Introduction

In cases involving employment discrimination, claims such as failure to promote or wrongful termination, economists are asked to calculate the difference between what the plaintiff(s) would have earned had the alleged discriminatory act not occurred offset by what the plaintiff is now expected to earn given that the alleged discriminatory action did occur. The economic losses are based upon the plaintiff’s past and future income specific to the employer who committed the alleged discriminatory act. By contrast, in cases involving personal injury and a permanent reduction in the plaintiff’s earnings capacity, the economic losses are usually based upon the plaintiff’s entire worklife expectancy, regardless of where the plaintiff would be working.

In discrimination cases, if there is some reason to believe that the plaintiff would not have worked for the defendant for the rest of his career, then the defendant may be liable for the plaintiff’s lost income only during the time that the plaintiff could reasonably be expected to work for the defendant. As such, the forensic economist must consider how long the plaintiff would have worked with that employer. When reliable data are available from the employer, the economist is able to examine the attrition rates or typical tenure of similarly-situated employees. The purpose of this paper is to provide a hypothetical case study to illustrate the methods and data considerations for cases involving the estimation of the plaintiff’s tenure with an employer.

The paper is organized as follows: Section II presents a brief discussion of the literature on firm-specific employee turnover. Section III describes the assumed characteristics of the employer and the employee in the case study. Using the case study, Sections IV and V present a discussion of the methodological and data considerations in the calculation of projected earnings with the employer. Section VI describes the methodology and data considerations for calculating mitigating earnings, while concluding comments are provided in Section VII.

Forensic Economics Literature on Firm-Specific Employee Turnover

Franz (1990) provides an overview of factors that should be considered indamage calculations for wrongful termination litigations. First he outlines the potentially relevant components of economic losses for a plaintiff. These components include:

• Back Pay
• Front Pay
• Fringe Benefits
• Prejudgment Interest
• Attorney Fees and Costs

Second, he points to the “make whole” standard of relief as a guide for computing damages (Tobias, 1987). This standard calls for economic losses to be offset or mitigated by the pay and fringe benefits that could have reasonably been obtained from alternative employment.

Franz points out that an important, and potentially challenging, step in computing back pay is the determination of the end date associated with the back pay period. Unless there is some kind of fixed term contract in place, the end date for back pay may not be obvious. According to Franz, the back pay period can be terminated by court decree or date of trial, or any of the following events: death, retirement, illness, conviction for a crime, voluntary separation, and an offer of reinstatement or other action that would be reasonable cause for discharge such as a reduction in force.

Each of the above factors is relevant to determining the end date for the front pay period. While Franz does not make specific suggestions as to how incorporate these factors into back and front pay calculations, he does make it clear that due to these factors it is “not advisable to compute front pay for young persons for the rest of their remaining work life…”. (p. 39) The possibil­ity of job turnover or attrition makes assumptions of this kind especially prob­lematic.

Trout (1995) specifically addresses the issue of job turnover in the calcula­tion of damages in wrongful termination litigation. He emphasizes the importance of considering the likelihood that, but for the allegedly wrongful termina­tion, the plaintiff employee would have remained working (or not remained working) for the defendant employer up until retirement. While a plaintiff’s attorney may be inclined to argue that their client would have continued to work with the defendant firm up until retirement, Trout shows that, on aver­age, it is not very likely that an individual will work with one particular em­ployer for a very long time.

The centerpiece of Trout’s paper is an econometric model of job turnover. He models the probability that, for a given employee, the current year em­ployer is identical to the prior year employer as a function of income, age, edu­cation, and employer tenure. The model is estimated using logistic regression methods and data from the Current Population Survey (CPS). Trout then uses the model and estimated coefficients to generate predicted attrition rates that are used in example economic loss calculations associated with wrongful ter­mination litigation.

