2. THEORETICAL BACKGROUND 6

ABSTRACT 2

1. INTRODUCTION 3

2. THEORETICAL BACKGROUND 6

2.1. T

T

D

T

6

2.2. E

T

F

8

3. LITERATURE REVIEW 12

4. METHODOLOGY 15

4.1. G

I

16

4.2. D

C

18

4.3. D

I

V

22

4.4. M

D

28

4.5. T

M

32

5. DESCRIPTION OF RESULTS 34

5.1. D

R

R

34

5.1.1 .D

R

R

E

2 35 5.1.2. D

R

R

E

3 37 5.1.3. D

R

R

38

5.2. D

T

M

42

5.2.1 D

T

R

E

2 42 5.2.2 D

T

R

E

3 44

6. ANALYSIS 46

6.1. A

R

R

E

2 46 6.2. A

R

R

E

3 49

7. CONCLUSION 55

8. APPENDIX 58

A

1: F

1980-2003 58

A

2: C

59

A

3: R

R

R

N

(

.2) 60 A

4: R

R

R

N

(

.3) 61 A

5: A

10: R

F

E

T

1955-2003 62 A

6: R

F

E

T

1996-2003 63

9. BIBLIOGRAPHY 64

ABSTRACT

In this master thesis the factors that have influenced the number of births in the Central and Eastern European countries between 1955 and 2003 are analysed.

Among the factors, are education per gender, employment per gender, wages per gender, GDP per capita, number of marriages, divorces and abortions, age at marriage and age at maternity. A multiple regression analysis and the economic theory on demography are used in order to find the factors that have an influence on the number of births. It appears that the results for GDP per capita, age at marriage, abortion, divorce, marriage, male employment, male education and female education are significant.

Keywords: number of births, economic theory on demography, social and economic

factors, opportunity costs, direct and indirect costs

1. INTRODUCTION

During the last couple of decades some countries of the world have experienced tremendous decreases in their fertility rates. The fertility rate is defined as the average number of children a woman will have during her lifetime.

By describing the

situation of the countries with respect to fertility one can group them into three different categories of fertility stages, namely in groups of countries with low fertility rates (less than 2 children per woman), very low fertility rates (less than 1.5 children per woman) and lowest low fertility rates (less than 1.3 children per women)

{Billiari(2005)}. In appendix 1 the total period fertility rates from 1980-2003 are presented. It can be seen that in the year 1980 there have already been countries with low fertility rates. Among those countries have been Belarus, Bulgaria and Croatia with fertility rates of 2.04, 2.05 and 1.92 respectively. With the exception of Germany and San Marino no other country has experienced very low fertility rates in 1980. This picture, however, has changed dramatically in the year 2003. While there have hardly been countries with very low fertility rates and no countries with lowest low fertility rates in 1980 many countries have experienced an enormous decrease in the fertility rate leading to many countries with very low and even lowest low fertility rates From appendix 1 one can see that Croatia, for instance has experienced low levels of

fertility in 1980. The fertility rate has decreased further and has reached a level below 1.5 children per woman by 2003. Moreover, with a rate of 1.33 children per woman the country has almost reached the lowest low fertility level. The fertility rate in Bulgaria has decreased even more than in Croatia and it has reached the lowest low fertility level with 1.23 children per women by 2003. Although the fertility rate has been described in this paragraph, the master thesis will focus on the number of births in the countries in question and not on the fertility rate itself.

After having had a short look at the fertility rate development of the last 23 years it becomes apparent that something has to be done in order to improve the number of births. But why is that so important? What might be the consequences for the countries which experience very low or even lowest low fertility rates? The most evident consequence will be that low fertility rates would lead to a different age

1

Youthink.worldbank.org/glossary.php

structure in the society which would mean that a higher proportion of people will be older than 65, for instance, and the proportion of younger people will decline. This change in the population structure could lead to a decline in the work force, for there would be more retirees and fewer workers. This then, according to Rogers (2000), can lead to a decline in output per capita and living standard if productivity level remains the same. With regard to national savings it can be argued that the demographic change will have a negative effect. It is said by Rogers et al (2000) that there are two reasons for a decline in the national savings rate to occur, namely the increase in the proportion of elderly and the fact that government spending is higher in social security and heath care, which contributes the elderly in the population. Since this share will have to increase, public savings will be reduced everything else (taxes, incomes) being equal. Although on the other hand due to the decrease in the proportion of younger people government spending on education will decrease, it might not be enough to offset the increase brought about by the elderly in the population. In addition, the pensions for the retirees are usually paid by the working population, for when the working population is smaller than the proportion of elderly there would be a miss match in what the working population can contribute and what the elderly have to obtain.

Next to the question on the consequences of low fertility rates another aspect appears

to be of importance, namely the factors that cause this development. There are several

theories that try to explain why fertility declines in some countries of the world. There

is the transition theory which mainly states that as a country becomes richer fertility

and mortality rates decline. The economic theory on the other hand tries to explain the

low fertility rates with opportunity costs and other economic terms. Further, there are

theories on psychological and biological factors that cause the fertility to decline. In

this thesis it will be mainly dealt with the first two theories, namely the transition and

the economic theory. However, in the analysis section some of the other theories will

be discussed as well. The aim of this master thesis will be to find factors that have

influenced the number of births in the Central and Eastern European Countries during

1955 and 2003. The relationship between the number of births and the variables

chosen in this masters thesis, such as the number of abortions, GDP per capita,

education per gender, the employment rate per gender, number of marriages, number

of divorces, mothers age at maternity and the age at marriage will be examined by

using multiple regression analysis. The countries that will be focused on are the following: Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, Slovenia and Romania. The contribution of the master thesis is that it focuses on the development of the number of births in the Central and Eastern countries and examines a number of economic and social variables that have a role to play in this development. Several other researches that will be illustrated in the literature review, deal either with economic or with social factors. Further, those researches deal mainly with one country or a relatively short time span. Finally, the value of the master thesis is that it uses several countries, several variables from both economic and social area and a time span of roughly five decades. Therefore, the research question that will be answered in this master thesis is:

“What are the factors that have influenced the decline in the number of births in Central and Eastern European Countries between 1955 and 2003”

The master thesis is structured as follows: In section two relevant theories on fertility and demography such as the demographic transition theory will be illustrated. After having described the background and the relevant literature the methodology will be elaborated, the dependent and independent variables will be described, and

hypotheses will be formulated. This section will then be followed by section 5 which

deals with the presentation of the results obtained from the regression analysis. The

section on results will then be extended by the analysis of those and finally in section

7 there will be a conclusion.

