Social Infrastructure and Economic Growth

Abstract

This dissertation has had a focus on the relationship and links between social infrastructure and economic growth in the context of Sub-Saharan countries. Predictability in development and growth is oftentimes linked to various endogenous conditions that can ultimately enhance or detract from the overall potential of a modern nation. The object of this research was represented by social and economic indicators in 23 Sub-Saharan countries. The aim was to find out whether the social infrastructure has a statistically significant impact on economic growth in this region. In order to reach this aim, the method of regression analysis has been implemented. The study has covered a wide range of social and economic variables observing them for a period from 1980 to 2008. The results of the study revealed that only population growth, life expectancy and savings rates are statistically significant determinants of economic growth in Sub-Saharan countries. This finding has supported the assumption that social infrastructure is an important factor that effect economic growth and development. However, the research has been limited by the lack of information on all 33 countries in Sub-Saharan region. Therefore, the sample was reduced to 23 countries. Furthermore, some social indicators such as Gini coefficient and mortality rates were not available for some of the older years in the sample. The study ends with recommendations to policy makers and discussion of implications.

Dedication and Acknowledgment

This dissertation is dedicated to my parents who I love unconditionally. I would like to express my gratitude to the University staff for the knowledge they shared with me and inspiring me to think critically.

Author’s Declaration

I declare that the research project has been independently prepared by myself and represents an original work with no plagiarism. All external ideas and quotations have been properly referenced. The full list of references is contained at the end of the dissertation.

Chapter 1: Introduction

1.1 Background

As economic theory has evolved over the past several decades, a variety of variables have begun to infiltrate the standard models of growth and development. Roseta-Palma et al. (2010), for example, recognise that human capital has become an increasingly important variable in growth modelling, suggesting that the force behind such capital can radically alter the shape and potential of industry and markets. However, there is an inherent expectation of support, one which is based on the conceptualisation of the social infrastructure that leads the vocal masses to expect national investment in their wellbeing. To perform within a developing nation, society must be supported. The support must include effective health care and improved educational standards. The perpetuation of economic performance within diverse marketplaces ultimately relies upon the sustainability of such practices, leveraging human capital and contributing to market development.

The theoretical background of this dissertation is represented by the elements of the economic theory that explains the growth and expansion as well as the role and influence of social factors. Econometric models put forward by the UN will be of particular importance. The research project will also review the arguments of famous contemporary economists such as Stiglitz (2009) and Jones and Klenow (2010). This will serve a useful theoretical background to the wider analysis, which is required for answering the research questions. The literature review will also cover the mainstream development theories such as dependency theory and social justice theory. From a conceptual perspective, researchers such as Newman and Tomson (1989) provide a precedence of focusing on social factors in economic development. They argue that social infrastructure is an essential element in sustainable long term growth of the economy. This theory may only be accepted as valid if it is statistically supported using the case studies of the economies. The testing has previously conducted by Jones and Klenow (2010). The researchers indeed supported the theory by finding that economic growth was boosted by increasing life expectancy in many countries. However, these researchers have not found such support of the theory for African countries in Sub-Saharan region. Therefore, this dissertation will attempt to study this region in more details and find statistical dependency between the social infrastructure and economic growth. Using the method of multiple regressions, a sample of 23 countries in Sub-Saharan region will be explored. The study will cover a period from 1980 to 2009.

1.2 Aims and Objectives

In African countries, economic growth is a function of a wide range of variables such as foreign aid, foreign direct investments (FDI), policy reformation and liberalisation and others. This investigation seeks to examine a link, which is frequently overlooked in this dynamic and evolving economic system: social infrastructure. There is an innate reciprocity between social infrastructure and economic growth, one which requires further definition within the context of African evolution in order to determine the true order of events. In order to limit the scope and breadth of this study, the following aims and objectives were established:

• To determine whether social infrastructure is a fundamental determinant of economic growth.
• To explore statistical significance of social infrastructure as a determinant of the economic growth in Sub-Saharan countries.
• To recommend strategic policy implications for the transition economies in Sub-Saharan region that would help them to grow and expand.

1.3 Research Questions

Based on the aforementioned aims and objectives, particular research questions were defined. They focus on investigation of the relations between the African social infrastructure and economic growth experienced by various nations within this region. It may be argued that the socio-economic aspects, which contributed to successes for many nations, remain inconsistent and non-definable today. Based on this supposition, the following research questions were defined prior to engaging in the investigative process:

• What are the primary threats/pitfalls associated with economic growth in Sub-Saharan Africa today?
• Does social infrastructure have a statistically significant impact on economic growth in Sub-Saharan countries?

1.4 Chapter Layout

In order to standardise the research project, it was important to create a clear structure and presentation format. The study is structured in a way that would allow for progressing from more general information regarding the Sub-Saharan countries to more specific information regarding the variables that have a direct impact on the growth of the transition economies. The rest of the research project has the following structure.

Chapter 2: Literature Review. This chapter focuses on a broad range of theoretical and empirical data that has been retrieved from a variety of academic sources. This literature review explores the determinants of the social infrastructure for developing nations, focusing on Sub-Saharan Africa.

Chapter 3: Methodology. This chapter highlights the research methodology chosen during the collection and analysis of empirical data. Based on the precedence established by the past researchers, econometric modelling is used as the main method of the research. This chapter also discussed the strategies and approached that were used with their justification.

Chapter 4: Data Presentation. This chapter reveals the main findings and results of the research. The historical statistical data is presented and analysed. Correlation and regression analysis is applied to the data. The main results are summarised in tables and figures.

Chapter 5: Discussion and Analysis. In this section, a synthesis of academic and empirical data is presented. The discussion is focused on the original research questions and objectives. They are compared to the findings achieved by previous researchers. Similarities and differences are analysed and explained.

Chapter 6: Conclusions and Recommendations. The final chapter provides the final insight into the relations between social infrastructure and economic growth in Sub-Saharan countries. Recommendations for future research are offered since the research project has encountered particular limitations that have to be addressed in the future. Furthermore, policy implications are recommended in this chapter.

Chapter 2: Literature Review

2.1. Measuring Economic Growth

While it is widely recognised that the measurement of economic growth provides an accurate picture of development and achievement in transition nations, the inherent value of such metrics has been questioned during the last decade because of several pitfalls. Hoogvvelt (2001:8) argues that in early development models, all emphasis was placed on strategic enhancement of the transition economies with impoverished nations. A traditional indicator of economic growth is represented by GDP. While conceptually indicative of growth and economic expansion, this indicator has been recently challenged as an effective measurement of sustainable national development. In fact, researchers such as Stiglitz (2002, 2007) and Collier (2007) have offered the arguments on the fact that GDP fails to represent an accurate picture of national economic welfare. It is argued to limit the identification of economic inequality and circumvent such influential social indicators as mortality rates, GNI per capita, education levels, etc. Other researchers such as Thakur (2006) have suggested that the United Nations Human Development Index (HDI) should be used as an alternative measure of economic growth besides the GDP.

2.2. Growth Models

The economic theory provides different growth models that explain the factors of economic growth and help to determine what cause an economy to expand. Among the well-established theories of growth are the neo-classical models suggested by Solow (1956: 65) and Ramsey (1928: 543). However, there are also alternative models that have recently been proposed. The most notable example is the endogenous growth model.

2.2.1. Neo-classical Growth Model of Solow and Ramsey

The exogenous growth model has been originally presented by Solow (1956: 65). It is an extension of the previously formulated Harrod-Domar growth model. The latter suggests that the rate of economic growth is a function of the productivity of the country’s capital and the savings rates. Solow (1956: 65) has improved the Harrod-Domar growth model by differentiating between the new capital that emerged from the use of new technology and old capital. Diminishing returns started playing an important role in the exogenous growth model.

