International Economics and Business Master Thesis HIPC AND EDUCATION EXPENDITURE An Empirical Analysis of the impact of the HIPC debt initiative on education expenditure Sylvester Agyemang

(1)

International Economics and Business

Master Thesis

HIPC AND EDUCATION EXPENDITURE

An Empirical Analysis of the impact of the HIPC debt initiative on education

expenditure

(2)

International Economics and Business Rijksuniversiteit Groningen

(3)

Abstract

This paper examines recent debt relief and its effects on public education expenditure. The focus is on the Heavily Indebted Poor Countries (HIPC) initiative. One of the objectives of the HIPC is to offer eligible countries additional resources for investments in education. The results are compared to other developing countries not classified as HIPC. Controlling for some of the major factors in the literature, this study shows that debt relief has no positive effect on public education expenditure in countries classified as HIPC. It even seems to have a detrimental effect on public education expenditure. This is very surprising and from a policy point of view very disappointing bearing in mind the amount of debt relief HIPC countries receive. On the other hand, Non-HIPC countries do not increase their education expenditure when they receive debt relief.

(4)

If the misery of the poor be caused not by the laws of nature, but by our institutions, great is

our sins. Charles Darwin

Education is the most powerful weapon which we can use to change the world.

(5)

Abbreviations

HIPC = Heavily Indebt Poor Countries IMF = International Monetary Fund OLS = Ordinary Least Squares

BLUE = Best Linear Unbiased Estimator US = United States

NPV = Net Present Value

UNCTAD = United Nations Conference on Trade and Development G7 = Group of Seven

NGO = Non-governmental organization

(6)

Since the early 1980s, there has been an ongoing debate relating to the debt problem of some of the poorest nations in this world. This led to the launch of the Heavily Indebted Poor Countries (HIPC) debt relief initiative by the World Bank and IMF in October 1996. Since the establishment of these two institutions, it was the first time that they both had contemplated on a strategy that induced the cancellation of debts owned to them, proving the increasing severity of the debt issue facing these nations. Countries, like most individuals, have to deal with some kind of debt. The national debt of the US has continued to increase on average by $1.42 billion per day since September 29, 2006.1 Despite this, the US is not considered a low income country, rather a high income country and one of the wealthiest nations on this planet. The main difference between the debt of the US and that of Honduras, i.e. low income countries, is the fact that the US debt is by all means considered sustainable. In other words, it is manageable. The main goal of the HIPC is to help countries achieve debt sustainability by writing off some of the owed debts. Intuitively, this should free-up additional recourses for human capital that should be used in order to reduce poverty in these nations. The main reasoning behind this is very basic. The public, particularly in the more developed western world, reason that one of the main reasons developing countries are lagging behind is due to the large amount of debts they owe. Cancelling these debts, usually accumulated by former dictators and kleptocratic regimes, will provide resources that can be used for needed sectors such as education or health care.2 Human capital includes aspects such as social capital, health and education (Barro, 2001).

The focus of this paper is on education since research has showed that education is a major determinant of economic growth (Barro, 2001).

The severity of debt facing HIPC countries is graphically showed in figures 1 and 2.

1

www.brillig.com/debt_clock

2

(8)

--- Insert figure 1 here --- ---

Insert figure 2 here ---

Public education expenditure (% GDP) has traditionally been lower in HIPC countries vis-à-vis higher income countries. Much more disturbing is the fact that HIPC lags vis a vis other developing countries that are not eligible to receive relief under the HIPC initiative. A comparison of both figures also reveals that HIPC donors spend a greater deal of their export on servicing debt rather than spending it on for example education, healthcare, infrastructure etc. A practice that has been declining since the beginning on the HIPC initiative (1996) thus making it more interesting to find out if the freed up resources were allocated to education or not. Table 1 and 2 gives the absolute values used in figures 1 and 2.

--- Insert table 1 here --- ---

Insert table 2 here ---

Table 3 gives a data profile of HIPC vis-à-vis Non Hipc countries and reveals that where the total debt burden of HIPC countries is decreasing, for Non-HIPC countries, the cost of debt has increased and might even imply a substitution effect

(9)

1.1 Purpose

The purpose of this paper is to identify the effects of the HIPC debt relief programme on public education expenditure in a sample of developing countries.

This paper tends to answer the following:

• Is there a positive relation between debt relief and public education expenditure in HIPC countries?

• Is there a positive relation between debt relief and public education expenditure in non-HIPC countries?

• What other variables are of importance in determining the scope of public education expenditure HIPC and Non HIPC countries?

• What are the possible explanations for any possible difference in HIPC and non-HIPC countries relating debt relief and public education expenditure?

1.2 Importance in the Literature

(10)

1.3 Literature Review

Most studies relating to debt relief and in particular the HIPC initiative deals with the meaning of sustainability3. There has been a lot of debate regarding the topic but not all in this particular setting. This paper builds on several strands of literature. It is mostly closest to the literature on the determinants of public education expenditures. The literature has produced a lot of mixed results regarding the determinants of public education expenditures while using different estimation techniques. These conflicting results pose a hazardous roadblock to policy makers. Most of these studies found that family income, measured by GDP per capita, is the variable that is the most important determinant of public education expenditure (Nord 2001).

Other variables that have consistently shown positive or negative signs are property value and the percentage of school age children within a district. [Barr and Davis (1966), Brazer (1959), McMahon (1970)].

Looking at the publishing dates of these studies, it suggest me to mention that is seems that the literature has remained quite silent relating to the topic. This has also been noticed in a more recent study by Nord (2001) as he concludes that these conflicting results are misleading because of incorrect regression estimates which produce heteroscedastic residuals (Kmenta, 1971, pp. 254-55). In order to remove this barrier, the procedure recommended by Klein (1962, pp. 196-97) and Goldfeld and Quant (1965, pp. 540-43) has to be followed in order to deal with heteroskedasticity. Most of these studies employ Ordinary Least Squares (OLS) regression analysis on cross sectional data. A major problem with these studies is that cross sectional studies cannot control for time variant and invariant effects. These may be controlled for by the use of panel estimation techniques.

