THE EFFECT OF CHINESE AID ON FEMALE LABOR FORCE
PARTICIPATION
By ELVIRA DE VRIES S2745615 University of GroningenFaculty of Economics and Business
Master’s Thesis International Economics and Business
Supervisor: Dr. Anna Minasyan
Co-assessor: Prof. Dr. Erik Dietzenbacher
Abstract
The benefits of foreign aid have been under recent severe scrutiny. While some scholars argue that foreign aid is one of the most efficient practices to reduce or eliminate poverty, others claim that aid is unlikely to have positive transformative effects and that it may even worsen economic development in recipient countries. China positioned itself in the midst of this controversy after becoming one of world’s largest and most influential donors. Using data on aid allocations from China to 119 recipient countries over the 2000-2014 period, this paper investigates whether Chinese aid affects female labor force participation in recipient countries. The empirical results suggest that, in its broadest sense, Chinese aid has no effect on female labor force participation. Surprisingly, when using different definitions for Chinese aid, by seperating Official Development Assistance and Other Official Flows, the latter becomes positive and significant with female labor force participation. Moreover, subsample analysis by region reveals that the insignificant results for Official Development Assistance may to some extent be subject to heterogenous treatment effects.
Table of contents
1. Introduction ... 4
2. Conceptual framework ... 5
2.1 China as an emerging donor ... 5
2.2 The link between female labor force participation and Chinese aid ... 6
3. Data and descriptive statistics ... 10
1. Introduction
Development assistance began in earnest in the 1960s, when rich countries started to feel that they had to do something to reduce the massive development deficits developing countries were facing (Doucouliagos and Paldam, 2009). Although development assistance is still used as a vehicle for poverty reduction, as well as trying to reduce other sources of welfare loss, it has always been surrounded by many doubts. For a long period of time, literature on the effectiveness of foreign aid was mainly focused on aid allocations by traditional donors only. In recent years, the increased amount of aid from China as an emerging donor has been a source of growing speculation and debate. Opponents of Chinese aid claim that it is mainly driven by self-interest and therefore not conducive to economic development (Naím 2007; Lengaurer 2011). However, while it has already been intensively discussed whether Chinese aid is effective in alleviating poverty and stimulating sustainable economic growth, considerably less attention has been paid to other economic outcomes of aid. Given that China’s influence on foreign aid policy is likely to increase even further, more research on the effects of their aid practices is needed. This paper attempts to fill in this gap and addresses the following research question: Does Chinese aid have an effect on female labor force participation in recipient countries?
There exist two underlying motivations for this question. First, equal opportunities for men and women and employment and decent work for all have become part of the Sustainable Development Goals (SDGs) and can therefore be seen as an important source to achieve sustainable economic development. Second, particularly China’s increased interest in achieving global political and commercial goals – as importer of natural resource endowments – has been the reason for a critical change in criticisms concerning their aid practices. The latter may have serious consequences for female employment when their investments give rise to the more industrial sectors, as these are often dominated by men.
To test the hypotheses, panel data on Chinese aid allocations for 119 recipient countries from 2000 to 2014 is analyzed. Using fixed-effects regressions, I test the relationship between Chinese aid and female labor force participation, taking into account a broader definition of aid. The results suggest that Chinese only has significant effects on female employment for projects that lack development intent or the level of concessionality that is required. Additionally, projects that can be qualified as Official Development Assistance only show a significant correlation with female labor force participation when I experiment with subsamples by region to explore whether the results are driven by heterogeneous treatment effects. These findings shed new light on the much debated outcomes of Chinese aid and can therefore be a contribution to the Chinese aid effectiveness literature.
The remainder of this paper is structured as follows. Section 2 presents a conceptualization of the hypotheses supported by a discussion of the literature on other economic outcomes of Chinese aid. Section 3 provides details on the dataset and descriptive statistics on Chinese aid and female labor force participation. In section 4, the methodology will be introduced, and the results will be provided and discussed. The paper ends with a conclusion in section 5.
2. Conceptual framework
2.1 China as an emerging donorDespite several unprecedented efforts to meet the needs of the world’s poorest, to date the world still faces global challenges related to poverty, inequality, climate, environmental degredation, prosperity, peace and justice. In order to address those challenges, in 2015 the United Nations introduced seventeen Sustainable Development Goals (SDGs) that have to be achieved by 2030. One way to accomplish those goals is through Official Development Assistance (ODA), which can be defined as government-to-government financial aid transfers designed to promote the economic development and welfare of developing countries (OECD, 2019). The aid effectiveness literature distinguishes three motives for the allocation of aid, depending on i) the needs, ii) the quality of policies and institutions, and iii) the donor’s political or commercial self-interests (Dreher and Fuchs, 2012). Furthermore, even though aid can take several forms, it mostly comprises the transfer of money, products and/or services from one country to another (Lengauer, 2011).
a brief period of time; although until recently the country was relying on foreign aid, it now rivals the US in becoming worlds’ largest aid donor (Dreher, Fuchs, Parks, Strange and Tierney, 2017).
According to Naím (2007) and Lengaurer (2011), aid from China is not always guided by the needs of developed countries and can therefore be characterized as ‘rogue aid’. Instead, the Chinese government rather seems to abuse their power to benefit their own political interests and – most frequently mentioned as commercial motive for its aid – to attract access to natural resources. As worlds’ largest energy consumer, China can not rely on its own oil and gas reserves to sustain economic growth, and therefore most of Chinese aid donations are invested in countries that are rich in natural resources (Naím, 2007; Lengaurer, 2011). Others contradict these findings by arguing that political and commercial considerations are indeed an important determinant of China’s aid allocations, but no reason to believe that their allocations are distorted by this. Compared to traditional donors, China does not pay significantly more attention to their own interests such as the attraction of natural resource endowments ( Dreher and Fuchs, 2012). One could say that, where the explosion of Chinese aid allocations is followed by concerns over its donor practices on the one hand, others prais e the Chinese government for its willingness to help recipient countries and its ability to get things done in a very short amount of time (Dreher, Fuchs, Hodler, Parks, Raschky, Tierney, 2015; Isaksson and Kotsadam, 2018).
