J Labor Res (2008) 29:272284

DOI 10.1007/s12122-007-9023-0

Recent Evidence on Factors Influencing the Female

Labor Force Participation Rate

Richard J. Cebula & Christopher K. Coombs

Published online: 29 June 2007

# Springer Science + Business Media, LLC 2007

Abstract This study seeks, using state-level data, to identify key factors that help to
explain recent trends of labor force participation among women. Adult females are
treated as attempting to maximize utility subject to a variety of budgetary and non-
budgetary constraints. Among the findings obtained is a positive impact from the
level of public assistance, i.e., the greater the extent of public assistance to adult
females in the forms of Supplemental Security Income, Food Stamps, and so forth,
the higher the female labor force participation rate (FLFPR). Other factors
contributing to observed FLFPRs include age, the presence of young children,
family income, educational attainment and disability status. In addition, we also find
evidence that an increase in the proportion of the population that is non-native to the
U.S. has a negative effect on the FLFPR.

Keywords Female labor force . Participation rate . State-level data

Introduction

One of the most fundamental changes in the workplace in the U.S. over the last
several decades, indeed over the last century, has been its changed gender
composition. As shown in Vedder (1976: 353354), the female labor force
participation rate (FLFPR) nearly doubled between 1920 and 1965 from 22.7% to
over 39%. Over this period, Vedder (1976: 353) observes females as ...abandoning
their traditional household role. Vedder (1976: 353) further observes that This may
well reflect changing attitudes towards womens roles in society and a growing
liberation of women...[including] reduced female household responsibilities.
More recently, over the period from 1965 to 2000, the FLFPR exploded from
39.3% to 59.9% (Council of Economic Advisors (2006, Table 39)), for an average

R. J. Cebula (*) : C. K. Coombs

Armstrong Atlantic State University, Savannah, GA 31419, USA
e-mail: [email protected]
J Labor Res (2008) 29:272284 273

annual growth rate of 1.19%. Since the year 2000, however, the observed FLFPR,
expressed as a percentage of the female population considered part of the labor force (for
females age 16 and older), has actually been very slowly declining at an average annual
rate of 0.17%. Moreover, there exists a surprisingly wide geographic (e.g., interstate)
variation in the level of the FLFPR. For example, in 2004, the FLFPR ranged from a low
of only 49.1% (for West Virginia) to a high of 69.4% (for South Dakota).
Earlier research analyzed factors influencing the decline in male labor force
participation (especially among the young and the old) that was coinciding with the
increase in female labor force participation.1 More recently, Hotchkiss (2006)
explains that the decline in male labor force participation among all age groups is, at
least in part, a function of increased female labor force participation. In this case,
when the income effect dominates the substitution effect, the decision to work made
by his spouse results in a higher family income, thereby causing the husband to
decrease his labor supply.
The labor market behavior of females has also received considerable empirical
attention.2 In a study analyzing its impact on economic outcomes, Nieswiadomy et al.
(1991) reported evidence that earnings inequality was lower in states with higher
female labor force participation in 1980. Also, Price and Yandle (1987) found
evidence in 1984 that a state is less likely to have a Blue law, the higher the FLFPR
is.3 Examining the effect of state laws on the labor supply decisions of females, Kalist
(2004) found that the existence of a state abortion law (making abortions in that state
legal) positively affected the labor force participation rates of females in that state.
Most studies analyzing the labor force participation of females generally include
characteristics of women as well as changes in their behavior. Recently, Hotchkiss
(2006) investigated the changes in the FLFPR for women aged 25 to 54 for the
period 1975 to 2005. The author finds that for the latter part of this period, when
FLFPR was falling on average, weaker labor market conditions (i.e. higher
unemployment rates) provided downward pressure. Other characteristics negatively
affecting female labor force participation in the Hotchkiss (2006) study included a
lower percent of high school graduates and women having more children under the
age of six.
Using state-level data, the purpose of this empirical study is to identify key factors
that influence the FLFPR. Insight into these factors may provide indirect insights
into how variations in the FLFPR might actually influence state-level unemployment
rate differentials (Vedder and Gallaway (1996)) and/or state-level voter participation
rate differentials (Cebula and Toma (2006)). To make the analysis contemporary, the
study year is 2004. Using this study year, we hope to be able to account for more of

