Should Sub-Saharan African countries spend more on education?

(1)

SHOULD SUB-SAHARAN

AFRICAN COUNTRIES SPEND

MORE ON EDUCATION?

Student: Anne van Ravenhorst

Student number: 10587470

Supervisor: Dhr. R.M. Teulings

University of Amsterdam

Faculty Economics and Business

Bachelor in Economics

January 27, 2017

Corruption: A barrier preventing public

spending efforts from improving student

outcomes

By Carlota Cores

Supervised by: Andro Rilović Faculty of Economics and Business Student number: 10464034

29th of June, 2015

Abstract

Recent empirical research has highlighted the importance of efficiency regarding national education spending in order to improve student outcomes. The majority of studies linking corruption and education mostly conclude an adverse effect of corruption on educational performance and attainment. The aim of this study is to assess the effectiveness of secondary education expenditure in improving educational results under the presence of corruption. Student outcomes are approximated by PISA scores. Government expenditure on secondary education is found to positively impact PISA scores, whilst public education spending is estimated to be less effective in improving student results in more corrupt countries. These findings emphasize the need for transparency in the public sector if improved national student outcomes are to be achieved.

(2)

Statement of Originality

This document is written by Anne van Ravenhorst who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

Developing countries are in need for more economic growth to reduce poverty, especially Sub-Saharan African countries since this is the poorest region in the world. Most academics agree that education has a positive effect on economic growth, but not many have looked at the relationship between public education expenditure and economic growth with a sample of only developing countries. The aim of this paper is to investigate whether an increase in public education expenditure will have a positive effect of GDP per capita growth. This will be done by looking at two effects, first the effect of public education expenditure on growth and second on the composition effect of public expenditure. The results for the first effect show a negative but insignificant coefficient for education expenditure, only when 10-year lags are used the coefficient becomes positive, although still insignificant. This could be because spending on education will take many years until it affects economic growth. The coefficient for education is again negative in the composition effect, only spending on health will have an immediate positive effect on economic growth.

(4)

The effect of education spending on economic growth has been studied extensively by academics. This is because prior research has shown a positive relationship between education and economic growth. Artadi and Sala-i-Martin (2003) found that if primary school enrolment rates in Africa had been as high as those in OECD countries, the average growth rate of GDP per capita would have been 2.37% instead of 0.9% per year. This could have resulted in per capita incomes that would have been two and a half times larger than they actually are today. Other researchers like Hanushek and

Woessmann (2008) argue that not school attainment or the quantity of schooling is the key driver for economic development, but it is cognitive skills that have positive effects on individual earnings, on the distribution of income, and on economic growth.

However, empirical results on the relationship between spending on education and economic growth are mixed. And despite the fact that developing countries are the ones in need for more economic growth, since increased growth rates lead to poverty reduction (Dollar & Kraay, 2004), not many studies have looked at this relationship using a sample consisting only of developing countries (Bose, Haque, & Osborn, 2007). Therefore, additional research should be done looking at the relationship between education spending and economic growth in developing countries. The aim of this paper is then to investigate whether an increase in public expenditure on education will have a positive effect on economic growth in Sub-Saharan African countries. The focus is on Sub-Saharan African countries because they can be seen as the poorest developing countries. In the list of Least Developed Countries published by the United Nations, more than 70% of the 48 countries in total, are Sub-Saharan African countries (United Nations, 2016). Also when looking at the lowest per capita growth rates from 1965 to 1995, 18 out of 20 countries are Sub-Saharan African countries (Barro, 2003).

Since education has a positive effect on economic growth and increased growth rates reduce poverty, developing countries have an incentive to invest in education, both private and public. Average returns to investment in education are highest in Sub-Saharan Africa, especially for primary and secondary education (Psacharopoulos & Patrinos, 2004). Private investments give higher returns than social investments, but since data on private expenditure on education is very limited, this paper only focuses only on public expenditure.

In order to answer the research question, I will look at two different effects. First I look at the effect that raising public education expenditures has on growth per capita. Second, I look at the composition effect of public expenditure and thus whether it is preferred to choose to spend on education instead of any other component of public expenditure, like health or agriculture. For the composition effect, I use the method by Devarajan, Swaroop and Zou (1996).

The remainder of the paper is structured as follows; section 2 will discuss the related literature on economic growth, education and Sub-Saharan African countries. In section 3 the model and data

(6)

will be outlined. The empirical results will be discussed in section 4. And the last section of this paper contains the conclusion, limitations and implications for future research.