2. THEORETICAL BACKGROUND

In this section some theories on fertility rates will be summarized. Among those theories the theory of demographic transition and economic theories of fertility will be presented. Those theories try to explain which factors might influence the

demographic development and are the theories most often used in the literature. There are of course several other theories regarding demographic development, such as the evolutionary and the psychological theory. However, those theories will not be dealt with in this section for describing the different theories is beyond the scope of this thesis. Those theories might be used in the analysis section if some results are obtained that are not explained by the theory described in this section.

2.1. The Theory of Demographic Transition

The first version of Malthusian Theory that population grows faster than food could not be supported by the developments in the nineteenth and the first half of the twentieth century. In contrast to Malthusian theory the population has grown at a lower rate than production and has already been decreasing in the beginning of the twentieth century. This has led to the development of the theory of demographic transition which at first has served the purpose of describing the demographic development. This new theory has been categorized into three phases by Landry (1934, 1945), namely the primitive phase, with both high fertility and high mortality rates (in this phase the main determinant of population growth has been mortality), the second phase the so called intermediate phase which is characterized by declining mortality and fertility rates and finally the modern phase characterized by both low fertility and mortality rates. In the last phase the population growth rate is, according to theory, not influenced by economic factors.

However, there is also another way to distinguish the different phases. According to

Blacker (1947) there are five phases of transition namely (1) the high stationary

phase, with high mortality and fertility rates, (2) the early expanding phase in which

mortality starts declining with fertility rates being unchanged, (3) the late expanding

phase when fertility rate also starts declining (4) the low stationary phase when

fertility and mortality rates reach the same level and finally (5) the diminishing phase in which fertility falls even below the mortality rate which leads to the diminishing of the population. However, Blacker´s and Landry’s theories are mainly descriptive and do not answer the question why fertility falls.

Referring to the reasons for fertility and mortality decline several explanations are possible. Reasons for mortality decline can be, according to Andorka (1978) of economic nature such as a higher living standard and the improvement of the health system. Fertility decline on the other hand can be explained by the decline of the mortality rate. Since infant mortality declines people do not need to have many children in order to overcome the possible death of a child who in former years has served as the parental supporter, when those become old. This explanation is

supported by Knodel (1974) who has found that in Germany 41% of the decrease in the fertility rate between 1871 and 1939 can be indeed explained by the decrease in child mortality. Child mortality, however, is not the only factor causing fertility to decline. Other factors mentioned are economic and social changes brought about by economic development. While mortality falls relatively fast after an economic change (for instance improvement of the health system) fertility rates do not decline

immediately after such a change due to religious and social constrains. First social institutions have to change in order to bring about a decline in the fertility rate.

In this paragraph the social change caused by economic development will be described a bit more detailed. It is said, by Andorka (1978), that the small family structure has arisen in the urban society. Causes for this development cannot be clearly stated but many factors have a role to play. The new mobility of young people coupled with the anonymity of the city has led to a reduction of the pressure to live the traditional life, expected from the family. Furthermore, people have needed more skills and new opportunities have arisen which have increased the costs of having children. In addition, these costs of having children have also increased by better education. Finally, women have gained more independence due to employment different from house keeping. All those changes have led to a change in believes and values of the people which in turn has led to a decrease in the number of children.

Although this argumentation has been supported by several authors, no real evidence

could have been found on the influence of education and urbanization on the decline in the fertility rate.

In spite of the advantage of this theory which is the ability to predict the level of fertility and mortality rates the major disadvantage mentioned by Coal (1976) is that the level of modernization that leads to the demographic transition is not defined properly. Furthermore, there are several examples which challenge this theory. Such examples are, Bulgaria, which entered the transition period (referred to demography and not economics) during a time of little economic development or Bangladesh, one of the least developed countries in the world that experienced a decline in the fertility rate.

2.2. Economic Theories on Fertility

Referring to economic theory on fertility two major assumptions can be stated:

1. Couples behave in a rational way when they decide how many children they will have

2. Parents perceive children as consumption goods

The choice of consumption of families is determined, according to consumption

theory, on factors such as the preferences for the goods offered on the market, their

prices and the parents´ income. Figure 1 below illustrates this relationship. On the

horizontal axis there is the number of children wanted and on the vertical all other

goods are presented. The slope of the so called indifference curves illustrates how

much a person is willing to give up from one commodity in order to receive more of

the other commodity, the so called marginal rate of substitution. The straight lines

drawn in the figure represent the so called budget line, which shows the combinations

of the two “goods” that can be bought with a certain amount of income. In this respect

this theory, according to Andorka (1978), leads to the conclusion that couples with

higher incomes will have more children than couples with low incomes.

Figure 1.

Source: Determinants of fertility in modern societies

Although the conclusion seems to be very nicely stated it can be criticised that in reality there are also cases where people with higher incomes have fewer children and people with lower incomes have more children {Andorka (1978)}. In order to explain this dilemma new formulations of the economic theory have been stated. Among those new explanations is the explanation given by Becker (1960) which basically states that the reason for this apparent negative relationship between fertility and income is due to the level of information on birth controlling methods. According to Becker, people with low incomes do not have the necessary information on

contraceptives which leads to more children. Furthermore, Becker argues that when all groups in society are equally informed the positive relationship will return.

