Building the Chief Data Officer Toolkit

Building the Chief Data Officer Toolkit Consumer Financial Protection Bureau Chief Data Officer Financial Services Conference March 28, 2017 Linda Powell, CDO CFPB This presentation is being made by Consumer Financial Protection Bureau representatives on behalf of the Bureau. It does not constitute legal interpretation, guidance or advice of the Consumer Financial Protection Bureau. Note: This document was used in support of a live discussion. As such, it does not necessarily express the entirety of that discussion nor the relative emphasis of topics therein.

Overview of the CFPB We educate, enforce, and study  Promote financial education  Conduct rulemaking, supervision and enforcement with respect to the Federal consumer financial laws  Handle consumer complaints and inquiries  Research consumer behavior, and  Monitor financial markets for risks to consumers Data-driven analysis The CFPB is a data-driven agency. Our aim is to use data purposefully, to analyze and distill data to enable informed decision-making in all internal and external functions. 2

What infrastructure do you need to do data driven analysis? Maslow’s /Data Hierarchy of Needs Self actualization Esteem

portal BI Analytics

Love/belonging

Tool kits (Stata)

Where most data

Masterdata

users try to start

Security & QA (data & analysis) safety

Governance & Standards PRA & Privacy

Food and water

Metadata (dictionaries, catalog) Data Platform (servers, warehouse)

DAMA Best Practices

4

Data Governance  The exercise of authority, control, and shared decisionmaking over the management of data assets. (DAMA Dictionary)  Data Governance Structure at the CFPB: 

Information Governance Policy • Defines scope of data and delegation of responsibilities • Empowers the Data Governance Board • Provides guidelines for information intake, centralized storage, and access rights • Provides an overview of Discusses information disclosure



Data Governance Board



Data Intake Group



Data Governance Team 5

Data Architecture The business needs should drive the architecture Questions to ask yourself: • How much data do I expect to have? • How often do I need to update the data? • Do updates need to be real time or next day? • How do I want to access data? • How complex are my data? • Is my data structured or unstructured? • Do I need to see history or do I only need the current data? • How long do I need to keep data?

6

Data Operations  Start with the answers to the architecture questions  Develop intake processes  Develop quality assurance processes 

Ensure data manipulation doesn’t introduce errors



Determine the business needs related to quality • Is precision to $1,000 sufficient or is precision to .0001 necessary?

Evaluate if it is helpful to have a quality rating  Document processes 

 Maintain metadata 7

Business Intelligence and Analytics • Start with the business need • Choose the right tool for the problem • Partner with the business to continuously refine requirements • Continuously iterate and improve 8

Educate The work doesn’t end when you write the policy or document.

9

Working with Internal Stakeholders Understand the Business

 The Industry  Your organization’s role in the industry  Management’s goals and objectives Increase market share  Maximize shareholder profit  Do good in the world  Support an agenda  Meet the organization’s mission  What risks are management sensitive to  Market changes  Economic conditions  Internal issues 

 Speak the same language 

Translate technical jargon to business jargon

10

Talking to the Business  Business lines don’t generally think in terms of data  Many business lines think in terms of processes or applications  What data are needed to make the process or application effective  Help define what is the goal to determine the data inputs needed  Put things in context  How much data is a lot? 1 terabyte of data is about equal to the Library of Congress’ print collection.  Why is my system so slow? We are pushing a watermelon through a garden hose. We cut it up, feed it through, & reassemble it.  Why don’t we just ask for whatever data they have? - Imagine I asked everyone to give me 1 paragraph on their summer vacation. Then, if I tried to compare where people went, the dates they traveled, and who they traveled with – I didn’t ask for what I need so my analysis 11 will be difficult, inaccurate, or impossible.

Working with External Stakeholders

 Use industry accepted data standards to make data exchanges are more efficient  Minimize or eliminate paper exchanges to improve timeliness and efficiency  Minimize the use of physical media to simplify data exchanges and make exchanges more secure

12