2. Theoretical framework

2.1 Economic growth and human capital

To try and understand economic growth, we first look at the Solow growth model (Solow, 1956). This model is built around two equations, a production function and a capital accumulation equation. The assumptions the model makes are diminishing returns to capital, a closed economy and perfect competition. The Cobb-Douglas production function is given by:

𝑌 = 𝐹 𝐾, 𝐿 = 𝐾!_𝐿!!!_, ₍₁₎ where 𝛼 is between 0 and 1, so there is constant returns to scale. Output, Y, is made out of two inputs, capital, K, and labour, L. For this research we are interested in per capita output, so we will rewrite the production function in terms of output per worker, y=Y/L, and capital per worker, k=K/L. This results in:

𝑦 = 𝑘!_. ₍₂₎ The other main equation in the model is the capital accumulation equation:

𝐾 = 𝑠𝑌 − 𝛿𝐾, (3) where 𝐾 is the change in capital stock, sY is the amount saved or invested, and 𝛿𝐾is the amount of depreciation. The change in capital also depends on the growth rate of the labour force and the exponential growth equation is as follows:

𝐿 𝑡 = 𝐿!𝑒!". (4) The labour force growth rate is then included into the capital accumulation equation and this equation is written in per worker terms:

𝑘 = 𝑠𝑦 − (𝑛 + 𝛿)𝑘, (5) where 𝑘 is the change in capital per worker, sy is the savings per worker, which equals investment per worker and increases k, nk is population growth and reduces k, and the last term 𝛿k is the depreciation of capital, which again reduces k. The saving/investment rate, s, the population growth rate, n, and the depreciation rate, 𝛿, are determined exogenously.

The model assumes that an economy will move towards its steady state. The steady-state value of capital per worker 𝑘∗, is when equation (5) equals zero. Capital per worker is now constant over time. Since 𝑦 = 𝑘!, 𝑘 = 𝑠𝑘!− 𝑛 + 𝛿 𝑘. It then follows that:

𝑘∗_{= (} !

!!!)!/(!!∝). (6)

The production function (2) and the steady-state value of capital per worker can then determine the steady-state value of output per worker, 𝑦∗= 𝑘∗!. So then:

(7)

𝑦∗ = (_!!!! )∝/(!!∝). (7) The model can now explain what happens to an economy when it experiences a shock. When for example the investment rate, s, increases, the economy will no longer be at its steady state. Now capital per worker will increase until it reaches its new and higher steady-state value 𝑘∗∗. This will result in a higher steady-state value of output per worker, 𝑦∗∗, which means that the economy is now richer than it was. The same line of argument holds for the population growth rate, but when this increases the result will be a lower steady-state value of output per worker and so a poorer economy. These results can also been seen from equations (6) and (7), they show what happens to the steady-state values when either s or 𝛿 changes.

The two determinants of growth according to the Solow model are investment and population growth. More investment and low population growth will result in an accumulation of capital per worker and thus an increase in labour productivity. This has a positive effect on output per worker and so the economy becomes richer. But due to diminishing returns to capital, this growth will stagnate once the economy becomes richer.

Where Solow says there are only two determinants of economic growth, Mankiw, Romer and Weil (1990) suggest that international differences in income per capita are best understood when human capital is added to the Solow model. They argue that when only physical capital is included, the estimated effects of investment and population growth on income are too large. Their results show that the augmented Solow model, where education is used as a measurement for human capital, is better at explaining differences between rich and poor countries. Thus not only investment and population growth are important determinants of economic growth, but education is as well.

The augmented growth model predicts that a country’s per capita growth rate is inversely related to the initial level of income per capita (Barro, 2003). This means that poorer countries tend to grow faster than richer countries. Mankiw et al. (1990) already criticized Solow’s theory about convergence on the fact that the estimated speed of convergence is much slower than predicted by the Solow model. Research done by Barro (2003) shows that this relation between the initial level of income per capita and a country’s growth rate, as explained by Solow, is not completely true, a poor country does tend to grow faster than a rich country, but only for a given quantity of human capital. Again, the influence of human capital on economic growth is shown by the theory.

2.2 The importance of education for development

Interest in the economic value of education started growing during the 1960s (Psacharopoulos & Woodhall, 1985). This interest was picked up by the World Bank, who provide financial and technical help to developing countries, and their policy began to reflect this view that education is a productive investment in human capital and they started several education projects in developing countries. And since the 1960s educational attainment grew rapidly in developing countries (Pritchett,

(8)

2001). But data show that the contribution of education to growth was much less than would have been expected by the augmented Solow growth model. Though this should not mean that we stop investing or invest less in education since economic growth is not the only goal we hope to achieve when investing in education. Besides that, most societies belief that education is a basic human right and that it is a basic component of social development (Psacharopoulos & Woodhall, 1985).

Many academics have focused on the effect of education on economic growth and development. In an extensive analysis of 67 explanatory variables in a cross-country economic growth regression, the primary school enrolment variable has the strongest influence on GDP per capita growth (Sala-i-Martin, Doppelhofer, & Miller, 2004). Krueger and Lindahl’s (2000) cross-country regressions indicate that the change in education and economic growth are positively associated. Other researchers, Temple (2001) and Benhabib and Spiegel (1994), stress the importance of the level of schooling rather than the growth rates of schooling in explaining economic growth. Recent research argues that not schooling but cognitive skills are the key driver for improving economic development (Hanushek & Woessmann, 2008). Most academics agree on the fact that education, whether it is measured in years of schooling or by someone’s cognitive skills, has a positive effect on economic growth.