Solow (1956:65) has also added labour to the determinants of the economic growth. The researcher argues that more than one factor of production should be included in the growth model. These factors are capital and labour. The researcher also emphasises the role of the technological progress in the economic growth. However, the model may be criticised for failing to provide the explanation of how and why the technology develops. In addition, the exogenous growth model may be criticised for neglecting the factor of entrepreneurship, which is argued to have a strong impact on economic growth (Braunerhjelm, 2008: 51; Audretsch et al, 2006: 119).

Mathematically, the exogenous growth model may be presented as follows:

Where Y is the output of the country; K is total capital (both new and old); L stands for labour; A represents technological development.

The exogenous growth model is heavily reliant on the indicators estimated per capita. Hence, it places a significant emphasis on the role of the population growth in the economic growth. The capital per worker is argued to be growing only if the savings rates exceed the rate of population’s growth and the level of depreciation of the capital. The exogenous growth model also suggests that the savings rate would be steady in the long run and have a positive correlation with the economic growth, i.e. the countries with higher savings rates will be expected to have higher economic growth. However, this notion was criticised by Ramsey (1928: 543) who proposed an alternative neo-classical model of growth.

In his model the savings rates are assumed to be varying and not constant. The Ramsey model has changed the way the capital is modelled. Mathematically, it is represented as follows:

Where k is capital; c is consumption; δ is the rate of depreciation of capital; f (k) is the value of total production. Since the savings rates are not viewed as constant, the level of consumption is also considered as a varying process since it is tightly connected to savings. Since neither Ramsey nor Solow model of growth included the factor of entrepreneurship and explained technological progress as an endogenous process, an alternative model has been developed. It is called endogenous growth model (Barro and i-Martin, 2004: 205).

2.2.2. Endogenous Growth Model

The previously discussed exogenous growth models suggested that a country’s GDP is a function of the savings rate and technological advances. Nonetheless, these exogenous growth models failed to show how savings are determined and how the technological changes are driven.

These limitations are effectively solved by the endogenous growth model. It suggests that savings rates are simply a function of the utility maximising actions of the economic agents. Given the financial constraints, companies would aim to maximise their net income while consumers will tend to maximise their utility (Romer, 1986: 89).

The endogenous growth model also explains technological progress as a result of the favourable policies from the government that do not restrict innovations and changes in the industries. It is valid to argue that in developing countries the governments may attempt to put certain restrictions on changes and innovations in order to protect the key sectors of the economy. The endogenous growth theory suggests that such actions would lead to a slowdown in the economic growth in the longer term. The theory also views company investments in the research and development as the way to technological progress and faster economic growth. Hence, the theory explains the growth of the economy with microeconomic elements (Aghion and Howitt, 1992: 323).

However, the model has also been criticised in the economic literature. For example, Parente (2001: 51) argues that the endogenous growth model, even though being more complex, still fails to explain why there is a divergence in the national income per capita in emerging economies and developed countries.

2.3. The Social Factors, Economic Development and Equality

A widespread academic research on social equality demonstrates that impoverished nations have traditionally failed to achieve healthy social infrastructure, which can sustain development amongst all groups of the population. Sebitosi and Pillay (2005:2045), for example, argue that poverty “is largely due to failure by society to productively deploy human resources” (Sebitosi and Pillay, 2005: 2045). The researchers argue that the governments of the countries with transition economies and policymakers cannot actively engage every individual in economic activities. In many cases, funding welfare programmes that were fiscally unsustainable has had minimal impact on the social welfare of the national inhabitants (Sebitosi and Pillay, 2005:2045). This is also illustrated by the efforts made by the African National Congress (ANC) in the late 1990’s and early 2000’s.

Ultimately, it is the strategic utilisation of national resources that will allow for perpetuated social stability, gradual reduction of poverty over and improvement of social infrastructure. Sebitosi and Pillay (2005:2048) argue that availability of resources and the specifics of the culture determine the social infrastructure in a country. This, in turn, plays a role in the economic growth and development.

Equality is a term used for describing the gap between the rich part of the population and the poor. This term is also expanded to describe the difference in rights between males and females, young and old, native and foreign ethnic groups, etc. Researchers such as Morvaridi (2008) and Houtzager (2005) argue that the merits of equality should be used as indicators of long term sustainability and economic growth of a nation. Anderson and Cavnagh (2009) have presented empirical evidence on the existence of income inequality and gender inequality that negatively impact the economic growth and development. Other academics (e.g. Sen, 2001) suggest that innate human rights must play a fundamental role in the development discourse, emphasising deficiencies within the national infrastructure that interrupt widespread equality.

Accessibility and availability of resources and the level of social equality in developing nations are frequently identified as primary indicators of social development. Researchers such as Moradi and Baten (2005) have modelled social inequality according to anthropometric data. The model is focused on the level of development of social groups over the past decade.

Their evidence highlights two different phenomena that have implications for policymakers in the future. First, the authors argue that evolution of the food supply has a direct and measurable impact on the physical characteristics of the population. Second, the marked increase in the social inequality has a direct impact on the resource accessibility and, subsequently, on the growth pattern of the surveyed nations (Moradi and Baten, 2005:1254). The implications of such evidence transcend the limitations of the model itself. The researchers recommend the governments to provide favourable external conditions for redistribution of wealth and resources in order to achieve higher rates of economic growth and development.

2.4. Resources, Social Determinants of Development and Opportunities

In economic analysis of national development, indicators of sustainable growth are oftentimes linked to the advancement of technology, resources, and industrial activity. From a social standpoint, it is the access to resources and provision of more advanced amenities that allow researchers to effectively measure progress. Buys (2009:1496), for example, explored a widespread diffusion of cellular phones throughout Sub-Saharan Africa, modelling competitive networks according to the population concentration and government policy measures. Their time-scale representation of progress in cell-phone usage throughout this continent suggests that strategic policy reform has provided the most significant opportunity for widespread distribution of such technologies (Buys, 2009:1497).

Improved competition amongst providers led to the spread of a sustainable low cost technology across the countries. This evidence suggests that opportunities play an important role in social and economic development. These opportunities, however, should be provided by the government and policy makers.

Social factors in the sustainability of economic growth can oftentimes be overlooked in academia. Researchers focus instead on more tangible variables, attempting to model economic growth using purely economic variables and neglecting social factors. Chou (2006:910) demonstrates how social capital, as a strategic resource, can have a measurable and long term impact on the growth of a nation and its economic development. Essentially, as policymakers provide the resources for social capital to develop and expand, the infrastructure will simultaneously expand, allowing individuals to use the skills they have developed in a more effective and productive way. Over the long term, Chou (2006) suggests that technology and favourable policies of the government will lead countries to sustainable economic growth and stronger social infrastructure.

Other models of social infrastructure have focused on the more practical composition of this expanding network. They emphasised such factors as the progress in transportation and population movement patterns. Porter (2002:286), for example, suggests that sustained improvements in both rural and urban transportation signal development progress in African nations. In particular, the author argues that economic recession of the 1980’s and 1990’s in African countries was reinforced by the poor condition of roads, transport and weak infrastructure. The deterioration of the transport infrastructure would reduce transport efficiency for the exchange of goods and services, resulting in a downward spiral in commercial activities (Porter, 2002:287).

Porter (2002:296) argues that one of the methods to provide sustainable economic development is to stimulate the ‘scaling up’ of the national economy through the installation and evolution of social institutions.

So, focusing on inequity in social development and the limitations imposed on infrastructure development and sustainability, the reviewed academics demonstrate how the consequence of restrictive social development is ultimately the deterioration of economic growth. The following empirical investigation will attempt to model such occurrences in modern Sub-Saharan Africa, highlighting those key variables that affect economic development.

Chapter 3: Research Methodology

3.1. Research Model

Researchers such as Moradi and Baten (2005:1234) argue that anthropometric models are fundamentally beneficial in the studies of national development, providing valuable insight into particular social factors that are indicative of long term development. In their analysis of Sub-Saharan African development, the authors used such models to analyse the data on accessibility of resources (i.e. nutritional and health inputs), providing a bounding metric by which they were able to evaluate inequality in the region (Moradi and Baten, 2005:1236).

