Can foreign aid increase the economic growth of a country?

15-07-2016

Can foreign aid increase the economic growth of a

country?

1

Statement of Originality

This document is written by Student Maria Frederieke Toppen, 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.

2

This paper examines the relationship between foreign aid and the economic growth of a country. A more up to date time base is used in this paper. The outcomes of other studies ranged from a positive effect of foreign aid on the economic growth, to no or even a negative effect. This study shows no clear positive or negative effect. The effect differs between the countries selected, and is highly depended on the set of factors chosen for each country.

1. Introduction

The main purpose of foreign aid is to stimulate the development of the economy and the population in a country, especially when focused on third world countries (OECD, 2016). A longer term goal is to put an end to poverty in the world (R. Williamson, 2009). Although the purpose of foreign aid seems clear, it is not clear to what extent foreign aid can reach these goals. Some researchers think of foreign aid as a failure and are of the opinion that most foreign aid is wasted (Easterly (2002), Djankov Montalvo and Reynal-Querol (2006), Boone (1996)).

These researcher that think of foreign aid as a waste argue that foreign aid only increases the unproductive public consumption (Alesina & Dollar, 2000). An example of the failure of foreign aid is a Canadian aid project in Lesotho taking place from 1975-1984 called the Thaba-Tseka Project. This project was meant to help farmers who were living in the mountains of Lesotho. It was supposed to give the farmers access to markets and develop better methods for the production of crops. The problem was that those farmers from Lesotho already knew that their crops production was not competitive enough because of the poor conditions in that region. And instead of providing these local farmers access to other

markets, farmers from other areas with better conditions would bring in their crops and drive out the local farmers from Lesotho. Despite the construction work and the expensive road work, the project did not lead to the expected economic growth in Lesotho (Ferguson, 1994).

There are also some researchers that argue that foreign aid is not wasted and instead could increase the economic growth of a country (Dalgaard, Hansen & Tarp, 2004). An example of successful foreign aid is the Marshall Plan, also known as the European Recovery Program. The Marshall Plan started three years after the end of the second World War. The USA gave more than 4 percent of their GDP to foreign aid (Sogge, 2002). It was meant for the recovery of the many countries in Europe that were affected by the second World War. According to Wood (1986) the Marshall Plan has been the most successful program in the history of foreign aid. It achieved all of its goals within the time span of four years.

3

This indistinctness about the effectiveness of foreign aid causes a lot of discussions. The OECD organized four events, called the High Level Fora on Aid Effectiveness, and these events where meant to discuss why foreign aid did not result in the development goals

everyone expected. The last High-level event, which was held in Busan, South Korea in 2011, ended in more than 100 countries who agreed to endorse a set of principles which would help to maximize the effectiveness of foreign aid (OECD).

Although the discussion regarding the effectiveness of foreign aid seems to be less of a priority after the last High-level event, it is still a relevant subject as illustrated by the

Chinese investment in Africa. It was only at the end of 2015, on the 4th of December, that Xi Jinping, the president of China, announced to donate 60 billion dollars towards foreign aid in Africa (Prisco, 12-04-2015, CNN). In 2012 the amount they gave to Africa as financial support was only 20 billion dollars. The amount of foreign aid China is giving to Africa now is more than Africa received in total from all countries together in 2013 (Witteman,

Volkskrant, 12-06-2015).

The problem with foreign aid is that it is not clear when foreign aid can increase the economic growth in a country and when it cannot. Notwithstanding this unclearness, still big cash flows of foreign aid goes around in the world. And if the economy of a country did increase, it is hard to tell if this is because of foreign aid or if it is because of other factors. There is not a clear model that can predict how foreign aid influences the economic growth. Partly because agencies do not give high priority to evaluate a project (Easterly, 2003). According to Easterly, the World Bank only reviews five percent of its loans with as purpose development after three to ten years after the last expenditures. If the World Bank, and other organizations, review their aid projects more sooner, they can learn from their past projects and learn how it affected, or did not affected, the development and growth of a country.

In my thesis I will explore if foreign aid can increase the economic growth of a country. I already mentioned that researchers differ about the fact if foreign aid can increase the economic growth. In the next section, empirical research that already have been done on this subject will be considered and the difference with my intentions will be described. In the third section the methodology is explained. The methodology starts with the cases (countries) for which the model is implemented and will be followed by the variables that will be used in the regression. The methodology ends with an explanation about the type of regression that is used. In the fourth section the dataset is reviewed. The fifth section investigates the result of my research, and I will conclude with an answer to the research question, can foreign aid increase the economic growth of a country.

4 2. Literature review

The effectiveness of foreign aid has been subject to extensive research.Although there are multiple models about the relation between foreign aid and economic growth, it is not clear which model is the right one. The conclusions of the studies that have been done differ

substantially. In this review I will first elaborate on some studies that were concluded negative on the role of foreign aid or are mixed in their appraisal for foreign aid. In the second part some studies are described that were concluded positive on the role of foreign aid.

Foreign aid does not help

Easterly (2007) mentions in his article that there are almost no cases where foreign aid led to development. He describes that the top quarter of aid recipients received 17 percent of their GDP in aid over 42 years, but the growth per capita was almost equal to zero. Easterly (2007) mentioned that it is not clear what actions can lead to development, but according to him foreign aid does not make those actions happen. However, he mentions that foreign aid is not per se a bad thing. Foreign aid can help with smaller tasks for which there is a huge demand. Examples are providing clean drinking water or help children going to school. ‘It could seek to create more opportunities for poor individuals, rather than try to transform poor societies’ (Easterly, 2007, pp. 331).

Djankov, Montalvo and Reynal-Querol (2006) also show empirical evidence that foreign aid is not effective. One of the reasons for this is that foreign aid causes an increase in the government consumption, but the investment will decrease. However, especially an increase in investment is important for economic growth. According to Djankov, Montalvo, and Reynal-Querol (2006) it would be better to give out a loan instead of a grant. While a grant is not used wisely, a loan would be more effective, because the loan has to be paid back, so the receiving country will use the loan more wisely.

Another research about foreign aid is done by Boone (1996). He investigated the effectiveness of aid and he relates aid effectiveness to political regimes. As variable of foreign aid he used the net Official Development Assistance (ODA), which he retrieved from the OECD. The ODA includes only aid that has the purpose of improving the economic and human welfare. Boone used the ODA cash flows to 96 countries. His sample consisted of four time periods of each 5 years, with as last period 1986 till 1990. Boone (1996) concludes that foreign aid does not have a positive effect on the economic development. Foreign aid does not

5

have a significant influence on the investment or the benefits of the poor in a country. Besides that, it does not matter for the effectiveness of the aid if the country is repressive or liberal democratic (Boone, 1996). So the effectiveness of aid does not relate to the political regime.