Education may affect economic growth in at least three different ways (Hanushek & Woessmann, 2008). First, education impacts the human capital in an economy, which increases the labour productivity and moves the economy towards a higher steady-state level of output per capita, as was explained by the augmented Solow growth model. Second, the innovative capacity of the economy may be increased by education and this promotes growth. Third, education improves knowledge on how to process and implement new information and technologies, which again has a positive effect on economic growth.

Some researchers have investigated the effects of public education expenditure on economic growth. Results of Jung and Thorbecke (2003) suggest that an increase in public expenditure on education can contribute to economic growth and poverty alleviation. Research done by Devarajan et al. (1996) focuses on the composition effect of public expenditure, that is for example when a country tries to stimulate economic growth but the budget can not be increased, which component of public expenditure contributes most to economic growth and spending on this component should thus be preferred over the others. Their main finding is that growth per capita and the ratio of current to total expenditure, where education is a part of, have a positive and statistically significant relationship and current expenditure should thus be preferred over capital expenditure. But their results also show a negative, statistically insignificant, effect of public education expenditure on growth per capita. Only when disaggregating education expenditure into three different types of expenditure, one is positively and significantly related to growth. Results of Sylwester (2000) also show that public expenditure on education has a negative effect on economic growth, but that short lags are insignificantly positive and

(9)

that longer lags are significantly positive. One reason for these short-term negative effects is delayed benefits. He argues that the benefits of public education expenditure might not appear for at least a decade. Another reason he gives is that these expenditures are financed by distortionary taxation, which has a negative effect on the current growth rate. There could also be factors that influence the relationship and make public expenditure less efficient. Rajkumar and Swaroop (2008) results show that increasing public spending on education is unlikely to lead to better outcomes if countries have poor governance, where governance is measured by a corruption index and a bureaucratic quality index.

But even when results show a small, insignificant or even negative relationship between public education expenditure and economic growth, there are still many other reasons to invest in education. For example, self-discipline and problem-solving capabilities and other benefits like good health and improvement of child welfare (Stacey, 1998).

2.3 Education and growth in Sub-Saharan Africa

Sub-Saharan Africa is one of the regions in the world where the absolute number of poor almost doubled between 1981 and 2005 (Thorbecke, 2013). This persistent poverty can partly be explained by historical reasons. Global slave trade, colonial occupation, exploitation of their natural wealth, and little institutional, infrastructural, and human capital when independence was achieved (Barrett, Carter, & Little, 2006). Prolonged conflicts and the HIV/AIDS pandemic too have left large parts very poor.

During almost the same period, from 1980 to 2000, the average GDP per capita growth rate was negative, -0.6%, and the primary school gross enrolment rate declined, from 80% to 77%, in Sub-Saharan African countries (Glewwe, Maïga, & Zheng, 2014). This relationship between education and economic growth can again be seen in a later period, from 2000 to 2010, but then positive. The average GDP per capita growth rate in this period was approximately 2.5% and the primary gross enrolment ratio in 2010 was 100% (Glewwe, Maïga, & Zheng, 2014). Glewwe et al. have examined different recent studies and found strong empirical support for the causal impact of education on economic growth. Their main result is that in Sub-Saharan African countries, the low quality of education has a large negative effect on economic growth. So improving the quality of schooling could play an important role in increasing economic growth in these countries.

This paper only focuses on seven Sub-Saharan African countries due to data availability. Some key numbers about these countries are given in table 1 and were extracted from the World Bank World Development Indicators database. These countries are very poor when you compare them to for example the Netherlands. In 2009, the Netherlands had a GDP per capita of 51.900 dollars, a net enrolment rate of 99.5%, and life expectancy at birth of 81 years.

(10)

Table 1: The seven Sub-Saharan African countries in 2009

Population GDP per capita ($) Net enrolment rate (%) Life expectancy

Botswana 1.930.431 5115 85 62 Kenya 37.250.540 943 82.8 58 Malawi 13.498.377 432 97.5 55 Mauritius 1.239.630 7318 96.7 73 Nigeria 147.152.502 1092 64 51 Swaziland 1.134.853 3047 84.7* 48 Uganda 29.991.958 566 90.2 55

* is the number for 2007 since 2009 was not available

3. Methodology

3.1 The Model

This paper tries to assess the effect of public education expenditure on economic growth. A pooled regression model will be used to estimate the hypothesis, whether an increase in public expenditure on education will have a positive effect on economic growth. Since we are looking at two different effects, the single effect of public education expenditure on growth and the composition effect of different components of public expenditures on growth, two estimating equations will be tested. The regressions are estimated by Ordinary Least Squares. The model assumes that the error term has conditional mean zero, that the variables are identically, and independently distributed, that large outliers are unlikely, and finally that there is no perfect multicollinearity (Stock & Watson, 2012).

