- BEING ? D OES RECEIVING DEVELOPMENT AID INCREASE SUBJECTIVE WELL

D

OES RECEIVING DEVELOPMENT AID

INCREASE SUBJECTIVE WELL

-

BEING

?

A cross-country analysis of 113 developing countries over the years 2005-2019

Charlotte Strijbos | S3529207 University of Groningen

A

Although aid’s social and economic implications have been studied extensively, researchers have previously ignored an important outcome: subjective well-being. This paper fills this knowledge gap by being one of the first studies to examine this link. Such an analysis is justified because subjective well-being is linked to economic indicators, such as productivity and longevity. Using a sample of 113 developing countries over the years 2005-2019, I test for a direct relationship between aid and subjective well-being, and an indirect one, through channels including income, education, and health. Results indicate no significant relationship between aid and subjective well-being. However, this changes when controlling for heterogeneity in terms of level of income and level of perceived corruption. I find that for the lowest income countries among the sample, receiving aid positively contributes to subjective well-being. For higher income countries, the effect is negative. Also, the effect of aid on subjective well-being is negative for countries with the highest level of perceived corruption. The results of this study are important because it shows that in the way aid is organized now, questions rise about the effectiveness of aid as a development tool overall.

Keywords: Development Aid, Subjective Well-being, Developing Countries.

MSc Economic Development & Globalization Faculty of Economics and Business

June 16th, 2020

2

1. I

High-income countries have been giving development aid to the poorer countries for many decades already. From 2000 until 2017, total official development assistance (ODA) exceeded $2 trillion (World Bank Group, 2019). To date, most studies have focused on evaluating the welfare consequence of aid in recipient countries in terms of gross domestic product (GDP) per capita. Even though a focus on the effect of aid on macroeconomic growth is necessary, it is insufficient. It leaves out important indicators such as non-market activity, sustainability, inequality, and human well-being (Fraumeni, 2017). Also, certain economic activities such as military conflicts or natural disasters are counted in GDP as well, while they are clearly not adding to well-being. Therefore, to consider the full impact of aid, non-monetary aspects influencing this relationship should be taken into consideration as well.

In this study, I examine whether receiving aid is also related to subjective well-being, measured by two dimensions of well-being: evaluative and hedonic. Evaluative well-being measures provide an overall reflection on one’s life, where hedonic well-being measures are reflections of (positive and negative) experiences at a particular point in time (Nikolova, 2016). A focus on subjective well-being is useful, as it is (causally) linked to different economic indicators such as productivity (Drewnowski, 2020), problem solving and creative ideation (Pannells & Claxton, 2008), and to increased longevity (Carstensen et al, 2011). Subjective well-being is shaped by many factors, for example unemployment, health, and income (Clark & Oswald, 2002). Receiving aid could result in an increase in physical and human capital, and improve indicators of social welfare (Arndt, Jones & Tarp, 2015). Besides increases in income, studies find that aid has effects in terms of education (i.e. by funding school-building and teacher training; Arndt et al, 2015) and health (i.e. by providing medicine and access to clean water; Mishra & Newhouse, 2009). These may also translate to subjective well-being benefits.

In this study, I find an answer to the following research question: does receiving development aid increase subjective well-being? I examine this effect directly and indirectly – through mechanisms such as income, education, and health. Also, I explore the role of income and corruption as moderators. I do this by testing for heterogeneity by examining whether the relationship between aid and subjective well-being affects countries with divergent levels of income or perceived levels of corruption differently.

3 receiving both aid and remittances are positively related to happiness. My study substantively differs from Arvin & Lew (2011). I do not include remittances, because the effects of remittances have been studied elsewhere (Ivlevs, Nikolova, Graham, 2019). This study rather focuses on the direct and indirect effects of aid on subjective well-being. Also, the time frame differs, as Arvin & Lew (2011) study data until 2008, where this study covers the more recent years 2005-2019.

I study the relationship between aid and subjective well-being by using fixed effects regressions based on panel data. Results show an insignificant relationship between subjective well-being and development aid. By accounting for different levels of economic development of countries, results show that for the lowest income countries among the sample, income is positively associated with subjective well-being. For higher income quartiles, the effect is negative. Furthermore, I found that aid negatively affects subjective well-being in countries where the level of perceived corruption is high. The results of this study contribute to the literature by showing that the benefits arising from aid in terms of subjective well-being are limited for most developing countries. A focus on the well-being implications of aid is recommended, as this could have a major economic impact on countries that receive aid.

This study uses a sample that consists of low-and middle-income countries. According to the World Bank Group (2020) this includes countries with a real gross national income (GNI) per capita below $12,376 and consists of 138 countries (Appendix 1 and Appendix 2), measured using the Atlas conversion factor.1 It is converted to U.S. dollars by applying the Atlas conversion factor to a country’s GNI in local currency.

This study is divided into 5 sections. The next section discusses relevant literature and states the hypotheses. Section 3 describes the data and methods used to assess the relationship between subjective well-being and aid, followed by the empirical results in section 4. This study finalizes in 5, providing both an answer to the research question and a discussion of the limitations and implications of this study.

1 _{The Atlas conversion factor is a method used by the World Bank to reduce the impact of exchange rate fluctuations. It is}

4

2. L

2.1. Literature review aid

A large number of studies focused on the incentives and implications of aid, but results seem to differ among researchers. The most pronounced critic to aid has arguably come from Easterly (2006), stating that aid efforts rarely succeed because the money does not reach the poor. He argues that this is caused by a lack of two key elements: accountability and feedback. As responsibilities are shared between a bureaucracy and other agencies, incentives for locals to find out what works is low. Also, the feedback system is weak as feedback comes from self-evaluation of donors instead of from the people at the bottom – the recipients of aid. In his book, Easterly (2008) states that the best way to reduce poverty is to let “searchers” (firms in private markets and democratically accountable politicians) explore solutions in a spontaneous way by trial and error, get feedback, and expand the solutions that work. He argues that, therefore, large inflows of aid will not end poverty. Others are more positive but state that aid only works if the recipient countries implement good policies (Burnside & Dollar 2004) or if institutional quality is good enough to ensure there is no corruption (Arvin & Lew, 2012). The majority of studies find positive implications of aid. Clemens, Radelet, Bhavnani & Bazzi (2012) and Hansen & Tarp (2001) find a positive relation between foreign aid and economic growth of the recipient country. According to Yiew & Lau (2018), the positive impact of ODA contributes to development in social infrastructure, economic infrastructure and services, and production sector. It creates more job opportunities in the market, leading to higher incomes and sustainable economic growth. Mekasha & Tarp (2019) extend previous meta-analysis and conclude that aid indeed has a positive impact on growth, and this result appears to be true for different time horizons. Sachs (2006) proposes another view to the effectiveness of aid. He argues that poor countries are stuck in a “poverty trap”, which refers to a set of self-reinforcing mechanisms whereby countries start poor and remain poor (Azariadis & Stachurski, 2005). Sachs (2006) argues that in order to end extreme poverty, a “big push” is needed, requiring wealthy G8 countries to invest in poor countries by increasing aid.

