THE
REPRESENTATION OF WOMEN IN STORE
MANAGEMENT
AT WAL-MART STORES, INC.
Marc Bendick, Jr., Ph.D.
January 2003
____________________________
Marc Bendick, Jr., Ph.D.
January ____, 2003
Bendick
and Egan Economic Consultants, Inc.
phone
(202) 686-0245 fax
(202) 363-4429
bendickegan@mindspring.com www.bendickegan.com
I. Introduction
1. I am a principal in Bendick and Egan Economic Consultants, Inc., located at 4411 Westover Place, NW, Washington, D.C. 20016. I earned a Ph.D. in economics from the University of Wisconsin in 1975 and have engaged in the full-time practice of economics, specializing in employment and related issues, for approximately 30 years. I am the author, co-author, or editor of more than 100 pieces of scholarly research, including books, monographs, articles in refereed journals, and Congressional testimony. My resume is provided as Attachment A to this report.
2. Since 1980, my professional activities have included litigation on employment discrimination, including analysis of the availability of job-seekers in different demographic groups; processes for recruiting, hiring, training, assigning, evaluating, compensating, promoting, and discharging employees; voluntary and involuntary initiatives to manage demographically-diverse workforces; and monetary damages associated with denial or diminution of employment opportunities. Attachment B to this report lists the approximately 114 cases in which I have participated, representing both employers and employees. That list indicates that my expert testimony has been accepted in 14 federal courts and 7 state courts or administrative tribunals.
3. In the case of Dukes et al. v. Wal-Mart Stores, Inc., I was requested by plaintiffs’ counsel to analyze the employment of women among employees, especially in-store management employees, at Wal-Mart, Inc.[1]
4. In preparing the analyses presented in this report, I have utilized modes of analysis, computational procedures, information sources, and standards of care identical to or comparable to those I use throughout my scholarly research, and I apply theories, models, concepts, reasoning, and assumptions that command general acceptance among my professional peers. I hold the opinions I present in this report to a reasonable degree of scientific certainty.
5. For my work in this case, I am being compensated at the rate of $225 per hour, plus out-of-pocket expenses.
6. In part, the present report represents an updated, refined version of my earlier report entitled The Representation of Women among Managers at Wal-Mart: A Preliminary Analysis Based on EEO-1 Data (June 2001). As in that previous report, in preparing the present report, I have relied on information available as of the date of this report. If at a future date this information is refined, corrected, or supplemented, I would like the opportunity to modify this report to reflect this additional information.
7. In preparing my analysis, I have reviewed the information sources listed in Attachment C to this report.
8. The core of the analysis presented in this report involves “EEO-1” data collected by the federal Equal Employment Opportunity Commission annually from large private employers. This mandatory survey covers all private sector establishments with 100 or more employees and documents the employment representation of different demographic groups, such as women, among 43 million employees in nearly 200,000 workplaces nationwide. As essentially a census of workers employed by the nation’s largest firms, it is a widely used, authoritative source of information on employment patterns from which results are published annually by the EEOC.[2] EEO-1 data are widely used by the EEOC and other EEO enforcement agencies to analyze employers’ compliance with anti-discrimination laws, by researchers and public policy analysts to measure demographic patterns in the U.S. labor market, and by employers reviewing their employment practices and developing affirmative action or workforce diversity management initiatives.
9. At the time this report was prepared, the latest EEO-1 data released by the EEOC for use outside the Agency is for the year 2000, and only partial data has been released for that year.[3] Therefore, my primary focus for detailed analysis is the most recent year for which apparently complete data has been released, 1999.
10. Attachment D presents some basic facts about the Wal-Mart work force based on EEO-1 data for 2000.[4] As is well known, Wal-Mart is a very large firm -- in fact, the largest private employer in the nation. According to the first table in Attachment D, in its EEO-1 data for 2000, the company reported 967,247 employees, and that figure was growing at an annual rate of 8.3 per cent. Not surprisingly for a retail firm, Wal-Mart’s largest category of workers is Sales Workers, accounting for 58.0 percent of all employees in 2000.
11. The firm’s 43,489 managers in 2000 represented 4.5 percent of all its employees. EEO-1 data do not subdivide this category, so it is not possible to distinguish, for example, between senior executives and middle managers or, within stores, between store managers and assistant store managers. Hence, the present analysis examines only the representation of women among managers as a group, as Wal-Mart as defined them in filing its EEO-1 reports. It is my understanding that in-store employees reported as in-store managers in Wal-Mart’s EEO-1 data consist of salaried employees typically bearing the job titles of store manager, co-manager, or assistant manager.[5]
12. Attachment D also reveals that, in 2000, Wal-Mart employed 600,569 women company-wide, constituting 62.1 percent of its total work force. Within this total, women accounted for 60.6 of the firm’s non-managerial employees and 33.6 percent of its managers.
III. Benchmarking Wal-Mart
Against Comparator Companies
13. The analysis in this report involves “benchmarking” -- that is, comparing -- employment patterns in Wal-Mart to those of comparable large retail firms. The practice of benchmarking a company’s performance against those of similar firms is a standard management technique used throughout the private sector, not only with respect to employment matters but also with respect to a variety of financial and operational concerns.[6] Applying this procedure to human resource management, companies very commonly participate in multi-firm salary information pools, read industry-wide surveys on fringe benefits, and study and adopt “best practices” in human resource management based on practices successfully implemented by other firms. Comparing the representation of different demographic groups in their own workforce to that in the workforces of comparator firms is a very common way for firms to estimate the expected representation of that group among its own employees, for example in preparing affirmative action plans or developing workforce diversity management initiatives.
14. Like virtually all large American corporations, Wal-Mart commonly engages in benchmarking its human resource management practices and outcomes. In a deposition taken for this case, Mr. Coleman Peterson, Wal-Mart’s Executive Vice President-People Division, described Wal-Mart’s activities as follows:[7]
Best practices, in my definition, … means to make efforts to really understand what practices, be they in finance or information systems or logistics or human resources, understanding what is it that people are doing in their businesses that seem to be successful processes of programs… We haven’t done a benchmark at one of these companies and immediately run back and say hey, that’s something we’ll want to go and do. I do think what is helpful is to kind of reaffirm the areas of focus that they are working on, the challenges they are facing, things of that nature, are things that kind of reinforce. Are we on the right track? Are we looking at the right things? Are we working on the right things?…It’s kind of like running a race, okay. You look to the left or the right and just kind of have a feel for, you know, where other people are. It kind of encourages you to run faster…[Benchmarking] can be used interchangeably with the term best practices.
