Alleged Gender Discrimination in the Financial Services Industry

On Tuesday, March 30, 2010 three women filed a lawsuit against Bank of America Corporation and Merrill Lynch alleging gender discrimination. According to a New York Times article, the suit was filed in the United States District Court in Brooklyn and accused Bank of America and Merrill Lynch of giving male counterparts of the three employees bigger bonuses and better opportunities. The women also said that the companies sought to punish them when they complained about perceived inequalities.

An article appearing on investmentnews.com indicates that the plaintiffs claim they were discriminated against as financial advisers in the opportunities made available to them, including account distributions, pay, and the professional support they were provided. The plaintiffs are seeking injunctive and declaratory relief, an award of back and front pay, and compensatory and punitive damages. The New York Times article also indicates that the complaint also asks the court for class action status.

A spokeswoman for Bank of America, Shirley Norton, denied the allegations, stating: "Bank of America has a strong track record of hiring and developing associates and has been recognized for its success in creating and supporting a diverse and inclusive workplace. We do not tolerate discrimination and discrimination of the type alleged in the complaint violates the bank's policies and values. Bank of America is regularly recognized as one of the top companies for women for its diversity policies."

Allegations of gender discrimination within the financial services industry is not new; there have been numerous cases filed dating back to at least the mid-1990s. While the plaintiffs and defendants have changed, they share a common set of claims. In my experience as an economic and statistical consultant on these matters, the central issues relate to the "production" of financial advisors, however the particular financial institution measures it. Plaintiffs contend that the production of female financial advisors is lower than that of their male counterparts, leading to lower compensation, because of discrimination by the employer. Employers have argued that while the production of some female financial advisors may be below that of their male counterparts, the difference is not attributable to discrimination by the employer.

The issues raised by these claims are difficult to study because they involve factors that are hard to measure. There are several possible non-discriminatory explanations, ranging from attachment to the labor force to differentials in selling and negotiation skills to customer preference. Exploring these possible explanations requires a multi-disciplinary approach to the analysis. The answer to the question cannot be found by a simple comparison of the production of male and female financial advisors, and concluding that any difference must be attributable to gender discrimination by the employer.  The real answer is likely to be found outside of the traditional realm of econometrics.

Why Employment Attorneys Should Care About Statistics

Why should employment attorneys care about statistics? In many cases, a lack of direct evidence or a "smoking gun" may mean that statistical analysis is the only evidence available. Attorneys with an understanding of these statistical analyses and the inferences drawn from these analyses are better positioned to advise their clients on the merits of the matter.

As noted by Bessey, Gilmartin and Stancavage in their paper "A Review of Statistical Books For Use In Employment Discrimination Lawsuits", this need is not satisfied by hiring an expert specializing in the use of statistics in employment discrimination matters. They state "[w]e are not advocating that these experts be replaced by technically sophisticated attorneys. Rather, we are arguing that attorneys will become more effective collaborators if they possess some knowledge of quantitative methods."

Bessey, Gilmartin and Stancavage make an excellent argument for why attorneys should have an understanding of quantitative methods:

"[T]rial attorneys who are knowledgeable about statistical methods are more effective during the pretrial preparation phase of a case, because they can take a more proactive stance when working with labor economists, industrial psychologists, and other experts to plan data collection, data analysis, and rebuttal activities. They understand what proof is needed for their case-in-chief and know that the analyses will be appropriate and provide them with the information they need. They can speculate about what analyses might be presented by the opposing side (and have their own expert carry out parallel analyses prior to rebuttal), and they can anticipate (and therefore prepare for) possible attacks on the analyses that they present. All of these activities lead to a more coherent and effective presentation at trial."

As an economic and statistical consultant specializing in employment issues, I couldn't agree more. While employment attorneys need not be quantitative experts in their own right, a basic familiarity with common statistical concepts and techniques as applied to law will allow them to better serve their clients and to participate in meaningful discussions with their experts.

The Statistics Police

The Statistics Police

I came across an interesting news bit in this month's issue of Wired: "Minister of Truth - The UK's data cop protects the public from lies, damn lies, and statistics". The piece, which appears on page 10, was written by Mathew Honan:

"Did you know that 62 percent of all cited statistics are bogus? OK, we made that up. But after a 2007 poll found that barely a third of all British citizens trust published stats, Parliament formed a math-police squad to investigate. The top cop in the UK Statistics Authority is Richard Alldritt, a expert in how governments fudge numbers."

In his current position, Mr. Alldritt monitors statistics from approximately 200 public agencies. Mr. Alldritt has been quoted as saying that "no set of statistics has has a completely clean bill of health."

