Labor Force Participation and Fertility in Young Women. fertility rates increase. It is assumed that was more women enter the work force then the

Transcription

1 Robert Noetzel Economics University of Akron May 8, 2006 Labor Force Participation and Fertility in Young Women I. Statement of Problem Higher wages to female will lead to higher female labor force participation while fertility rates increase. It is assumed that was more women enter the work force then the fertility rate decreases, an inverse relationship. However in the past 20 years, The United States have seen a positive correlation, as more women enter the work force, more women have children as well. This paper will research this for women in the 20s, because this is the age where they are finishing school, making career choices and starting families. II. Review of Literature 2.1 Dolores Ferrero Martinez and Amaia Iza Martinez and Iza state that is the past twenty years on the United States, there has been a positive relationship between female labor supply and total fertility rates, which are different in preceding years pattern. They created a generation model with a general equilibrium capable of generating a relationship between fertility and female labor supply. They argue that skill premiums have increased because of technological changes and higher demands for skilled employees, will therefore decreased the relative cost of child care services, which is unskilled. With an increase in female mean wages there is a positive effect on fertility rates, induced by the lower relative cost of child care.

2 2.2 Robert McNown In Robert McNown s paper, he uses cointergration methods to estimate and test nonstationary data applied to United States time-series data on age-specific fertility rates, female labor force participation rates, women s wages, unemployment rates and educational attainment, and male relative incomes. One finding that McNown had was that when one expects to see an increase in fertility because of a reduce in opportunity cost of childbearing, because of an increase in female unemployment, he finds that there is in increase unemployment, that there is a decrease in fertility because the excepted female income is now lower. There is a negative income effect of fertility, as income is decreased, and then fertility is decreased. He also finds that a rise in unemployment rates is associated with an increase of female labor force participation rates. The added worker effect could be the cause of this, because of the fact that women increase their market labor activity to offset their husband s diminished employment prospects. Women s education attainment also seems to have a positive effect on fertility, as stated by McNown. This is because the expected lifetime income has increased producing a positive income effect of fertility. In short, unemployment and education attainment have income effects of fertility. 2.3 Massimiliano Bratti One paper that I had read was written by Bratti. He looks at how education plays a role in determining both female labor force participation rates and fertility rates, and vice-versa, in Italy. He states that in most paper are focused on how a woman receives more years of education, that she is more likely to participate in the labor force, and less likely to have children. However he replies that there might be reverse effects too; if a

3 woman has a child or decides that she wants to work, that this could affect the number of years of education. Bratti uses an economic model that I am not familiar with, and after reading through it multiple times, the model is out of my realm. He uses a multi-nominal logit model and binary dependant variables for both female labor force participation rates and fertility rates and for the independent variable he uses education, amongst others. He sets it up by into age groups and education groups. The age groups are 21-24, 25-29, 30-34, and The education groups are primary, lower secondary, upper secondary, and tertiary. He then finds the probably of a woman in a certain age group with a certain level of education to have children and to work. Some of the results are as follows, among women that have had only five years of education, or primary education, and there is only percent fertility. Out of the age groups of women, the group that sees the highest level of employment is aged from 35 to 39 at percent and has a total fertility rate of If women that have a lower secondary education, or 8 years, then there is a percent probably that she will have a child. In the case of having a higher secondary education, or 13 years, the total fertility rates re increased to percent. For tertiary education, 17 years, the total fertility rates once again drop to percent. At any level of education the age group that seems to have the largest level of employed is the group of years. In tertiary education, that age group has an employment of percent of the women. III. Formulation of the Model The model that will be form is going to be one that has a dependant variable that is a probit, finding the probability of whether or not a young woman is going to work.

4 Some of the variables that will play a factor in the outcome would be the woman s income, the woman s spouse s or partner s income, and years of education. The belief is that if a young woman has more years of education she is more likely to have a higher income, through her wages, and therefore she is more likely to work. If her husband has a higher income or if she has a higher level of unearned income then she is not as likely to work. However if she has a lower income and years of education then she is not as likely to work. The regression with be as followed: Employed = β 0 + W_Income β 1 + H_Income β 2 + Edu β 3 + U It is assumed that as the woman s income increase or the years of education increase that the likelihood of her working is increased. However if the woman s spouse s or partner s income is increased or her level of unearned income is increased, then her likelihood of working is going to decrease. This regression is also going to be compared to another regression that is going to be set up the same; however the dependant variable is going to change, although it s still a probit: Child = β 0 + W_Income β 1 + H_Income β 2 + Edu β 3 + U Even though the equation looks the same, the signs and coefficients will change. It is believed that if a woman s husband is making more and her income is low and has less years of education, then she won t be likely to work yet is more likely to have child. Therefore, as a woman s income increases, or years of education increases, then her likelihood of having a child decreases. On the other hand, if her husband s or partner s income is increased, or her unearned income is increased, then her likelihood of having a child is also increased.