However, despite the fact that a wide range of existing literature already focused on the direct growth effects of Chinese aid allocations, the question whether and to what extent Chinese aid is contributing to a (de-)feminization of the labor force remains unanswered. Interestingly, while earlier studies suggested that gender inequality in education and therefore also in employment might increase economic growth (Barro and Lee, 1994; Barro and Sala-i-Martin, 1995), more recent work has shown that a reduction of the gender gap in the labor force can be seen as an important step fo rward in the process of economic development (Dollar and Gatti, 1999; Klasen and Lamanna, 2009; Forbes, 2000). With respect to the latter, a gender gap in employment might actually distort the economy as it reduces the range of talent employers can choose from, thereby reducing the average ability of the workforce (Klasen and Lamanna, 2009). Therefore, it might be interesting to see if, and to what extent, Chinese aid contributes to female labor force participation in recipient countries. Below I briefly report recent empirical findings on various economic outcomes driven by Chinese aid allocations.
2.2 The link between female labor force participation and Chinese aid
that Chinese aid projects “play a positive role in expanding the national economies of the recipient countries and improving the material and cultural life of the people in these countries” (Dreher and Fuchs, 2012). Moreover, with respect to the fact that Chinese aid is a lightning rod for critisism by scholars and policymakers stating that China abuses its power for its own political interests, the Ministry of Commerce of China admits that grants are used to coordinate diplomatic work and that this can be accompanied by great political outcomes. However, “they will never be used as a means to seek political privileges for itself” (Dreher and Fuchs, 2012). The Chinese ministry claims to help recipient countries developing their economy by projects that are intented to foster economic development for both China and these countries . In contrast to other Western donors, the quality of policies and institutions do not seem to matter in the allocation of aid by China (Dreher and Fuchs, 2012). However, only theoretical reasons are not enough, and those words by the Chinese ministry should be suited with its actions.
That Chinese aid indeed may have positive economic effects, becomes evident from the following studies. According to Bluhm et al. (2018), Chinese-financed development projects produce positive economic spillover effects that can be translated into a more equal distribution of economic activity in the concerned localities. By showing that African regions receiving financial support from China experienced a reduction in their Gini coefficient, which means that the distribution of income across regions has become more equal, their findings confirm that aid from China helps developing countries to reduce inequality through a more efficient spatial equilibrium. Moreover, findings by Dreher et al. (2017) suggest that Chinese aid not only tends to reduce the level of inequality, but that it is also positively related to economic growth. Particularly the latter findings are quite interesting for this topic of research. It has already been mentioned that female labor force participation may increase economic growth, but studies suggest that it can be other way around as well (Seguino, 2000; Standing, 1989, 1999).
change since it enables firms to enhance their productivity by adopting more advanced technologies or inducing a better allocation of production factors ( Gaddis and Pieters, 2017). If this technological change is skill-biased, and women and men differ in their core competences, they will become non-substitutable in the field of production which in turn will generate relative demand shifts by gender.
Although a clear relationship between Chinese aid and female labor force participation is still missing in existing literature, empirics have shown a positive relationship between aid allocations from China, economic inequality and growth, of which the latter is also positively correlated with the female share of the labor force (Seguino, 2000; Standing, 1989, 1999). These findings correspond with the findings by Dreher, Gehring and Klasen (2015), where they investigated whether donors give more aid to countries that experience a larger gap in education, health and women’s rights. They found some empirical evidence to believe that donors adjust their allocation of aid to reflect gender gaps in recipient countries.
Taking all the above together, it could be argued that these outcomes detected by Bluhm et al. (2018) and Dreher et al. (20152, 2017) should reduce the fears that scholars and policymakers have expressed about China acting as a “rogue aid” donor. In fact, Dreher et al. (2018) even state that China places much more emphasis on investments that complement the social and productive sectors than other Western donors, which actually may ease key constraints to economic growth and accelerate further growth (Dreher et al., 2017). Therefore, one could hypothesize that Chinese aid may also increase female labor force participation in recipient countries.
However, if we are to believe the other part of the literature, the positive effects of aid only take place under specific conditions or – even worse – not at all. A study on the income effects of aid depending on the quality of donors by Minasyan , Nunnenkamp and Richert (2017) has shown that aid will have stronger effects when donors provide their assistance in higher quality. The quality of aid can be improved in several ways, for instance by allocating aid in a more selective way that favors the poorer, better-governed recipients or penalizing project proliferation and tying aid (Roodman, 2012; Minasyan et al., 2017).
aid does not fulfill the conditions that are necessary to improve the quality of aid. Hence, one could hypothesize that Chinese aid hampers female labor force participation rather than reducing the gendered labor gaps when it is allocated to sectors that are dominated by men.
So far, it can be generally stated that evidence on the economic outcomes of aid is mired in controversy. Whereas some studies on Chinese aid show evidence of positive effects between the two (e.g. Bluhm et al., 2018 and Dreher et al., 2017) , others find that aid in general can accelerate growth only under specific conditions (e.g. Minasyan et al., 2017 and Burnside and Dollar, 2000). By contrast, there is also a widespread of literature that have found no relationship between aid and economic growth at all (Rajan and Subramanian, 2008; Doucouliagos and Paldam, 2009; Dreher and Langlotz, 2017).
To explain such null effects, one could argue that aid is measured imprecisely, or that the statistical power is too low for the estimators to find any significant effect (Dreher and Langlotz, 2017). Another reason one can think of when trying to explain the insignificant results, is that the positive and negative effects cancel each other out. As stated by Dreher and Langlotz (2017), donors pursue multiple objectives when granting aid, and economic growth is just one of them. If China pursues different priorities for each recipient country, the outcomes of the “true” aid effects may be affected. For instance, if we assume that China uses its development finance to gain access to natural resource endowments, countries that are rich in natural recources may experience an increase in female labor force participation, while others experience a decrease. If both effects take place at the same time, one might expect null effects. Therefore, it can also be hypothesized that there is no significant relationship between Chinese aid and female labor force participation.
3. Data and descriptive statistics
The full panel dataset in this study consists of data on 119 recipient countries and territories in East Asia and the Pacific, Europe and Central Asia, Latin America and the Caribbean, the Middle East and North Africa, South Asia and Sub -Saharan Africa for the years 2000–2014. The limiting factor is the allocation of aid from China which is not available for more years or countries.
Table 1 shows the number of recipient countries in the final dataset categorized by region and income, where the latter is classified based on an identification of the World Bank. As one can see, most recipient countries lie in Sub-Saharan Africa, which suggests that particularly countries in this region face massive development deficits. The high-income group is respresented by the lowest number of recipient countries, which can be explained by the fact that rich countries are more self-sufficient than countries that belong to the lower-income level and therefore less dependent on development assistance. Striking is that the dataset still contains a few high-income countries. With traditional aid, high-income countries generally do not belong to the group of countries that receive aid, but this seems to be different for Chinese aid2. Therefore, a list of recipient countries per income category is shown in the appendix (table A1). The rest of this section introduces the main variables that are used to construct the final dataset supported by some relevant descriptive statistics.
