Abstract

US public diplomacy efforts claim to promote democracy and peace via direct engagement with foreign publics. Aid initiatives that target civil society aim to increase direct cultural and educational ties between host societies and the US, and thereby contribute to democratic socialisation and democratic norm infusion in the developing world.

In this paper, we analyze whether US’ public diplomacy-related aid contributes to improving the level of democracy, human development and peace in the recipient countries. We employ panel data analysis taking into account alternative explanations for determinants of democratization and control variables. Our findings shed light on the efficacy of these public diplomacy-oriented channels of aid that deviate from the traditional forms of economic, disaster and military aid that already populate the literature.

We also uncover the connection between public diplomacy aid and the humanitarian-development-peace nexus literature.

Liberal Peace theory

The proposed paper will examine the role of US civil society-oriented aid and whether it has an effect on democratic outcomes.

Additionally, the paper will examine whether there is a regional difference in the efficacy of civil society across different world regions.

Since Immanual Kant, liberal theories argue that as states internalise and socialse to democratic rule, they will seek out like-minded states that organise in a similar way. This logic argues that these like-minded countries will in turn create a ‘league of peace’ that ensures countries are more secure in their international interactions.Subsequently, these interactions can act as a vehicle to further spread these desirable norms of peace, democracy and integration.

We see the lineage of this logic present in US President Woodrow Wilson’s post-WWII Fourteen Point plan that aimed to spread democratic rule across the world (especially in post-war countries) and to formalise the idea of international integration via the League of Nations. All these plans would thus result in a Kantian peace. In more recent years democratic peace theory has become popularised by works that empirically support the pacifying effects of democratic rule and the peace the characterised interactions between democratic states (Russett & Oneal, 1999).

US foreign policy strategy: USAID and civil society aid to promote democracy

The premise of democracy promotion as a foreign policy strategy is not new to the post-Cold War era (see Smith 1994), nor is it exclusive to the United States (see Burnell 2000; Youngs 2002; Collins 2009). However, in the last two decades, democracy promotion has become an increasingly prominent component of US foreign policy (Carothers 1999; Ikenberry 2000).

Since the advance of “the third wave” of democratization (Huntington 1991) since 1975 have led to greater attention to the kind of regimes the United States promotes and supports, with issues of human rights and democracy becoming increasingly important political factors (see Meernik, Krueger, and Poe 1998).

Democratic sponsor states such as the United States have compelling reasons for promoting democracy abroad, including the implications of the democratic peace research program and democracy’s better outcomes in terms of human rights and economic performance (Art 2003).2 Building on such rationales, post-Cold War American presidents from Bill Clinton to Barack Obama have all embraced the goal in one way or another.

As Finkel, Perez-Linan, Seligson, and Azpuru (2006:26) detail, USAID has allocated increasingly larger amounts to its “Democracy and Governance” initiatives around the world.

Carothers (2009) outlines two overall approaches to democracy support (see also Carothers 1999, 2015). On the one hand, the political approach, associated especially with US democracy assistance, proceeds from a relatively narrow conception of democracy—focused, above all, on elections and political and civil rights—and a view of democratization as a process of political struggle in which democrats work to gain the upper hand over nondemocrats in society. It directs aid at core political processes and institutions—especially elections, political parties, and politically oriented civil society groups—often at important conjunctural moments and with the hope of catalytic effects (p. 5).

The U.S. Agency for International Development (USAID) identifies two democracy-specific development objectives within their “Democracy, Human Rights, and Governance” strategy framework. The first objective is to “promote participatory, representative and inclusive political processes and government institutions” (United States Agency for International Development, 2013). This includes activities that support the implementation of participatory political processes by state institutions, including advising, training, and financial support for electoral management boards. The second objective is to “foster greater accountability of institutions and leaders to citizens and to the law” (United States Agency for International Development, 2013). This objective focuses on activities that support citizen participation, such as voter registration, and develop of institutions and systems that promote political competition through institutional reform.

The nascent literature on democracy promotion has not been especially sanguine about its beneficial effects on recipient countries. Early work by Abraham Lowenthal, for example, tended to express skepticism about the motivations of the United States in attempting to “export” democracy. Larry Diamond argued that USAID is often not flexible enough and powerful enough vis-a-vis the rest of the U.S. foreign policy establishment to program assistance where it is needed most. Thomas Carothers and his colleagues, in the most extensive body of evaluative work on the topic, suggest that democracy promotion can work when done well, although much of the time political blinders, misguided beliefs in the “inevitability” of democratic transitions, and a “one size fits all” mentality have undermined USAID’s effectiveness.

