1
Development Aid and Income inequality:
The importance of the quality of the
institutional environment
MSc Thesis International Economics and Business
Author: Melanie Hekwolter of Hekhuis Student ID number: 1768867
E-mail: m.s.hekwolter.of.hekhuis@student.rug.nl Date Thesis: 08-07-2013
2 Content
Abstract ... 3
1. Introduction ... 4
2. Literature review ... 5
2.1. Determinants of income inequality ... 5
2.2. Aid and inequality ... 7
2.3. The role of the institutional environment ... 8
3. Data and Methodology ... 12
3.1. Methodology ... 12
3.1.1. Cross-section approach ... 12
3.1.2. Panel data approach ... 13
3.2. Data on dependent and independent variables ... 14
3.3. Data on control variables ... 16
3.4. Sample ... 17
4. Empirical results ... 17
4.1. Cross-section approach ... 17
4.1.1. Democracy model ... 17
4.1.2. Robustness checks Democracy model ... 21
4.1.3. Corruption model ... 23
4.1.4. Robustness check Corruption model ... 25
4.2. Panel data approach ... 25
4.2.1. Democracy model ... 25
4.2.2. Corruption model ... 27
4.3. Discussion of the results ... 27
5. Conclusions ... 29
References ... 31
3
Abstract
This MSc thesis explores the influence of official development assistance (ODA) on income inequality. It is argued that the quality of the institutional environment in the ODA-receiving countries, proxied by the level of democracy and corruption, is of crucial importance in the relationship between ODA and income inequality. While the direct effect of institutional quality on income inequality is found to be zero to positive, it is confirmed that ODA only leads to a lower income inequality in the presence of a relatively high level of democracy. Hence, ODA is only efficiently spent if it is donated to relatively democratic countries.
4
1 | Introduction
Numerous rich countries donate large sums of money to developing countries with the objective of reducing poverty in those countries. However, the effectiveness of this development aid is a topic of debate. Many scholars have researched the effect of official development aid on economic growth. However, aid might also have other effects, for example an effect on income inequality. In contrast to the large literature on the growth effects of development aid (Svensson, 1999; Burnside & Dollar, 2000; Dalgaard & Hansen, 2001; Hansen & Tarp, 2001; Lensink & White, 2001; and others), the distributional effects of aid have received only limited attention in previous empirical research. In the few papers that research these effects, the existence of a clear research gap is acknowledged: the effect of foreign aid on income inequality is a research area that has remained largely untouched (Chong, Gradstein, & Calderon, 2009; Layton & Nielson, 2009). This is surprising, since the most important goal of development aid is to eradicate
extreme poverty and hunger. The exact target is to halve, between 1990 and 2015, the percentage of people living on less than $1 a day (UN General Assembly, 2000). Income inequality
complements economic growth as an indicator to measure the progress in reaching this target, since it is possible that the proceeds of economic growth are mainly reaped by the a priori richer part of the country, leaving the percentage of people living on less than $1 a day unaltered. Since a more equal distribution of income within a country would mean that less people suffer from extreme poverty, it is desirable to examine the effect of aid on income inequality as well. In this research, I will attempt to bridge the research gap by concentrating on the relationship between development aid and income inequality. Bridging this gap is of crucial importance, since this might lead to new insights in how aid can be allocated most effectively. The focus will be on the effect of official development assistance (ODA) on income inequality within aid-receiving countries.
5 It is argued that institutions are of crucial importance in the relationship between ODA and
income inequality. Whether or not the government is able to allocate development aid correctly depends on the quality of governance. Corruption and democracy can be seen as proxies for the quality of institutions in a country. Therefore, the effect of corruption, as well as the effect of the level of democracy will be examined. I argue that, rather than a direct effect, the main effect of aid is that it exacerbates the impact of institutions on inequality.
This thesis is organized as follows. In the next section, the existing literature in the field of income inequality and development aid will be reviewed. Furthermore, the theoretical model and hypotheses are outlined. In section III the methodology and data are presented. The empirical results follow in section IV and section V concludes.
2 | Literature review
2.1. Determinants of income inequality
A broad literature has tried to estimate the determinants of income inequality (Ahluwalia, 1976; Chong et al., 2009; Kuznets, 1955; Li, Squire, & Zhou, 1998; and others). Perhaps the best-known determinant of income inequality is economic development. As early as in 1955, Kuznets pointed out that the relationship between economic development and income inequality is
represented by an inverted U-shaped curve. In early levels of development, income inequality increases as GDP per capita increases. When higher levels of development are reached, income inequality decreases with an increase in GDP per capita. According to Kuznets (1955), this pattern is caused by the dynamics of the dual economy.
6 In an exploratory study, Ahluwalia (1976) finds that except the population shifts between sectors, more widespread education and a lower population growth can explain part of the inverted U-shaped curve found by Kuznets (1955). In particular, these processes seem to explain part of the decrease in income inequality that is taking place in later stages of development, while they cannot explain the noticeable increase in income inequality in earlier stages of development (Ahluwalia, 1976).
Income inequality does not only differ between the agricultural and the modern sector, it also differs within these sectors. Kuznets (1955) posited that inequality within the agricultural sector is less than inequality within the modern sector, as the agricultural sector consists of enterprises of relatively similar size, while the modern sector is composed of economic units of very
different sizes (Nielsen, 1994). The percentage of the labor force in agriculture is thus expected to have a negative influence on inequality, as this variable reflects a greater weight of the more equal sector in the overall income distribution. Nielsen (1994) indeed finds a significant negative effect of labor force in agriculture on inequality. Rural economies can therefore be expected to exhibit lower income inequality than urban economies.
Furthermore, it is established in the literature that demographics play a role in determining income inequality. In particular, countries associated with high birth rates, and therefore younger populations, tend to experience higher income inequality (Simpson, 1990; Shafiullah, 2011). High population growth generates inequality by expanding the proportion of the national
7 In addition, the income distribution is affected by the level of human capital accumulation of the population. If improvements in the educational characteristics and skill endowments of the labour force take place, the income distribution tends to equalize. This is the case because human capital accumulation inevitably means diffusion of education across a larger part of the population, since there are limits to how much human capital one person can accumulate (Ahluwalia, 1976).
Moreover, more widespread education across the population dilutes the value of education for the elite and lowers the inequality in the level of education attained (De Gregorio & Lee, 2002). A more educated population has the possibility to exert influence on the policy-making process. This constrains the capacity of the rich minority to lobby for policy that will lead to increased income inequality (Li et al., 1998). The income equalizing effect of educational expansion is repeatedly found to be very significant (Bourguignon & Morrisson, 1990; Koechlin & León, 2007; Nielsen & Alderson, 1995; Simpson, 1990; and others).
2.2. Aid and inequality
Only a few papers have researched the effects of development aid on income inequality. Layton and Nielson (2009) find that foreign aid has a zero to weakly positive effect on income
inequality, as increases in foreign aid are still related to limited increases in income inequality after controlling for all other factors. Bjørnskov (2010) finds that in foreign aid-receiving
democracies, a relatively larger share of total income is held by the elite. However, other scholars have found that foreign aid improves the distribution of income. Chong et al. (2009) e.g. find that development aid is linked to decreased income inequality in countries displaying a low level of corruption, although this effect is not robust. Shafiullah (2011) concludes that foreign aid has a small, but robust, equalizing effect. Furthermore, he discovers that aid received in the previous year influences inequality in the current year more than current aid does.