1
See Quinn (1977), Burkhauser (1979), Parsons (1980), and Fuchs (1982) for insight into the retirement
decisions of older male workers. At least in part, the reason for the increase in interest in older male
workers and their labor force participation is due to a slowing-down in the growth of the male non-aged
labor force over this time period. See Bowen and Finegan (1969) for an examination of the labor force
participation rates of young males.
2
For earlier articles on womens labor supply in the U.S. and other countries, see the Journal of Labor
Economics, January 1985 supplement, and Pencavel (1998).
3
A Blue law is an ordinance that prohibits retail selling and other forms of commercial activity on
Sunday.
274 J Labor Res (2008) 29:272284

what was previously left unexplained. That is, because the overall FLFPR was on
average falling during the 2000 to 2005 period, in addition to the customary
regressors, we will also include other factors, factors previously unobserved over this
period that we consider having a potentially significant impact on female labor force
participation.
This study includes an investigation into the potential role of public assistance
since it is a form of public policy. Indeed, public assistance has been found in a
variety of studies of earlier periods to significantly influence labor market behavior
not only through labor force participation effects (Blau and Robins (1988), Durbin
(1973), Gensler (1996), Levy (1979), Miller and Xiao (1999), Moffit (1986, 1992),
Robins and West (1980), Schultz (1990)) but also through changing the pattern of
geographic labor mobility (Carlson and Cebula (1981), Cebula (1978, 1991), Glantz
(1975), Pack (1973), Renas (1980), Vedder (1976), Vedder and Cooper (1974)).
Most of these studies analyze income maintenance programs, a specific form of
public assistance (most commonly, AFDC, aid to families with dependent children)
for their potential roles in creating budget constraints that encourage people to drop
out of the labor force. In other words, welfare programs have historically taken the
form of a de facto guaranteed annual income. Due to the growing evidence of
decreases in labor force participation created by the impact of these original
programs on the disincentive to work, other programs have been devised and
implemented to offer a solution to the problem of work disincentives: welfare
reform is introduced. Specifically, the Earned Income Tax Credit (EITC) program
makes income tax credits available to low-income families with at least one worker.
Because the earnings subsidy goes to only those who work, the EITC potentially
serves as an income maintenance program that preserves work incentives. For policy
implications, it would be important to estimate not only the significance but also the
magnitude of the labor supply responses to the EITC program. Although earlier
versions of the EITC were shown to have only modest effects on labor supply, recent
studies suggest that the program is having a positive impact. In fact, Meyer and
Rosenbaum (1998) show that expansion of the EITC induced more single mothers to
join the labor market. Also, Eissa and Hoynes (1998) found that qualifying for the
EITC was related to only a slight reduction in labor force participation and hours
worked among married mothers. Aggregating these impacts suggests to some that
the EITC is a highly effective means to provide income maintenance without
distorting the incentive to work.
We also include an analysis into the potential impact that legal immigration into a
state has on female labor force participation. One of the arguments for restrictive
immigration policy is that when immigration increases the supply of labor, both the
wages and the employment levels of native workers are reduced.4 To the extent that the
native workers employed as laborers are the target of anti-poverty efforts, the influx of
immigrants could frustrate such efforts by reducing their wages or employment levels.

4
In theory, some native workers leave the market in response to the reduced wage, and those who stay
earn less. See Borjas (1995) for a theoretical framework showing that immigration generates a sizeable
redistribution of wealth in an economy, reducing the incomes of natives who are competing with
immigrant workers in the labor market.
J Labor Res (2008) 29:272284 275

The point is highlighted when considering that native single females with children are
relatively more likely to be hovering around the poverty threshold.5 However, in a
now 12-year-old survey of the literature, Friedberg and Hunt (1995) revealed that
empirical estimates in a variety of settings, using a variety of approaches, had shown
no evidence of significant reductions in native employment, even for those natives
who should be the closest substitutes with immigrant labor.
Our analysis proceeds as follows. First, we discuss the specific variables included
in the model and the associated hypotheses. Next, the empirical results are presented
and discussed. In the final section, we present our conclusions and briefly offer
insights into extending the analysis in the future.

