Abstract
Can female participation in the labor force contribute to economic growth? Is there a causal relationship between female participation in the labor force and economic growth? This paper addresses these questions using the vector error correction model. For this study, an annual time series data for the period 1970 to 2012 for the United States has been used. The results show that there exists a co-integration between female participation and GDP per capita, which is a proxy measure for economic development. The result does not support a two-way causality between the variables. However, a causality running from GDP per capita and education to female participation was observed.
Introduction
Women have always played significant roles in their communities, especially by providing labor for specific tasks including those that are restricted to them. Typically, women have been found to work extensively in the agriculture sector. Women take care of the household chores by preparing the essential needs of living, such as food for the whole family. They also take care of the education of their children. Off late, the inclusion of women in the paid workforce is regarded as one of the most prominent changes in the socio-economic structures of the 21st century. Even though there is still significant variation in the female labor force participation rates (FLFPR) between countries, there is a consensus that their participation is crucial to the economic growth of individual national economies. However, the contribution of FLFPR to the measured economic activity and growth levels is still low compared to that of males; a fact that may have serious macroeconomic consequences (Elborgh-Woytek et al., 2013). The presence of women in economic activity, a phenomenon that offers many benefits to a nation, can be achieved by giving the females equal and efficient opportunities to participate in the process of national development, policy making and implementation to serve the welfare of the women in particular, and the nations in general.
One way in which FLFPR relates to economic development is that increased level of women participation increases the production of a country, which in turn leads to a higher national output. This is well supported by the fact that when females were more involved in unpaid jobs, it reflected in lower per capita GDP and the national output realized was low. This partially explains why developed countries have higher GDP than developing countries, as the FLFPR is higher in the former than in the latter.
Unlike literature which examines the U-Shape hypothesis that describes the relationship between FLFPR, economic development, and gender gap, the objective of this work is to address the relationship between FLFPR and economic development with the inclusion of the most important determinants that influence female participation in the labor market namely, educational attainment, unemployment rate, urban population, and fertility rate. In addition, this paper will evaluate the causal relationship between FLFPR, economic development, and educational attainment. To investigate these relationships, analytical framework will be constructed using nonlinear Least Squares Estimation and Granger Causality Test using Vector Error Correction Model (VERM). This paper analyzes the period starting in 1970 and finishing in 2012 for the United States.
Literature Review
Considerable work has been done to investigate the impact of labor force participation of women on economic development. Literature has primarily focused on two paths to explain the relationship between FLFPR and economic development. While some of the literature available focuses on examining the U-Shape relationship, others consider gender equity.
The U-Shape hypothesis focuses on critical issues affecting women's labor participation and it is important to note that their increased participation is a key driver of growth. Thus, the U-Shape hypothesis illustrates the correlation that exists between women's labor participation and economic development, such as structural shifts within economic activities of women as well as changes occurring in the household labor supply with perceptions concerning females working outside the households. Therefore, the U-Shape hypothesis states that women participation rate tends to be very high in poor countries since they are more engaged in subsistence activities. In comparison, it is lower in the middle-income nations where there tends to be a transition of men towards industrial jobs that fetch better salaries and improve the standard of living thereby allowing the women to focus more on household activities. The women participation rate again rises in the high-income nations owing to better education levels and lower fertility rates. This implies that with an improvement in their educational levels and a reduction in the fertility rates, females in the high-income countries are able to participate in the labor force and help to meet the increased demand, particularly within the service sector.
A number of studies have put forth empirical evidence to suggest that FLFPR follows a U-shaped curve during the course of economic development, and general results confirm the fact that the estimated pooled data exhibits the U-shaped hypothesis relationship in terms of sectoral employment share, wages, education attainment, unemployment rates, and marital status. It is important to consider that higher rates of economic development promote FLFPR through the promotion of work opportunities for women. However, there appears to be a systematic U-shaped relationship occurring between women's LFPR and level of economic development. Upon further examination of the relationship between the economic structure and women's economic activities, it is noteworthy that the rate of economic growth is more important as compared to the composition of economic growth. Studies suggest that the aggregate alterations in the proportion of women participating in the labor force are mostly linked to the movements existing in the workforce across different sectors rather than variations in the percentage of women in the workforce within sectors (Lam & Duryea, 1999).