8 resource curse can also originate from the receipt of development aid. Therefore, it can be stated that the resource curse literature is dealing with a similar problem.
One of the most renowned problems arising after the discovery of natural resources is the
appreciation of the real exchange rate, leading to a phenomenon called “Dutch Disease.” The rise in the value of natural resource exports leads to an appreciation of the exchange rate, which makes other exports less competitive. Because of low-cost imports, domestic producers face fiercer competition in the local market. Furthermore, prices of local labor and assets increase, since they are needed to process the natural resources. The combination of these effects lead to a situation in which the natural resource and nontradeable sectors are in advantage, while the traditional exports (manufacturing, agriculture) are crowded out (Morrison, 2012). In a similar vein, the receipt of aid can also lead to the occurrence of Dutch Disease. The inflow of
development aid pushes up the nominal exchange rate, which dilutes the competitiveness of the traded goods sector if wages do not decline (Rajan & Subramanian, 2005). The economic decline caused by the Dutch Disease augments income inequality, since resource rents are mainly
allocated to the rich while benefits from economic growth are spread more evenly (Lam & Wantchekon, 2002). Hence, according to the resource curse literature, foreign aid can lead to an increase in income inequality because of the occurrence of the Dutch Disease.
The review of the literature on the direct effect of aid on inequality shows that no consensus on either the existence or direction of the effect has been reached. The effect of development aid on income inequality is therefore less clear than the effects of the variables mentioned in the
previous subsection. Because of this, no a priori hypothesis on the direct effect of aid can be given. In fact, instead of a direct effect, an indirect effect of aid might be more important. This indirect effect might be linked to the role of institutions, which I will discuss next.
2.3. The role of the institutional environment
9 their behaviors increasingly towards the needs of the poor, which constitute a majority in
developing countries (Reuveny & Li, 2003). Timmons (2010) reinforces this political
competition argument. When politicians have to compete for the support of voters, more public services are produced for any given level of revenue. If the poor benefit more than proportionally from government operations such as public education, democracy should cause a decrease in income inequality. Hence, it is expected that the prevalence of a higher level of democracy would lead to a lower level of income inequality. Reuveny and Li (2003), Li et al. (1998) and Gradstein & Milanovic (2004) indeed find evidence for this claim, although some other authors conclude that the level of political democracy as measured at one point in time is not generally linked to lower levels of income inequality (Sirowy & Inkeles, 1990).
A second institutional variable that is found to determine income inequality is corruption. Many scholars have found that the prevalence of corruption increases income inequality. Gupta,
Davoodi and Alonso-Terme (2002) state that corruption increases poverty and income inequality since it biases tax systems towards the rich and well-connected. Furthermore, because of
corruption social programs are being targeted poorly and resources are used to lobby government for policies that perpetuate inequality in asset ownership. Mauro (1998) finds that corruption reduces government spending on education. Because of unequal access to education, income inequality is worsened (Gupta et al., 2002). Hence, corruption has a direct effect on income inequality.
In this thesis, the institutional variables mentioned above will be used to proxy institutional quality. The level of corruption and democracy prevalent in a country will be suitable indicators, as earlier literature showed that these two institutions have a profound effect on the functioning of the government (e.g. Reuveny & Li, 2003; Mauro, 1998). In fact, it is likely that the quality of the institutional environment is problematic when a high level of corruption prevails or when the discretion of the government is indefinite, since the allocation of resources and the delivery of services will be less than optimal (Chong & Calderon, 2000).
10 development assistance flows constitute an increase in the funds available to the government for redistribution. As explained earlier, reelection-oriented democratic leaders adjust their behaviors towards the needs of the poor, as they are held responsible by the voters (Reuveny & Li, 2003; Timmons, 2010). Hence, democratic leaders are inclined to distribute the official development assistance flows equally, just like they are inclined to distribute regular government funds equally. A democratic government will therefore invest ODA in projects that will increase the welfare of the poor. This leads to a reduction in income inequality within their country. In contrast, authoritarian leaders are only accountable to a rich minority (Reuveny & Li, 2003). Owing to this, authoritarian leaders will most likely transfer the development aid to projects or policies benefiting this minority. These projects and policies will increase income inequality. Kosack (2003) finds that aid donated to a democracy can lead to improvements in life quality. The author argues that aid will be effective at improving life quality if it causes an augmentation of the resources assigned to projects which raise life quality. In addition, it is found that aid is not effective in autocracies.
The second proxy of institutional quality is corruption. As elaborated on earlier, corruption increases income inequality because social programs are being targeted poorly (Gupta et al., 2002). Official development aid receipts can be seen as government revenues. If corruption is prevalent, these assistance receipts will therefore be targeted as poorly as any other government revenues. In the absence of corruption, however, ODA will be targeted properly. After executing OLS estimations with cross-country data, Chong et al. (2009) state that aid appears to be related to lower income inequality in countries exhibiting a low level of corruption.
The resource curse literature has discovered the importance of institutions as well. Indeed, it is notified that Dutch Disease does not always bring negative consequences after new resources are found or when aid is received. In fact, whether or not these occur depends on how the
11 Yet, the Dutch Disease effect is not the only problem originating from the resource curse that can be prevented by good institutions. Mehlum, Moene and Torvik (2006) conclude that the quality of institutions is decisive for the resource curse to occur at all. The authors notice that resource abundant countries with “producer friendly” institutions can benefit from these resources, while resource abundant countries with “grabber friendly” institutions (e.g. a high level of corruption) experience low growth. This difference is caused by the fact that resource abundance attracts entrepreneurs into successful production in high quality institutional environments, while resource abundance attracts entrepreneurs into unproductive enterprises in low quality
institutional environments. Moreover, Robinson, Torvik and Verdier (2006) state that countries with institutions that promote the accountability and capacity of the state will profit from resource abundance, as the negative political incentives created by resource abundance are mitigated. In contrast, countries without good institutions are likely to be hit by the resource curse.
To conclude, whether the government allocates ODA flows effectively depends on the quality of institutions. This institutional quality can be proxied by the level of democracy and the level of corruption prevalent in a country. These considerations lead to the following hypotheses: Hypothesis 1: In a high quality institutional environment, proxied by the level of democracy, Official Development Assistance flows influence income inequality negatively.
Hypothesis 2: In a low quality institutional environment, proxied by the level of democracy, Official Development Assistance flows do not influence income inequality negatively. Hypothesis 3: In a high quality institutional environment, proxied by the level of corruption, Official Development Assistance flows influence income inequality negatively.
12
3 | Data and Methodology
3.1. Methodology
3.1.1. Cross-section approach
The relationship between ODA, quality of institutions and income inequality will first be analyzed by performing a cross-section ordinary least squares (OLS) regression analysis using the statistical software package Stata IC 12.