A second strand of literature tends to find the effectiveness of debt relief vis-à-vis foreign aid.

3

(11)

While using both Two-stage least squares and OLS, Burnside and Dollar (2000) find that aid raises growth in a good policy environment. A re estimation of the Burnside and Dollar model by Easterly et al. suggests that the authors model is statistically not robust at all. The study by Burnside and Dollar is also sensitive to sample selection and specification and may also suffer from the omitted variable bias (Hepp, 2005). Hansen and Tarp (2001) criticized Burnside and Dollar claiming that the issue of endogeneity has not been properly dealt with within their study. Hansen and Tarp uses the Arellano-Bond generalized method of moments (GMM) estimator to deal with endogeneity. This estimator takes care of the country fixed effects. The authors also find that aid exhibits diminishing returns with respect to growth.

Rajan and Subramanian (2005a) explain in a study the channel through which aid can hurt growth. First, aid inflows may lead to overvalued exchange rates within a flexible exchange rate regime leading to a loss of competitiveness within the tradable sector. Secondly, aid inflows may lead to the phenomenon known as “the Dutch Disease”4. Another implication of large aid inflow is the fact that it does not create an incentive for governments to make improvements regarding tax collecting and propagate a civilization dependent of aid. This makes debt relief a vivid alternative to aid. It does not lead to an overvalued exchange rate while indirectly providing additional resources through a reduction of the debt (service) burden. This means that debt relief increases resources for public spending.

This leads me to another strand of literature, the pro-poor spending (human capital) and its effects on economic growth within an endogenous model. The concept of human capital as described by its originator Arthur Cecil Pigou5, states that it is a means of production. Any additional investment (through ex. education) within it yields additional output. The concept was further promulgated by Theodore Schultz in his ground breaking paper Investment in Human Capital:

4

The theory is that an increase in revenues from natural resources will deindustrialise a nation's economy by raising the exchange rate, which makes the manufacturing sector less competitive.

5

(12)

The Role of Education and of Research in 1971 for which he received the Nobel Prize in economic sciences. Schultz demonstrated that the yield on human capital (better education) was larger than that based on physical capital. There is a widely held belief that education can play a critical role in development. The macro impact of education on economic growth has been researched by economists such as, Barro (1991) and Sen (1999), who views education as an intrinsic good argues that in order for economic growth to be achieved, improvements in education must precede economic reform. Barro (1991) wrote the most informative paper exploring the role of human capital as proxied by educational attainment in explaining cross-country dissimilarity in economic growth rates. Barro finds that educational attainment growth is positively related to a country’s GDP per capita. Barros’ evidence is consistent with the exogenous growth model or Solow growth model and with the endogenous growth model developed by Lucas (1988) emphasizing the importance of human capital externalities. Countries that invest enormous into education such as South Korea grow extremely rapidly and generate much more equal income distributions than other countries that put less emphasis on education. The “Kuznets curve”6 gives a more detailed explanation of the narrowing of income inequality. The empirical support of the theory assigning education expenditures a key role in growth is quite mixed.7 It is, however beyond the scope of this paper to examine if the HIPC initiative has generated growth in the participating countries through an increase in educational expenditures. In a study by Blankenau et al., the authors state that the mixed results have to do with the fact that most empirical investigations fail to control for the method of finance. The authors estimate a growth equation in which the regression accounts for the general adjustment to the taxes levied in support of education. Their model suggests that taxation can alter the positive growth effects from increased public education expenditure. They run a series of regression using OLS while

6

See Van Zanden, J.L. (1995)

7

(13)

controlling for other variables that potentially affect the relationship between growth and education expenditures. Musila et al. (2004) uses time-series technique to show that education expenditure per worker has a positive and significant impact on economic growth both in the long run and short run. Musila et al. uses a panel cointegration8 technique to show that a 1% increase in average education expenditure per worker will increase output by 0.6% in the long run for the case of Uganda. Jung and Thorbecke (2003) apply and calibrate for two HIPC countries a multisector Computable General Equilibrium (CGM) model. The results obtained by the authors suggest that education expenditure can raise economic growth. Even though relations are often found between education and economic growth, it usually entails “mere” correlation. Causality is more often implicit than proved, with ad hoc studies widespread over more robust longitudinal data. Although the studies described above don’t provide significant evidence of causality, they all identify the possible impacts of education on growth which are consistent with the intuitive reasons for their occurrence. Barro (1991), while looking at GDP, education levels, life expectancy, fertility rate, government consumption, rule of law, democracy, terms of trade, and inflation tries to isolate causation of growth from correlation with growth by analyzing the mentioned variables roughly five years prior to a per capita growth measurement. Suher (2004), criticize Barro’s method by stating that a five year lag is not enough to completely separate causation from correlation. More specifically, the author exemplifies that individuals’ expectations of future growth may alter the factors Barro’s studies. A paper by Kiu (2005) proves the existence of Granger causality between human capital as proxied by primary education and economic growth through cointegration. The result shows that primary education is cause of economy while higher is not significant. On the other hand, economic growth is cause of high education other than primary or junior. The above indicates that empirical evidence on the

8

(14)

relationship between education expenditure and economic growth is not conclusive and further empirical research is needed on the topic.

Finally, the pro-poor literature suggests that well targeted public spending might reduce poverty. Investments in education are considered to be pro poor. Although it is not in the scope of this paper to link the effects of debt relief on growth and poverty reduction, it is worthwhile mentioning the fact that there has been facile reasoning linking inputs (public spending) and outcomes (absolute poverty levels). Paternostro et. al (2005) claims that the link between growth and poverty reduction is a false dichotomy that has been falsely taken as a fact. The author responds to the concern about the lack of conceptual foundations and empirical basis for the growth – poverty reduction linkage by proposing a framework combining principles of both public economics and to a lesser extend growth theory.

1.4 Reliability and Validity of Study

It is in my belief that the study conducted here is of a high standard of reliability and validity. The articles and books read and analyzed for this research are all articles that have been written by renowned researches. These articles are thoroughly screened before publication. Data collected for this study is of a high quality since it is gathered from highly reliable sources. A more detailed explanation of the data and its sources is presented in Chapter 3.