Panel data consisting of observations of seven Sub-Saharan African countries (Botswana, Kenya, Malawi, Mauritius, Nigeria, Swaziland and Uganda) from 1982 to 2011 is used in this analysis. The estimating equation for the first effect is:

𝑔𝑑𝑝𝑔𝑟!" = 𝛽!𝑒𝑑𝑢!"+ 𝛽!𝑔𝑑𝑝!,!!!+ 𝛽!𝑔𝑑𝑝𝑔𝑟!,!!!+ 𝛽!𝑤𝑜𝑟𝑙𝑑𝑔𝑑𝑝!!+ 𝛽!𝑙𝑖𝑓𝑒𝑒𝑥𝑝!"+ 𝛽!ℎ𝑢𝑚𝑎𝑛𝑐𝑎𝑝!" + 𝛽!𝑝𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙𝑟!" + 𝛽!𝑜𝑝𝑒𝑛𝑛𝑒𝑠𝑠!"+ 𝛼!+ 𝜀!",

and the estimating equation for the composition effect is:

𝑔𝑑𝑝𝑔𝑟_!" = 𝛽_!𝑡𝑜𝑡𝑎𝑙𝑒𝑥𝑝_!"+ 𝛽_!𝑎𝑔𝑟𝑖𝑐_!" + 𝛽_!𝑒𝑑𝑢_!" + 𝛽_!ℎ𝑒𝑎𝑙𝑡ℎ_!"+ 𝛽_!𝑔𝑑𝑝_!,!!!+ 𝛽_!𝑔𝑑𝑝𝑔𝑟_!,!!! + 𝛽_!𝑤𝑜𝑟𝑙𝑑𝑔𝑑𝑝_!"+ 𝛽_!𝑙𝑖𝑓𝑒𝑒𝑥𝑝_!"+ 𝛽_!ℎ𝑢𝑚𝑎𝑛𝑐𝑎𝑝_!"+ 𝛽_!"𝑝𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙𝑟_!"

+ 𝛽!!𝑜𝑝𝑒𝑛𝑛𝑒𝑠𝑠!" + 𝛼!+ 𝜀!".

The dependent variable, 𝑔𝑑𝑝𝑔𝑟!", is the GDP per capita growth rate and is used to measure economic growth. The variable of interest is 𝑒𝑑𝑢!" and is the percentage of public education expenditure in total public expenditure. Public expenditure on education is supposed to have a positive effect on economic growth, although this could be delayed for some years. A 1-year lag of the log of GDP per capita, 𝑔𝑑𝑝!,!!!, was taken so it reflects the initial level of GDP per capita and the coefficient

(11)

will then represent the rate of convergence. According to Barro (2003) convergence will occur once accounted for human capital as well, which has been done in this research by adding an index for human capital per person. If initial levels of GDP per capita are already high, a country’s growth rate will be lower compared to countries with lower initial levels of GDP per capita. The log of world GDP per capita, 𝑤𝑜𝑟𝑙𝑑𝑔𝑑𝑝!", is included to control for time fixed effects within the model. The time fixed effects could not be included in the model because due to the small sample of countries, there are not enough observations to estimate the time fixed effects precisely. World GDP per capita is used instead because it also captures common shocks that affect all countries’ GDP.

The variable 𝑙𝑖𝑓𝑒𝑒𝑥𝑝!" means life expectancy at birth and is expected to have a positive effect on economic growth since better health predicts higher economic growth (Barro, 2003). The index of human capital per person, ℎ𝑢𝑚𝑎𝑛𝑐𝑎𝑝_!", is also expected to have a positive effect on economic growth according to the earlier discussed literature. As a measure for democracy, a rating for a country’s political rights, 𝑝𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙𝑟!", is used. Expected is that having more political rights will enhance economic growth, which is supported by Rajkumar and Swaroop (2008) as they argue that public spending on education will be more effective in countries with good governance. The openness ratio, 𝑜𝑝𝑒𝑛𝑛𝑒𝑠𝑠_!", is included in the model and is expected to have a positive effect on economic growth. Harrison (1996) found that greater openness is associated with higher economic growth.

The percentage of total public expenditure in total GDP, 𝑡𝑜𝑡𝑎𝑙𝑒𝑥𝑝_!", is used to control for level effects (Devarajan, Swaroop, & Zou, 1996). Due to the addition of this variable we can look at the composition effect of the other public expenditure components. The two other components of public expenditure in the model were chosen due to data availability. These are public agriculture expenditure, 𝑎𝑔𝑟𝑖𝑐!", and public health expenditure, ℎ𝑒𝑎𝑙𝑡ℎ!".