Table 1: Recipient countries by region and income category
Region Countries Income category Countries East Asia and Pacific 18 High income 9
Europe and Central Asia 20 Upper middle income 44 Latin America and Caribbean 23 Lower middle income 37 Middle East and North Africa 9 Low income 29
South Asia 7
Sub-Saharan Africa 42
3.1 Dependent variable
To empirically test whether and to what extent Chinese aid contributes to female employment in recipient countries, I use data on female labor force participation (FLFP) from the World Bank database as the dependent variable. This database captures annual data on the female economically active population for 264 countries and economies from 1990 to 2018. Fortunately, this means that the data on female labor force participation is highly overlapping with the data on China’s aid allocations.
2 According to the dataset, the high-income countries Argentina, the Bahamas and Seychelles receive
The World Bank defines economically active as “all people who supply labor for the production of goods and services during a specified period” (World Bank, 2018). This includes people who are currently employed, people who are curr ently unemployed but seeking a job, and first-time job-seekers. Unpaid workers, students and family workers are often omitted, as well as armed forces for some countries. It should be mentioned that the definitions of employment age may differ between countries. In most countries, the working age is set at 15 years and older, but in some countries children younger than 15 years are also economically active, which means that they are part of the estimates. If this is the case, the calculated female labor force participation rate may over- or underestimate the actual rates (World Bank, 2018). Table 1 has already shown that most recipient countries lie in Sub-Saharan Africa. To see if the female labor force employment rate also differs between regions, the share of females employed per region over the years 2000-2014 is depicted in figure 1. First of all, it shows that the percentage of economically active females was highest in Sub-Saharan Africa for the entire period. However, even more interesting is that the female labor force participation rate has slightly decreased in Europe and Central Asia, Ea st Asia and the Pacific and Sub-Saharan Africa. Only the Middle East and North Africa have shown an overall increase in female employment that is worth mentioning, but this is accompanied by some large increases and decreases.
Figure 1: Average changes in labor force participation per region, 2000-2014
20 30 40 50 60 70 F em al e la bo r fo rc e pa rt ic ip at io n ra te 2000 2005 2010 2015 Year
3.2 Independent variable
For the independent variable, I use the AidData Global Chinese Official Finance Dataset by Dreher et al. (2017) to collect data on China’s aid allocations. This dataset covers 5467 projects of official financing – including foreign aid and other forms of concessional and non-concessional state financing – from China to 138 recipient countries and territories in Africa, Asia and the Pacific, Central and Easte rn Europe, the Middle East and Latin America and the Carribean for the years 2000 –2014. A list of all countries receiving aid from China can be found in the appendix (table A2). The paper uses a broader definition of Chinese aid by including data on proje cts regarding Official Development Assistance as well as Other Official Flows, hereafter ODA and OOF. According to Dreher et al. (2017), “ODA-like” projects in the dataset can be qualified as ODA because they have the intention to promote economic or social development and are provided at concessional levels that meet the ODA criteria settled by the OECD-DAC. In contrast, Chinese-financed projects that are coded as “OOF-like” either have a non-developmental purpose or are not provided at sufficient concessional levels and can therefore not be qualified as ODA (Dreher et al., 2017). The reason why I include projects that are both ODA-like and OOF-like for the dependent variable, is because projects that are coded as OOF-like may also contain infrastructure projects that can be relevant for this research, even though they do not meet the official ODA requirements. For instance, OOF-like projects do not have a grant element of 25%. This creates the opportunity for China to secure political and commercial advantages – if we assume that their allocation of aid is primarily motivated by a desire to pursue its own interests.
To get rid of so-called umbrella projects and projects with a status of pledged, cancelled or suspended, only projects that are recommended for research are taken into account. It is important to exclude those kind of projects as they could represent double counting or would lead to including projects that never actually reached the official state of commitment (Dreher et al., 2017). Furthermore, for different regressions it could be interesting to see whether the results change when I distinguish between two different types of Chinese official finance. That is to say, by decomposing Chinese official finance and analyzing Chinese ODA and OOF seperately. With respect to projects that are coded as “Vague Official Finance”, it is unclear whether they qualify as ODA or OOF, and therefore they are excluded from the analysis.
development and population size, so that AIDCHN,i,t = AIDCHN,i,t GDP/capita,i,t
3. GDP per capita refers to the Gross Domestic Product divided by midyear population and is obtained from the World Bank. According to several researchers, a low level of GDP per capita is negatively related to female labor force participation (Brown, 2004; Dollar and Gatti, 1999). The reason for this is that, in countries or regions with a relatively low level of GDP per capita, families are more relying on the income of every family member, and therefore it is not affordable for the female family members to attend good education.
One of the main reasons why China’s financial actions receive so many attention is because of its rapid expansion of engagement in developing countries. Therefore, the total aid expenditures by the Chinese government over the years 2000-2014 are plotted in figure 2. It confirms that the Chinese aid expenditures have increased massively since the 2000s, with an enormous peak between 2008 and 2009.
Another reason for the increased attention is that scholars and policymakers are, to say the least, worried about the motives of Chinese aid as it will dampen the growth prospects of its recipient countries. To see if the allocation of aid by China differs from other donors, a comparison between the sectoral distribution of aid by China and traditional donors that are member of the Development Assistance Committee, from now on DAC, is presented in table 2. For instance, when Chinese aid is hampering female labor force participation in recipient countries, one might expect it to be allocated to the more male-dominated sectors, and vice versa.
Depending on their OECD 2-digit sector code, sectors are divided into three categories: social sectors, private sectors and other sectors. In the appendix, a list of the sectors belonging to each category is attached (table A3). As one can see, sectors that are considered as male-dominated, such as transport and storage, industry, mining and construction and energy generation and supply, are classified as private sectors. Furthermore, besides the total share of aid allocations by category for Chinese aid (i.e. ODA+OOF) and DAC, the categorical share of aid allocations is also presented for ODA and OOF seperately.
