A Regression Analysis Study on Rwanda and a Demographic Study on the Philippines.

Authors Avatar

ECON 3710 – Term Paper

A Regression Analysis Study on Rwanda and a Demographic Study on the Philippines

By: Aaron Tepperman ()

For: Professor Oystein Kravdal

Due: March 4, 2004

1) Regression Analysis - Rwanda

Model 1

It is important to note that these findings do not represent facts, only predictions based on a data set.  I ran a regression that included years of education and age.  The calculated intercept was 4, 583072596. This means that a women living in Rwanda with 0 years of education would, on average, have 4, 5 children.  To calculate the average amount of children for a 35 year old women living in Rwanda with 0 years of education, I added the intercept to the coefficient calculated for the age group 35-37, which was 1, 30237771.  This in turn explains that a 35 year old women living in Rwanda with 0 years of education would have 5, 8 children.  

Following the lecture on the importance of women’s education on fertility levels, it would seem quite reasonable to believe this first regression model – that there is an inverse relationship between years of education and number of children.  An explanation could contend that education, perhaps in the most likely scenario, causes women to delay fertility.  And this is why we see the inverse relationship.  Moreover, it is logical to trust the data that interpret a positive relationship between age and number of children.  This can simply be explained by the fact that younger women have a greater possibility for more education because, so to speak, the education ship has sailed for the aged.  And as mentioned above, more years of education, keeping everything else constant, will act to decrease fertility.  It is also possible to infer that a large degree of education could cause women’s reproductive years to pass without giving birth to a child.  

Model 2

In the second model I ran a regression that included the variables from the first model, but I also included three important dummy variables: urban? Muslim? Traditional religion?  I found that a women living in an urban area would have a coefficient of -0, 132983566.  This certainly accords with the lecture notes because in most rural areas it is significantly more difficult to obtain contraceptives and family planning advice.  Thus, it is a problem of availability and knowledge, unless, community based programs have volunteers that appear at your door, as we have seen in Bangladesh.  As well, rural areas may be more agriculturally based and therefore, families desire more children as they can act as laborers on the farm.  It also could be that in rural areas there is a higher infant mortality rate and as such, parents have more children to ensure they meet their desired amount of offspring.  Consequently, an urban women would have 0, 132 less children than a rural women.  Therefore, by subtracting the new intercept (4, 587088502) from the urban coefficient, I found that a women living in an urban area in Rwanda, with 0 years of education, is likely to have 4, 454 children.  On the other hand, Muslim women would, on average, have 0, 22 more children than the new intercept.  After learning that other Muslim countries have initiated widespread family planning programs and contraceptive distribution initiatives, it is unlikely that the Muslim religion acts as the reason for having more children.  Perhaps, it is the dominating influence of the husband and his refusal to accept contraceptives as legitimate, or it could be the belief of the woman that contraceptives are harmful to one’s health.  Conversely, a women of traditional religion would be predicted to have 0, 89 less children.  

However, it is important to note that all three of these dummy variables were insignificant.  For each, a t-stat was obtained that was less than the absolute value of 2.  Moreover, each variable has yielded a lower and an upper 95 percentile that are of opposite signs.  I can then determine that there is the possibility of having a value of 0, and as I have learned, this would disqualify the results.  Lastly, all of these dummy variables have p-values that are well above 0, 05.  The further the p-value is above of 0, 05, the more likely the results were computed by coincidence.

Model 3  

Including all of the variables from the previous two models, I added wealth as another dummy variable, and marriage age as a constant variable.  Wealth is an important variable as it can offer some insight into the socioeconomic status of the individual.  With wealth there is usually a significant level of education, and a greater ability to move away from norms in society, such as 4-10 children.  Subsequently, I would predict wealth to decrease the amount of children.  However, in the regression that I ran, this was not the case.  Wealth yields a positive coefficient.  Perhaps a greater income allows parents to afford more children.  Because I included marriage age as a constant variable, it is difficult to infer what effect it might have.  However, from the readings and the lectures, I believe that the younger a woman gets married, the greater the amount of children she will have.  This could be due to a longer reproductive period and/or unwanted births.  

It is in this final model that I would like to run two hypothetical situations.  Starting from the intercept (10, 33) I will say this individual women has 11+ years of education (-0, 89); is between the age of 38-40 (2, 21); lives in an urban area (-0, 14); is wealthy (0, 12) and her age at marriage was 28 (-0, 29 * 28 = -8, 12).  Running a regression that controls for all these factors yields a coefficient of 3.51.  Thus, it could be said that a women bearing all of these characteristics has 3.51 children.  A second hypothetical situation begins from the original intercept (10, 33); but this women has only 3-6 years of education (-0, 07); is older than 47+ (3, 75); is Muslim (-0, 47) and her age at marriage was 21 (-0, 29* 21 = -6, 09).  Before I calculate the finding, I predict that this woman will yield a coefficient that is not far from the intercept.  Indeed, in this situation this woman would have 7, 45 children.  As I have learned, both the lack of education, and being older than age 47 have contributed to this relatively high number of children.  

2) Philippines

Source: http://www.wpro.who.int/chips/chip02/phl.htm

A) Fertility

The Philippines is today experiencing one of the most rapid increases in population in Asia and the Far East. Demographers have monitoring the trend in fertility in the Philippines since the 1960’s.  The first national survey that included detailed measurements of reproductive behavior and fertility desires was the 1968 National Demographic and Health Survey (NDHS).  Fertility estimates from the 1978 national survey showed that fertility decline had accelerated during the 1970’s, around the same time that the introduction of family planning services were introduced.  But later, surveys revealed that a rapid pace of decline was not maintained.  Instead, brief bursts of rapid decline were followed by longer stretches of stagnation.  At the beginning of the 1990’s and though to the mid 1990’s the TFR was approximately 3, 5 births per women, a substantial distance above replacement level (2.1).  Unfortunately, recent projections by the UN Population Division (UN, 2001) show replacement level fertility not attained until the period 2015-2020.  Moreover, the NDHS estimates show, if anything, a decrease in the rate of decline in the 1990’s as compared to the 1980’s.  

Join now!

The 1998 Philippines NDHS shows an urban TFR of 3.0, that is, even in the segment of the population that I hypothesize would be more inclined towards small families, fertility remains substantially above replacement.  As well, rural TFR’s in the 1990’s appear to be relatively stagnant at a high level (4.7). Moreover, desired fertility remains some distance from replacement level: the wanted fertility rate in 1998 was 3.3, exactly the same rate as in 1993.

As in many countries, there are significant differences in fertility levels by region.  For example, fertility is more than twice as high in Eastern Visayas and ...

This is a preview of the whole essay