2.2. Hypotheses development

5 subjective well-being. Arvin & Lew (2011) only focus on evaluative subjective well-being, and therefore fail to capture the effect of aid on daily emotions of people in recipient countries. Also, the time frame differs, as Arvin & Lew (2011) study data until 2008, where my study covers the years 2005-2019 by using newly published data by the World Happiness Report (2020). The scarcity of this subject in the literature is surprising because happiness of individuals is becoming an increasingly important economic indicator (Heinberg, 2011). As Heinberg wonders justly: “After all, what good is increased production and consumption if the result isn’t increased human satisfaction?” (p. 257). De Neve, Diener, Tay, & Xuereb (2013) find that subjective well-being leads to benefits in three categories: it causes increased health and longevity, higher income, productivity, and organizational behavior, and finally leads to better individual and social behavior. This implies that having happy individuals in a firm or country has a major economic impact and should thus be included as one of the implications of aid. I expect that the effect of aid is more likely to be visible in terms of life evaluations compared to hedonic well-being, because aid is mostly aimed at improving development in the long run. I therefore expect that the effects on daily emotions will be less evident. This leads to the first hypothesis:

H1: Development aid has a positive direct effect on subjective well-being of residents in

recipient developing countries.

In the second hypothesis, I argue that the effect of aid on subjective well-being might also run indirectly, through channels including income, health, and education. I discuss the three channels by first showing the link between the individual channels and aid, followed by the link between the individual channels and subjective well-being.

2.2.1. Income channel

Receiving aid is related to an increase in income, as shown by Yiew & Lau (2018). They state that the positive impact of ODA contributes to development in social infrastructure, economic infrastructure and the services and production sector. It creates more job opportunities in the market, leading to higher incomes and sustainable economic growth.

6 income and wealth, they gain purchasing power, which expands their bundle of affordable goods. This, again, leads to increased consumption, and, ultimately to improved subjective well-being.

As aid is positively related to income, and income again increases subjective well-being, I expect that income is one of the three channels through which aid affects subjective well-being.

2.2.2. Health channel

Aid increases health outcomes in several ways. Greenhalgh, Kristjansson & Robinson (2007) find that aid that is directed to school feeding programs, causes significant improvements in cognitive performance and growth of disadvantaged children in developing countries. Mishra & Newhouse (2009) find more positive effects of aid on health outcomes, resulting from the fact that aid could provide medicine and access to clean water.

In turn, health is an important determinant of subjective well-being. Layard (2005) states that good health gives a person the opportunity for a long and enjoyable life. When a person is free of human suffering caused by pain, distress and fear, a foundation for happiness is created. Ngamaba, Panagioti, & Armitage (2017) find that health is strongly related to evaluative subjective well-being and conclude that improving people’s health status may be one means by which governments can improve the subjective well-being of their citizens.

Given this evidence, I assume that aid has a positive effect on health. As health is also linked to subjective wel-lbeing, I expect that health is one of the three channels through which aid affects subjective well-being.

2.2.3. Education channel

Studies find a positive relationship between aid and educational outcomes. Arndt et al (2015) argue that receiving aid could lead to an increase in physical and human capital and improve indicators of social welfare. They find that aid directly improves a country’s level of education – both quantitatively (by funding school-building) and qualitatively (by teacher training).

7 As aid affects education, and education is related to subjective well-being, I hypothesize that education is one of the three channels through which aid affects subjective well-being.

Therefore, I hypothesize that aid positively affects income, health and education outcomes and that income, health and education positively affect subjective well-being. I conclude that besides a direct relationship (hypothesis 1), also an indirect relationship exists, leading to hypothesis 2:

H2: The positive effect of development aid on subjective well-being also runs indirectly

through three channels including income, health, and education.

I furthermore expect that the strength of the effect of aid on subjective well-being depends on the level of income. The law of diminishing marginal utility of income suggests that as income increases, each additional dollar contributes less additional satisfaction. This is also described by the Easterlin Paradox (Easterlin, 1973), stating that over the long term, happiness does not increase as a country’s income rises. According to Veenhoven (1991) his “basic needs” theory, the relationship between wealth and happiness is curvilinear. First, rising wealth allows for the satisfaction of basic needs (food, shelter, safety), causing happiness to increase. However, further increases will only add little to people’s subjective well-being. Howell & Howell (2008) study the relationship between the average economic status and subjective well-being for developing countries. They find that this effect is the strongest among low-income developing countries and for samples that were least educated, and that this effect is the weakest among high-income developing countries and for highly educated samples. Therefore, I expect that the law of diminishing marginal utility is also be visible in terms of subjective well-being, resulting in the third hypothesis:

H3: The marginal effect of aid on happiness is the highest for the lowest income countries

8 welfare costs of corruption consist of for example the time and effort required to cope with corrupt behaviour and the psychological costs associated with a general climate of unlawfulness. Therefore, I expect that aid does not increase subjective well-being in countries where corruption is high. I test for heterogeneity between countries with different levels of corruption in the final hypothesis:

H4: In countries where the perceived level of corruption is high, aid does not cause

improvements in subjective well-being.

3. M

3.1. Sample and data collection

The analysis sample used for this study consists of 113 developing countries (Appendix 1). There are no universally agreed criteria for what makes a country developing versus developed. Therefore, this study follows the definition of the World Bank (Khokhar & Serajuddin, 2015), referring to developing countries as low- and middle-income countries. For the current 2020 fiscal year, low- and middle-income countries are those countries with a real GNI per capita below $12,376 (World Bank Group, 2020). Classifying economies by income is useful as it correlates well with several other indicators commonly used to assess the progress of countries, and because data are well available (Fantom & Serajuddin, 2016).

9

3.2. Variables and measurements

3.2.1. Dependent variables

Subjective well-being is defined by the OECD (2013) as “Good mental states, including all of the various evaluations, positive and negative, that people make of their lives and the affective reactions of people to their experiences.”. This definition encompasses three related but separate dimensions: evaluative, hedonic, -and eudaimonic (OECD, 2013; Graham & Nikolova, 2015). Evaluative well-being provides an overall reflection of one’s life as a whole. Hedonic well-being reflects a person’s feelings or emotional states, measured at a particular point in time. This dimension contains of a negative affect (emotions such as worry and stress) and a positive affect (emotions of pleasure, enjoyment, and happiness). The third dimension, eudaimonic well-being, is related to good psychological functioning. It focuses on realizing human potential – having the means and freedoms to fulfil one’s true life purpose (Graham & Nikolova, 2015). However, little research has been done on this dimension and data is scarcely available. I therefore do not consider this dimension in my analysis.