Moreover, Becker states that people with higher incomes also want quality (which means better educated children) and that they are willing to pay this extra amount of money.

In contrast to Becker, Okun (1958) does not agree with the consumption theory because in his opinion a distinction has to be made between children and

consumption. The reason for this distinction is due to the different prices faced by

different income groups. This is in contrast to the nature of commodities for which the

prices are equal for all income groups. The rationale behind Okun´s idea is that minimum cost of the child is lower for people with lower incomes than for people with higher incomes. This is the case because children of low income parents might get lower education as well which reduces the costs of children. However, children of high income parents will need to have higher education which is considered to

increase the cost of children. This might explain the situation why people with higher incomes decide to have fewer children, and moreover can explain why fertility falls with income over time. Summing up due to increased costs of children parents decide to substitute other commodities for them.

The first to establish the new fertility rate theory has been Leibenstein (1957 1974) who has tried to explain the relationship between income and fertility by using benefits and costs of children. Leibenstein has defined three types of utility which illustrate why people want to have children. The first is the consumption utility which states that couples get children in order to have pleasure. The second utility, the so called work or income utility represents the value that the children, after entering the labour market can add to the household. The last utility is the security utility which reflects the ability of the child to care for the parents, especially when they become old. Next to these benefits that parents can derive from having children, there are also costs associated with them, namely the direct costs which consist of determinants such as nutrition, education, clothes and housing and secondly the indirect costs which are brought about by opportunity costs. An example of such indirect costs might be that the mother has to leave her job, stay at home and take care for the child. Leibenstein mentions also three factors that might influence the benefits and the costs and which explain why parents decide to get a third, fourth or fifth child (Leibenstein has not been interested in the question why parents decide to get the first two children). Those factors are increases in per capita income, the decline in mortality and changes in occupational structure.

Referring to an increase in income Leibenstein states that not all three utilities would

change, rather the consumption utility will be unchanged and the other two work and

security utilities will decrease due to the longer education of children at a higher level

of per capita income which will lead to a late entrance in the labour market. In terms

of direct and indirect costs it can be stated that when per capita income rises the direct

costs rise. With regard to indirect costs it can be said that they are important for high per capita income for higher amount of money is given up in order to raise children.

With respect to the decline in the infant mortality rate results are obtained which state that this leads to increases in the utilities due to the lower number of children who die before entering the labour force. Finally, changes in the occupational structure lead to higher direct costs due to the higher costs of education.

Extension to the role of indirect costs has been made by Mincer (1963) who has stated that consumption of goods, children in particular brings opportunity costs such as time spend on the children. In this respect the mother has to give up her job and the wage earned in order to take care of the child. Mincer has come up with the following equation for the demand of children:

X

= β

(X

– X

) + αX

+ β

X

+ µ

Where X

represents fertility, X

and X

represent the husband´ s and the wife´ s potential earnings respectively. Therefore X

+ X

stand for the potential income of the family and X

is the husband´ s education. In this context Mincer has stated that according to economic theory on fertility the parameters should be as follows: β

> 0, α < 0 and β

> 0. In order to test his hypothesis he has used average data for cities and areas in 1940 and 1950 and has found the following results: the husband´ s wage is related positively to fertility on the one hand, and on the other hand the mother´ s potential wages is related negatively to fertility. So in conclusion the fertility rate is lower the higher the mother´ s potential wage and the lower the husband´ s wage.

These results are important for this thesis for among the variables also the wages per gender are taken into consideration when testing the determinants of the number of births.

With respect to the theories mentioned above it is noteworthy that although the

theories appear to be contradicting, they have some things in common. They try to

find the reasons why couples with lower incomes often have more children than their

richer counterparts. In their attempt to find an explanation the authors mentioned in

this section use similar approaches, such as opportunity, indirect and direct costs.

Therefore one can see that the theories are not contradicting but rather extensions of each other.

3. LITERATURE REVIEW

After having discussed the demographic theories relevant for this thesis there will be a summary section of several articles related to the topic of determinants of fertility.

Some articles will use a similar approach as the one used in this master thesis namely multiple regression analysis. The first article, however, does not use the same

approach but does test for some variables also used in this master thesis which explains why this article is used in this section. Next to the particular articles the contribution of the master thesis will be discussed in this section.

In the article “Analysis of Determinants of fertility decline in Czech Republic” the authors apply the multi period model of birth process to the family and fertility survey 1998 in order to find the determinants of fertility decline. The authors are able with their model to estimate which variables lead to a postponement or a speeding up of births. The results obtained are very interesting for this thesis for first it tries to explain the determinants of fertility decline and second it tries to explain the factors for a country which will be dealt with in this paper. For instance the influence of education will be also tested in this thesis and it is possible that although the

approaches are different, for in this thesis multiple regression will be used, the results might be the same. The results that have been obtained from Klasen and Launov are the following: With regard to the education level they have found out that the higher the level of education is, the later the first child will be born. This is also a hypothesis that will be tested in this master thesis. When looking at cohort that are older, that means the cohorts that got children during the socialist time it has been found out that education has not postponed the first birth as much as it has for the younger cohort.

According to Klasen and Launov women with higher education have postponed birth between half and two years during the socialist era while women with the same level of education postpone births between two and three years during the transition.

Another consequence of the socio-economic changes is that women do not just

postpone birth longer but they also do not give birth to as many children as women with the same level of education have done during the communist era. Another variable that is being tested in the paper by Klasen and Launov is the employment intensity of women. For this variable the authors have got the expected results namely that higher employment intensity will lead to a postponement of birth and fewer children. However, in this case the situation between the older and the younger cohorts are not as obvious as before and employment intensity is not a major

determinant. In the case of postponement there is a difference in timing between the older and the younger cohort implying that combining the full-time job with children has become increasingly difficult with transition. The third variable in the article is ownership of the dwelling. The result that has been obtained is that ownership of the dwelling does not influence the fertility decline during transition.