In a research model that was focused on a similar issue regarding the social determinants of economic growth, Newman and Tomson (1989:464) used World Bank databases to identify particular social indicators and statistically connect them to economic development. The methods and models of this dissertation are based on the research methodology of Newman and Tomson (1989: 464) and Jones and Klenow (2010). The econometric models will be represented by several equations that start from simpler ones and progress to the more complicated, which include additional variables and dummies. The list of equations that will be used is provided below.

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + εij (1)

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + εij (2)

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + ωhexpij + εij (3)

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + ωhexpij + ρdummyrij + εij (4)

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + ωhexpij + ρdummytij + εij (5)

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + ωhexpij + ρdummyrij + τdummytij + εij (6)

Among these models, the best one will be selected with the Akaike information criterion. Random and fixed effects will be used in the panel regression models to investigate, for example, the impact of the geographical location on the economic growth and other factors.

A general form of the panel regression model with fixed effects will be as follows:

yij = α + β’Xij + uij,
where the error term u is assumed to be a sum of the fixed effect and another error term:
uij = μi + νij.
The random effect model will be different from this one in how it explains μi and νij . These terms are assumed to be completely independent. Furthermore, they random variable effect implies that these terms are normally distributed, i.e.

The choice of the methodology is consistent with the theoretical concepts of the growth models reviewed in the literature and supported by such economists as Solow (1956), Romer (1986) and Barro and i-Martin (2004). The theory of economic growth expressed by these economists mainly suggests that a country’s GDP is a function of both economic variables and social. In particular, it has been seen in the literature review that exogenous growth model connects GDP with the savings rates and technical progress. The theory of Solow (1956) and the growth theories in Barro and i-Martin (2004) also suggest that GDP is related to the population (social variable) because the latter determines the amount of capital and labour as factors of production. Hence, the core of the econometric model has been built on the exogenous growth theory proposed by Solow (1956) and explained in Barro and i-Martin (2004). However, it was found in literature review that this theory was also criticised. The models has been enhanced by inclusion of additional variables to make it more complicated and create a representation of social infrastructure, which is a key focus of the research.

3.2. Research Instruments, Approach and Sampling

Based on the research model presented by Moradi and Baten (2005) and Newman and Tomson (1989), this investigation is focused on the changes in economic growth as a result of a number of social and economic variables that have been described. Researchers Thomas (2003) and Creswell (2009) provide models of empirical research, emphasising a unique link between both quantitative (statistical, data-driven) and qualitative (phenomenological, experience-driven) data streams. Their mixed method research approach places one of these two methods in a primary position over the other, allowing the subsequent research to serve as a validation mechanism.

The data used in the dissertation is entirely based on economic development statistics within the Sub-Saharan African counties. However, the various phenomena, which contribute to such development, are of primary concern for the relevance and validity of this investigation. Therefore, a mixed method research approach was chosen for the study. Using this method, statistical findings will be achieved and later compared with various economic and social phenomena across the surveyed nations.

Because there are 33 different nations currently associated with Sub-Saharan Africa, this research has chosen a sample of the top 23 countries in terms of population, attempting to retrieve data that is directly relevant to the conceptualisation of the long term sustainable growth and the impact of the social infrastructure on this process. Non-probability sampling technique has been implemented in choosing the countries. This decision may be justified by the fact that total population represented by the 33 countries in Sub-Saharan region is quite small and could be used without picking a sample. However, sampling was needed since a limited amount of data was available for the countries. Historical statistics have been gathered from Penn World Table, International Monetary Fund (IMF, 2010) and World Bank (2010) database. These sources provided information for only 23 counties in the Sub-Saharan region.

3.3. Strategy of Research

While all of Africa could have provided very general information relative to the development of these nations as a conglomerate, it was important to evaluate the social infrastructure of these nations to narrow the scope of the research. The case study research strategy has been employed in order to investigate the social and economic situation in all the companies within the chosen sample. The case study strategy, which was popularised by Yin (2009), allows the researcher to extract particular data from complex problems and identify those variables, which are most significant. Furthermore, this strategy allows for effective exploration of both the statistics and context of the problem (Saunders et al, 2007, p.119).

Yin (2009) presents a model of the investigative case study, suggesting that the breadth and focus of research questions will ultimately define the methods employed during the study. His validation of the case study strategy as a valuable tool within academic research is based on the depth and scope of the data generated from such investigation (Yin, 2009:14). Following such case study guidance and the mixed method approach previously discussed, this research was conducted in an effort to determine whether or not the social infrastructure has a direct and measurable impact on overall economic performance of the countries in Sub-Saharan region.

The data sources were retrieved from two globally respected sources: The World Bank and the International Monetary Fund (IMF). Economic indicators were also gathered from Penn World Table. These databases have compiled specific economic and social data on the majority of the nations in the world, providing a resource for academics and policymakers. While the World Bank (2010) remained the primary source of the data, the IMF (2010) database was used for comparative purposes and in order to identify several variables not found within the World Bank annals. All analysis was conducted using Microsoft Excel and Eviews 6 statistical package.

3.4. Limitations

As previously mentioned, the scope of the research in this empirical case study was limited to the top 23 countries in Sub-Saharan region. This limitation arose from the lack of economic and social data for the rest ten countries in the region. World Bank (2010) provided most but not all information that was needed.

Another important limitation of the research, which is worth noting, is the lack of observations for several social indicators. It was noted previously that the sample of data covers 23 countries with the time range from 1980 to 2009. While many of the economic variables such as GDP were available for this period, some social indicators such as mortality rate were available only for a period of up to 5 years. Therefore, the overall sample will have to be shrunk to run the regression with these variables that have fewer observations. This is expected to have a negative impact on the accuracy of the study and estimated statistics.

Chapter 4: Data Presentation and Analysis

World Bank (2010) has provided economic and social data for twenty three countries in Sub-Saharan region. However, most of the data contained missing points. In order to avoid the problem of missing points, sixteen Sub-Saharan countries have been selected to be analysed for which more complete data was available.

The data ranges from 1980 to 2008. However, some of the social indicators such as health expenditure and mortality rate were available only for a limited time period. The health expenditure indicator was available only for a period from 2003 to 2007. Mortality rate indicator was available only for a period from 1998 to 2008. Due to the differences in the time period of data several panel regressions will be run and the best model will be chosen by means of the Hausman test.

Panel regression analysis has provided a number of advantages to the research project. First of all, it has allowed for gathering a large number of observations that totalled 4,250. If only time-series analysis was used, there would have been fewer observations. Similarly, in a cross sectional analysis the number of observations would solely depend on the number of countries included. Panel regression analysis has allowed for combining both time and cross sectional dimensions making the analysis more advanced. Secondly, another advantage of using the panel data analysis was higher degrees of freedom. This is a result of the more observations that the method has provided. Degrees of freedom are statistical values that may vary and change. Finally, panel data has helped to reduce the problem of correlation among the independent variables in the regression model. It is also called the problem of multicollinearity. As a result, the paned data analysis has increased the efficiency of the calculations and improved econometric findings.

The first panel regression model that has been created is based on the analysis of the relationship between the nominal GDP expressed in current US dollars as a dependent variable and a list of independent variables that consists of the population, population over 65 years old, and life expectancy and savings rates. These variables have the longest range and no missing points are present. The first panel regression model is as follows:

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + εij (1)

where i is the cross sectional dimension, i.e. country; j is the time dimension, i.e. year; pop is population; pop65 is the older population of 65 years and over; lif is life expectancy; sav is savings rates; ε is the error term.

The estimates of the model have proved the hypothesis of the exogenous growth model reviewed in the literature, which suggests that an increase in savings rates is associated with faster growth of the economy. Indeed, the estimate coefficient of the savings rates was positive indicating a positive relationship with GDP and the t-test has shown that the savings rates are statistically significant determinants of the GDP. Similarly, life expectancy and population were found to have a positive effect on the economic growth. All independent variables were found to be statistically significant at a 99% confidence (Table 1).