In this study, evaluative subjective well-being is measured using the average life evaluation scores for each country in the GWP. The WHR (2020), whose statistical appendix serves as a data source for this paper, describes how the score of life evaluations in the GWP is measured. It is the national average response to the question of life evaluations. This question is formulated as follows:

“Please imagine a ladder, with steps numbered from 0 at the bottom to 10 at the top. The top of the ladder represents the best possible life for you and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time?”

This measure is referred to as Cantril’s life ladder (Cantril, 1965). It makes well-being assessment possible through a simple visual scale, and it has been used by a wide variety of researchers since its development. Levin & Currie (2014) show that this measure is reliable and valid.

10 negative affects of hedonic subjective well-being, this issue is partly accounted for. The positive affect is defined as the average of three positive affect measures in GWP, which are responses (yes or no) to the following three questions2:

1. “Did you experience feelings of happiness during A LOT OF THE DAY yesterday?” 2. “Did you smile or laugh a lot yesterday?”

3. “Did you experience feelings of enjoyment during A LOT OF THE DAY yesterday?” The negative affect is defined as the average of three negative affect measures in GWP, which are responses (yes or no) to the following three questions:

1. “Did you experience feelings of worry during A LOT OF THE DAY yesterday?” 2. “Did you experience feelings of sadness during A LOT OF THE DAY yesterday?” 3. “Did you experience feelings of anger during A LOT OF THE DAY yesterday?”

3.2.2. Independent variables

This study uses four key independent variables: aid, health, education, and income. This paragraph describes these variables and their measurements. I do not include lags for health, education, and income, as I expect that these variables affect subjective well-being in a short period of time. Similar studies (Arvin & Lew, 2011) also do not use this for measuring the effect of these variables on subjective well-being. Also, Hagerty & Veenhoven (2003) investigated the timing of changes in wealth relative to happiness and found no difference between current and lagged correlations. A lag of one year for aid is used more frequently (Mekasha & Tarp, 2019; Muller, 1985). I follow these studies and therefore include a lag of one year for aid.

- Net ODA received per capita (PPP, ln, t-1)

Aid is measured by net ODA received per capita in current US$ (World Bank Group, 2019a), which is calculated by dividing net ODA received by the midyear population. Because prices in developing economies are lower, the purchasing power of aid spent in the local economy is greater than the purchasing power of the same amount spent in the sending country (The World Bank, 2008). Therefore, it is necessary to adjust ODA per capita to the PPP (purchasing power parity) price level index. I do this this by multiplying the net ODA received per capita (World

11 Bank Group, 2019a) with the price level ratio of PPP conversion factor (World Bank Group, 2019c) of the corresponding years. In this way, I am able to provide a measure of aid that better reflects the costs of living. Lastly, data is transformed into a logarithmic scale. This transformation is preferred as it causes the data to be distributed more normally, leading to smaller standard errors in a regression analysis.

ODA consists of different flows. First, disbursements of loans (made on preferential terms – loans with a grant element of at least 25%). Furthermore, it includes grants by official agencies of members of the Development Assistance Committee (DAC), by multilateral institutions, and by non-DAC countries. In addition to financial flows, it includes the value of technical cooperation, expenditures for peacekeeping under UN mandates, and concessional funding to multilateral development banks. Flows are transfers of resources, either in cash or in the form of commodities or services – measured by their monetary value. The goal of this aid is to promote economic development and welfare in recipient countries.

- Life expectancy at birth, total (years)

This study measures health according to life expectancy. I chose this measure, because life expectancy is an important indicator of health status in a country. Life expectancy is one of the most frequently used health status indicators and is also a good measure to compare socioeconomic development across countries (World Bank Group, 2019d). A rise in life expectancy can be attributed to several factors, such as rising living standards, improved lifestyle, and greater access to quality health services (OECD, 2020). It is measured by the average number of years a new-born infant is expected to live if mortality patterns at the time of its birth remain constant throughout its life.

- Education index

Education index is an average of mean years of schooling (of adults) and expected years of schooling (of school-age children), both expressed as an index3. This index, as is explained in the Human Development Report (UNDP, 2019), provides a good measure of education in a country, as it allows for perfect substitutability between mean years of schooling and expected years of schooling. This seems to be right given that many developing countries have low school

3 _{The index is obtained by applying the dimension index (} 𝑎𝑐𝑡𝑢𝑎𝑙 𝑣𝑎𝑙𝑢𝑒−𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒

𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒−𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒 ) to each of the two

12 attainment among adults but high primary and secondary school enrolment among children. Data is provided by UNDP (2019a), but originate from multiple underlying sources (UNESCO Institute for Statistics (2019); Barro and Lee (2018); ICF Macro Demographic and Health Surveys, UNICEF Multiple Indicator Cluster Surveys, and OECD (2018)).

- GNI per capita (PPP, ln)

GNI per capita (World Bank Group, 2019d) is the sum of value added by all resident producers plus net receipts of primary income from abroad, divided by the midyear population. GNI per capita is based on PPP, meaning that an international dollar has the same purchasing power over GNI as a U.S. dollar has in the United States. Data are in constant 2011 international dollars. I transformed the variable to a logarithmic scale, for similar reasons as for ODA per capita.

3.2.3. Control variables

The WHR (2020) decomposes a country’s average happiness score in six underlying components which explain 75% of the cross-country variation in life evaluations: GDP per person, healthy life expectancy, social support, perceived freedom to make life choice, generosity, and perception of corruption. Life expectancy is already included as an independent variable. Also, as GDP per capita and GNI per capita are correlated (rho=79), I only include GNI per capita. Therefore, the control variables in this study consist of four determinants, measuring different aspects of social environments. I explain the variable definitions and their measurements shortly below.

- Social support

13 - Perceived freedom to make life choice

Following WHR (2020), freedom to make life choices is the national average of responses to the GWP question “Are you satisfied or dissatisfied with your freedom to choose what you do with your life?” Autonomy and the freedom to make life choices are connected to subjective well-being. If a country is able to provide individuals a sense of agency, freedom, and autonomy, this plays a role in explaining subjective well-being (WHR, 2020). Verme (2009) even finds that freedom of choice is one of the variables that predicts life satisfaction better than any other known factor such as subjective health, employment, or income rank. The importance of the freedom to make life choices has also been demonstrated by Inglehart et al (2008), who argue that this sense of freedom is the result of three factors: economic development, democratization, and increasing social tolerance. This in turn has led to higher levels of happiness around the world.

- Generosity

The WHR (2020) measures generosity by residual of regressing national average of response to the GWP question “Have you donated money to a charity in the past month?” on GDP per capita. Generosity is a marker for a sense of positive community engagement, and a central way that humans connect with one other. WHR (2019) finds that people derive happiness from helping others, especially when they feel free to choose how to help, when they feel connected to the people they are helping, and when they can see how their help is making a difference. However, this measure has several methodological limitations, one of them being reverse causality: donating might increase happiness, but it is also possible that happier people are more likely to give to charity.