Avery M. Guest (1974), in contrast to the former article uses multiple regression analysis in order to find the relationship between the crude birth rate and illegitimacy, the marriage rate, legitimate fertility and sex and age composition. This article

therefore does not just use the similar statistical tool but also tries to find the impact of marriages on the fertility rate. In his paper Guest analyses 79 countries and tries to explain the social/economic influences on the fertility rate. The results that have been obtained by Guest are that the marriage rate, the illegitimacy and the legitimate fertility rate have a strong influence on the crude birth rate, while the sex and the age composition does not influence the dependent variable significantly. The result obtained for the marriage rate is interesting for this paper for Guest has been

analysing the countries in the 1960s and during this time marriage has been important.

However, has this impact changed during the last five decades? This will be found out in this master thesis. Guest further has found that although the marriage rate and the legitimate fertility rate are in all regions among the three variables that influence the crude birth rate, the effect of illegitimacy varies across the regions. In Europe, for instance, illegitimacy has not a strong effect on the crude birth rate and in Asia has not had an effect at all. However, in American-African sample illegitimacy is the most important factor that influences the crude birth rate.

A further article presented in this section is the article “Determinants of Fertility in a

Developing Society: The Case of Sierra Leone”, written by Suhas L. Ketkar (1979).

This article like the previous one uses multiple regression analysis in order to estimate the determinants of fertility. Ketkar (1979) uses the household model which is similar to the economic theory mentioned in the last section. This approach states basically that the more expensive a child gets, the fewer children will be born. In addition, the time spent on children and the price of other goods determines the price of the child, according to this model and according to the economic theory. Ketkar (1979) uses this household model and modifies it because according to him also other components have to be included into the model. With respect to Sierra Leone there has been a family system which has reduced the cost of childbearing. Therefore, Ketkar (1979) has assumed that the more members a family has the higher will be its fertility. The second component that is important, according to Ketkar (1979) is child mortality.

Although Sierra Leone has a different development stage compared to the countries in question in this master thesis the paper just described is interesting for this thesis because Ketkar (1979) also uses variables that are used in this paper as well. Those variables are the age at marriage, the wife’s expected life income which might be compared to the wage rate and the wife’s level of education.

The results obtained by Ketkar (1979) are the following: child mortality increases the fertility rate which is quite reasonable due to the fact that the families in developing countries rely on their children to care for the parents when those become older. The wife’s present age and the size of the family also have a positive influence on the fertility rate. With respect to the husbands level of education and its expected income the influence is also positive. This is also in line with the expectations. Changes in the education level and the monthly income of the husband might have a substitution and an income effect. Since Sierra Leone is a developing country and men do not spend much time on child rearing it might be expected that the main effect is the income effect. With respect to the wife’s education level and the expected monthly income Ketkar (1979) has found out that these variables together with the age at marriage has a negative effect on the fertility rate. This is a result which is also expected for the regression analysis conducted in this thesis.

Another article that is relevant for the literature of this thesis is the article “Fertility and Women’s Employment: A Meta-Analysis”, written by Matysiak and Vignoli.

This article is important because it uses one of the variables, namely women’s

employment and tests the relationship between this variable and the fertility rate. The authors question the theory that women’s employment decreases fertility rate. In their argumentation they point to the 1980s when the countries with highest women’s employment have shown to have higher fertility rates. The authors have assumed that the negative relationship between women’s employment and the fertility rate is due to the missing of places in kindergartens or other facilities where children can be cared for while the women are working. The results that the authors have obtained are consistent with their assumptions. The results have varied with the regimes, the support and the attitudes toward working mothers. Further, they suggest that the impact of employment is less significant in social democratic states, due to the liberal attitude toward working mothers and the possibility to combine work and

childbearing.

The relevant articles show that there are many different ways in order to investigate the determinants of fertility. In order to determine the factors that led to a lower number of births and its relationship to other variables such as employment per

gender, GDP per capita, wages per gender, education per gender, number of abortions, mothers´ age at birth and the number of marriages and divorces regression analysis will be used. Further, this master thesis, unlike the previously mentioned articles, will examine eleven countries and will use more variables.

4. METHODOLOGY

After having described the theory of demography and summarized some researches on

this topic the model used will be established in this section in order to answer the

research question “What are the factors that have influenced the number of births in

Central and Eastern European Countries between 1955 and 2003” .The methodology

part consist of five subsections, entailing data collection, the description of dependent

and independent variables, formulation of hypothesis and derivation of the model

used. Finally, there will be a section where the model will be tested on whether

heteroscedasticity and autocorrelation are present and whether the model is

appropriate.

4.1. General Information

The first step to establish a model that is used in order to test the research question and the hypotheses stated in a later part of this section is to explain which countries have been chosen and which period will be examined. The countries in question are the following: Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia,

Lithuania, Poland, Romania, Slovakia and Slovenia. The period for which the data is used is the period from 1955 until 2003 for social and 1996 until 2003 for economic factors. However, the fact that the countries in question have been organized

differently than they are nowadays and that they have formed other countries, for instance, led to the fact that also data on social factors for countries such as Croatia, Czech Republic, Slovakia, Slovenia, Estonia, Latvia, Lithuania is available from the late eighties and early nineties onwards. Before they became independent they

belonged to Yugoslavia (among others Croatia and Slovenia), former Czechoslovakia (Czech Republic and Slovakia) and former Soviet Union (among others Estonia, Latvia and Lithuania). Therefore, Czechoslovakia (Czech Republic and Slovakia), Yugoslavia (among the countries that belonged to the former Yugoslavia have been Slovenia and Croatia), Poland, Bulgaria, Romania and Hungary are used for the period 1955-1993 and after 1993 until 2003 Czechoslovakia and Yugoslavia are compensated by Czech Republic, Slovakia, Croatia and Slovenia. Due to the

availability of economic data such as wages per gender, employment per gender and education per gender from 1996 this data will just be used for the period 1996-2003 for the countries Hungary, Latvia, Slovakia and Slovenia. It might seem that the group taken for the economic factors is rather small and perhaps not representative but at the second sight it is. Slovenia can represent the Yugoslavian countries (for the social factors Croatia and Slovenia are taken and those two countries are relatively similar in size), Latvia can represent the group of Baltic countries, Hungary can stand for the Central European Countries and Slovakia can stand also for Czech Republic. To have a better overview of the countries used please consider the table 1 below.

