Pay Secrecy and the Gender Wage Gap in the United States

Abstract: Legislators and advocates claim that pay secrecy perpetuates the

gender wage gap and that the FLSA should be amended to outlaw these practices. Using

a difference-in-differences fixed effects human capital wage regression, I find that

women with higher education levels who live in states that have outlawed pay secrecy

have higher earnings, and that the wage gap is consequently reduced. State bans on pay

secrecy and federal legislation to amend the FLSA to allow workers to share information

about their wages may improve the gender wage gap, especially among higher-educated

women.

Marlene Kim, Professor

Department of Economics

University of Massachusetts Boston

Boston, MA 02478

Voice: 617/287-6954

Fax: 617/287-6976

[email protected]

Acknowledgements: Many thanks to the US Women’s Bureau staff and former Director for

their support, the staff in many of the states that were interviewed for this article, and to the

legislators and activists in these states for their helpful information and comments. Thanks to

Reagan Baughman, Michael Carr, Bill Dickens, Andrew Houtenville, Ju-Chin Huang, Emily

Weimers, Catherine Weinberger, Paul Wolfson, Linus Yamane and the University of New

Hampshire’s Economics Department Seminar for their help, insights and suggestions. Thanks

also to Jesse Rothstein and two anonymous reviewers who helped improve, tighten, clarify, and

find all the typos and errors in this paper. All remaining errors are the author’s.

There has been a sea-change in the workforce since the Fair Labor Standards Act (FLSA)

was passed in 1938. One of the more notable changes is that few women worked for pay when

the law was passed, while today, most do.

To address the needs of the modern workforce,

legislation has amended the FLSA to account for these changes. These include the Equal Pay

Act of 1963, which mandates equal pay for equal work, regardless of sex. But there have been

other proposed amendments that purport to meet the needs of working women. This paper

examines one of the least known policy proposals -- amending the FLSA in order to outlaw pay

secrecy.

In this paper, I discuss pay secrecy, its extent, proposed legislation to amend the FLSA in

order to prohibit these practices, and the likely effect of amending the FLSA to outlaw pay

secrecy on earnings and the gender wage gap. Using a natural experiment of states that prohibit

pay secrecy compared to those that do not, I examine whether states that outlaw pay secrecy

reduce the gender wage gap. I find that in states that have outlawed pay secrecy, earnings for

college educated women are greater, reducing the gender pay gap.

Pay Secrecy, Its Prevalence, and the Law

Pay secrecy includes rules, policies and practices that prohibit workers from discussing or

sharing information about their earnings (Gely and Bielman, 2003; Bielman and Gely, 2004;

Edwards, 2005). These include formal policies written in employee handbooks and informal

policies conveyed to workers sometime during their employment (Gely and Bielman, 2003).

Advocates and legislators who have proposed to amend the FLSA by outlawing pay secrecy

argue that pay secrecy perpetuates the gender wage gap: if women don’t know what other

workers are paid, gender discrimination in earnings can continue.

Lilly Ledbetter illustrates this argument. For twenty years, Ledbetter was the only female

supervisor among sixteen male supervisors for Goodyear Tire in Alabama. She earned less than

all these men, including some who had less seniority. Yet she did not know that she was

underpaid because her workplace prohibited employees from discussing their pay. It was only

after she received an anonymous note that revealed the earnings of some of these male managers

that she realized she was underpaid (Greenhouse, 2007; National Women’s Law Center, 2013).

Ledbetter is not alone. In the United States, most employees are prohibited from

discussing their earnings. According to a survey conducted in 2010, 61 percent of private sector

workers are either formally forbidden or informally discouraged from discussing their pay with

their colleagues (Institute for Women’s Policy Research, 2010). About one-third of private

sector workers are explicitly forbidden from doing so because of formal rules or policies not to

discuss their pay, with another third informally discouraged from doing so (Bielman and Gely,

2004; Colella et al., 2007; Edwards, 2005; Gely and Bielman, 2003; Bamberger and

Beloglolovsky, 2010).

Yet, for most of these workers, these pay secrecy policies are illegal. Section 7 of the

National Labor Relations Act (NLRA) protects workers in “concerted activities for the purpose

of collective bargaining or other mutual aid or protection.”

The National Labor Relations Board

(NLRB), which enforces the NLRA, has consistently ruled that discussions of wages are a form

of “protected concerted activity” and thus prohibiting discussions of earnings is illegal (Bielman

and Gely, 2004; Gely and Bielman, 2003). It views sharing information about pay as integral to

organizing workers into unions, even if a union organizing campaign is not in progress.

Dissatisfaction due to low wages is the grist on which concerted activity feeds. Discord

generated by what employees view as unjustified wage differentials also provides the

sinew for persistent concerted action. (Jeannette Corp v. NLRB, 532 F.2d 916, 919 (3d

Cir. 1976), cited in Gely and Bierman, 2003: 128).

Thus, the NLRB has ruled very broadly that employers are in violation of the law if they

discourage or prohibit workers from sharing information about their pay (Bielman and Gely,

2004; Gely and Bielman, 2003). Such prohibitions include informal or formal pay secrecy

policies, even if not enforced. Policies that in any way restrict employees from sharing

information about their earnings is forbidden, including employers’ preventing employees from

opening their paychecks among other workers. The NLRB has ruled that these prohibitions

hamper employee’s rights under the NLRA (Bielman and Gely, 2004; Gely and Bielman, 2003).

These protections extend to both unionized and non-union workers.

Why is pay secrecy so widespread if it is illegal? First, many employees are not covered

by the NLRA: supervisors and managers are excluded. Hence if Lilly Ledbetter had violated

her company’s policy and had asked how much her male counterparts earned, she could have

been fired with no legal recourse. Second, most employees do not know that pay secrecy is

illegal (Gely and Bierman, 2003). Third, the penalties from violating the NLRA are mild, so that

employers commonly break the law (Gely and Bierman, 2003; Freeman and Medoff, 1984;

Brofenbenner, 1984). Fourth, many employees favor pay secrecy policies, in part because the

culture in the US dictates that one does not discuss one’s earnings (Colella et al., 2007).

Fifth, employers favor pay secrecy (Gely and Bierman, 2003). They believe that morale

and productivity would decline, relationships among workers would be strained, and conflict will

occur if workers knew how much others earned (see Colella, 2007, for a review of this

literature). As one employer states, “jealously and strife among employees” would prevail if

employees knew what others were paid but not the justifications for these wage differentials

(Bierman and Gely, 2004). Pay secrecy may also prevent employees from leaving their

employers for better paid companies (Danziger and Katz, 1997; Colella, 2007). Thus, since most

managers believe that pay secrecy is a good policy, the unwritten rule in most workplaces is that

employees must keep their mouths shut about their pay (Bierman and Gely, 2004; Gely and

Bierman, 2003).

