How Graph Technology & AI Helped a Global 50 Bank Lend in Latin America

Challenge

A two-year stint of negative growth in the LatAm region happened due to a number of
factors: changes in Chinese economic activity, a strong U.S. dollar and a sharp decline
in commodity prices. These were compounded with domestic political instability,
macroeconomic fragilities and corruption scandals.

On top of negative growth, the private sector was experiencing lower credit provisions.
Factors included differences in GDP per capita, weaker legal rules to enforce creditor rights
and high information asymmetry due to inadequate debtor data.

Marionete was already working with the Global 50 LatAm bank to help minimize exposure
to credit risk. However, the towering quantity and expanding diversity of data spanning the
bank’s siloed databases presented an extreme challenge.

“Beyond effectively managing the data,” said Ricardo Miranda, Big Data Engineer at
Marionete, “we couldn’t generate the real-time insights necessary to quickly identify and
respond to threats of credit risk from borrowers.”

“If a bank is unable to interpret data quickly to discover the relationship between customer
credit risk, as well as if debt interest is paid on time (if at all), you’re incapable of evaluating
and understanding the exposure level of each and every one of its borrowers,” he added.

The bank’s legacy technology simply couldn’t manage such sweeping amounts of data.

“With various data sources in use, many based on traditional relational database technology
(RDBMS), the volume and diversity of data the bank generated couldn’t be harnessed to
render real-time analytics and recommendations,” Miranda said.

Solution

Marionete addressed the bank’s data challenge by integrating its various databases, including
RDBMS, using the Neo4j graph database. With this level of insight, the bank was equipped to
mitigate credit risk and influence bank charges and interest rates.

“With Neo4j graph technology,” said Miranda, “we were able to gain a deeper understanding
of the bank’s borrowers, such as their relationships with other economic agents like
suppliers, financial intermediaries and customers.”

The digital transformation project was twofold. Marionete created the master data “blueprint”
that integrated all of the bank’s disparate data. Marionete also implemented an extensive inhouse
training program across several of the bank’s divisions.

“With Neo4j installed, we were able to create a data model that relays borrower’s credit
scores,” said Miranda. “At the same time, the graph database solution provided the bank with
a complete view of master data, all made available in real time.”

Now, Neo4j has become the cornerstone of powerful insight into the bank’s data, helping to
reduce credit risk, empower decision making and identify new business opportunities.