Since the values of the level of democracy change slowly and it is to be expected that the influence of the independent variables is only noticeable after a period of multiple years, it is examined whether the averages of the level of democracy and the corresponding interaction variable and control variables over the years 1990-2000 influence the average level of inequality in 2000-2011. For the Corruption model, I use the same approach as above, but because of the more limited data on the level of corruption, the independent variables are averaged from 2002-2006 only. Then, the influence on average income inequality over the years 2007-2011 is examined.
The considerations above lead to the following models: Democracy model:
Income inequalityi = αi + β1‘Net ODA received’i + β2‘Level of Democracy’i + β3 ‘Level of
Democracy*Net ODA Received’i + β4‘Youth Population’i + β5‘Agriculture, Value Added’i +
β6‘Years of Schooling’i + β7‘GDP per Capita’i + β8‘GDP per Capita Squared’i + εi
In which Income inequalityi is the average income inequality in country i from 2001-2011, αi is a
constant, ‘Net ODA received’i is the average official development assistance received from
1990-2000, ‘Level of Democracy’i is the average level of democracy over the period 1990-2000, and
‘Level of Democracy*Net ODA Received’i is an interaction variable of the average level of
democracy and the average amount of official development assistance received. ‘Youth
Population’i , Agriculture, Value Added’i, ‘Years of Schooling’i, ‘GDP per Capita’i and ‘GDP per
Capita Squared’i are control variables, of which the average over the years 1990-2000 is
13 Corruption model:
Income inequalityi = αi + β1‘Net ODA received’i + β2‘Corruption’i + β3 ‘Corruption*Net ODA
Received’i + β4‘Youth Population’i + β5‘Agriculture, Value Added’i + β6‘Years of Schooling’i +
β7‘GDP per Capita’i + β8‘GDP per Capita Squared’i + εi
In which Income inequalityi is the average income inequality in country i from 2007-2011, αi is a
constant, ‘Net ODA received’i is the average official development assistance received from
2002-2006, ‘Corruption’i is the average level of corruption over the period 2002-2006,
‘Corruption*Net ODA Received’i is an interaction variable of the average level of corruption and
the average amount of official development assistance received. ‘Youth Population’i ,
‘Agriculture, Value Added’i , ‘Years of Schooling’i , ‘GDP per Capita’i and ‘GDP per Capita
Squared’i are control variables, of which the average over the years 2002-2006 is included.
3.1.2. Panel data approach
The behavior of the different countries over time can be described by panel data models. Data on multiple countries and multiple years allow me to control for country differences, as well as to study adjustment over time (Hill, Griffiths, & Lim, 2008). Therefore, the cross-section analysis will be complemented by a panel data approach. However, the cross-country models remain preferred, as the quality of the institutional environment within countries only changes slowly over time. Most of the variation is expected to be cross-country. Hence, the panel data models will only pick up the time trend.
With respect to the Democracy model, I again assume that the democracy and ODA need sufficient time to influence income inequality. Therefore, 10-year lagged values of these variables are used in the specification. Data is over the years 1990-2011.
Limited data availability on the variable ‘Corruption’ restricts the dataset used for the Corruption model to 2002-2011. Five-year lagged values of ‘Net ODA Received’, ‘Corruption’ and the interaction variable are used to account for the lagged impact of corruption and ODA on income inequality.
14 The considerations above lead to the following panel data models:
Democracy model:
Income inequalityit = αi + β1‘Net ODA received’it + β2‘Level of Democracy’it + β3 ‘Level of
Democracy*Net ODA Received’it + β4‘Youth Population’it + β5‘Agriculture, Value Added’it +
β6‘GDP per Capita’it + β7‘GDP per Capita Squared’it + εit
Corruption model:
Income inequalityit = αi + β1‘Net ODA received’it + β2‘Corruption’it + β3 ‘Corruption*Net ODA
Received’it + β4‘Youth Population’it + β5‘Agriculture, Value Added’it + β6‘GDP per Capita’it +
β7‘GDP per Capita Squared’it + εit
I use fixed effects models to control for omitted country-specific factors. Breusch-Pagan tests show that individual heterogeneity is present (p=0.000) in both the corruption model and the democracy model, while Hausman tests indicate that in both cases the random effects model is not appropriate (p=0.000). Hence, the fixed effects model is the model that fits the data best. Moreover, results with robust standard errors will be presented. The robust standard errors help to correct for heteroskedasticity and autocorrelation. Autocorrelation is expected as the variables only change slowly over time.
3.2. Data on dependent and independent variables
Data on the dependent variable income inequality is obtained from the World Bank. In particular, the variable ‘Gini index’ will be used as a measure of the level of income inequality. The Gini coefficient is the most widely used measure of income inequality in research within this area. It is derived from the renowned Lorenz curve. The Lorenz curve depicts the relationship between the cumulative percentage of a population and the cumulative percentage of income belonging to the same population. In the case of a completely egalitarian income distribution, the Gini coefficient is 0, while it approaches 1 if the total income of a country is acquired by only one individual (UNU-WIDER, 2008b).
15 starts in the year 1960. However, many country-year observations are missing. Another widely used database on Gini coefficients is the UNU-WIDER database. This database contains Gini coefficients of 159 countries or areas, but the most recent Gini coefficients are from 2006. Moreover, in this database many country-year observations are lacking as well (UNU-WIDER, 2008a). Therefore, it is chosen to use the more recent database, i.e. the database of the World Bank (2013).
The main independent variables are the interaction variable consisting of the level of democracy and the amount of ODA received, and the interaction variable consisting of the level of
corruption and the amount of ODA received. The effects of the level of democracy, the level of corruption and the amount of ODA received are also examined in isolation.
The data utilized to measure the amount of ODA received by the developing countries is from the World Bank (2013). This database contains an indicator called “Net ODA received (% of GNI)”. It comprises official financial flows which promote the economic development and welfare of developing countries. The ODA flows consist of grants and concessional loans by government agencies to developing countries in the Development Assistance Committee (DAC) list of ODA recipients (ibid.). This data is available for an extensive amount of countries and from 1960 onwards.
16 The Polity concept simultaneously examines autocratic and democratic authority qualities of governments (CSP, 2013). Among the qualities of democratic and autocratic authority included in the measure are competitiveness of executive recruitment, openness of executive recruitment and regulation of participation. The combination of these measures leads to a range of possible outcomes. The “Polity2 Score” ranges from -10 (hereditary monarchy) to +10 (consolidated democracy). In case of a transition, the change in the scores before and after the transition is prorated across the span of the transition. This enables time-series analysis (Jaggers & Marshall, 2012b).
The extent of corruption is measured by the Corruption Perception Index (CPI) from Transparency International (2012). The CPI measure is used in prominent papers relating
corruption to income inequality (e.g. Gupta et al., 2002) and is the most widely used indicator of corruption worldwide (Transparency International, 2012). The Corruption Perceptions Index assigns scores to countries on the basis of expert views on how corrupt the public sector of the country under consideration exactly is. The views of these experts are bundled to arrive at one composite score ranging from 0.0 (highest level of corruption) to 10.0 (no corruption) (ibid.). The database of Transparency International contains information on a large number of states.