A complete integration of females into the economy should be considered as a desirable goal in relation to equity and efficiency implications. Equity considerations mean that the participation of women in the labor force market should enhance their relative economic position. This would in turn boost the overall economic efficiency of the countries along with their potential to develop. Nevertheless, the low levels of female participation in the labor force causes variations that conflict with efficiency and equity goals. Policy makers should examine the labor market policies, initiatives, and special programs and develop them in order to ameliorate the negative situation and to enhance the labor market participation of females (Adler, Clark & Ramsbey, 1991).
Investigations into the relationship between FLFPR and economic development have focused greatly on the gender gap found in the employment sector and reveal that gender gap has a significant effect on this relationship. Economic growth of a nation significantly impacts women's participation in employment, and therefore in the context of the U-shaped hypothesis, the association between FLFPR and economic development can be analyzed in terms of the GDP per capita in relation to the correlation between women's participation in employment and the state of the economic growth (Tam, 2011).
Females are more likely to work in the white-collar service sector. According to Goldin (1994), white-collar jobs have become feminized in the United States since the 1920s. Therefore, when the U-shaped hypothesis is feminized, this allows the use of data concerning female participation in the sectors that are female-oriented or female-dominated. Although feminized, such a U-shaped hypothesis can result in a better estimation of the correlation between female participation in the economy and economic growth since the impact of economic development upon women is looked at from the perspective of more complex determinants than what the existing literature is revealing (Rangarajan & Padma, 2011).
As the society continues to develop, female participation in the labor force increases. There is a corresponding improvement in educational attainment of women in relation to the continued rise in the white-collar service sector, which provides more attractive employment and new opportunities for females. Therefore, these new opportunities are not subject to stigmatization and the restrictions placed earlier on access to employment continually erode over time. The decline in fertility rates in relation to increases in full-time and part-time jobs as well as better access to childcare facilities has enabled women to combine the tasks of raising children and working outside their households. Therefore, feminization of the U-shaped hypothesis has had a great impact on the recent theoretical work, which leads to the development of a model where exogenous processes of globalization bring about gender-specific labor demand (Verme, 2014).
The increasing participation of women in the labor force in United States of America is associated with a strong positive correlation between the proportion of uneducated females and female participation rates in the United States since 1900. However, between 1900 and 1940, things took a new turn given the acceleration of economic development, which enhanced the demand for educated women. In developed economies such as the United States, a strong positive correlation exists between female participation in the labor force and female level of education (Kottis, 1990). In addition, a study by Goldin (1994), using a time series analysis for the United States, confirmed the U-shaped hypothesis and concluded that as economic growth continues, educational attainment of women tends to increase while the fertility rate tends to decrease. As a result, women's participation in the labor force rises. Another time series study conducted in Turkey has also confirmed the U-shaped hypothesis. The same study also observed a negative effect of unemployment and a positive effect of education on female participation in the labor force (Tansel, 2001).
Time series data has revealed that the trend of women participating in the economy has increased over the past decades. When analyzing the time series data for explaining the relationship between economic development and women's participation in the labor force within the framework of the U-Shaped hypothesis, other determinants must be considered such as unemployment rates, educational attainment, industry mix, and the total fertility rate. The linear and quadratic econometric models of economic development consider these variables, the coefficients of which can either take positive or negative forms, which further confirm the U-Shape hypothesis. Other determinants of the time series data analysis are statistically significant and the coefficients of the econometric models applied make educational attainment the most important variable in measuring women's participation in the economy (Gaddis & Klasen, 2013).
Briefly put, female participation in the labor market is considered as one of the most constant and important drivers of social and economic transformations. Women who have higher education and fewer family responsibilities show higher rates of economic participation.