Foreign aid does help

One of the most well-known studies about foreign aid is Burnside and Dollar (2000). They studied the relationship between foreign aid, economic policies, and growth of real GDP per capita. Their definition of foreign aid is that it acts as an income transfer, and that it can produce growth but this depends on if aid is used to invest or used to consume. If it is used to invest, foreign aid will be effective, but it will depend on the policy in the receiving country. Burnside and Dollar retrieved their date from the World Bank and used a panel of 56

countries for six four-year time periods with as last time period 1990-1993. Their hypothesis was that the effectiveness of aid depends on the policies that affect growth. They run a number of regression models with the growth rate of GDP per capita as dependent variable and the most important variables were a number of institutional and political variables and a policy index of the inflation rate, the budget surplus and a dummy for the openness of a country. The results of the research of Burnside and Dollar were that aid has only a small impact on growth and that for a country where the policy environments were good the impact of aid was bigger. Another finding was that the last years, the policy environments of poor countries were getting better, what means that the effectiveness of aid will increase in those countries (Burnside & Dollar, 2000). This differs from the research of Boone (1996). In the research of Burnside and Dollar (2000) they argue that foreign aid could be effective but it depended on the policy in a country while Boone (1996) argues that foreign aid is not effective, and that it does not depend on the policy in a country.

According to the research done by Dalgaard, Hansen and Tarp (2004) foreign aid has a significant positive influence on the productivity in a lot of countries, but that foreign aid cannot reduce poverty directly. The aid can only stimulate the process of reducing poverty. Another finding of their research was that the effectiveness of foreign aid depends on the climate of the country. Aid seems to be less effective in tropical areas. This research was not their first work on the effect of foreign aid. In 2001 Dalgaard and Hansen discussed the

research of Burnside and Dollar, which I described above (Dalgaard, Hansen, 2001). Dalgaard and Hansen used the same data as Burnside and Dollar, but used a different growth model. They found evidence that the influence of aid on productivity is positive, but that the positive effect diminishes as the aid inflow increases. With regard to the interaction between aid and

6

the policy in a country, the coefficient of this interaction is significant positive according to Burnside and Dollar (2000), but according to Dalgaard and Hansen (2001) this coefficient is insignificant. Dalgaard and Hansen (2001) found a positive impact of foreign aid on economic growth, no matter what kind of policy the country had. This is because they conditioned on policy in the regression before they evaluated the effect of foreign aid, which Burnside and Dollar (2000) did not do.

Relevance for my research

I am not the first one who wants to investigate the effectiveness of foreign aid, but most of the studies on this topic have been done more than 15 years ago. Despite the many studies, there still is not a clear answer to the question if foreign aid can cause economic growth. My study will not be innovative in how I do my empirical research, but it will use a recent time sample. My hypothesis is that countries have learned from the last years why foreign aid is not always as effective as they hoped it would be, and that foreign aid is more effective nowadays. I want to use the research of Burnside and Dollar as basis, because their study is very well-known and highly quoted. I will use a different time sample, which is more up to date, and I will do research if there are other important variables which have influence on the economic growth of a country.

3. Methodology

In this section I will start with explaining on which countries I will base my research and why. Next I will describe the factors that are important for the economic growth of a country and provide the chosen growth equation for this study. The methodology ends with an explanation about the choice of method.

The countries

Africa is a continent that is well known for their lack of development. Although there are parts that are quite wealthy, for example Tunisia with a GDP per capita at PPP (current international $) of $11,436, or South-Africa with a GDP per capita PPP (current international $) of $13,049 (World Bank, 01-07-2016), still a part of Africa is underdeveloped, mainly Sub-Saharan Africa. Because of this, a substantial part of foreign aid is going to Sub-Saharan Africa. The GDP per capita at PPP (current international $) in Sub-Saharan Africa was $3,514 in 2014 compared to a GDP per capita at PPP (current international $) of $18,222 in Middle

7

East & North Africa (World Bank, 01-07-2016). In Europe and Central Asia the GDP per capita at PPP (current international $) was $29,057 in 2014 (World Bank, 01-07-2016). To make a better comparison for the effectiveness of foreign aid, I decided to choose five

countries in Sub-Saharan Africa that are almost equal in their GDP per capita at the beginning of the time series (table 1), but differ in the amount of foreign aid they received and differ in their real GDP per capita at the end of the time series. These five countries are Comoros, Ghana, Kenya, Madagascar, and Togo. It is interesting to see that the amount of foreign aid received differ substantially between these countries. According to Alesina and Dollar (2000) the reason why the amount of foreign aid received differ between countries is because of political and strategic considerations. Besides that, the colonial past of a country can also influence the amount of aid they receive. Countries will give more aid to their former colonies (Alesina and Dollar, 2000).

Looking at the overall growth rate of real GDP per capita (table 2) and amount of foreign aid received (table 3), there is not a clear answer about the effect of foreign aid. Ghana and Kenya received, compared to the other countries, a high amount of foreign aid (table 3), and their real GDP per capita showed the highest increase (table 2). However, the real GDP per capita of Madagascar decreased with 34.1% between 1980 and 2014, although it also received quit an amount of foreign aid. Graph 1 shows that the real GDP per capita is not stable for each country. For Comoros, Madagascar, and Togo the real GDP per capita moves between the 250 and 400 dollar. Only the real GDP per capita for Ghana and Kenya grew substantially, compared to Comoros, Madagascar and Togo, for which the real GDP per capita decreased between 1980 and 2014 (table 2). Because of these differences in the amount of foreign aid received and the real GDP per capita growth for each country, I hope to get a clear view on the effect of foreign aid and why it is more effective in one country than in another country.

8

Graph 1. Development real GDP per capita, Comoros, Ghana, Kenya, Madagascar and Togo, 1980-2014

The GDP per capita (current US$) are originated from the World Bank. The real GDP per capita is calculated as follows:

1. Calculate the exchange rate in 1980 for each country with the following equation: 𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎 (1980)(𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑙𝑜𝑐𝑎𝑙 𝑐𝑢𝑟𝑟𝑒𝑛𝑐𝑦)

𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎 (1980)(𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑈𝑆$)

2. Change the GDP deflator for each country to the same base year, 1980 GDP deflatort (base year=1980)=

𝐺𝐷𝑃 𝐷𝑒𝑓𝑙𝑎𝑡𝑜𝑟 (𝑡) 𝐺𝐷𝑃 𝐷𝑒𝑓𝑙𝑎𝑡𝑜𝑟 (1980)

3. Real GDP per capitat (constant 1980 US$) =

𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎(𝑡)(𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑙𝑜𝑐𝑎𝑙 𝑐𝑢𝑟𝑟𝑒𝑛𝑐𝑦) 𝐺𝐷𝑃 𝐷𝑒𝑓𝑙𝑎𝑡𝑜𝑟 (𝑡) 𝐸𝑥𝑐ℎ𝑎𝑛𝑔𝑒 𝑟𝑎𝑡𝑒 1980 0 100 200 300 400 500 600 700 800 900 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 20 06 20 08 20 10 20 12 20 14 G D P p e r c a p i t a Year

Development real GDP per capita

1980-2014

9

Table 1. GDP per capita (N) (current US$) and real GDP per capita (R) (constant 1980 US$), Comoros, Ghana, Kenya, Madagascar and Togo, 1980-2014

Country / Year 1980 1990 2000 2014 Comoros (N) 400 602 372 810 Comoros (R) 400 397 367 353 Ghana (N) 412 403 265 1442 Ghana (R) 412 376 445 763 Kenya (N) 447 366 409 1358 Kenya (R) 447 462 419 548 Madagascar (N) 462 267 246 449 Madagascar (R) 462 369 321 305 Togo (N) 418 430 266 635 Togo (R) 418 333 321 330