However, it does not have a large time span, as for most countries data only is available from 2001 onwards.
3.3. Data on control variables
As indicated in the literature review, certain variables have been shown to influence income inequality within a country. These are GDP per capita, youth population, human capital
17 Furthermore, the amount of human capital accumulation is described by the indicator “Barro-Lee: Average years of secondary schooling, age 15+, total” which is retrieved from the World Bank Education statistics database (2013). This database contains data on 144 countries from 1970-2010.
3.4. Sample
Due to differences in data availability for the variables ‘Corruption’ and ‘Level of Democracy’, two different samples will be used for the two different models. The need for data on ‘Income inequality’ and ‘Level of Democracy’ limits the sample to 79 countries. If the full specification including all control variables is used, data is available for 67 countries. The countries included in the sample are printed in appendix 1. When the panel data approach is used, the sample is
reduced to 65 countries. Jamaica and Niger are no longer part of the sample in this case. Limited data on the variable ‘Corruption’ reduces the sample used to estimate the Corruption model to 49 countries for the cross-section model, and 47 countries for the panel data model. The list with countries included in the sample of the Corruption model can be found in appendix 2.
4 | Empirical results
4.1. Cross-section approach
4.1.1. Democracy model
Summary statistics are in table 1. Moreover, the correlation coefficients for all variables are reported in appendix 3. Almost all of the correlation coefficients are well below 0.8, which is the common threshold value for multicollinearity. Only the variables ‘GDP per Capita’ and ‘GDP per Capita Squared’ are highly correlated. As the latter variable is just the squared version of the former, this is completely logical.
18
Table 1: Summary Statistics Democracy model
Variable Obs Mean Std. Dev. Min Max
Cross-country data
Income inequality 81 43.17 8.67 29.83 65.77
Net ODA Received 81 8.05 9.50 0.03 50.75
Level of Democracy 79 1.93 5.54 –9.27 10.00
Net ODA * Democracy 79 2.66 48.14 –178.55 185.02
Years of Schooling 70 1.75 1.23 0.09 5.52
Youth Population 80 38.67 7.34 18.32 50.96
Agriculture, Value Added 78 23.97 14.29 3.81 57.66
GDP per Capita 81 1524.59 1802.51 117.15 8429.85
GDP per Capita squared 81 5.533e+6 1.23e+7 1.372e+4 7.11e+7
Panel data
Income inequality 488 45.11 9.17 26.82 74.33
Net ODA Received 776 7.77 10.42 –0.47 94.95
Level of Democracy 797 2.10 6.16 –10.00 10.00
Net ODA * Democracy 776 2.95 76.97 –569.68 432.36
Youth Population 1474 36.24 8.16 13.89 53.02
Agriculture, Value Added 1439 20.34 13.05 2.27 61.81
GDP per Capita 1449 1,776.77 2,076.25 122.09 13,836.19
GDP per Capita squared 1449 7.465e+6 1.79e+7 1.491e+4 1.910e+8
The panel data variables ‘Net ODA Received’, ‘Level of Democracy’ and ‘Net ODA * Democracy’ are 10-year lags of the original variables.
19
Table 2: Regression results cross-country Democracy model
Income inequality (1) (2) (3) (4)
Level of Democracy 0.525*** 0.514*** 0.806***
(0.172) (0.177) (0.219)
Net ODA Received 0.126 0.194
(0.134) (0.117)
Net ODA * Democracy –0.044**
(0.019)
Years of Schooling –0.89 –0.776 –0.706 –0.458
(0.677) (0.583) (0.613) (0.615)
Youth Population 0.445** 0.609*** 0.564*** 0.559***
(0.196) (0.153) (0.163) (0.150)
Agriculture, Value Added –0.104 –0.112 –0.142 –0.209*
(0.110) (0.101) (0.107) (0.119)
GDP per Capita 0.006*** 0.006*** 0.006*** 0.005***
(0.002) (0.002) (0.002) (0.002)
GDP per Capita squared –6.88e – 07*** –6.61e – 07*** –7.06e – 07*** –6.59e – 07***
(2.10e – 07) (1.94e – 07) (2.05e – 07) (2.06e – 07)
Constant 23.949*** 17.363** 18.435** 19.863***
(9.057) (7.310) (7.411) (7.065)
Observations 68 67 67 67
R2 0.353 0.477 0.485 0.518
* Significant at 10% ** Significant at 5% *** Significant at 1%. Robust standard errors in parentheses. Independent and control variables are averaged over the years 1990-2000, income inequality is averaged over the years 2001-2011.
However, this effect is not significant at conventional significance levels. This means that ODA in the aggregate does not influence income inequality. Nevertheless, the interaction variable between ‘Net ODA received’ and ‘Level of Democracy’ has a negative influence on the level of income inequality (p=0.021). This indicates that ODA, donated to countries with a relatively high level of democracy, is indeed related to a decreased level of income inequality in those countries. Combined with the insignificant and positive effect of ‘Net ODA received’, this means that ODA is associated with lower income inequality if and only if it is donated to countries that are
20 However, this effect is not caused by an aggravation of the income-equalizing effect of a high level of democracy. As expected, the level of democracy significantly influences the level of income inequality occurring later (p=0.001). However, the effect is counter-intuitive.
Surprisingly, the level of democracy seems to increase the level of income inequality in the long term, instead of having the expected lowering direct effect on income inequality. This effect is visible throughout all regressions, i.e. even when ‘Net ODA received’ and the interaction between ‘Net ODA received’ and ‘Democracy’ are not included. Hence, countries with a higher level of democracy in 1990-2000, tend to experience a higher level of income inequality over the period 2001-2011.
To examine whether this effect is due to the influence of less democratic but socialist countries included in the sample, I perform two regressions in which a dummy variable indicating whether countries are considered to be socialist or not is included. In the first regression (appendix 4, column 1), countries which are considered to be socialist at the moment of writing receive the value “1”. All other countries receive the value “0”. While the dummy variable has a negative sign and is significant at a 10% level, the positive sign of ‘Level of Democracy’ remains significant (p=0.000). In the second regression (appendix 4, column 2), all countries that were considered to be socialist until at least 1990 receive the value “1”. The dummy variable is now significant at a 5% level, but the effect of the level of democracy has not changed (p=0.000). Hence, the occurrence of less democratic socialist countries in the sample cannot explain the unexpected positive effect of the level of democracy on income inequality.
Still, the sample used in this research might be the explanation for this remarkable result, although with a different reason. Chong and Calderon (2000) find that institutional quality displays a significant and robust quadratic relationship with income inequality. Hence, they find that the quality of the institutional environment is positively associated with income inequality in poor countries, while it is negatively related to income inequality in richer countries. Since the sample employed in this research obviously principally includes poor countries, the positive effect of the level of democracy on income inequality is consistent with the results found by Chong and Calderon (2000).