1.5 Delimitation

(15)

2 Debt and HIPC

In this chapter, a detailed explanation is given with respect to debt, debt relief and the HIPC programme.

2.1 Debt

Debt is not a new phenomenon; borrowers struggling to repay debt were first recorded the fourth century B.C. when 13 Greek cities defaulted on their loan from the Delos Temple9. Debt is the accumulation of previous borrowing. In development economics, it particular concerns the third world debt. If an individual is not able to pay his/her debts, this individual will be bankrupt. Beyond this “line”, people are not allowed to fall. This line is not applicable for nations, when nations become indebted, they fall into economic ruin. This is what has happened to mostly African and South American nations. These nations have continuously borrowed large amounts at high interest rates hoping that the loans would help them achieve faster growth rates through higher investments. When they did not provide these growth rates, more had to be borrowed in order to fulfil their debt obligations.

2.1.1 Debt Overhang

The concept of debt overhang was originated in a research by Meyers (1977). The author defines debt overhang as: “situations in which a firm’s debt is so large that any earnings generated by new investment projects are entirely appropriated by existing debt holders and hence even projects with a positive net present value cannot reduce the firm’s stock of debt or increase the value of the firm (Myers, 1977).” At the highlight of the debt crisis in the 1980s, this concept was applied to development economics in order to find answers to the debt problems facing

9

(16)

developing nations. Krugmann (1988, 1989) and Sachs (1989) argue that in the presence of debt overhang, a high debt stock acts like a high marginal tax on investment.

Any incentive to invest at home is distorted since any future return will be fully used to repay debts. The concept of debt over hang is represented in a debt laffer curve10.

--- Insert figure 3 here ---

According to Krugman, developing countries find themselves on the wrong side of this curve where larger debt stocks are associated with lower returns and debt repayments. The theory of debt overhang supports an economic rationale for debt relief to countries suffering from debt overhang since dept relief will reduce debt stock and will increase the NPV of debt repayments. The existence of debt overhang has been found in some empirical studies. While using a panel data of 93 developing countries Pattillo et al. (2002) find evidence of debt overhang suggesting that the effect of debt on per capita growth becomes negative at a debt stock of 35-40 percent of GDP.

This result is confirmed by Clements et al. (2003). The authors also find evidence that debt service has a crowding out effect on public investment. Chowdhury (2001) uses extreme bound analysis to find evidence that debt service and debt stock both have a negative effect on growth in his sample set. The author distinguishes between HIPC countries and other severely indebted countries not eligible for HIPC assistance. The focus of that paper is whether the set of HIPC countries should be extended. His results suggest the existence of debt overhang in both set of countries which could be “removed” by debt relief. He concludes that not only HIPC countries suffers from debt overhang, the set of countries eligible should be extended beyond its present scope. The existence of debt overhang in low income countries is questioned by several authors. [Kletzer and Wright (2000), Cordella et al. (2002), Bulow and Rogoff (1988, 1989)].

10

(17)

Bird and Milne (2003) suggest that low income countries do not suffer from debt overhang since the transfer of official assistance exceeds their debt service repayments. The authors argue that this removes the distortionary incentives for domestic investments.

Kletzer and Wright (2000) argue that debt beyond the maximum point of a countries willingness to pay (debt overhang) is irrelevant. Cordella et al. (2002) additions that where traditional view only encompass the present value of debt stock, debt service is of greater importance than debt stock. They present a model where only debt service raises welfare and whereas debt stock does not.

2.1.2 Debt Relief

Debt relief as defined by the World Bank is the partial or total reduction of debt owed by individuals, corporations or nations. Despite the recent attention debt relief is getting, it is not a new concept. In 1967, UNCTAD declared that “debt-service payments have risen to the point at which a number of countries face a critical situation11.

After a series of official UNCTAD meetings during 1977-1979, official creditors wrote off USD 6 billion debt to 45 countries. After this, The World Bank Africa reports of 1981, 1984, 1986, and 1991 repeatedly called for debt relief for particularly African nations south of the Sahara. The G7 summits of 1988, 1989 and 1991 also called for additional debt relief which eventually led to concessional lending and more debt relief. Pre-HIPC, debtor countries had to deal with the Paris Club12 and the London Club13 in order to renegotiate official debts and commercial debts. In the mid 1980s, the then US secretary of Treasury, Nicolas Brady, launched a debt relief plan in order to lengthen maturity, reduce interest payments and forgive

11

Quoted from Easterly (2002, p. 1678)

12

the group of wealthy donor nations which also belong to the (OECD).

13

(18)

principal interest for developing nations. The Brady plan at that time was the largest private sector debt relief effort ever & plan was triggered by the debt crisis that started on august 12, 1982 when Mexico announced that is was not capable of fulfilling its debt obligations.

(19)

2.2 HIPC

(20)

fast. An increase in US interest rates caused a hike in interest rates and coupled with a further increase in oil prices caused developing countries to fall in a trap. They were earning less on exports and paying more on their loans and imports. More money had to been borrowed in order to even pay off the interest alone. These countries became indebted and the trap was set.

2.2.1 Goal HIPC

The HIPC initiative is aimed at reducing the net present value of debt to what the creditors (bilateral and multilateral) deem a more manageable level: a maximum of 150 percent of the value of countries annual exports and 250 percent of their government revenue.

The amount of debt relief under HIPC is different for each HIPC country but aims to bring the debt to a sustainable level. In the words of the World Bank, “it is to ensure deep, broad and fast debt relief and thereby contribute toward growth, poverty reduction, and debt sustainability in the poorest, most heavily indebted countries.” The rationale of the initiative is to fund debt relief for all eligible countries by reducing external debts owed by HIPC governments.

2.2.2 HIPC and Education

Education per se is not a key factor for the creditors in granting debt relief under the HIPC program but eligible countries have to develop a Poverty Reduction Strategy Paper (PRSP) that identifies “pro-poor” budget expenditure and show how they would use the funds released by the debt relief to mitigate poverty. A thorough review of the various PRSP reveals that all of the HIPC countries list health care and education as “pro poor”14. It is therefore reasonable to hypothesize that debt relief under the HIPC program will lead to an increase in education expenditure in the eligible countries.