The countries are interesting to examine because of their past in which they have had

a centrally planned economic system and then turned to market economies. Those

countries have been able to catch up quite fast and with the exception of Croatia all

countries mentioned above have entered the European Union in 2004 or 2007. The

relevance of the research question has already been discussed in the introduction part and explains why this topic has been chosen for the master thesis. The reason

mentioned is that due to an aging population, financial problems and problems with the social security system might arise. Therefore, governments, by knowing which factors influence the fertility rate, might introduce policies in order to improve the low fertility rate and prevent the countries from this kind of scenarios.

Summing up it can be said that two regression equations will be used. One for the social factors (1955-2003) and one for the economic factors (1996 and 2003). Due to the fact that Croatia, Slovenia, Slovakia, Czech Republic, Estonia, Latvia and

Lithuania have belonged to other countries and have become independent in the late eighties and early nineties, there will be a replacement of the former countries by the newer ones from the time of independence in the case of social factors. For an even more detailed overview of the countries used and the time span when they are used please consider appendix 2. In the following section the data collection procedure will be described.

Table 1: Country Selection

Countries used in the sample from 1955 Countries used in the sample after the collapse of the communist system

Bulgaria Bulgaria

Czechoslovakia (until 1990) Croatia (1989)

Hungary Czech Republic (1992)

Poland Estonia (1988)

Romania Hungary

Yugoslavia (until 1989) Latvia (1988) Lithuania (1988) Poland

Romania

Slovakia (1988)

Slovenia (1988)

4.2. Data Collection

The data for the variables has mainly been collected from the Demographic Yearbook published by the United Nations. Among this data is the data on the number of births, number of divorces and marriages. Also included in these yearbooks has been the number of abortions. These data has been collected from 1955 until 2003. The employment rate per gender and the youth education attainment level per gender which is used for education and are used for the years 1996 until 2003 have been collected from Eurostat. The wage rate per gender has been collected from the National Statistic Offices for the period from 1996 to 2003. However, data on wages is available just for Slovenia, Slovakia, Latvia and Hungary. With regard to GDP per capita it can be said that this data has been collected from the Groningen Growth and Development Centre Historical Statistic for the World Economy edited by Angus Maddison. It is noteworthy that data that will be used in the regression analysis later on is panel data which consist of time series and cross-section data. In the next paragraphs there will be a more detailed description of the measurements and the sources of the data used in this master thesis.

The dependent variable used in this thesis namely the yearly number of births is also obtained from the Demographic Yearbook for the period 1955-2003 issued by the United Nations. The United Nations have obtained the data from sending

questionnaires to the countries and collecting their responses. The number of births provided by the yearbook is related to the live births and grouped by year of

occurrence. Live births are defined as follows: “complete expulsion or extraction from its mother of a product of conception, irrespective of the duration of pregnancy, which after such separation, breathes or shows any other evidence of live such as beating of the heart, pulsation of the umbilical cord, or definite movement of voluntary muscles, whether or not the umbilical cord has been cut or the placenta is attached: each product of such a birth is considered live-born irrespective of gestational age.”

Further, not only births that occur in marriage but also illegitimate births are included

while stillbirths are excluded. The data used from the yearbook is the number of births

and not a percentage the so called birth rate which is calculated as number of births

per 1000 mid-year population. However, in order to see whether the results obtained

from the regression analysis differ when the number of births or the birth rate is taken,

there will be an additional regression with the birth rates. However, the main focus of this thesis will be on the number of births. In this respect in the yearbook there is also a section on the birth rate and the number of births related to the mothers´ age, which is one of the independent variables used in this thesis. This data is measured as the number of live births of a mother in a certain age group. To be better able to include this variable in the regression analysis an average age has been calculated for all years on the mothers' age at birth.

Beginning with the description of the independent variables it can be said that GDP per capita is obtained for the period 1955 until 2003 from Development Centre studies

“Monitoring the World Economy” edited by Angus Maddison. Further, it is

noteworthy that some adjustments had to be made before using the GDP stated by the former communist countries. Usually GDP is calculated by using the value added approach and including the whole economy. With regard to this approach the GDP can be verified in three ways (1) by looking at the production side and checking whether it is the same as the sum of value added in different sectors, such as industry, agriculture and services, (2) by looking at the sum of final expenditures which is then the GDP minus imports and plus exports and (3) by looking at the income side and comparing the sum of profits, rent and wages with the GDP (it has to be the same).

The need for adjustment of the GDP provided by the communist countries has been due to the fact that those countries used different ways of measuring the economic performance. Unlike the other countries, communist countries have excluded many service activities and the activities that have been included in the calculation of GDP have been calculated by using a different method instead of using value added measures. This often led to double counting in those countries. In order to improve this Abram Bergson (who is mentioned by Angus Maddison in “Monitoring the World Economy) has established the so called adjusted factor cost framework for measuring the Soviet output by adjusting the prices for goods to equal average cost.

This adjustment has led to the opportunity to compare GDP of the Soviet Union with

that of Western countries. The same adjustment has been done for the other Eastern

European countries, such as Czechoslovakia, Bulgaria, Romania, Poland and

Yugoslavia from the beginning of the 1960s.

When referring to the number of marriages there, again, appears to be difficulties on how to measure marriages. In general just marriages are counted that are recognized by the country’s laws and regulations and this can differ from country to country. In some countries just the official marriage might be important in order to be recognized while in other countries the religious marriage is the only one which is recognized.