15. Continuing the description just quoted, Mr. Peterson identified large, nationwide retail chains as key comparators for Wal-Mart benchmarking, specifically mentioning Target, JC Penney, Home Depot, and Kmart as companies with which Wal-Mart has had discussions concerning employment best practices.[8] Mr. Peterson also discussed a presentation he had made to Wal-Mart’s Board of Directors on August 15-16, 2001 that included an analysis of profit per employee in which Wal-Mart compared itself against 11 other large retail chains: Home Depot, Costco, Lowe’s, Gap, Dollar General, Target, Sears Roebuck, Nordstrom, Dillard’s, Kmart, and JC Penny.[9] Mr. Peterson also described an effort by Wal-Mart in 2000 to initiate an exchange of information on the demographic characteristics of their workforces among 15 of the largest retailers in the nation: Comp USA, Dayton Hudson, Home Depot, May Department Stores, Toys-R-Us, Circuit City, Costco, Dollar General, Dollar Tree, Kmart, Office Depot, Sears, Service Merchandise, Shopko, and Montgomery Ward.[10] In reporting to the Compensation and Nominating Committee of the Wal-Mart Board on March 4, 1999, Mr. Coleman compared the demographic characteristic of the Wal-Mart workforce to that of all workers in the U.S. labor force and all workers in the retail industry as reported in EEO-1 data.[11]
16. In examining issues such as the demographic characteristics of a firm’s workforce, benchmarking using comparator firms addresses the important issue of ”labor supply” -- the availability of persons in the demographic group being examined (here, women) who are interested in and qualified to hold the positions in question (here, Wal-Mart in-store management). The logic in benchmarking is that, if retail chains comparable to Wal-Mart are successfully employing women at some rate, then women are presumably available, interested, and qualified to hold comparable positions at Wal-Mart at a similar rate. Therefore, if the representation of women among store managers at Wal-Mart is substantially lower than at the comparator large retail chains, the cause of the “shortfall” is to be found in how Wal-Mart hires, trains, selects, assigns, treats, promotes, or retains women employees rather than lack of available, interested, qualified women job candidates.
17. Wal-Mart itself has benchmarked the level of representation of women among its management employees. For example, through a memorandum to all division heads dated March 29, 1999,[12] the company compared itself to five large retail chains – Comp USA, Dayton Hudson, Home Depot, May Department Stores, and Toys-R-Us – and concluded that “…Wal-Mart’s women in management percent (32.4%) is significantly behind several of the other retailers reporting (43.2% to 65.3%).” Using nationwide average EEO-1 data for 1997, this same report also concluded, “…Our Wal-Mart percent of women trails both the retail industry and workforce averages…”
18. Such benchmarking of Wal-Mart’s representation of women among managers using five retail chains and simple nationwide averages represent a relatively crude benchmarking analysis. In this report, I refine Wal-Mart’s procedures in three important ways:
o First, Wal-Mart’s analysis was limited to data readily available to Wal-Mart: national averages published by the EEOC and data provided by five firms that had agreed to share workforce demographic information. My analysis begins by defining appropriate comparator firms as those large, multi-state retail chains that are widely recognized by Wal-Mart itself and by standard sources of business and financial information as comparable to Wal-Mart in such terms as products, services, market niche, size, scope, operational style, and expected financial and operational results. I then systematically search EEO-1 data to identify firms matching this definition. The approximately 20 comparator firms identified through this process represent a more comprehensive, more conceptual set of comparators than Wal-Mart itself sought out.
o Second, Wal-Mart’s analysis simply examined the representation of women among managers wherever they work within the company. My analysis separates Wal-Mart’s retail stores from its other establishments, such as corporate headquarters and distribution centers, and I do the same for the 20 comparator firms. This separation allows more precise comparison of the representation of women in in-store managerial positions, which I understand to be the current focus of this case.
o Third, Wal-Mart’s analysis simply examined the average representation of women among managers nation-wide. My analysis separately examines 375 local labor markets, thereby removing from the comparison differences between Wal-Mart stores and their comparators attributable to local labor markets where the participation of women in paid employment is higher or lower than the national average.[13]
All three of these refinements are designed to ensure that this benchmarking analysis presents an “apples to apples” comparison. In such a comparison, differences in the representation of women in management between Wal-Mart and its comparators can appropriately be attributed to differences in Wal-Mart’s employment policies and practices, not flaws in the comparisons.
19. To accomplish the first of these three refinements, I need to identify those firms that are appropriate comparators to Wal-Mart. As noted in paragraph 15 above, Mr. Peterson identified large retail chains as Wal-Mart’s key comparators, and he repeatedly relied on data from such chains in his own work. Echoing this corporate positioning, Wal-Mart’s 10K/A Annual Report to the Securities and Exchange Commission (SEC) for the fiscal year ended January 31, 2001, filed April 17, 2001, pages 4 and 6, states:
Wal-Mart discount stores
compete with other discount, department, drug, variety, and specialty stores,
many of which are national chains.
Wal-Mart Supercenters compete with other super-center type stores,
discount stores, supermarkets, and specialty stores, many of which are national
or regional chains….the Wal-Mart Stores segment ranked first among all retail
department store chains and among all discount department store chains….[The
SAM’S Club segment] competes with other warehouse clubs, as well as with
discount retailers, wholesale grocers and general merchandise wholesalers and
distributors.
20.
Consistent with this description, in its 1999 EEO-1
report to the EEOC, Wal-Mart places itself in Standard Industrial Code (SIC)
53, General Merchandise Stores. This
category encompasses department stores, variety stores, and miscellaneous
general merchandise stores. Some EEO-1
reports filed for individual Wal-Mart’s retail establishments place the
establishment in SIC 531 (Department Stores), others in SIC 533 (Variety
Stores), and still others in SIC 539 (Miscellaneous General Merchandise
Stores).
21. Taking my lead from Wal-Mart’s own description, I compare Wal-Mart to a set of large, multi-state chains of general merchandise stores identified by applying the following criteria to EEO-1 data for each year I examine: (1) The firm was not Wal-Mart; (2) the EEO-1 “parent company” record placed the firm in a retail industry other than eating and drinking places: Standard Industrial Codes (SIC) 52, 53, 54, 55, 56, 57, or 59; and (3) the parent company had at least 100 separately-reporting establishments in the General Merchandise Stores industry (SIC 53).