World bank, World Development Indicators

Table 2. The percentage growth of the real GDP per capita, Comoros, Ghana, Kenya, Madagascar and Togo, 1980-2014

Country Percentage growth

between 1980-2014 Comoros -11,7 Ghana 85,4 Kenya 22,6 Madagascar -34,1 Togo -21,0

10

Table 3. Net official development assistance and official aid received (x 1.000, current US$), Comoros, Ghana, Kenya, Madagascar and Togo, 1980-2013

Country/ Year 1980 1990 2000 2013 Comoros 43.250 44.880 18.710 81.900 Ghana 190.810 559.720 598.170 1.330.510 Kenya 394.790 1.181.290 512.720 3.236.280 Madagascar 229.550 396.960 320.200 499.760 Togo 90.450 258.240 69.550 220.530

World Bank, World Development Indicators

Another reason why these countries are chosen is because they not only differ in their real GDP per capita and the amount of foreign aid received, but they also differ in their policies. A difference in policy can mean many different things, for example that in one country is more corruption than in another country, or a difference in the regulation of trade. The Country Policy and Institutional Assessment index, also called the CPIA index, is a measure of policy indicators in a country (Collier and Dollar, 2002). This index depends on a set of 16 criteria, which are related to economic management, structural policies, policies for social inclusion and equity or public sector management and institutions (World Bank, 2016). A CPIA score is between 1 and 6, and a high CPIA score means a good policy (Dalgaard, Hansen & Tarp, 2004). The overall CPIA scores in 2015 ranged from 2.2 to 4.3 (graph 2). There was no data available of the CPIA scores for years before 2004. Since 2004, the CPIA scores of Kenya and Togo increased substantially. The CPIA score of Madagascar increased between 2004 and 2008, but decreased in the following years. The CPIA score of Comoros was quite stable between 2004 and 2015, and for Ghana the CPIA score was increasing since 2004, but declined since 2012. Looking at the year 2015, Kenya is the country with the best policy according to the overall CPIA index, followed by Ghana, Comoros, Madagascar and Togo (African Development Bank Group, 07-06-2016) .

11

Graph 2. Development of the overall CPIA scores, Comoros, Ghana, Kenya, Madagascar and Togo, 2004-2015.

African Development Bank Group

Choice of variables

Economic growth depends on many factors, not only on foreign aid. If foreign aid even has an effect on the economic growth, which is of course the main question of this study. Besides a variable for foreign aid, I will include other variables in the growth equation. The dependent variable in this equation is the economic growth as measured by the growth of the real GDP per capita.

To control for the variable foreign aid, other variables has to be included, for example the initial real GDP per capita. This variable is included to capture convergence effects (Burnside and Dollar, 2000). Convergence effects, also called the catch up effect, is the idea that economies with a lower per capita income will tend to grow at faster rates than the economies with an higher income per capita. Because of these higher growth rates for poorer economies, and lower growth rates for economies that are richer, the income per capita for countries tend to converge (Lindauer, Perkins & Radelet, 2012).

The second variable which is included is M2 as a percentage of GDP. This variable is included to take the development of the financial markets into account (Dalgaard, Hansen &

12

Tarp, 2004). According to Chauvet and Guillaumont (2001) it is a good measure of the initial distortions in the financial system or the financial depth.

According to many scientists, a main reason why foreign aid fails sometimes is because of the policy of a country. Burnside and Dollar (2000) described that foreign aid only can cause economic growth if the macroeconomic policy in a country is good. Collier and Dollar (2002) included multiple policy related variables in their regression for economic growth. They included financial development as variable, which was calculated by the liquid liabilities of the financial system divided by the GDP. Also the government budget was taken into account, which was measured by the government surplus as part of the GDP. The last policy variable was the distortion variable which was estimated by the black market exchange rate premium. Each of these policy variables were found to be significant. Burnside and Dollar (2000) also included in their regression a control variable for the policy in a country included in their regression. They made their own policy variable, which is based on the inflation rate, the budget surplus and a dummy for the openness. This dummy for the

openness is developed by Sachs and Warner (1995) and measures how open a country is for trade. According to them a country has a closed trade policy if it has one of more of the following characteristics:

1. Nontariff barriers covering 40 percent or more of trade; 2. Average tariff rates of 40 percent or more;

3. A black market exchange rate that is depreciated by 20 percent or more relative to the official exchange rate, on average, during the 1970s or 1980s;

4. A socialist economic system; 5. A state monopoly on major exports. (Sachs and Warner, 1995, p. 22)

This policy index is significant at the five percent statistical significance level, which means that the policy in a country has an effect on the economic growth.

Another measure of policy, which is used by Collier and Dollar (2002) and is already mentioned earlier in this section is the CPIA index. In their regression the dependent variable was growth rate of per capita GNP, and the variable CPIA is significant for all regression at the 10 percent statistical significance level of 10 percent.

It is clear that a control variable has to be included in the equation that measures the policy. Because there is some criticism on the CPIA index as policy variable, it will not be

13

used as a measure of the policy index in this research. Kaufman and Kraay designed the CPIA index and, as cited by Alexander (2009), also say themselves that this index has a high margin of error. For example, the CPIA do not take exogenous shocks into account, which means that the CPIA can punish governments for factors, although the government could impossible control these factors (Alexander, 2009). As the research of Burnside and Dollar (2000) is one of the most cited contributions in this field, their choice of policy variable will be used in this regression analysis.

For simplicity, Burnside and Dollar (2000) decided to make an overall measure of the policy instead of the three different measures for inflation, budget surplus and the openness dummy. This policy index should weight these three policy variables according to their correlation with growth. This policy index is included in the growth equation (Burnside and Dollar, 2000, pp. 855).

(1) Policy = α + Inflation * β1 + Budget Surplus * β2 + Openness * β3

As already mentioned, the effectiveness of foreign aid can fail sometimes because of the policy in the receiving country. Because of that I will include a control variable in the

equation to control for this. Namely, (aid/GDP) * policy. This interaction term means that the effect of aid is depended on the policy in a country. Even when the amount of foreign aid received is high, the effect on the economic growth can be lower as expected because of a low policy index. Or if the policy index in a country is high, the effect of foreign aid has a larger effect on the economic growth of a country.

If foreign aid has a positive influence on the economic growth, most of the times this positive influence diminishes (Dalgaard and Hansen, 2001). Because of this diminishing return of foreign aid, Burnside and Dollar (2000) include a variable (aid/GDP)2 * policy. In their research this variable improved the R-squared of their regression, what means that the data is closer to the fitted regression line. Because of this, I consider it to be an important factor to include in the equation in this study.

