Ensuring Data Completeness and IntegrityBY MATTHEW JACOBSEN

Achieve Better Measurement, Performance, and Member Service

As data storage, computation, and analysis technology continues to advance, many financial institutions continue to leverage this evolution to improve performance and customer service. At a minimum, this has included better overall analysis to improve risk/return measurements. Beyond the deeper overall risk/return analysis, many institutions have realized that through better data mining and analysis, they can even become better predictors in such areas as member product/service preferences, price point adoption, and credit performance at a more finite level to name a few. Although many credit unions will not have the budget to obtain the level of data mining and analysis available at large banks, the continuing technological evolution has made a greater level of analysis much more affordable for any size institution. In fact, in today’s world, increasing an institutions data analysis in certain areas is essential to remaining competitive in addition to meeting ever-changing industry and regulatory requirements.

Given the above, what should a credit union be thinking about in preparation for taking the next step toward greater use of their data? Two areas, to begin with, should be performing an assessment of the credit union’s data completeness (including potential future needs) and data integrity. In fact, this should really be a standard practice that is done on a periodic basis. In performing asset and liability management analysis, modeling, and consulting, we work with a lot of credit union data as it relates to these functions. Periodically, we see deficiencies in data completeness for the expected analysis or data integrity issues in loan, deposit, and investment files. Although we are very adept at cleaning, fixing, and appending additional data to data files, this is not always possible nor the best solution. So what should a data completeness and integrity assessment include?

Data element completeness

To begin with, the first assessment of data element completeness should consider all possible needs in the different areas of the institution. One subset of this assessment could be financial reporting/ analysis, operations, and compliance. Another subset to consider could be products/services, member relationship support and analysis, regulatory requirements, financial analysis (including risk analysis), and financial performance.

The next step of this assessment should include not only the current data element needs but potential future needs. For example, this assessment should be done for those institutions that foresee increasing the depth of their loan and credit analytics on the horizon with the goal of increasing competitiveness and/or performance. Even if this is not on the credit union’s horizon, CECL (for example) is on the horizon and an assessment of data element needs should be done with loan data at a minimum for this purpose. In addition, the credit union should consider current and future processes to a refined enough degree and in their entirety. For example, a credit union may be able to produce historical loan charge-off data elements, but can these be produced by loan type and charge-off date? Can historical recoveries be identified with specific loan types?

Following the needs assessment, the next step would be a review of all the data elements being captured. This means across all systems and processes where data elements are captured including those of an outsourced service provider (ex. credit card processor). In addition, this should also include data element capture in loan participations, particularly where the credit union is not the service provider. This review should also identify how long each data element is retained, how is it stored (including its ease of access), and is it backed up. Once you have identified all of the data elements being captured, how does it match up with your needs?

Data Integrity

According to BusinessDictionary.com, data integrity is defined as “The accuracy and consistency of stored data, indicated by an absence of any alteration in data between two updates of a data record. Data integrity is imposed within a database at its design stage through the use of standard rules and procedures, and is maintained through the use of error checking and validation routines.”

In addition to having a complete set of data elements, the data needs to meet a number of expectations to address data integrity. For our purposes, we focus on only three of those expectations in this article. These are accuracy, consistency, and being able to identify the capture time of a data element. These expectations can be ensured by sound procedural and system controls. If manual input processes are involved, are the procedures themselves accurate, consistently followed, reviewed for revisions, and are operators periodically checked to ensure accuracy and consistency? Are appropriate system controls established where possible and applicable? For example, are data fields correctly identified as optional or required. At times we will see certain relevant data fields randomly populated that are used for ALM analysis and as such are clearly identified as optional in the system. In certain cases, we can employ “cleansing” methods to accurately populate the empty fields, but at other times we cannot. Another example is the use of validation rules where possible to ensure data accuracy. There are numerous possible types and combinations of validation rules that can be incorporated to ensure data capture accuracy, consistency, and time identification of the data element capture.

Finally, after all of this is done, what are the benefits to my credit union?

  1. More accurate and/or in-depth analysis including ALM/IRR modeling and CECL/loan analytics modeling results.
  2. Improved decision making and performance by having a complete and more accurate picture.
  3. Improved competitiveness and member service.
  4. Being able to meet industry and/or regulatory requirements.

Periodically going through this process is a strong practice, particularly as the credit union evolves and the business environment changes. Two areas of change in the industry are the upcoming CECL requirement and the strive at institutions for a higher level of data analysis “big data”, particularly with loan analytics.

What if I’ve done these assessments and have gaps in my data for CECL or broader loan analytics? Don’t worry, we’ve got answers for those gaps. At Mark H. Smith, Inc. we understand the needs of the credit union and will be offering affordable loan analytic and CECL solutions to help our clients succeed. We’re here to help!