Trout correctly incorporates two key requirements in the application of at­trition rates to an economic loss forecast. First, he appropriately compounds the attrition probabilities over the calculated back pay and front pay periods. Consider an example. Assume there is a 3% probability that an individual will leave their employer in any given year. What is the probability that the indi­vidual will still be with the employer at the end of year one? The answer is (1­.03), or 97%. What is the probability that the individual will still be with the employer at the end of year two? The answer is (1-.03)*(1-.03), or 94.1%, not 97%. Ignoring the effect of compounding will inappropriately understate the impact of job-turnover and thus overstate the value of the economic losses.

Second, Trout appropriately attempts to apply attrition rates to the plain­tiff damage calculations that are derived from data on similarly-situated employees. His econometric model controls for differences in a variety of personal characteristics (such as age) in the calculation of attrition rates. Attrition rates vary across individual, job and industry characteristics. The more of these fac­tors that can be controlled in the calculation the more reliable the calculations will be. As Trout points out, his ability to control for many potentially relevant factors is limited by the quality of the available data.

As an alternative to estimating attrition rates based on aggregate data, fo­rensic economists may calculate attrition rates based on the defendant’s own records. When the data are of sufficient quality and quantity, these firm-spe­cific attrition rates are preferred to the estimated attrition rates based on aggregate data. In the next section we develop a detailed case study that takes advantage of firm-specific attrition data in the calculation of economic losses in wrongful termination litigation.

Case Study Background

The point of departure for the analysis presented in this paper is a hypothetical case study of a single-plaintiff wrongful termination case, where estimations are based on the assumption that liability has already been established. Full information regarding the plaintiff’s pre-termination and post-termination career paths and earnings history is assumed to be available. Similarly, defendant-specific attrition rate information is assumed to be available and reliable.¹

The fictional plaintiff considered in this case study is assumed to have similar qualifications for the job as his/her peers, to have worked full years with defendant, and to have been wrongfully terminated at the end of the seventh year of employment. The plaintiff is also assumed to work full years in alternative employment, which began immediately after the incident with defendant. The assumption of full years simplifies the calculations and prevents partial years from distracting the reader from the focus of the analysis.

The career path for employees of defendant in the occupation of our hypothetical plaintiff is based on a seniority progression composed of four levels, namely, Level 1 through Level 4. The employees who are believed to be similarly-situated to the plaintiff in this hypothetical case are hired at the entry level position (Level 1) and, generally, employees in this occupation with defendant have been promoted to Level 2 at approximately the beginning of the fifth full year of employment, to Level 3 around the beginning of the 10th year of seniority and to Level 4 at about the beginning of the 15th year of employment. Promotions to Level 2 through Level 4, however, are not automatic, but based on successful performance in the prior attained level(s). Retirement eligibility with defendant is granted at the end of the 30th year of service.²

Raises are given to employees at the beginning of each employment year and are based on cost of living changes and the financial performance of the organization. Non-promotional raises of plaintiff averaged 3.23%. Promotional raises, however, are pre-established as company policy to be 8%. Combining promotional and non-promotional raises yields an average growth rate of plaintiff with defendant in the order of 4%. The historical earnings of our hypothetical plaintiff with our defendant are presented in Table 1.

The plaintiff’s post-termination employment is assumed to be similar innature and to require similar qualifications as the job with defendant. The growth rate of earnings with the alternative employer is assumed to be similar to the average of all raises with defendant (promotional and non-promotional raises combined). However, earnings levels with the alternative employer are assumed to be lower than with defendant.³

In order to complete the information set necessary for the estimation of economic losses, we have assumed the back pay interest rate, the front pay discount rate, and the unemployment rate to be constant throughout time. The interest and discount rates are set to be 6%, while the occupation and location specific unemployment rate is set to be 7% for all years.⁴

There are at least two broad categories of methods that can be used to generate estimates for the relevant variables specific to the wrongfully terminated employees. These relevant variables include factors such as promotion indicators, expected earnings, retention rates, and mitigating earnings. The two categories of estimation methods include:

• Taking simple averages over a group of similarly-situated employees
• Specifying and estimating stochastic models

The first approach has two things going in its favor. First, it makes relatively modest demands on the available data. Second, it is much easier to explain to non-technicians. Econometrically sophisticated methods such as survival analysis (for estimating and predicting retention rates) are often either not feasible (due to limited data) or unadvisable (due to the need to convey the analysis in simple terms). For these reasons, we have focused on the more straightforward methods in our case study.