The Framework: Specific Model Variables and Hypotheses

Let us consider the specific variables used to examine female labor force participation
and explain how these variables are expected to influence a females willingness to
work outside the home.
It is assumed that the rational adult female seeks to maximize her utility subject to
a host of real world constraints, including (initially) a budget constraint (inclusive of
the cost of living, which reflects a vector of commodity prices faced by the females
family unit), the presence of dependent children, spousal income plus income of
other males in the family unit, her own expected income, her health/disability status,
her marital status, her age, her educational preparation, and public assistance. Labor
market conditions that alter the likelihood of obtaining a job or change the expected
value of wages are also considered. We focus on human capital characteristics,
behavioral factors, and labor market conditions
Clearly, in a family unit, a budget constraint must be considered in real-world
terms. Among its dimensions would be the overall cost-of-living confronting the
family unit. The higher the overall cost of living for a given family unit, the greater
may be the need for the adult female or adult females in that family unit to seek
gainful employment in order for the family to meet its budget constraint, i.e., cover
its budgetary requirements, ceteris paribus. That is, the FLFPR is an increasing
function of cost-of-living, ceteris paribus. To control for cost-of-living differences,
which may affect the choice to seek employment, we normalize all variables with
dollar values by the cost-of-living index for 2004. Within this context, we now
present the variables in the analysis.
1. The percentage of families in a state with annual income in excess of $200,000
(HIGHINC): Consistent with studies by Gerner and Zick (1983), Schultz
(1990), and Miller and Xiao (1999), it is hypothesized here that the higher the
family income earned, e.g., by a male spouse [and/or other male family
member(s)] contributing to the family budget, the lower may be the need for adult
females in the family unit, e.g., the female spouse, to hold or seek employment for

5
In 2004, 28.3% of families headed by a female adult and without a spouse present were below poverty.
They are more than five times as likely to live in poverty as a family headed by a married couple. Source:
U.S. Bureau of the Census.
276 J Labor Res (2008) 29:272284

purposes of earning additional budgetrequired family income per se, ceteris

paribus. In this study, the family income measure is the variable HIGHINC, which
is the percentage of the families that is considered to be in the highest income
bracket, i.e., above $200,000 per year. Accordingly, it is hypothesized that the
FLFPR is a decreasing function of the HIGHINC, ceteris paribus. In effect, the
variable HIGHINC reflects the high-income status of the family unit. To some
degree, this hypothesis might be interpreted as consistent in principle with the
finding by Lundberg (1985) regarding the impact of a husbands unemployment
on his spouses labor supply decision.
2. The percentage of the states female population age 16 and over that has a
disability (DISABLED): Naturally, ones health status can impose constraints
on the ability to effectively seek and secure gainful employment, i.e., to
participate in the labor force. For example, despite the Americans with
Disabilities Act of 1990 and the employment doors opened thereby and despite
the willingness and official policy on the part of certain firms and various
government agencies to openly seek to employ the disabled, those persons who
are truly disabled on the average are less able to accept and therefore
pragmaticallyare less able and less willing even to seek many forms of
employment. Clearly, in some cases, they are unable to accept any form of
employment. Hence, the greater the percentage of the total female population
age 16 years and older that is classified as disabled, the lower the FLFPR,
ceteris paribus.
3. The percentage of the states female population age 16 and over with children
whose children are under the age of six (CHILDRENLT6): We also control for
a behavioral aspect of a females labor supply decisions across different stages
of her life by including whether she has relatively younger children at home.
Traditionally, despite the phenomenon in recent years in which the male spouse
allegedly assumes greater responsibility for childcare than was the case
historically (Vedder (1976, p. 353)), the overall male labor force participation
rate is still significantly higher than the FLFPR and the percentage of males
with full-time employment still also far exceeds the percentage of females with
full-time employment (Council of Economic Advisors (2006, Tables B-39, B-
40)). Thus, the female spouse more often than the male spouse assumes
primary responsibility for child rearing and household management, at least at
the childrens younger ages. This phenomenon would of course (on the
average) be reinforced by the fact that, in the formal labor market, male
earnings potential exceeds female earnings potential (even in the 21st century).
Thus, having relatively younger children at home becomes a de facto constraint
for women in the pursuit of utility maximization via labor force participation.
Accordingly, the FLFPR is hypothesized to be a decreasing function of
CHILDRENLT6, ceteris paribus. Clearly, because we use the states female
population with any children at home as the base, the sign of the coefficient on
this variable will be relative to females with children older than six.
4. The percentage of the states female population age 16 and over that is married
with spouse present (SPOUSEPRESENT): The expected influence of this
variable is unclear. To the extent that an extra hour of market work by both
husband and wife can offset the hour of lost home production, we could expect
J Labor Res (2008) 29:272284 277