Pay Secrecy and the Gender Wage Gap

Some scholars argue that pay secrecy can contribute to the gender pay gap (Eisenberg,

n.d.) since it can “avoid perceptions of unfairness when pay inequities do exist and can minimize

claims of discrimination” (Colella et al., 2007). Hence, lack of knowledge about what others

earn can contribute to the existence of pay discrimination and thus to the gender wage gap.

Eisenberg (n.d.) argues that pay transparency is important so that the employer is motivated to

establish fair pay systems, and so that employees can monitor, complain about and remedy any

unfair pay. Although market wages are supposed to discipline both workers and employers in

compensation, much information is unavailable to workers (such as what their colleagues earn),

so market discipline may not work. Instead, without standard salary scales, and with salaries

based on previous salary, the gender pay gap can perpetuate (Eisenberg, n.d.). In addition, with

salaries open to negotiation, women may be underpaid because they do not negotiate as hard as

men for higher salaries, in part because if they do, they are seen as too demanding and

unpleasant to work with. Thus, for a woman, negotiating over salary may lead to loss of a job

offer. Men, however, are able to negotiate for higher salaries without such adverse consequences

(Eisenberg, n.d.).

With salaries based on previous earnings, and with women earning less than men on

average and unable to negotiate for higher pay, women will continue to earn less than men

(Eisenberg, n.d.). Salary transparency, however, would enable women to know what others are

earning and to negotiate for similar pay. It would also allow women to know if they are

underpaid compared to similar men and to correct these disparities, either informally or through

the court system (Eisenberg, n.d.).

Federal Legislation to Outlaw Pay Secrecy: Amending the FLSA

Recognizing the importance of sharing information on pay for women’s earnings,

Congress has introduced 22 pieces of legislation that would amend the FLSA to outlaw pay

secrecy. None of these have passed. These include the Paycheck Fairness Act (introduced 17

times), the Fair Pay Act of 2001 and 2011, the Enhancing Economic Security for America's

Working Families Act in 2001, the Fairness and Individual Rights Necessary to Ensure a

Stronger Society: Civil Rights Act of 2004, and the Wage Awareness Protection Act in 2000.

The Wage Awareness Protection Act was the only bill in which pay secrecy was the sole content;

all the other bills included broader legislation to reduce the gender wage gap (such as mandating

comparable worth or harsher penalties for findings of discrimination). These bills would have

amended Section 15 of the FLSA—the Prohibited Acts—to make unlawful any policies or

actions against employees who share information about their earnings.

For example, the Wage Awareness Protection Act would have prevented employers from

taking any adverse employment action against employees who inquire about or discuss wages, or

to “make or enforce a written or oral confidentiality policy that prohibits an employee from

inquiring about, discussing, or otherwise disclosing the wages of the employee or another

employee.” (Bierman and Gely, 2004: 186). Various Paycheck Fairness Acts introduced in

Congress, most recently in January 2013, would amend Section 15 so that it would be unlawful

for any person:

“to discharge or in any other manner discriminate against, coerce, intimidate, threaten, or

interfere with any employee or any other person because the employee inquired about,

discussed, or otherwise disclosed the wages of the employee or another employee.” (See

S. 3772, 9/13/2000; S. 182 4/19/2005; HR 12 1/6/2009).

The motivation for federal legislation was to reduce the gender wage gap (see FDCH Press,

2009; Gely and Bierman, 2003). For example, Congresswoman Eleanor Homes Norton states

that it is important

To keep employers from gagging employees by threatening them with sanctions for

freely discussing and learning the wages of their coworkers, enabling women to engage

in self-help to demand wage increases where appropriate….(cited in Gely and Bierman,

2003: 132)

Because repeated attempts to outlaw pay secrecy at the federal level failed, on April 8, 2014,

President Obama issued an Executive Order banning pay secrecy for federal contractors. Only

one out of five workers are covered by this provision, however,

but managers and supervisors

are included.

State Laws on Pay Secrecy

Seven states have passed their own laws banning pay secrecy: Michigan (1982),

California (1984), Colorado (2008), Illinois (2004), Maine (2009), Vermont (2005) and most

recently, New Jersey (2013). These laws vary in terms of which employees are covered and

under which circumstances (see Table 1). In general, states that prohibit pay secrecy through

their labor laws (such as California, Michigan, and Colorado) commonly include only private

sector workers, with Colorado further limiting employees to those covered by the NLRA (so that

supervisors and managers are excluded from these protections). In contrast, states that prohibit

pay secrecy in their Equal Employment laws, such as Illinois, New Jersey and Maine, do so for

both public and private sector workers but only when employees are investigating unequal pay

claims. Vermont is the only exception to this pattern, covering both private and public sector

workers without limiting them to investigating unequal pay claims. Vermont also allows

workers to file claims anonymously, by sending in employee manuals that reveal an employer’s

pay secrecy policy. (See Table 1 for a summary of state pay secrecy laws.) Although New York

State has introduced legislation several times that would ban pay secrecy, this legislation has

never passed.

In states without laws prohibiting pay secrecy, little is known about the extent to which

workers face reprisals for sharing information about their earnings. However, research suggests

that workers don’t seem to share information about their wages out of fear of such punishment

(Bierman and Gely, 2004; Gely and Bierman, 2003). In the states that have outlawed pay

secrecy, not much is known about the extent to which workers avail themselves of these laws or

know about them. Interviews in some states indicate that few charges are filed regarding pay

secrecy violations (Deputy Labor Commissioner, 2011; Hernandez, 2011; Moy, 2011; Bass,

2011; Maine Department of Labor, 2011), and that few workers may know about these laws

(Bass, 2011; Moy, 2011), so that more needs to be done about publicizing them (Bass, 2011).

However, others discuss workers who, knowing that they were protected by these laws,

inquired about their colleagues’ pay, and when they discovered they were underpaid because of

their gender, they complained to their human resource departments and demanded higher pay

(Meric, 2011; Everett, 2011). Of course, it could be that in some states the laws are not

publicized adequately and that many workers do not know about them, but that those who are

informed use the laws to investigate their pay and remedy any discrepancies, if warranted.