22.
In 1999, 20 firms met these criteria and are used as
comparators.[14] Selected characteristics of these 20 firms,
as of 1999, are reported in the following table:
|
Firm |
SIC for the Parent
Company |
Industry
for the Parent
Company |
Total Employees |
Establishments Nationwide |
|
A |
531 |
Department Stores |
279,255 |
1,168 |
|
B |
531 |
Department Stores |
271,171 |
2,111 |
|
C |
531 |
Department Stores |
264,581 |
1,254 |
|
D |
569 |
Misc. Apparel &
Accessories Stores |
190,063 |
1,030 |
|
E |
531 |
Department Stores |
150,942 |
415 |
|
F |
531 |
Department Stores |
133,906 |
434 |
|
G |
531 |
Department Stores |
85,421 |
356 |
|
H |
531 |
Department Stores |
63,285 |
340 |
|
I |
565 |
Family Clothing Stores |
58,228 |
214 |
|
J |
539 |
Misc. General Merchandise
Stores |
54,845 |
247 |
|
K |
539 |
Misc. General Merchandise
Stores |
54,171 |
336 |
|
L |
531 |
Department Stores |
37,152 |
244 |
|
M |
531 |
Department Stores |
37,039 |
297 |
|
N |
531 |
Department Stores |
34,600 |
391 |
|
O |
531 |
Department Stores |
18,837 |
167 |
|
P |
531 |
Department Stores |
15,895 |
198 |
|
Q |
539 |
Misc. General Merchandise
Stores |
15,693 |
187 |
|
R |
531 |
Department Stores |
14,512 |
109 |
|
S |
539 |
Misc. General Merchandise
Stores |
12,943 |
102 |
|
T |
531 |
Department Stores |
10,547 |
105 |
|
Wal Mart |
533[15] |
Variety Stores |
887,383 |
3,027 |
23.
Because the EEO-1 data publicly released by the EEOC
do not include firms’ names, I do not individually identify firms A through T
by name. However, standard analyses of
Wal-Mart repeatedly mention a limited number of large firms as Wal-Mart’s
direct counterparts. In its 2001
“Fortune 1000” list of the nation’s largest firms, Fortune magazine
places Wal-Mart in the category of “General Merchandise Stores” with 15 other
companies: Sears Roebuck, Kmart, Target,
JC Penny, Federated Department Stores, May Department Stores, Dillard’s, Saks,
Kohl’s, Nordstrom, Dollar General, Ames Department Stores, Family Dollar
Stores, Belk, and Value City. The online investment guide Forbes.com
places Wal-Mart in the “retail/stores/discount” group and compares it to: Ames
Department Stores, Big Lot, Bradlees, Dollar General, Dollar Tree,
Duckwall-Alco Stores, Family Dollar Stores, Freds, Kmart, Liquidation World,
Mazel Stores, 99 Cents Only Stores, Shopko Stores, Denninghouse, Target, TJX
Companies, Toys-R-Us, and Value City.
The on-line business information system Hoovers.com lists Wal-Mart’s
top three competitors as Costco, Kmart, and Target. Many of these same firms appear in Mr.
Peterson’s description of Wal-Mart’s own benchmarking, discussed in paragraphs 15 and 17 above. I am confident that the comparator firms in
my EEO-1 analyses substantially overlap these lists.
24. The second refinement listed in paragraph 18 is to subdivide the 3,027 establishments for which Wal-Mart’s provided separate EEO-1 reports in 1999 into six mutually- exclusive groups, so that they can be benchmarked separately:
o
Corporate headquarters (1 Wal-Mart
establishment), identified as the firm’s headquarters in the company’s EEO-1
report.
o Stores (2,909 Wal-Mart establishments), in which more than 50% of each establishment’s staff are Sales or Service Workers.[16] [17]
o Non-Store Establishments–Managerial/Professional (4 Wal-Mart establishments), in which more than 50% of employees are Officials and Managers or Professionals.
o Non-Store Establishments–Blue Collar (72 Wal-Mart establishments), in which more than 50% of employees are Craft Workers, Operatives, or Laborers.
o Non-Store Establishments–Other (41 Wal-Mart establishments), which fall into none of the previous four categories.
o Employees not Locationally-Identified These employees appear in the firm’s count of all employees but not in any of the establishments for which a separated EEO-1 report has been filed.
25. The third refinement listed in Paragraph 18 is to replace Wal-Mart’s comparisons of national averages with more precise comparisons between Wal-Mart and its comparators in 375 local labor markets across the nation.
26. The diagram in Attachment E illustrates my procedure for analyzing each local labor market, using 1999 EEO-1 data for the Bakersfield, California Metropolitan Area as an example. The diagram reports that, in the Bakersfield area in 1999, Wal-Mart operated four stores (represented by shaded bars), while its comparator chains operated 18 stores (represented by white bars). At the Wal-Mart stores, the proportion of women among managers ranged from 0.0 percent to 25.0 percent, while at the comparators, it ranged from 0.0 percent to 83.3 percent. If Wal-Mart had utilized women as managers in its four stores at the same rate as the average of those 18 comparators stores, which was 54.7 percent, then Wal-Mart would have employed 14.6 more women managers in the Bakersfield area in 1999 than it actually did. More conservatively, if Wal-Mart had utilized women among managers at the same rate as they were utilized in the total “large chain general merchandise industry” in the Bakersfield area -- the 18 comparator stores plus the four Wal-Mart stores -- then Wal-Mart would have had 12.0 more women managers in the Bakersfield area than it actually did. Throughout this report, I present figures such as this latter one as my conservative estimate of Wal-Mart’s shortfall in women managers.[18]
27. To develop my estimate of Wal-Mart’s shortfall nationwide, I first compute the shortfall in each local labor market in which Wal-Mart and at least one comparator operate retail stores. In analyzing EEO-1 data for 1999, I made such comparisons for 375 separate locations, including 327 Metropolitan Statistical Areas as defined by the U.S. Census Bureau (for example, the Bakersfield, California Metropolitan Area) and 48 “balance of states” (the areas within each state outside all Metropolitan Statistical Areas).