to see a positive impact of this variable on FLFPR. Additionally, if both

spouses are complementary in the household production of commodities (and
leisure activities), ones increased labor supply may tend to increase the labor
supply of the other.6 However, if both husband and wife are substitutes in the
household production of commodities, ones increased labor supply might tend
to decrease the labor supply of the other. There is evidence, however, that the
incentives for married females to work for pay have increased more rapidly in
the postwar period than before.7 While we are reluctant (despite the earlier
findings by Kniesner (1976)) to hypothesize a net sign for the coefficient on
this variable, we shall pay close attention to either of these possibilities.
5. and 6. Female median age and female median age squared (FMEDAGE and
FMEDAGESQ): To control for labor market experience, we include the states
female median age as a proxy. Female median age squared is also included to
capture the concavity of the experiencelabor force participation profile.
Following conventional analysis (Hotchkiss (2006)), we expect that a females
labor force participation is positively related to the females age but the positive
effect is diminishing, ceteris paribus.
7. Female (age 16 and over) median earnings (FMEDEARN): The opportunity
cost of staying at home and engaging in household production and/or leisure
is the salary earned from market work. We expect that an increase in the
opportunity cost will have a positive effect on female labor force participation.
That is, we use FMEDEARN in a state in 2003 to capture the expected
earnings of females in 2004, and we hypothesize this will have a direct
relationship with their labor supply decisions in 2004.
8. The percentage of the states female population age 16 and over that has a
college education (COLLEGEGRAD): We assume that a female makes
decisions on her education level mindful of the expected returns associated
with different levels of education. To the extent that she expects a relatively
higher return with a college education, the choice to acquire this level of
education makes it more likely that she will be in the labor force seeking to
receive this return. Furthermore, in the absence of significant gender
discrimination, greater levels of educational attainment theoretically afford
females a greater range of employment possibilities and a greater choice of
superior (more desirable) employment, e.g., higher paying jobs, thereby
creating an incentive to enter the labor force. Indeed, previous studies by
Gerner and Zick (1983), Rexroat (1990) and Miller and Xiao (1999) have
found that higher educational attainment exercises a positive impact on female
labor force participation. Therefore, we hypothesize a positive relationship
between the percentage of a states female population with a college degree and
the FLFPR within that state.

6
We are assuming that the husband is working. Also, this variable is relative to a females spouse not
being present or not being married. Therefore, relative to not being present or a female that is not married,
when husband is present, this is considered an increase in labor supply of the husband.
7
See Kniesner (1976) for evidence that husband and wife are complementary in home production and
consumption.
278 J Labor Res (2008) 29:272284

9. The states expenditures on public assistance medical care benefits and income
maintenance benefits per capita (ASSIST):8 Based on the conventional wisdom,
it is expected that leisure is a normal good (Gensler (1996, p. 420)), and,
according to the classic article by Brehm and Saving (1964, p. 1003) the
...demand for government assistance (welfare) programs may be looked upon
as a special case of the demand for leisure. In addition, Brehm and Saving
(1964, p. 1018) conclude that recipients of government assistance (welfare) ...
are like the remainder of consumers in that they react to economic incentives.
Therefore, it could be argued that the greater the degree to which female head-
of-household units receive welfare/public assistance (ASSIST) such as Food
Stamps and/or Supplementary Security Income (SSI), the greater the degree to
which those household units experience an enhanced family income.
Accordingly, the greater the amount a state provides in welfare expenditures
to female head-of-household units, the greater the degree to which the female
heads of those households do not participate in the labor force (and consume
leisure instead). Several prior studies have yielded findings consistent with this
behavior (Blau and Robins (1988), Durbin (1973), Levy (1979), Miller and
Xiao (1999), Moffit (1986, 1992), Robins and West (1980), Schultz (1990)).
Indeed, historically this behavior has been reinforced because SSI recipients
and Food Stamps recipients face restrictions on earned income in order to
maintain eligibility. In any case, the funds impounded under the ASSIST
umbrella can potentially be viewed as actual or potential components of the
family unit budget constraint. Alternatively, however, because programs such
as the Earned Income Tax Credit (EITC) program, which makes income tax
credits available to low-income families with at least one worker, could
provide an incentive to seek gainful employment, we would expect to observe a
positive relationship between the FLFPR and ASSIST. Without hypothesizing
a sign, we will also pay careful attention to the outcome of this relationship.
10. The number of admitted immigrants by state of intended residence as a
percentage of the states population (NONNATIVE): To the extent that
immigration is harmful to the interests of native female workers, we would
expect a negative impact from an increase in immigration to a state on the
FLFPR. In this case, the marginal native female worker anticipates a lower
wage due to the increase in supply of labor. For this female worker, the lower
wage is now below her reservation wage, causing her to drop out of the labor
force. However, if our results do coincide with what is hypothesized, we have
not necessarily uncovered a harmful impact of immigration on native female
labor force participation. Because we use the states population as a base,
immigrants admitted in the state are included. Therefore, if results do show a
negative effect of NONNATIVE on the FLFPR within a state, it could be that