Overall, table 2 presents that most Chinese aid is allocated to the sectors that are coded as private, while DAC donors place much more emphasis on investments that complement the social sectors. The seperate columns that represent the share of ODA by China and OOF by China to each category show that more than 90% of the investments coded as OOF are allocated to private sectors, while this is a bit more even distributed for ODA. This confirms that it may be interesting to use seperate
measures as well for ODA and OOF, and to explore whether this will have different effects on female labor force participation.
Figure 2: Total aid expenditures by China, 2000-2014
Table 2: Sectoral distribution of aid
Chinese aid Traditional aid
Category ODA OOF Total DAC Social 14.5% 3.3% 7.6% 32.4% Private 56.5% 90.4% 83.6% 18.2% Other 29.0% 6.3% 8.8% 49.4%
3.3 Control variables
This section describes the control variables that are taken into account as they may influence the relationship between Chinese aid and female labor force participation. The summary statistics and pairwise correlations for all variables, including the
dependent and independent variable, are displayed in table 3 and 44. More details on variable definitions and sources are provided in the appendix, table A4.
With respect to the pairwise correlations, table 4 shows that there is a very small positive correlation between Chinese aid and female labor force participation, however, the size is almost negligable. A graph that shows how Chinese aid is correlated with the female labor force participation rate in different regions is demonstrated in figure 3. Although for most regions the correlations seem to fluctuate over the years, it remains relatively stable and close to zero for Sub -Saharan Africa. The first control variable that is commonly used in literature on gendered -labor effects is fertility rate, and is expressed by the total amount of births per women. According to Bussmann (2009), fertility rate is an important control variable as it not on ly controls for population growth, but also gives an indication of the extent to which women are busy with raising their children, implying that they will have less time for a job or to attend school.
Second, to proxy changes in the proportion of a country’s population that is employed, annual data on the employment to population ratio for each country is included. It is important to control for the employment rate as changes in a country’s employment population may also affect the female share of the labor force. Important to note is that the correlation between employment rate and female labor force pa rticipation is relatively strong, which means that I should take this variable with some caution. Next, data on the poverty headcount ratio controls for the degree of poverty within a country. For instance, in countries with a high level of poverty, it may be more necessary that all females have a job than in countries with a lower poverty headcount ratio. According to table 4, there is a moderate correlation between the poverty headcount ratio and female labor force participation.
The third control variable is the political regime type. This variable controls for the influence of institutional characteristics on the female share of the workforce, where the index ranges from -10 for pure autocratic institutions and +10 for pure democratic institutions. Literature suggests that democratic institutions tend to increase female labor force participation because they have a more open recruitment process that allows access for female politicians, who in turn create opportunities for other women once they are in function (Bussmann, 2009).
as an upper bound, because there is usually a large gap between male and female education in developing countries. Data on the expected years of schooling for males can be used as a way to control for the upper bound level of education in a country. Lastly, data on aid allocations by traditional donors is included because some of the effects may be explained by aid allocations from DAC donors instead of allocations by China only. As with the dependent variable, DAC is divided by GDP per capita, and projects that contain negative numbers are dropped from the analysis because it is unclear why these are negative.
Table 3: Summary statistics
Figure 3: Correlation between Chinese aid and female labor force participation (by region)
4. Econometric analysis
4.1 Methodology
My empirical approach follows existing literature on other economic outcomes of foreign aid by estimating a linear regression model where female labor force participation is the dependent variable. This regression is displayed in equation (1), which represents the main econometric model of this paper:
(1) FLFPi,t = 0 + 1AIDCHN,i,t-3 + 2controlsi,t(-3) + i + t + i,t,
where FLFPi,t is the female labor force participation rate of country i in year t, 1AIDCHN,i,t-3 is the amount of aid from China divided by GDP per capita that is
allocated to country i in year t-3, 2controli,t(-3) denotes the set of seven control
variables for country i in year t (and year t-3 for DAC), i and t represent country-
and year-fixed effects, respectively, and is the error term.
The dataset consists of two levels of cross-sections: countries i=1,…,N and years t=2000,…,2014. To conduct a more precise analysis and to control for both country specific and time invariant characteristics, I use the effects model. The
fixed--1 -. 5 0 .5 1 C hi ne se a id a nd f em al e la bo r fo rc e pa rt ic ip at io n 2000 2005 2010 2015 Year
effects model controls for all unobserved time-invariant characteristics that are constant over time and can not be captured by the control variables, i.e. unobserved heterogeneity at the country level. For instance, some religions that are more present in one country than another prohibit women to have a job. Not controlling for this unobserved heterogeneity will bias the results if it is correlated with one of the variables on the right-hand side of the equation. Furthermore, the fixed-effects model also controls for differences across time. By including time-fixed effects, my model captures the influence of crises or other global effects on female labor force participation. Without controlling for such time invariant characteristics, again, results may appear to be biased or inconsistent.
A methodological issue could be an endogeneity problem, which means that the independent variable is endogenous and correlated with the error term . In general, endogeneity can occur through three potential causes: reverse causality or simultaneity, omitted variables and a measurement error. Arguably, reverse causality does not seem to be a threat in this paper as there is no reason to assume that female labor force participation affects Chinese aid. However, it could be argued that foreign aid projects need some time to be realized and to achieve an impact. By lagging Chinese aid with three years, both reverse causality concerns will be addressed. In other words, the dependent and independent variable can not simultaneously affect each other, because the independent variable is measured three years before the allocation appears.
With respect to the omitted variable bias, there may be some omitted variables that can not be captured by the fixed effects. The fixed-effects model only controls for country-level omitted variables that do not change over time. This could mean that subnational differences are not taken into account, or that time -varying and non-linear effects for each country over time are not included in the model. For instance, Chinese aid may have led to a sort of structural transformation within countries when it is allocated to specific sectors and more people start working in this particular sector. Not being able to control for such non-constant, unobserved country-specific confounders can be a problem as it could lead to null effects or biased results, which is a limitation for most cross-country studies.
4.2 Results
Although there are no other studies that investigated the direct relationship between Chinese aid and female labor force participation so far, one could argue that this consistent with the null-effect findings by Rajan and Subramanian (2008) and Dreher and Langlotz (2017) on other economic outcomes of aid. Both papers found no significant relationship between aid and economic growth. If Chinese aid is also insignificant with economic growth, then it is expected to see null effects for female labor force participation as well because of its positive correlation with economic growth (Klasen and Lamanna, 2009; Seguino, 2000).