IV. Wal-Mart’s Shortfall of Women
in Store Management in 1999
28. Attachment F presents, in its column (m), the estimate of Wal-Mart’s shortfall in female managers in 1999 that emerges when the procedure illustrated for Bakersfield is applied company-wide. Row (2) of this table focuses specifically on managers employed within the 2,909 Wal-Mart establishments I have identified as retail stores.[19] According to this table, in 1999, Wal-Mart had a shortfall of 4,004 female in-store managers, compared to the expected number based on the representation of women among in-store managers in the large chain retail industry consisting of Wal-Mart itself and 20 comparator large retail chains.
29. Column (q) of the table presents a
second, alternate estimate of Wal-Mart’s shortfall in female in-store mangers,
based on a benchmark that includes the 20 large comparator chains but not
Wal-Mart itself. As in the Bakersfield
example, this alternative estimate is larger than the estimate based on a
benchmark including Wal-Mart itself. Row
(2) of the table reports a total shortfall in column (q) of 6,445 female
managers, 61.0 percent more than the shortfall of 4,004 reported in column (m).
30. The first,
more conservative of these two estimates, -- 4,004 female in-store managers --
can be restated several different ways.
First, while Wal-Mart actually employed about seven women out of
every 20 in-store managers, it would be expected to have employed ten. Second, for every seven women managers
Wal-Mart actually employed, three additional women managers were
“missing.”
31. One reason that this estimated shortfall is so large is that Wal-Mart is such a large employer that even a modest percentage difference in female representation would generate a large shortfall; as row (2) of column (d) of the table in Attachment F reports, in 1999, Wal-Mart employed 29,462 in-store managers.[20] However, Wal-Mart’s size is not the only source of this large shortfall. Columns (i) through (k) of the table report that the percentage difference between the representation of women among managers at Wal-Mart and its comparators is itself very large; across all Wal-Mart retail stores in 1999, the difference is 22.0 percent of all Wal-Mart in-store managers – the gap between Wal-Mart’s in-store managerial workforce with 34.5 percent women and its comparators’ corresponding 56.5 percent.
32. Could a shortfall as large as 4,004 have arisen by chance alone? In many research contexts as well as legal proceedings, a figure such as this shortfall is considered “statistically significant,” and therefore not attributable to chance, if the figure corresponds to a number larger than two to three when translated into statisticians’ units called “standard deviations.” In the present case, column (o) of row (8) of Attachment Three reports that a shortfall of 4,004 corresponds to 47.0 standard deviations, much larger than two to three. Indeed, this figure is so large that the probability that this result could be due to chance alone would be written with several hundred zeros after the decimal point, reflecting a probability very many times less than one chance in many billions. A shortfall as large as 4,004 definitely did not arise by chance alone.
33. When Wal-Mart conducted its own benchmarking, it also observed a shortfall in its female managers. However, the company discounted the meaningfulness of its own comparisons. As the company’s most senior human resources executive, Mr. Peterson, stated in his deposition, “…[T]he reason it’s so difficult to compare apples to oranges is that a number of the comparators we are looking at put department managers in their management category when they are talking about performance. And had we [Wal-Mart] put department managers in these numbers we would be almost without exception better than every last one of the retailers that is reflected here.”[21]
34. Mr. Peterson bases his statements about the comparator firms’ reporting practices on his long experience in the retail industry, including 16 years as a senior human resource executive at the May Company;[22] May Company operates department stores, where, according to Mr. Peterson, including department managers in the EEO-1 category of managers is common practice.[23] In addition, Mr. Peterson had held discussions with various comparators firms over the years, including Target, Kmart, and Home Depot.[24]
35. Apparently, Mr. Peterson believes that even large differences in the representation of women among in-store managers can be explained by differences in firms’ reporting practices. For example, in his deposition, he was asked to comment on a comparison, which he had presented to the Wal-Mart Executive Committee, in which the Target Corporation was estimated to have 52 percent women among its managers, compared to 33 percent for Wal-Mart.[25] Mr. Peterson replied that, although the Target discount stores owned by the Target Corporation were similar to Wal-Mart in terms of their product lines and store size,[26] the figure for Target Corporation was high because it included data from department stores operated by the Target Corporation, not just Target discount stores,[27] and such department stores tended to report department managers as managers, while Wal-Mart does not. Because of such differences in reporting, Mr. Peterson felt that such a comparison, “…absolutely positively…” does not fairly reflect what should be conveyed about the relative performance of Wal-Mart and its comparators.[28]
36. To test whether Mr. Peterson’s rejection of these comparisons make sense, I calculated the average number of managers per store reported in EEO-1 data for 1999 by Wal-Mart and each of the 20 comparator chains. Among the 20 comparators, this figure averaged 10.2 managers per store, almost identical to Wal-Mart’s 10.1 managers per store. Wal-Mart appeared “right in the middle of the pack,” reporting a larger number of managers per store than 11 of the chains and a smaller number than 9 of them. Some comparators reported as few as 3.7 managers per store, strongly suggesting they were not reporting as Mr. Peterson claimed. However, one comparator reported 23.6 managers per store, suggesting that it might doing what Mr. Peterson described.
37. To take account of this consideration, I performed the benchmarking comparisons documented in Attachment G. In section A of Attachment G, I divided the 20 comparators into groups with a similar number of managers per store and total employees per store. In section B, I divided the 20 comparators into groups with a similar number of managers per 100 non-managerial employees and total employees per store. I then benchmarked Wal-Mart against only those comparator chains that closely matched Wal-Mart on these measures, ignoring comparators that appeared possibly to be defining in-store management more broadly or more narrowly than Wal-Mart. In both section A and section B of the attachment, the same two comparator chains turned out to be the most direct matches to Wal-Mart. [29]
38.
The representation of women among in-store managers at
these two closely matched comparator chains averaged 57.4 percent, a figure
22.9 percentage points higher than Wal-Mart’s 34.5 percent. Thus, closely matching comparators to
Wal-Mart in terms of apparent EEO-1 practices did not eliminate Wal-Mart’s very
large shortfall of women in-store managers, as Mr. Peterson implied it would.
39.
In fact, it did the opposite. The 57.4 percent benchmark based on the two
closely matching chains is 1.1 percentage points higher than the
benchmark representation of females among in-store managers based on all 20
comparator chains, which was 56.3 percent.