There is at least one other simplicity-complexity trade-off that a potential expert will consider when generating these kind of damage estimates. What is the appropriate method to forecast the relevant variables into the front-pay period? Again, there are two broad categories of forecasting options. First, one could use relatively simple methods of extrapolation that rely simply on historical values of the variable in question. Alternatively, one could use more sophisticated econometric models estimated over the historical period for the variable in question to forecast future values. The choice between extrapolation and forecasting based on a stochastic model is similar in nature to the choice that all forensic economists make when deciding how to project earnings streams, work life expectancy, and discount rates. While we recognize the possibility that stochastic modeling may in some cases enhance the attrition estimates, this paper focuses on the extrapolation technique, which in our experience is the most common of the two approaches used by forensic economists to estimate economic losses.

The Attrition Rate

In cases involving the estimation of economic losses associated with wrong­ful employment actions, the calculation of projected earnings with the defen­dant incorporates assumptions regarding the expected growth rate of earningsand the date at which losses should stop. When such information is available,the most common approach in the estimation of projected earnings incorpo­rates the plaintiff’s pre-incident history of earnings and organization-specificinformation regarding salary increase policies. With regards to the time when losses stop, it is common to see analyses that project such expected earningswith defendant either (1) through an assumed time when alternative employ­ment earnings reach parity with expected earnings with defendant, (2) through the expected retirement age of plaintiff, or (3) a date determined by the judge or jury.

Projections of earnings with defendant that extend either through the timeof mitigating earnings parity or through the plaintiff’s expected retirement age rely on the underlying assumption that retirement is the only factor affectingthe time span of plaintiff’s likelihood of continued employment with defen­dant.⁵ However, factors such as the possibility of a future lay-off, or a plaintiff’s decision to voluntarily leave, among others, could affect such likelihood. When reliable organization-specific information is available that would allow the es­timation of the likelihood of continued employment with the defendant at eachpoint in time, such information should be incorporated into the projections ofplaintiff’s earnings. Not incorporating probabilities of continued employment(or a proxy for such) in the estimation of projected earnings with defendant willlikely lead to overestimation of economic losses.

In our hypothetical case, information regarding the likelihood that employ­ees similarly-situated to the plaintiff will remain employed with defendant each year is assumed to be available. It is important at this point to highlightthe meaning of “similarly-situated” in the context of our case study. In thiscase, similarly-situated employees are those employees of defendant who pos­ses similar qualifications and experience and who entered a particular level ofthe set career path at a similar point in time.⁶

Once the definition of “similarly-situated” has been established, the calculation of the attrition rate has two components. The first component is the number of similarly qualified employees whose employment with defendantterminated (voluntarily or involuntarily) during the particular time unit of the analysis. The second component is constituted by those employees with similar qualifications and experience who were active as of the end of the prior time unit of the analysis. This second component of the attrition rate should also exclude those employees of defendant who were promoted, demoted or trans­ferred during the course of the time unit of analysis.

