Law360 is providing free access to its coronavirus coverage to make sure all members of the legal community have accurate information in this time of uncertainty and change. Use the form below to sign up for any of our weekly newsletters. Signing up for any of our section newsletters will opt you in to the weekly Coronavirus briefing.
Sign up for our Banking newsletter
You must correct or enter the following before you can sign up:
Thank You!
Law360 (November 6, 2020, 1:26 PM EST) --
 |
| Virginia Adams |
 |
| Scott Jarrell |
 |
| Ranjana Ramchandran |
The highest percentage of matters and spending are attributable to labor and employment and consumer fraud actions. With such an increase in matters and the virtual environment that we are all working in, it is more important than ever for counsel to leverage all of the tools at their disposal, including technology and, more specifically, data analytics.
It is conventional wisdom that tapping into the power of data holds the mythical key to everything. But what does this mean in practical terms in the context of litigation, when preparing for and resolving potential class actions?
Anticipated class actions may involve disputes of lost wages and benefits in employment litigation, as well as the more recent allegations of discriminatory practices in disbursing COVID-19-related federal funding, such as Paycheck Protection Program loans. Calculation of potential damages is essential preparation even prior to class certification.
These calculations can be used in negotiations with plaintiffs counsel, where they may help mitigate the magnitude of potential class actions. Damages calculations rely on data.
Whether your organization or client is dealing with a small or large class of plaintiffs, developing scalable data models in the initial phases of litigation offers critical advantages for defense attorneys. These data models can quickly and efficiently analyze multiple data sources, automatically apply mathematical formulas, and produce reports that can be customized by the counsel at will, as results support evolving legal strategy.
In this article, we discuss a three-phased approach toward building a good data model and describe how such models can be leveraged in the current climate of class actions to support your legal strategy.
1. Data Model Design
The first step in analyzing data is model design — the who, what, when, where and how of the analysis. This step is critical in creating a scalable, repeatable analysis you can defend. To define the elements of the data model, start with the legal strategy and ask the following questions:
- Who are the plaintiffs, and what is their relationship to you or your client — employee, customer, third party, etc.? Is the strategic focus on one plaintiff, a group or the class?
- What are the allegations and the potential scenarios to analyze — wage and hour, discriminatory practices, etc.?
- When did the allegations take place, and what time period of data needs to be analyzed?
- What are the relevant data sources to be considered? What business systems can provide the relevant data?
- How do you want to calculate damages based on different scenarios, and what are the relevant data sources? Think broadly about potential strategies and sources, and remember to plan for different local, state and federal regulations, when appropriate:
- Determine the math. What are the relevant formulas and variables to examine? Consider, for example, how overtime wages are calculated, how time is classified as work time, leave, etc., and what the relevant pay rates are.
- Define the output and reporting, considering the potential use cases for the model. For example, are tables, narratives or graphs that show lost wages for a group of plaintiffs most appropriate?
- Determine the math. What are the relevant formulas and variables to examine? Consider, for example, how overtime wages are calculated, how time is classified as work time, leave, etc., and what the relevant pay rates are.
2. Data Collection, Model Development and Analysis
Once the data model is designed according to the established framework, it provides a clear direction for the data collection and model development. Model development means translating the design framework into a working analytics solution that can turn data into insights.
As a rule of thumb, the more detailed the design is, the more robust and reusable the model is expected to be. Such models enable straightforward application to new or future class members with minimal to no changes in assumptions and design. The process of data collection and model development typically includes the following steps.
Data Collection and Normalization for Analysis
The first step of the model development process begins with data collection. Quality and exception review are critical elements of this step.
Given that the model may require information from various business systems, this step is typically time- and effort-intensive and requires a good understanding of relevant business data. This step is also required to confirm that the data collected is analyzable in the model.
For example, if reports are the primary source of relevant data sets, those reports may need to be converted to machine-readable form where necessary so the underlying data can be converted into consumable data for the model.
Quality Review and Exception Identification
Depending on how far back in time the data goes, additional factors may need to be considered, such as legacy business systems, lack of certain data fields, or data purges based on record retention policies. As such, a thorough quality review helps to identify data gaps and business rule exceptions — i.e., can we apply the data model to the data we have available, and if we cannot, is there another way to get to the same results?
It is crucial to resolve data gaps and exceptions as they have the potential to skew damages estimations, leading to inaccurate projections.
Resolution of Data Gaps and Exceptions
Data gaps are typically resolved through identification of alternate data sources to replace unavailable data. We often see that transactional business systems do not contain full records of historical data because, as data ages, it is migrated to the organization's archive or official system of record.
We can often find missing data by referring to the archive systems. However, exceptions and nonconforming data elements should be reviewed closely with business systems owners and then normalized on a case-by-case basis.
Development of the Mathematical Model
Once the data is standardized, the model can be developed based on the design established.
Advanced tools and languages like SQL or Alteryx offer analysts with relevant skills the ability to build programmable models in a short span of time. These tools and languages enable translation of the design into executable code that can be both robust enough to handle millions of records and enable reuse for multiple plaintiffs or classes.
Review Results of Model and Revise as Necessary
For specific plaintiffs and outliers, a deep dive into the results helps evaluate the model for accuracy prior to applying it across a full class. In most cases, this step works in tandem with the legal strategy.
For example, you may adjust the assumptions to provide a more realistic or conservative picture of exposure as the strategy dictates.
3. Deployment and Further Expansion
As the model is developed, the results extracted from it will effectively drive and support class certification, extrapolation and settlement:
- Data models developed by the above framework can be applied to a full class quickly and efficiently with very minor rework.
- The model can be refined to effectively account for variations in local and state regulations.
- As discovery and the legal strategy develop, new data sets identified can be integrated into an existing model to accommodate new assumptions.
Complex data analytics damages estimation models help companies think proactively about data modeling in preparation for anticipated litigation. Data models developed in support of legal strategies often help to uncover key data findings that can lead to class reductions during the certification process and highlight a specific range of exposure for the company that may ultimately drive settlement negotiations.
Given the unprecedented amount of funding the Coronavirus Aid, Relief and Economic Security Act and PPP have provided in the U.S., as well the speed in which these funds were disbursed, the anticipated wave of class action litigation will likely require analysis of large data sets through leveraging data models.
Undoubtedly, consistency and oversight of data models will be paramount in these matters. Many companies are already starting to think proactively about data modeling in preparation for anticipated CARES Act and PPP litigation, including the creation of enterprisewide data lineage and process-mapping efforts across different lines of business to reduce potential data gaps and anomalies.
Virginia Adams is a partner, Scott Jarrell is a principal and Ranjana Ramchandran is a senior manager at Ernst & Young LLP.
The opinions expressed are those of the author(s) and do not necessarily reflect the views of the firm, its clients, or Portfolio Media Inc., or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.
[1] Erica Rutner, "The Class Action Fallout from COVID-19: a Proliferation of New Disputes," American Bar Association, April 7, 2020.
For a reprint of this article, please contact reprints@law360.com.