Increasing the benchmark to 57.4 percent benchmark increases the
estimated shortfall of women in-store managers in 1999 to 4,328. [30]
40. In his deposition, Mr. Peterson stated, “[H]ad we [Wal-Mart] put department managers in these [in-store manager] numbers we would be almost without exception better than every last one of the retailers that is reflected here.”[31] This statement is equivalent to a frank concession by Mr. Peterson that Wal-Mart’s employment exhibits a very strong “glass ceiling” for women at the department manager level. A glass ceiling is signaled by a sharp decrease in the representation of women from one level in a hierarchy to the next one up. The only way that inclusion of department managers could raise a Wal-Mart 34.5 percent representation of women managers to the comparators’ 56.3 percent is if the representation of women among Wal-Mart’s department managers is well above 56.3 percent and therefore far above Wal-Mart’s 34.5 percent representation among assistant managers/co-managers/store managers.
41. In any case, for Wal-Mart to add its department managers to its other in-store managers and then compare the results to in-store manager figures reported by other retailers is clearly inappropriate benchmarking, mixing apples and oranges. Such a procedure would make sense only if the practice of including department managers was widely or universally practiced by the comparators. Mr. Peterson himself asserted only that “some” of the comparators follow the practices, and our data analysis suggests that their number is probably not large. For example, in 1999, only four of the 20 comparator chains reported 15 or more managers per store. The chain that reported 23.6 managers per store appears to be more of an isolated instance rather than typical of the industry as a whole.
VI. Consistency of Shortfall Nationwide
42. The 2,909 retail stores that Wal-Mart reported on in its 1999 EEO-1data were located in hundreds of locations in all 50 states.[32] Is the firm’s shortfall in female managers limited to only some of these locations, or is it universal across the nation? This question is addressed in Attachment H.
43. One piece of information relevant to this question is provided
in column (h) of that attachment.
There, I report that in 1999, a shortfall in female managers was present
in 79.5 percent of all Wal-Mart stores, some 2,255 stores.[33]
With shortfall present in four
out of five stores, it is virtually impossible for the pattern to be
geographically localized.
44. A more refined way to examine this question is to tabulate shortfall
separately for each of the 50 states.
This is done in column (g) of Attachment H, which reports state-by-state
shortfall for 1999. The column reports
shortfall in 49 of the 50 states.[34]
The largest shortfalls are found
in Texas (473 “missing” female in-store managers), Florida (322), and
California (276).
45. Column (h) of Attachment H also reports
that the proportion of stores with a shortfall in women managers is at least 50
percent in every one of the 50 states.
The proportion ranges from 100 percent (in Rhode Island and North
Dakota) to 50.0 percent (in Vermont).
Attachment I tabulates Wal-Mart stores’ shortfall in female managers in
1999 by a different geographical dimension – whether stores are located inside
or outside of Metropolitan Areas defined by the U.S. Census Bureau. I examine this question separately in each
of the four large regions into which the Census Bureau divides the United
States: Northeast, Midwest, South and
West. According to column (g) of the
attachment, substantial shortfall is found both inside and outside Metropolitan
Areas in all four of these Census regions.
Furthermore, according to column (h) of the attachment, the proportion
of stores with shortfall is well above 50 percent both inside and outside Metropolitan
Areas in every region.
46. As noted in paragraph 27 of this report, overall counts of shortfall, such as are presented in Attachments H and I, total the results of comparisons between Wal-Mart stores and comparators conducted separately for 375 local labor markets located across the country. Attachment E presented one example of a local comparison, for Bakersfield, California. To illustrate the point that under-representation of women among Wal-Mart in-store managers is not limited geographically, Attachment J provides four additional examples of local-area comparison demonstrating Wal-Mart under-representation, each drawn from a different major region among the four into which the Census Bureau divides the nation. An example from Portland, Maine, is provided to represent the Northeast. It is joined by Eau Claire, Wisconsin (Midwest), Fayetteville-Springdale-Rogers, Arkansas (South), [35] and Pueblo, Colorado (West).
47. In all four graphs, the shaded bars representing Wal-Mart stores are closer to the left axis of the graph than all, or nearly all, of the white bars representing comparators. The closer to the left axis a bar is located, the smaller the proportion of managers in that store who are women.[36] The four graphs in Attachment J demonstrate that examples of clear Wal-Mart–comparator differences in the representation of women among managers can be found in widely-separated locations in the nation.
48. It is my understanding that the focus of this case is the representation of women among managers in Wal-Mart’s retail stores, not its headquarters or other non-retail establishments such as distribution centers. Nevertheless, Attachment F includes a benchmarking analysis for Wal-Mart’s non-retail establishments parallel to that reported in paragraphs 28 and 29 for its retail establishments. I do so because patterns of employment of women throughout an organization can suggest whether a shortfall in female employment observed in part of the organization -- such as retail stores -- reflects attitudes and practices pervasive throughout the company and deeply embedded in the organization’s corporate culture,[37] or isolated or transitory happenstance.
49. Attachment F presents an estimate of Wal-Mart shortfall in 1999 of females among managers for each of the six categories of establishments defined in paragraph 24 of this report: corporate headquarters (row 1); retail stores (row 2); 4 managerial/professional non-store establishments (row 3); 72 “blue collar” non-store establishments (row 4); 41 other non-store establishments (row 5); and the set of Wal-Mart employees reported in corporate total employment but not allocated to a specific establishments (row 6). According to column (m) of the table, in 1999, Wal-Mart had a shortfall of women among managers in all six categories, including a shortfall of 466 women managers at corporate headquarters, 144 at managerial/professional non-store establishments, 144 in “blue collar” non-store establishments, 107 in other non-store establishments, and 97 among employees outside separately-reporting establishments. According to column (o), each of these shortfalls is “statistically significant,” at a level between 6.7 and 15.9 standard deviations.
50. These findings are clearly inconsistent with an interpretation of Wal-Mart’s shortfall in women managers as isolated to any particular part of the corporation. Instead, they suggest that Wal-Mart’s shortfall of women among in-store managers reflect the firm’s company-wide corporate culture and attitudes, perceptions, policies and practices that embody that culture.
VIII. Persistence of Shortfall, 1975-2002
51. A second indicator that under-representation of women in management is deeply rooted in an organization’s corporate culture is that the pattern has persisted over time. Attachment K repeats the benchmarking analysis for Wal-Mart in-store managers for every year from 1998 through 1995 and every fifth year from 1995 to 1975.
