A former consumer lending and risk officer at Citibank is launching a business to bring U.S. consumer loans to Asian investors, by connecting peer-to-peer networks using technology developed in China. (UPDATE: her company was later named FinEx Asia.)
Maggie Ng, a two-decade veteran at Citi in Hong Kong, is leveraging technology developed by Shanghai-based P2P platform Dianrong to access four large, U.S.-based consumer-lending networks.
She is also about to raise a seed round of financing for the new, as-yet unnamed business; it should be ready to accept customers in June or July, pending regulatory approval.
She is co-founder along with another person who has yet to formally leave her job as chief technology officer at an asset-management company.
“U.S. lending assets offer high yields, low volatility and resiliency, but there is no vehicle to introduce them to Asian investors,” she told DigFin. “Fintech is a bridge.”
Ng, who left Citi in April as its managing director for cards and unsecured lending for Asia, Europe and the Middle East, says U.S. consumer, credit-card and unsecured loans yield anywhere from 5% to 12% net of fees on an annualized basis. Moreover they are governed by U.S. regulation and established credit scoring from FICO (a private, California-based data analytics company).
This is in contrast to China, which despite its vast and raucous P2P industry, has only nascent efforts at establishing a widely used credit bureau for rating loans, as well as question marks regarding data integrity and security, Ng says.
Leveraging Dianrong’s tech
What China does have, though, is technology.
Ng has set up a holding company, Sunday Finance, through which she has acquired a Hong Kong brokerage and its advisory and investment licenses (though these still await regulatory approval for her to use).
This company is leasing technology from Dianrong, an online marketplace lender. Dianrong’s founder, Soul Htite, is also serving as an advisor to Maggie Ng’s business.
She will use Dianrong’s blockchain to connect to U.S. platforms, for data protection, and to select loans. The low costs of using technology to enable these activities means she can pass savings on to Asian investors.
For now she will not be targeting retail investors, but intermediaries such as mutual-fund and insurance companies could package her loans for broader consumption. She intends to sell her wares to wealthy people and professional investors at family offices and private banks.
She has already signed agreements with four online lenders in the U.S. to connect their loans to Asian investors. For the U.S. groups, she represents a chance to diversify their investor bases.
Marketplace lenders in the U.S. generally charge borrowers far lower interest rates than traditional banks, but they do not have depositors and continue to rely on institutional investors for capital.
Ng’s platform will initially bring Asia-based institutions, but at some point it could expand to allow individuals to access loans, which would help U.S. online lenders diversify their investor base.
She declined to identify partnering P2Ps in the U.S.; the biggest, Lending Club, was itself co-founded by Soul Htite before he moved to Shanghai.
Putting machine learning to work
Two aspects of Dianrong’s technology are operational: APIs to enable it to communicate with software among U.S. online marketplaces, and blockchain for securing data and providing a clear audit trail. A third is a work in progress: machine learning for developing algorithms that build a portfolio.
The A.I. is meant to learn how to avoid loans at higher risk of default. Dianrong engineers are currently training their algos in China using data from a U.S. P2P.
(They are doing this via gradient boosting, a machine-learning technique for regression and classification problems that is often used to develop algorithms operating as a function of a factor’s cost.)
So far the team has already worked out a first generation model for asset allocation, based on correlations that are easy for a machine to record but would evade a human’s scan.
Ng and her sales team will provide a human overlay of credit expertise. More algos for credit risk are in development. The idea is that as the machine learns, its algos will improve, affording investors ever-more precise allocations to loans that best fit their suitability profile.
When asked why, then, a client should invest now instead of waiting for the A.I. to mature, Ng says the initial model will be sufficient, and that net yields averaging 8% are already a pull.
The platform will also use a robo-advisor to determine customer suitability and product needs. For example, a customer in Japan may be happy with a yield of 4% to 5%, along with credit enhancements; an investor in Hong Kong or Taiwan might want double-digit returns. The Dianrong tech can segment customers and create tranches in the portfolio to match their needs, all done automatically and on mobile devices. It can also provide real-time performance reporting.
The analog barrier of KYC regulation remains. However, Ng says that once a customer is onboarded onto her company’s platform, they have access to multiple U.S. lenders, as opposed to signing up to those platforms individually, and filling out the paperwork multiple times.