The formulation for the calculation of attrition rates is as follows:

(1) AR = (ni,y/Ni,y-1)

Where:

ni,y = Number of employees in job’ i’ who left the employer during year ‘y’

Ni,y-1 = Number of employees in job ‘i’ at the end of year ‘y-1’

Both components of the attrition rate demand careful attention. In a wrongful termination case the attrition rate should provide a clear picture of the likelihood than any given employee would leave the employer (voluntarily or involuntarily). Employees who leave a particular job due to promotion, demotion or transfer remain employed with defendant, and as such, do not provide an accurate reflection of the likelihood of termination from the employer. As a result, they should not be included in the numerator of the attrition rate. These same employees were actively employed in the job in question at the end of the previous pay period (and therefore could have potentially left the employer voluntarily or involuntarily during the current pay period), and as such should be included in the denominator of the attrition rate.⁷

As discussed in the next section, the use of retention rates in calculation tables becomes more convenient for intuitive purposes than the attrition rate, and it is calculated as follows:

(2) RR = 1 – AR

Table 2 reports hypothetical information regarding the retention rate of employees similarly-situated to our fictional plaintiff. The information in this table was constructed under the premise that attrition rate information was readily available from defendant and that it was produced during discovery. It is also assumed that employees in each level of the fictional career path possess similar qualifications. Moreover, any potential differences in prior experience or other relevant qualifications between plaintiff and other employees of defendant who are in the same job level as plaintiff at any point in time, are assumed to be negligible.

In order to calculate meaningful attrition rates, one needs to consider several factors, the most important one being the definition of similarly-situated employees. In many cases, similarly-situated employees are those employees in the same occupation with the same employer and whose qualifications for the job do not vary significantly. For qualifications not to vary significantly, factors such as education, training, prior experience and seniority with defendant often need to be taken into consideration according to the relevance of each of these factors for the performance of the job in question and for the particular organization.

Another factor that may need to be considered is the economic situation of the firm itself. Suppose that the time period in question occurs when the employer is in a state of decline with constant layoffs. By incorporating data for this time period only, the economist is implicitly assuming that the employer will remain in a state of decline and will continue to have layoffs. This may be a valid assumption if the time period for which the economist is making projections is relatively short. However, if the projections cover many years, then some modifications to the attrition rate may be necessary.⁸

It has been our experience that information regarding attrition rates is often not readily available from the employer. In these cases, it is the job of the economic expert to make an effort to obtain relevant information (such as work history data) in order to construct the appropriate attrition rates.⁹

Expected Earnings with Defendant

Click here to download Table 3 and Table 4.

To illustrate how attrition rates may affect the projected earnings of plaintiff with defendant, this concept is applied to the calculations of economic losses in our hypothetical case study. Pre-trial and post-trial calculations of economic losses associated with the wrongful termination of our hypothetical plaintiff are presented in Table 3 and Table 4, respectively.

The estimations are conducted incorporating the retention rates for each of the levels of plaintiff’s career path presented in Table 2. As presented in this table, the likelihood that the plaintiff would have remained employed with defendant varies across years and across career Levels 1 through Level 4. Moreover, as discussed in Section III, it is assumed that plaintiff would have been promoted to Level 3 at the beginning of the 10th year of employment with defendant and to Level 4 at the beginning of the 15th year of employment with defendant.

Both the retention rate and the expected promotions of plaintiff in the absence of wrongful action by defendant are incorporated in the estimation of economic losses and presented in Tables 3 and 4. In both of these tables, the column titled ‘Compounded Retention Rate’ reflects the changes in the retention rate from year to year and the expected promotions of plaintiff in years 10 and 15.

The likelihood that any given employee will remain employed an entire year is contingent on the fact that the employee was actually active the year before. The concept of conditional probability is applied in column ‘Compounded Retention Rate’ of Tables 3 and 4 by calculating the joint probabilities that the plaintiff would have remained employed with defendant each year after the incident, conditional on the likelihood that he/she was employed each prior year. The likelihood that our hypothetical plaintiff would have remained employed with defendant in year 13, is then conditional on his/her having remained employed though year 12, which in turn is conditional of his/her having remained employed through year 11, and so on.