52. According to column (g) of the table in Attachment K, Wal-Mart had a shortfall of female managers in its retail stores in every year examined, ranging from 4,004 in 1999 (when Wal-Mart was a nationwide retailing giant with 2,909 stores) to a shortfall of 168 in 1975 (when Wal-Mart was a relatively modest chain with 106 stores). Dividing these shortfalls by the number of stores operated by Wal-Mart in each year demonstrates that the shortfall per store has remained quite stable for nearly a quarter century -- 1.6 per store in 1975, 1.4 per store 25 years later.
53. Underlying these persistent shortfalls is the persistent gap between the representation of women among managers at Wal-Mart and its comparators, expressed in percentage terms. The graph in Attachment K makes clear that, although Wal-Mart’s representation of women among in-store managers has increased over time, the representation at its comparators has also increased over the same period, and Wal-Mart has never caught up. In the graph, a large and persistent gap is clearly visible between the comparators’ line (marked by c’s) and Walmart’s line (marked by w’s). Most dramatically, the w representing Wal-Mart in 1999 stands at 34.5 percent and thus is lower than the c representing the comparators in 1975, 38.4 percent. That is, in 1999, Wal-Mart had not yet achieved the level of representation of women among managers in its stores that was common to its comparators a quarter of a century earlier.
54. Such long-term persistence is further evidence that the shortfall in female employment observed in 1999 reflects attitudes and practices deeply embedded in the organization’s corporate culture.
55. If Wal-Mart’s employment of women managers continues to converge to that of its comparators at the same rate it has over the past quarter century, 1999’s gap in in-store managers between Wal-Mart and its comparators will require about 29.5 years to disappear.[38] Thus, assuming past historical tends continue in the future, the Wal-Mart and comparator lines in the graph would converge sometime in 2028.
56. However, this estimate is almost certainly far too optimistic. The .75 rate of closure of the gap between Wal-Mart and its comparators is based on all years from 1975 through 1999. In the more recent years within that period, the rate of closure has slowed dramatically. Between 1990 and 1999, it fell to .6 percentage points per year, and between 1995 to 1999, to .25 percentage points per year. At that final rate, convergence between Wal-Mart and its comparators would take about 88 years.
57.
If the EEOC follows its historic schedule for
releasing EEO-1 data, information for 2002 will not be available from the EEOC
until late 2003. However, I can
conservatively estimate Wal-Mart’s shortfall of female in-store managers in
2002 by combining several pieces of information. First, according to the more conservative
estimate presented in Attachment F, Wal-Mart’s shortfall of female in-store
managers in 1999 was 4,004. Second,
according to Wal-Mart’s annual report for 2002, the company’s total employment
in 2002 was 1,383,000, or 152.0 percent of the company’s employment of 910,000
in 1999. Third, assuming the optimistic
.75 percentage points per year rate of closure in the gap in female in-store
management representation between Wal-Mart and its comparators continued
through 2002, the percentage gap would decline by about 2.24 percentage points
between 1999 and 2002, leaving it at 89.8 percent of its 1999 level.[39]
Multiplying 4,004 by 1.52 and then
by .898 yields a conservatively- estimated shortfall in 2002 of 5,465 female
in-store managers.
IX. An Estimate of Shortfall
XI. Summary and Conclusions
66. In his description of benchmarking quoted in paragraph 14 of this report, Wal-Mart’s Executive Vice President-People Division Coleman Peterson stated that benchmarking helps a company determine, “Are we on the right track?” When Wal-Mart benchmarked the representation of women among its in-store managerial employees, evidence of substantial under-representation warned that the firm was not on the right track. After carefully refining, expanding, and updating that benchmarking in the ways discussed in report, I conclude that warning was correct.
67. Specifically, based on the analyses presented in paragraphs 10 through 66 of this report, I reach thirteen conclusions supporting this general finding:
o Conclusion One: The representation of women among in-store managers in Wal-Mart is substantially smaller than would be expected based on the availability of qualified, interested women in the local labor markets in which Wal-Mart operates store, as measured by the utilization of women in in-store management in those labor markets by large retail chains widely recognized as comparable to Wal-Mart.
o Conclusion Two: In 1999 – the year for which I have the most recent complete data – a conservative estimate of the extent of this shortfall is 4,004 female in-store managers.
o Conclusion Three: Reasonable alternative estimates of this shortfall in 1999 include: 6,445 female in-store managers (based on the same method as the conservative estimate but excluding Wal-Mart itself in setting the benchmark), 4,911 female in-store managers (based on benchmarking Wal-Mart’s managerial/non-managerial employee ratios), and 4,328 (based on comparing Wal-Mart to the two comparator chains most closely matching it in terms of managers per store).[42]
o Conclusion Four: When conservatively updated to 2002, the conservatively-estimated shortfall of 4,004 women in-store managers increases to 5,465.
o Conclusion Five: Because the rate at which comparator firms employ women in-store managers takes account of the rate at which women are available, interested, and qualified to hold in-store management positions comparable to those at Wal-Mart, the causes of Wal-Mart’s shortfall are likely to be found in how Wal-Mart hires, trains, selects, assigns, treats, promotes, or retains women employees, not lack of interested, qualified women candidates.
o Conclusion Six: Wal-Mart’s shortfalls are not simply artifacts of differences between Wal-Mart and its comparators in how they classify and report in-store managerial employees. To the extent these differences exist, they do are not sufficiently widespread to invalidate the benchmarking I have performed or to justify an alternative benchmarking procedure in which Wal-Mart would broaden its counts of in-store managers.
o Conclusion Seven: The estimated shortfalls are “statistically significant” and cannot reasonably be attributed to chance alone. The conservative estimated shortfall of 4,004 corresponds to 47.0 standard deviations, representing a probability that a figure this large could have been estimated by chance of less than one chance in many billions.
o Conclusion Eight: Shortfalls of women among Wal-Mart’s in-store managers are pervasive among Wal-Mart retail establishments, appearing in approximately four out of five of the firm’s stores.
o Conclusion Nine: Shortfalls of women in Wal-Mart’s in-store management occur nationwide. They are found in virtually every state in the nation, in every region, and in both urban and non-urban areas.
o Conclusion Ten: Women are also substantially under-represented among managers throughout Wal-Mart’s non-retail establishments, including corporate headquarters, “white collar” facilities, and “blue collar” facilities such as distribution centers.
o Conclusion Eleven: Substantial shortfalls of women among Wal-Mart’s in-store managers were present in every year between 1975 and 1999, and they are likely to have persisted from 1999 through 2002.
o Conclusion Twelve: Throughout its history, Wal-Mart has maintained a degree of centralization of it managerial staff at corporate headquarters that is strikingly higher than its comparator firms.
o Conclusion
Thirteen: The scale, pervasiveness,
persistence, and consistency of under-representation of women among Wal-Mart’s
managers suggests that such under-representation is deeply rooted in the
organization’s corporate culture and the company-wide employment attitudes,
policies and practices that reflect and maintain that culture.