8
The variable ASSIST includes the following expenditures: Public assistance medical care benefits
(Medicaid and other medical vendor payments), Supplemental security income benefits (SSI), food
stamps, temporary assistance for needy families (TANF), expenditures for food under the supplemental
program for women, infants, and children, refugee assistance, foster home care and adoption assistance,
energy assistance, and earned income tax credits (EITC).
J Labor Res (2008) 29:272284 279

nonnative females are not supplying their labor in that state. We will pay close
attention to the outcome of this relationship and offer potential remedies should
the correct sign result.
11. Female unemployment rate (FUNEMPRATE): To control for labor market
conditions and the expectation of obtaining gainful employment in 2004, we
include the female unemployment rate in each state in 2003. Relative to using
the unemployment rate for both genders, we posit that this measure provides the
marginal female with a more precise likelihood of gaining employment to the
extent that there still exist occupations that are predominantly performed by
women. To the extent that unfavorable labor market conditions exist in the current
period, discouraged females will drop out of the labor force in the next period.
Therefore, we hypothesize a negative relationship between FUNEMPRATE and
the FLFPR.

Empirical Results

Data were gathered for all 50 states; Washington, D.C. was excluded from the study.
To normalize each dollar variable the cost-of-living data were obtained from
ACCRA (2004). The variable FLFPR for 2004 was obtained from the U.S. Census
Bureau (2006, Table 581). Data for the variables DISABLED, SPOUSEPRESENT,
FMEDEARN, FMEDAGE, CHILDRENLT6, and COLLEGEGRAD were obtained
from the U.S. Census Bureau (2003, American Community Survey). The data for the
variable HIGHFAMINC was obtained from the U.S. Census Bureau (2003, Tables
P025, P031, P047), whereas the data for variable ASSIST was from the U.S. Census
Bureau (2005, Tables 527, 550, 557, 559). The data for FUNEMPRATE was gathered
from the U.S. Census Bureau (2004, Table 579). Finally, data on the variable
NONNATIVE was obtained from the U.S. Department of Homeland Security (2003,
Table 11). All explanatory variables used in this analysis, for their potential effects on
female labor force participation in 2004, were obtained for 2003; this lagging is
adopted in order to mitigate concerns of simultaneity bias. Table 1 provides the means
and standard deviations for all of the variables in the analysis.
The OLS results of the basic model are presented by model (1) in Table 2. The basic
model includes those variables used in the previous literature and are considered
conventional. In this model, all of the nine estimated coefficients are significantly
different from zero and all hypothesized signs are exhibited. Moreover, five of these
nine coefficients are statistically significant at the 1% level or beyond and the other
four are significant at or beyond the 5% level. The coefficient of determination
(adjusted) is 0.84, so that the model explains over four-fifths of the variation in the
FLFPR. Finally, the F-statistic is significant at far beyond the 1% level, attesting to
the overall strength of the model.
The estimated coefficient on the HIGHFAMINC variable is negative, as hypothe-
sized, and significant at the 1% level, a result consistent with Gerner and Zick (1983),
Schultz (1990), and Miller and Xiao (1999). Apparently, in the present study, the
higher the percentage of families with an annual income in excess of $200,000, the
280 J Labor Res (2008) 29:272284

Table 1 Descriptive statistics

Variable Mean SD

FLFPR 60.65 4.19

FUNEMPRATE 5.26 1.04
FMEDEARN 20,417.84 3,148.54
HIGHFAMINC 2.54 1.30
DISABLED 9.95 2.04
SPOUSEPRESENT 52.45 3.37
CHILDRENLT6 29.72 1.81
FMEDAGE 37.51 2.25
FMEDAGESQ 1,411.65 161.61
COLLEGEGRAD 15.22 2.68
ASSIST 1,353.23 363.41
NONNATIVE 0.17 0.12

less the necessity of adult females to be labor force participants. The estimated
coefficient on the DISABLED variable is negative, as expected, and significant at
beyond the 5% level. Clearly, the greater the percentage of the adult female population
that is disabled, the lower the FLFPR because of the employment restrictions that are
implied by the disability status. The estimated coefficient on the CHILDRENLT6
variable is negative and significant at far beyond the 1% level. Thus, it appears that the
likelihood of a female participating in the labor force declines when there are
dependent relatively younger children in the household. Presumably, this finding
suggests that the mother is more likely to remain in the household as the primary
family caretaker until the child is at least 6 years old. Given that, on the average, male
earning capacity exceeds female earning capacity, the result more often than not would
be for the male spouse to enter the labor force and for the female spouse not to; the
result is a lower FLFPR.9
The estimated coefficients also exhibit the expected signs for the variables
FMEDEARN, COLLEGEGRAD, FMEDAGE, and FMEDAGESQ and are statisti-
cally significant at the 5% level or beyond in all four cases. It appears that relatively
higher expected earnings induce an increase in female labor force participation, all
else held constant. Also, we find evidence that females are relatively more likely to
seek the returns to education, as well as superior employment, when they are
prepared at the college level. Our proxies for experience and the concavity of
experience also exhibit the expected signs. That is, an increase in experience has a
positive effect on labor force participation but the positive effect gets smaller at
larger levels of experience.
Because the estimated coefficient on FUNEMPRATE exhibits the hypothesized
negative sign and is significant at well beyond the 5% level, we have evidence that
the higher average female unemployment rate in 2003 put downward pressure on
female labor force participation. Finally, the coefficient on SPOUSEPRESENT is