If a country does not spend the received foreign aid well, what means that the foreign aid is not used for development purposes, but for example is used for expanding

nonproductive expenditures like military capability, or is used for inefficient investment, it can be a reason for the failure of foreign aid (Griffiths and Wall, 2007). The economic growth

14

depends on how the foreign aid is used. If the foreign aid is consumed, aid will not be

effective. But if the foreign aid is invested, the domestic output can increase, what can lead to economic growth (Burnside and Dollar, 2000). Sometimes the government of the receiving country does not spend the foreign aid wisely on purpose. According to Pedersen (2001) some countries that receive foreign aid know that they will get more aid donated by organizations if these organizations observe more poverty, what will lead to no incentives for those poor countries to improve their poor situation. This is called the Samaritan’s Dilemma

(Williamson, 2009). According to Buchanan, as cited by Williamson (2009), the receiving country has different incentives, which are not in line with increasing the economic growth. Their incentive to consume more, instead of investing or saving it, will make them dependent on their donors. In the growth equation it is important to include a variable that measures the investment. Because economic growth depends on how much of the foreign aid is invested.

With these variables, the growth equation will be as follows:

Growth rate of GDP per capita = β1 *log(Initial GDP per capita) + β2 * M2/GDP + β3 * policy index + β4 * Aid/GDP + β5 * (Aid/GDP) * policy + β6 * (Aid/GDP)2 * policy + β7 * investment/GDP

Choice of method

In this study time series analyses will be done. I will make use of the method of ordinary least squares. The assumption is made that the error term variance is not constant, which mean that heteroskedastic errors will be used in the regression analysis.

4. Dataset

In this section will be described how the different variables included in this research are measured.

The dependent variable, the growth of real GDP per capita will be measured as followed: (real GDP per capitat – real GDP per capitat-1) / realGDP per capitat-1. The real

GDP per capita is measured in constant 1980 US$.

A summary of independent variables is provided in table 4. One of the independent variables is initial GDP. This is measured as the logarithm of the real GDP per capita at the beginning of the period, measured in constant 1980 US$.

15

The variable M2 is measured by the total of money and quasi money divided by the total GDP of the specific country.

In line with Burnside and Dollar (2000) the policy index formula that will be included in the growth equation will look as follows:

Policy index = α + β1 * Inflation + β2 * Budget Surplus + β3 * Trade

An explanation about how this policy index is calculated is given at the end of this section. Foreign aid is measured by the net official development assistance and official aid received, in current US dollars. To get to the aid variable this number is divided by the total GDP. Two other variables that are included are (the amount of foreign aid received / total GDP) times the policy index and (the amount of foreign aid received / total GDP)2 times the policy index.

Investment is measured by the total amount of invested money as percentage of the total GDP.

16 Table 4. Summary of the variables

Variable name as used in the regression analysis

Measured by Source

Growth (real GDP per capitat – real GDP per capitat-1) / real

GDP per capita

t-World Bank

Initial GDP ln (real GDPt-1 per capita World Bank

M2/GDP M2/GDP (lagged) Indexmundi

Aid/GDP net official development assistance and official aid received (in current US$) / GDP (in current US$)

Aid: World bank GDP: IMF (Aid/GDP) * policy (net official development assistance and official aid

received (in current US$) / GDP (in current US$)) * policy

(Aid/GDP)2 * policy (net official development assistance and official aid received (in current US$) / GDP (in current US$))2 * policy

Investment/GDP Investment / GDP (in current US$) IMF

Inflation Ln (1 + change in average consumer prices) IMF Budget surplus ((government revenue / GDP (in current US$)) –

(government expenditures/GDP (in current US$))) / 100)

IMF

Trade (import + export) / GDP World Bank

Before starting with the regression analysis, the variable policy index has to be calculated. This has to be done for each country. An example of the calculation of this policy index for Kenya is given below. The data is for the country Kenya, for 1982-2012. First a regression analysis will take place for the variable growth on all the variables except the variables that are related to the foreign aid received. This gives the following results:

17 Table 5. Regression output, Kenya

(1) VARIABLES Growth M2/GDP 0.350*** (0.109) Initial GDP -0.203* (0.112) Inflation -0.210** (0.0848) Budget surplus -0.215 (0.206) Trade 0.00423 (0.0462) Investment/GDP 0.382*** (0.120) Constant 1.062 (0.656) Observations 31 R-squared 0.608

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table 6. Mean estimations, Comoros, Ghana, Kenya, Madagascar and Togo

VARIABLES Comoros Ghana Kenya Madagascar Togo

M2/GDP 0.240*** 0.224*** 0.346*** 0.203*** 0.320*** (0.0116) (0.0117) (0.00793) (0.00539) (0.0150) Initial GDP 5.919*** 6.073*** 6.088*** 5.827*** 5.734*** (0.0134) (0.0341) (0.00924) (0.0208) (0.0142) Inflation 0.0305*** 0.237*** 0.110*** 0.132*** 0.0391*** (0.00947) (0.0294) (0.0134) (0.0145) (0.0128) Budget Surplus -0.00787 -0.179*** -0.0364*** -0.0603*** -0.0331*** (0.00463) (0.00832) (0.00557) (0.00717) (0.00528) Trade 0.587*** 0.620*** 0.566*** 0.532*** 0.843*** (0.0158) (0.0522) (0.0112) (0.0296) (0.0283) Investment 0.116*** 0.149*** 0.185*** 0.166*** 0.153*** (0.00583) (0.0119) (0.00819) (0.0132) (0.00840) Aid/GDP 0.164*** 0.0552*** 0.0546*** 0.103*** 0.0837*** (0.0168) (0.00440) (0.00381) (0.00932) (0.00852) Observations 30 34 31 34 25

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

18

The results for the policy index equation for all countries are below:

Comoros

Policy index = -1.269444255 - 0.2526021 * inflation - 0.0424574 * budget surplus - 0.0402312 * trade

Ghana

Policy index = 0.202870282 - 0.0807315 * inflation - 0.109395 * budget surplus + 0.040601 * trade

Kenya

Policy index = -1.043889366 - 0.2103184 * inflation - 0.2152141 * budget surplus + 0.0042255 * trade

Madagascar

Policy index = -1.299650764 - 0.0315665 * inflation - 0.2002563 * budget surplus + 0.1194669 * trade

Togo

Policy index = -0.430387061 + 0.5254391 * inflation + 0.5387922 * budget surplus - 0.0053373 * trade

To calculate the policy index equations that are given above, an OLS regression on economic growth will be calculated, with all variables included except the variables Aid/GDP,

(Aid/GDP) * policy and (Aid/GDP)2 * policy. The results of this regression for Kenya are given in table 5.

The openness dummy in this policy index equation will be different from the data of Burnside and Dollar (2000), because this data was not available. Instead of the openness dummy of Sachs and Warner, the openness of a country is measured by the trade, what is the sum of the import and export of the country, as percentage of their GDP.

To calculate the budget surplus, the data of the general government revenue as part of total GDP and the general government total expenditure as part of total GDP will be used as follows: (Government revenue as part of GDP – government total expenditure as part of GDP) / 100.

The inflation variable is measured by the average consumer prices and calculated as ln(1+ change in average consumer prices).

19

The coefficients for the budget surplus, inflation and the trade factor that resulted from this OLS regression will be used as coefficients for the policy index. To the policy index equation, an constant will be added. This constant can be interpreted as a country’s predicted growth rate, given its budget surplus, inflation rate, and trade openness, assuming that it had the mean values of all other characteristics (Burnside and Dollar, 2002, pp. 855).