Consider Year 9 of our plaintiff’s expected employment with defendant as an example. Throughout this year, our hypothetical plaintiff is expected to have been employed in Level 2 of the career path. According to Table 2, the retention rate for Level 2 employees at the end of Year 9 is 0.9833. The compounded retention rate for Year 9 is calculated as the retention rate for Level 2 employees at the end of Year 9, compounded by the likelihood of having remained employed through Year 8.10 As reported in Table 3, the compounded retention rate for Year 9 is 0.9779 (0.9833*0.9945)

In order to illustrate the effect of promotions in the calculation of the compounded retention rate, consider Year 10 as an example. Our fictional plaintiff is expected to be promoted to Level 3 of the career path at the beginning of this year. The retention rate applicable to employees in Level 3 during this 10th year of employment is 0.9899. The compounded retention rate for Year 10 will then be the result of compounding the retention rate of Level 3 in Year 10 by the already compounded retention rate of Year 9. As reported in Table 3, this calculation yields the compounded retention rate 0.9680 (0.9899*0.9779) for year 10.

The ‘Expected Earnings’ column in Tables 3 and 4 reflects the expected promotions of plaintiff, had he/she remained employed with defendant, as well as the history of non-promotional and promotional raises received during his/her employment with defendant. Years when a promotion is not expected to occur reflect expected earnings resulting from applying the average non-pro- motional growth rate of earnings to the earnings of the prior year. The expected earnings of plaintiff for years other than 10th and 15th (expected promotion years) reflect the historical 3.23% average increases received by plaintiff during non-promotional years of employment with defendant. Expected promotion years, on the other hand, reflect the 8% earnings increase established through company policy.

To illustrate this point, let’s consider Year 9 and Year 10. The plaintiff is not expected to receive a promotion at the beginning of Year 9. This being the case, the earnings for this year are expected to have been $59,353, which equals the earnings in Year 8 plus 3.23%. With regard to Year 10, however, the plaintiff is expected to have been promoted to Level 3 of his/her career path had he/she remained employed with defendant. This being the case, the expected earnings in Year 10 are expected to have been $64,102, which equals the expected earnings of Year 9 plus the 8% promotional raise policy of defendant.

The effect of incorporating attrition rates in the estimation of economic losses is reflected in the column titled ‘Adjusted Expected Earnings’. The figures in this column result from adjusting the ‘Expected Earnings’ figures by the figures reported in the column ‘Compounded Retention Rate’. This adjustment takes place by simply multiplying the expected earnings by the compounded retention rate. Continuing to use Year 9 as the illustrative example, the earnings figure in the absence of attrition is $59,353, while the attrition adjusted expected earnings for this same year reach the amount of $58,041.

For comparison purposes, Tables A1 and A2 in the Appendix report the expected earnings figures of a scenario that does not account for employee attrition. This comparison is discussed further in Section VII.

Mitigating Earnings

The forensic economist must estimate mitigating earnings in the front pay period, and in some cases, the back-pay period as well. There is typically a legal finding as to whether or not the plaintiff has or has not made adequate efforts to mitigate losses over the back-pay period by pursuing alternative employment. If there is a finding of a failure to mitigate on the part of the plaintiff, then it will be necessary to estimate mitigating earnings over the back-pay period. Mitigating earnings in the front-pay period are always unobserved and must be estimated.

Estimates for the level of mitigating pay will depend on a variety of factors. The skill set, education level, and experience level of the plaintiff typically play an important role in the wage that the plaintiff can demand in his or her best employment opportunity. Demand and supply conditions within the occupation in which the plaintiff is employed will also impact the level of mitigating pay. Economists often rely upon input from other experts, including Industrial Psychologists and Vocational Experts, when generating estimates of earnings capacity.

While attrition rates are very important to the calculation of front pay and back pay, the unemployment rate is crucial to the calculation of mitigating earnings. Back and front pay calculations require us to determine the plaintiff’s expected earnings at a particular employer (the defendant). Mitigating earnings calculations, on the other hand, call for the determination of expected earnings at any employer. Hence, for mitigating earnings, the primary concern is whether the plaintiff is employed, not where the plaintiff is employed.