9
Included in this measure are females who are head of household with no male present. Therefore,
relative to children older than six, single females are also relatively more likely to stay at home and care
for the child who is less than 6 years old.
J Labor Res (2008) 29:272284 281

positive and significant at the 1% level. We did not hypothesize a sign for this
variable because the theory allowed for opposite predictions. However, relative to
the evidence shown in the previous literature, we have similar results that imply the
husband and wife consider themselves complements in household production and
leisure activities.
In model (2) of Table 2, all of the variables from model (1) are retained but we
include the additional variable ASSIST. As shown in model (2) of Table 2, all of the
originally estimated coefficients are statistically significant at the 1% level or
beyond, and the basic model variables retain the same signs. Overall, the adjusted R2
indicates that model (2) explains approximately 86% of the variation in the degree of
labor force participation.
The estimated coefficient on the variable ASSIST is positive and significant at
beyond the 5% level. Apparently, an increase in welfare expenditures (as now
defined) provided by government lessens the degree to which leisure is consumed in
lieu of female labor force participation. This is inconsistent in principle with the
hypotheses and findings set forth in the classic article by Brehm and Saving (1964)
and empirical findings in more recent studies, including Schultz (1990) and Miller
and Xiao (1999). For example, Miller and Xiao (1999, p. 417) find that Receiving
public financial assistance...reduced the probability of [female labor force]
participation. Our results, by contrast, offer support that contemporary income
maintenance programs, i.e., as they are established today, are much more effective at
providing welfare assistance without distorting/reducing the incentive to work.
Arguably, at least in this sense, welfare reform has worked.
The third model in Table 2 includes the variables from model (1) and includes the
variable NONNATIVE. Once again, all of the basic model variables retain the same

Table 2 Regression results

Variable (1) (2) (3) (4)

FUNEMPRATE 0.82 (2.63)** 0.812 (2.72)*** 0.623 (2.02)** 0.628 (2.10)***

FMEDEARN 0.0004 (2.52)** 0.0005 (2.91)*** 0.0005 (2.96)*** 0.001 (3.31)***
HIGHFAMINC 1.868 (4.98)*** 1.845 (5.04)*** 1.528 (4.95)*** 1.528 (5.30)***
DISABLED 0.501 (2.50)** 0.619 (3.02)*** 0.653 (3.37)*** 0.752 (3.88)***
SPOUSEPRESENT 0.444 (3.62)*** 0.577 (4.24)*** 0.396 (3.08)*** 0.521 (3.68)***
CHILDRENLT6 0.641 (3.36)*** 0.587 (3.35)*** 0.588 (3.07)*** 0.542 (3.01)***
FMEDAGE 3.533 (2.64)** 3.889 (2.75)*** 3.556 (3.25)*** 3.881 (3.31)***
FMEDAGESQ 0.057 (3.02)*** 0.063 (3.13)*** 0.058 (3.73)*** 0.063 (3.76)***
COLLEGEGRAD 0.865 (6.85)*** 0.87 (8.21)*** 0.893 (7.58)*** 0.896 (8.52)***
ASSIST 0.021 (2.55)** 0.02 (2.93)***
NONNATIVE 9.095 (3.34)*** 8.532 (3.38)***
Constant 3.15 20.296 2.722 18.453
0.12 0.71 0.12 0.75
Observations 50 50 50 50
R2 0.87 0.89 0.89 0.91
Adjusted R2 0.84 0.86 0.87 0.88
F-statistic 42.5 44.93 56.38 57.14

Dependent variable: female labor force participation rate.