To compute the constant term in the policy index formula, we take the coefficients of the variables initial real GDP per capita, investment/GDP and M2/GDP that resulted from the regression. These coefficients will be multiplied with the average value of the variables. The coefficients of the policy index equation, trade, inflation and the budget surplus, will be set on zero. This will give the following equation:

Constant term policy index = predicted growth rate of GDP per capita = βlgdp * Average ln(Initial real GDP per capita) + βinvestment * Average (Investment / GDP) + βm2 * Average (M2/GDP)

According to the theory of Burnside and Dollar (2000), the coefficients of the budget surplus and the trade openness should be positive. The higher the import and export is of a country, the more open the country is for trade. This has a positive influence on the policy index. Inflation has a negative effect on the policy of a country, so should have a negative coefficient. For the overall policy index it means that the higher the policy index is for a specific year, the better the policy is in that country.

In the case for Kenya (see for the numbers table 5 and 6):

Constant term policy index = 0.3500303 * 0.3456194 – 0.2029404 * 6.088097 + 0.3818028 * 0.1850542 = -1.043889366.

The constant of the policy index formula is -1.043889366. The policy index for each year is calculated as follows for Kenya:

Policy index = -1.043889366 - 0.2103184 * inflation - 0.2152141 * budget surplus + 0.0042255 * trade

20 5. Results

This section will start with a brief background summary for each country. This will make it easier to interpret the results. Second, an overview of the results of each country separately will be given. Next, the main growth equations for all countries will be compared with each other.

Comoros

The population of Comoros in 2015 is 788.000 with a GDP (at market prices, in current US$) of $623.8 million in 2014, which is a GDP per capita of $810.1 (World Bank). Comoros is known as one of the poorest countries in the world (Metz, 1994). It has a poor management of macroeconomic policies (The Heritage Foundation, 2016). According to the Heritage

Foundation this has hindered overall economic development. Since 2009 the trade in Comoros is moderately to mostly free, but the economy is in general still closed. This makes Comoros vulnerable to external shocks. The Heritage Foundation also gives a number for the freedom from corruption. From 60% or higher a country is considered to be free from corruption. Comoros has in 2016 a freedom from corruption percentage of 26, what means that there is a lot of corruption in the country. The Heritage Foundation also mentions that Comoros

depends heavily on foreign aid.

Comoros has a low value of freedom of investment (The Heritage Foundation, 2016). This means that there are different constraints on the flow of investments in Comoros. These constraints can be for example constraints on payments or capital transactions.

21 Table 7. Growth regression Comoros

(1) (2) (3) (4)

VARIABLES growth Growth growth growth

Initial GDP -0.172 -0.192 (0.131) (0.120) M2/GDP (lagged) -0.315 -0.322 -0.361* (0.208) (0.198) (0.211) Policy index 2.360 1.523*** 1.496*** 1.314*** (1.530) (0.408) (0.492) (0.399) Investment/GDP 0.802*** 0.772** 0.890*** 0.358* (0.268) (0.286) (0.296) (0.179) Aid/GDP -5.696 -0.847** -1.018*** -0.679*** (8.538) (0.304) (0.328) (0.202) (Aid/GDP) * policy -3.748 (6.646) (Aid/GDP)2 * policy -1.233** -1.322*** -1.368*** -1.033*** (0.455) (0.388) (0.405) (0.295) Constant 4.145** 3.178*** 2.029*** 1.726*** (1.893) (0.930) (0.646) (0.505) Observations 30 30 30 30 R-squared 0.546 0.539 0.484 0.357

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

To start with the main question, does foreign aid has a positive influence on the GDP growth in Comoros. According to each regression (table 7) foreign aid has a negative influence, although this negative influence is in regression 1 not statistical significant. The least statistical significant variable in regression 1 was the variable (Aid/GDP) * policy. The sign of the coefficient of this variable was also not in line with the theory. Besides that, there is multicollinearity in the model, and with dropping the (Aid/GDP) * policy variable the multicollinearity in the model should decrease.

In regression 2 the variable (Aid/GDP) * policy is dropped. Compared to regression 1, in regression 2 the coefficients of Aid/GDP and the policy index are statistical significant now, which they were not in regression 1. The coefficient of Initial GDP is negative, with is in line with the theory. The higher the initial GDP is, the lower the economic growth of real GDP per capita is. However, this coefficient is not statistical significant different from zero, and will be dropped.

In regression 3, besides the (Aid/GDP) * policy variable, also the initial GDP variable is dropped. Each variable in regression 3 is statistical significant for at least a significance level of 10%. The coefficient of M2/GDP is negative, which is not similar to the findings of

22

Burnside and Dollar (2000). A negative coefficient means the higher the amount of money supply is, the less economic growth there is. The Aid/GDP variable is negative, which means that foreign aid does not increase the economic growth in Comoros. A reason for this can be the corruption in Comoros. Corruption can lead to spending foreign aid in ways that are not efficient, or spending it in ways that it only helps the government. The coefficients of the policy index and investments are both positive and in line with the theory. The better the policy is in Comoros, the higher the economic growth, and the more investment there is in Comoros the higher the economic growth. The coefficient of (Aid/GDP)2 * policy is negative, which is also in line with the theory. This means that the higher the amount of foreign aid Comoros receives, the influence of foreign aid is getting smaller.

To conclude for Comoros, the variables that have a statistical significant impact on the economic growth of the country are M2/GDP, Investment, Policy index, (Aid/GDP) and (Aid/GDP)2 * policy. However, according to this regression analysis, foreign aid has a negative effect on the economic growth for Comoros.

Ghana

The population of Ghana is 27.41 million in 2015 with a GDP (at market prices, in current US$) of $38.62 billion in 2014, which is a GDP per capita of $1441.6 (World Bank). Although the economic expansion in Ghana was slowing, Ghana was able to decrease the poverty in the country (The Heritage Foundation). After a stable average of $375.81 between 1980-2005, in 2006-2007 Ghana’s GDP per capita increased from $501.72 to $929.73. After 2007, it continues increasing, with sometimes a slight decrease (World Bank).The poverty reduction is partly due to relatively sound institutional and legal frameworks (The Heritage Foundation, 2016). But last year there were economic troubles because of a government deficit and a growing public debt (The Heritage Foundation, 2016). The management of public finance was poor and the soundness of fiscal policy has to be restored according to The Heritage Foundation. The freedom of the economy in Ghana is moderately free in 2016. The freedom of trade for the last couple of years was on average moderately free (The Heritage Foundation, 2016). Untill 1996, Ghana was free from corruption. But since then this freedom from corruption decreased, and is repressed. In 2016 their freedom from corruption score is 48%. Above a percentage of 60, a country is considered to be free from corruption. Ghana is not totally free from corruption, there is still some insecurity in economic relationships (The Heritage Foundation, 2016).