[1]Unless otherwise noted, throughout this report, the phrase “Wal-Mart” refers to the entire firm Wal-Mart Stores, Inc. rather than the subset of that firm’s retail establishments that operate under the Wal-Mart brand.
[2]For general background on EEO-1 data, see the instructions, reporting form, and latest published data from this reporting system at www.eeoc.gov. See also Marc Bendick, Jr., Using EEO-1 Data to Analyze Allegations of Employment Discrimination, A Presentation to the Section on Labor and Employment Law, American Bar Association Annual Meeting, New York, July 10, 2000, at www.bendickegan.com.
[3]In releasing EEO-1 data for 2000, it appears that the EEOC deleted more data than it had deleted in releasing comparable data for 1999 and earlier years. In particular, a record for Wal-Mart’s headquarters establishment was included in 1999 but not in 2000. Similarly, in 1999, separate records were provided for 3,026 Wal-Mart establishments, but in 2000, for only 764 establishments.
[4]This attachment is based on data from the firm’s company-wide “consolidated record,” which appears to be complete for 2000, even if records for individual establishments are not.
[5]Unless otherwise noted, throughout this document, phrases such as “store managers” and “store management” refer to all in-store employees holding any of these titles, not just the one person within each store holding the title “store manager.”
Wal-Mart’s inclusion of
assistant managers in the EEO-1 manager category and exclusion of support
managers from that category is confirmed by the company in Defendant’s
Answers and Objections to Plaintiffs’ Second Set of Interrogatories,
September 13, 2002, Interrogatories 36 and 37.
[6]In one example of retail benchmarking selected almost at random, the website www.bizstats.com/realworld.htm reports three common measures of retail operations -- average sales per square foot, gross profit margins, and inventory turnover -- for Wal-Mart and other large retail chains such as Costco, Home Depot, and Nordstrom.
In another example also encountered almost at random, The Wall Street Journal (February 19, 2002), page A-2, reported findings from a nationwide study of consumer satisfaction conducted by the University of Michigan and the American Society for Quality. The article compared Wal-Mart’s Stores scores in this study to Target discount stores, Target department stores, Nordstrom, Sears, Dillard’s, JC Penney, May Department Stores, AAFES (Army and Air Force post exchanges), Kmart, and Federated Department Stores. It compared Sam’s Club to Costco, Lowe’s, and Home Depot.
[7]Deposition of Coleman Peterson taken December 18, 2002, p. 49 09:07 24 – p. 50 09:07 4; p. 52 09:10 12-20; p. 104 10:38 6-9; p.54 09:12 3-4; italics added.
[8]Deposition of Coleman Peterson taken December 18, 2002, p. 51, 09:09 8-20.
[9]Deposition of Coleman Peterson taken December 18, 2002, p. 76, 09:53 1 – 09 57 9; Exhibit 4 to that deposition, p. 9.
[10]Deposition of Coleman Peterson taken December 18, 2002, p. 104 10:38 10-p. 105 10:39 13; Exhibit 6 to that deposition, Bates WMHO 502704; WMHO510433 (also marked WMHO51043); deposition of Rebecca Carter, a Wal-Mart Corporate Human Resources Analyst, taken June 13, 2002, WMHO160229.
[11]Deposition of Coleman Peterson taken December 18, 2002, Exhibit 6, WMHO 502704.
[12]WMHO510430-WMHO510438.
[13]For example, in 2000, women accounted for 48.5 percent of the civilian labor force in Vermont but 43.4 percent in Utah (U.S. Bureau of the Census, Statistical Abstract of the United States, Table 572, at www.census.gov/statab).
[14]In earlier years, the number of comparator large retail chains varies somewhat: 21 in 1998, 22 in 1987, 23 in 1986, 18 in 1995, 15 in 1990, 19 in 1985, 18 in 1980, and 14 in 1975
[15]Individual Wal-Mart retail establishments that variously report themselves in SIC 531 (Department Stores), 533 (Variety Stores), or 539 (Miscellaneous General Merchandise Stores) do not appear to be distinct from each other in obvious ways. For example, in 1999, total employees per store averaged 294 for those in SIC 531, 272 in SIC 533, and 284 in SIC 539, all within a few percentage points of each other. Moreover, between 1996 and 1999, more than 700 Wal-Mart establishments changed their SIC one or more times.
[16]2,837 of the Wal-Mart establishments that I categorize as stores are reported by Wal-Mart in some category within SIC 53, the General Merchandise Stores industry. 71 other Wal-Mart establishments that I categorize as stores are reported by Wal-Mart in SIC 421, Trucking and Courier Services. My justification for counting such establishments as retail stores is that their staffing reported in 1999 EEO-1 data are very similar to other Wal-Mart stores and entirely unlike other establishments in Trucking and Couriers Services industry, as the italicized rows in the following table show:
|
|
Other |
Wal-Mart “Stores” |
SIC 421 |
|
Job |
Wal-Mart |
Establishments |
Not |
|
Group |
“Stores” |
in SIC 421 |
Wal-Mart |
|
Managers |
3.6% |
3.5% |
6.5% |
|
Professional |
1.0% |
0.9% |
1.9% |
|
Technical |
0.0% |
0.0% |
0.6% |
|
Sales |
64.1% |
64.0% |
1.8% |
|
Clerical |
4.3% |
3.8% |
7.4% |
|
Craft |
8.5% |
9.6% |
5.0% |
|
Operatives |
0.6% |
0.0% |
46.0% |
|
Laborers |
15.8% |
17.1% |
30.3% |
|
Service |
2.2% |
1.0% |
0.6% |
|
TOTAL |
100.0% |
100.0% |
100.0% |
The 72nd Wal-Mart establishment from outside SIC 53 that I call a “store” was reported in SIC 549, Miscellaneous Food Stores.
[17]In describing its retail
operations, Wal-Mart often distinguishes its Wal-Mart operating segment from
its SAM’s Clubs segment or refers to other categories of retail establishments
such as Supercenters or Neighborhood Markets.