21 induced by an increased level of democracy, such as improved tax collection, lead to additional costs for workers in the underground economy. This sector commonly represents a large fraction of the less-developed economies, and the people working in this sector are among the poorest inhabitants of the country. Therefore, improvements in institutional quality might lead to even lower incomes for the poorest, resulting in higher income inequality. While the informal sector suffers, the formal sector benefits. Consequently, after the initial gains disappear and the informal economy has adapted to the new situation, income inequality starts to decrease (ibid.).
Considering the positive effect of a higher level of democracy on income inequality, the result that ODA donated to democracies is related to decreased income inequality seems surprising. However, it is evident that ODA needs good institutions before it can be spent efficiently. Hence, in countries exhibiting lower institutional quality, as proxied by a lower level of democracy, ODA will not make a difference, while in countries displaying higher institutional quality ODA exercises an income equalizing effect.
The control variables ‘Youth Population’, ‘GDP per Capita’ and ‘GDP per Capita Squared’ are highly significant and influence income inequality as expected. Countries facing a larger share of youth population experience more income inequality in the long term (p=0.000). Furthermore, ‘Agriculture, Value Added’ is significant at a 10% level, indicating that, typically, an agrarian economy exhibits less income inequality than an urban economy. Years of schooling seem to matter less. The variable has the expected negative sign (more years of schooling are associated with less income inequality) but it is insignificant (p=0.459).
4.1.2. Robustness checks Democracy model
The results above might be biased, if the amount of aid donated is influenced by the quality of institutions or by the level of income inequality in the aid-receiving country. To test whether this is the case, a regression with ‘Net ODA received’ as the dependent variable, ‘Level of
Democracy’ as the independent variable and ‘GDP per Capita’, ‘GDP per Capita Squared’, ‘Income inequality’, ‘Youth Population’, ‘Agriculture, Value Added’ and ‘Years of Schooling’ as control variables is carried out (appendix 5). All variables are averaged over the years 1990-2000. The R2 is 0.54 and most importantly, ‘Level of Democracy’ and ‘Income inequality’ are
insignificant. In fact, the p-value associated with ‘Level of Democracy’ is even 0.932.
22 low (-0.3011), while the correlation between ‘Net ODA received’ and ‘Income inequality’ is even lower (-0.1049). This entails that the average quality of institutions (as proxied by the level of democracy) and the average income inequality in 1990-2000 do not influence the amount of ODA received over this period. Hence, the results above are not biased due to any influence of the quality of institutions or the level of income inequality on the amount of ODA received. Furthermore, it is verified whether the choice of the years included in the regression does not bias the results. In the base model, the independent variables are averaged over the years 1990-2000. To check for robustness, I verify whether the results are similar when 5-year averages are used. As can be seen in appendix 6 (column 1), the result that ODA donated to democracies is related to lower income inequality holds, albeit it is now only significant at a 10% significance level. This might reflect the fact that a considerable amount of time is needed before the effect of ODA on income inequality is noticeable. Furthermore, some of the control variables have become insignificant. Most probably, the independent variables need ample time before effects on income inequality are observable as well.
Next, a change of the start year is applied (appendix 6, column 2). To rule out any influence of the turbulence many countries encountered in the year 1990, a regression with starting year 1991 is executed. The results are qualitatively the same.
23
Table 3: Summary Statistics Corruption model
Variable Obs Mean Std. Dev. Min Max
Cross-country data
Income inequality 57 43.04 8.89 29.75 65.77
Net ODA Received 81 6.18 7.16 –0.13 31.45
Corruption 79 3.04 1.02 1.54 7.38
Net ODA * Corruption 79 16.27 18.92 –0.44 82.00
Years of Schooling 70 2.15 1.35 0.16 5.68
Youth Population 80 35.05 8.38 14.52 48.81
Agriculture, Value Added 81 19.42 12.82 2.72 57.28
GDP per Capita 80 1,786.79 2,127.14 130.57 11,545.74
GDP per Capita squared 80 7.661e+6 1.85e+7 1.705e+4 1.33e+8
Panel data
Income inequality 257 44.55 9.03 28.15 67.40
Net ODA Received 331 5.99 7.65 –0.69 55.09
Corruption 295 3.20 1.12 1.20 7.50
Net ODA * Corruption 291 13.61 17.64 –2.28 82.99
Youth Population 670 33.83 8.40 13.89 48.99
Agriculture, Value Added 647 17.30 11.13 2.39 59.71
GDP per Capita 653 1,998.78 2,307.41 128.30 13,836.19
GDP per Capita squared 653 9.311e+6 2.23e+7 1.646e+4 1.91e+8
The panel data variables ‘Net ODA Received’, ‘Corruption’ and ‘Net ODA * Corruption’ are 5-year lags of the original variables.
4.1.3. Corruption model
Summary statistics of the corruption model are in table 3. Moreover, the correlation matrix can be found in in appendix 7. The correlations and VIF-scores are mostly lower than the
24
Table 4: Regression results cross-country Corruption model
Income inequality (1) (2) (3) (4)
Corruption –0.004 –0.150 0.889
(1.444) (1.548) (1.323)
Net ODA Received 0.09 1.756
(0.267) (1.114)
Net ODA * Corruption –0.590
(0.357)
Years of Schooling –0.796 –0.796 –0.791 –1.029
(0.793) (0.792) (0.804) (0.893)
Youth Population 0.454** 0.454** 0.434** 0.396*
(0.188) (0.185) (0.205) (0.209)
Agriculture, Value Added 0.03 0.03 0.023 –0.038
(0.264) (0.274) (0.297) (0.318)
GDP per Capita 0.009** 0.009** 0.009** 0.009**
(0.003) (0.004) (0.004) (0.004)
GDP per Capita squared –9.17e – 07** –9.18e – 07** –9.70e – 07** –9.88e – 07**
(4.01e – 07) (4.37e – 07) (4.22e – 07) (4.41e – 07)
Constant 18.876 18.879 19.129 19.119
(11.736) (12.063) (12.814) (13.551)
Observations 49 49 49 49
R2 0.356 0.356 0.357 0.393
* Significant at 10% ** Significant at 5% *** Significant at 1%. Robust standard errors in parentheses. Independent and control variables are averaged over the years 2002-2006, income inequality is averaged over the years 2007-2011.
25 As in the Democracy model, all of the control variables have the expected sign and most of them are significant. The only difference is that in this model the control variable ‘Agriculture, Value Added’ is not significant at conventional levels.
4.1.4. Robustness check Corruption model
To verify the results, I check whether changing the years under consideration changes the results. Since the data on corruption is very limited, changing the 5-years averages used for the
independent variables from 2002-2006 to 2001-2005 means a smaller amount of observations is available (N=52). However, the results are still qualitatively equal (see appendix 8).
4.2. Panel data approach
4.2.1. Democracy model
The results of the fixed effects model are published in table 5. The effect of ‘Level of
Democracy’ is now negative, but highly insignificant. It seems that over time, within countries, the level of democracy does not matter for income inequality. As the level of democracy only shows limited changes within countries over time, this result was to be expected. Furthermore, again ‘Net ODA Received’ is not significant by itself.
The effect I am mostly interested in is the effect of ‘Net ODA * Democracy’. In the fixed effects model, this interaction variable influences income inequality negatively, in consonance with the cross-country model. While the coefficient is significant at the 10% level if normal standard errors are used, it loses its significance when robust standard errors are applied.