23 Table 8. Growth regression Ghana

(1) (2) (3) (4) (5)

VARIABLES Growth Growth growth growth growth

Initial GDP 0.0743 0.0629 0.0759* 0.0483 (0.0745) (0.0575) (0.0421) (0.0441) M2/GDP (lagged) -0.0721 -0.0613 -0.116 (0.112) (0.110) (0.0842) Policy index 0.770 0.819 0.568* 0.505 0.860*** (0.808) (0.737) (0.323) (0.303) (0.193) Investment/GDP -0.0339 (0.0970) Aid/GDP 3.071* 3.101* 1.957** 2.162** 1.797** (1.698) (1.672) (0.929) (0.817) (0.698) (Aid/GDP) * policy -5.879 -6.902 (14.98) (13.88) (Aid/GDP)2 * policy -58.68 -52.73 -67.08** -77.90*** -70.63*** (64.87) (56.91) (30.67) (26.03) (24.75) Constant -0.631* -0.573** -0.597** -0.442* -0.216*** (0.335) (0.253) (0.231) (0.234) (0.0409) Observations 34 34 34 34 34 R-squared 0.653 0.652 0.649 0.640 0.618

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Looking at the first regression (table 8) with all variables included, only the variable (Aid/GDP) is statistical significant. This variable Aid/GDP is positive, which means that foreign aid has a statistical significant positive influence on the economic growth in Ghana. Remarkable is that both the variables (Aid/GDP) and Policy index are positive, but the interaction term of these variables is negative. This negative coefficient means that when the policy in Ghana is better, which means a higher policy index number, foreign aid has a more negative effect on the economic growth, compared to when the policy index is lower.

In the second regression, the variable investment is dropped, because this was the variable that was the least significant different from zero. Because there were still some variables not statistical significant, in regression 3 the variable (Aid/GDP) * policy is also dropped. In regression 3 each variable is statistical significant, besides M2/GDP. However, the variable Initial GDP is positive, which is not in line with the theory. According to the theory this variable should be negative. A positive value means that the higher the initial real GDP per capita was, the higher the economic growth is, but according to the convergence effects, the higher the initial real GDP per capita, the lower the economic growth. According

24

to regression 3 the economic growth of Ghana is depended on the policy. The better the policy is, the higher the economic growth.

To conclude for Ghana, foreign aid has a positive effect on the economic growth.

Kenya

The population of Kenya is 46.05 million in 2015 with a GDP (at market prices, in current US$) of $60.94 billion in 2014, which is a GDP per capita of $1358.3 (World Bank).

Although the economic freedom in Kenya is low, which means that the government does not allows labor, capital, and goods to move freely, the economy of Kenya is one of the most developed in Africa (The Heritage Foundation). Graph 3 shows that overall the trade freedom score has improved since 1995, in 2016 it is moderately free with a score of 65.6%, but the score is not stable, as the graph shows that the score is switching between increasing and decreasing each year.

Graph 3. Trade Freedom Kenya, 1995-2016

The Heritage Foundation, 2016

A problem in Kenya is corruption. Its freedom of corruption number is 25%, and that is the highest since 1998 for Kenya, which means that it has always been a country that faces major corruption problems. Because of this corruption, doing business in Kenya is less attractive than in countries where the freedom of corruption number is higher. Since 2006 the amount of foreign aid Kenya received increased substantially. In 2006 it was $946,7 million, and in 2013 it was about $3,236 billion. Although corruption can decrease the effectiveness of foreign aid, the real GDP per capita is since 2003, besides the year 2008, only increasing.

25 Table 9. Growth regression Kenya

(1) (2) (3) (4)

VARIABLES growth growth growth Growth

Initial GDP -0.267*** -0.271*** -0.270*** -0.251*** (0.0883) (0.0838) (0.0876) (0.0805) M2/GDP (lagged) 0.410*** 0.428*** 0.416*** 0.375*** (0.0927) (0.0703) (0.0771) (0.0773) Policy index 0.461 1.199*** (1.107) (0.320) Investment/GDP 0.402*** 0.417*** 0.375*** 0.373*** (0.109) (0.0971) (0.0984) (0.110) Aid/GDP 12.31 18.87*** 16.70*** 0.230 (14.62) (3.754) (3.817) (0.158) (Aid/GDP) * policy 12.05 18.53*** 15.48*** (14.31) (3.806) (3.477) (Aid/GDP)2 * policy -5.386 -7.767 (6.296) (4.559) Constant 1.906 1.445*** 1.420** 2.589*** (1.150) (0.489) (0.510) (0.626) Observations 31 31 31 31 R-squared 0.643 0.639 0.616 0.630

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

For Kenya, all coefficients for the regression analysis are consistent with the theory (table 9). However, in the first regression the policy index has the least influence on the economic growth in Kenya. Because of this, this variable is dropped in the second regression. The only variable in regression 2 that is not statistical significant (at the 1% level), is (Aid/GDP)2 * policy. Dropping this variable would also decrease the problem of multicollinearity in the model. The final growth equation for Kenya is regression 3. Each variable is statistical significant at the 1% level. The initial real GDP per capita has a negative influence on the economic growth, because the higher the initial real GDP per capita is, the lower the growth of it is. Investment has a positive influence on the economic growth, the higher the amount of investment in Kenya, the higher the economic growth is. Foreign aid has a positive influence on the economic growth of Kenya, and the effect of foreign aid is depended on the policy in Kenya. Is the policy good in Kenya, then the effect of foreign aid on the economic growth will be higher.

To conclude, foreign aid has a statistical significant positive effect on the economic growth in Kenya.

26 Madagascar

The population of Madagascar is 24.24 million in 2015 with a GDP (at market prices, in current US Dollars) of $10.59 billion in 2014, which is a GDP per capita of $449.4 (World Bank). Poverty is still a problem in Madagascar. Also looking at GDP per capita, the poorest country of the five countries I took into account is Madagascar. About 92% of the population in Madagascar is living on less than 2 dollars a day (World Bank). Long-term economic development will be difficult because of political instability and a deteriorating rule of law (The Heritage Foundation). For the last 20 years there is no freedom from corruption. The highest percentage in those years was 34%, and in 2016 it is 28%. However, the last years Madagascar has stabilized and because of this foreign donors have restored ties, that had been broken after the 2009 coup. This coup was announced by Andry Rajoelina, the capital city’s mayor, who accused Marc Ravalomanana, the president of Madagascar, of being a dictator. According to Rajoelina, Ravalomanana did not care about the people in Madagascar (Silva, 2009). Although there is a lot of corruption in Madagascar, the economy of Madagascar is moderately free and their trade is mostly free. Besides corruption, property rights makes it also less attractive to do business in Madagascar (Lindauer, Perkins & Radelet, 2012), which has a negative influence on the economic growth.