** Table 1 about here **

The motivation for passing these state laws was the same as with the federal proposals--to

close the wage gap. Legislators and activists claim that for women’s earnings to increase,

women need to be able to discover if they are underpaid (Hayden, 2011; Meric, 2011, Bass,

2011; Harris, 2011; Donovan, 2011). As one advocate elaborates:

“As part of pay equity, workers need to know how their pay compares to other workers in

order to understand if they are paid fairly, and if not, take action. Few workers know that

sharing their wages is a “concerted activity” protected by the NLRA. We wanted a clear

statement in Colorado law so that employees knew they could share wage information

without reprisals.” (Meric, 2011)

For Maine, the Lilly Ledbetter case was also a motivating factor in passing legislation (Bass,

2011). But was Lilly Ledbetter the exception, or are women indeed underpaid like these

advocates and legislators believe--because women simply don’t know that they are paid less than

men?

Pay Secrecy and Wages: An Overview and the Data

If advocates and legislators are correct in that women feel empowered to inquire and

remedy any gender differences in pay, state laws outlawing pay secrecy would increase wages

for women and reduce the gender wage gap. I use the March Supplement of the Current

Population Survey (CPS), also known as the Annual Social and Economic Supplement, from

IPUMS, to investigate this. Data included 1977 to 2012 because prior to 1977, not all the

variables were available. Civilian workers between 25 and 64 years of age, who were wage and

salary workers, and who earned positive earnings (and worked positive numbers of weeks) were

included in the sample. With these restrictions, the sample contained over 2.1 million

observations: approximately 1.1 million men and 1 million women.

A cross sectional examination of the data in 2011 and 2012, when pay secrecy had been

outlawed in six of these seven states,

indicates that women’s wages are higher in states that

have outlawed pay secrecy—but men’s wages are higher, too. As Table 2 shows, men earn an

average of $30.56/hour in states that ban pay secrecy, compared to $28.45 in states without these

bans. Similarly, women earn $24.18 in states that ban pay secrecy, and a lower amount,

$22.10/hour, in states that do not have such bans (these differences are statistically significant).

Of course, there are two possible reasons for these outcomes. First, employers’ worst fears may

be realized when pay secrecy is illegal: wages may creep higher once employees know what

others earn, because employers increase wages that are too low. Second, however, is that

workers in these states are quite different, leading to higher wages.

*** Table 2 about here ***

Notice that the gender wage gaps persist whether one is in a state with or without laws

banning pay secrecy. These gender wage differences are all statistically significant at the 1%

level. Most likely, these persistent gender gaps result from many factors (more on this later).

The gender wage gap is smaller in states that have banned pay secrecy, although these

differences are not statistically significant. The statistical insignificance may be due to the low

sample size (six each year) or to other factors that are influencing wage rates (more on this

below).

Obviously, however, many factors can explain these findings. Workers living in states

that have banned pay secrecy may be different than workers in other states, for example, with

higher education or work experience levels. States that have banned pay secrecy may also have

passed other laws that can increase wages, such as higher minimum wages or laws that allow

advocating for labor unions. As Table 3 shows, those living in states that have outlawed pay

secrecy are more likely to live in metropolitan areas and in central cities. These states also have

fewer African American workers, more Hispanic and Asian American workers, and fewer

workers with less than high school degrees. These differences are likely to explain the higher

wage rates of both men and women in these states.

** Table 3 about here **

Similarly, the smaller (albeit insignificant) wage gaps in states with laws that ban pay

secrecy may be due to two entirely different factors. First, it may be true that states with pay

secrecy laws increase the pay for women and thus lower the gender pay gap because women are

informed if they are underpaid compared to men and remedy this problem. In this scenario, the

law is working as proponents claim.

A second explanation, however, may be self-selection. States that pass pay secrecy laws

may care more about non-discrimination and thus have greater enforcement about workplace

discrimination. They may also be more aggressive in other areas that affect the gender wage

gap, such as in affirmative action and gender neutral education. In this scenario, the gender pay

gap may be lower compared to states without such laws, but this may be due to other factors

besides the pay secrecy law, including a culture more supportive of working women and ending

employment discrimination, a legislature more supportive of women’s rights, stronger laws on

gender pay discrimination, or stronger enforcement of such laws. Thus, banning pay secrecy

may not be what increases pay for women, but rather a larger culture that supports women. In

this scenario, passing pay secrecy laws in other states would not lower the gender pay gap;

instead, changing the social climate around women’s pay may be effective.

Research Methods

In order to examine these competing explanations, control for factors that may explain

these patterns, and increase the sample size, I ran a difference-in-difference-in-difference (DDD)

human capital wage regression. Human capital regressions are commonly run to examine gender

wage differentials, since this allows for key variables to be measured separately from the effect

of other characteristics that affect wage rates, such as higher education levels, work experience,

race, and ethnicity (see Kim, 2013). Difference-in-difference-in-difference (DDD) analysis is

used to examine the effect of policy changes on specific groups of people—in this paper, women

(see Pischke, 2005; Imbens and Woodridge, 2007). Following Pischke (2005), the DDD

specification I ran was:

1. Ln(w

i,s,t

) = α

i,s,t

+ α

i,s,t

+ α

(ps

s,t

i,s,t

) + α

i,s,t

) + α

i,s,t

)

+ α

)+ e

i,s,t

For each individual i in state s and time t, the dependent variable is the natural log of the

real (in 2012 dollars) hourly wage, w

i,s,t

Independent variables (X

i,s,t

) were included to account

for factors other than gender that may affect wages. These include educational attainment, race,

potential work experience, its square, living in a metropolitan area or central city, and marital

status, all which are typical controls in wage regressions. The number of children, the presence

of children under age 5, and broad industry and occupational dichotomous variables were also

included, since differences in wages by gender can be explained by these.

A female dummy

variable, f

i,s,t

, is also included to capture the effect of underpaying women workers, controlling

for all these other factors.

A matrix of fixed effects by year (I

) controls for economic conditions that vary over

time, such as the business cycle. A matrix of state effects (I

) controls for variations in the cost

of living, economic conditions, business climates, and state laws. This matrix interacted with the

female dummy variable (f

i,s,t

) captures political climates toward women or other state laws

that may affect the pay for women and the gender pay gap across states. A dichotomous

variable, ps

s,t

,indicates those living in a state s in which pay secrecy laws were in effect in year t.