Absolute values of T statistics in parentheses. *, **, and *** denote significance at the 0.10, 0.05, and 0.01
levels, respectively. All estimations include Whites (1980) correction for heteroskedasticity. All dollar
values are inflated/deflated to 2004 values using cost-of-living index.
282 J Labor Res (2008) 29:272284

sign and all are statistically significant. Also, similar to an improvement in explanatory
power of model (2) relative to model (1), model (3) explains relatively more (87%, as
indicated by the adjusted R2) of the variation in the degree of labor force participation.
The sign on the coefficient for the variable NONNATIVE is negative and highly
significant, implying that an increase in immigration exercises downward pressure
on the participation rates of females.10 The underlying forces of this effect, however,
are indeterminate. One possible explanation is that an increase in the number of
immigrants increases the working-age population but the female non-native
population does not join the labor force. Although this might have other implications
in which the native-born population can suffer from immigration, if the negative
effect we observe is explained by this phenomenon, female native workers would
not necessarily be harmed by immigration.11 On the other hand, if the non-native
female population that is admitted into a state is working, this could be causing a
harmful impact on the native female worker. The effect is an increase in the supply
of labor. To the extent that the two types of labor are substitutes in the production
process, we would expect to observe a decrease in the wages paid and the response
of some native female workers would be to leave the market. The results here
strongly suggest that there is indeed a net negative effect of immigration on the
FLFPR. However, we are mindful of how it could be erroneous to conclude that this
negative effect is borne solely at the expense of the female native population.
Finally, in model (4), we include both ASSIST and NONNATIVE to the variables
included in model (1). Interestingly, all variables retain their signs from the previous
models, and all are significant at the 1% level. Coupled with the fact that the size of
the coefficients on each variable does not exhibit much variation across the
estimations, we comfortably maintain that multicollinearity is not a problem in this
analysis. Finally, the adjusted R2 suggests that model (4) explains 88% of the
variation in the degree of labor force participation. Indeed, observing across the
models, both the adjusted R2 and F-statistic increased, further attesting to the overall
strength of the model.

Conclusion

Using state-level data, this study seeks to provide contemporary evidence as to key
factors that determine the female labor force participation rate. The perspective for
the analysis is that females seek to maximize utility subject to a variety of economic,
family, health, and institutional constraints/factors. The empirical results find that
there appear to be at least 11 such factors that influence the FLFPR.

10
The number of immigrants admitted into each state in 2003 includes both males and females. We were
not able to disaggregate by gender. For the U.S. in 2003, however, 55% of the total number of immigrants
admitted are female. Of those immigrants who are female, 41% are not married.
11
One way in which the native-born population, in the aggregate, can be harmed by immigration is if the
immigrants obtain government services or payments in excess of the taxes they pay. To the extent that
immigrants are being admitted to a state but they are not working in that state, the harmful effect on the
native-population could be realized.
J Labor Res (2008) 29:272284 283

On the one hand, there is strong evidence that the FLFPR is directly impacted by:
greater education (having a college degree), being married with spouse present,
expected earnings, experience, and welfare assistance. On the other hand, there also
is strong empirical evidence that the FLFPR is negatively impacted by: having
relatively young children in the household, the percentage of the female population
that is disabled, the percentage of females in households with very high incomes,
higher rates of unemployment, and increases in immigration. Whereas the pattern of
empirical results shown in this study/analysis appear consistent and robust, we are
mindful not to draw excessively strong conclusions regarding the possible
implications of our results for the ASSIST and NONNATIVE variables. Ideally,
further evidence consistent with these particular results should be obtained before
stating strong or potentially definitive conclusions. In any case, we consider the
results here as complementary to those found in previous research. Future research
should be aimed at using measures and techniques that more closely identify
potential effects of these factors on the labor supply decisions of females. Finally,
perhaps subsequent research might benefit from examining these results when the
FLFPR is disaggregated by race and/or age.

Acknowledgement The authors are indebted to the Editor and an anonymous referee for very helpful
insights on this project.

References

ACCRA (2004) The Council for Community and Economic Research

Blau D, Robins P (1988) Labor supply response to welfare programs: a dynamic analysis. J Labor Econ
4:82105, January
Borjas GJ (1995) The economic benefits from immigration. J Econ Perspect 9:322, Spring
Bowen WG, Finegan TA (1969) The economics of labor force participation. Princeton University Press,
Princeton, NJ
Brehm CT, Saving TR (1964) The demand for general assistance payments. Am Econ Rev 54:10021018,
December
Burkhauser RV (1979) The pension acceptance decision of older workers. J Hum Resour 14:6375, Winter
Carlson L, Cebula RJ (1981) Voting with ones feet: a brief note on the case of public welfare and the
American Indian. Public Choice 37:321326
Cebula RJ (1978) An empirical note on the TieboutTullock hypothesis. Q J Econ 92:705711, March
Cebula RJ (1991) A brief note on welfare benefits and human migration. Public Choice 69:345349
Cebula RJ, Toma M (2006) Determinants of geographic differentials in the voter participation rate. Atl
Econ J 34:3340, March
Council of Economic Advisors (2006) Economic report of the president, 2006. U.S. Government Printing
Office, Washington, D.C.
Durbin E (1973) Work and welfare: the case of aid to families with dependent children. J Hum Resour
8:103125, Supplement
Eissa N, Hoynes HW (1998) The earned income tax credit and the labor supply of married couples.
Working Paper No. 6856. National Bureau of Economic Research, Cambridge, MA
Friedberg RM, Hunt J (1995) The impact of immigrants on host country wages, employment, and growth.
J Econ Perspect 9:2344, Spring
Fuchs VR (1982) Self-employment and labor force participation of older males. J Hum Resour 27:
339357, Summer
Gensler H (1996) The effects of welfare guarantee levels and tax rates on male labor supply. Appl Econ
Lett 3:417422, June
Gerner JL, Zick CD (1983) Time allocation decisions in two-parent households. Home Econ Res J
12:145158
284 J Labor Res (2008) 29:272284