Involuntary separations associated with layoffs and downsizing are relevant to the calculation of mitigating earnings over the back- and front-pay period. One typically controls for the possibility of these involuntary separations by adjusting the expected mitigating earnings in each period downwards by the appropriate non-compounded unemployment rate. To the extent possible and appropriate, industry and occupation specific unemployment rates should be used. Furthermore, industry, occupation, and plaintiff-specific characteristics will dictate whether or not a local, regional, or national unemployment rate is appropriate.

For the purposes of this case study we have assumed that the plaintiff’s best employment alternative after the alleged wrongful termination has a salary level of $45,000. Furthermore, we assume that this alternative salary will increase by 4% each year. As discussed earlier, mitigating earnings must be adjusted downward each year by the relevant unemployment rate and we have selected a rate of 7%. The calculations of mitigating earnings based on these assumptions are detailed in the ‘Mitigating Earnings’ and ‘Adjusted Mitigating Earnings’ columns of Tables 3 and 4. Identical mitigating earnings calculations for back pay and front pay associated with the “No Attrition” alternative calculation are presented in Tables A1 and A2.

Summary and Conclusion

The forensic economics literature makes a clear case that job turnover or attrition is a critical component in the calculation of back pay and front pay in a wrongful termination damages calculation. Franz’s survey calls attention to the issue and Trout successfully incorporates attrition rates into damage calculations using estimated attrition rates based on aggregate data. In this paper we have developed a hypothetical case study of damages in a wrongful termination litigation that utilizes calculated attrition rates based on defendant specific data.

We stress two important requirements in our application of attrition rates to back- and front-pay estimates. First, it is necessary that the attrition rate used in each period is generated using a set of similarly-situated (relative to the plaintiff) individuals. Second, it is necessary that the attrition rates be appropriately compounded over the back- and front-pay period. Failure to adequately implement either of these steps can seriously bias the damages estimate.

While our case study is designed to highlight the relevance of job turnover or attrition in wrongful termination damage calculations, we address other key issues in the model. The case study allows for promotions and associated pay increases. We also construct mitigating earnings estimates that incorporate an unemployment rate adjustment. Within this damage modeling framework we generated economic loss calculations based on a selected set of parameters.

As one might expect, accounting for the possibility of job turnover or attrition has the potential to have a dramatic impact on the plaintiff’s estimated economic loss in each period and overall. In Table 4 it is revealed that, when accounting for attrition the plaintiff’s adjusted expected earnings with the defendant firm fall below the plaintiff’s expected mitigated earnings at 28 years of service. In other words, the defendant’s estimated economic loss associated with the wrongful termination fall to zero in the 28th year of service. On the other hand, Table A2 reveals that when attrition is ignored the estimated economic loss associated with the wrongful termination remain positive all the way up to and through the plaintiff’s retirement age.

The impact of the adjustment is even more apparent when contrasting the cumulative discounted economic loss results. Table 5 summarizes the cumulative discounted damages associated with the two scenarios. The table reveals that, for the set of parameters we have selected for this case study, the failure to account for job turnover or attrition will overstate plaintiff damages by 70%.

While our analysis does address the core issues in generating damages estimates in wrongful termination litigation, there are potential complexities we do not address here. Chief among these are the complexities associated with fringe benefits. Valuing various employer provided insurance packages presents a unique set of challenges. An even more potentially complex set of issues is raised by defined benefit pension plans in which the level of benefits is typically a function of years of service. Another complexity of this analysis is the wide variation in the data made available by different employers. Due to data availability and the specific issues of a particular case, there is no single correct method for incorporating attrition rates into an analysis of economic losses. The analysis presented in this paper assumes that the relevant data are available and reliable. If either of these conditions is not met, the forensic economist may need to take a substantially different approach.