Educational inequality, lesser access to resources, and employment inequality are the most important factors when talking about gender inequality. There is extensive evidence in the literature concerning the impact of inequality on economic development. A series of research work by Klasen (1999, 2002, 2006) has explained two points: first, inequality in education can harm economic development by decreasing the human capital. Unequal education opportunities for women can also negatively affect the fertility rate and mortality rate of children, which indeed motivate economic development. Second, employment inequality can have a negative impact on the economic development by reducing the average workforce availability. A more recent study by Klasen and Lamanna (2009) has shown that education inequality and employment inequality negatively impact economic development. Similarly, a strong positive effect on economic development was found along with a reduction in the gap between female and male education by Thévenon et al. (2012). Thevenon et al. also conclude that the effect of FLFPR on economic development depends on the convergence of FLFPR to the male LFPR. Similarly, Ince (2010) concludes that higher education for women is associated with higher FLFPR. Furthermore, a slow economic development process is a result of not taking into account policies and barriers that reduce the gender gap (Agénor & Canuto, 2105).
The relationship between women activities and economic growth is multi-faceted, which suggests that gender discrimination at the workplace has a negative impact on economic development. Many studies have suggested that the implications of gender inequality on the economy are a consequence of the gap in educational opportunities provided to men and women, a scenario that is present in many countries including the United States of America. Women's participation in the labor force tends to increase with economic development, although, this association is not clear. Women's participation in the workplace is affected by a wide range of personal factors, labor market factors, such as unemployment, income, fertility rate, and educational attainment, and a number of other social factors.
3.1 Unemployment Rate
Literature has found an ambiguous effect of the unemployment rate on FLFPR that can be clarified by the "added-worker effect" and the "discouraged-worker effect" hypotheses. The added-worker effect hypothesis explains the situation when husbands of married women become unemployed. In this case, the supply of women in the labor force increases in order to compensate the shortage of family income that is associated with the husband leaving the labor market. Thus, in this case the women act like primary workers (Tansel, 2001). Therefore, the added-worker effect has a positive effect on FLFPR. In contrast, the discouraged-worker effect hypothesis explains the situation in which women leave the labor force when unemployment rate is high because the cost of finding a job is high and the probability of finding a job is low. Thus, the discouraged-worker effect has a negative effect on FLFPR. Looking at the literature, the discouraged-worker effect is most likely to dominate. In this study, to test both "added-worker effect" and "discouraged-worker effect" hypotheses, total unemployment, female unemployment, and the male unemployment rate has been used.
3.2 Income
The total income of a family depends on the income of both husband and wife. When the husband has a high income, in general wives prefer to stay at home and do not participate in the labor force because their basic necessities are easily met by the husband’s income (Chen, el all, 2014). Thus, the high income of the husbands has a negative effect on FLFPR.
In terms of women’s wages, the higher the female income is, it is more likely that the woman participates in the labor force rather than enjoying more leisure time thereby substituting her leisure time for working hours, which is the substitution effect. Thus, the substation effect has a positive effect on FLFPR. On the other hand, the income effect comes into force when women have a higher income and they spend less time on work and more on leisure. In this case, the income effect dominates and the relationship between FLFPR and female wages is negative. Usually, the substitution effect is dominant (Tansel, 2001). For this paper, neither husband’s wages, nor wives’ income will be used because per capita GDP will take into consideration both male and female income.
3.3 Fertility rate
Family burdens usually increase with the size of the household and that affects the decision of the female participating in the labor force. As household size increases, women tend to stay at home to take care of their families and thus, are less likely to participate in the labor force.
Generally, economic literature agrees on the inverse correlation between fertility rate and FLFPR. The effect of fertility on FLFPR strongly affects younger women more than older women. The fertility rate depends on the cost of child rearing and the women’s age. The cost of child rearing in rich families is higher than that in the poor families because rich families focus primarily on the quality of the children they raise, while poor families care more about quantity of children they bring up. (Becker,1960). Thus, unlike in developing countries, developed countries with a high income will have a lower fertility rate and consequently, higher female participation.
Moreover, women’s age also influences fertility rate. An additional child has been associated with 10-15 percent decline in the participation of the younger women and 5-10 percent decline in the participation of the older ones (Bloom, et al., 2009). Although some studies have found that fertility rate is a less important variable for explaining women’s participation in the labor force (Amador et al., 2013), fertility rate will be used as a dependent variable in this study.