In the remaining models of table 5, the seven control variables are added one by one. While some of the variables are insignificant at conventional levels, the coeff icients on the employment rate, poverty headcount and political regime type show positive signs and are at least significant at the 10% level in most models. Secondary education for men is only significant in one of the models. Either way, even after including all seven control variables, the overall results remain the same compared to the basic model in column (1). In other words, I find no significant evidence that Chinese aid is related to female labor force participation in recipient countries.
As stated in section 3, normally aid goes only to developing countries. This is bec ause they are most in need of assistance to obtain sustainable development, whether achieved through economic growth or other means. High-income countries, on the other hand, generally have a relatively high GDP per capita and receive only small amounts of aid. Therefore, one might suspect the results to be biased by the high-income countries that are included in the analysis. This is not the case, however. In the first column of table A5 in the appendix, I exclude high-income countries from the sample of recipient countries to see if they may have influenced the baseline estimations. Apart from some small changes in the size of the coefficients, no significant differences were found compared to the baseline regressions presented in table 5.
The analysis continues in table 6 where I re-estimate the baseline regressions by using seperate measures for ODA and OOF. Table 2 in section 3 already showcased that the sectoral distribution of aid differs between ODA and OOF. Thus, one could argue t hat it may be interesting to explore if the outcomes change when I use a more strict definition of aid. The first column of table 6 shows that the estimation parameter for ODA is positive but insignificant, which means that there is no significant evidence to assume that ODA from China affects female labor force participation. Interestingly, the estimation parameter for OOF in column (2) reflects a positive relationship with female labor force participation and is significant at the 1% significance level. The coefficient shows that, when Chinese OOF increase by 1%, female labor force participation increases on average by 1.68%, ceteris paribus.
Furthermore, though the impact of aid by traditional donors is not the main topic of interest in this paper, the third column in table 6 shows the same regressions as for column (1) and (2), but with DAC per capita as explanatory variable and Chinese aid as control variable. When I would have measured the effects for Chinese aid only, no conclusions on China’s intentions specifically can be drawn as it can not be compared with the effects for traditional donors. Therefore, in column (3), table 6 I test whether the results are different for traditional donors. The insignificant coefficients reveal that traditional aid has, as with Chinese aid, no significant impact on female labor force participation5. In other words, I find no empirical evidence that the effects for Chinese aid are different from that of traditional donors.
The robustness of the main results in column (2), table 6 is tested in several ways. First of all, I consider different delayed effects of Chinese OOF on the female labor force participation rate in recipient countries. Table A6 in the appendix shows the results without lags in column (1), with lags of one year in column (2) and two years in column (3). The estimation results reveal that the effects of aid on female labor force participation were generally stronger when using three-year lags; only column (2) shows significant results at the 5% level. Important to note is that the size of the coefficient decreases from 1.68% in column (2), table 6 to 0.48% in column (2), table A6. In other words, a 1% increase in Chinese OOF increases female labor force participation to a lesser extent when using one-year lags instead of three-year lags. Arguably, the variation in significance and significance levels for different lags indicate that it may take a few years before aid-projects achieve an impact and a visible effect on female employment in recipient countries.
Second, the regression in the last column of table A6 is basically the sam e as before, but now I estimate the model using random effects instead of fixed effects. Compared to the fixed-effects model, the random-effects model has one additional assumption: individual-specific effects are uncorrelated with the independent variables.
5Even after replacing the negative values on traditional aid allocations, which were initially dropped
Importantly, column (4) shows that the positive relationship between Chinese OOF and female labor force participation continues to be statistically signi ficant at the 5% level. In other words, I still find that Chinese OOF is likely to have positive effects on female labor force participation in recipient countries when using the random -effects model.
Finally, for the same reasons as described earlier, the latter two robustness checks in table A6 test whether the results remain robust when high-income countries are excluded from the analysis (column (5)) or when the sample size is less restricted by poverty headcount as a control variable (column (6)). While the positive results continue to be positively significant in column (5), albeit only at the 10% significance level, this is less so for column (6) in which I do not control for poverty headcount. Important to note is that the R2 declines considerably in the last column, which means that less variance in the dependent variable can be explained by the independent variables collectively.
The results in column (2), table 6, which to some extent remain robust after several robustness tests, contradict the hypothesis that Chinese aid decreases female employment when it is allocated to private sectors that are often male -dominated. An explanation for these unexpected findings could be as follows. The increase of investments in private sectors could have led to industrialization. That is to say, people who were initially working in argicultural sectors, now move to jobs related to industry and services. It could be argued that, while OOF from China is mainly allocated to male-dominated jobs in industry sectors, this also generates new jobs in other services sectors through sectoral adjustment, which I couldn’t capture. If those services sectors contain a lot of females participating in the labor force, one could argue that OOF from China promotes female employment.
Table 6: Re-estimation table 5, seperate regressions for ODA and OOF + DAC FLFP (1) (2) (3) CH ODA by GDP/capita (-3) 0.090 (0.380) CH OOF by GDP/capita (-3) 1.677*** (0.405) DAC by GDP/capita (-3) -0.038 -0.554* -0.050 (0.172) (0.240) (0.143) Fertility rate -0.791 -2.066* 0.183 (0.908) (0.905) (0.811) Employment rate 1.170*** 1.523*** 1.160*** (0.072) (0.065) (0.068) Poverty headcount 0.094*** 0.261** 0.107*** (0.020) (0.072) (0.025) Political regime type 0.536*** -0.092 0.432*** (0.139) (0.228) (0.123) Secondary educ(f) -0.047 -0.112 -0.017 (0.050) (0.056) (0.054) Secondary educ(m) 0.113* 0.027 0.078 (0.052) (0.030) (0.048) CH aid by GDP/capita (-3) 0.288 (0.289) Constant -23.311*** -31.875*** -25.277*** (3.827) (5.297) (4.128) Adj R2 0.90 0.96 0.89 Observations 95 50 114
Year dummies Yes Yes Yes
Fixed effects Yes Yes Yes
Note: ***, **, * denote statistical significance at the 1%, 5% and 10% level, respectively. All models involve OLS regressions that include clustered standard errors at the country level, which are presented in brackets.
Lastly, in table 7, I conduct a subsample analysis to explore if the insignificant results for Chinese ODA as presented in column (1), table 6 are driven by other factors6. As can be seen in the left-hand column of table 1, the number of recipient countries per region is considerably diverging. To test if the results are subject to any heterogeneous treatment effects across different regions, I conduct the same regressions as with equation (1) but for each region seperately.