However, it is valid to argue that these results could be misleading because the model has not yet been tested for serial correlation and heteroskedasticity. Furthermore, adding new independent variables could also change the outcome. The first problem that has been dealt with is serial correlation in residuals. Breusch-Godfrey test has been run to detect serial correlation. The outcome of the test allowed for rejection of the null hypothesis that there is no serial correlation. Hence, the problem exists and corrections are needed. One of the ways to fix the problem is to add the lagged residuals to the list of independent variables and run a new regression.

The new model has demonstrated that the p-values of the t-test decreased, thus making the estimated coefficients more significant. Furthermore, the Durbin-Watson statistic has shown the absence of significant autocorrelation. The fit of the model has also been improved when the lagged residual were added. It is also valid to note that after the correction of the model, the role of population in the economic growth has increased while the contribution of life expectancy and savings rates seemed to have slightly diminished. Nonetheless, all independent variables remained statistically significant (Table 1).

In order to run the robustness test on the regression, Newey-West procedure has been used. It estimates the regression coefficients with correction for the heteroskedasticity problem. The estimator of the Newey-West procedure is calculated using the following equation:

(2)
Where T is the total observations; k is the number of independent variables.

The results of the new regression output are presented in the following table and compared to the estimates of the previous models.

1 gdpij = α + βpopij + γpop65ij + λlifij + δsavij + εij (1980 – 2008; 464 observations)

	Model 1	T-test	Model 1 + resid(1)	T-test	Model 1 + Newey-West	T-test
pop	906	15.54	977	40.45	977	19.76
pop65	7,330,000,000	3.29	9,020,000,000	9.80	9,020,000,000	6.41
lif	960,000,000	3.10	1,160,000,000	9.08	1,160,000,000	9.43
sav	2,500,000,000	5.38	2,010,000,000	10.50	2,010,000,000	8.50
resid(-1)	n/a	n/a	1	47.32	1	10.59
R-squared	0.362795		0.891714		0.891714
DW	0.184517		1.75736		1.75736
AIC	51.1937		49.42481		49.42481

The table shows that inclusion of the Newey-West estimates has not changed the value of the coefficients. Only standard errors were changed. The significance of all independent variables remained high.

This regression model has supported the theoretical exogenous growth model in the context of Sub-Saharan economies. However, the full impact of social infrastructure on economic growth may only be investigated when other social variables are included in the model. Among the social variables that are available for the African countries are the mortality rate and healthcare expenditure. It has been said that these variables were available only for a limited time range. Hence, the whole sample will be shrunk to avoid the missing points. The next model that will be estimated will include only mortality rate for the period from 1998 to 2008. It is expressed as follows:

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + εij (3)

where mortf is the mortality rate of females and mortm is the mortality rate of male population.

2 gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + εij (1998-2008; 176 obs.)

	Model 2	T-test	Model 2 + resdi(-1)	T-test	Model 2 + Newey-West	T-test
pop	1,188	11.45	1,333	24.81	1,333	14.56
pop65	11,500,000,000	1.93	18,000,000,000	6.02	18,000,000,000	5.52
lif	2,390,000,000	2.55	293,000,000	0.54	293,000,000	0.45
sav	7,380,000,000	5.89	7,390,000,000	11.72	7,390,000,000	7.42
mortf	-400,000,000	-3.25	-240,000,000	-3.89	-240,000,000	-3.50
mortm	371,000,000	3.03	193,000,000	3.12	193,000,000	2.60
resid(-1)	n/a	n/a	1	21.98	1	8.12
R-squared	0.476984		0.870946		0.870946
DW	0.302307		1.71449		1.71449
AIC	51.62488		50.24289		50.24289

Originally, the model has demonstrated that both male and female mortality rate helped to explain the changes in economic growth, i.e. they had a statistically significant impact on the countries’ GDP. The previous variables such as population, life expectancy and savings rates also remained statistically significant. However, the model was found to have the problem of serial correlation and therefore lagged residuals have been added to the regresses and a new regression was run.

The outcome has improved. Statistical significance of the population over 65 years old has improved. The transformation of the model has also improved statistical significance of the mortality rates as determinants of economic growth. However, life expectancy was not estimated as a significant determinant of the GDP after this transformation of the mode. Finally, Newey-West procedure was once again implemented to solve the problem of heteroskedasticity. The coefficients are seen to remain unchanged. It may be concluded that adding mortality rates has increased the degrees of freedom and eliminated life expectancy from the list of statistically significant determinants of GDP.

Since health care expenditure information was not available for the same period as the mortality rates and economic variables, a new regression had to be run in order to take the health issue into account. The new regression has had fewer observations and, as a result, fewer degrees of freedom. It is valid to argue that this puts a limitation on the model and weakens it. However, without including the health care expenditure to the list of independent variables, it would be impossible to fully assess the role of social infrastructure in the economic growth of the Sub-Saharan countries. The new regression model has had the following form:

gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + ωhexpij + εij (4)

Inclusion of the health expenditure in the list of independent variables has negatively affected the significance of other variables as it can be seen from Table 3.

Table 3 gdpij = α + βpopij + γpop65ij + λlifij + δsavij + σmortfij + νmortmij + ωhexpij + εij (2003-2007; 79 obs.)

	Model 3	T-test	Model 3 + resid(-1)	T-test	Model 3 + Newey-West	T-test
pop	1,573	12.40	1,717	18.28	1,717	12.63
pop65	-4,350,000,000	-0.45	8,610,000,000	1.22	8,610,000,000	1.15
lif	6,340,000,000	3.89	6,010,000,000	5.17	6,010,000,000	3.98
sav	8,890,000,000	5.43	8,270,000,000	7.06	8,270,000,000	5.41
mortf	-403,000,000	-2.15	-310,000,000	-2.32	-310,000,000	-2.54
mortm	678,000,000	3.18	550,000,000	3.60	550,000,000	3.77
hexp	10,300,000,000	5.11	10,500,000,000	7.27	10,500,000,000	5.22
resid(-1)	n/a	n/a	0.712738	8.22	0.712738	6.75
R-squared	0.767906		0.885114		0.885114
DW	0.657115		1.856486		1.856486
AIC	51.25138		50.58384		50.58384

The role of population over 65 has diminished in explaining the economic growth of the Sub-Saharan countries. Male mortality rates were found to be more significant determinants of economic growth than female mortality rates. The health care expenditure was also found to be a statistically significant variable to explain the changes in GDP. Hence, this new model has made all variables statistically significant except the population over 65. However, there was found a problem of serial correlation, which was eventually solved by running a new regression with the added lagged residuals.

As a result, the standard errors of all variables decreased and statistical significance was improved (Table 3). Running the Newey-West procedure has not had an effect on the estimated coefficients.

These models have been tested using fixed effects and random effects. The output of the random effect testing is shown in Appendix B while the fixed effect testing is demonstrated further in the chapter. In order to run a fixed effect regression model, two dummy variables were introduced. They are unchangeable and therefore represent a fixed effect. The regional dummy variable was represented by geographical location of the Sub-Saharan countries. The region was divided into southern countries and central countries. The southern countries were assigned the value 1 and the central countries were assigned the value 0. South Africa dominates in terms of economic growth in one region and Nigeria dominates in another region. The aim of using the regional dummy variable is to explore whether geographical location has an impact on the economic growth in Sub-Saharan Africa.

Another dummy variable that was introduced is the time variable to represent the periods of economic recession in the countries. The period of positive growth has been expressed with the value 1 and the period of negative annual economic growth was expressed with the value 0. The first fixed effect regression model included only regional dummies. The second fixed effect model included only time dummies. The third model included both types of dummies. The results of the regressions are compared and contrasted in the following table.

Table 4 Model 4 with added dummy variables (2003-2007; 79 obs.)