This pay secrecy variable is interacted with the female dummy variable (ps

s,t

i,s.t

) to measure the

effect of outlawing pay secrecy on women’s wages in particular. The coefficient of interest is α

Interacting the female dummy variable with year effects (f

i,s,t

) controls for trends

particular to women, such as the general tendency for the wage gap to decline over time.

Interacting state and year effects (I

) controls for unobservable factors by state-year cells. The

last argument in equation 1 is the individual specific error term. All reported standard errors are

clustered at the state level.

Pay secrecy probably affects some workers more than others. Those with higher

educations are more likely to work in professional jobs in which negotiating pay is more

common. These jobs are more likely to have wider variation in skill levels and that also allows

for more discretion in pay-setting by employers. Research confirms that college educated

workers have higher variation in their pay (Lemieux, 2006; Chay and Lee, 2000), and that this

variation has increased over the time period examined in this study (Lemieux, 2006). Thus,

separate regressions are also run on those with college degrees and those without, to see the

effects of outlawing pay secrecy by educational differences. The expectation is that laws

outlawing pay secrecy will have a greater effect on higher educated workers.

Definitions of the variables and their means and standard deviations are included in

Appendix A.

Research Results

To assess the effects of outlawing pay secrecy, Table 4 first shows the regression results

using a simpler difference-in-difference specification. Here, only human capital variables

(including gender) and state and year fixed effects are included, omitting all interaction

variables.

The policy variable is shown without any interactions with gender in order to see if

the policy had any effect at all. As specification #1 in this table indicates, the policy does not

seem to affect wages, as the results are statistically insignificant. Separate regressions by gender

also yield insignificant results.

This confirms that the higher average wages in the states that outlawed pay secrecy in

Table 2 results from the different characteristics of workers in these states: their higher

education levels, more likelihood of living in metropolitan areas, and fewer African-Americans.

Enacting pay secrecy laws does not appear to increase wages for all employees, as employers

feared.

The next row (#2) in Table 4 keeps the previous specification, but adds the policy

interacted with being female. Here, the policy by itself is once again insignificantly different

from zero. The coefficient on the interaction term of this policy and being female is positive and

statistically significant, however. On average, women’s wages increased 4-5 percent after pay

secrecy laws were passed, leading to an increase in the gender wage ratio (the ratio of women’s

to men’s pay) of 3 to 3.5 percentage points.

The next results (#3) uses the DDD specification in equation 1 for all education levels.

Notice that in this specification, the effect on outlawing pay secrecy for women is much smaller.

Women’s wages increase by only one percent, and this is not even statistically significant for

full-time year-round workers. But when examined by education level, the results are quite

different: Wage increases for women with low education levels, although positive, are

statistically insignificant (see #4). In contrast, women with college degrees increased their pay 3

percent. Thus these results show some support for the claims of advocates that laws outlawing

pay secrecy increase pay for women, especially those with college degrees.

** Table 4 about here **

As a check on these results, I use a method to measure race or gender discrimination (see

Verdugo, 1992; for examples, see McGuire and Reskin, 1993; Green and Ferber, 2005; Yamane,

2002; Mar 2000; Kim, 2009) and calculate the gender wage gap in states with and without pay

secrecy laws in effect. I then use a difference-in-difference (DD) model to see if this gender

wage gap is lower in states in which pay secrecy is outlawed.

To do this, first, a wage regression was run only for men from 1977 to 2012, using the

same human capital controls, X, in equation 1 and state fixed effects:

2. Ln(w

i,s,t

) = β

i,s,t

+ β

+ e

i,t

This regression provides estimates of the coefficients for men, β

and β

, after

controlling for the same education attainment, geography variables, and demographic variables

as in equation 1. Using these estimated coefficients, a predicted wage, p, was estimated for every

women by using the characteristics of each of the women (the X’s) but the estimated coefficients

(β

and β

) from the regression of men:

3. p

i,s,t

= β

i,s,t

+β

This predicted wage is the wage that women would earn if they had the same returns to

their characteristics as do men. It is often a measure of a non-discriminatory wage—i.e. what

women would earn in the absence of discrimination, because it measures their earnings as if they

were treated as men (Verdugo, 1992). The actual real log wage of each woman was then

subtracted from this predicted wage to estimate the amount of discrimination in wages each

woman faced—i.e. what they would have earned as a man minus what they actually did earn:

4. D

i,s,t

= p

i,s,t

– ln (w

i,s,t

)

where i is the ith woman in year t and state s. This discrimination variable, D, measures the

adjusted gender wage gap (adjusted for human capital and other characteristics). I used a

difference-in-difference specification to examine if laws outlawing pay secrecy policies reduced

this wage gap:

5. D

i,s,t

= γ

pse

+ γ

s,t

+ e

i,s,t

Whether or not a state ever had prohibited pay secrecy at any time is now a control (pse), as

these states may be different from those that never passed such laws. Year effects are included

in this specification. The coefficient of interest is γ

, which measures the effect of the policy on

lowering the pay gap for women.

The first three rows in Table 5 show the results from equation 5. Notice that although the

adjusted wage gap, or D, is indeed lower for women in states in which pay secrecy laws are in

effect, these are not statistically significant. But these effects are once again very different by

education level. Those with college degrees receive 5-6 percent reductions in the gender wage

gap. In contrast, those without college degrees have statistically insignificant changes in the

wage gap.

A variation of this specification was run with state fixed effects instead of the pse

variable in equation 5. The measure of discrimination, D, was constructed similarly in equations

2 and 3 but using year instead of state fixed effects.

Rows 4-6 in Table 5 indicate that in this

specification, the effects are even greater, with reductions of the adjusted wage gap of 8-11

percent. The results by education level are also consistent with previous findings: Women with

college degrees face much higher declines in the gender wage gap—12-15 percent. In

comparison, women without such degrees experienced 6-8 percent decreases in the wage gap,

and these are statistically significant.

Taken together, these results show support that state laws that outlaw pay secrecy

increase earnings for women relative to men, especially among college educated women, and

that the gender wage gap falls among this population as well. Thus pay secrecy laws appear to

help women determine if they are underpaid compared to men and may be useful to reduce the

gender wage gap, especially among the higher educated.

Conclusion

When the FLSA was first passed, gender differences in wages were accepted and

perfectly legal (Kim, 1999). But to be relevant, the FLSA may need to be amended from time to

time in order to meet the needs of the current workforce. Today, the labor force is vastly

different than in 1938, in that it is comprised of many more women. Currently, women earn 82%

of what men earn,

and although the gender wage gap has narrowed over several decades,

advocates feel that with women now attending college at higher rates than men, the gender wage

gap should be much smaller.