Next to the crude number of marriage there is also, like with the birth rate, an age specific marriage rate, which is calculated as “the number of marriages of brides and grooms of a specific age per 1000 marriageable persons at this age.”

Marriageable refers to widows, singles and divorced people that are older than 15. In order to be better able to use the data for the regression the average year at marriage has been calculated from the data provided by the United Nations. Here, like with the dependent variable not the marriage rate but the actual number of marriages will be taken. And, again like in the case of birth, the marriage rate will be taken to compare the results of the regressions for the number and the particular rate.

With respect to the data for divorces, which is also obtained from the demographic yearbook issued by the United Nations, the number of divorces will be taken for the main regression and the divorce rate which is measured as the number of divorces per 1000 persons will be taken for an additional regression in order to compare the results and to see whether there might be differences. Divorces are defined by the United Nations as “final, legal dissolutions of marriage which confer on the parties the right to remarry”. The rates are the actual percentages which improves the ability to compare the countries on those characteristics.

Another variable used in this thesis is the number of abortions. This data, like the previous data is collected from the Demographic Yearbook issued by the United Nations. The abortion rate is going to be used from the 1966 onwards and is described in the yearbook as “legally induced abortions by age and number of previous live births of women”.

This data, like other data from the Demographic Yearbook has been collected by the means of a questionnaire. However, the limitations of this data are that just abortions that have been done in hospitals are recorded and other sources obtained from the countries might be incomplete.

2

Demographic Yearbook 1949-1950 p. 26

3

Demographic yearbook 1971 page 39

For a better overview of the variables used, their abbreviation, the source and the measurement, please look at table 2 below.

Table 2: Data Collection

Variable Source Measurement

GDP = GDP per capita “Monitoring the World Economy” Angus Maddison

Value added approach (in $)

Birth =number of births UN: Demographic Yearbook

Number of live births (whether occurred in or outside marriages) Marriages= number of

marriages

UN: Demographic Yearbook

Number of marriages that are recogized by the country’s laws

Empm= male employment Eurostat Number of men between

15 and 64 divided by the male population of the same age

Empf= female employment

Eurostat Number of females

between 15 and 64 divided by the female population of the same age

Wm= male wages Eurostat Gross monthly wages in

Euro

Wf = female wages Eurostat Gross monthly wages in

Euro Abo= number of abortions UN: Demographic

Yearbook

“legally induced abortions by age and number of previous live births of women”

Divorce=number of divorces

UN: Demographic Yearbook

Number of divorces

Edum= male education Eurostat Males aged 20-24 that attained at least secondary education/by the male population of the same age Eduf= female education Eurostat Feales aged 20-24 that

attained at least secondary education/by the female population of the same age

4.3. Dependent and Independent Variables

The dependent variable in this master thesis is the number of births of the countries in question. This variable will be analysed and the factors that might have an influence on it will be tested. The number of births among the migration and the mortality rate is one of the three parts of demography. The independent variables can be grouped in economic factors, such as employment per gender, GDP per capita, the school level per gender and the wage rate per gender and the so-called social factors such as, mothers mean age at birth, the number of abortions, the mean age at marriage (just the age of the bride not the age of the groom) and the number of divorces and marriages.

The first independent variable that is going to be described is the average age at maternity. Referring to this variable it is important to distinguish the timing of having a baby and the number of babies that one has. It might happen that women who become mothers for the first time at the age of 30 give birth to just one or two

children while women who become mothers at the age of 20 give birth to two or three children. In addition, it is more often observed according to David Coleman that women who become mothers at early ages give birth to more children than women who give birth for the first time at age of 30. Therefore, the age of the women might be an important factor in determining the number of children. A similar argument has been given in the literature review on fertility determinants in the Czech Republic.

The women that have postponed childbearing have also a lower number of children.

Therefore, the hypothesis formulated in this context is the following:

Hypothesis 1: Higher average age at maternity has a negative influence on the number of births.

Most of the children in advanced countries are born in families where parents are married. Therefore, the proportion of marriages and of divorces is playing an important role in determining fertility. However, there are also births outside

marriages. Marriage has been very important in former decades but the importance of this institution has changed. Further it is said that during the baby boom in the

twentieth century the age at marriage has declined and the proportion of married people has increased. So referring to this observation it is assumed that the number of marriages is positively related to the number of births. Furthermore, when talking about the number of marriages and their effect on the number of children, the age specific marriage rate is important. It has been assumed as described in section 2 that most children have been born within marriages. Therefore, postponing marriages might also have a negative effect on the number of births of the particular country.

The formulated hypotheses are then:

Hypothesis 2a: Marriages have a positive influence on the number of births.

Hypothesis 2b: Higher average age at marriages has a negative influence on the number of births

After having discussed the number of marriages we will now turn to the number of

divorces. Here several possibilities might occur depending on whether it is seen from

a woman’s or a man’s point of view and the child support policies in the particular

country. If a man gets divorced and he already has, for instance, two children, then if

child support is asked from him he might decide not to have more children when he

remarries. On the other hand, if the child support policies are not that strict it might be

that the man decides to remarry and to found a new family. In the case of a woman it

might be assumed that if she already has two children she will have difficulties in

finding a new husband and arranging her life so the chances that she will have

additional children after remarriage are quite low. However, this shows how difficult

it is to estimate whether the number of divorces will have a negative or a positive

effect on the number of births. Nevertheless, the hypothesis that is formulated in respect of the number of divorces is:

Hypothesis 3: Divorces have a negative influence on the number of births.

The next independent variable used in the model is abortion. Abortion can be

compared to contraceptives, although abortion is not used as birth control but is rather the last resort when someone does not want to have a child. However, the

consequences are the same for the contraceptives and abortion; the prevention of getting pregnant in the case of contraceptives and the prevention of giving birth to the child in the case of abortion. Abortion makes it possible for a couple to plan the family size and to become active in the case that the woman becomes pregnant

although that was not wished. However, in order to influence the number of births, the number of abortions has to be relatively high and legalized (in order to be measured).