- Perception of corruption

14 Table 1: Variables and data sources

Variable Description Source

swbev Subjective well-being: evaluative dimension Gallup World Poll swbpos Subjective well-being: positive affect Gallup World Poll swbneg Subjective well-being: negative affect Gallup World Poll lnodapc_lag Log of net ODA received per capita (PPP) in t-1 World Bank

lngnipc_ Log of GNI per capita (PPP) World Bank

education Education index UNDP

lifeexp Healthy life expectancy at birth (years) Gallup World Poll

social Social support Gallup World Poll

freedom Freedom to make life choices Gallup World Poll

generosity Generosity Gallup World Poll

corruption Perceptions of corruption Gallup World Poll

3.3. Empirical model

To analyse the relationship between subjective well-being and aid, I study panel data using time and country fixed effects. I use the following baseline regression model (Eq. 1):

𝑆𝑊𝐵_𝑖𝑡 = 𝑎 + β𝐴𝑖𝑑_{𝑖,𝑡−1}+ 𝑋_𝑖𝑡+ τ_𝑡+ ε_𝑖𝑡 (1) Where 𝑆𝑊𝐵_𝑖𝑡 is the subjective well-being (measured by either life evaluations (0-10), positive affect (0-1), or negative affect (0-1)) of country i in year t. The parameter of interest is β, which shows the logarithm of the amount of aid received, with a lag of one year. X is a matrix of country characteristics (income, education, life expectancy, social, freedom, generosity, and corruption), τ are the year fixed effects, and ε is the stochastic error term.

I test hypotheses 1 and 2 according to the base model above. I start with a bivariate relationship between subjective well-being and aid and add the controls and the channels sequentially. By looking at the magnitude coefficient estimate, I can see how the magnitude of the coefficient estimate on aid changes when I include the additional controls and channels. 4

15 In hypothesis 3, I explore the role of income as a moderator. I test for heterogeneity between countries of different levels of development by splitting the sample into 4 quartiles according to income per capita based on the cross-country distribution. I use interaction terms to see if additional effects of aid on subjective well-being exist for countries in the lowest income group.

In hypothesis 4, I test for a different source of heterogeneity, by examining how different levels of corruption affect the relationship between aid and subjective well-being. The method is similar to hypothesis 3: I split the sample into 4 quartiles according to the perceived level of corruption, and then I compare the coefficients to analyse how the effects of aid on subjective well-being differ for countries with different levels of corruption.

3.3.1. Diagnostic tests

16 threshold value of 10, suggesting that multicollinearity is not an issue in this study (Belsey, Kuh, & Welsch, 1980).

Next, I address possible threats to identifying a causal relationship. Endogeneity stemming from reverse causality would be possible if donor countries take into account information about citizen’s life satisfaction when distributing aid. Another way this relationship could be hypothesized is that happier people are more likely to make friends, and to derive benefits from such relationships (Graham, Laffan, & Pinto, 2018). If this results in happier countries being better able to negotiate and lobby for aid, reversed causality could occur. Lobbying can positively affect the amount of aid received (Montes-Rojas, 2013), however, Arvin & Lew (2009) find no statistically significant impact of happiness on the receipt of aid. Even though this causality between well-being and receiving aid might be uncommon, the causal interpretation of the results should be treated with caution.

17

4. R

4.1. Descriptive statistics

The analysis sample consists of 113 developing countries over a period of 15 years. To make sure all regressions and statistics in this chapter are based on a constant sample group, I only include common observations (N=866). Table 2 shows the descriptive statistics of the variables. The mean value of life evaluations of all countries is 4.9303 out of 10. The mean of the variable corruption is remarkably high (0.8032 out of 1), indicating that the perception corruption is highly present among the countries in this sample. This is not surprising, given that the sample consists of low- and middle- income countries. However, it could still be problematic for this study, as corruption might suppress the effects of aid.

Table 2: Descriptive statistics

Variable Mean St. Dev. Minimum Maximum

swbev 4.9290 0.9002 2.6931 7.6149 swbpos 0.6950 0.1085 0.4347 0.9436 swbneg 0.2766 0.0840 0.1094 0.5993 lnodapc_lag 4.3218 1.2675 -0.3423 6.7521 lngnipc 8.5396 0.9186 6.4198 10.1256 education 0.5574 0.1551 0.1490 0.8560 lifeexp 59.4604 6.8808 32.3000 71.3000 social 0.7647 0.1189 0.2902 0.9592 freedom 0.7022 0.1365 0.2575 0.9638 generosity -0.0126 0.1393 -0.2799 0.6270 corruption 0.8031 0.1131 0.0780 0.9436

N = 866 observations for 113 countries.

4.2. Regression results H1 and H2

Table 3 presents the results of the fixed effects analysis. The bivariate relationship between evaluative subjective well-being and aid is tested in model 1, where model 2 includes the controls. In both models, there appears to be no significant relationship between subjective well-being and aid.

18 core model including all variables, the adjusted R² is 0.1490. This indicates that adjusted for the number of variables in the model, the independent variables help to explain 14.90% of the variation in subjective well-being. The controls social (p ≈ 0.000), generosity (p ≈ 0.086), and corruption (p ≈ 0.007) are statistically significant at 1, 10 and 1% respectively. Also, the directions of their coefficients are as hypothesized: having social support and being generous in helping others positively affects subjective well-being and living in a corrupt institutional environment negatively affects subjective well-being. In model 5, where all independent variables and controls are included, the relationship between aid and subjective well-being remains insignificant.

Finally, models 6 and 7 show the results of the regressions where the positive and negative affect of hedonic well-being are used as dependent variables. As was predicted in the hypotheses, does receiving aid not significantly affect daily emotions. In model 6, freedom (p ≈ 0.055) significantly affects the positive affect of subjective well-being. In model 7, education (p ≈ 0.062), life expectancy (p ≈ 0.061), social (p ≈ 0.000) and freedom (p ≈ 0.046) significantly affect the negative affect of hedonic well-being. The difference between model 6 and 7 also becomes visible when comparing the values of the adjusted R² (0.0325 in model 6 and 0.3213 in model 7). This indicates that the independent variables are very helpful in explaining the variation in the negative affect of hedonic well-being, but that this is not the case for the positive affect of hedonic well-being. For the remainder of this study, I use evaluative well-being as a measure of subjective well-being. This measure provides the best reflection on one’s life and is used most commonly in subjective well-being studies (Arvin & Lew, 2011; Howell & Howell, 2008; Ivlevs, Nikolova, Graham, 2019). Ngamaba, Panagioti, & Armitage (2017) also prefer life satisfaction to happiness as a measure of subjective well-being.