14

(21)

2.2.3 Recipient of HIPC assistance

In order to be eligible for HIPC assistance, a country has to have unsustainable debt levels, be only eligible for assistance from the IDA. Once a country has proven it has made sufficient progress in fulfilling these criteria’s, The IMF and IDA decide whether it is eligible for debt relief or not. This is called the decision point. After a country has reached this decision point, it begins receiving interim relief. When a country receives full debt relief, it has reached its completion point. At the end of March 2007, there were a total of 40 countries that qualified for or were eligible to receive HIPC assistance.

(22)

2.2.4 Finance of HIPC

The total cost of providing debt relief assistance to the 40 HIPC countries is estimated to be about US$64 billion by the end of 2005 net present value terms. The amount is provided by both bilateral and multilateral creditors. The share of the cost provided by the IMF is primarily financed by the return on investments from off market gold sales in 1999. These returns were deposited in the IMF’s PRGF-HIPC Trust. Member countries also provided additional contributions to this trust.

2.2.5 Critique of HIPC

In the academic world, there are advocates and opponents with respect to the HIPC. Economist Jeffry Sachs is known for his Jubilee 2008 campaign to have the debt of developing countries dropped completely. In his 2005 book The End of poverty, Professor Sachs argues that

industrialised countries should have learn from the historical failure of German war reparations after World War I and the success of the grand system of the Marshall Plan after World War II. Sachs mentions that loans given to developing countries should have been grants rather than loans. In his book, Sachs also mentions 8 factors that in his point of view explains a nations lack of economic growth, Poverty Trap; Physical Geography; Fiscal Traps; Governance Failures; Cultural Barriers; Geopolitics; Lack of Innovation and Demographic Trap. As outlined in the summary, “Debt cancellation cannot be for a lark or a whim. It must not be a game to avoid past obligations. Debt cancellation must reflect true social, economic, and political realities. Under those circumstances, a negotiated cancellation of debt can give new hope and new economic opportunities to the debtor country” Sachs (2005).

(23)

panaceas is for growth alongside population control, investment in machinery and education. The strongest point Easterly makes is the economic principle that ‘people respond to incentives’ Dempski et. al. (1984). Easterly concludes that debt relief fails to provide incentives for developing nations governments to change their irresponsible behaviour. While providing statistics, Easterly argues that debt relief will because corrupt government will abuse the loan system for their own benefit and frustrate growth.

(24)

3 Empirical Model and Data

In order to find answers to the research questions, it is important to construct an econometric model in order to clearly understand the testing procedure. In this section, I describe the econometric model used in order to uncover the effects of debt relief on public education expenditure. Regression analyses were performed using the statistical programme EVIEWS. In order to conduct this study within a proper econometric framework, I have chosen to conduct a panel data analysis with country fixed effect making it possible to analyze a particular subject within multiple sites, periodically observed over a defined time frame Yaffee (2003). The combination of time series with cross-sections can enhance the quality and quantity of data in ways that would be impossible using only one of these two dimensions (Gujarati, 638). The panel data covers 1998 to 2004 and the analysis holds constant (fix) the average effect of every country thus reducing the threat of omitted variable bias.

3.1 Empirical Model

The main question being addressed is whether debt relief has a positive effect on public education expenditure. I divide the sample in two different groups: the first group includes countries defined as HIPC and the second group is defined as Non-HIPC countries. The econometric model created in order to realize the purpose of this study follows the methodology proposed by Hepp (2005). Hepp suggests an econometric model in finding the effects of debt relief on health expenditure. This model used is also explained in Yaffee (2003). Since this research follows the same subject, a similar model is used and is constructed as follows:

(25)

Where Yit is the dependent variable observed for country i at time t, Xit are the variables of

interest, Zit are the control variables, θi is a country fixed effect, λt is a time fixed effect and εit is

the error term. Yit is total public education expenditure (PEDex) observed for country i at time t.

Since the paper is not intended to add to the growth theory literature, PEDex is used instead of education attainment as suggested by Barro (1991). My interest is in how recent debt relief initiatives have altered education expenditures since the relieved debt is at least publicly guaranteed. PEDex includes only public expenditure on education and it is therefore not necessary to include a variable measuring public education expenditure as a share of total education expenditure in order to control for a possible shift in education expenditure from private to public or vice versa15.

Public expenditure on education consists of current and capital public expenditure on education plus subsidies to private education at the primary, secondary, and tertiary levels. The variables of interest, Xit, are guided by previous literature and also the OECD list of indicators for education

used to track the progress made in the Millennium Development Goals16. These include

debt relief, net enrolment ratio in primary education, literacy rate of 15-24(both sexes), aid per capita and debt service.

Essential to this study is the measurement used to capture debt relief. Two different variables are mentioned by Hepp (2005) to capture debt relief. The author makes a distinction between debt relief captured from the World Bank data set and from the OECD’s International Development Statistics data set. The World Bank has two different variables to measure debt relief. These data are derived from the World Bank data set and are called debt service relief and debt stock relief. According to Hepp(2005), the World Bank data are the ones that come closest to capturing the

15

See Leu (1986)

16

(26)

definition of debt relief as it is mentioned in the theoretical literature. Due to data availability, the debt relief variable used in this research is debt stock relief rather than debt service relief.

3.2 Data Collected

The following data was collected for years 1998 – 2004. • Public Education Expenditure

This consists of current and capital public expenditure on education plus subsidies to private education at the primary, secondary, and tertiary levels (EdStats).

• Debt Stock Relief

This is the change in debt stock due to debt forgiveness or reduction. It is derived by subtracting debt forgiven and debt stock reduction from debt buyback (World Bank, 2005a).

• Net enrolment ratio in primary education

(27)

• Literacy rate of 15-24 (both sexes)

The percentage of people ages 15-24 who can, with understanding, read and write a short, simple statement on their everyday life (WorldBank, 2005a).

• Debt Service

The sum of principal repayments and interest actually paid in foreign currency, goods, or services on long-term debt, interest paid on short-term debt and repayments (repurchases and charges) to the International Monetary Fund (IMF). Both long-term public and private debts are included (World Bank, 2005a).

