ICICI Bank will now use satellite data to determine whether or not to give out farm loans in India

• ICICI Bank's new machine learning (ML) algorithms use satellite data and images to determine whether a farmer is creditworthy or not.
• The bank has been training its algorithms for two years and announced the commercial launch today with plans to cover 65,000 villages within the next few months.
• ICICI Bank wants to use the solution to dole out more Kisan Credit Cards (KCCs), which already account for one-third of its rural loan book.

The same satellite data that predicts when the next storm is going hit will now be used by ICICI Bank to determine farmers’ creditworthiness.

One of India’s ‘Big Four’, ICICI Bank has trained its own machine learning (ML) algorithms so that it no longer has to travel to far reaches of the country to physically examine farmland.

“Satellites capture data remotely and accurately that, frankly, traditional methods cannot because of the travel involved and the analogous method,” said ICICI Bank’s executive director Anup Bagchi. According to him, the bank can now approve loans within a couple of days versus the industry-standard of at least 15 days if not longer.

It also brings down the cost of intermediation by removing commute, logistics and other manual aspects that used to go into credit assessment.

The availability of data isn’t the big issue — most it is publicly available online for free. However, using that data to give you the insights that you want is another story. For ICICI Bank, it’s not satellite data alone that will be in play but also other data sets — like information about soil — that are used to run an analysis.

“We have partnered with agri-fintech companies specializing in harnessing space technology and weather information for commercial usage. We’ve worked very closely with them and our internal team to develop 40 parameters,” said Bagchi.

Data is more reliable from a third-party
Weather, imaging, colour, water level and cropping level data sets are only some of the inputs that get fed into ICICI Bank’s algorithm. Because it’s third-party data, it’s more reliable than taking a farmer at his word, according to Bagchi.

“Assessment of credit is better in a decongested way. The data is known,” he said. The data is then analysed using 40 parameters as satellite images from around the world are run through the algorithm. The results from that are then put into the bank’s scoring models to create indices at a district level as well as for individual land.

“This provides an estimate of the past and future agricultural income, the timing of the harvest, and sources of incomes — key inputs for credit assessment,” Bagchi explained.


ICICI Bank wants to give out more Kisan Credit Cards
ICICI Bank spent two years just training its model. Only recently did the bank roll out its products across 500 villages in Maharashtra, Madhya Pradesh and Gujarat.”We have confidence now that it works well, which is why we are now commercialising it” said Bagchi.

The plan is to cover another 25,000 villages within the next month and hit a total of 65,000 villages in the near future.

ICICI Bank claims that its the first bank in India, among only a handful globally, that uses satellite data to power the credit assessment of farmers.

Currently, the bank’s rural loan book stands at ₹58,000 crore and Kisan Credit Cards (KCCs) account for one-third of it. Even though Bagchi was hesitant to promise any numbers, he did say that he expects an uptick in farmers being able to obtain KCCs and bring down costs for the bank.

SEE ALSO:
Without Huawei and ZTE in India, Airtel and Vodafone-Idea's battle against Jio to become tougher

Airtel wants to increase data prices to ₹100 per GB, Sunil Mittal asks subscribers to ‘prepare to pay a lot more’

India's startup founders are opting for money in the bank instead of high valuations as they look to survive this pandemic