The results indicate that the relationship between ODA from China and female labor force participation remains insignificant for all regions, except for Europe and Central Asia in the second column of table 7. In this region, the estimation coefficient turns out to be positive and significant at the 5% significance level. In marginal terms, there is evidence showing that a 1% increase in Chinese ODA may increase female labor force participation by 1.95% on average, ceteris paribus, in Europe and Central Asia. The findings in column (2), table 7 are quite surprising because figure 1 in section 3 revealed that the female labor force participation rate in Europe and Central Asia has
6 Unfortunately, I was not able to conduct a similar subsample analysis for OOF from China as there
slightly decreased over the 2000-2014 period. In other words, given its positive significant relationship with female labor force participation in Europe and Central Asia, it could be cautiously concluded that ODA from China is not the cause for the decrease in female employment in this region. However, when the data is combined, that is with all regions together, there are no significant effects. Noteworthy, the sign and the size of the insignificant coefficients vary considerably across regions. This corresponds to the hypothesis that the appearance of null effects can be subsequent to positive and negative effects cancelling each other out.
Table 7: Heterogeneity analysis by region7
FLFP (1) EA&P (2) E&CA (3) LA&C (4) ME&NA (5) SA (6) SSA CH ODA by GDP/capita (-3) -0.029 1.946** -5.383 122.123 0.614 -0.262 (0.286) (0.580) (40.417) (49.694) (0.282) (0.272) Fertility rate -1.214 4.415 -6.473 -1.783 2.862* -2.792 (1.589) (2.633) (14.613) (2.965) (0.736) (1.941) Employment rate 1.431*** 1.018*** 1.269*** 0.807*** 1.834* 0.925*** (0.137) (0.212) (0.228) (0.101) (0.527) (0.136) Political regime type -0.053 0.867 1.460 0.215 0.182 0.079 (0.072) (0.596) (3.205) (0.171) (0.069) (0.085) Secondary educ(f) -0.235 -0.010 0.161 0.257 -0.041 0.084 (0.164) (0.188) (0.176) (0.228) (0.013) (0.082) Secondary educ(m) 0.115 0.171 -0.071 -0.240 0.226 -0.114 (0.146) (0.134) (0.288) (0.275) (0.085) (0.063) DAC by GDP/capita (-3) 0.041 -5.861 0.437 2.106 -0.053 0.047 (0.686 (3.009) (0.631) (4.398) (0.039) (0.044) Constant -26.734** -25.491 -29.900 -4.166 -89.535 19.333 (7.383) (19.529) (20.779) (11.388) (35.065) (17.484) Adj R2 0.88 0.82 0.88 0.88 0.96 0.67 Observations 44 38 27 23 24 176
Year dummies Yes Yes Yes Yes Yes Yes Fixed effects Yes Yes Yes Yes Yes Yes Note: ***, **, * denote statistical significance at the 1%, 5% and 10% level, respectively. All models involve OLS regressions that include clustered standard errors at the country level, which are presented in brackets.
5. Conclusion
Compared to the extensively debated role of Chinese aid on economic growth, the question whether and to what extent Chinese aid contributes to other economic outcomes has received only scant attention. Therefore, this paper investigates the following research question: Does Chinese aid have an effect on female labor force participation in recipient countries? For this purpose, I estimate a fixed-effects linear regression model with female labor force participation as dependent variable and Chinese aid as independent variable. The full panel dataset covers data on Chinese aid allocations to 119 recipient countries and territories in five major world regions over the 2000-2014 period.
The analysis shows the following main findings. In its broadest definition, I find no epirical evidence to believe that Chinese aid is related to female labor force participation. Even when the sample size is less restricted by one of the control variables, or when high-income countries are not taken into account, the insignificant results hold. Importantly, additional tests on the same group of recipient countries have shown that aid from traditional donors is also not related to female employment. The results for China become different when I start using a more strict definition of Chinese aid by using seperate definitions for ODA and OOF. The latter analysis shows that, on average, a 1% increase in OOF from China leads to a 1.68% increase of female employment in recipient countries. Noteworthy, these findings only hold for some of the robustness checks. In addition, ODA from China remains insigni ficant at conventional levels. To test whether these insignificant results can be driven by heterogenous treatment effects, I conduct a subsample analysis for each region. The results reveal that ODA from China indeed does not affect female labor force participation in most regions, however, this differs for Europe and Central Asia. In this region, female labor force participation increases by 1.95% when ODA from China increases by 1%.
not be captured by the control variables. This could mean that the model still contains some non-constant country-specific unobserved confounders influencing the error term, which is a limitation for most cross-country studies. Third, the definition of employment age is somewhat inconsistent across countries. This means that the calculated female labor force participation rate might be slightly over - or underestimated for some recipient countries.
Additionally, this paper gives rise to further questions. Future research can address the question whether Chinese aid might have different effects for other economic outcomes. In comparison with traditional aid, Chinese aid has received mounting criticisms concerning their aid practices. While existing literature has already paid a lot of attention to the effects of Chinese aid on economic growth, more research on other outcomes of aid needs to be done before one can confirm or deny those criticisms. Especially since this study has found no empirical relationship between aid from traditional donors and female labor force participation either. Furthermore, the ability to vary with regressions in this paper is limited by the availability of data. When more data on Chinese development finance is available, further research could provide a fuller understanding of how and why Chinese aid has differential effects between regions or income groups.
7. References
Anderson, Edward (2005). Liberalization and inequality in developing countries: A review of theory and recent evidence. World Development, 33, pp. 1045-1063.
Berlin, M.P., Bonnier, E. and Olofsgård, A. (2018). The donor footprint and gender gaps. WIDER Working Paper #130.
Barro, R. and Lee, J. (1994). “Sources of economic growth”. Carnegie-Rochester Conference Series on Public Policy 40(1), pp. 1-46.
Barro, R. and Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw-Hill. Black, S.E. and Brainerd, E. (2004). Importing equality? The impact of globalization
on gender discrimination. Industrial and Labor Relations Review 57(4), pp. 540-559.
Bluhm, R., Dreher, A., Fuchs, A. Parks, B., Strange, A. and Tierney, M. (2018). Connective financing: Chinese infrastructure projects and the diffusion of economic activity in developing countries. AidData Working paper #64. Williamsburg, VA: AidData.
Brown, D.S. (2004). Democracy and gender inequality in education: Across-national examination. British Journal of Political Science 34, pp. 137-192.
Burnside, C. and Dollar, D. (2000). Aid, policies and growth. The American Economic Review 90(4), pp. 847-868.