Focusing on civil society strengthening

Strengthening or even ‘building’ civil society has become a preoccupation of international development actors working in post-conflict settings in the past decade (Van Leeuwen & Verkoren, 2012: 81)

‘Civil society building’ is considered a key component of democratisation and peacebuilding, as it might contribute to reforming state-society relations and fostering responsive and legitimate institutions that can effectively deal with conflict (Cousens et al 2001; Woodward 2007).

It was assumed that strengthening civil society would contribute to the cultivation of alternative political processes and institutions to authoritatively and legitimately manage group conflicts (Cousens et al 2001: 12; Woodward 2007).

In addition it was assumed that civil society could play an important role as the ears and eyes of the international community, monitoring human rights, advocating for disadvantaged groups and providing early warning (Barnes 2006).

Civil society had become the ‘imagined agent of development’ (Pearce 2005), being considered more effective than – and thus an alternative to – governments in providing development needs.

Despite a convergence in policy discourse that civil society has important roles to play in development, democratisation and peacebuilding, different emphases are being put that reflect different analytical traditions. Building on the classic work of De Tocqueville, contemporary authors like Putnam have argued how organised groups of citizens are an indispensable element of democracy as they create trust, promote citizens’ interests and foster democratic skills and civic values (De Tocqueville 1864; Putnam 1993).

The role of civil society then is to constantly renegotiate the mutual rights and obligations of state and citizens, critically assess dominant modes of representation and accountability, and press for alternative ways of policy making (Howell & Pearce 2001; White 2004; Paffenholz & Spurk 2006). Instead of ‘making democracy work’ (Putnam 1993), Gramscian/Habermasian civil society is about ‘making democracy happen’.

Civil society has increasingly come to be seen as the sphere where this social contract can be renegotiated (White 2004; Kaldor 2003).

Steele, Pemstein and Meserve (2021) found that while they could replicate earlier findings that show a positive relationship between democracy aid and aggregate levels of democratic quality and election quality, we find surprisingly little evidence that democracy promotion works by building the specific institutional capacities necessary for conducting clean and safe elections.

The models used by Steele et al. (2021) did not establish a robust association between USAID spending on democracy and governance aid and even basic technical capacities such as improved or more autonomous election management.

The mechanisms through which technical foreign aid promotes democracy remain, therefore, troublingly elusive.

While Steele et al. (2021) panel results do not represent a challenge to the core relationship between electoral aid and improved democracy posited by the literature, we nevertheless remain unable to pin down a robust connection between election aid and changes in the electoral institutional targets of that aid,

Disaggregating categories of USAID aid allocations

The study of the impact of aid on democratic change and institutional development has been mixed. This highlight how complex a phenomenon the impact of aid can be to study.

For example, some studies find that aid can be associated with higher levels of democracy under some conditions (Bermeo, 2011; Dunning, 2004; Finkel, Pérez-Liñán, & Seligson, 2007; Scott & Steele, 2011). Goldsmith (2001), for example, found that there was a positive impact of aid on democracy in Sub-Saharan Africa. Bermeo (2011) and Dunning (2004) find that, while aggregate aid is not associated with democratic change, aid from democratic donors has a positive impact on regime change and democratic improvements, especially in the post-Cold War era.

Other studies, however, have found that aid has relatively little effect (Knack, 2004). And some studies have found that aid can have a detrimental impact on democracy and democratic instutions (Bräutigam & Knack, 2004; Djankov, Montalvo, & Raynal-Querol, 2008; Knack, 2001; Licht, 2010) on democratic change and institutional development. In arguments which echo the resource curse comparative literature (Bueno de Mesquita, Alastair, Siverson, & Morrow, 2003; Haber & Menaldo, 2011; Robinson, 2006; Stasavage, 2003), the financial independence of the government created by foreign aid may help stabilize and entrench authoritarian governments (Djankov et al., 2008; Licht, 2010).

Rather than only look at total aid from the US, our paper will examine aid that specifically targets the promotion of civil society.

If democracy assistance is supposed to increase democracy practice, then researchers need to focus on that assistance per se and not aggregate it with programs designed to improve health, education, the environment, or economic growth.

Scott and Steele (2011) found that carefully tailored democracy assistance packages can positively affect democratization independent of factors that influence the decision to provide aid. Targeted democracy aid thus appears to provide “more bang for the buck” than other forms of assistance.

The theory behind this influx of aid in recent years is that a vibrant civil society can hold governments accountable and promote democratic norms. This links to the greater Kantian theory of liberal peace in the international system. Civil society can be a component in the Kantian triangle; namely that democratic norms domestically translate into peaceful interaction between democratic states in a system otherwise marked by anarchy.