19 Table 3: Regression results on the different dimensions of subjective well-being (SWB)

Variables SWB: Evaluative (0-10) SWB: Positive Affect (0-1) SWB: Negative Affect (0-1) (1) (2) (3) (4) (5) (6) (7) Aid -.0518 -0.0408 -0.0399 -0.0403 -0.0392 -0.0056 -0.0045 (0.0368) (0.0373) (0.0378) (0.0400) (0.0382) (0.0052) (0.0049) Income 0.0458 0.0438 0.0658 -0.0402 -0.0426 (0.3490) (0.3398) (0.3402) (0.0343) (0.0438) Education 0.0758 -0.0443 -0.2573 0.3603* (1.3502) (1.3535) (0.1798) (0.1908) Life Expectancy -0.0652*** -0.0039 0.0054* (0.0187) (0.0012) (0.0029) Social 1.4714*** 1.4669*** 1.4672*** 1.3466*** 0.0003 -0.2557*** (0.3664) (0.3673) (0.3672) (0.3503) (0.0518) (0.0390) Freedom 0.2851 0.2826 0.2803 0.2145 0.0537* -0.0646** (0.2683) (0.2657) (0.2682) (0.2584) (0.0277) (0.0319) Generosity 0.3488 0.3504 0.3518 0.4902* 0.0714** 0.0320 (0.2747) (0.2705) (0.2696) (0.2826) (0.0353) (0.0314) Corruption -0.8802** -0.8750** -0.8751** -0.9230*** -0.0019 0.0342 (0.3550) (0.3496) (0.3501) (0.3355) (0.0378) (0.0375) Adjusted R-squared 0.0512 0.1136 0.1126 0.1116 0.1490 0.0325 0.3213

Year & country fixed effects Yes Yes Yes Yes Yes Yes Yes

N 866 866 866 866 866 866 866

20

4.3. Regression results H3

In hypothesis 3, I test for heterogeneity between countries by splitting the sample into 4 quartiles according to average income per capita. Table 4 shows how the countries with different incomes are divided over the quartiles.

Table 4: Sample quartiles according to average GNI per capita

Quartile Observations Minimum income Maximum income

1 219 704.7963 2471.7910

2 221 2559.5280 5610.3970

3 213 5754.4520 11530.5400

4 217 11706.1300 20910.9900

Income is measured by GNI per capita in PPP (constant 2011 international dollar).

Table 5 presents the results of the regression. The second row shows that if I regress the quartiles separately, the effect of aid on subjective well-being is not significant for all four quartiles. This might be caused by the loss of statistical power. Therefore, I use an interaction term between the income quartiles and aid to preserve the number of observations (N=870). Results show that the effect of the interaction between aid and the income quartiles on subjective well-being is significant for all four quartiles. For the first quartile, the effect of aid on subjective well-being is positive. For the other, higher income quartiles, the effect of aid on subjective well-being is negative. This indicates that for the lowest income countries among the sample, receiving aid positively contributes to subjective well-being. This is in line with the “basic needs” theory of Veenhoven (1991), as for those countries, the satisfaction of basic needs (food, shelter, safety) causes happiness to increase.

The controls also affect subjective well-being differently along the income quartiles. For the first quartile, education (p ≈ 0.058) and social (p ≈ 0.004) are significant, suggesting that attending school and having social support are important indicators of subjective well-being among countries with the lowest income in the sample. Life expectancy (p ≈ 0.005) negatively affects subjective well-being. Also, the coefficient for corruption (p ≈ 0.030) is negative and significant, suggesting that corruption negatively affects subjective well-being in the lowest-income countries in the sample.

21 show that social support positively significantly affects subjective well-being in Q1 (p ≈ 0.004), Q3 (p ≈ 0.008), and Q4 (p ≈ 0.000).

The adjusted R² of the regression including Q1 is 0.2412, which is an increase compared to the core model (5) in table 3. The highest adjusted R² is reached in the fourth quartile (0.2788). This indicates that, adjusted for the number of variables included in the model, the independent variables help to explain 27.88% of the variation in subjective well-being in the countries that report the highest income per capita among the sample.

4.4. Regression results H4

In hypothesis 4, I test for heterogeneity between countries by splitting the sample into 4 quartiles according to average perceptions of corruption. Table 6 shows how the countries with different perceptions of corruption are divided over the quartiles.

Table 6: Sample quartiles according to average perceptions of corruption (0-1)

Quartile Observations Minimum corruption Maximum corruption

1 218 0.1765 0.7602

2 221 0.7611 0.8162

3 221 0.8163 0.8580

4 210 0.8643 0.9481

The method for this hypothesis is similar to hypothesis 3. First, I interact the corruption quartiles with aid using the complete sample (N=870), to examine if there are additional negative effects of aid on subjective well-being in countries with a high level of corruption. The results shown in table 7 indicate that for all quartiles, the effect of aid on subjective well-being is not significant. However, the results from the sample split report a significant negative coefficient on aid for the fourth quartile (p ≈ 0.083). This indicates that for the group of countries with the highest perceptions of corruption, receiving aid has a negative effect on subjective well-being. For the other country groups, no significant effect is shown.

The controls also affect subjective well-being differently along the different quartiles, but this happens mostly in an unstructured manner. Freedom (p ≈ 0.036) is significant only in the fourth quartile, suggesting that the freedom to make life choices positively affects subjective well-being in countries where the perceived level of corruption is high.

22

Coefficients are reported with robust standard errors clustered by country in parentheses. *p<0.1; **p<0.05, *** p<0.01.

Table 5: Regression results by different income groups

Variables Evaluative subjective well-being (0-10)

(Q1) (Q2) (Q3) (Q4)

Income quartile * aid 0.2645*** -0.3250** -0.3404*** -0.3421***

(0.0849) (0.1321) (0.1149) (0.0912) Aid 0.1156 -0.0118 0.0302 -0.0452 (0.0029) (0.0857) (0.0564) (0.0370) Income 0.0029 -1.1724 1.0773** 0.7216* (0.5392) (0.7941) (0.4578) (0.3671) Education 4.7989* -2.0250 2.7358 -4.2081*** (2.4180) (3.2524) (1.8457) (0.9194) Life Expectancy -0.0540*** -0.0513 -0.1097 -0.1502*** (0.0174) (0.0671) (0.0679) (0.0261) Social 1.6739*** 0.4585 1.9251*** 2.0514*** (0.5231) (0.7660) (0.6633) (0.4478) Freedom -0.9211 0.1716 0.6380 1.3952*** (0.7668) (0.3589) (0.5760) (0.3621) Generosity 0.3583 0.6987* -0.7018 0.6610* (0.6784) (0.4045) (0.6175) (0.3767) Corruption -1.4663** -1.1859 -0.8279* 0.0213 (0.6388) (0.7958) (0.4811) (0.6972) Adjusted R-squared 0.2412 0.1081 0.2534 0.2788

Year & country fixed effects Yes Yes Yes Yes

23

Coefficients are reported with robust standard errors clustered by country in parentheses. *p<0.1; **p<0.05, *** p<0.01.