Bussmann, M. (2009). The effect of trade openness on women’s welfare and work life. World Development 37(6), pp. 1027-1038.
Dollar D., and Gatti, R. (1999). Gender inequality, income, and growth: Are good times good for women? Policy Research Report on on Gender and Development Working Paper Series Working Paper #1. Washington, DC: World Bank. Doucouliagos, H. and Paldam, M. (2009). The aid effectiveness literature: The sad
results of 40 years of research. Journal of Economic Serveys 23(3), pp. 433-461.
Dreher, A., Fuchs, A., Hodler, R. Parks, B.C., Raschky, P.A. and Tierney, M.J. (2015). Aid on demand: African leaders and the geography of China’s foreign assistance. CESifo Working Paper #5439
Dreher, A. Fuchs, A., Parks, B.C., Strange, A.M. and Tierney, M.J. (2016). Apples and dragon fruits: The determinants of aid on other forms of state financing from China to Africa. International Studies Quarterly 52, pp. 182-194.
Dreher, A., Fuchs, A., Parks, B.C., Strange, A. M. and Tierney, M. J. (2017). Aid, China, and growth: Evidence from a new global development finance dataset. AidData Working Paper #46. Williamsburg, VA: AidData.
Dreher, A., Gehring, K. and Klasen, S. (20152). Gesture politics or real commitment? Gender inequality and the allocation of aid. World Development 70(1), pp. 464-480.
Dreher, A. and Langlotz, S. (2017). Aid and growth. New evidence using an excludable instrument. CESifo Working Paper #5515.
Forbes, K. (2000). A Reassessment of the Relationship between Inequality and Growth. American Economic Review 90(4), pp. 869–87.
Gaddis, I. and Pieters, J. (2017). The gendered labor market impacts of trade liberalization: Evidence from Brazil. Journal of Human Resources 52(2), pp. 457-490.
Isaksson, A. and Kotsadam, A. (2018). Chinese aid and local corruption. Journal of Public Economics 159, pp. 146-159.
Klasen, S. And Lamanna, F. (2009). The impact of gender inequality in education and employment on economic growth: New evidence for a penal of countries. Feminist Economics 15(3), pp. 91-132.
Lengauer, S. (2011). China’s foreign aid policy: Motive and method. Culture Mandala: The Bulletin of the Centre for East-West Cultural and Economic Studies 9(2), pp. 1-81.
Marshall, M. G., Gurr, T.R. and Jaggers, K. (2013). Polity IV Project: Political regime characteristics and transitions, 1800-2012. Vienna, VA: Center for Systemic Peace.
Naím, M. (2007). Rogue Aid. Foreign Policy 159, March/April, pp. 95-95.
OECD (2019). Net ODA (indicator). Available at: https://data.oecd.org/oda/net-oda.htm
Rajan, R. G. And Subramanian, A. (2008). Aid and growth. Review of Economics and Statistics 90(4), pp. 643-665.
Roodman, D. (2012). An index of donor performance. Working Paper #67. Center for Global Development. Washington, DC.
Seguino, S. (2000). Gender inequality and economic growth: A cross -country analysis. World Development 28(7), pp. 1211-1230.
Standing, G. (1989). Global feminization through flexible labor. World Development 17(7), pp. 1077-1096.
Standing, G. (1999). Global feminization through flexible labor: a theme revisited. World Development 27(3), pp. 583-602.
8. Appendix
Table A1: List of recipient countries included per income category
High income Upper middle income Lower middle income Low income
Antigua & Barbuda
Albania Angola Afghanistan
Argentina Algeria Bangladesh Benin
Bahamas Armenia Bolivia Burundi
Barbados Azerbaijan Cambodia Chad
Chile Belarus Cameroon Comoros
New Zealand Bosnia-Herzegovina Cape Verde Congo, Dem. Rep.
Seychelles Botswana Congo, Rep. Eritrea
Trinidad & Tobago Brazil Cote D’Ivoire Ethiopia
Uruguay Bulgaria Djibouti Guinea
Colombia Georgia Guinea-Bissau
Costa Rica Ghana Haiti
Cuba India Liberia
Dominica Indonesia Madagascar
Ecuador Kenya Malawi
Equatorial Guinea Kyrgyz Republic Mali
Fiji Laos Mozambique
Gabon Lesotho Nepal
Grenada Mauritiana Niger
Guyana Micronesia, Fed.
States of Rwanda
Iraq Moldava Senegal
Jamaica Mongolia Sierra Leone
Jordan Morocco Somalia
Kazakhstan Myanmar South Sudan
Lebanon Nicaragua Syria
Libya Nigeria Tajikistan
Macedonia, FYR Pakistan Tanzania Malaysia Papua New Guinea Togo
Maldives Philippines Uganda
Mauritius Sri Lanka Zimbabwe
Mexico Sudan Montenegro Timor-Leste Namibia Tunisia Nauru Ukraine Peru Uzbekistan Romania Vanuatu
Russia Viet Nam
Samoa Zambia
Table A2: List of countries receiving aid from China
Afghanistan, Albania, Algeria, Angola, Armenia, Azerbaijan, Bangladesh, Barbados, Belarus, Benin, Bolivia, Bosnia-Herzegovina, Botswana, Brazil, Burundi, Cambodia, Cameroon, Cape Verde, Central African Rep., Chad, Chile, Colombia, Comoros, Congo, Dem., Congo, Rep., Costa Rica, Cote D’Ivoire, Cuba, Djibouti, Dominica, Ecuador, Egypt, Equatorial Guinea, Eritrea, Ethiopia, Fiji, Gabon, Georgia, Ghana, Grenada, Guinea, Guinea-Bisseau, Guyana, Haiti, India, Indonesia, Iran, Iraq, Jamaica, Jordan, Kazakhstan, Kenya, Korea, Dem., Kyrgyz Republic, Laos, Lebanon, Lesotho, Liberia, Libya, Macedonia, Madagascar, Malawi, Malaysia, Maldives, Mali, Mauritania, Mauritius, Mexico, Federated States of Micronesia, Moldava, Mongolia, Montenegro, Morocco, Mozambique, Myanmar, Namibia, Nauru, Nepal, Nicaragua, Niger, Nigeria, Niue, Pakistan, Palestinian Adm. Areas, Papua New Guinea, Peru, Philippines, Russia, Rwanda, Samoa, Senegal, Serbia, Seychelles, Sierra Leone, Somalia, South Africa, South Sudan, Sri Lanka, St. Lucia, Sudan, Suriname, Syria, Tajikistan, Tanzania, Thailand, Timor-Leste, Togo, Tonga, Trinidad & Tobago, Tunisia, Turkey, Turkmenistan, Uganda, Ukraine, Uruguay, Uzbekistan, Vanuatu, Venezuela, Vietnam, Yemen, Zambia, Zimbabwe.