Although much research has been conducted on various policies to see how pay can

increase for women, no one has examined the effect of pay secrecy on women’s earnings.

Using a natural experiment of states that have outlawed employment practices that prevent

workers from discussing pay, I find that wages are higher for women in states that have outlawed

pay secrecy, especially among those with college degrees. These women increase their earnings

3 percent in states that have outlawed pay secrecy. Additional analyses finds that those with

college degrees reduce the gender wage gap by 5-6 percent, or by 12-15 percent, depending on

the specification and population of workers examined.

These results provide support for state laws outlawing pay secrecy. Thus prohibiting pay

secrecy in other states is likely to benefit college-educated women, increasing their pay and

lowering the gender wage gap. National legislation has been introduced to amend the FLSA,

including the Paycheck Fairness Act, to outlaw pay secrecy on a national level, but this has never

been passed by Congress. The results in this paper indicate that such legislation is likely to

improve women’s pay.

Thus, outlawing pay secrecy in the Fair Labor Standards Act and in state legislation

should be considered as another tool to lower the gender wage gap among higher educated

women. In this way, the Fair Labor Standards Act can be amended to address a problem that

was not seen as critical when it was first passed: the underpayment of women workers.

References

Bamberger, Peter and Elena Belogolovsky. 2010. “The Impact of Pay Secrecy on Individual

Task Performance.” Personnel Psychology 63 (winter): 965-996.

Bass, Peaches. 2011. Interview with Peaches Bass on Michigan pay secrecy law (November).

Bierman, Leonard and Raphael Gely. 2004. “ ’Love, Sex and Politics? Sure. Salary? No

Way’: Workplace Social Norms and the Law.” Berkeley Journal of Employment and

Labor Law, 25(1): 167-91.

Brofenbrenner, Kate. 1994. “Employer Behavior in Certification Elections and First-Contract

Campaigns: Implications for Labor Law Reform.” In Restoring the Promise of American

Labor Law, edited by Sheldon Friedman, Richard W. Hurd, Rudolph A. Oswald and

Ronald L. Seeber, pp. 75-89. NY: Cornell University (ILR) Press.

Budig, Michelle J. and Paula England. 2001. “The Wage Penalty for Motherhood.” American

Sociological Review 66 (April): 204-255.

Chay, Kenneth Y. and David S. Lee. 2000. “Changes in Relative Wages in the 1980s: Returns

to Observed and Unobserved Skills and Black-White Wage Differentials.” Journal of

Econometrics. 99 (November): 1-38.

Colella, Adrienne, Ramona L., Paetzold, Ashghar Zardkoohi, and Michael J. Wesson. 2007.

“Exposing Pay Secrecy.” Academy of Management Review, 32 (January): 55-71.

Correll, Shelly J., Stephen Benard and In Paik. 2007. “Getting A Job: Is there A Motherhood

Penalty?” American Journal of Sociology 112 (March): 1297-1338.

Danziger, Leif and Eliakim Katz. 1997. “Wage Secrecy as a Social Convention.” Economic

Inquiry, 35 (January): 59-69.

Donovan, Johanna Leddy. 2011. Phone interview with Johanna Leddy Donovan, one of the

sponsors of the legislation in Vermont (November).

Eisenberg, Deborah Thomson. N.d. “Money, Sex and Sunshine: A Market-Based Approach to

Pay Discrimination.” Mimeo. University of Maryland School of Law.

Everett, Sandy. 2011. Interview with Sandy Everett about Vermont’s pay secrecy laws.

Attorney General’s Office of Civil Rights.

Federal Reserve Board of St. Louis, 2013. “Labor Force Participation Rate –Women”

http://research.stlouisfed.org/fred2/series/LNS11300002. Accessed September 17, 2013.

Freeman, Richard and J. Medoff. 1984. What Do Unions Do? New York: Basic Books.

Futrell, Charles M. and Omer C. Jenkins. 1978. “Pay Secrecy Versus Pay Disclosure for

Salesmen: A Longitudinal Study.” Journal of Marketing Research 15 (May): 214-9.

Gely, Raphael and Leonard Bierman. 2003. “Pay Secrecy/Confidentiality Rules and the

National Labor Relations Act.” University of Pennsylvania Journal of Employment and

Labor Law, 6(1): 121-56.

FDCH Press. January 8, 2009. “Senator Clinton Reintroduces the Paycheck Fairness Act.”

Greenhouse, Linda. 2007. “Justices’ Ruling Limits Lawsuits on Pay Disparity,” New York

Times, May 30, 2007.

Green, Carole A. and Marianne A. Ferber. 2005. “Do Detailed Work Histories Help to Explain

Gender and Race/Ethnic Wage Differentials?” Review of Social Economy 77 (March):

55-85.

Hayden, Tom. 2011. Emails from Tom Hayden, November 30, 2011 and December 1, 2011.

Institute for Women’s Policy Research. 2010. Pay Secrecy and Paycheck Fairness.

Washington, DC.

Hegewisch, Ariane and Hanna Liepmann. 2013. “Occupational Segregation and the Gender

Wage Gap in the US.” In Handbook on Research on Gender and Economic Life, edited

by Deborah M. Figart and Tonia L. Warnecke, pp. 200-217. Northampton, MA: Edward

Elgar.

Hernandez, Nancy. 2011. Interview with Nancy Hernandez regarding Illinois’s Pay secrecy

law.

Imbens, GM and Jeffrey M. Wooldridge. 2007. Lecture Notes 10.

www.nber.org/WNE/Lect_10_diffindiffs.pdf.

Kim, Marlene. 2013. “Race and Ethnicity in the Workplace,” In Handbook on Research on

Gender and Economic Life, edited by Deborah Figart and Tonia Warneke, pp. 218-235.

Northampton, MA: Edward Elgar.

--- 2009. “Race and Gender Differences in the Earnings of Black Workers.” Industrial Relations,

48 (July): 466-88.

--- 1999. “Inertia and Discrimination in the California State Civil Service.” Industrial Relations,

38 (January): 46-68.

Lemieux, Thomas. 2006. “Increasing Residual Wage Inequality: Composition Effects, Noisy

Data, or Rising Demand for Skills?” American Economic Review. 96 (June): 461-498.

Maine Department of Labor. 2011. Interview on pay secrecy law.