• Debt Stock

This consist of public and publicly guaranteed long-term debt, private nonguaranteed long-term debt (whether reported or estimated by the staff of the World Bank), the use of IMF credit, and estimated short-term debt (World Bank, 2005a).

• Aid per capita

Aid per capita includes both official development assistance (ODA) and official aid, and is calculated by dividing total aid by the midyear population estimate (OECD, 2004).

• GDP capita

The annual per person output of an economy (World Bank, 2005a).

• Institutional quality.

(28)

3.3 Reliability and Validity Data

The reliability of the secondary data collected is believed to be of a very high quality.

The dataset used is an unbalanced panel of 137 developing countries over the period 1998-2004. The data set is restricted to this time frame due to data availability. These countries are classified by the United Nations as being low-income, lower middle-income and upper middle-income. These countries are further subdivided into 39 HIPC and 98 Non-HIPC.

(29)

3.4 Econometric Problems and Solutions

This research uses both time series and cross sectional data. The use of these kinds of data can give some problems. Two of the most common econometric problems will be discussed in this chapter. These complications are autocorrelation and heteroskedasticity.

3.4.1 Autocorrelation

Autocorrelation exists when the current error term contains not only the effects of current shocks, but also the carryover from previous shocks. This so called carryover will be correlated with the effects of earlier shocks. When these circumstances lead to error terms that are correlated, autocorrelation exist. The possibility of autocorrelation should therefore be always considered when dealing with time-series data (Hill, Griffiths and Judge p. 258). The authors also suggest using the Durban-Watson test to test for the existence of autocorrelation. The result of this test is explained in section 4.1.

3.4.2 Heteroskedasticity

(30)

4 Empirical Findings and Result

This section will present the findings of this study and the results obtained from the econometric model mentioned section 3. Before showing the final result, tests were conducted in order to establish the existence or non existence of both autocorrelation and heterosketdasticity.

4.1 Autocorrelation

Hill, Griffiths and Judge mention that a Durbin-Watson statistic of approximately 2 is taken as an indication that the models errors are not autocorrelated. A low Durbin-Watson statistic is an indication of the existence of autocorrelation. The hypotheses considered in the Durbin-Watson test are:

H0 : p = 0

(2) H1 : p > 0

Upper (dU) and lower (dL) critical values have to be found in order to test the null hypothesis of

no autocorrelation in the errors against the alternative hypothesis of positive autocorrelation the p-value. The following decision rules, known as the bounds test, were suggested by Durbin and Watson.

If d < dL reject H0 : p = 0

If d > dU do not reject H0 : p = 0 (3) If dL < d < dU test is inconclusive.

A visual inspection of the regression result in appendix II yields the following, in the HIPC regression, a Durbin-Watson statistic (d) of 2.47 is found and for the Non-HIPC regression, this statistic is 2.56.

For the HIPC, dL = 0.950, du= 2.018 and d = 2.47. Out of these, the second decision rule in

(31)

regards to the Non-HIPC data, dL = 0.950, du= 2.018 and d = 2.56. It is therefore also impossible

to reject the null hypothesis and concludes that also this data set does not exhibit autocorrelation.

4.2 Heteroskedasticity

Detecting heteroskedasticity is a case of whether using residuals plots or a statistical test known as the Goldfeld-Quant test or the White’s test (Hill, Griffiths and Judge p. 237). The existence of different variances is often encountered when using cross-sectional data (Hill, Griffiths and Judge p. 237). In the current data sample, any existence of heterosketdasticity will be unknown thus requiring the use of a technique demonstrated by Harold White in 1980. This White covariance estimator, is available in EVIEWS and is used in the regression to minimize heterosketdasticity. According to Bernhard et. al. (2005), the White covariance estimator is resistance to the heterosketdasticity problem.

4.3 Descriptive analysis

In order to better understand the data and differences between HIPC and Non-HIPC countries, a descriptive analysis has been performed and reported in tables 4 and 5.

--- Insert table 4 here --- ---

(32)

(33)

4.4 Regression results

In this section, the result of the regression analysis will be presented. The focus of this paper is whether debt relief, had a positive effect on public education expenditure. Two regressions were conducted distinguishing between HIPC and Non-HIPC. The regressions were conducted while using fixed effect estimation and White’s covariance estimator.

The results for the HIPC countries are presented in table 6. ---

(34)

literacy rate has a decreasing effect on public education expenditure (in % of GDP). An increase in these two ratios might be seen by HIPC governments as an indication that enough resources has been diverted to this sector and thus reducing the investments in education and diverting it to other sectors that might have falling behind such as health care or infrastructure. Both debt service and institutional quality have no significant effects in HIPC countries. GDP is negatively significant. Higher GDP levels may induce HIPC countries to cut public education expenditure with the belief that the population is able to provide the funds needed to send their children to private schools. Debts stock is also significant and has the expected negative sign for HIPC countries. An increase in debt stock of one percent will lead to a decrease in public education expenditure (in % of GDP). Aid in HIPC countries also contributes to higher public education expenditures.

With respect to Non-HIPC countries, the results are presented in table 7. ---

(35)

(36)

5 Conclusion

A fundamental goal of the HIPC debt relief initiative is to help poor countries move towards macroeconomic stability. The western donor community has been heavily debating whether the debt relief granted to HIPC’s has been used in a proper way in order to help eradicate poverty. To shed some light in this highly important issue, in this paper, I investigate the effects of debt relief granted through the HIPC initiative on public education expenditure between 1998 and 2004. The results are also compared with developing countries not classified as HIPC. I find that debt relief for countries classified as HIPC and non-HIPC did not positively affect public education expenditure significantly. It even seems to have a detrimental effect on public education expenditure in HIPC countries. This detrimental effect has to be taken with cautious since it is only negatively significant at the 10% significant level. There are a few possible explanations for this detrimental effect. First, there may still be debt overhang after the relief of debt. It is therefore hard to increase education expenditure under this circumstance. This lends support to Cordella et. al (2005). The authors suggest that more debt relief is required for HIPC countries since the debt facing them is too large. The authors also suggest using debt service relief rather than debt stock relief. A third explanation could be the existence of aid fungibility. Aid destined for education expenditure could replace government expenditure on education. In a panel data study by Feyzioglu, T. et. al (1998), the authors find that developing-country governments receiving earmarked concessionary loans for agriculture, education, and energy reduce their own resources going to these sectors and use them elsewhere. Debt relief itself may also be fungible.

