|dc.description.abstract||In this paper, my main purpose is to find out the relations between male total fertility rates and female total fertility rates. The basic idea of doing such a research is from the previous researches available, from which I see big potentials in the field of male fertility research. The lack of male fertility data has been keeping researchers away from doing male fertility research. Little work has been done on male fertility researches compared to that of female. Throughout previous literatures, the relations between male and female fertility is an important and major topic been discussed by demographic researchers. We pay attention to the relations between the two fertilities because it will help us to learn the differences between male and female fertility pattern, as well as male fertility transition model. Since the present fertility transition model and theory are both based on female fertility, we need to know whether they are also applicable to male fertility (Zhang, 2011). In recent years, some data sources such as the United Nations Demographic Yearbook have made male fertility data available, which enable us to predict more accurate relations between the two fertility rates.
My hypothesis that male total fertility rates are of linear relations with female total fertility rates is based on the research achievements of Paget and Timaeus in 1994, who believed that the male standard fertility schedule could be expressed by a linear model of female standard fertility schedule.
In order to find out the relations between the two total fertility rates, I collect data of male fertility rates and female fertility rates of as many countries as I could. Most data are from the 2008 demographic year book of the United Nations. I made attempts of four difference regression models. The first one is an extremely simple linear regression model that regressing TFRm (male total fertility rates) on TFRf (female total fertility rates). The first regression model is rough, but it gives me confirmed results that a linear model is possible. Through further analysis on outliers of the first regression, SR (sex ratio) was regarded as an important omitted variable. The second regression model improves a lot by including SR as a second independent variable. When SR included, the value of R-squared is about 0.889, the regression results are better.
The next important process is a series of simulation analysis, which aims to test the effect of SR on TFRm and the effect of Am (age structure of men) on TFRm respectively. The simulation analysis provided four important conclusions. Firstly, the effect of SR on TFRm will become larger as the fertility rates increase. The effect of SR on TFRm is stronger for higher fertility rates population. Secondly, when sex ratio is in a narrow range, a linear form of SR is acceptable, while for a large interval of sex ratio, SR should be in a curvilinear form. Thirdly, since the realistic interval of SR is not narrow enough, a curvilinear form of SR is preferred in the regression model. Fourthly, Am has little impact on male TFR, thus age structure of men is not an omitted variable in the regression model.
With the findings from simulation analysis, another two regression models are tested. They are still of linear regression form, but the form of each variable is changed from linear to curvilinear. The third regression model is not satisfying, but the fourth one is perfect, with the result: LN (TFRm) = -0.078 + 1.094*LN (TFRf) - 1.032*LN (SR). Approximately, TFRm is equal to 0.925 times of TFRf/SR, where we see a curvilinear form of 1/SR in the result. It is further confirmed by the empirical time series data of Norway from 2000 to 2010 that TFRm is equal to 0.939 times of TFRf/SR. Confirmed by the time series data of Norway, the fourth regression model is even more promising than the previous three. Moreover, for a balanced population with SR equal to 1, TFRm is approximately 0.925 times of TFRf. This means the male total fertility rate is about 92.5% of the female total fertility rate. The reason why male TFR is a little lower than female TFR is due to a longer reproduction age range for men than that for women.
The significance of doing this research lies in the following aspects. When we find out the mathematical relations between the two total fertility rates, we can calculate male total fertility rates on the basis of female total fertility rates for every country in the world as long as its female total fertility rates are available. The application of our results might improve the world total fertility data system to a big extent. As a newly introduced fertility research method, male fertility will help to throw more lights on the fields such as male fertility transition pattern, fatherhood in a family, marriage pattern, and even population change. There is no doubt that doing such a research is of great practical values and theoretical significance.||eng