In summary, it can be stated that evidence is found for the proposition that ODA donated to democratic countries is associated with lower income inequality ten years later, regardless of the exact model specification employed. This evidence is robust in the cross-country models, but not in the fixed effects model. As most of the variance in institutional quality is between countries instead of within countries over time, this is an expected result. Hence, the results of the cross-country model are more important for my analysis.
26
Table 5: Regression results Fixed Effects models
Income inequality Democracy Corruption
Level of Democracy –0.013
(10-year lag) (0.115)
Corruption 0.213
(5-year lag) (0.710)
Net ODA Received –0.010 0.535*
(10-year lag) (0.033) (0.289)
Net ODA * Democracy –0.007
(10-year lag) (0.005)
Net ODA * Corruption –0.229***
(5-year lag) (0.084)
Youth Population 0.184 0.261
(0.214) (0.357)
Agriculture, Value Added –0.179** 0.177*
(0.082) (0.096)
GDP per Capita –0.004*** –0.007**
(0.001) (0.003)
GDP per Capita squared 2.46e – 07* 3.54e – 07**
(1.24e – 07) (1.70e – 07) Constant 50.730*** 48.818*** (8.550) (15.164) Observations 277 89 Countries 65 47 R2 Within 0.199 0.331 R2 Between 0.029 0.062 R2 Overall 0.047 0.0673
* Significant at 10% ** Significant at 5% *** Significant at 1%.
Democracy: Fixed Effects Democracy Model. Robust standard errors in parentheses. Corruption: Fixed Effects Corruption Model. Robust standard errors in parentheses.
27
4.2.2. Corruption model
The results of the fixed effects model are in table 5. The core variable is ‘Net ODA *
Corruption’. In the cross-country model, this variable displayed the expected negative sign, but it was not significant at conventional significance levels. In the fixed effects model however, the coefficient is negative and significant at a 1% significance level if robust standard errors are used. Furthermore, in the random effects model, the same result holds. Therefore, it can be stated that the panel data results show that within countries with a relatively low level of corruption, ODA is related to lower income inequality over time. However, as the panel data approach only shows a time trend, this result bears less importance than the result of the cross-country model.
Furthermore, it was expected that an income equalizing effect would come about because ODA would strengthen the effect of lower corruption. However, this seems not to be the case. Lower corruption by itself is not associated with a lower level of income inequality. In fact, the coefficient of ‘Corruption’ is even positive, albeit not even close to significance. This result resembles the result from the cross-country model. Hence, the prevalence of corruption by itself does not help or harm income equality.
In line with the fixed effects Democracy model, ‘Youth Population’ exhibits a positive but insignificant effect on income inequality. Furthermore, the signs of ‘GDP per Capita’ and ‘GDP per Capita squared’ match the signs in the fixed effects panel Democracy model. This confirms the view that the GDP variables mostly show a time trend in these models. Furthermore,
‘Agriculture, Value Added’ exhibits a positive effect on income inequality, in contrast to the negative effect it displayed in the cross-section models. In the fixed effects panel model, weather-related fluctuations in yields and food prices on the world market are mainly responsible for the variation. Hence, if the share of agriculture in value added increases over time because of a rise in food prices, income inequality will increase, as the poor are relatively hit harder by a rise in food prices.
4.3. Discussion of the results
28 influence income inequality negatively when the level of democracy is relatively high.
Hypothesis 1 is thus confirmed. Furthermore, it is found that ODA donated to countries
exhibiting a low level of democracy, does not influence income inequality. Hence, hypothesis 2 is confirmed as well.
While ODA influences income inequality negatively in countries exhibiting a relatively high level of democracy, ODA does not seem to influence income inequality negatively in the case of a relatively low level of corruption. However, when effects within countries over time are examined, a negative effect is revealed. Hence, hypothesis 3 is partly confirmed. Lastly, hypothesis 4 is confirmed, as ODA donated to countries exhibiting a relatively high level of corruption does not influence income inequality.
Based on an extensive review of the literature, it was expected that the income equalizing effect of ODA donated to countries with high quality institutions would come about because the availability of ODA would strengthen the income equalizing effect of good institutions.
However, the results suggest otherwise. A higher quality of the institutional environment, notably a higher level of democracy and less corruption, has a zero to positive direct influence on income inequality. Hence, institutional quality does not exhibit the expected direct income equalizing effect. As explained above, better institutional quality might initially not have a profound effect or even increase income inequality in less-developed countries, as it will initially affect the informal sector most (Chong & Calderon, 2000).
29
5 | Conclusions
In this MSc thesis, I examined the nature of the effect of official development assistance (ODA) on income inequality. It is argued that the effect of ODA on income inequality depends on the institutional quality in the aid-receiving country. Whether or not the government is able to allocate development aid correctly depends on the quality of governance. Corruption and democracy have been employed as proxies for the quality of institutions in a country. Based on an extensive review of the existing literature, it was hypothesized that ODA would influence income inequality negatively in high quality institutional environments, while it would not do so in low quality institutional environments. It has been proven that ODA does indeed influence income inequality negatively in countries exhibiting a relatively high level of democracy, while it does not influence income inequality in countries displaying a relatively low level of democracy. However, even though I find some panel data evidence of a negative relationship between ODA and income inequality in less corrupt countries, my main model cannot prove that ODA donated to less corrupt countries is related to lower income inequality. Hence, the most important result is the robust negative relationship between ODA donated to countries with a relatively high level of democracy and income inequality.
30 institutional quality leads to a higher income inequality in less-developed countries or whether there are multiple explanations.
Notwithstanding the zero to positive effect on income inequality by itself, the main result of a negative relationship between ODA donated to relatively democratic countries and income inequality still holds. Hence, ODA is only efficiently spent when a country exhibits a relatively high level of democracy. This is an important result, with important implications for governments of both ODA-donating and ODA-receiving countries. ODA-donating governments should
reconsider their plans to donate funds to countries in which the level of democracy is low, if reducing income inequality in these countries is one of their main objectives. These donations might not reach the desired effect. Furthermore, this study reassures ODA-donating governments that donating ODA to countries displaying high institutional quality helps to decrease income inequality and therefore to eradicate poverty.
31
References
Ahluwalia, M. S. (1976). Inequality, poverty and development. Journal of Development Economics, 3(4), 307-342.
Alderson, A. S., & Nielsen, F. (1999). Income inequality, development, and dependence: A reconsideration. American Sociological Review, 606-631.
Bjørnskov, C. (2010). Do elites benefit from democracy and foreign aid in developing countries? Journal of Development Economics, 92(2), 115-124.
Bollen, K. A., & Jackman, R. W. (1985). Political democracy and the size distribution of income. American Sociological Review, 438-457.
Boulier, B. L. (1977). Population policy and income distribution. In C.R. Frank (Ed.), Income Distribution and Growth in the Less Developed Countries (pp. 159-213). Washington, DC: The Brookings Institution.
Bourguignon, F., & Morrisson, C. (1990). Income distribution, development and foreign trade: A cross-sectional analysis. European Economic Review, 34(6), 1113-1132.