Table 7: Regression results by different corruption groups

Variables Evaluative subjective well-being (0-10)

(Q1) (Q2) (Q3) (Q4)

Corruption quartile * aid -0.0543 0.0324 0.0971 -0.0593

(0.0941) (0.1104) (0.1212) (0.1221) Aid -0.0489 -0.0246 0.0523 -0.1277* (0.1141) (0.0436) (0.0812) (0.0700) Income -0.0983 -0.2128 -0.1892 0.3601 (0.4550) (0.5187) (0.8331) (0.6202) Education -1.2379 4.2544 -2.5437 0.1167 (1.9890) (2.6601) (3.1429) (2.0241) Life Expectancy -0.0556*** -0.0716* -0.0881** -0.0784* (0.0178) (0.0386) (0.0392) (0.0463) Social -0.3147 2.3406*** 2.4393*** 0.2539 0.5993 (0.4777) (0.3986) (0.7074) Freedom -0.0523 0.3665 -0.7577 1.2644** (0.4377) (0.4480) (0.5955) (0.5609) Generosity 0.9160* 0.7873 0.1219 0.6270 (0.4959) (0.5480) (0.7234) (0.4647) Corruption -0.4560 -0.8847* -1.7064* -1.1958 (0.4965) (0.4785) (0.8609) (1.0732) Adjusted R-squared 0.1123 0.2137 0.1674 0.2405

Year & country fixed effects Yes Yes Yes Yes

24

4.5. Robustness checks

To examine how the core regression coefficients estimates behave when the regression specification is modified, I conduct several robustness checks. Table 8 shows the results of the core model (similar to model 5 in table 3), and the results of the different robustness checks.

First, I remove observations of aid below the 1st percentile and above the 99th percentile, to make sure that the results are not driven by outliers. The differences of this model (8) compared to the core model are negligible.

Next, I look at different regions within the data. The Gallup data has excellent country coverage, especially of countries in Sub-Saharan Africa (SSH), that before were not in any of the surveys on subjective well-being. Therefore, it is possible that others do find some effects on subjective well-being because they primarily looked at other parts in the world and exclude SSH. To test if SSH affects the results, I split the sample into 7 regions (Appendix 9), as classified by the World Bank Group (2020) and exclude SSH in the regression. Results show that the effect of aid on subjective well-being remains similar. Even though the number of observations drops, the adjusted R² increases from 0.1490 in the core model to 0.1925 in model 8. Also, education (p ≈ 0.069) and freedom (p ≈ 0.011) become significant. Since the effect of aid on subjective well-being remains similar and insignificant, I conclude that SSH does not bias the results.

25 Table 8: Regression results of the different robustness checks

Variables Evaluative subjective well-being (0-10)

Core model (8) (9) (10) Aid -0.0392 -0.0381 -0.0638 0.0560 (0.0382) (0.0402) (0.0441) (0.0446) L1 0.0659 (0.1008) L2 -0.0626 (0.0813) Income 0.0658 0.0855 0.1849 0.4808 (0.3402) (0.3514) (0.4570) (0.4602) Education -0.0443 -0.0776 -2.4219* -1.7927 (1.3535) (1.3766) (1.3060) (2.3934) Life Expectancy -0.0652*** -0.0768*** -0.0493*** 0.0214 (0.0187) (0.0160) (0.0146) (0.0329) Social 1.3466*** 1.2598*** 1.6980*** 1.0359** (0.3503) (0.3580) (0.4755) (0.4974) Freedom 0.2145 0.1581 0.7138** 0.6476* (0.2584) (0.2661) (0.2727) (0.3352) Generosity 0.4902* 0.4855* 0.0483 0.4946 (0.2826) (0.2907) (0.3228) (0.3438) Corruption -0.9230*** -0.9492*** -1.0931*** -0.4761 (0.3355) (0.3378) 0.4084) (0.3944) Adjusted R-squared 0.1490 0.1564 0.1925 -

Year & country fixed effects Yes Yes Yes No

N 866 848 542 521

5. C

5.1. Conclusion

This study investigated the effect of receiving development aid on subjective well-being. Building on theories that support positive implications of aid, which in turn contribute to increased subjective well-being, I hypothesized that aid directly and indirectly positively affects subjective well-being. However, results of the core regression reported insignificant results on the relationship between aid and subjective well-being. I did find that social support and generosity positively affect subjective well-being, while the presence of corruption has a negative effect.

Furthermore, I tested for different sorts of heterogeneity. In hypothesis 3, I accounted for different levels of income per capita. Results have shown that for the lowest income countries among the sample, receiving aid positively contributes to subjective well-being. For the other quartiles, the effect is negative. This is in line with the “basic needs” theory of Veenhoven (1991), as for countries with a low level of development, the satisfaction of basic needs (food, shelter, safety) causes happiness to increase. Results have furthermore shown that for the higher income countries among the sample, income is positively associated with subjective well-being. This is in line with Easterlin (1974) and the economic utility theory. Education positively affects subjective well-being in the lowest income quartile, while it negatively affects well-being in the highest income quartile. This might be caused by the fact that in lower income countries, opportunities for education are limited. Having a safe place to teach children, in a building with electricity and running water, may not be a given. Attainment could increase subjective well-being in a country where education is scarce and decrease subjective well-well-being in a country characterized by a higher level of development, where it might bring more stress and worry.

27

5.2. Discussion

This study makes important contributions to the literature. It is the first study to link receiving aid with subjective well-being and test this by using a large sample of 113 developing countries for a period of 15 years. Even though development aid and subjective well-being have extensively been investigated separately, studies linking both topics are scarce. The results of this study are important because it shows that in the way aid is organized now, questions rise about the effectiveness of aid as a development tool overall. Only for the lowest income countries, aid positively contributes to subjective well-being. A focus on the well-being implications of aid is highly recommend for all developing countries, because having happy residents in a country or firm has several economic implications such as increased productivity and longevity.

There are several ways to explain the results in this study. First, this study does not account for different types of aid. As different types of aid have different purposes, aid should not be treated as a unitary concept. Also, the structure of aid has changed over the years. Sub-Saharan Africa used to borrow from official lenders, but since the global financial crisis in 2008, the share of official development assistance has declined while the share of private lending has increased (The Economist Group, 2014). This study only includes ODA from the year 2006, and therefore only captures the new developments in aid. It could also be that the effect of aid on subjective well-being might be visible for particular individuals, but that it is not possible to pick up this effect from country level data.

To explore the relationship between subjective well-being in more detail, I conducted some additional tests. I measured changes in subjective well-being and aid rather than levels, however, the relationship between the two main variables remained statistically insignificant. Furthermore, I tested for a non-linear relationship between subjective well-being and aid, but this was not the case. Also, I tested the extensive margin by splitting the sample according to the amount of aid received, but concluded that no matter the level of aid, the relationship between aid and subjective well-being remained insignificant. Finally, I tested whether the result was driven by the nature of the subjective well-being question. I ran the regression using the measure of life evaluation used by WVS but found that there were too few common observations to test this.