I have not subdivided Wal-Mart “stores” into such categories for two
reasons. First, I could not readily
assign Wal-Mart stores for which I had a separate EEO-1 reports to segments based
on information in their EEO-1 reports (such as employees per store; see
footnote 15). Second, I could not
readily assign the individual establishments of comparator chains to comparable
categories.
[18]Inclusion of Wal-Mart establishments in the standard of availability against which Wal-Mart’s employment is compared implicitly corresponds to the conservative assumption that the only persons available to enhance Wal-Mart’s number of female managers are women who have demonstrated their interest and ability to perform as Wal-Mart store managers by already being employed in comparable positions at Wal-Mart’s comparator firms. The alternative way of calculating shortfall, in which Wal-Mart is compared against the level of utilization at its comparators excluding itself, corresponds to an assumption that the sources from which Wal-Mart could expand its number of female store mangers are broader, so that Wal-Mart shortfall could be reduced without reducing the number of women managers at the comparator firms. These more diverse sources might include hiring new women Wal-Mart managers from outside the large chain retail industry. Most prominently, however, it includes promoting women to Wal-Mart management from among non-managerial Wal-Mart employees. According to the Wal-Mart annual report 2002, page 6, “More than 60 percent of our store managers were ‘grown’ through the Wal-Mart organization, starting as store Associates.”
[19]See paragraph 24 above.
[20]As previously noted in footnote 5, the term “in-store managers” here refers not to the one individual per store bearing the title store manager but all salaried Wal-Mart in-store managerial personnel, including store managers, co-managers and assistant managers.
[22]Deposition of Coleman Peterson taken December 18, 2002, p. 125 11:16 9 – 11:17 15.
[23]Deposition of Coleman Peterson taken December 18, 2002, p. 128, 11:20 9 – 11:20 16.
[24]Deposition of Coleman
Peterson taken December 18, 2002, p. 111 10:46 2 – 10:47 24. These discussions appear to have focused on
whether department managers in the retail industry should be classified as
hourly or salaried employees for purposes of the wages and hours laws, rather
than specifically on their classification in EEO-1 reports. The two issues are potentially related but
not identical.
[25]Deposition of Coleman Peterson taken December 18, 2002, p. 124 11:15 11-22. This comparison appears in a memorandum tiled “Equal Opportunity at Target,” dated December 20, 2000 from Cole Peterson to the Wal-Mart Executive Committee (WMHO512710).
[26]Deposition of Coleman Peterson taken December 18, 2002, p. 126 11:18 17-11:18 23.
[27]According to www.hoovers.com, among Target Corporation’s 1,475 retail stores, about 75 percent are discount stores (Target and related brands), while 25 percent are department stores (Mervyn’s and Marshall Fields).
[28]Deposition of Coleman Peterson taken December 18, 2002, p. 106 10:40 8-17.
[29]In terms of managers per store, one chain’s 9.1 and other’s 12.3 put them in the same category as Wal-Mart’s 10.1. In term of total employees per store, one chain’s 195 employees and the other’s 232 put them in the same category as Wal-Mart’s 279. In terms of managers per 100 employees, one chain’s 4.9 and other’s 5.6 put them in the same category as Wal-Mart’s 3.8.
One reason Wal-Mart has fewer in-store managers per store than both these comparators is that Wal-Mart’s managerial system assigns a much higher proportion of the firm’s managers to its headquarters in Bentonville, Arkansas than their comparators place at their headquarters; see Section X, below.
[30]1.1
percent of 29,462 in-store managers is 324 persons. 4,004 + 324 = 4,328.
[32]Wal-Mart had no stores in the District of Columbia.
[33]If the difference in female managers between Wal-Mart and its comparators were due to chance alone, we would expect that 50 percent of Wal-Mart stores would have a shortfall and 50 percent would not. What is the probability that, in a sample of 2,909 stores, we would observe a number as high as 79.5 percent if the true probability of a shortfall was 50 percent? That probability is so small that it would be written with 236 zeros after the decimal point, corresponding to 32.9 standard deviations. Thus, this difference is clearly not due to chance alone.
[34]In the only exception, Vermont, Wal-Mart operated four stores, the smallest number in any state in the nation. Shortfall of women in management arose in two of the four stores, and the statewide representation of women among managers exceeds the expected number by a fraction of one person.
[35]Wal-Mart’s headquarters in Bentonville, Arkansas is located within the Fayetteville-Springdale-Rogers Metropolitan Area. However, because the only establishments included in the graphs in Attachment J are retail stores; data from headquarters are not included.
[36]The four graphs correspond
to the following numbers:
| Census Region | Illustrative Local Area | Wal-Mart Stores | Comparator Stores | Female % of Managers - Walmart | Female % of Managers - Comparator | Conservative Shortfall |
|---|---|---|---|---|---|---|
| Northeast | Portland, ME | 3 | 7 | 17%-40% | 33%-78% | 59 |
| Midwest | Eau Claire, WI | 2 | 9 | 38%-45% | 50%-88% | 4.3 |
| South | Fayetteville, AR | 13 | 6 | 0%-40% | 20%-91% | 19.8 |
| West | Pueblo, CO | 3 | 8 | 11%-30% | 40%-83% | 7.6 |
[37]An organization’s corporate culture is defined as the interdependent system of beliefs, values, and ways of behaving that are common to a workplace. Less formally, it is simply, “the way things are done around here.” See Marc Bendick, Jr., and Mary Lou Egan, Changing Workplace Cultures to Reduce Employment Discrimination, A Presentation to the Conference on Low Wage Workers in the New Economy, Washington, DC, May 2000, at www.bendickegan.com.
[38]The percentage gap between Wal-Mart and its comparators took 24 years to reduce by the17.9 percentage points between 39.9 percent and 22.0 percent. To eliminate the remaining 22.0 percentage points at this closure rate of .75 percentage points a year would require about 29.5 years.
[39] (22.0-2.24)/22.0 = 89.8.
[40]Attachment D is based on Wal-Mart employees company-wide, not just in retail stores.
[41]Assuming the gap between Wal-Mart and its comparator continues to close at the same rate as between 1975 and 1999, the remaining gap will require about 22 years to eliminate.
[42]These alternative estimates -- especially those taking account of internal promotional paths to in-store management -- may be particularly relevant in calculating economic damages.