Burnside, C., & Dollar, D. (2000). Aid, Policies, and Growth. American Economic Review 90(4), 847–868.
Chong, A., & Calderon, C. (2000). Institutional quality and income distribution. Economic Development and Cultural Change, 48(4), 761-786.
Chong, A., Gradstein, M. & Calderon, C. (2009). Can foreign aid reduce income inequality and poverty? Public Choice, 140(1), 59-84.
CSP (2013). Polity IV Project: Home page. Retrieved from http://www.systemicpeace.org/polity/polity4.htm.
32 De Gregorio, J. D., & Lee, J. W. (2002). Education and income inequality: new evidence from cross‐country data. Review of Income and Wealth, 48(3), 395-416.
Freedom House (2013). Freedom in the World Comparative and Historical Data. Retrieved from http://www.freedomhouse.org/report-types/freedom-world.
Gradstein, M., & Milanovic, B. (2004). Does liberté = égalité? A survey of the empirical links between democracy and inequality with some evidence on the transition economies. Journal of Economic Surveys, 18(4), 515-537.
Granger, C. W. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica: Journal of the Econometric Society, 424-438.
Granger, C. W. (1980). Testing for causality: a personal viewpoint. Journal of Economic Dynamics and Control, 2, 329-352.
Gupta, S., Davoodi, H. & Alonso-Terme, R. (2002). Does Corruption Affect Income Inequality and Poverty? Economics of Governance 3(1): 23-45.
Hansen, H., & Tarp, F. (2001). Aid and growth regressions. Journal of Development Economics, 64(2), 547-570.
Hill, R. C., Griffiths, W. E., & Lim, G. C. (2008). Principles of Econometrics (Vol. 3). Hoboken, NJ: Wiley.
IMF (2003). External Debt Statistics: Guide for Compilers and Users – Appendix III, Glossary. IMF, Washington DC.
Jaggers, K., & Marshall, M. G. (2012a). Political regime characteristics and transitions, 1800- 2012. George Mason University and Center for Systemic Peace. Retrieved from
http://www.systemicpeace.org/inscr/inscr.htm.
Jaggers, K., & Marshall, M. G. (2012b). Polity IV project: Dataset users’ manual. George Mason University and Center for Systemic Peace. Retrieved from
33 Koechlin, V., & León, G. (2007). International remittances and income inequality: an empirical investigation. Journal of Economic Policy Reform, 10(2), 123-141.
Kosack, S. (2003). Effective aid: How democracy allows development aid to improve the quality of life. World Development, 31(1), 1-22.
Kuznets, S. (1955). Economic growth and income inequality. The American Economic Review, 45(1), 1-28.
Lam, R., & Wantchekon, L. (2002). Political Dutch Disease. (Working paper). New Haven, CT: Yale University.
Layton, T. & Nielson, D. (2009). Aiding Inequality: The Effect of Foreign Aid on Income Inequality, Brigham Young University. Retrieved from
https://pedl.byu.edu/Documents/Aiding%20Inequality%20_%20The%20Effect%20of%20Foreig n%20Aid%20on%20Income%20Inequality.pdf.
Lensink, R., & White, H. (2001). Are there negative returns to aid?. Journal of Development Studies, 37(6), 42-65.
Li, H., Squire, L., & Zou, H. F. (1998). Explaining international and intertemporal variations in income inequality. The Economic Journal, 108(446), 26-43.
Lipset, S. M. (1959). Some social requisites of democracy: Economic development and political legitimacy. American Political Science Review 53: 69-105.
Mauro, P. (1998). Corruption and the Composition of Government Expenditure. Journal of Public Economics 69: 263-279.
Mehlum, H., Moene, K., & Torvik, R. (2006). Institutions and the Resource Curse. The Economic Journal, 116(508), 1-20.
Morrison, K. M. (2012). What Can We Learn about the “Resource Curse” from Foreign Aid?. The World Bank Research Observer, 27(1), 52-73.
34 Nielsen, F. (1994). Income inequality and industrial development: Dualism revisited. American Sociological Review, 654-677.
Nielsen, F., & Alderson, A. S. (1995). Income inequality, development, and dualism: Results from an unbalanced cross-national panel. American Sociological Review, 674-701.
Plümper, T., & Neumayer, E. (2010). The level of democracy during interregnum periods: Recoding the polity2 score. Political Analysis, 18(2), 206-226.
Rajan, R. G., & Subramanian, A. (2005). What undermines aid's impact on growth?. IMF Working Paper 05/126.
Reuveny, R. & Li, Q. (2003). Economic openness, democracy, and income inequality: An empirical analysis. Comparative Political Studies 36(5), 575-601.
Robinson, J. A., Torvik, R., & Verdier, T. (2006). Political foundations of the resource curse. Journal of development Economics, 79(2), 447-468.
Sachs, J. D. (2007). How to handle the macroeconomics of oil wealth. Humphreys, M./Sachs, JD/Stiglitz, JE (Ed.), 173-193.
Sachs, J. D., & Warner, A. M. (2000). Natural resource abundance and economic growth. In Meier, G.M. & Rauch, J.E. (Eds.), Leading Issues in Economic Development. Oxford University Press.
Shafiullah, M. (2011). Foreign aid and its impact on income inequality. International Review of Business Research Papers, 7(2), 91-105.
Simpson, M. (1990). Political rights and income inequality: A cross-national test. American Sociological Review, 682-693.
Sirowy, L., & Inkeles, A. (1990). The effects of democracy on economic growth and inequality: A review. Studies in Comparative International Development, 25(1), 126-157.
35 Timmons, J.F. (2010). Does Democracy Reduce Economic Inequality? British Journal of
Political Science, 40(04), 741-757.
Transparency International (2012). Corruption perception index. Retrieved from http://www.transparency.org.
UN General Assembly (2000). United Nations Millennium Declaration, Resolution Adopted by the General Assembly.
UNU-WIDER (2008a). World income inequality database. Retrieved from http://www.wider.unu.edu/research/Database/en_GB/database/.
UNU-WIDER (2008b). World income inequality database user guide and data sources. Retrieved from http://website1.wider.unu.edu/wiid/WIID2c.pdf.
World Bank (2013). World Development indicators. Retrieved from
36
Appendices
Appendix 1: Sample Democracy model (N=67)
Albania Egypt, Arab Rep. Mali Senegal
Argentina El Salvador Mauritania Slovenia
Armenia Gambia, The Mexico South Africa
Bangladesh Ghana Moldova Sri Lanka
Bolivia Honduras Mongolia Swaziland
Brazil India Morocco Tajikistan
Burundi Indonesia Mozambique Tanzania
Cambodia Iran, Islamic Rep. Namibia Thailand
Cameroon Jamaica Nepal Tunisia
Central African Republic Jordan Nicaragua Turkey
Chile Kazakhstan Niger Uganda
China Kenya Pakistan Uruguay
Colombia Kyrgyz Republic Panama Venezuela, RB
Costa Rica Lao PDR Paraguay Vietnam
Cote d'Ivoire Lesotho Peru Yemen, Rep.