(37)

(38)

References

Anderson, S., (2006), The Debt Boomerang 2005: How Americans Would Benefit from

Cancellation of Impoverished Countries Debts. The Institute for Policy Studies discussion paper march 2006.

Arslanalp, S. and Henry, P.B., (2005), Is Debt Relief Efficient? Journal of Finance, Vol. 60, No. 2, 1018-1051

Auboin, M., (2004), The Trade, debt and Finance Nexus: at the Cross-road of Micro- and

Macroeconomics. WTO Discussion paper No. 6, WTO Publications.

Barr, J.L., and Davis, O.A., (1966), "A Political and Economic Theory of the

Expenditures of Local Governments". Southern Economic Journal, Vol. 33 ,149-165

Barro, J.R., (1991), Economic Growth in a Cross Section of Countries. Quarterly Journal of Economics, v. 106, no. 2, pp. 407-43

Barro, J.R., (1998), Education as a Determinant of Economic Growth: A Cross-Country

Empirical Study. Journal of Comparative Economics, Vol. 26, No. 4, 822-824 Barro, J.R. and Jong-Wha, L., (2000). International Data on Educational Attainment:

Update and Implications. CID Working paper no. 42, Harvard University, US

Belassi, W., and Musila, J.W., (2004), The Impact of Education Expenditures on Economic

Growth in Uganda: Evidence from Time Series Data. The Journal of Developing Areas, Vol. 38, No. 1, 123-133

Belloc, M. and Vertova, P., (2004), How does Public Investment Affect Economic Growth in

HIPC? An Empirical Assessment. Working paper no. 416, University of Siena, Italy Bernhard, L., Philipp J.S. and John W., (2005), Volatility Modelling – From ARMA to

ARCH. Paper presented at the Global Finance Center, France.

(39)

Goals. In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt Relief, Myths and Reality: 97-108. The Hague: FONDAD

Blackmore, S. and Greenhill. R., (2002), Relief Works: African proposals for debt

cancellation – and why debt relief works. Paper presented at the new Economics Foundation, London.

Blankenau, W.F. and Simpson, N.B. (2004), Public Education Expenditures and Growth. Journal of Development Economics, Vol. 73, No. 2, 583-605

Boote, A, R. and Thugge K., (1997). Debt Relief for Low-Income Countries and the

HIPC Initiative. IMF Working Paper 97/24, Washington, DC.

Bird, G. and Milne, A., (2003), Debt Relief for Low Income Countries: Is It Effective and Efficient? World Economy, v. 26, no. 1, pp. 43-59

Brazer, H. E., (1959). City Expenditures in the United States. New York: National Bureau of Economic Research.

Bullow, J. and Rogoff, K., (1988). Sovereign Debt: Is To Forgive To Forget? National Bureau of Economic Research, Inc, NBER Working Papers: 2623

Burnside, C. and Dollar, D. (2002), Aid, Policies and Growth. American Economic Review, v. 90, iss. 4, pp. 847-68

Business Monitor International. 2006. Debt Relief Funds to Help Reduce Poverty. Vol. 11, No. 1, January 2006

Chowdhury, A. and Verbina, I., (2004), What Determines Public Education Expenditures in

Russia? Economics of Transition, Vol. 12, No. 3, 489-508

Clements et. al., (2003), External Debt, Public Investment, and Growth in Low-Income Countries. International Monetary Fund, IMF Working Papers: 03/249

(40)

Centre Technical Paper No.166, September.

Cordella, T. and Dell’Ariccia, G., (2002), Limits of Conditionality in Poverty Reduction Programs. International Monetary Fund, IMF Working Papers: 02/115

Demski, J., and D. Sappington (1984), “Optimal Incentive Contracts with Multiple Agents,” Journal of Economic Theory, 33, 152-171.

Dijkstra, G., (2004), Debt Relief from a Donor Perspective: The Case of the Netherlands. In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt Relief, Myths and Reality: 109-131. The Hague: FONDAD

Easterly, W (2005), How did highly indebted poor countries become highly indebted after two decades of debt relief efforts? World Bank Working paper No. 2225.

Fernandez, R. and Rogerson, R., (1997), The Determinants of Public Education

Expenditures: Evidence from the States, 1950 – 1990.

NBER working paper no. W5955

Feyzioglu T., Swaroop, V. and Zhu M. (1998), A Panel Data Analysis of the Fungibility of

Foreign Aid, The World Bank Economic Review, Vol. 12, No. 1, 29-58

Gilman, M. and Mitchell, W., (2004), Achievements to Date and Challenges Ahead: A View from the IMF. In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt Relief,

Myths and Reality: 72-96. The Hague: FONDAD

Goldfeld, S.M. and Quandt, R.E., (1965), Some Tests for Homoskedasticity. Journal of the American Statistical Association 60, 539–547

Hansen, H. and Tarp, F., (2001), Aid and Growth Regressions. Journal of Development Economics, April 2001, v. 64, no. 2, pp. 547-70

(41)

University of California, Santa Cruz, CA.

Hepp, R., (2005), Can Debt Relief Buy Growth? Working paper, University of California, Santa Cruz, California.

Hjertholm, P., (1999), Analytical History of Heavily Indebted Poor Country (HIPC) Debt

Sustainability Targets. Paper presented at the joint World Bank/Nordic Working Seminar, Santa Cruz, California.

Hsiao, C., (2005), Why Panel Data? Institute of Economic Policy Research working paper no.05.33, University of Southern California, CA.

Independent Evaluation Group. 2006. Debt Relief for the Poorest: An Evaluation Update of

the HIPC Initiative. April 11, 2006.

Jubilee USA Network. The Debt Relief Obstacle Course: Understanding The Heavily

Indebted Poor Country Initiative.