3.2 Data

As was mentioned before, the sample of the analysis consists of seven Sub-Saharan African countries; Botswana, Kenya, Malawi, Mauritius, Nigeria, Swaziland and Uganda. The time period used is from 1982 to 2011. These choices were made due to data availability. Data for public expenditures was extracted from the IFPRI’s Public Expenditure for Economic Development (SPEED) database. The data for almost all the other variables were gathered from the World Bank World Development Indicators database, only the index for human capital per person and political rights are not. The index for human capital is from "The Next Generation of the Penn World Table" (2013) by Robert C. Feenstra, Robert Inklaar and Marcel P. Timmer, and the ratings for political rights are from the Freedom House. See Appendix A for a clear overview.

Some explanation on how the variables are constructed is necessary. According to Barro (2003), convergence will only happen when human capital is included in the model. Human capital is based on education and health. In this research we take the same measure for health as Barro (2003),

(12)

life expectancy. While he uses life expectancy at age one, we use life expectancy at birth due to data availability. As a measure for education, Barro uses average years of school attainment. Again due to data availability another measure had to be taken. An index for human capital per person is used instead and is based on years of schooling and returns to education. For the political rights of a country, a rating from 1 to 7 is used (Freedom House, 2015). Countries with a rating of 1 have a wide range of political rights including free and fair elections. Countries with a rating of 7 have few or no political rights because of government oppression, sometimes in combination with civil war. These seven categories have been adjusted to a zero-to-one scale where 0 can be seen as a democracy and 1 as a totalitarian system (Barro, 2003). The openness ratio is measured by the ratio of exports plus imports to GDP.

Following Barro’s ‘’Determinants of Economic Growth’’(2003), there are many other variables that influence GDP growth but these are not included in this model because this created an unbalanced sample. The fertility rate is the only variable that was excluded from the model not because it was unbalanced, but to prevent high multicollinearity with variables measuring health, and life expectancy therefore causing increasing standard errors. Other variables that are not included but do have an effect on economic growth are the government consumption ratio, the rule of law, inflation, terms-of-trade and the black market premium. Consequently, there is omitted variable bias.

3.2 Tests performed

To find the results, a panel data pooled regression is estimated. A benefit of using panel data is that you can control for unobservable country and time specific effects. Two techniques used for panel data are the fixed effects and the random effects model. The fixed effects model controls for country specific characteristics, as for example the trade policies of a country or its culture, which influences the other variables. The random effects model on the other hand assumes that these variations across countries are random and uncorrelated with the dependent and independent variables

(Torres-Reyna, 2007). To test whether the fixed effects or random effects model should be used, a Hausman test is performed and this tests whether the null hypothesis, that the preferred model is the random effects model, should be rejected or not. The Hausman test statistic is 14.03 with a p-value of 0.0507. With a 5% significance level, the null hypothesis should not be rejected. But since this is very close to the 5% level and it reasonable to think that the country specific characteristics in the model are not random, the fixed effects model is preferred in this analysis.

Besides only country fixed effects, the model can also control for time fixed effects. These differ over time but are the same for all countries, as for example a change in the oil price. The model controls for both the unobserved country and time specific effects and thus partly eliminates the omitted variable bias (Stock & Watson, 2012, p. 402). A joint test is performed to test whether the null hypothesis is rejected or not, so whether the coefficients for all years are jointly equal to zero. The F-statistic is 2.64 with a p-value of 0.0001, so with a significance level of 5% the null hypothesis is

(13)

rejected and time fixed effects are needed in this model. Although the time fixed effects regression model should be used, most of the coefficients in the model are insignificant. As was explained earlier, this could be due to the small sample of countries and for this reason world GDP per capita is included in the model instead of using the time fixed effects.

4. Results

4.1 The effect of public education expenditure

The results of the first model, the single effect of public education expenditure on GDP per capita growth, are presented in table 2, regression 1. Our coefficient of interest, edu, is negative but statistically insignificant. This is not consistent with our predictions. We expected a positive effect of public education expenditure on the growth rate of GDP per capita. Although the coefficient is very small and insignificant, a reason for a negative effect of public education expenditure on economic growth could be that investments in education take up a long time until they actually affect economic growth. Investments in education this year will only pay off over a couple of years, and consequently have a negative effect on economic growth this year. This is because the money used for the investments in education cannot be used for other components, which will have a direct effect on growth. To check if this might be the case, a 1-year, 5-year and 10-year lag of public education expenditure is used; see regression 2, 3, and 4 in table 2. Regression 2 and 3 show that the coefficient becomes less negative but also less significant. It is only after 10 years that the coefficient is positive, although still insignificant. Due to the loss of observations, we loose too much degrees of freedom leading to high standard errors. Using a longer time period could solve this problem.