	Model 4 + Dum.R	T-test	Model 4 + Dum.T	T-test	Model 4 + DumR + DumT	T-test
pop	1,793	11.63	1,711	11.67	1,794	11.53
pop65	18,100,000,000	2.09	7,720,000,000	0.96	18,000,000,000	2.07
lif	3,850,000,000	2.17	5,980,000,000	3.64	3,880,000,000	2.13
sav	9,200,000,000	5.23	8,200,000,000	4.92	9,180,000,000	5.17
mortf	-162,000,000	-1.21	-318,000,000	-2.44	-163,000,000	-1.22
mortm	248,000,000	1.41	561,000,000	3.48	251,000,000	1.42
hexp	11,300,000,000	5.33	10,300,000,000	4.70	11,300,000,000	5.29
dummyr	22,500,000,000	3.02			-3,180,000,000	-0.54
dummyt			-5,890,000,000	-1.26	22,300,000,000	2.98
resid(-1)	0.704172	6.94	0.690678	6.02	0.703179	6.86
R-squared	0.887125		0.876875		0.88722
DW	1.870822		1.698689		1.863594
AIC	50.5915		50.67842		50.61598

Adding the regional dummy variable has made mortality rates insignificant in the model. Population, life expectancy, savings and health expenditure remained significant. It is valid to note that regional dummy variable also shows statistical significance. This may be interpreted as a sign that economic growth and development are linked to the geographical territory.

The time dummy variable was not found to be statistically significant. All other variables except population rate over 65 remained statistically significant. When both dummies were finally included in the model, the findings appeared similar to the first model in which mortality rates lost their significance as determinants of economic growth. A more detailed transformation of the models with additional regressions that were used in order to achieve these final three outputs is presented in the Appendix A. They show transformation of the model and correlation for serial correlation and heteroskedasticity. Overall, twenty one regressions have been run.

The random effect tests have been conducted as well. Full output is presented in Appendix B and C and the summary is provided below.

Table 5 Random Effect Models (2003-2007; 79 obs.)

Variable	Model 3 + resid(-1)	T-test	Model 4 + Dum.R	T-test	Model 4 + Dum.T	T-test	Model 4 + Dum.R + Dum.T	T-test
POP	1,827	16.70	1,849	23.98	1,833	15.55	1,882	18.25
POP65	10,800,000,000	1.61	30,100,000,000	5.62	10,800,000,000	1.49	16,600,000,000	2.56
LIFEEXP	4,220,000,000	3.67	1,520,000,000	1.82	4,250,000,000	3.45	2,920,000,000	2.57
SAVINGS	9,350,000,000	6.96	9,950,000,000	10.59	9,350,000,000	6.49	10,000,000,000	7.93
MORTALFEM	-191,000,000	-1.71	37,578,645	0.40	-191,000,000	-1.59	-132,000,000	-1.25
MORTALMALE	366,000,000	3.10	-85,000,000	-0.73	368,000,000	2.90	176,000,000	1.42
HEALTH	9,990,000,000	6.20	12,700,000,000	11.11	10,000,000,000	5.80	10,900,000,000	7.17
DUMMYT	n/a	n/a	n/a	n/a	-7,070,000,000	-0.43	-3,510,000,000	-0.25
DUMMYR	n/a	n/a	34,400,000,000	5.42	n/a	n/a	27,300,000,000	3.35
RESID01(-1)	1	6.48	1	10.26	1	6.03	1	6.94

The Hausman test was run to check whether the estimated regression models had valid random effects.

Table 6 Hausman Test

Variable	Model 2	Chi-Sq. Statistic	Chi-Sq. d.f.	Prob.	Model 1	Chi-Sq. Statistic	Chi-Sq. d.f.	Prob.
C	-440,000,000,000	242,000,000,000	-1.820	0.070	-219,000,000,000	2,740,000,000	-79.859	0.000
POP	2,220,000,000,000	760,000,000,000	2.925	0.004	596,000,000,000	24,200,000,000	24.650	0.000
POP65	98,200,000,000	70,500,000,000	1.393	0.165	65,200,000,000	777,000,000	83.913	0.000
LIFEEXP	903,000,000	306,000,000	2.950	0.004	520,000,000	15,107,134	34.439	0.000
SAVINGS	587,000,000	186,000,000	3.159	0.002	-291,000,000	11,004,547	-26.419	0.000
MORTALFEM	35,255,181	133,000,000	0.265	0.791
MORTALMALE	69,447,329	167,000,000	0.417	0.677

The null hypothesis of the test is that misspecification is not present in the model. Misspecification could occur if the independent variables were correlated with the random effects. The results of the test demonstrate that in both the model that uses only population, life expectancy and savings as independent variables and the regression that includes mortality rates, the null hypothesis was supported. No evidence was found against it (Appendix C). The p-value of the test may be interpreted as a strong sign of the absence of misspecification in all models. However, Hausman test also had limitations in this case. It was not possible to apply the test to the regression models that had health expenditure among the independent variables. The reason is that there were few observations. Since the Hausman test showed that all both the short and longer version of the model are valid, the final choice has to be made with the Akaike information criterion.

The models that have been achieved were tested with the Akaike information criterion in order to detect the best one. The comparisons of the tests are presented in the following tables.

7 AIC Test of Models

Model	AIC
Model 1	49.42481
Model 1 + resid(-1)	49.40855
Model 1 + Newey-West	49.41114
Model 2	49.39721
Model 2 + resid(-1)	50.2296
Model 2 + Newey-West	50.19959
Model 3	50.2374
Model 3 + resid(-1)	50.2071
Model 3 + Newey-West	50.65575
Model 4 + Dum.R	50.5915
Model 4 + Dum.T	50.67842
Model 4 + Dum.R + Dum.T	50.61598

The analysis demonstrates that Model 2 can be chosen as the explanation of economic growth in Sub-Saharan Africa according to the test statistics, which is the lowest one. This model justifies the use of both regional and time dummies, indicating that they improve the total fit. All social and economic variables in the model have been found statistically significant.

Chapter 5: Discussion

After revealing the results and findings of the research it is valid to make a synthesis of both academic and empirical evidence, discussing the growth and development of Sub-Saharan Africa from a socio-economic context. The acceptance of slowing GDP in accordance with the IMF model would ultimately result in a tragedy for the African people, as high levels of poverty and extensive inequity in systems and opportunities limit any potential for expansion. Therefore, this chapter addresses growth and prosperity from an opportunistic perspective. It focuses on those economic indicators identified during the data presentation and incorporates modern economic theory into the research.

5.1. Social Infrastructure and Sustainability

The overwhelming challenge for policymakers is adapting to conditions of transition and economic growth in an appropriate and adequate manner. Newman and Thomson (1989:461) recognise that as nations begin to develop, there is an inherent difficulty for policymakers in satisfying the needs of the poorest members of the society while simultaneously sustaining and expanding domestic economic growth. While industrialisation and the rapid technological acceleration of some social conduits may provide an opportunity for individuals to accelerate their personal development and access newfound wealth, the disparity between the rich and the poor in developing economies will remain significant. This under-development of the social infrastructure and the lack of commitment to social evolution will most likely result in long term economic stagnation in the region. It is valid to argue that stagnation will not be simply expressed in terms of GDP but also in other social factors that have been found to be a problem in Sub-Saharan region.

From an educational perspective, it is evident that Africa is currently deficient in this field. As education becomes an increasing priority, Fosu (2007) argues that domestic budgets will gradually begin to reflect greater investment in such programmes. One particular challenge in this historical model of educational provision is that higher debt levels ultimately reduce investment in education, detracting from progress that might have otherwise sustained such practices (Fosu, 2007:711). By eliminating the debt servicing constraint, Fosu (2007:711) proposes that Africa could circumvent many of the pitfalls associated with monetary allocation and education needs, sustaining such programmes in a more efficient and effective way.

5.2. Economic Development and Growth Prediction

One of the particular challenges associated with revitalising the national economic environment in the nations such as Sudan or Kenya is that business operations continue to remain opportunistic in nature. Smith (1978:61) argues that as self-interests are propagated during economic growth, the resultant dissatisfaction with the economic system and marketplace leads the public to withhold support. Even as the household income increases due to greater economic opportunities, the equality of the distribution of products and sustaining the economic process is an unrealistic expectation (Smith, 1978:62).