It's unlikely that this problem is limited to the 200 UK public agencies being monitored. If the same level of scrutiny was applied to the US, I would bet that we would find a similar situation.

This raises an interesting issue for employment litigation. Assume that a firm is being sued for failing to hire female applicants for a given position. Further assume that an expert conducts a statistical analysis in which the gender characteristics of hired individuals is compared to the gender characteristics of individuals "available" for the position, and that Census data is used to construct the "available" population. Finally, assume that this availability analysis indicates a statistically significant shortfall of female hires for the position(s) in question.

This analysis, and the finding of a statistically significant disparity, may contribute to a finding of gender discrimination. But what if the Census data is wrong? What if the Census data contains errors?

Let me be clear - I am not advocating the abandonment of Census data (or other government data) for any analytical purpose based on the assumption that it may be incorrect. In many cases, Census data is the best available data for the analytical question at hand. I do think, however, that this is an issue that deserves some consideration. I don't know whether the formation of a US Statistics Police is the answer, or whether this would eliminate all data error with 100% accuracy. We, as consumers of data, should educate ourselves about the data we're using in our analyses and question anything that doesn't pass the smell test.

'Dagger Through The Heart of Statistics' in Reduction-In-Force Cases

In a post on February 4, 2010, Jonathan Hyman, a partner at Kohrman Jackson & Krantz and author of "Ohio Employer's Law Blog", discussed the Sixth Circuit's decision in Schoonmaker v. Spartan Graphics Leasing. Mr. Hyman states that this decision "clarifies what a laid-off employee has to prove to establish age discrimination following a reduction in force."

In his post, Mr. Hyman states:

In Schoonmaker v. Spartan Graphics Leasing, the plaintiff claimed that the fact her employer retained younger employees in her position, and that her employer RIFed the two oldest employees, satisfied the "additional evidence" necessary to overcome the employer's economic justification for the RIF. The Sixth Circuit correctly rejected this assertion, and in doing so put a dagger through the heart of the use of bald statistics of small samples in RIF cases".

The Sixth Circuit held that "statistical evidence may satisfy the fourth element in a work force reduction case... [b]ut such a small statistical sample is not probative of discrimination."

This decision highlights a recurring question regarding the use of statistics in employment litigation: what is a small sample size? Statisticians and courts alike have refused to articulate the "magic number" which, if exceeded, transforms a sample from "small" to "large". No such threshold exists. There is a statistical threshold for sample sizes below which statistical analysis is not possible (such as in multiple regression analysis, when the number of explanatory variables exceeds the number of cases being studied), but the boundary between "small" samples and "not small" samples is not discrete.

Sample size also has implications for statistical significance. Ramona Paetzhold and Steven Willborn note that "the presence of statistical significance in small sample sizes should not be discounted, however, because statistical significance is relatively difficult to obtain in low power situations" (Statistics of Discrimination, Chapter 4, p. 47). Paetzhold and Willborn continue:

Courts should not erroneously discount such results [from small sample analyses], either directly or indirectly. For example, courts should not dismiss them as unreliable or less probative because of the small sample size, nor should they substitute their own re-analysis (such as noting that small changes in the numbers would eliminate statistical significance). See, e.g., Murray v. District of Columbia.

...The only reason not to accept statistically significant results based on small samples at face value is that there is evidence that the assumptions underlying the statistical model or process that produced the test are not met... Further, it is unreasonable to assume that small sample size alone could cause the assumptions not to be met. The burden should be on the challenger to demonstrate the manner in which the assumptions of the model have been violated and that the circumstances causing the assumptions to be violated would be likely to suggest statistical significance erroneously.

Paetzhold and Willborn argue that statistically significant analyses based on "small samples" are not unreliable or less probative because of the sample size.

I agree with Mr. Hyman that Schoonmaker has the potential to require plaintiffs to provide more than pure statistics to move forward with discrimination claims in reduction in force matters with "small" sample size. However, I do not think that Schoonmaker has put a dagger through the heart of the use of statistics of small samples in RIF cases. "Small sample size" is amorphously defined and is somewhat of a moving target. It will be interesting to see if other cases accept the Schoonmaker argument, or if they will find that statistical analyses based on "small samples" do in fact have probative value.

 (Mr. Hyman's post can be found here).

Workplace Class Action Litigation: Pro-Active Planning 'Critically Important'

On January 12, 2010, Seyfarth Shaw announced the availability of their Sixth Annual Workplace Class Action Litigation Report.  Information about this report can be found here:

http://www.seyfarth.com/ClassActionReport/

A press release issued by Seyfarth Shaw highlights the increasing number of class action and collective action matters filed, as well as the financial exposure these matters present to employers:

"Since we began publishing this annual report six years ago, both the number of cases filed and the financial exposure that they pose to companies has increased exponentially," said J. Stephen Poor, Chair and Managing Partner of Seyfarth Shaw.  "As plaintiffs' attorneys bring increasingly sophisticated litigation against employers that combine claims under multiple statutes, the financial exposure is only going to become greater for businesses."