Croatia Malawi Philippines Zambia
Dominican Republic Malaysia Rwanda
37
Appendix 2: Sample Corruption model (N=49)
Albania El Salvador Panama
Argentina Honduras Paraguay
Armenia Jordan Peru
Bangladesh Kazakhstan Philippines
Bolivia Kyrgyz Republic Rwanda
Brazil Lao PDR South Africa
Cambodia Malaysia Sri Lanka
Cameroon Mali Swaziland
Central African Republic Mauritania Tajikistan
Chile Mexico Tanzania
Colombia Moldova Thailand
Costa Rica Mongolia Turkey
Cote d'Ivoire Morocco Uganda
Croatia Mozambique Uruguay
Dominican Republic Nepal Vietnam
Ecuador Niger
Egypt, Arab Rep. Pakistan
38
Appendix 3: Correlation matrix cross-country Democracy model
Income inequality Net ODA Received Level of Democracy Years of Schooling Youth Population Agriculture, VA GDP per Capita GDP per Capita sq. Net ODA * Dem. Income inequality 1
Net ODA Received –0.105 1
Level of Democracy 0.346 –0.301 1
Years of Schooling –0.180 –0.358 0.209 1
Youth Population 0.110 0.567 –0.45 –0.546 1
Agriculture, Value Added –0.385 0.647 –0.401 –0.261 0.504 1
GDP per Capita 0.236 –0.529 0.473 0.246 –0.653 –0.700 1
GDP per Capita squared 0.079 –0.379 0.402 0.209 –0.592 –0.515 0.939 1
39
Appendix 4: Does the inclusion of a “Socialist” dummy change the story?
Income inequality (1) (2) Level of Democracy 0.864*** 0.833*** (0.233) (0.225) Socialist –3.577* –4.105** (2.009) (1.783)
Net ODA Received 0.168 0.248**
(0.117) (0.120)
Net ODA * Democracy –0.049*** –0.040**
(0.018) (0.0178)
Years of Schooling –0.637 –0.745
(0.614) (0.629)
Youth Population 0.497*** 0.473***
(0.159) (0.157)
Agriculture, Value Added –0.203* –0.226*
(0.118) (0.116)
GDP per Capita 0.004** 0.004**
(0.002) (0.002)
GDP per Capita squared –6.04e – 07*** –6.02e – 07***
(2.07e – 07) (1.98e – 07) Constant 23.774*** 25.295*** (7.408) (7.699) Observations 67 67 R2 0.535 0.549
* Significant at 10% ** Significant at 5% *** Significant at 1%. Robust standard errors in parentheses. Independent and control variables are averaged over the years 1990-2000, except for the dummy variable “Socialist”. Income inequality is averaged over the years 2001-2011.
(1) Socialist dummy: Countries receive the value “1” if they are considered to be socialist at present.
(2) Socialist dummy: Countries receive the value “1” if they were considered to be socialist until at least 1990.
40
Appendix 5: Does the quality of institutions or the level of income inequality influence the amount of ODA received?
Net ODA received Democracy model Corruption model
Level of Democracy 0.015 (0.171) Corruption 1.019 (0.800) Income inequality 0.15 –0.0353 (0.101) (0.089) GDP per Capita –0.004* –0.001 (0.002) (0.001)
GDP per Capita squared 4.56e – 07* 1.38e – 07
(2.31e – 07) (1.03e – 07)
Youth Population 0.236 0.317**
(0.187) (0.136)
Agriculture, Value Added 0.251** 0.273***
(0.098) (0.085) Years of Schooling –0.484 0.12 (0.756) (0.632) Constant –9.805 –10.317 (8.866) (6.880) Observations 67 64 R2 0.540 0.588
* Significant at 10% ** Significant at 5% *** Significant at 1%. Standard errors in parentheses.
41
Appendix 6: Robustness checks cross-country Democracy model
Income inequality (1) (2) (3) (4)
Level of Democracy 0.852*** 0.678*** 9.857*** 1.518***
(0.284) (0.251) (2.186) (0.561)
Net ODA Received 0.091 0.212* 0.243** 0.490*
(0.284) (0.120) (0.106) (0.270)
Net ODA * Democracy –0.054* –0.040** –0.457** –0.076*
(0.031) (0.019) (0.213) (0.045)
Years of Schooling –0.362 –0.688 –0.275 –0.750
(0.804) (0.661) (0.648) (0.705)
Youth Population 0.613*** 0.444** 0.590*** 0.448**
(0.199) (0.193) (0.144) (0.209)
Agriculture, Value Added –0.085 –0.239* –0.151 –0.190
(0.275) (0.124) (0.115) (0.122)
GDP per Capita 0.005 0.005** 0.005*** 0.005**
(0.003) (0.002) (0.002) (0.002)
GDP per Capita squared –6.76e – 07 –6.19e – 07*** –6.32e – 07*** –6.54e – 07***
(4.05e – 07) (2.17e – 07) (1.90e – 07) (2.11e – 07)
Constant 16.038 26.302*** 15.019* 18.573
(12.097) (9.900) (7.574) (11.458)
Observations 54 67 67 67
R2 0.468 0.489 0.549 0.490
* Significant at 10% ** Significant at 5% *** Significant at 1%. Robust standard errors in parentheses.
(1) Five-year averages: Independent and control variables are averaged over the years 2001-2005, level of income inequality is averaged over the years 2006-2010.
(2) Ten-year averages: Independent and control variables are averaged over the years 1991-2000, level of income inequality is averaged over the years 2001-2010.
(3) Dichotomized dependent variable: All countries receiving an average Policy2-score of +6 or higher receive the value 1, others receive the value 0. Independent and control variables are averaged over the years 1990-2000, level of income inequality is averaged over the years 2001-2011.
42
Appendix 7: Correlation Matrix cross-country Corruption model
Income inequality Net ODA Received Corruption Years of Schooling Youth Population Agriculture, VA GDP per Capita GDP per Capita sq. Net ODA * Dem. Income inequality 1
Net ODA received –0.188 1
Corruption 0.297 –0.323 1
Years of Schooling –0.189 –0.354 0.174 1
Youth Population 0.065 0.609 –0.389 –0.603 1
Agriculture, Value Added –0.299 0.691 –0.504 –0.420 0.565 1
GDP per Capita 0.342 –0.593 0.657 0.192 –0.579 –0.680 1
GDP per Capita squared 0.248 –0.449 0.565 0.127 –0.502 –0.506 0.957 1
43
Appendix 8: Robustness check cross-country Corruption model
Income inequality (1)
Corruption 1.051
(1.228)
Net ODA Received 0.452
(0.992)
Net ODA * Corruption –0.077
(0.376)
Years of Schooling –1.064
(0.871)
Youth Population 0.441**
(0.195) Agriculture, Value Added –0.404*
(0.216)
GDP per Capita 0.004
(0.003)
GDP per Capita squared –5.06e – 07
(3.65e – 07) Constant 28.596*** (9.577) Observations 52 R2 0.449
* Significant at 10% ** Significant at 5% *** Significant at 1%. Robust standard errors in parentheses.