Another explanation for a negative coefficient of education could be that the government is inefficient at generating actual education (Landau, 1986). Actual education is measured by enrolment ratios and this is positively related with economic growth rates, but the level op public expenditure on education is not. And another possible reason for the negative coefficient is omitted variable bias (Stock & Watson, 2012, p. 222). When the regressor is correlated with a variable that is not included in the analysis and that determines, in part, the dependent variable, the estimator will have omitted variable bias. The overview of variables given earlier that were excluded from the model because they created an unbalanced sample, all effect GDP per capita growth and not including them thus creates omitted variable bias.

The coefficient of initial per capita GDP growth, 𝑔𝑑𝑝!!!, is negative and statistically significant and thus shows conditional convergence as was predicted. The convergence is conditional in that a lower level of initial per capita GDP predicts higher growth only if the other explanatory

(14)

variables are held constant (Barro, 2003). Worldgdp is very positive and significant. This means that a country its GDP is very dependent on the GDP of the world.

The lifeexp coefficient shows a very small but significantly positive effect on the GDP per capita growth. This confirms the theory in which is stated that better health predicts higher economic growth. Humancap is positive but insignificant.

Politicalr is negative but insignificant in regression 1, although significant in regression 4. A

negative coefficient means that if a country with very few political rights is moving towards a lower rate, and thus the political rights are enhancing, the negative effect on economic growth is becoming smaller. So only when a country has a wide range of political rights and thus can be seen as a democracy, the effect on economic growth is zero. But when political rights are diminishing, this results in a negative effect on economic growth.

The last coefficient, the degree of openness of a country, is very negative and significant. This is in sharp contrast to what theory predicts. Yanikkaya (2003) found results that could support this negative coefficient. In his research, all measures of trade barriers except one are positively and significantly correlated to growth, which provides some evidence that restrictions on trade can promote economic growth.

(15)

Table 2: The effect public education expenditure on economic growth

Dependent variable: GDP per capita growth rate

(1) (2) (3) (4) 𝑒𝑑𝑢 -0.007 -0.004 -0.003 0.006 (-1.49) (-0.80) (-0.62) (1.43) 𝑔𝑑𝑝!!! -0.759*** -0.703*** -0.811*** -1.049*** (-5.71) (-5.57) (-5.77) (-5.72) 𝑔𝑑𝑝𝑔𝑟!!! 0.131* 0.119 -0.125* 0.074 (1.79) (1.63) (1.67) (0.88) 𝑤𝑜𝑟𝑙𝑑𝑔𝑑𝑝 0.924*** 0.887*** 1.037*** 1.430*** (5.30) (5.14) (4.40) (4.59) 𝑙𝑖𝑓𝑒𝑒𝑥𝑝 0.006* 0.006 0.006 -0.004 (1.69) (1.65) (1.58) (-0.89) ℎ𝑢𝑚𝑎𝑝𝑐𝑎𝑝 0.037 -0.001 0.022 0.284 (0.30) (-0.00) (0.15) (1.51) 𝑝𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙𝑟 -0.142 -0.124 -0.096 -0.200* (-1.61) (-1.41) (-1.09) (-1.90) 𝑜𝑝𝑒𝑛𝑛𝑒𝑠𝑠 -0.484*** -0.437*** -0.469*** -0.585*** (-3.64) (-3.41) (-3.47) (-3.73) 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 -0.991** -1.046** -1.343** -2.113*** (-2.19) (-2.31) (-2.32) (-2.79) 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠 196 196 175 140 t-statistics in parentheses

* Statistically significant at 10% level ** Statistically significant at 5% level *** Statistically significant at 1% level

4.2 The composition effect

Table 3, regression 5, shows the results of the composition effect. This time the coefficient for educ is significant but although still negative. Agric too is negative but insignificant. Health is the only significantly positive coefficient. This means that spending on health now will have an immediate positive effect on GDP per capita growth. In other words, if a government wants the GDP per capita growth to be increased soon, spending on health is preferred over agriculture and education. Looking at the other variables, there are no real changes in the coefficients only some small differences in the significance. Regression 6 shows the results of the composition effect when 1-year lags are taken for the expenditure variables. The only difference is that the coefficient for agric is now positive, although still insignificance. The same conclusion can be drawn as in regression 4, for a direct positive effect on economic growth, public expenditure on health is preferred over expenditures on agriculture and education.

(16)

In regression 7, 5-year lags are used for the public expenditure variables, and in regression 8 10-year lags are used. In the last regression, edu is now positive but insignificant. In regression 7 and 8, all the public expenditure variables have become insignificant so no real conclusions can be drawn from these results. This is not a real problem because when a government wants to stimulate growth by changing the composition of their expenditures, they will look at the short-term effects.