5 These assumption is based of the models and research written by others using European data. This paper will focus on United States data and different results might be found. IV. Data Sources and Description Variables Employed Child W_Income H_Income Edu W3_income W4_income W5_income H3_income H4_income H5_income Description The likelihood of a woman working The likelihood of a woman having a child The income that a woman receives The income that a woman s husband or partner receives Number of years of education that a woman has completed Woman s income that is below $30,000 Woman s income that is Between $30,000 and $50,000 Woman s income that is above $50,000 Husband s income that is below $30,000 Husband s income that is Between $30,000 and $50,000 Husband s income that is above $50,000 Percentage probability Percentage probability In 10,000s 1 = $10,000 In 10,000s 1 = $10,000 All these data can be found through the NLS website: This data was compiled from the National Longitudinal Survey (NLS) for Young Women. The Young Women group is comprised of women in their teens and 20s, they are finishing school, making job decisions and career choices, and starting families. V. Model Estimation and Hypothesis Testing

6 To run these models, SAS version 9.1 was used, with compiled data from the National Longitudinal Survey (NLS) of Young Women. The following is a table of variables with the number of observations, the mean, minimum and maximum of each variable. N Mean Minimum Maximum Employed Binary (0 or 1) Child Binary (0 or 1) W_Income In 10,000s H_Income In 10,000s Edu The following are not used in the model, but are used to run the results W3_income W4_income W5_income H3_income H4_income H5_income VI. Interpretation of the Results Regressors Employed Intercept (β 0 ) Woman s Income (β 1 ) Husband/Partner Income (β 2 ) Years of Education (β 3 ) This model, Employed = Ф ( W_Income H_Income Edu), will find the z-score. Using the normal table, one can find the probability of a young woman having a job. For example, if a woman s income is $30,000, her husband s income is $50,000 and she was a 4 year degree (16 years), the probably of her working is 92.1 percent Employed = Ф ( (3) (5) (16)) Employed = Ф (1.4132) use this z-score to find the probably on the normal table

7 Employed = 92.1% Regressors Children Intercept (β 0 ) Woman s Income (β 1 ) Husband/Partner Income (β 2 ) Years of Education (β 3 ) This model, Child = Ф ( W_Income H_Income Edu), will find the z-score. Using the normal table, one can find the probability of a young woman having a child. For example, if we use the same woman as before, where her income is $30,000, her husband s income is $50,000 and she was a 4 year degree (16 years), the probably of her having a child is 6.6 percent. Child = Ф ( (3) (5) (16)) Child = Ф ( ) Child = 6.6% VII. Conclusions and Limitations of the Study 7.1 Probability in a Table The following is a table that shows the probability that a young woman is likely to work and to have a child. This is broken into in three income groups for both the woman s income and for the husband s income. The groups are: less then $30,000; $30,000 to $50,000; and greater then $50,000. I found an average of both the woman s and her husband s income in each group, hence the other variables that were not used in the model, and used those number to find that probability of both being employed and having a child. Also, it is broken down by the years of education, assuming that a woman has graduated high school (12 years), graduated college with a 4 year degree (16 years), or the woman with advance degrees (18 years).

8 Husband s Income (number used, in 10000s, found by mean) Less then $30,000 (.8379) $30,000 - $50,000 (3.908) $50,000+ (10.715) Education Woman s Income (Probability of being employed in percent, probability of having a child in percent) High school Less then $30,000 ( 90.7, 10.6) ( 91.1, 15.9) ( 92.2, 33.7) Diploma $30,000 - $50,000 ( 96.6, 7.4) ( 96.9, 11.7) ( 97.3, 26.8) (12 years) $50,000+ ( 99.3, 4.5) ( 99.4, 7.5) ( 99.5, 19.5) 4 year Degree (16 years) Less then $30,000 ( 85.3, 4.2) ( 86.0, 6.9) ( 87.5, 18.4) $30,000 - $50,000 ( 94.1, 2.7) ( 94.4, 4.7) ( 95.1, 13.6) $50,000+ ( 98.6, 1.5) ( 98.7, 2.7) ( 98.9, 8.9) Advance Less then $30,000 ( 82.1, 2.4) ( 82.9, 4.3) ( 84.6, 12.7) Degree $30,000 - $50,000 ( 92.2, 1.5) ( 92.6, 2.8) ( 93.7, 9.0) (18 Years) $50,000+ ( 98.0, 0.8) ( 98.2, 1.5) ( 98.5, 5.6) Number used for Woman s income, less then $30,000 (1.531), $30,000 - $50,000 (3.854), and greater then $50,000 (6.785). 7.2 Expected Results Some of the results were as expected. As the husband s income increases, the probability of her having a child increases. This being that the husband can provide more for the family and there is less pressure on the woman to work. Also, as the woman s income is increased, the probability of having a child is decreased. If a woman is offered a higher income, then her opportunity cost of having a child has increased. Therefore she is less likely to forego the high wages to have a child. It is also expect that as a woman s income has been increased, the probability of her working is also increased. As a firm offers a higher wage to a woman, then the woman is drawn away from other factors in her life, children or school. This proves to be the case in these models that I ran using young woman. 7.3 Unanticipated Results

9 It is expected that as a woman s education increases, that her probability of her going into the work force is higher. However in this model, this does not prove to be the case. In fact the opposite is occurring. As she attains more education, the probability of her going into the work force decreases. This could be because of the fact that the data that is being used is of young women, women in their twenties. There are taking the NLS survey while they are still in school, and not in the work force. As a woman graduates high school, then she has a choice, she can enter the work force of go to college. If she decides to go to college, then she is not in the work force. Therefore, we see that if a woman has only a high diploma, she is more likely to work, because she would not be in college at the time of the survey. Moreover, if a woman decides to get an advance degree, she has prolonged her time away from the work force and therefore decreases the chances of her work, because she is in school. 7.4 Other Results The results also show that as a woman s husband s income increases, she is more likely to work as well. This show that their incomes are complements for each other. This could because they are both at a younger age and have decided to offset having children to build up financial stability for their future and their children s future. VIII. References McNown, Robert. "A Cointegration Model of Age-Specific Fertility and Female Labor Supply in the United States." Southern Economic Journal 70.2 (2003): Bratti, Massimiliano. "Labour force participation and marital fertility of Italian women: The role of education." Population Economics 16 (2003):

10 Martinez, Dolores Ferrero, and Amaia Iza. "Skill Premium Effects on Fertility and Female Labor Force Supply." Journal of Population Economics 17.1 (2004): 1-16.