Mar, Don. 2000. “Four Decades of Asian American Women’s Earnings: Japanese, Chinese and

Filipino American Women’s Earnings 1960-1990.” Contemporary Economic Policy 18

(April): 228-237.

Matthew, Edward. 2005. “The Law and Social Norms of Pay Secrecy,” Berkeley Journal of

Employment and Labor Law, 26(1): 41-63.

McGuire, Gail M. and Barbara F. Reskin. “Authority Hierarchies at Work: The Impacts of Race

and Sex.” Gender and Society 7(December): 487-506.

Meric, Linda. 2011. Interview with Linda Merick on Colorado law (November).

Moy, Doris. 2011. Interview with Doris Moy on Illinois law (November).

National Women’s Law Center. 2013 Lilly Ledbetter Fair Pay Act.

http://nwlc.org/resource/lilly-ledbetter-fair-pay-act-0.

O’Neill. June. 2004. “The Gender Gap in Wages, Circa 2000.” American Economic Review 93

(May): 309-14.

Pische, Jorn-Steffen. 2005. “Empirical Methods in Applied Economics.” Lecture Notes.

London School of Economics. http://econ.lse.ac.uk/staff/spischke/ec533/iv_handout.pdf

Smith, Ralph E. 1979. “The Movement of Women into the Labor Force.” In The Subtle

Revolution: Women at Work, edited by Ralph E. Smith, pp. 1-29. Washington DC: The

Urban Institute.

Verdugo, Richard R. 1992. “On the Cost of Being a Minority Worker, 1972-1987.” Social

Science Quarterly 75(September): 663-673.

US Bureau of Labor Statistics. 2013a. Economic News Release. Table 1. Median usual weekly

earnings of full-time wage and salary workers by sex, quarterly averages, seasonally

adjusted. http://www.bls.gov/news.release/wkyeng.t01.htm. Accessed December 16,

2013.

---. 2013b. Table 2, Employment Status of the Civilian Noninstitutional Population 16 years

and over by Sex, 1972 to Date,” http://www.bls.gov/cps/cpsaat01.pdf, accessed

September 17, 2013.

---. 2012. Highlights of Women’s Earnings in 2011. Report 1038.

www.bls.gov/cps/cpswom2011.pdf.

Yamane, Linus. 2002. “Native-Born Filipina/o Americans and Labor Market Discrimination.”

Feminist Economics 8 (July): 125-144.

Table 1. State Legislation on Pay Secrecy

State

Date

Scope of Law Workers Covered

Passed Effective

________ ____________________ __________________

California 1984 Jan. 1985 Policies and retaliation;

Wage disclosure only

Private sector

Colorado April

2009

August

2008

licies and retaliation;

wage inquiry, disclosure,

Comparisons, discussions

Private sector workers

covered by the NLRA

Illinois May

2003

Jan.

Cannot pro

hibit wage

Disclosure or inquiry but

Only when exercising Equal

Pay laws.

private and public sector

Maine Jun

2009

Sept.

2009

Cannot prohibit wage

disclosure or inquiry but

Only when exercising Equal

Pay laws

private and public sector

Michig

an 1982 March

1983

Policies and retaliation;

wage disclosure only

private sector

Vermont 2005 July

2005

Policies

and retaliation;

wage disclosure only

private and public sector

New Jersey 2013 2013 Retaliation for wage sharing

And inquirie

Investigation of discrimination

only

Table 2: Hourly Earnings for Full-Time Year-Round Workers, 2011-12

(Standard errors are in parentheses)

_________________In the States

_________________

Banning Pay Secrecy No Ban in States

A. Hourly Earnings

Men 30.15***

(.2906)

28.32

(.1291)

N 11,388 46,396

Women

24.20***

(.2355)

21.97

(.0962)

N 8892 39,053

B. Women’s Earnings Compared to Men’s Hourly Earnings

Ratio of Women’s/Men’s Earnings 0.7985

(.0111)

0.7783

(.00544)

N 12 90

Note: Data are from the Current Population Survey, March Supplement 2011-2012 data. Full-time year-

round workers. Wages are in 2012 dollars.

Statistically different between states with and without laws banning pay secrecy at the:

*10% level; **5% level; ***1% level

Table 3 Means by States With and Without Pay Secrecy Laws

No Pay Secrecy Laws

Pay Secrecy Outlawed

Mean

s.d.

Mean

s.d.

Ln real wage

2.8997632

0.6747491

2.9789798

0.6876831

number of

children

1.0198004

1.1864069

1.037327

1.2225702

metroarea

0.7686778

0.4216779

0.8935097

0.3084642

central city

0.2391953

0.4265922

0.3104914

0.4626948

Hispanic

0.0788398

0.2694886

0.1699448

0.3755843

female

0.4718978

0.4992096

0.4566614

0.4981182

black

0.1242244

0.3298374

0.0823992

0.2749719

Other race

0.038352

0.1920446

0.0814843

0.2735774

high school

0.220942

0.4148815

0.1851082

0.3883853

some college

0.2404129

0.4273342

0.2690724

0.4434777

college

0.2133254

0.4096556

0.2327202

0.4225654

advanced

degree

0.0690064

0.2534649

0.0704794

0.2559533

never marry

0.1726349

0.3779313

0.1989451

0.3992067

married

0.6610127

0.4733655

0.6384964

0.480436

sep/div/wid

0.1663524

0.3723966

0.1625585

0.3689624

pay secrecy

0.6238283

0.4844239

female*pay

secrecy

0.2862372

0.4520016

Work

experience

23.6157704

10.7683

23.2651797

10.6324109

Work

experience

673.660898

556.689613

654.316749

545.627074

Child under 5

0.1517406

0.3587692

0.1589455

0.3656252

Number of

observations

1,685,259

419,730

Table 4. Regression Results: Difference-in-Difference and DDD results

(cluste

red standard erro

rs are in parentheses)

All wor

kers

Full-tim

e, year-round

workers

_________

1. DD model: state and year F.E.

Policy

in effect (PS)

-0.00502

02338)

-.00710

(.02327)

2. Policy

in effect (PS)

-0.02735

02090)

-0.02408

(.01779)

Policy

in effect female *

0.04810*

(.02090)

0.04045**

(.0186)

DDD model

: (see equation #1 in text) in

cludes all interaction terms (female state,

female year, year state) **

Coeffici

ent on pay secrecy policy in effect

(PS) female for *

3. All edu

cation levels

0.01267*

(.00578)

0.0102

008)

4. No Col

lege Degree

0.00958

(

.0235)

0.0116

02195)

5. College

egree

0.02665*

(

.01318)

0.02469***

(.0095)

Note:

Regressions are on log of real (2012) h

ourly earnings, 1977-2012. Data are from March CPS,

Annual Social and Economic Supplement, 1977-2012 from IPUMS. Sample includes wage and salary

earners with positive earnings and weeks worked, between 25-64 years of age. See text for control

variables used. N=1,540,179 for the full sample (1,094,816 men and 99,605 women) and 2,083,421

(892,921 men and 647,258 women) for full-time year-round earners.