Table 3: The composition effect of public expenditure on economic growth Dependent variable: GDP per capita growth rate

(5) (6) (7) (8) 𝑡𝑜𝑡𝑎𝑙𝑒𝑥𝑝 -0.010*** -0.009** -0.001 0.001 (-3.01) (-2.54) (-0.23) (0.17) 𝑎𝑔𝑟𝑖𝑐 -0.001 0.006 0.006 0.001 (-0.13) (1.13) (0.92) (0.10) 𝑒𝑑𝑢 -0.011** -0.009* -0.003 0.008 (-2.35) (-1.94) (-0.64) (1.53) ℎ𝑒𝑎𝑙𝑡ℎ 0.010** 0.012*** -0.006 -0.009 (2.40) (2.88) (-1.08) (-1.07) 𝑔𝑑𝑝!!! -0.754*** -0.759*** -0.826*** -1.061*** (-5.80) (-6.12) (-5.56) (-5.42) 𝑔𝑑𝑝𝑔𝑟!!! 0.099 0.086 0.124 0.075 (1.38) (1.19) (1.64) (0.88) 𝑤𝑜𝑟𝑙𝑑𝑔𝑑𝑝 0.941*** -0.862*** 1.240*** 1.504*** (5.28) (4.47) (4.39) (4.14) 𝑙𝑖𝑓𝑒𝑒𝑥𝑝 0.006 0.003 0.005 -0.004 (1.73) (0.71) (1.29) (-0.81) ℎ𝑢𝑚𝑎𝑝𝑐𝑎𝑝 0.016 0.079 -0.010 0.275 (0.12) (0.61) (-0.07) (1.41) 𝑝𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙𝑟 -0.166* -0.177** -0.076 -0.203* (-1.87) (-2.03) (-0.84) (-1.86) 𝑜𝑝𝑒𝑛𝑛𝑒𝑠𝑠 -0.568*** -0.546*** -0.480*** -0.593*** (-4.24) (-4.24) (-3.51) (-3.72) 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 -0.707 -0.503 -1.946*** -2.329** (-1.48) (-1.04) (-2.73) (-2.50) 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠 196 196 175 140 t-statistics in parentheses

* Statistically significant at 10% level ** Statistically significant at 5% level *** Statistically significant at 1% level

(17)

5. Conclusion

A lot of empirical literature has examined the effect of education on economic growth, and some of them focused on public expenditure on education. But not many investigated the effect of public education expenditure on economic growth in a sample with only developing countries. Even though developing countries are exactly the ones needing more growth for poverty reduction. Especially Sub-Saharan African countries, since they can be seen as the poorest region in the world. That is why this paper aimed at answering the question whether an increase in public expenditure on education has a positive effect on economic growth in Sub-Saharan African countries.

To answer the question, two estimating equations were used. The first one looks at the effect of public education expenditure on economic growth and the second equation looks at the composition effect of public expenditure to see what component contributes most to economic growth.

Data was gathered for seven Sub-Saharan African countries from 1982 to 2011. The panel data pooled regressions were estimated using a fixed effects model. The variable worldgdp was included in the model instead of using a time fixed effects model because there are not enough observations to estimate the time fixed effects precisely.

The results for the first effect, the effect of public education expenditure on GDP per capita growth, showed a negative but statistically insignificant coefficient for edu. According to the related literature discussed, this was not expected. To check whether this was due to the fact that spending on education now will only effect growth in a couple of years, 1-year, 5-year, and 10-year lags were used to see if this was the case. Only after 10 years the coefficient has become positive, but is still insignificant. Other reasons for the negative coefficient could be that public expenditure is inefficient to improve actual education, and also because of omitted variable bias and thus not enough degrees of freedom.

The results for the composition effect again show a negative, but this time significant, coefficient for edu. The component of public expenditure that positively and significantly affects economic growth is health. So if a government wants to increase economic growth this year, spending on health should be preferred over the other components. These results do not change when 1-year lags are taken for the expenditure variables. Using 5- and 10-year lags, all the expenditure coefficients become insignificant, so no real conclusions can be drawn.

Results in this paper should be interpreted with caution due to some limitations. First, the sample of only seven countries is very small. Due to missing data for the dependent variable, many countries had to be removed from the sample. This selection process can introduce correlation between the error term en the independent variables, leading to sample selection bias (Stock & Watson, 2012, p. 365). Second, due to missing data again but this time for the independent variables, omitted variable bias occurred and this might make the estimators inconsistent (Stock & Watson,

(18)

2012, p. 224). Third, there is the possibility of reverse causality. A negative coefficient could also mean that slow-growing countries increase their expenditures on education in an attempt to grow faster (Devarajan, Swaroop, & Zou, 1996). This is called simultaneous causality and makes the estimators biased and inconsistent (Stock & Watson, 2012, p. 366). When more data becomes available, future research could try to overcome some of these limitations by using a longer time frame and test by how much and after how many years public expenditure on education will positively affect economic growth.