In a more holistic view of economic growth and market advancement, England (2000:430) argues that there are inherent limitations in any marketplace because of a relative scarcity of natural capital and underdevelopment of technologies. The consequence of such system deficiencies can result in a slow growth rate or ultimately, in the stagnation of the local economy.

The findings from this dissertation have supported the theoretical arguments provided in literature review. Furthermore, the results of the research have been consistent with a similar but broader study conducted by Jones and Klenow (2010). This dissertation and the study of Jones and Klenow (2010) both showed that Sub-Saharan region is deficient in terms of life expectancy and income equality. Furthermore, the research project has been consistent with the works of Jones and Klenow (2010) in finding that life expectancy and population play an important role in economic growth of the Sub-Saharan countries.

The growth theory that has been tested in the empirical part of the dissertation is closely related to the exogenous growth model provided by Solow (1956). This theoretical model views the economic growth of a country is a function of the two main factors of production: capital and labour with added factor of technological progress and development. Although, the model does not explain with microeconomic variables how technological progress occurs, it provides an initial basis for creating econometric models of economic growth. The basis was seen in using the population and savings rates as the main independent variables. The exogenous growth model suggested that a society with higher savings rates would eventually demonstrate higher economic growth. The theorists such as Barro and i-Martin (2004) have shown previous example of how the countries with higher savings rates outperformed the countries with lower savings rates in terms of economic growth. This dissertation has supported these theoretical arguments in the context of the Sub-Saharan countries, which also showed this dependency between the economic growth and savings rates.

However, it is valid to argue that this research project has also deviated from the neo-classical exogenous growth model. It has used additional variables such as mortality rates, life expectancy and health expenditure. Although these variables were not implies by the theoretical model, they help to explain changes in the population and labour, which are important elements in the growth model. Mortality rates determine the length of human life. The latter is also determined by the expenditures on health care. Hence, population growth may be considered endogenous to the mortality rates and health care expenditure. Since the population growth partly explains GDP growth, these social variables of mortality rate and health care spending will also determine GDP growth. Nonetheless, the model that included these two social variables was deemed less significant and stable than the model with fewer variables. This phenomenon may be explained by the loss of degrees of freedom when adding new social variables because the latter had fewer observations due to the limitations with data. The loss of degrees of freedom could explain why more complex models were found to be less stable.

It is also valid to note that in spite of the fact that all variables in the model were found to be statistically significant, the model has not covered all countries in the Sub-Saharan region. Nearly half of the countries provided statistical data with many missing points, which had to be excluded from the research. Exclusion of these countries and variables such as GINI coefficient for the lack of sufficient number of observations has also determined the outcome of the regression analysis. The use of regional and time dummy variables has also been found to be beneficial in this research since these two additional variables have made the model more stable and better explained the economic growth in the African countries.

Chapter 6: Conclusions and Recommendations

6.1. Conclusions

Similar to the economic analysis presented by Newman and Thomson (1989), this investigation has focused on the determination of whether or not the social infrastructure has a direct and measurable impact on sustainable economic growth. On the basis of the findings in this empirical case study, it can be concluded that this is a mutually symbiotic relationship, resulting in increased need for economic growth to invest in the social infrastructure and an increased need for the social infrastructure to expand in order to stimulate economic growth. This difficult developmental paradox essentially requires a measured, strategically-oriented focus on growth and development, allowing nations to progress steadily while simultaneously investing in the social well-being of their citizens.

The interconnection between the social infrastructure and economic growth may be seen in how the health care expenditure programs work. The governments of the Sub-Saharan countries may spend more on health care only if the national income allows for it. Hence, if GDP demonstrates growth, this implies that the economy has higher output, more funds allocated to the social needs and improved infrastructure. At the same time, the portion of the national income invested in the health care helps to reduce mortality rates, increase life expectancy and improve the labour in the country. Labour, in turn, is deemed to be one of the main factors of production and determinants of economic growth as suggested by neo-classical growth theory. So, an increase in the proportion of the GDP invested in health care may be viewed as an investment of the government in future economic growth.

The findings of the dissertation have shown that the most significant social variables that have a strong impact on the economic growth in the region are population growth and life expectancy. This is theoretically explained by the fact that higher number of the population would increase total consumption in the country. Furthermore, larger population would increase the total labour force and the volume of production. Life expectancy is an indicator of how long the present population may be expected to live and work. Life expectancy is also linked to the mortality rates. High mortality rate were found to have a statistically significant impact on the GDP only in several models. The reduction of male population due to their death in relatively young age leads to the decrease of the labour force and weaker production. Although, some of the models that were run showed that mortality rates may be statistically significant factors affecting economic growth in Sub-Saharan Africa, the final choice of the model has led to the exclusion of this factor from the list of the regressors since the model appeared to be more stable without this social indicator.

The first objective of the study was to determine whether social infrastructure is a fundamental determinant of economic growth. This objective has been successfully reached by finding that at least two of the chosen social variables have a statistically significant impact on economic growth of the countries. Therefore, social infrastructure is found to determine the economic growth. However, many of the social variables such as health expenditure cannot become strong without economic growth of the countries. This has been emphasised in the discussion chapter. An improvement in health of the population requires more investments to be made in the health care sector. If the economy of a country grows, there will be more opportunities to expand the funding of the health care sector. However, if the economic growth is negative, the country often has other priorities and the spending on health care may be reduced, which would have a negative impact on the future economic growth.

The second objective was to explore statistical significance of social infrastructure as a determinant of the economic growth in Sub-Saharan countries. Among the social infrastructure variables only population and life expectancy were found to have a strong level of significance. Mortality rate was found to be a statistically significant determinant of economic growth but its significance was lower and present only in other models that were analysed in the empirical part of the dissertation. Other social variables had to be excluded from the model because they made the final model less stable from the econometric and statistical point of view. Besides the social variables, the research project has shown that economic variables such as savings rates play a significant role in GDP growth. This finding supported the postulates of the exogenous growth theory that linked savings rates to the economic growth suggesting that countries with higher savings rates would eventually demonstrate higher economic growth rates.

The final objective of the dissertation was to recommend strategic policy implications for the transition economies in Sub-Saharan region that would help them to grow and expand. These are provided and summarised in the last part of the chapter.

6.2. Recommendations

Based on the findings from this research project, it is valid to recommend that education and healthcare are the two important variables, which policymakers should embrace when seeking to revitalise the social infrastructure in the countries. The primary of these two variables, education, provides the necessary knowledge to advance industries and make the country competitive on a global level. The second variable, healthcare, ensures that all individuals remain healthy and that the workforce is protected against adverse conditions. The finding from half of the models that mortality rate has a strong negative effect on GDP suggests that policy makers should make the improvement of health care and reduction of mortality rate their priority.

This dissertation has identified several key findings that provide impetus for future research. First, the value of healthcare and education cannot be effectively quantified using such a broad scale research method as that introduced in this analysis. Therefore, it is recommended that future studies narrow the focus to a few social indicators of economic growth and study them in depth. The following recommendations represent several additional focal points that could be pursued in the future research:

• To evaluate why Sudan has experienced such aberrant growth patterns in relation to other African nations.
• To identify those variables that continue to detract from greater development and highlight opportunities for improvement in the Sudanese infrastructure.
• To explore the South African path to development, focusing on key social drivers that enabled the improvement of economic performance in the country and its transition to the list of developed economies.
• To identify the effect of foreign direct investments on the economic growth and development in the Sub-Saharan region and evaluate whether or not foreign capital can sustain the economic growth in a single African nation.