28 though it is desirable to do so. Also, due to a lack of data availability, eudaimonic well-being is not included. This means that not all dimensions of subjective well-being are considered. Besides that, data about ODA are measured in current US$, which might cause discrepancies with data if it were measured in constant US$.

29

R

L

Arndt, C., Jones, S. & Tarp, F. (2015). Assessing Foreign Aid’s Long-Run Contribution to Growth and Development. World Development, 69: 6-18.

Arvin, M. & Lew, B. (2009). Happiness and Foreign Aid. Atlantic Economic Journal, 37: 325-326.

Arvin, M. & Lew, B. (2011). Are foreign aid and Migrant Remittances Sources of Happiness in Recipient Countries? International Journal of Policy, 7(4/5/6): 282-300.

Arvin, M. & Lew, B. (2012). Development Aid, Corruption, and the Happiness of Nations: Analysis of 118 Countries over the Years 1996-2009. Applied Econometrics and International Development, 12(2): 69-78.

Azariadis, C. & Stachurski, J. (2005). Poverty traps. Handbook of economic growth 1(A): 295-384.

Barro, R.J., & Lee, J.W. (2018). Dataset of Educational Attainment, February 2016 Revision. Accessed (by UNDP) 15-06-2019 from www.barrolee.com.

Belsey, D. A., Kuh, E., & Welsch, R. E. (1980). Regression diagnostics: Identifying influential observations and sources of collinearity. New York: John Wiley and Sons.

Cantril, H. (1965). The Pattern of Human Concerns. New Brunswick: Rutgers University Press.

Carstensen, L. L., Turan, B., Schiebe, S., Ram, N., Ersner-Hershfield, H., Samanez-Larkin, G. R., Brooks, K. P., & Nesselroad, J. R. (2011). Emotional experience improves with age: evidence based on over 10 years of experience sampling. Psychology and Aging, 26(1): 21-33.

Clark, A. E. & Oswald, A. J. (2002). A simple statistical method for measuring how life events affect happiness. International Journal of Epidemiology, 31: 1139-1144.

30 De Neve, J. E., Diener, E., Tay, L., & Xuereb, C. (2013). The objective benefits of

subjective well-being. In Helliwell, J. F., Layard, R., & Sachs, J. (Eds.), World Happiness Report 2013, 2: 54-79. New York: UN Sustainable Network Development Solutions Network.

Drewnowski, A. (2020). Impact of nutrition interventions and dietary nutrient density on productivity in the workplace. Nutrition Reviews, 78(3): 215-224.

Easterlin, R. A. (1973). Does Money Buy Happiness? Public Interest, 30: 3-10.

Easterlin, R. A. (1974). Does economic growth improve the human lot? Some empirical evidence. Nations and Households in Economic Growth: Essays in Honour of Moses Abramowitz. New York and London: Academic Press.

Easterly, W. (2006). The White Man’s Burden: Why the West’s Efforts to Aid the Rest Have Done So Much Ill and So Little Good. New York: The Penguin Press.

Easterly, W. (2008). Reinventing Foreign Aid. London: The MIT Press.

Fantom, N. & Serajuddin, U. (2016). Classifying countries by income: A new working paper. accessed 02-04-2020 from https://blogs.worldbank.org/opendata/classifying-countries-income-new-working-paper

Fraumeni, B. M. (2017). Gross Domestic Product: Are Other Measures Needed? IZA World of Labor, 368.

Gallup. (2009). World Poll Methodology, Technical Report. Washington, DC.

Greenhalgh, T., Kristjansson, E. & Robinson, V. (2007). Realist Review to Understand the Efficiacy of School Feeding Programs. British Medical Journal, 335: 858-861.

Hagerty, M. R. & Veenhoven, R. (2003). Wealth and happiness revisited: Growing wealth of nations does go with greater happiness. Social Indicators Research, 64: 1-27

Hansen, H. & Tarp, F. (2001). Aid and growth regressions. Journal of Development Economics, 64(2): 547-570.

Heinberg, R. (2011). The End of Growth: Adapting to Our New Economic Reality. Gabriola Island, Canada: New Society Publishers.

31 Helliwell, J. F., Layard, J. R, Sachs, J., & De Neve, J. E. (2020). World Happiness Report 2020. New York: Sustainable Development Solutions Network.

Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10: 227-237.

Howell, R. T. & Howell, C. J. (2008). The Relation of Economic Status to Subjective Well-Being in Developing Countries: A Meta-Analysis. Psychological Bulletin, 134(4): 536-560.

ICF Macro. (Various years). Demographic and Health Surveys. Accessed (by UNDP) 15-04-2019 from www.measuredhs.com

Inglehart, R., Foa, R., Peterson, C., & Welzel, C. (2008). Development, freedom, and rising happiness: A global perspective (1981–2007). Perspectives on Psychological Science, 3(4): 264–285.

Ivlevs, A., Nikolova, M., & Graham, C. L. (2019). Emigration, Remittances and the Subjective Well-Being of Those Staying Behind. Journal of Population Economics, 32(1): 113-151.

Khokhar, T. & Serajuddin, U. (2015). Should we continue to use the term ‘developing world’? accessed 02-04-2020 from https://blogs.worldbank.org/opendata/classifying-countries-income-new-working-paper

Layard, R. (2005). Happiness. Lessons from a new science. London: Allen Lane.

Levin, K. A. & Currie, C. (2014). Reliabillity and Validity of an Adapted Version of the Cantril Ladder for Use with Adolescent Samples. Social Indicators Research, 119: 1047-1063.

Mekasha, T. J. & Tarp, F. (2019). A Meta-Analysis of Aid Effectiveness: Revisiting the Evidence. Politics and Governance, 7(2): 5-28.

Michalos, A. C. (2017). Education, Happiness, and Wellbeing. In: Connecting the Quality of Life Theory to Health, Well-being and Education. Springer, Cham.

32 Montes-Rojas, G. (2013). Can Poor Countries Lobby for More US Bilateral Aid? World Development, 44: 77-87.

Muller, E. N. (1985). Dependent Economic Development, Aid Dependence on the United States, and Democratic Breakdown in the Third World. International Studies Quarterly, 29(4): 445-469.

Ngamaba, K. H., Panagioti, M., & Armitage, C. J. (2017). How strongly related are health status and subjective well-being? Systematic review and meta-analysis. The European

Journal of Public Health, 27(5): 879-885.

Nikolova, M. (2016). Minding the happiness gap: Political institutions and perceived quality of life in transition. European Journal of Political Economy, 45: 129-148.

OECD. (2013). OECD Guidelines on Measuring Subjective Well-being. Paris: OECD Publishing.

OECD. (2018). Education at a Glance 2018: OECD Indicators. Paris. Accessed (by UNDP) 15-06-2019 from www.oecd-ilibrary.org/education/education-at-a-glance-2018_eag-2018-en.