(19)

References

Artadi, E. V., & Sala-i-Martin, X. (2003). The economic tragedy of the XXth century: growth in Africa. NBER Working Paper , 9865.

Barrett, C. B., Carter, M. R., & Little, P. D. (2006). Understanding and reducing persistent poverty in Africa: Introduction to a special issue. The Journal of Development Studies , 42 (2), 167-177.

Barro, R. J. (2003). Determinants of Economic Growth in a Panel of Countries. Annals of Economics

and Finance 4 , 231-274.

Benhabib, J., & Spiegel, M. M. (1994). The role of human capital in economic development: Evidence from aggregate cross-country data. Journal of Monetary Economics , 34, 143-173.

Bose, N., Haque, M. E., & Osborn, D. R. (2007). Public Expenditure and Economic Growth: A Disaggregated Analysis for Developing Countries. The Manchester School , 75 (5), 533-556.

Devarajan, S., Swaroop, V., & Zou, H.-f. (1996). The composition of public expenditure and economic growth. Journal of Monetary Economics 37 , 313-344.

Dollar, D., & Kraay, A. (2004). Trade, Growth, and Poverty. The Economic Journal , 114 (493), F22- F49.

Freedom House. (2015). Freedom in the World Methodology. Retrieved from https://freedomhouse.org/report/freedom-world-2015/methodology

Glewwe, P., Maïga, E., & Zheng, H. (2014). The Contribution of Education to Economic Growth: A Review of the Evidence, with Special Attention and an Application to Sub-Saharan Africa.

World Development , 59, 379-393.

Hanushek, E. A., & Woessmann, L. (2008). The Role of Cognitive Skills in Economic Development.

Journal of Economic Literature , 46 (3), 607-668.

Harrison, A. (1996). Openness and growth: A time-series, cross-country analysis for developing countries. Journal of Development Economics , 48, 419-447.

Jung, H.-S., & Thorbecke, E. (2003). The impact of public education expenditure on human capital, growth, and poverty in Tanzania and Zambia: a general equilibrium approach. Journal of

Policy Modeling , 25, 701-725.

Krueger, A. B., & Landahl, M. (2000). Education for growth: Why and for whom? NBER Working

Paper , 7591.

Landau, D. (1986). Government and Economic Growth in the Less Developed Countries: An

Empirical Study for 1960-1980. Economic Development and Cultural Change , 35 (1), 35-75.

Mankiw, N. G., Romer , D., & Weil, D. N. (1990). A contribution to the empirics of economic growth.

NBER Working Papers , 3541.

Pritchett, L. (2001). Where Has All the Education Gone? The World Bank Economic Review , 15 (3), 367-391.

(20)

Psacharopoulos, G., & Patrinos, H. A. (2004). Returns to investment in education: a further update.

Education Economics , 12 (2), 111-134.

Psacharopoulos, G., & Woodhall, M. (1985). Education for Development: An Analysis of Investment Choices. New York: Oxford University Press for the World Bank.

Rajkumar, A. S., & Swaroop, V. (2008). Public spending and outcomes: Does governance matter?

Journal of Development Economics , 86, 96-111.

Sala-i-Martin, X., Doppelhofer, G., & Miller, R. I. (2004). Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach. American Economic Review ,

94 (4), 813-835.

Solow, R. M. (1956). A Contribution to the Theory of Economic Growth. The Quarterly Journal of

Economics , 70 (1), 65-94.

Stacey, N. (1998). Social Benefits of Education. The Annals of the American Academy of Political and

Social Science , 559, 54-63.

Stock, J. H., & Watson, M. M. (2012). Introduction to Econometrics . Boston: Pearson.

Sylwester, K. (2000). Income inequality, education expenditures, and growth. Journal of Development

Economics , 63, 379-398.

Temple, J. R. (2001). Generalizations that aren't? Evidence on education and growth. European

Economic Review 45 , 905-918.

Thorbecke, E. (2013). The Interrelationship Linking Growth, Inequality and Poverty in Sub-Saharan Africa. Journal of African Economies , 22, i15-i48.

Torres-Reyna, O. (2007). Panel Data Analysis Fixed and Random Effects using Stata. Retrieved from https://www.princeton.edu/~otorres/Panel101.pdf

United Nations. (2016, May). List of Least Developed Countries. Retrieved from http://www.un.org/en/development/desa/policy/cdp/ldc/ldc_list.pdf

Yanikkaya , H. (2003). Trade openness and economic growht: a cross-country empirical investigation.

Journal of Development Economics , 72, 57-89.

(21)

Appendix A

Table A1: data sources

gdp World Bank educ SPEED dataset lifeexp World Bank

humancap Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer (2013), "The Next Generation of the Penn World Table"

politicalr Freedom House openness World Bank totalexp SPEED dataset agric SPEED dataset health SPEED dataset

Should Sub-Saharan African countries spend more on education?