In the case that abortion would not be present the number of birth would be higher than in the case when abortion is indeed allowed. Therefore, the following hypothesis is formulated in respect to abortion:

Hypothesis 4: Abortion has a negative influence on the number of births.

After having discussed the variables that can be categorized as social variables the

economic variables are discussed. With regard to GDP per capita it can be said that

this indicator is important for this thesis for it can be used in order to test the

demographic transition theory which states that when countries’ incomes increase

then after a certain point the fertility and the mortality rate start to fall. Therefore,

GDP per capita can be seen as indicator for an improvement of the economy of the

countries in question. When GDP per capita of a country rises it means that the

population becomes wealthier. In return when the income of a population rises people

can consume more and take advantage of their improved situation. It might happen

that when GDP per capita increases people are able to consume some goods but they

have to make compromises when doing so because the income is not yet so high that

they could afford all or a high proportion of the goods they have wanted. In this

situation it might happen that, one has to decide on whether to buy a car, a house, go

on holidays or get a child. When GDP per capita increases it could be expected that

the number of births starts to decrease because the previously poor country is now able to consume goods that it was not able to consume before. Another argument might be that as a country becomes richer, education might play a more important role, increasing the direct costs of children. Therefore, it is expected that the number of births decreases with the increase in GDP per capita. This can be supported by the example of the number of births in developed and developing countries. While in several developing countries the number of births is pretty high it is rather low in advanced societies such as the countries in question in this master thesis. Although GDP per capita is non-monotonic over time the hypotheses relating to GDP per capita can be stated as follows:

Hypothesis 5: GDP per capita has a negative influence on the number of births.

Another independent variable used in order to explain the fertility decline is the wage rate per gender (1996-2003) which might help to test the economic theory on fertility.

In the literature review section it has been said that the higher the opportunity costs are the fewer children the women will have. This would mean that women with lower wages have more children than women with higher wages. In this respect the wage rate would have a negative impact on the number of births due to higher opportunity costs. Next to the higher opportunity costs also the direct costs, as explained in the literature review section are important in this regard. Women with higher wages might wish that their children attend better schools which would also raise the costs of having children and which might have a negative impact on the fertility rate.

However, although higher wages imply that the opportunity costs rise it might also have another effect on the number of births. Women with higher wage rates are better able to cope with their children. They have more money to spend on clothes, nutrition and schooling and this might also be a factor that increases the number of children.

Some mothers might be concerned with the lives their children would have instead of

the pursuit of being successful in the job. Here, the argument would work in the

different direction. Another aspect that is important when talking about wages and

number of births is the wage the husband is earning. We assume that when the

husband´ s wage is high, the fertility might be higher than in a family in which the

man earns less. The reason for this is that the husband earns enough for the children

and the wife so that the fact that she does not work but stays at home with the children

does not have bad consequences for the family. Therefore the following hypotheses are formulated:

Hypothesis 6a: Wages earned by women have a negative influence on the number of births.

Hypothesis 6b: Wages earned by men have a positive influence on the number of births.

The next independent variable described is education per gender. As mentioned in the literature review women that have a relatively high level of education tend to

postpone childbearing for several years and in some cases decide not to have children at all. Furthermore, as also mentioned in the literature review the number of children that a woman with a higher level of education has is lower than a women with a lower level of education. A reason for this finding might be that women with higher levels of education are more interested in their careers. They want to take advantage of their higher education level and they might be more interested in fulfilling their dreams in terms of money and living standard than women with lower level of education. More educated women might not consider children as self realization but have other

opinions on what is important in their lives. Another reason for fewer children born by more educated mothers is that they might have better information on contraceptives and on birth control in general. Women with lower levels of education might lack this knowledge and have other beliefs and values. It might be that they see their reason for existence in being mothers. Being a mother might, for women with lower education be one of the few things that they might accomplish in their lives. Besides a negative relationship between fertility and education level there is a assumed u-shaped

relationship, indicating that the birth rate declines with the education level until it

reaches the level of secondary education and then starts rising at above secondary

education level {Andorka (1978)}. It appears from researches that a u-shaped

relationship exists but on the other hand also evidence for a negative relationship

exists. Referring to the education level of men on fertility it might be assumed that

more educated men are also willing to pursue their careers. It might also be the case

that being a successful businessman or lawyer, for instance, would also imply to have

a family which they might show in the public. The private live of a more educated employee might be a very important factor for the employer and might improve the image of the employee. For many people it is important to show that their private lives are going well in order to impress people or to have a good reputation. It is important, with respect to the different educational levels to define high and low educational level. Due to the data that is used in this master thesis, low education is defined as the attendance of the primary and higher educational level is defined as all other levels of education that are higher than the primary, starting with the secondary level. Therefore, in the case of the master thesis the data is just used for the more education people with an education level at least secondary or higher. With regard to all those arguments the following hypotheses can be formulated:

Hypothesis 7a: Education with respect to women has a negative influence on the number of births.

Hypothesis 7b: Education with respect to men has a positive influence on the number of births.

Next the influence of employment will be discussed. According to the book

“Determinants of Fertility in Advanced Societies” written by Rudolf Andorka employment of women can have different effects depending on the particular circumstances. In cases when women have full-time employment they have fewer children and therefore there is a negative impact on the fertility rate caused by employment of women. Furthermore, the impact of women employment depends on the fact of whether employment and family is compatible. In other words, when a woman can combine family and career without serious income or position losses she will have as many children as an unemployed woman. However, when this is not given, then employment of women might have a negative effect on the number of births. Since no closer look is going to be taken at the child care and child support services of the countries in particular the assumption that employment of women has a negative effect on the number of births will be taken. On the other hand the

employment of the man plays also an important role when deciding on whether to

have children and how many to have. Since in the countries in question the man is still

seen as the primary bread-earner the higher the employment of men will have a positive effect on the fertility rate.

Hypothesis 8a: Employment of women has a negative influence on the number of births.