Glantz FB (1975) The determinants of the intermetropolitan migration of the poor. Ann Reg Sci 8:2539, July
Hotchkiss JL (2006) Changes in behavioral and characteristic determination of female labor force
participation, 19752005. Federal Reserve Bank of Atlanta Economic Review 91:120, 2nd quarter
Kalist DE (2004) Abortion and female labor force participation: evidence prior to Roe v. Wade. J Labor
Res 25:503514, Summer
Kniesner TJ (1976) The full-time workweek in the United States, 19001970. Ind Labor Relat Rev 30:3
15, October
Levy, F (1979) The labor supply of female household heads or AFDC work incentives dont work too
well. J Hum Resour 14:7697, Winter
Lundberg S (1985) The added worker effect. J Labor Econ 3:1137, January
Meyer BD, Rosenbaum DT (1998) Welfare, the earned income tax credit, and the employment of single
mothers. Working Paper No. 2. Joint Center for Poverty Research, Northwestern University, Chicago, IL
Miller CF, Xiao JJ (1999) Effects of birth spacing and timing on mothers labor force participation. Atl
Econ J 27:410421, December
Moffit R (1986) Work incentives in the AFDC system: an analysis of the 1981 reforms. Am Econ Rev
76:219223, May
Moffit R (1992) Incentive effects of the U.S. Welfare System: a review. J Econ Lit 30:162, March
Nieswiadomy M, Slottje DJ, Hayes K (1991) The impact of unionization, right-to-work laws, and female
labor force participation on earnings inequality across states. J Labor Res 12:185195, Spring
Pack JR (1973) Determinants of migration to central cities. J Reg Sci 13:249260, August
Parsons DO (1980) The decline in male labor force participation. J Polit Econ 88:117134, February
Pencavel J (1998) The market work behavior and wages of women: 19751994. J Hum Resour 33:
771804, Fall
Price J, Yandle B (1987) Labor markets and Sunday closing laws. J Labor Res 8:407414, Fall
Quinn JF (1977) Microeconomic determinants of early retirement: a cross-sectional view of white married
men. J Hum Resour 12:329346, Summer
Renas SM (1980) An empirical note on the TieboutTullock hypothesis: comment. Q J Econ 94:619623,
May
Rexroat C (1990) Race and marital status differences in the labor force behavior of female family heads:
the effect of household structure. J Marriage Fam 52:591601, August
Robins P, West R (1980) Program participation and laborsupply response. J Hum Resour 15:499523,
Summer
Schultz TP (1990) Testing the neoclassical model of family labor supply and fertility. J Hum Resour
25:599634, Autumn
U.S. Census Bureau (2003) 2003 American community survey summary tables. U.S. Government Printing
Office, Washington, DC
U.S. Census Bureau (2004) 2004 American community survey summary tables. U.S. Government Printing
Office, Washington, DC
U.S. Census Bureau (2005) Statistical abstract of the United States, 20042005. U.S. Government Printing
Office, Washington, DC
U.S. Census Bureau (2006) Statistical abstract of the United States, 2006. U.S. Government Printing
Office, Washington, DC
U.S. Department of Homeland Security (2003) Yearbook of immigration statistics. Government Printing
Office, Washington, DC
Vedder RK (1976) The American economy in historical perspective. Wadsworth, Belmont, CA
Vedder RK, Cooper D (1974) Nineteenth century English and Welsh geographic labor mobility: some
further evidence. Ann Reg Sci 8:131139, June
Vedder RK, Gallaway LE (1996) Spatial variations in U.S. unemployment. J Labor Res 17:445461,
Summer
White H (1980) A heteroskedasticity-consistent covariance matrix and a direct test for heteroskedasticity.
Econometrica 48:817838, May