In conclusion, the dissertation has successfully reached its aim and objectives by conducting econometric analysis of the Sub-Saharan region. The future studies are needed to narrow the focus of the research on individual social variables and to provide more recommendations to the policy makers of this African region. Future investigations should also solve some of the limitations of this research, in particular those connected to the missing points of data. Some researchers may have access to more complete data bases that provide a larger set of economic data on the countries in Sub-Saharan region. The use of more observations will provide more degrees of freedom and the more complex models that were eventually rejected in the dissertation could appear to be efficient and stable. Furthermore, future researchers may be recommended to enhance the model with additional social variables that could be tested for significance. In particular, a suggestion may be made to include the social inequality indexes, wages of workers, literacy rates and others. The exogenous growth model has also emphasised the role of technology in the economic growth. Inclusion of this factor in the econometric model may be difficult because a proper approximation has to be found. One of the ways to account for the technology is to use total expenditure on research and development. These additional variables will further improve the growth model tested in this dissertation.

References

Aghion, P., and Howitt, P. (1992) ‘A model of growth through creative destruction’, Econometrica, Vol. 60, pp. 323–351

Audretsch, D., Keilbach, M. and Lehmann, E. (2006) Entrepreneurship and economic growth, Oxford: Oxford University Press, pp.105 – 121

Barro, R. and i-Martin, X. (2004) Economic Growth, 2nd ed., MIT Press

Braunerhjelm, P. (2008) ‘Entrepreneurship, Knowledge and Economic Growth’, Foundations and Trends in Entrepreneurship, Vol. 4, No. 5, pp. 1-81

Buys, P., Dasgupta, S., Thomas, T.S. and Wheeler, D. (2009). ‘Determinants of a Digital Divide in Sub-Saharan Africa: A Spatial Econometric Analysis of Cell Phone Coverage.’ World Development, Vol. 37, No. 9, pp. 1494-1505.

Chou, Y. (2006). ‘Three Simple Models of Social Capital and Economic Growth.’ The Journal of Socio-Economics, Vol. 35, pp. 889-912.

CIA (2010). World Factbook. Central Intelligence Agency, [online] Available at: https://www.cia.gov/library/publications/the-world-factbook/ [Accessed 20 August 2010]

Cochrane, F. (2008). Ending Wars. London: Polity, pp.88-167.

Collier, P. (2008). The Bottom Billion. Oxford: Oxford University Press, pp.19-77.

England, R.W. (2000) ‘Natural Capital and the Theory of Economic Growth.’ Ecological Economics, Vol. 34, pp. 425-431.

Fosu, A.K. (2007) ‘Fiscal Allocation for Education in Sub-Saharan Africa: Implications of the External Debt Service Constraint.’ World Development, Vol. 35, No. 4, pp. 702-713.

Hertz, N. (2005). IOU: The Debt Threat and Why We Must Defuse It. London: HarperPerennial, pp.7-21.

Hoogvelt, A. (2001). Globalisation and the Postcolonial World: The New Political Economy of Development, Baltimore: John Hopkins University Press, pp.52-69

Houtzager, P. (2005). Changing Paths: International Development and the New Politics of Inclusion, Michigan: The University of Michigan Press, pp.13-41

IMF (2010) ‘World Economic Outlook Database.’ International Monetary Fund, [online] Available at: http://www.imf.org/external/pubs/ft/weo/2010/01/weodata/index.aspx [Accessed 25 August 2010]

Jones, C. and Klenow, P. (2010) ‘Beyond GDP? Welfare across countries and time’. Stanford: Stanford University Working Paper

Krueger,D. and Perri, F. (2006). ‘Does Income Inequality Lead to Consumption Inequality? Evidence and Theory’, Review of Economic Studies, Vol. 73, No. 1, pp. 163-193.

Kularatne, C. (2006). Social and Economic Infrastructure Impacts on Economic Growth in South Africa. Cape Town: University of Cape Town.

Moradi, A., Baten, J. (2005) ‘Inequality in Sub-Saharan Africa: New Data and New Insights from Anthropometric Estimates.’ World Development, Vol. 33, No. 8, pp. 1233-1265.

Morvaridi, B. (2008). Social Justice and Development, London: Palgrave Macmillan, pp.96-147.

Newman, B.A., Thomson, R.J. (1989). ‘Economic Growth and Social Development: A Longitudinal Analysis of Causal Priority.’ World Development, Vol. 17, No. 4, pp. 461-471.

Parente, S. (2001) ‘The Failure of Endogenous Growth’, Humanity, Social Sciences and Law, Vol. 13, No. 4, pp.49-58

Porter, G. (2002). ‘Living in a Walking World: Rural Mobility and Social Equity Issues in Sub-Saharan Africa.’ World Development, Vol. 30, No. 2, pp. 285-300.

Ramsey, F. (1928) ‘A mathematical theory of saving’, Economic Journal, vol. 38, no. 152, pp. 543–559

Romer, P.M.,(1986) ‘Increasing Returns and Long Run Growth’, Journal of Political Economy, Vol.13, pp. 89-101

Roseta-Palma, C., Ferreira-Lopes, A., Sequeira, T.N. (2010). ‘Externalities in an Endogenous Growth Model With Social and Natural Capital.’ Ecological Economics, Vol. 69, pp. 603-612.

Saunders, M., Lewis, P. and Thornhill, A. (2007). Research Methods for Business Students, 4th ed., London: Prentice Hall, pp. 115 – 210

Sebitosi, A.B., Pillay, P. (2005). ‘Energy Services in Sub-Saharan Africa: How Conducive is the Environment?’ Energy Policy, Vol. 33, pp. 2044-2051.

Smith, K.V. (1978). ‘Economic Growth—Meritorious or Meretricious: A Review of Social Limits to Growth.’ Socio Economic Planning Sciences, Vol. 12, pp. 61-65.

Solow, R. (1956) ‘A Contribution to the Theory of Economic Growth’, Quarterly Journal of Economics, Vol. 70, No. 1, pp. 65–94

Stiglitz, J. (2002). Globalisation and its Discontents, London: Penguin, pp.36-110.

Stiglitz, J. (2007). Making Globalisation Work, London: Penguin, pp.21-85.

Stiglitz, J., Amartya, S. and Fitoussi, J. (2009). Report by the Commission on the Measurement of Economic Performance and Social Progress, Discussion Paper.

Taylor, I. (2005). ‘Globalisation studies and the developing world: making international political economy truly global’. Third World Quarterly, Vol. 26, No. 7, 1025-1043.

Thakur, R. (2006). The United Nations, Peace and Security: From Collective Security to the Responsibility to Protect. Oxford: Oxford Publishing House, pp.9-21.

Wendt A. (1999). Social Theory of International Relations, Cambridge: Cambridge University Press, pp.44-72

WHO (1978) ‘Declaration of Alma Ata.’ World Health Organization [online] Available at: http://www.who.int/publications/almaata_declaration_en.pdf [Accessed 20 August 2010]

World Bank (2010) World Development Indicators [online] Available at: http://databank.worldbank.org/ddp/home.do [Accessed 26 August 2010]

Appendices

Appendix Ai: Regressions 13 – 15

	pop	pop65	lif	sav	dummyr	dummyt	resid(-1)
Reg 13 coeff.	811.6437	4.08E+09	-8.20E+08	2.63E+09	2.02E+10	-1.47E+09	n/a
St. Error	53.88518	2.13E+09	1.75E+08	4.60E+08	3.07E+09	3.08E+09	n/a
T-test	15.06247	1.913854	-4.67297	5.721071	6.586072	-0.47869	n/a
p-value	0	0.0563	0	0	0	0.6324	n/a
Reg 14 coeff	860.2809	5.41E+09	-9.06E+08	2.24E+09	2.01E+10	2.92E+09	0.910837
St. Error	23.08754	9.09E+08	74818623	1.96E+08	1.31E+09	1.32E+09	0.020048
T-test	37.2617	5.947583	-12.1084	11.40062	15.32283	2.216414	45.43333
p-value	0	0	0	0	0	0.0272	0
Reg 15 coeff.	860.2809	5.41E+09	-9.06E+08	2.24E+09	2.01E+10	2.92E+09	0.910837
St. Error	46.85048	1.25E+09	1.14E+08	2.28E+08	1.54E+09	1.52E+09	0.087526
T-test	18.36226	4.3156	-7.94043	9.824733	12.99804	1.925279	10.40643
p-value	0	0	0	0	0	0.0548	0