Jung, H.S., and 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 Modelling, Vol. 25, No. 8, 701-725

Kamenta, J. and Gilbert, R.F. (1970), Estimation of Seemingly Unrelated Regressions with Autoregressive Disturbances. Journal of the American Statistical Association, March 1970, v. 65, no. 329, pp. 186-97

Kao, C et. al. (1999), On the Estimation and Inference of a Cointegrated Regression in Panel Data. Working paper no.2. Center for Policy Research, Maxwell School, Syracuse University.

Kletzer, K.M. and Wright, B.D., (2000), Sovereign Debt as Intertemporal Barter. American Economic Review, v. 90, no. 3, pp. 621-39

(42)

research working Paper Series 3200.

Krugman, P. (1988), Financing vs. Forgiving a Debt Overhang. Journal of Development Economics, v. 29, no. 3, pp. 253-68

Kuteesa, N. and Nabbumba, R (2004), HIPC Debt Relief and Poverty Reduction Strategies: Uganda’s Experience. In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt Relief,

Myths and Reality: 48-56. The Hague: FONDAD

Martin, M., (2004), Assessing the HIPC Initiative: The Key Policy Debates.

In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt Relief, Myths and Reality: 11-47 The Hague: FONDAD

Maruping, M., (2004), Lessons from Eastern and Southern Africa. In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt Relief, Myths and Reality: 57-71.

The Hague: FONDAD

McMahon, W.W., (1970), An Economic Analysis of Major Determinants of Expenditures on Public Education. Review of Economics and Statistics, August 1970, v. 52,

no. 3, pp. 242-52

Meyers, S. (1977), Determinants of Corporate Borrowing, Journal of Financial Economics, 5, 147-75

Morrissey, O., (2002), Making Debt Relief Conditionality Pro-Poor. UNU-WIDER Research Paper DP2002/04

Musila, J.W. and Belassi, W., (2004), The Impact of Education Expenditures on Economic Growth in Uganda: Evidence from Time Series Data. Journal of Developing Areas, v. 38, no. 1, pp. 123-33

(43)

Osei, R. and Quartey, P., (2001), The HIPC Initiative and Poverty Reduction in Ghana. WIDER Discussion Paper No. 2001/119, United Nations University.

Paternostro, P., Rajaram, A. and Tiongson, E.R., (2005), How Does the Composition of

Public Spending Matters? World Bank policy research working paper no. 3555, World Bank, Washington, DC

Pattillo, C., Poirson, H. and Ricci, L., (2002), External Debt and Growth. IMF Working Paper WP/02/69, International Monetary Fund, Washington, DC.

Rajan, R. and Subramanian, A.,(2005), Aid and Growth: What Does the Cross-Country Evidence Really Show? International Monetary Fund, IMF Working Papers: 05/127 Sachs, J.D., (1989), The Debt Overhang of Developing Countries. World Institute for

Development Economics Research. Debt, stabilization and development: Essays in memory of Carlos Diaz-Alejandro, 1989, pp. 80-102

Sachs, J.D., (2002), Resolving the Debt Crisis of Low-Income Countries. Brookings Papers on Economic Activity, Vol. 2002, No. 1, 257-286

Sachs, J.D. and Warner, A.M., (1997), Sources of Slow Growth in African Economics. Journal of African Economics, Vol. 6, No. 3 335-376

Sen, A., (1999), Economic Policy and Equity: An Overview. Economic policy and equity, 1999, pp. 28-43

Telatin, M., Thomas, B. and Pettifor, A. (2001), HIPC – Flogging a dead process: The need

for a new, independent and just debt work-out for the poorest countries. Jubilee Plus Research Report, London.

Teunissen, J.J., (2004), Introduction. In J.J Teunissen, and A. Akkerman (Eds.), HIPC Debt

Relief, Myths and Reality: 1-10. The Hague: FONDAD

(44)

Tanzania. WIDER Discussion Paper No. 2001/53, United Nations University. United Nations. 2005. The Millennium Development Goals Report 2005. May DPI/2390 United Nations. 2006. The Millennium Development Goals Report 2006. June 2006 World Development Movement. 2003. Debt and the G8. June 2003.

(45)

Books

Cooper, D.R. and Schindler, P.S., (2003), Business Research Methods, 8th edition, McGraw Hill Easterly, W. (2001), The Elusive Quest for Growth, 2nd edition, MIT, USA

Gujarati, D., (2003), Basic Econometrics, 4th edition, New York: McGraw Hill, pp. 638 - 640. Hill, C.R., Griffiths, W.I. and Judge, G.G., (2001), Undergraduate Econometrics, 2nd edition,

John Wiley & Sons Inc., USA

Mankiw H.G., (2003), Macroeconomics, 5th edition, Worth Publishers, USA

Klein L.R., (1962), An introduction to Econometrics, Englewood Cliffs, New York: Prentice Hall, USA

Sachs. J (2005), The End of Poverty, 1st edition, MIT, USA

(46)

(47)

Appendix 1

List of Sample Countries

HIPC Non-HIPC

Benin Albania Algeria

Bolivia Angola Argentina

Burkina Faso Armenia Azerbaijan

Burundi Bangladesh Barbados

Cameroon Belarus Belize

Central African Republic Bhutan Bosnia

Chad Botswana Brazil

Comoros Bulgaria Cambodia

Congo Cape Verde Chile

Congo, Dem. Rep. China Colombia

Côte d'Ivoire Costa Rica Croatia

Eritrea Djibouti Dominica

Ethiopia Dominica Republic Ecuador

Gambia Egypt El Salvador

Ghana Equatorial-Guinea Estonia

Guinea Fiji Gabon

Guinea-Bissau Georgia Grenada

Guyana Guatemala Hungary

Haiti India Indonesia

Honduras Iran Jamaica

Kyrgyz Rep. Jordan Kazakhstan

Liberia Kenya Laos

Madagascar Latvia Lebanon

Malawi Lesotho Lithuania

Mali Macedonia Malaysia

Mauritania Maldives Mauritius

Mozambique Mexico Micronesia Moldavia

Nepal Mongolia Morocco Namibia

Nicaragua Nigeria Oman Pakistan

Niger Palau Panama Papua Guinea

Rwanda Paraguay Peru Poland

São Tomé Príncipe Philippines Romania Russia

Senegal Samoa Serbia Seychelles

Sierra Leone Slovakia Solomon Isl. South Africa

Sudan Sri Lanka Saint Lucia Saint Vincent

Tanzania Surinam Swaziland Syria

Togo Tajikistan Thailand Timor-Leste

Uganda Tonga Trinidad Tunisia

Zambia Turkey Turkmenistan Ukraine

Uruguay Uzbekistan Vanuatu

(48)

