According to a recent Experian Data Quality study, 90% of financial institutions believe increasing regulation has driven their need for better data analytics and management. So how do you boost data accuracy - especially when it comes to reporting quality data to the bureaus? This deck reveals best practices, as well as solutions to consider when striving to improve your data reporting.
Every year, Experian Data Quality releases a research report that benchmarks data management trends. This year, we found that:
90% of financial institutions in the US believe increasing regulation has driven their need for better data analytics and management
76% of financial institutions in the US believe inaccurate data is undermining their ability to provide an excellent customer experience
29% of US companies feel an inadequate data strategy contributes to a lack of data accuracy
And 28% of financial institutions suffer from regulatory reporting challenges in the US
As you can see here and as you are all aware, accuracy in data reporting is a huge challenge.
CU Debate: The CFPB issued a bulletin Wednesday warning banks and credit unions that if they fail to meet accuracy obligations when reporting negative account histories to credit reporting companies, the result could be bureau action.
Auto lender failed to provide accurate, positive credit information that it promised consumers it would supply to the credit reporting companies. The CFPB’s investigation found that the companies inaccurately reported information for more than 84,000 accounts on a widespread and systemic basis. The CFPB is ordering the companies to cease their illegal activities and pay a $6,465,000 civil penalty
Examples of payday proposal: #1 A full-payment test that would require lenders to determine whether borrowers can afford to make each repayment on time and still cover basic living expenses #2 A provision that would enable consumers to borrow a short-term loan up to $500 without requiring a full-repayment test. Lenders could not offer the option to consumers with outstanding short-term or balloon-payment loans.
CFPB announced it is accepting complaints from consumers encountering problems with loans from online marketplace lenders. The Bureau is also releasing a consumer bulletin that provides an overview of marketplace lending and outlines tips for consumers who are considering taking out loans from these types of lenders
FCRA Section 623
Provide accurate information
Prohibited from reporting information with known errors
Provide notice of a dispute
Duty to investigate and respond to a dispute
Reg V:
Furnishers must implement written policies and procedures ensuring accuracy and integrity of information delivered to the credit bureau
Policies and procedures must be appropriate relative to furnisher’s size, complexity and nature of business
Policies and procedures must be reviewed periodically ensuring continued effectiveness
I want to take you through some feedback we received from someone who went through a CFPB exam and lived to tell about it. Previous to his current role in Experian’s Global Consulting Practice, he held the role of vice president of internal audit at a top ten national bank, and among his responsibilities was managing regulatory exams on behalf of the bank.
Having gone through many regulatory exams in the past, they thought they were prepared and knew what to expect. They were ready to provide policies and procedures related to how they report their data and manage and answer their disputes. They had robust monitoring of their metrics in place. They expected to provide samples of reporting and disputes.
What they encountered was very different from what they expected.
The process, at least for large institutions, is to receive an entrance memo, followed by fact finding questionnaire which is incredibly detailed.
Not only were they surprised at the level of detail being requested, such as resumes and org charts, board meeting minutes, data ownership and controls, and training, they were very surprised at the approach, and the extent of the information being asked and in very specific detail. Instead of throwing out wide net and narrowing in as other regulators had in the past, the CFPB had specific targets they go for.
In this person’s opinion, the manner in which the CFPB gathered their initial information did not take into account the way the bank was organized. The specifics of the requests and the way they are formatted created an immediate conflict because the CFPB made assumptions based on previous exams that this bank was organized in the same way as another bank they had examined. They would ask for information in a manner that the bank could not easily produce and so they had to pull in resources to create ad hoc reports and information to meet the needs and requirements of the CFPB.
Once data is provided back, no one is prepared. Responses are quick, and turnaround is expected to be quick. They don’t want to hear you don’t have the information. And here was a key point: once they found a couple of errors in the data, they wanted to go through every account to provide proof. If the data is not good up front, they will pull the thread all the way through.
They tend to have high demands and unrealistic expectations. A key takeaway from the conversation was that if your data is good from the start, they will not pull that thread.
Speak to Partnership with Experian: Credit reporting is our core competency.
Experian Pandora: Dashboard allows to easily track and measure results - Thresholds and alerts can be customized to your specific business rules - Validate and assess improvements or identify new areas of opportunity
Functionality : Ability to upload dispute information and to monitor and compare against Metro 2® submission - Identify root cause of errors
Results driven: Identify and investigate relevant data, assess and improve data quality, and control this process over time.
Regardless of which data quality methodology you choose, typically the first stage is to assess you data holistically, this is where analysts can view 100s of metadata attributes to identify inaccuracies such as
Completeness of record
Formats of values
Datatypes
Null analysis as well as look at min and max values
One of the most important aspects to this method of analysis is interactivity with the data, for anything that doesn’t look right, you should be able to dig into the data itself and see where these violations are occuring
With this you can identify issues immediately as well as generate reports that can be used to highlight data for remediation such as the example here where we find alphabetic characters within data fields.
This is an interactive approach with subject matter experts who are familiar with the data who can spot trends or values that are not fit for purpose. The main aim of this exercise is to identify critical data attributes that can be measured and monitored on an ongoing basis
We can take this a step further and look at outlier reports too, this is where we measure the trend in the values for each column and flag up any records that deviate from the norm. Using this standard deviation technique we can highlight abnormalities where attributes like SSN, account numbers, date and monitory values don’t follow the accepted trend of the data.
Examples of this are:
Unusually rare formats or values
Unexpected frequent values or duplicates where there shouldn't be any
Or extremely long or short values
Essentially, identifying these issues is the first step but the goal is to produce a dashboard of results, here you can see checks that look at the accuracy of the data compared to the definitions of Metro2 guidelines and we have layered in addition checks for illogical conditions.
This can be as simple as field validation, such as Account status codes being of the appropriate length, datatype and that the values correspond to a list of expected and valid codes
Or this can be a little more complicated by assessing the Amount past due values, performing calculations on the fields and comparing them to the expected status codes
When monitoring these conditions it is key to be able to perform root cause analysis by investigating rows that have failed as well as compare the results over time to identify any trending data
This robust data monitoring essentially automates the process in a controlled and governed fashion, the idea being that we don’t want to re-invent the wheel every time but we do want to build on the knowledge and insights we gather and formalize this in an ongoing process
And finally, Alert systems should be in place so that during any error or failure, the appropriate individual or team is notified, and dependent on your organizations process, a roster can be created so that if a notification hasn’t been acknowledged, then subsequent notifications are sent to the next person or group on the list, this way we are ensuring governance and a process has been put in place so that the people, process and technology are working hand in hand to ensure the accuracy of data and that compliance is being met