27 Table 10. Growth regression Madagascar

(1) (2) (3) (4)

VARIABLES growth growth growth growth

Initial GDP -0.204** -0.203** -0.208** -0.226*** (0.0867) (0.0864) (0.0806) (0.0781) M2/GDP (lagged) -0.730* -0.730* -0.710** -0.721** (0.398) (0.381) (0.326) (0.331) Policy index 0.333 0.323 (1.626) (1.258) Investment/GDP -0.00282 (0.170) Aid/GDP 10.05 10.05 12.97** 9.704 (8.241) (8.047) (6.311) (5.873) (Aid/GDP) * policy 7.893 7.897 10.24* 8.004 (6.659) (6.499) (5.100) (4.847) (Aid/GDP)2 * policy 1.358 1.357 1.534* (0.869) (0.851) (0.780) Constant 1.717 1.703 1.331** 1.461*** (1.951) (1.482) (0.499) (0.483) Observations 34 34 34 34 R-squared 0.393 0.393 0.390 0.365

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

In Madagascar, foreign aid also seems to have a positive effect on the economic growth (table 10). Although, the variable (Aid/GDP) is only statistical significant in regression 3. In each regression the variable M2/GDP is negative, which is not in line with the theory. Also the coefficient of investment/GDP is negative, although according to the theory investment should have a positive impact on the economic growth of a country. However, the variable investment/GDP is dropped in regression 2, because it statistically does not have an effect on the economic growth.

Regression 2 has the same R-squared value as regression 1, which means that for both models the data is replicated by the model even well. Dropping the variable investment/GDP seems to have not influenced the model. All the coefficients are the same, or only slightly changed. For example the coefficient of (Aid/GDP)2 * policy only changed from 1.358 to 1.357. Because dropping the variable investment did only slightly changed the model, the variable that is least statistical significant will also be dropped, the policy index.

In regression 3 the variables investment and policy index are dropped. In this model all variables are statistical significant. The variable (Aid/GDP)2 * policy is positive, however

28

according to the theory this should be negative, because of the diminishing return of foreign aid.

To diminish the multicollinearity in the model, I decided to drop the variable (Aid/GDP)2 * policy in regression 4, however the variables (Aid/GDP) and Aid/GDP)* policy are not significant anymore in regression 4.

To conclude, foreign aid seems to have a positive effect on the economic growth of Madagascar, and this effect is dependent on the policy in the country.

Togo

The population of Togo is 7.305 million in 2015 with a GDP (at market prices, in current US Dollars) of $4.518 billion in 2014, which is a GDP per capita of $635 (World bank). The GDP per capita was stable for a long time, around $350. But since 2008 it is increasing more

rapidly. According to The Heritage Foundation this is partly due to the increased investment in infrastructure. The economy in Togo is mostly unfree, which means that the government does not allows labor, capital, and goods to move totally free. The trade is on average

moderately free. The federal government of Togo implemented an electronic customs system to simplify the importing and exporting process, which should stimulate the trade freedom of the country (The Heritage Foundation). Corruption is a problem in Togo. Till 2006 the

freedom from corruption was as low as 10%. Since then it has increased but it is still as low as 29% (The Heritage Foundation).

29 Table 11. Growth regression Togo

(1) (2) (3) (4) (5) (6)

VARIABLES growth growth growth growth growth growth

Initial GDP -0.316 -0.271 (0.261) (0.221) M2/GDP (lagged) -0.291 -0.293 -0.139 -0.0964 (0.169) (0.175) (0.166) (0.132) Policy index 1.541 0.867* 1.253*** 1.148*** 1.155*** 1.143*** (1.991) (0.475) (0.252) (0.177) (0.192) (0.189) Investment/GDP 0.777** 0.797** 0.578* 0.471* 0.386 0.402* (0.322) (0.329) (0.284) (0.264) (0.240) (0.228) Aid/GDP -2.980 (7.863) (Aid/GDP) * policy -12.64 -5.003 -3.319 -0.404 -0.232 (22.17) (3.695) (3.694) (0.360) (0.309) (Aid/GDP)2 * policy 28.08 23.80 16.02 (26.24) (19.03) (19.25) Constant 2.351 1.811 0.437*** 0.440*** 0.431*** 0.432*** (1.920) (1.085) (0.0808) (0.0829) (0.0767) (0.0745) Observations 25 25 25 25 25 25 R-squared 0.589 0.584 0.554 0.533 0.525 0.519

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

To start with the main question, does foreign aid has a positive effect on the economic growth? In this regression analysis (table 11) foreign aid has a negative effect. However, the coefficient of (Aid/GDP) is 2.980, with a standard error of 7.863, which gives a Tvalue of -0.38. Because of this low T-value the variable (Aid/GDP) is dropped from the model. A variable for foreign aid is still included in regression 2, namely the interaction term of (Aid/GDP) and policy.

The variable (Aid/GDP) * policy is negative, -5.003 in regression 2. This negative value differs from the findings of Burnside and Dollar (2000). According to them this variable should be positive. A negative value implies that, if for example you keep the amount of foreign aid received the same, if the policy is better in a country, the effect of foreign aid is more negative.

In regression 2 most of the variables are not statistical significant. Because of that more variables will be dropped, to improve the significance of variables that seems to have more influence on the economic growth.

30

However, dropping variables does not increase the significance of other variables. In each regression analysis only the variables policy index and investment/GDP are statistical significant. In Togo, if the policy is good in the country, the economic growth is higher. Investments also have a positive influence on the economic growth in Togo. The higher the amount of investment, the higher the economic growth.

To conclude, foreign aid does not have a statistical significant effect on the economic growth in Togo. A reason for this can be the corruption in Togo. As already mentioned above, the freedom of corruption number is not higher as 29%, which means that corruption is a problem in Togo. Corruption can decrease the positive effect of foreign aid.

Comparison

This regression analysis was done for multiple countries separately instead of countries combined. This was done to understand why some factors have an important influence on the economic growth in a country, but have in another country almost no influence. These

different regression analyses should also make it more clear why foreign aid has a positive influence on the economic growth in a country, and a negative influence in another country.

For the countries in this study the results were different. There is no base model were for each country the same variables were statistical significant. Because of that I choose to compare the growth equation with the highest R-squared overall for the countries, namely the first regression. This is the model where all variables are included.

Table 12. Signs of the coefficients Countries / Variables Initial GDP M2/GDP Policy index Investment / GDP Aid/GDP Aid/GDP * policy (Aid/GDP)2 * policy Constant Comoros - - + +*** - - -** +** Ghana + - + - +* - - -* Kenya -*** +** + +** + + - + Madagascar -** -* + - + + + + Togo - - + + - - + + According to the theory - + + + + + - *** p<0.01, ** p<0.05, * p<0.1

In table 12 the signs of the coefficients of the variables are given, and also what signs the coefficients should have according to the theory. Although in this base model most of the variables are not statically significant, the sign of the coefficient did not change for countries separately in the different regression analyses. This makes it more easy to compare the

31

different countries. The country that is most in line with the theory is Kenya, each variable has the right sign according to the theory.

First we look at some remarkable differences for the variables that should have had a negative effect on the economic growth of a country, and next we are looking at the variables that should have a positive coefficient.

Initial real GDP per capita should have a negative sign. This is for each country, except Ghana. A reason for the positive sign for Ghana is maybe the fact that the real GDP per capita has only increased since 1984, and for the other countries there were still some periods of negative growth till 2014.

The variable (Aid/GDP)2 * policy should have a negative coefficient too. This is not the case for Madagascar and Togo.

The policy index should have a positive coefficient, because the better the policy is in a country, the higher the economic growth is. For each country this coefficient is positive, which is in line with the theory. For each country the value of this coefficient is not higher as 1, besides for Comoros. The coefficient has a value of 2.360. Looking at the CPIA score (graph 2), what is a different measure for the policy in a country, Comoros has low CPIA scores compared to the other countries. Because of this low CPIA score, a little increase in the policy in Comoros can increase their economic growth substantially more than in other

countries, where the CPIA score is already on average.