Hypothesis 8b: Employment of men has a positive influence on the number of births.

4.4. Model Development

In the previous section the hypotheses have been formulated which have to be tested by the model that will be developed in this section. Topics that are discussed in this section are the characteristics of the data, whether the fixed, the random effects model or the common effects model is used and finally the formulation of the regression equation.

The data used is panel data for countries are observed on different sections such as abortion, number of births and GDP per capita for a longer time period. Using panel data might have benefits but also disadvantages. One advantage of using panel data, according to Baltagi is that it can be controlled for individual heterogeneity. In the case of time series or cross-section data the heterogeneity is not taken into account and therefore biased results can occur. Further, when using panel data it is possible to have more informative data, variability, more degrees of freedom, more efficiency and on the other hand less collinearity among the variables used. For instance, when time series are used more multicollinearity might occur. Finally, panel data enables one to construct and test more complicated models than do time-series and cross- section data. On the other hand, as mentioned there are also disadvantages concerning panel data such as problems with data collection due to non response, problems of coverage and frequency of interviewing. According to Baltagi (1995) another

disadvantage is distortion of measurement errors. These are the errors that might arise due to incorrect responses, which can arise when the questions are not well stated.

Finally, selection problems might arise such as non response and attrition. At first

sight these disadvantages do not seem to be relevant for this paper and for the master

thesis itself due to the nature of the data (the fact that the data has not been selected by using questionnaires by me). However, the disadvantages do indeed matter for both this paper and the master thesis. The reason is that the data obtained from the United Nations has been collected by sending questionnaires to the different countries.

To be able to test the hypotheses multiple regression analysis will be used. With this regression analysis the relationship between the dependent variable namely number of births and the chosen independent variables can be tested. By using multiple

regression researchers might find answers to the questions “what is the best predictor of..?” However, what is important when using multiple regressions is that no causality can be implied, that establishing a model has to be guided by theory. Moreover, regression just describes linear relations. The form of the regression equation that will be used in this model is the following:

Y

= C + β

A

+ β

B

+ β

D

+ u

(1)

The term C and u represent the constant and the one way error component,

respectively. U

µ

ν

. µ

is the unobservable individual specific effect and ν

is the remainder disturbance. The first term is time invariant and it entails all the individual effects that are not included in the regression model. The second term on the other hand depends on time and individuals. The subscripts i and t stand for the country and the years respectively. The independent variables are represented by A, B, C and the parameters are represented by the letters α, ß and γ. The coefficients, α, ß and γ are unknown and represent the effect of a change in A, B, C and D on Y. There are several assumptions made for the multiple regression model. Those assumptions are the following:

MR.1 Y

= β

β

x

+ ….β

x

+ e

t= 1,…..T MR.2 E(y

) = β

β

x

+ ….β

x

E(e

) = 0

MR.3 var(y

) = var(e

)=σ²

4

Undergraduate Econometrics p. 150

MR.4 cov(y

,y

) = cov(e

,e

)=0

MR.5 The values of xtk are not random and are not exact linear functions of the other explanatory variables

Assumption MR.2 means that each of the random errors has a probability distribution with zero mean. Although some error might have a postive and ohter might have a negative sign this will be averaged to zero with the increasing number of

observations. This assumption means that one assums the model to be correct on average when specifiying the model. MR. 3 states that the variance of the random error denoted by σ² measures the uncertainty that might be presented in the econometric model. It is further assumed that this variance is the same for each observation. In the case that this is true, the model is homoskedatic while if that is not true the problem with heteroskedasticity is present. Heteroskedasticity will be

discussed in the next section, when it comes to testing the model. Referring to the assumption MR.4 one can say that the error pairs are assumed to be uncorrelated.

Before stating the equation that is going to be used in this master thesis, one further decision has to be made. The decision is whether to use the random, the fixed or the common effects model. There are some differences between the fixed effects also called least squares dummy variables and the random effects model. The former model is used when one focuses on particular N firms or like in this case countries while the random effects model is used when the countries are chosen randomly.

Further, the fixed effect model is used where one expects that each firm, country or other piece of observations has a fixed effect that shifts yet up or down. Since specific countries are used in this master thesis that are not chosen randomly and fixed effects shifting yet up or down the fixed effects model is going to be used. In the fixed effects model, according to Altai (1995) it is assumed that µ is fixed and that the disturbances independent and identically distributed. The fixed effects model might suffer from a loss of degrees of freedom brought about by too many dummies that might cause the problem of multicollinearity among the regressors. The random effect overcomes this problem by allowing v to be random. However, due to the fact that the loss of degrees of freedom occurs when N is large and due to the fact that in the case of this master

5

Undergraduate Econometrics p. 149

thesis N represents the countries, amounting to eleven to thirteen the fixed effects model will be used here.

The regression equation in this master thesis would then be:

Log(birth) = C + β

log(gdp)+ β

log(marriage) + β

log(divorce) + β

log(mothers_age) + β

log(marriage_age) + β

log(abo) + u

(2)

The terms GDP, mothers_ age and marriage_age are GDP per capita, mothers' age at maternity and the age at marriage, respectively. Marriage, divorce and abo stand for number of marriages, number of divorces and the number of abortion. This equation is going to be used for the period 1955 until 2003 for all countries. As can be seen it is a log-log function meaning that the logarithm is used for both the dependent and the independent variable This model can just be used when the values used are all positive which here is indeed the case.

The second equation used is described below. It is important to note that in this equation the economic and the social factors are combined but just for the period between 1996 and 2003 due to the lack of economic data.

Log(birth) = C + α

log(empw) + α

log(empm) + α

log(edum) + α

log(eduf) + α

log(wf ) + α

log(wm) + u

(3)

Empw and Empm represent the employment rate of women and men respectively.

Further, eduw and edum represent the level of education that women and the level of

education men have, respectively. The variables represented by wf and wm are the

wages earned by women and men respectively. According to the hypotheses the