OECD. (2020). Life expectancy at birth (indicator). Accessed on 29-04-2020 from https://data.oecd.org/healthstat/life-expectancy-at-birth.htm

Pannells, T. C. & A. F. Claxton. (2008). Happiness, Creative Ideation, and Locus of Control. Creativity Research Journal, 20(1): 67-71.

Preacher, K. J. & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36(4): 717-731.

Quibria, M. G. (2017). Foreign aid and Corruption: Anti-Corruption Strategies Need Greater Alignment with the Objective of Aid Effectiveness. Georgetown Journal of International Affairs, 18(2): 10-17.

Sachs, J. D. (2006). The End of Poverty. New York: The Penguin Press.

The Economist Group (2014). The coming African debt crisis. Accessed 06-06-2020 from https://www.economist.com/news/2014/11/13/the-coming-african-debt-crisis

33 UNESCO Institute for Statistics. (2019). Data Centre. Accessed (by UNDP) 11-04-2019 from http://data.uis.unesco.org

UNDP. (2019). Human Development Report 2019. New York: AGS. UNDP. (2019a). Education Index. Accessed 04-05-2020 from

http://hdr.undp.org/en/indicators/103706

UNICEF. (Various years). Multiple Indicator Cluster Surveys. New York. Accessed (by UNDP) 15-04-2019 from http://mics.unicef.org

Veenhoven, R. (1991). Is happiness relative? Social Indicators Research, 24: 1-34. Verme, P. (2009). Happiness, Freedom and Control. Journal of Economic Behavior & Organization, 71(2): 146-161.

Welsch, H. (2008) The welfare costs of corruption. Applied Economics, 40(14): 1839-1849. World Bank Group. (2008). World Development Indicators 2008. Washington.

World Bank Group. (2019). Net official development assistance and official aid received (current US$). Accessed 07-02-2020 from

https://data.worldbank.org/indicator/DT.ODA.ALLD.CD

World Bank Group. (2019a). Net ODA received per capita (current US$). Accessed 03-04-2020 from https://data.worldbank.org/indicator/DT.ODA.ODAT.PC.ZS

World Bank Group. (2019b). GNI per capita, PPP (constant 2011 international $). Accessed 05-05-2020 from https://data.worldbank.org/indicator/NY.GNP.PCAP.PP.KD

World Bank Group. (2019c). Price level ratio of PPP conversion factor (GDP) to market exchange rate. Accessed 05-05-2020 from

https://data.worldbank.org/indicator/PA.NUS.PPPC.RF

World Bank Group. (2019d). Life expectancy at birth, total (years). Accessed 03-04-2020 from https://data.worldbank.org/indicator/SP.DYN.LE00.IN

World Bank Group. (2020). World Bank Country and Lending Groups. Accessed 02-04-2020 from https://datahelpdesk.worldbank.org/knowledgebase/articles/906519

34

Appendix 1: Low- and middle-income countries in sample

Underlined countries represent low-income countries according to the World Bank classification.

Afghanistan, Albania, Algeria, Angola, Argentina, Armenia, Azerbaijan, Bangladesh, Belarus, Belize, Benin, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brazil, Bulgaria, Burkina Faso, Burundi, Cambodia, Cameroon, Central African Republic, Chad, China, Colombia, Comoros, Congo, Dem. Rep., Congo, Rep., Costa Rica, Cuba, Djibouti, Dominican Republic, Ecuador, Egypt Arab Rep., El Salvador, Ethiopia, Gabon, Gambia, Georgia, Ghana, Guatemala, Guinea, Guyana, Haiti, Honduras, India, Indonesia, Iran, Iraq, Jamaica, Jordan, Kazakhstan, Kenya, (South) Korea, Kosovo, Kyrgyz Republic, Lao PDR, Lebanon, Lesotho, Liberia, Libya, Madagascar, Malawi, Malaysia, Maldives, Mali, Mauritania, Mauritius, Mexico, Moldova, Mongolia, Montenegro, Morocco, Mozambique, Myanmar, Namibia, Nepal, Nicaragua, Niger, Nigeria, (North) Macedonia, Pakistan, Paraguay, Peru, Philippines, Romania, Russian Federation, Rwanda, Senegal, Serbia, Sierra Leone, Somalia, South Africa, South Sudan, Sri Lanka, Sudan, Suriname, Syrian Arab Republic, Tajikistan, Tanzania, Thailand, Togo, Tunisia, Turkey, Turkmenistan, Uganda, Ukraine, Uzbekistan, Venezuela, Vietnam, Yemen, Rep., Zambia, Zimbabwe.

Appendix 2: Low- and middle-income countries excluded from sample (due

to a lack of coverage in the GWP)

35

Appendix 3: Hausman test

Appendix 4: Test for heteroskedasticity

Appendix 5: Test for correlation

Appendix 6: F-test for year dummies

Prob>chi2 = 0.0000 = 68.97

chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B) Test: Ho: difference in coefficients not systematic

B = inconsistent under Ha, efficient under Ho; obtained from xtreg b = consistent under Ho and Ha; obtained from xtreg corruption -.7213624 -.4008457 -.3205167 .1282226 generosity .3819545 .3400578 .0418967 .0905608 freedom .3916352 .4374532 -.045818 .0964866 social 1.160902 1.626638 -.4657358 .117553 education .9821366 .1886038 .7935328 .6080203 lifeexp -.0402801 -.000865 -.039415 .0064359 lngnipc .4888646 .4075187 .081346 .1677039 lnodapc_lag -.0245139 -.0825448 .0580309 .0198758 fe re Difference S.E. (b) (B) (b-B) sqrt(diag(V_b-V_B)) Coefficients Prob>chi2 = 0.0000 chi2 (98) = 1.1e+30

H0: sigma(i)^2 = sigma^2 for all i in fixed effect regression model

Modified Wald test for groupwise heteroskedasticity

Prob > F = 0.0000 F( 1, 81) = 56.403 H0: no first-order autocorrelation

Wooldridge test for autocorrelation in panel data

36

Appendix 7: VIF

Appendix 8: Scatterplot evaluative SWB and ln ODA per capita lagged

Appendix 9: Sample split according to region

Mean VIF 2.14 corruption 1.21 0.824671 generosity 1.26 0.795623 swbneg 1.30 0.770144 lnodapc_lag 1.48 0.674308 freedom 1.57 0.638449 social 1.88 0.530726 swbpos 2.01 0.497534 lifeexp 2.87 0.348395 education 3.62 0.276409 lngnipc 4.21 0.237765 Variable VIF 1/VIF

Region number Region name Observations Countries

1 East Asia and Pacific 62 9

2 Europe and Central Asia 174 17

3 Latin America & the Caribbean 199 18

4 Middle East and North Africa 40 8

5 North America 0 0 6 South Asia 67 7 7 Sub-Saharan Africa 328 39 3 4 5 6 7 8 E v a lu a ti v e s u b je c ti v e w e ll-b e in g 0 2 4 6 8