(Aid/GDP) * policy should have a positive coefficients. This is only the case for Kenya and Madagascar. Remarkable is that for Ghana both the coefficient of the policy index and for Aid/GDP is positive, but together in the (Aid/GDP) * policy variable the coefficient is negative.

This section will end with looking at the (Aid/GPD) variable, which is most important for answering the main question, can foreign aid increase the economic growth of a country. If foreign aid has a positive effect on the economic growth of a country, this variable should be positive. This variable is positive for Ghana, Kenya and Madagascar. For Ghana it is even significant at the 10% level in regression 1. In other regression models this variable is also statistical significant for Kenya and Madagascar, as given above. For Comoros and Togo foreign aid has a negative effect on the economic growth. An explanation can be the low CPIA scores (graph 2) for both countries compared to the other countries. Also for both countries the (Aid/GDP) * policy variable is negative, which is in line with the negative value of the coefficient of (Aid/GDP).

To conclude, there are some similarities between the growth equations of the

countries, which are often in line with the theory. But the factors that influence the economic growth of a country can be quite different from each other.

32 Limitations

There have been some limitations in this empirical study. For each country I had maximum 34 observations, and for Togo only 25 observations. The results of the regression analyses would have been more reliable if more observations were used. The reason I choose to not use more observations is because for some variables the data was not available.

Another limitation is one of the OLS assumptions that is violated, namely

multicollinearity. Multicollinearity can limit the research conclusion we can draw from the analysis. In the growth equation in this study the variables aid, policy, aid * policy and aid2 * policy were included. Because of the interaction terms between aid and policy

multicollinearity was a problem in these analyses.

Another OLS assumption that is violated is that data should be independent and identically distributed. In this study a time series is used, where the ordering matters of observations. This means that the data is not independent and identically distributed.

6. Conclusion

In this paper I looked for the effect of foreign aid on the economic growth. In a number of relevant papers, the answers on this subject are different. Some authors found evidence that foreign aid had a positive effect on the economic growth of a country, other a negative effect, or sometimes no effect. I hoped for some clear results in this study, but unfortunately it is still not clear if foreign aid has a positive effect on the economic growth of a country. For three countries, Ghana, Kenya, and Madagascar, foreign aid had a statistically significant positive influence on the economic growth of the country. For Togo and Comoros this effect was negative, however for only Comoros this variable was statistically significant. If foreign aid has an effect, the size of the effect depends on multiple variables, an important variable is the policy in a country.

Because the effect of foreign aid differs between countries, we cannot say something about whether giving foreign aid is a good or bad thing. However, there have been some success stories about foreign aid, which I mentioned in the introduction part. Because of this, I think that giving foreign aid to countries who need the money is a good thing. But it is important to give aid to countries where it will be used efficiently.

7. Discussion

Although it is still not clear if foreign aid has a positive influence on economic growth, the question is also what an increase in economic growth means. Economic growth does not necessarily means that the population of a country becomes wealthier. For a next research, I suggest to study the effect on the economic development, which is a better measure for the wealth of a country. The main purpose of foreign aid is to increase the development of a

33

country. And although economic growth is needed for economic development, it is not sufficient. So economic growth is maybe not the best measure for the effectivity of foreign aid.

The variables that have been included in the regression were according to me most important. However, there are more variables that have influence on the economic growth of a country. Burnside & Dollar (2000) also included ethnic fractionalization and assassinations in their regression for example. The variable to control for geographic location was also

excluded from this research, although it can have effect on the economic growth of a country. For a next research, more variables should be included.

This research was, compared to other research, modest. Only five countries were included with a maximum of 34 data points for each a country. It is difficult to find data for more years, so maybe it is an idea for a next research to do a panel data regression on all countries combined.

34 References

African Development Bank Group, https://cpia.afdb.org/, consulted on 07-06-2016 Alesina, A., & Dollar, D. (2000). Who gives foreign aid to whom and why?. Journal of

economic growth, vol. 5, No. 1, pp. 33-63

Alexander, N. (2009) The World Bank’s Scorecard for Rating Performance of its Recipient Governments. G-24 Policy Brief No. 34

Boone, Peter. (1996). Politics and the Effectiveness of Foreign aid. European Economic Review, 40(2), pp. 289-329

Burnside, Craig and David Dollar. (2000). “Aid, Policies, and Growth.” American Economic Review. September, 90:4, pp. 847–868.

Chauvet, L. & Guillaumont, P. (2001) Aid and performance: a reassessment, The Journal of Development Studies, Vol. 37, Iss. 6, 66-92

Collier, Paul and Dollar, David. (2002) Aid allocation and poverty reduction. European Economic Review, Vol. 46, Iss. 8, pp. 1475-1500

Dalgaard, C.-J., Hansen, H., (2001). On Aid, Growth and Good Policies, The Journal of Development Studies, 37:6, 17-41

Dalgaard, C.-J., Hansen, H. & Tarp, F.. (2004). On the Empirics of Foreign Aid and Growth. The Economic Journal,114(496), F191–F216. Retrieved from http://www.jstor.org/stable/3590060

David, S. (2002). Give and Take: What’s the Matter with Foreign Aid? Zed Books Ltd, New York

Djankov, Simeon, Garcia-Montalvo, Jose and Reynal Querol, Marta (2006) Does Foreign Aid Help?

Dwight Perkins, Steven Radelet and David Lindauer, Economics of Development, 7th ed, 2012

Easterly, William. (2003). Can Foreign Aid Buy Growth? Journal of Economic Perspectives,17(3): 23-48.

Easterly, William. (2007). Was development assistance a mistake? American Economic Review, 97(2), 328– 332. Easterly, W., Le

Ferguson, James. (1994). The Anti-Politics Machine: “Development,” Depoliticization and Bureaucratic Power in Lesotho. Minneapolis: University of Minnesota Press

Griffiths, A., and Wall, S., Applied Economics, 11th ed.

35 OECD, www.oecd.org

The Heritage Foundation, www.heritage.org, consulted on June 15th, 201

Metz, Helen C ed. (1994) Comoros: A Country Study. Washington: GPO for the Library of Congress.

Pedersen, K. (2001). The Samaritan’s Dilemma and the Effectiveness of Development Aid. International Tax and Public Finance. 8, 693-703

Sachs, J. D., and Warner, A. M.. (1995). Economic Reform and the Process of Global Integration. Brookings Papers on Economic Activity. (1), pp. 1-118

Silva, J. (2009) . Mayor Declares a Coup in Madagascar. New York Times, pp. A6 Witteman, J., (2015). China voert steun aan Afrika op: 60 miljard ontwikkelingshulp.

Volksrant

Williamson, Claudia R. (2010). Exploring the failure of foreign aid: The role of incentives and information. The review of Austrian Economics, Vol. 23, No. 1, pp 17-33

Wood, Robert E. (1986). From Marshall Plan to Debt Crisis: Foreign Aid and development Choices in the World Economy. California, United States of America: University of California Press