Appendix Aii: Regressions 16 – 18

	pop	pop65	lif	sav	mortf	mortm	dummyr	dummyt	resid(-1)
Reg 16 coeff.	1283.335	1.28E+10	-2.02E+09	7.32E+09	-2.73E+08	1.89E+08	2.75E+10	-1.30E+09	n/a
St. Error	104.1126	5.79E+09	3.79E+08	1.21E+09	1.28E+08	1.32E+08	7.52E+09	6.87E+09	n/a
T-test	12.32641	2.204159	-5.32644	6.040876	-2.13193	1.439164	3.650909	-0.18949	n/a
p-value	0	0.0289	0	0	0.0345	0.152	0.0003	0.8499	n/a
Reg 17 coeff	1415.217	1.71E+10	-2.23E+09	7.22E+09	-2.02E+08	94429134	3.04E+10	8.39E+09	0.867059
St. Error	52.46098	2.82E+09	1.85E+08	5.95E+08	62416439	64144657	3.66E+09	3.43E+09	0.038359
T-test	26.97657	6.065661	-12.0888	12.14211	-3.23975	1.472128	8.315994	2.445139	22.60367
p-value	0	0	0	0	0.0014	0.1429	0	0.0155	0
Reg 18 coeff.	1415.217	1.71E+10	-2.23E+09	7.22E+09	-2.02E+08	94429134	3.04E+10	8.39E+09	0.867059
St. Error	85.81063	3.37E+09	2.90E+08	8.84E+08	62248708	63682279	3.85E+09	4.48E+09	0.096315
T-test	16.49233	5.071493	-7.69657	8.174235	-3.24848	1.482817	7.896737	1.873589	9.002346
p-value	0	0	0	0	0.0014	0.14	0	0.0627	0

Appendix Aiii: Regression 19 – 21

	pop	pop65	lif	sav	mortf	mortm	hexp	dummyr	dummyt	resid(-1)
Reg 19 coeff.	1700.922	3.14E+10	-3.74E+09	1.26E+10	-7.2E+07	-2.18E+08	1.24E+10	4.91E+10	-1.18E+10	n/a
St. Error	150.2831	1.03E+10	6.17E+08	1.79E+09	2.24E+08	2.56E+08	2.33E+09	1.14E+10	2.08E+10	n/a
T-test	11.31812	3.064535	-6.06426	7.014015	-0.32146	-0.84901	5.334019	4.305237	-0.56708	n/a
p-value	0	0.0031	0	0	0.7488	0.3987	0	0.0001	0.5725	n/a
Reg 20 coeff	1824.227	4.33E+10	-4.07E+09	1.13E+10	26534798	-3.57E+08	1.21E+10	5.10E+10	-2.96E+09	0.705825
St. Error	110.1653	7.42E+09	4.39E+08	1.28E+09	1.60E+08	1.83E+08	1.66E+09	8.08E+09	1.48E+10	0.086512
T-test	16.55899	5.831217	-9.25975	8.758661	0.16613	-1.95134	7.286171	6.307227	-0.19975	8.158739
p-value	0	0	0	0	0.8685	0.0551	0	0	0.8423	0
Reg 21 coeff.	1824.227	4.33E+10	-4.07E+09	1.13E+10	26534798	-3.57E+08	1.21E+10	5.10E+10	-2.96E+09	0.705825
St. Error	196.1694	9.32E+09	6.75E+08	2.29E+09	1.47E+08	1.71E+08	2.55E+09	9.71E+09	5.98E+09	0.119448
T-test	9.29924	4.646874	-6.02833	4.905499	0.180965	-2.08801	4.733331	5.249189	-0.49502	5.909075
p-value	0	0	0	0	0.8569	0.0405	0	0	0.6222	0

Appendix Bi: Random Effect Models

	Variable	Coefficient	Std. Error	t-Statistic	Prob.
Model 1	C	-6.02E+11	6.88E+10	-8.75293	0
	POP	1827.471	109.4255	16.7006	0
	POP65	1.08E+10	6.75E+09	1.607396	0.1082
	LIFEEXP	4.22E+09	1.15E+09	3.670189	0.0003
	SAVINGS	9.35E+09	1.34E+09	6.963969	0
	MORTALFEM	-1.91E+08	1.12E+08	-1.7097	0.0875
	MORTALMALE	3.66E+08	1.18E+08	3.095711	0.002
	HEALTH	9.99E+09	1.61E+09	6.19996	0
	RESID01(-1)	0.663904	0.102415	6.482512	0
Model 2	C	-4.16E+11	5.22E+10	-7.97175	0
	POP	1849.031	77.09593	23.98351	0
	POP65	3.01E+10	5.35E+09	5.618438	0
	LIFEEXP	1.52E+09	8.34E+08	1.82322	0.0685
	SAVINGS	9.95E+09	9.40E+08	10.58997	0
	MORTALFEM	37578645	94622194	0.397144	0.6913
	MORTALMALE	-8.5E+07	1.17E+08	-0.72964	0.4657
	HEALTH	1.27E+10	1.14E+09	11.1096	0
	DUMMYR	3.44E+10	6.35E+09	5.419461	0
	RESID01(-1)	0.716872	0.069852	10.26277	0
Model 3	C	-5.99E+11	7.42E+10	-8.06766	0
	POP	1832.707	117.8511	15.55104	0
	POP65	1.08E+10	7.23E+09	1.492654	0.1357
	LIFEEXP	4.25E+09	1.23E+09	3.447513	0.0006
	SAVINGS	9.35E+09	1.44E+09	6.493885	0
	MORTALFEM	-1.91E+08	1.20E+08	-1.59106	0.1118
	MORTALMALE	3.68E+08	1.27E+08	2.902868	0.0038
	HEALTH	1.00E+10	1.73E+09	5.798977	0
	DUMMYT	-7.07E+09	1.64E+10	-0.43192	0.6659
	RESID01(-1)	0.661765	0.109827	6.025527	0
Model 4	C	-5.09E+11	6.96E+10	-7.32195	0
	POP	1881.706	103.1296	18.24602	0
	POP65	1.66E+10	6.50E+09	2.555147	0.0107
	LIFEEXP	2.92E+09	1.14E+09	2.567118	0.0104
	SAVINGS	1.00E+10	1.26E+09	7.928632	0
	MORTALFEM	-1.32E+08	1.05E+08	-1.2516	0.2109
	MORTALMALE	1.76E+08	1.24E+08	1.423171	0.1549
	HEALTH	1.09E+10	1.52E+09	7.168843	0
	DUMMYT	-3.51E+09	1.42E+10	-0.24724	0.8048
	DUMMYR	2.73E+10	8.16E+09	3.350335	0.0008
	RESID01(-1)	0.660439	0.095137	6.941955	0

Appendix Ci: Hausman Test Model 1

Hausman Test Summary
Test Summary		Chi-Sq. Statistic	Chi-Sq. d.f.	Prob.
Cross-section random		0	4	1
C	-2.19E+11	2.74E+09	-79.8588	0
POP	5.96E+11	2.42E+10	24.64955	0
POP65	6.52E+10	7.77E+08	83.91342	0
LIFEEXP	5.20E+08	15107134	34.4388	0
SAVINGS	-2.91E+08	11004547	-26.4186	0

Appendix Cii: Hausman Test Model 2

Hausman Test Summary
Test Summary		Chi-Sq. Statistic	Chi-Sq. d.f.	Prob.
Cross-section random		0	6	1
C	-4.40E+11	2.42E+11	-1.82041	0.0703
POP	2.22E+12	7.60E+11	2.925257	0.0039
POP65	9.82E+10	7.05E+10	1.393362	0.1652
LIFEEXP	9.03E+08	3.06E+08	2.949871	0.0036
SAVINGS	5.87E+08	1.86E+08	3.159266	0.0018
MORTALFEM	35255181	1.33E+08	0.265375	0.791
MORTALMALE	69447329	1.67E+08	0.416961	0.6772