Table A3: Sectors per category
Category Sectorname Sectorcode
Social Education 110
Health 120
Population Policies / Programmes and Reproductive Health 130
Water Supply and Sanitation 140
Government and Civil Society 150
Other Social infrastructure and services 160
Private Transport and Storage 210
Communications 220
Energy Gernation and Supply 230
Banking and Financial Services 240
Business and Other Services 250
Agriculture, Forestry and Fishing 310
Industry, Mining and Construction 320
Trade and Tourism 330
Other General Environmental Protection 410
Women in Development 420
Other Multisector 430
General Budget Support 510
Developmental Food Aid / Food Security Assistance 520
Non-food commodity assistance 530
Action Related to Debt 600
Emergency Response 700
Support to Non-governmental Organizations (NGOs) and Government Organizations
920
Table A4: Variable definitions and sources
Variable Definition Source
Female labor force participation
“The female proportion of the population aged 15 and older that is economically active: all females who supply labor for the production of goods and services during a specified period.”
The World Bank
Chinese aid “The known universe of officially-financed Chinese projects in 5 regions of the world from 2000-2014, referring to concessional and non-concessional sources of funding from Chinese government institutions (including central, state or local government institutions) with development, commercial, or representational intent.”
AidData (Dreher et al., 2017)
GDP per capita “The gross domestic product divided by midyear population. Data are in current U.S. dollars.”
The World Bank
Fertility rate “The number of children that would be born to a woman if she were to live to the end of her childbearing years and bear children in accordance with age-specific fertility rates of the specified year.”
The World Bank
Employment rate “The proportion of a country's population that is employed. Employment is defined as persons of working age who, during a short reference period, were engaged in any activity to produce goods or provide services for pay or profit, whether at work during the reference period or not at work due to temporary absence from a job, or to working-time arrangements. Ages 15 and older are generally considered to the working-age population.”
The World Bank
Poverty headcount ratio
“The percentage of population living on less than $1.90 a day at 2011 international prices.”
The World Bank
Political regime type “The political structure that make up the country. The data combines various institutional characteristics of a political system to an index ranging from -10 for pure autocracies and +10 for pure democracies.”
Secondary school enrollment rates (f/m)
“The the ratio of total enrollment, regardless of age, to the population of the age group that officially corresponds to the level of education shown. Secondary education completes the provision of basic education that began at the primary level, and aims at laying the foundations for lifelong learning and human development, by offering more subject- or skill-oriented instruction using more specialized teachers.”
The World Bank
Aid from traditional donors
“Goverment aid designed to promote the economic development and welfare of developing countries. May be provided bilaterally, from donor to recipient, or channelled through a multilateral development agency such as the United Nations or the World Bank. Aid includes grants, “soft loans” (where the grant element is at least 25% of the total) and the provision of technical assistance.”
Table A5: Robustness tests for Table 5, without high-income countries (1) and poverty headcount (2)
Note: ***, **, * denote statistical significance at the 1%, 5% and 10% level, respectively. All models involve OLS regressions that include standard errors clustered at the country level, which are presented in brackets.
FLFP (1) (2) Ch aid by GDP/capita (-3) 0.301 0.037 (0.289) (0.131) Fertility rate 0.232 -0.850 (0.801) (0.884) Employment rate 1.153*** 1.028*** (0.068) (0.092) Poverty headcount 0.105*** (0.026)
Political regime type 0.422** 0.069 (0.130) (0.056) Secondary educ(f) 0.001 0.017 (0.064) (0.061) Secondary educ(m) 0.068 -0.033 (0.054) (0.057 DAC by GDP/capita (-3) -0.024 0.049 (0.150) (0.038) Constant -25.466*** -2.880 (4.033) (6.753) Adj R2 0.89 0.61 Observations 106 365
Year dummies Yes Yes
Table A6: Robustness tests for Table 5, using different time lags8 (1-3), random effects (4) and without high-income countries (5) and poverty headcount (6)
FLFP (1) (2) (3) (4) (5) (6) CH OOF by GDP/capita 0.721 0.480** -0.081 1.330** 1.065* 0.019 (0.864) (0.145) (0.153) (1.203) (0.470) (0.193) Fertility rate -1.655 1.179*** -0.440 -1.261 -1.894* 0.581 (3.191) (2.355) (2.311) (1.139) (0.952) (0.907) Employment rate 1.238*** 1.065*** 1.178*** 1.306*** 1.365*** 1.169*** (0.168) (0.150) (0.137) (0.089) (0.063) (0.071) Poverty headcount 0.162 -0.309*** -0.037 0.289* 0.278*** (0.081) (0.050) (0.070) (0.116) (0.083) Political regime type 0.050 0.176*** 0.484** -0.348* 0.044 0.106 (0.310) (0.034) (0.163) (0.156) (0.223) (0.097) Secondary educ(f) 0.176 0.203 -0.013 -0.113 -0.029 0.086 (0.140) (0.111) (0.102) (0.069) (0.052) (0.090) Secondary educ(m) -0.151 -0.142 -0.035 0.121 0.025 -0.065 (0.121) (0.088) (0.110) (0.071) (0.062) (0.097) DAC by GDP/capita -0.092 3.705*** 0.671*** -0.436 -0.284 -0.002 (0.571) (0.417) (0.158) (0.247) (0.300) (0.083) Constant -26.201 -29.251*** -15.870** -26.939*** -29.031*** -20.247*** (13.639) (1.839) (4.943) (6.698) (5.940) (4.969) Adj R2 0.91 0.98 0.95 0.97 0.96 0.69 Observations 64 52 56 50 48 134
Year dummies Yes Yes Yes Yes Yes Yes Fixed effects Yes Yes Yes No Yes Yes
Random effects No No No Yes No No
Aid lagged No Yes Yes Yes Yes Yes
Aid twice lagged No No Yes No Yes Yes Aid thrice
lagged No No No Yes Yes Yes
Note: ***, **, * denote statistical significance at the 1%, 5% and 10% level, respectively. All models involve OLS regressions that include standard errors clustered at the country level, which are presented in brackets.