*significan

t at 10% level; **significant at 5

% level; *** significant at 1% level

Table 5: Coefficient Estimates for Pay Secrecy Laws in Effect (PS)

(Depe

ndent Variable is the Adjusted Gender Wage Gap; clustered standard errors are in parentheses)

All Wor

kers

Full-tim

e Year-Round

Model with pse variable in equation 5

(ever had pay secrecy outlawed in state)

1 All Education Levels -.0232

(.0232)

-.0217

(.0212)

2 No Co

llege Degree

-.0086

(.0209)

-.0089

(.0203)

3 College Degree

-.0578**

0273)

-.0485***

(.0209)

del with state fixed effects instead of pse variable

4. All Education levels -.1068***

(.0304)

-.0830***

(.0312)

5. No Col

lege Degree

-.0785***

(.0224)

-.0561**

(.0282)

6. College

Degree

-.1468**

0550)

-.1198***

(.0418)

Note: See t

ext for an explanation of variables used.

*Significant

at 10% level; **significant at 5% level; *** significant at 1% level

Appendix A. Variable Definitions and Means

All Workers

Full-Time Year-Round

Men

Women

Men

Women

Variable

Definition

Mean

(S.D.)

Mean

(S.D.)

Mean

(S.D.)

Mean

(S.D.)

Ln re

al wage ln of (2012) real wage

3.0599

(0.67)

2.7554

(0.651)

3.1196

(0.616)

2.836

(0.574)

number of

children

Number of own children 1.0053

(1.222)

1.0445

(1.162)

1.0529

(1.223)

0.9255

(1.100)

metroarea 1 i

f metro area

0.7962

(0.403)

0.7961

(0.403)

0.8026

(0.398)

0.8106

(0.392)

central city

1 if central city 0.2535

(0.435)

0.2565

(0.437)

0.2471

(0.431)

0.2677

(0.443)

Hispanic

1 if Hispanic

0.1104

(0.313)

0.0859

(0.28)

0.1048

(0.306)

0.0857

(0.2799)

Female

1 if female

n/a

.4685

(.499)

n/a

0.4171

(0.493)

Black

1 if black

0.1013

(0.302)

0.1306

(0.337)

0.0955

(0.294)

0.1407

(0.348)

other race

0.0476

(0.213)

0.0482

(0.214)

0.0469

(0.211)

0.0506

(0.219)

high school

1 if high school ed

0.2125

(0.409)

0.2136

(0.41)

0.2117

(0.409)

0.2193

(0.414)

some college

1 if some college

0.2315

(0.422)

0.264

(0.441)

0.2352

(0.424)

0.2702

(0.444)

College

1 if college degree

0.2207

(0.415 )

0.2141

(0.41)

0.2362

(0.425)

0.2258

(0.418)

advance degree

1 if advanced degree

0.0683

(0.252)

0.0704

(0.256)

0.0733

(0.261)

0.0777

(0.268)

never married

1 if never married

0.192

(0.394)

0.163

(0.369)

0.1704

(0.376)

0.1807

(0.385)

Married

1 if married

0.6877

(0.463)

0.6202

(0.485)

0.7171

(0.451)

0.5856

(0.493)

sep/div/wid

1 if separated, divorced, or

widowed

0.1203

(0.325)

0.2168

(0.412)

0.1124

(0.316)

0.2337

(0.423)

pay secrecy

1 if in state where pay secrecy

is outlawed

0.1399

(0.347)

0.1345

(0.341)

0.137

(0.344)

0.1337

(0.340)

Work

exper

ience

potential work experience

23.422

(10.79)

23.671

(10.68)

23.664

(10.572)

23.8562

(10.557)

Work

experience

above squared

664.94

(558.5)

674.47

(549.6)

671.75

(547.04)

680.569

(539.73)

Child under 5 1 if presence of child under 5

0.1706

(0.376)

0.1337

(0.34)

0.1762

(0.381)

0.1074

(0.310)

Mean and s.d. for female is computed across both gender groups.

Endnotes

In 1948, 32% of women were in the labor force; in 1938, even fewer were (Smith, 1979; Federal Reserve Board of

St. Louis, 2013). In contrast, in 2012, 58% of women participated in the labor force (US Bureau of Labor Statistics,

2013b).

29 U.S.C. § 157 (2003).

An NLRB Board member confirms that “The right of employees to talk to each other about pay is as fundamental

as any activity intended to receive NLRA protection, given that pay discussions among disgruntled employees are

often at the heart of unionization activity.” (Bielman and Gely, 2004: 169, citing John E. Higgins, an NLRB Board

member.)

The only employer prohibitions the NLRB has allowed involve revealing the entire pay structure, since pay

structures are viewed as proprietory. See Gely and Bierman (2003) and Bierman and Gely (2004)

See for example, S. 71: Paycheck Fairness Act. 105th Congress (introduced 1/21/1997)

http://www.gpo.gov/fdsys/pkg/BILLS-105s71is/pdf/BILLS-105s71is.pdf; H.R. 2023 : Paycheck Fairness Act

(

introduced 6/24/1997) http://www.gpo.gov/fdsys/pkg/BILLS-105hr2023ih/pdf/BILLS-105hr2023ih.pdf; S 74, 106

ngress Paycheck Fairness Act (introduced 1/19/1999) http://www.gpo.gov/fdsys/pkg/BILLS-106s74is/pdf/BILLS

106s74is.pdf; H.R. 541 Paycheck Fairness Act (introduced 2/3/1999) http://www.gpo.gov/fdsys/pkg/BILLS-

106hr

541ih/pdf/BILLS-106hr541ih.pdf; H.R. 2397 Paycheck Fairness Act (introduced 6/30/1999)

http://www.gpo.gov/fdsys/pkg/BILLS-106hr2397ih/pdf/BI

LLS-106hr2397ih.pdf; S. 77 Paycheck Fairness Act

(introduced 1/22/2001) http://thomas.loc.gov/cgi-bin/query/z?c107:S.77; S. 8 Enhancing Economic Security for