Endnotes

*Dr. White is a Director of ERS Group, Washington, DC. Dr. Tranfa-Abboud, formerly with ERS Group is a Consultant in New Jersey and Visiting Professor, Rutgers University (Newark). Dr. Holt is a Research Economist with ERS Group, Washington, DC. We gratefully acknowledge the comments of two anonymous referees.

¹A discussion of cases involving imperfect or incomplete data is presented later.

²Although promotions across the different levels of the career path of plaintiff are assumed to be contingent on good performance on prior attained levels, the estimation of economic losses presented in this paper assumes certainty of future promotions of the plaintiff at the beginning of the typical years when such promotions have historically taken place. In cases where promotions are uncertain even in the absence of wrongful actions by defendant, probabilities of promotion need to be incorporated in the analysis according to the availability of information necessary for the estimation of such probabilities. Our assumption of promotion with certainty is a simplifying one, and it is introduced with the purpose of maintaining focus on the effects of the attrition rates on the estimation of economic losses.

³The assumption that the mitigating earnings growth rate is equal to the pre-termination growth rate is utilized in this hypothetical case for ease of analysis. Of course, it would be the economist’s job to determine whether the post-termination earnings growth rate is greater than, less than, or approximately the same as the pre-termination rate.

⁴It should be noted that these assumptions are made for illustrative purposes only. With regard to the back-pay interest rate, the case’s jurisdiction may not allow back-pay interest at all. Furthermore, if back-pay interest is allowed, the case’s jurisdiction may mandate the type of instrument and the time period to be used. With respect to the unemployment rate, the authors encourage the use of an unemployment rate that is as specific to the plaintiff as possible (e.g., same gender, occupation, industry, geographic location, time period of measurement, etc.). It is often the case, however, that unemployment rates that account for all these factors are not available. Of course, any assumptions made in actual cases need to be tailored to the specific case and tested for their appropriateness based on the particular situation. However, adjustments in the assumptions can be introduced to reflect the reality of an actual case without a significant impact on the methodology discussed in this paper.

⁵One complicating factor arises when a plaintiff is already close to or beyond his/her expected retirement age given the behavior of similarly-situated employees. When this situation arises it may be necessary to incorporate mortality tables into the forecasts of attrition probabilities.

⁶Special care must be taken when there are a small number of employees determined to be similarly- situated to the plaintiff. Attrition rate calculations in these cases are more subject to extreme values, further highlighting the importance of the forensic economist understanding the underlying process used to calculate the attrition rates.

⁷Another aspect that may need to be considered is whether any of the similarly-situated employees were also wrongfully terminated in the past. If these people can be identified, then they should be excluded from the attrition rate calculation, under the argument that the plaintiff’s damages should not be reduced because of the defendant’s past discrimination of others. The inclusion of other wrongfully-terminated employees would lead to the attrition rate being overstated.

⁸The authors are indebted to a referee for raising this point.

⁹It may also be the case that the information could potentially be available from the employer, but for various possible reasons has not been produced during discovery. As such, the economist may need to consider relying on tenure or turnover statistics available from government surveys, such as the one reported by the Bureau of the Census (U.S. Department of Commerce, Bureau of the Census: Current Population Survey: Job Tenure and Occupational Mobility, 2000, Computer file, Washington, DC). Furthermore, it is often the case that the employer cannot provide data to calculate annual attrition rates, but is able to determine either (a) the average seniority of similarlysituated employees, or (b) the average years of service at the time employees leave the company.

¹⁰Recall that the plaintiff worked with the defendant until the end of Year 7.

References

Franz, Wolfgang, “Wrongful Employment Termination and Resulting Economic Losses,” Journal of Forensic Economics, 1990, 3(2), 31-47. Tobias, Paul, Litigating Wrongful Discharge Claims, Deerfield, Illinois: Caloghan & Co, 1987. Trout, Robert, “Duration of Employment In Wrongful Termination Cases,” Journal of Forensic Economics, 1995, 8(2), 167-177.