OCC looks for data’s context, ‘not just numbers, not just ones and zeros’

At the Office of the Comptroller of the Currency, there’s no job that can’t be handled through better data analysis.

The OCC’s primary function is to ensure that the country’s “banking system operates in a sound and safe manner,” said Katherine Tom, director of Data and Analytical Solutions. And that’s where the most data analysis takes place. The OCC uses bank statement data, loan level data, financial market data, economic and legal data to provide the government with both holistic, macroeconomic views and the ability to dive deep into a particular subject.

But that’s not the only thing the agency uses data for.

“On top of that, we also use data analytics to streamline our internal operations, such as human resource allocation, financial planning, system performance,” Tom said. “Even though that’s not directly related to bank supervision, it’s important to have a strong back office to support our examiners.”

And the agency is always looking to improve its ability to analyze data of any sort. Currently, one issue it has is unstructured data, which is anything that is not organized in a predefined manner. This frequently includes large amounts of text, and is difficult to represent in a database or spreadsheet.

“We have a simpler text searching function in OCC for unstructured data right now, but we are in the middle of getting a more advanced text analytics tool in house,” Tom said.

Tom said OCC uses a wide array of tools like this, ranging from basic business intelligence tools originating in the 1950s and 1960s, to more advanced statistical packages that provide functions like data mining, predictive modeling, simulations, advanced mathematical statistical analysis.

Most of these tools can run off the analysts’ standard Windows PCs. But for more advanced computations and analysis, Tom said OCC has a system specifically designed for that.

“For our complex and large data analytics, we have what we call the high performance computing platform,” Tom said on Data Analytics Month. “That’s on a LINUX operating system. We use parallel processing computing technology to process large data within a small period of time.”

The analysts themselves are embedded in their business lines to help them better understand the specific needs and processes, to understand the context behind the data.

“It’s not just numbers, not just ones and zeros,” Tom told the Federal Drive with Tom Temin. “They have to have business meanings to it.”

Most of this data is either publicly available, or collected through data sharing with other agencies, especially those involved in the Financial Stability Oversight Council.

“There’s no such thing as going to one data source to come up with a holistic report,” Tom said. “It’s always multiple data bases, multiple data sources, merging the data and then look at the results.”

Tom said the first thing analysts at OCC do is to figure out what types of relevant data the agency already has. Then they run basic reports and data distributions to get a wider view and see where a closer look is required.

Then the analysts dig into the data and collaborate on the various reports. The main goal is to answer the question that prompted the analysis, and any “what-ifs” that might arise.

“Data analysis always starts with questions asked,” Tom said. “Questions could be, ‘What are the risky assets these days? How banks are doing in terms of their profit margin, in terms of their loss mitigation. Do they hold enough capital? What is the macrocondition on commercial real estate?’ These are all sorts of the various questions we get and some of those requires you to spit out a report. Some of those are really on the fly analysis, we slice and dice the data in multiple ways, or we put in data models, try to have a prediction, and match the reality to see what’s going on.”

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.