Ame

rica's Working Families Act (introduced 1/22/2001) http://www.gpo.gov/fdsys/pkg/BILLS-107s8is/pdf/BILLS-

107s8is.pdf; H.R. 781 Paycheck Fairness Act (introduced 2/28/2001) http://www.gpo.gov/fdsys/pkg/BILLS-

107hr

781ih/pdf/BILLS-107hr781ih.pdf; S. 76 Paycheck Fairness Act, 108th Congress (introduced 1/7/2003)

http://www.gp

o.gov/fdsys/pkg/BILLS-108s76is/pdf/BILLS-108s76is.pdf; 10. H.R. 1688 Paycheck Fairness Act

ntroduced 4/9/2003) http://www.gpo.gov/fdsys/pkg/BILLS-108hr1688ih/pdf/BILLS-108hr1688ih.pdf; H.R. 1687

ycheck Fairness Act (introduced 4/19/2005) http://www.gpo.gov/fdsys/pkg/BILLS-109hr1687ih/pdf/BILLS-

109hr

1687ih.pdf; S. 841 Paycheck Fairness Act (introduced 4/19/2005) h

ttp://www.gpo.gov/fdsys/pkg/BILLS-

109s841is/pdf/BILLS-109s841is.pdf; H.R. 1338 Paycheck Fairness Act (introduced 3/6/2007)

http://www.gpo.gov/fdsys/pkg/BILLS-110hr1338ih/pdf/BILLS-110hr1338ih.pdf; H.R. 12 Paycheck Fairness Act

(in

troduced 1/6/2009) http://www.gpo.gov/fdsys/pkg/BILLS-111hr12ih/pdf/BILLS-111hr12ih.pdf;

15.

S. 182 Paycheck Fairness Act (introduced 4/19/2005) http://www.gpo.gov/fdsys/pkg/BILLS-

109s841i

s/pdf/BILLS-109s841is.pdf and http://www.gpo.gov/fdsys/pkg/BILLS-111s182pcs/pdf/BILLS-

111s182p

cs.pdf; S. 3772 Paycheck Fairness Act (introduced 9/13/2010) h

ttp://www.gpo.gov/fdsys/pkg/BILLS-

111s3772p

cs/pdf/BILLS-111s3772pcs.pdf; S. 2966 Wage Awareness Protection Act (introduced 7/27/2000)

tp://www.gpo.gov/fdsys/pkg/BILLS-106s2966is/pdf/BILLS-106s2966is.pdf; Fair Pay Act of 2011. Introduced as HR

1493 in the House of Representatives on April 12, 2011 http://www.gpo.gov/fdsys/pkg/BILLS-

112hr

1493ih/pdf/BILLS-112hr1493ih.pdf and as S 788 on April 12, 2011 in the Senate:

http://www.gpo.gov/fdsys/pkg/BILLS-112s788is/pdf/BILLS-112s788is.pdf;

ir Pay Act of 2001. Introduced as HR 1362 on April 3, 2001 http://www.gpo.gov/fdsys/pkg/BILLS-

107hr1362ih/pdf/BILLS-107hr1362ih.pdf

and S684, on April 3, 2001 . http://www.gpo.gov/fdsys/pkg/BILLS-107s684is/pdf/BILLS-107s684is.pdf;

Fairness and Individual Rights Necessary to Ensure a Stronger Society: Civil Rights Act of 2004. Introduced on

February 11, 2004 as HR 3809: http://www.gpo.gov/fdsys/pkg/BILLS-108hr3809ih/html/BILLS-108hr3809ih.htm

and on Feb 12, 2004 as S 2088 in the Senate: http://www.gpo.gov/fdsys/pkg/BILLS-108s2088is/html/BILLS-

108s2088is.htm.

See https://www.hrc.org/resources/entry/an-important-step-toward-workplace-equality-an-executive-order-on-

federal-c; http://thinkprogress.org/economy/2014/04/06/3423399/obama-secrecy-salary/.

Since the data end in 2012, I could not include New Jersey.

This was calculated as the annual wage or salary earnings in the previous year divided by the product of the usual

hours worked per week and the number of weeks worked in the previous year.

Men usually have positive coefficients on these children variables, while women usually have negative

coefficients. The explanation for this difference varies from women not working as hard as men (O’Neill, 2004) to

discrimination against mothers (Correll et al., 2007; Budig and England, 2001). In addition, some explain the

occupational and industry variables as accounting for job preferences or unmeasured worker or job characteristics

(O’Neill, 2004); while others explain these variables as capturing job segregation by gender (Hegewisch and

Liepmann, 2013). Thus including these variables may underestimate wage penalties for women if job segregation

by gender and the coefficients on the children variables are in part determined by employer discrimination rather

than the qualifications, productivity and preferences of workers.

Those covered by unions are also more likely to have standardized pay and less room to negotiate pay than

those without unions, consequently having lower variance in earnings (Freeman and Medoff, 1984). However,

data on union coverage begins only in 1990, which would omit the effect of two states that implemented their

laws before this time period. In addition, not all surveyed respondents were asked about being covered by a

union, leaving much of the sample in the “missing variable” category even after 1990.

The regression results for the control variables (not shown) are consistent with previous findings in the research

literature. Those living in metropolitan areas, who are not racial or ethnic minorities, and who had more potential

work experience earn more. Higher education levels and being married (as opposed to never married) also earn

more, and women earn less.

With average wages of $19.41 for women and $27.33 for men in the data, the wage ratio is 71.02%. For full-

time year round workers, average wages were $19.74 for women and $27.17 for men, for a wage ratio of 72.65%.

If wages increased 4% for full-time year round female workers and 5% for all workers, the wage ratios would be

74.57% for all and 75.56% for FTYR workers, for an increase of 3.55 and 3 percentage points, respectively. Wages

were calculated as real wages, 2012=100.

Thus for this specification, D

i,s,t

= φ

+ φ

s,t

+ e

i,s,t

(similar to equation 5),

equation 2 is now: ln(w

i,s,t

) = ψ

i,s,t

+ ψ

+ e

i,t ,

and

equation 3 is now p

i,s,t

= ψ

i,s,t

+ψ

D is still calculated as D

i,s,t

= p

i,s,t

- ln(w

i,s,t

) in equation 4.

Author’s calculations from US Bureau of Labor Statistics, 2013a. The third quarter in 2013 reports weekly

earnings of $860 for men and $706 for women for full-time workers.