3.4 Education
An important factor influencing the FLFPR is the level of education. Education when regarded as an investment on human capital, increases wages and provides opportunities to a woman to participate in the labor market. It can also promote consumption behavior, if the cost of not working is high (Tansel, 2001). The effect of schooling on FLFPR can be ambiguous as it depends on whether the income effect or the substitution effect is dominant. If education increases the opportunity of a higher earning, the income effect dominates and the effect of education on FLFPR will be positive. On the other hand, if higher earning encourages the worker to allocate time between work and leisure, the substitution effect dominates and education will have a negative effect on FLFPR. Based on the literature, the substitution effect usually dominates (Tansel, 2001).
Data and Selections of the Variables
Methodology
A significant body of literature has discussed the factors that affect the participation of women in the labor force. The econometric model used in this paper is following the model used in the literature where the female labor participation is the only dependent variable, and the GDP per capita and its quadratic form are used as a measurement of economic development. This paper will also consider some of the variables that may influence female participation according to Tsanial (2013). Four models will be estimated in this study in order to test the robustness of the main model. For all the models, the log of female participation in the labor force is the only dependent variable and log of GDP per capita and its quadratic form, education, and fertility rate are the common independent variables. For the first and second model, an unemployment rate will be included, with the inclusion of urban population for the second model only. In order to evaluate the "added-worker effect” and the “discouraged-worker effect” hypotheses, different measurements of unemployment rate namely, female unemployment rate and male unemployment rate will be considered. The third and fourth models will have female and male unemployment rates respectively, with the exclusion of the urban population.
Where t denotes time, and ε is a classical error term.
For the U-shaped hypothesis to hold, the signs for the logGDP and its quadratic form are expected to be positive (β1>0) and negative (β2<0), respectively. Educational attainment is expected to increase the chance of women to participate in the labor force (β3>0). For the unemployment rate, the discouraged-worker effect is expected to dominate in all models (β4<0). As the unemployment rate is high, women leave the labor market because the probability of finding a job is low and cost of finding a job is high. With more responsibilities resulting from a larger family size, women are less likely to participate in the labor force. Thus, fertility rate is expected to be negative (β5<0) and as the urban population increases, more women are likely to participate in the labor market (β6<0).
Econometrics Techniques and Empirical Results
Before estimating the above model, stability diagnostics will be conducted to determine the most appropriate functional form for the analysis of the FLFPR and economic development using Ramsey’s RESET test, the CUSUM test, and the CUSUM square test. Ramsey’s RESET test usually tests for omitted variables and improper functional form, while CUSUM test and CUSUM squares test examine the stability of the model by plotting a line between a pair of 5 percent level critical lines. If the line is within the 5 percent critical lines, it means the variance of the residual is stable.
A non- stationary series usually appears in a time series analysis. Therefore, to ensure that the standard assumptions for asymptotic analysis are valid, the Unit Root Test for stationary will be applied using Augmented Dickey-Fuller (ADF) to test the null hypothesis that there is a unit root in the series against the alternative hypothesis that there is no unit root in the series.
Based on the results shown in Table 2 below for ADF, I fail to reject the null hypothesis that there is a unit root for all independent variables and the dependent variable in the time series level, except for the unemployment rate and fertility rate. However, all the non-stationary variables are stationary in first difference; these variables are integrated of order one I(1), unemployment and fertility rate are integrated of order I(0).
