

A trusted third-party provider: Many fintech service providers which used to position themselves as pure technological service providers have been enticed by lucrative lending businesses after gaining more and more user data, but Tongdun sticks to its third-party positioning and does not release their list of prospective customers or directly refer any traffic to financial institutions. By only focusing on the AI-driven accurate marketing, customer segmentation and customer sentiment analytics, we believe that the company will not be a rival to its financial clients or intrude upon user privacy at the same time it engages in lowering regulatory risks.

A global visionary: Tongdun's global vision also sets it apart. We learned that some entrepreneurs and business leaders in Southeast Asian countries visit China with Tongdun as their research and diligence focus. After five-years of operating history in the domestic market, the company began overseas expansion in March 2018 targeting emerging markets with a large population base, low credit card usage but high smartphone penetration (i.e. Indonesia, the Philippines, Vietnam and Thailand). In less than a year and a half, Tongdun has secured contracts with 300 foreign companies in 12 countries. Representative clients are Bank of the Philippine Islands and Shopee, the largest e-commerce platform in Southeast Asia.