Total debt servicing (% of Export) 2 7 12 17 22 27 32 71 74 77 80 83 86 89 92 95 98 01 04 HIPC Non-Hipc Appendix 2

Figure 1. Public education as percentage of GDP

(Constructed on the basis of data from the World Bank education statistics, EDStats)

Figure 2. Total debt service as percentage of Export

(Constructed on the basis of data from the World Banks Global Development Finance, GDF 2005) Public education expenditure

(49)

Figure 3. The Debt “Laffer curve”

(50)

Appendix 3

Table 1 Public education expenditure (% GDP)

1991 1998 1999 2000 2001 2002 2003 2004 2005 HIPC 2.80 3.40 3.40 3.20 3.70 3.60 4.10 3.70 3.80 Non-HIPC 4.50 4.20 4.45 4.35 4.50 4.50 4.80 4.75 4.60 Developed 4.80 4.85 4.90 5.05 5.35 5.45 5.90 4.50 5.10

Table 2 Total debt service (% of export)

1991 1998 1999 2000 2001 2002 2003 2004 2005

HIPC 20.9 18.6 15.7 12.9 10.3 8.5 8.75 7.1 7

Non-HIPC 12.6 9.1 10.25 11.45 10.02 11.8 12.3 11.7 10.07

(51)

Table 4 Descriptive Statistics HIPC Countries

Variable Observations Mean Std. Dev. Minimum Maximum

Public Education

Expenditure * 114 3.42 1.42 1.00 8.00

Debt Stock Relief * 263 2.99 8.48 0.00 66.49

Net enrollment Ratio

Primary Education 172 62.30 18.82 24.00 99.00 Literacy rate of 15-24 165 66.11 19.08 21.70 99.70 Debt Service * 273 3.47 3.27 0.00 18.00 Debt Stock * 273 140.30 120.59 20.00 699.00 AID ** 267 48.90 42.29 2.61 253.35 GDP ** 273 376.45 251.75 84.00 1118.00 Institutional Quality 108 3.35 3.57 1.40 4.90 Notes: * = as a percentage of GDP ** = in per capita terms

Table 5 Descriptive Statistics Non-HIPC Countries

Variable Observations Mean Std. Dev. Minimum Maximum

Public Education

Expenditure * 342 5.19 4.87 0.60 16.50

Debt Stock Relief * 675 1.21 17.41 -0.63 63.00

Net enrollment Ratio

(52)

Appendix 4

Table 6. Regression HIPC Countries. Dependent variable: Public Education Expenditure Fixed Effects

Debt Stock Relief -0.017501*

(-2.056545) Net Enrolment Ratio Primary Education -0.045984***

(-4.795840) Literacy rate of 15-24 -0.027922*** (-4.889276) Debt Service 0.035884 (1.767303) Debt Stock -0.007178*** (-4.547946) AID 0.004532** (3.407216) GDP -0.002261** (-2.548133) Institutional Quality 0.121388 (0.524231) Observations 31 Adjusted R-squared 0.98 Durban-Watson stat. 2.47

(53)

Appendix 5

Table 7. Regression Non-HIPC Countries. Dependent variable: Public Education Expenditure Fixed Effects

Debt Stock Relief 0.020784

(1.027377)

Net Enrolment Ratio Primary Education 0.019941

(1.102517) Literacy rate of 15-24 -0.046171* (-1.696675) Debt Service 0.046901* (1.734748) Debt Stock -0.005194*** (-3.411156) AID 0.021611* (1.661064) GDP 3.78E-05 (0.397275) Institutional Quality 0.469921** (1.999447) Observations 136 Adjusted R-squared 0.90 Durban-Watson stat. 2.56

(54)

Appendix 6

Table 8. Summary of selected Empirical Studies: Debt relief

Author Period Country (ies) Analytical Method Summary of Findings Kuteesa and

Nabbumba 1991-200 Uganda Descriptive analysis

The additional resources from the HIPC initiative have been instrumental in

reducing poverty but have not resulted in debt sustainability Hepp 1998-2001 Developing

countries Panel analysis

Debt relief has surprisingly no effect on health

expenditures in HIPC countries.

Greenhill and

Blackmore 1998-2002 HIPC Countries Descriptive analysis

Debt relief has a clear impact on government budgets.

Cohen 1965-1987 Developing

countries Investment equations

High debt has a negative impact on growth for Latin America countries. Pattillo et al. 1969-1998 93 developing

countries

Panel regressions (specifications with debt dummies, a quadratic specification, and a spline function)

(55)

Appendix 7

Table 9. Summary of selected Empirical Studies: Public Expenditure and Poverty Author Period Country (ies) Analytical Method Summary of Findings Dabla-Norris

and Matovu 1996 Ghana Dynamic CGE

Increasing primary and secondary education has significant macroeconomic benefits, even if these come at the expense of infrastructure investment.

Jung and

Thorbecke 1995 Zambia CGE

Well-targeted education expenditures can be effective for poverty alleviation. To maximize benefits, education spending needs to be complemented by sufficient public investment.

Dollar and

Kraay 1950-1999 Cross country (92 countries) Regression Analysis (system of equations) Overall government spending is negatively related to poverty. Health and education spending are insignificant.

Musila et. al

1965-1999 Uganda Error Correction Model and cointegration estimates

Education expenditure per worker has a positive and significant impact on

economic growth both in the long run and short run

Barro

1960-1995 100 countries OLS

International Economics and Business Master Thesis HIPC AND EDUCATION EXPENDITURE An Empirical Analysis of the impact of the HIPC debt initiative on education expenditure Sylvester Agyemang