While large banks have become heavily reliant on big data when offering loans to small businesses, Zhejiang Tailong Commercial Bank Co Ltd still highlights the importance of on-site investigations to screen qualified applicants for loans.

"Only a very limited amount of data is available on micro and small enterprises. We cannot assess their real conditions simply via big data analytics, so we combine online data with offline investigations," said Ding Lei, director of the center of risk measurement at Tailong Commercial Bank, a city bank specialized in serving small businesses.

Unlike large State-owned commercial lenders that focus their small loan business on micro and small enterprises with a total credit line of up to 10 million yuan ($1.4 million) per borrower, Tailong Commercial Bank targets smaller enterprises with a total credit line below 5 million yuan per borrower.

"It's hard to collect data from these types of small businesses, not to mention that the data including financial statements provided by the owners of these businesses are usually not in accord with industry norms," Ding said.

To uncover more information, client managers at the bank conduct on-site investigations and meet with small business owners, their employees and neighbors. Apart from getting a deeper understanding of their products, collateral and integrity－as well as checking water meters, electricity meters and customs declaration forms－client managers also analyze and cross-check firms' credit reports, transaction records and government data including business registrations, tax payments and any judicial proceedings.

In addition, Tailong Commercial Bank purchases blacklists from commercial credit reporting institutions to weed out businesses with poor credit histories.

This year's Government Work Report said large commercial banks should increase inclusive lending to micro and small businesses by more than 40 percent. As large banks expand their clientele to include high-quality small businesses by offering loans at lower interest rates, smaller banks such as Tailong Commercial Bank have to tap smaller enterprises with higher risks to position themselves differently from larger banks, Ding said.

"We combine human efforts and experiences with financial technologies－such as artificial intelligence, big data and cloud computing－to build different models suitable for different types of businesses and clients and improve the accuracy of our risk evaluation of micro and small enterprises that have never borrowed from banks before," he said.

From the beginning of this year to the end of April, Tailong Commercial Bank issued loans to 1,155 micro and small enterprises that had never borrowed from banks. The number of such businesses increased by 279 from the same period in 2019. The average outstanding balance of these loans was 1.12 million yuan per borrower. Among the total, 86.3 percent of first-time small business borrowers received collateral-free loans.

By the end of April, the bank's nonperforming loan ratio for micro and small enterprises was a mere 1.24 percent.

"Our bank's nonperforming loan ratio kept falling over the past few years as financial technologies and our client screening model played a big role in the improvement of our risk control capacity," Ding said.

"In the past, we used to completely rely on client managers to select qualified clients. Using big data, we have now set a threshold to eliminate low-quality clients automatically. There are people at our bank to review whether the loan application packages and on-site investigation documents submitted by client managers conform to regulations and whether client managers have conducted due diligence. This will improve the authentication and accuracy of the information they acquired."

Drawing on the experience of client managers, the bank established more than 1,000 rules and built a client screening model based on these rules. Only those clients that meet the requirements can be approved for small business loans. Using this model, client managers, whether experienced or not, will be able to control risks using the same standards.

With the adoption of big data and other financial technologies, Tailong Commercial Bank has greatly improved efficiency. Each client manager at the bank is now able to handle about 200 micro and small business clients, up from about 150 in the past. The model will also tell client managers which of their clients have higher risks and require more post-loan visits.