Recently, Big Data emerged as the “next new thing” in forensic accounting analytics. Big Data is a popular term for describing the exponential growth and availability of data created by information technology and the Internet. It is an all-encompassing term for any collection of data sets so large and complex that they become difficult to process using traditional data processing applications. Its use is based on a hypothesis that more information leads to more accurate analysis. Its attraction is the additional information available from analysis of a single large set of related data, allowing correlations that are not otherwise easily deduced.
Big Data usually includes data sets with sizes beyond the abilities of commonly used software tools to efficiently and accurately recognize, organize, manage, and process. This is especially applicable for larger complex commercial litigations where its use can hypothetically lead to more accurate analyses that in turn lead to rendering more accurate opinions. And, hopefully, more accurate opinions should lead to less controversial court decisions.
While commercial litigation damages tend to turn on an analysis of presumably objective data sets, there are occasions in which juries are not aware that the data sets used by testifying experts rely on manipulated presentations of their provenance. In other instances, the goal of winning may affect a testifying expert’s manipulation of available data sets.
Rule 61H1-21.002 Integrity and Objectivity of the Florida Rules Governing the practice of public accounting requires that “a certified public accountant shall not knowingly misrepresent facts, and, when engaged in the practice of public accounting, shall not subordinate his/her judgment to others including but not limited to clients, employers or other third parties….”. There are occasions when licensed testifying CPAs do not take this seriously enough.
I was retained to testify as an accounting expert for the husband during a divorce matter litigated in Broward County. The testifying CPA accounting expert for the wife testified that hundreds of cash deposits amounting to many hundreds of thousands of dollars into a personal account of the husband over a three-year period constituted additional unreported income to the husband in determining his income stream. The accounting expert had no objective information on which to base his opinion. Furthermore, he ignored a similar number of cash withdrawals from the husband’s Sub-Chapter S business bank accounts over the same time period and their classification in the business tax returns of the husband’s business as distributions of already taxed income. The cash withdrawals were similar in total to the hundreds of cash deposits into his personal account. When, during my direct testimony, I explained to the court that the spouse’s expert was inappropriately manipulating the data sets involved, his testimony in this area was struck.
The same issues can arise during an economic crime prosecution. We were hired to assist criminal defense counsel with defending a Florida sales tax fraud matter.
The defendant was a dealer of mobile homes. The State of Florida investigator who calculated the state’s alleged economic loss used the records of the Florida Department of Motor Vehicles (DMV) universe of recorded sales and the sales records of the defendant dealer as the state’s basis for arriving at an economic loss of $273,344.97 in unremitted sales taxes plus $253,729.43 in penalties and interest. The prosecutor assigned to the case calculated that if the defendant was convicted at trial, the state’s sentencing guidelines called for a sentence of up to 30 years of incarceration. The data sets involved covered a three-year period. The state’s criminal investigation agent audited 100 percent of the sales during the relevant period. The way in which the sales tax was calculated by the agent was designed to maximize a loss to the State of Florida and ignored the other relevant data contained within the Florida Administrative Code and related State court cases bearing on the matter.
When we completed our analysis and correlated the correct sections of the Florida Administrative Code with the sales records of the DMV, the defendant’s bank records and the defendant’s sales records, the actual loss to the State of Florida amounted to $20,269.20 plus applicable penalties and interest. In this case, had the criminal investigator for the State of Florida correlated all of the available relevant data, the financial cost of the defendant’s defense could have been reduced significantly. After several meetings with the prosecutor, an admission by the state’s investigator that the state’s findings were inaccurate and skillful negotiating by defense counsel Bruce Fleisher, the prosecutor accepted a check in the amount of $20,000 from the defendant and null processed the case.
While the data sets used in these two local litigations seemed large and were vulnerable to unsuitable manipulation they were not Big Data sets and the sets are at even greater risk for exploitation. This is because the reality of Big Data is misunderstood. Public confidence in accounting rests on objectively and truthfully using historically classified numbers to arrive at pictures of numerical flows of information and snapshots of a particular entity or individual. When those numerical data flows are stratified into industry groups and further stratified into sub-groups within each business model type, there is a general belief that those larger data sets (Big Data sets) carry a predictive value.
In the U.S., data is aggregated on local, statewide and national scales. Within each U.S. geographic area, the accounting data is distinct. On a larger regional scale, the data changes significantly. It is accounted for by different economic conditions. On a worldwide scale, differing economic conditions affect the Big Data sets even more deeply. Unless real “on the ground” conditions are factored into how Big Data sets are used, an accounting expert could use them to draw inaccurate similarities between “apples and oranges.”
In a litigation environment, the characteristics of Big Data merit increased caution by the judicial system. Without understanding exactly how relational database management software analyzes Big Data sets, it may not be possible for judges and juries to determine the accuracy of an accounting expert’s results.
South Florida Legal Guide 2015 Edition