Regarding this year's report, Gerald Maatman, Jr., Co-Chair of the firm's Complex Discrimination Litigation Practice Group states:

"One certain conclusion drawn from this year's report is that employment law class action and collective action litigation is becoming ever more sophisticated and will continue to be a source of significant financial exposure to employers well into the future... [G]iven the enormous financial stakes, pro-active planning and legal compliance programs - to get ahead of class action risks - are critically important for businesses in 2010."

Developing an employment litigation risk assessment and management plan is an important component of your overall business strategy.  Minimax Consulting has more than a decade of experience in the statistical analysis of wage and hour issues and employment decisions, including hiring, promotion, termination, and compensation.  If you are interested in incorporating a formal statistical analysis as part of your compliance and litigation risk management program, please contact me for a consultation.

Synopsis of "Manager Race and the Race of New Hires"

I recently came across an article in the Journal of Labor Economics that examines the relationship between manager race and the race of new hires.  The central question of this analysis is whether or not the race or ethnicity of the hiring manager a determinant of the racial and ethnic composition of new hires.  The authors of the article - Laura Giuliano, David Levine, and Jonathan Leonard - examine this question using "personnel data from a large U.S. retail firm".  Based on their analysis, the authors conclude:

• Non-black managers (i.e., whites, Hispanics, and Asians) hire more whites and fewer blacks than do black managers;
• Hispanic managers hire more Hispanics and fewer whites than white managers in locations where Hispanics make up at least 30% of the population.

Three primary explanations are offered by the authors.  First, managers may use racially segregated social networks when recruiting applicants. To test for managerial hiring networks, the authors "assume that there is some congruence between social networks and residential proximity, and we ask whether managers tend to hire employees who live near them".

Second, if manager-employee similarity improves employee productivity, managers may hire racially similar employees for efficiency purposes (referred to as 'statistical' discrimination).  Finally, both managers and employees may engage in 'taste-based' discrimination (indulging personal preferences).  The authors note that these explanations are not mutually exclusive and "it is difficult to distinguish between them empirically".

This analysis may have potential implications for the use of statistical analysis in race discrimination litigation.  If the authors' conclusions are correct and one cannot differentiate empirically between 'statistical' and 'taste-based' discrimination, one could not determine empirically whether the observed racial patters are attributable to "efficiencies" or personal preferences.

Fisher's Exact Test: Alternative to 80% Rule

Considerable time has been spent discussing the shortcomings of the 80% Rule. We know:

1. The 80% Rule is insensitive to sample size, which can lead to Type I and Type II errors
2. The 80% Rule is insensitive to the size of the disparity
3. The 80% Rule is highly sensitive to the framing of the question
4. The 80% Rule is subjective and lacks a statistical basis

What alternatives are available? For "employment decisions" questions (i.e., hiring, promotion, termination, etc.) one alternative is the Fisher's Exact Test.

The Fisher's Exact Test is a test of statistical significance used to study categorical data when employees are classified two different ways, such as (1) "protected" or "non-protected" and (2) "selected" or "not selected". Assume that we have a population of 100 and that this population is one-half "protected" and one-half "non-protected". Further assume that of the 50 "protected" individuals, 10 are "selected" and of the 50 "non-protected" individuals, 30 are "selected":



The underlying assumption is that under a "protected status-neutral" selection process, the selection rates for "protected" and "non-protected" individuals would be the same. In this example, the overall selection rate is 40%. We see, however, that the protected selection rate is 20% and the non-protected selection rate is 60%. Can we infer from this that the selection process was not "protected status-neutral"?

To answer this question, we can use the the Fisher's Exact Test to compare the actual and expected selections of "protected" individuals. Under a "protected status-neutral" selection process, we would expect that the selection rate among "protected" individuals would be 40%. In other words, the likelihood of being selected, regardless of protected status, is 40%. We statistically compare the expected 40% selection rate with the 20% actual selection rate via the Fisher's Exact Test, and find that there is a shortfall of 10 "protected" individuals selected (20 expected selections minus 10 actual selections). This shortfall is statistically significant at 3.77 units of standard deviation.

The Fisher's Exact Test is based on exact probabilities from the hypergeometric distribution, rather than relying on large sample approximations as in the Chi Square test.  This makes the Fisher's Exact Test particularly useful for small samples.