6.2 Estimation of the models
In order to estimate the models, a stationary series is required. Thus, all non-stationary series will be transformed into stationary series by taking the first difference. The models will be estimated by the Least Squares estimation with the Newey-West covariance method to overcome serial correlation and heteroskedasticity in the data. The result represented in Table 3 below show that the variables explain more than 65 percent of model 1 and model 2, and more than 75 percent of model 3 and model 4. The results show that not only does economic development influence female participation but also educational attainment, fertility rate, and unemployment rate. Urban population is not found to be significant in model 2. Also, all the coefficient signs of the variables match the predictable signs. However, in model 3, educational attainment was found to be negative, but not significant. In model 4, the sign for all measurements of unemployment rate are negative and significant as expected. The result of female unemployment rates is obvious as with more women participating in the labor market, the unemployment rate for women declines. Using the total unemployment rate, the discouraged-worker effect dominates as expected. It is very costly for women to find a job when the probability of finding a job is low. Thus, in such cases they choose to leave the labor market. An explanation for the negative sign of the male unemployment rate could be vast availability of job vacancies in the labor market, which indeed presents an opportunity for both women and men to participate in it. Finally, GDP per capita and its quadratic form are found to positively and negatively correlated, respectively with FLFPR, which in turn is found to be associated with it, and conforms to the U-shaped hypothesis. Thus, as the nations develop, FLFPR increases.
Co-integration Test
For fulfilling the second objective of the paper, testing the causality between female participation and economic development and testing for co-integration will be applied to the dependent variable (FLFPR), GDP per capita and EDU. Since the FLFPR, GDP per capita and EDU have unit root and are integrated of the same order I(1), it is expected that a long-standing equilibrium may exist between the variables. Co-integration test can be applied using the Johansen co-integration test. The null hypothesis is that there is one co-integrating equation. Table 4 below represents the results of the Johansen co-integration test using Trace and Maximal Eigenvalue tests. Both tests conclude that there is one co-integrated equation between the variables (at most one). Since FLFPR and LNCGDP are co-integrated, then a Vector Error Correction Model (VECM) exists.
6.4 Vector Error Correction Model (VECM) and Granger Causality Test
VECM will be used to detect the causality between FLFPR, LNCGP and EDU. It can be specified based on the following models:
Δ denotes the first difference of the variables FLFPR, log GDP per capita and education. ϵt-1, εt-1and ωt-1 are the error correction terms. The optimal lag order (p) was determined using the model with the lowest AIC. The result of the VECM is displayed in Table A.6, given in the appendix. The results suggest that there exists a long-standing relationship between the variables since the coefficients of the error correction terms in the equations 5, 6 and 7 have a negative sign.
The causality test between the variables is presented in Table 5 below. The test result shows that FLFPR does not cause GDP per capita or EDU. However, GDP per capita does cause FLFPR and EDU as I reject the null hypothesis at 1 percent level. For education, I reject the null hypothesis that EDU does not Granger Cause FLFPR, and does not Granger Cause LNGDP at 10 percent level and at 1 percent level, respectively.
Conclusion
This paper examines the relationship between female participation in the labor market and economic growth in the United States using the annual data for the period 1970 to 2012. The econometrics results proved the U-Shaped hypothesis between female participation and economic growth. Additionally, statistically significant result was found with respect to all independent variables used in this paper including the log of the GDP per capita and its quadratic form, educational attainment, unemployment rate, female unemployment rate, male unemployment rate, and fertility rate. A long-standing relationship between female participation, economic development, and educational attainment was found from the VEC Model. However, only causality running from GDP per capita and educational attainment to female participation was found. No causality running from female participation to either GDP per capita or to educational attainment was observed.
One essential factor that both the United States and developing countries have in common is that improvement in the education of females in these nations has enabled the women to have as good an access to better education as the males, which has had a positive effect on the economy of these nations as a whole. Indeed, it has made the overall economy of these nations robust. The other thing that the United States and the developing countries have in common is the high participation rate of females in the labor force, which has also strengthened the respective national economies. However, the United States, being a developed and high-income nation is located at the upward part of the U-shape hypothesis, whereas the mid to low-income developing countries are located at its downward part.
Not only is the participation rate of females important, but the quality of females participating in market also matters. Therefore, critical policy issues for improving the access of women to quality employment by enhancing their educational attainment should be adopted. Women, especially those from poor families, should be made aware of the importance of their participation in the labor market as it benefits the macro-economy as a whole. In addition, policy-makers should take it upon themselves to formulate and implement policies that favor increased participation of females at all income levels so as to positively influence overall economic growth. It is also important to note that including females in the labor force will act as a step toward empowering the women so that they become self-sufficient and productive members of the society.