One drawback to a plurality of these studies is that they ignore the qualitatively distinct regional differences. The present study will examine the impact of aid at regional level to test whether aid had different impacts in different.

Examining Regional Differences

Since the 1990s, there has been a renewed theoretical reflection concerning the more decentralized and regionalized international system of the post-Cold War period.

The writings of Buzan, Waever (2003) under the rubric of the Copenhagen securitization school focus on the patterns of amity and enmity define the character of distinct geographical regions around the world. These different patterns are based on Wendtian idea of the social structure of anarchy – Hobbesian, Lockean or Kantian or in other words, based on enemy, rival or friend relations, which prevail in the system. This leave us with a complex of conflict formation, security regime or security community.

The positivist, large-N research enterprise that hopes to find empirical evidence for Kantian liberal peace theory (that is to say that the three ‘Kantian’ triangle variables, democracy, international organization membership, and international trade interdependence, are strongly and causally associated with peaceful interstate relations. The effects of these variables, with roots in liberal philosophy (e.g. Kant, 1795/1999), are assumed to be universal, and studies typically include pooled data for all states in the international system (e.g. Oneal & Russett, 1999; Cederman & Rao, 2001; Russett & Oneal, 2001; Cornwell & Colaresi, 2002; Gartzke, Li & Boehmer, 2001; Jungblut & Stoll, 2002; Oneal, Russett & Berbaum, 2003).

There are many new updates on and revisions of this global level analysis which appreciates the distinct regional differences when it comes to difference in conflict (Buzan & Wæver, 2003; Holsti, 1996; Kacowicz, 1998; Solingen, 1998).)

These differences thus would theoretically preclude the benefits of large-N research which has been the norm in recent positivist scholarship.

Many researchers in recent years have focused on the patterns in Asia highlighting a realist-orientation in the region.

Goldsmith (2007) argues that there is a shift in research that builds on exploring the significance of regional variations and path-dependencies

Preliminary findings show that while in aggregate, total USAID amounts can have a negative relationship with HDI, if we disaggregate these aid categories, we can see that different aid strategies show different relationships to HDI.

Polacheck, Robst and Chang (1999) found a negative relationship between bilateral trade volumes and the frequency of interstate military conflict.

In contradiction to this logic, Martin et al. (2008) investigated multilateral trade openness - that is to say, global trade openness - increases the probability of military conflicts.

Martin et al. (2008) argue that countries more open to global trade have a higher probability of dyadic conflict because multilateral trade openness reduces bilateral dependence to any given country and thus lowers the opportunity cost of military conflict.

Recent scholars have echoed the Buzanian argument about regionalism. They argue that that Asian international relations are qualitatively different from those in other parts of the world and, thus, inappropriate for study in the framework of general theories (e.g. Acharya, 2003; Kang, 2003).

One fundamental expectation of some analysts (e.g. Kang, 2003; Katzenstein, 2000) is that the international politics of Asia are ‘different’ from those in other parts of the world, especially ‘the West’. Culture and history combine in a path-dependent process of interaction among states in a given region to create sui generis and empirically meaningful patterns of behaviour and conflict (Goldsmith, 2007: 7). This contradicts the assumption of international relations generalists who commonly use pooled data for all states to test general hypotheses. If the effects of important influences on peace and conflict can be shown to differ significantly in Asia, this will support the arguments for an Asian difference.

The present study will test the idea that there are regional difference concerning the efficacy of different channels and categories of USAID.

How effective is externally-provided aid after we control for internal factors?

In general, studies of democratization and democratic transitions have emphasized internal factors in political transformations (Geddes 1999; Bunce 2000).

Among these internal factors, social and economic developments have been considered crucial to democratization and consolidation. Analysts often stress wealth, economic growth, and education as key correlates of democracy (see Lipset 1959; Hadenius 1992; Rowen 1995; Przeworski, Alvarez, Cheibub, and Limongi 2000).

Therefore our proposed study will control for level of wealth in the form of GDP per capita.

Kantian triangle also claims that increased dependence on international trade will have a pacifying and democratising effect. This paper therefore controls for level of trade as a percentage of GDP

Moreover, the literature on “democratic diffusion” suggests that democracies tend to “cluster” within regions and across time (see Starr 1991; Gleditsch and Hegre 1997; Starr and Lindborg 2003; Cederman and Gleditsch 2004; Gleditsch and Ward 2006).4 The diffusion argument implies that the borders of states are permeable to outside influences—specifically the influence of neighbors—and that democracy is a relatively powerful contagion.

The present paper will therefore control for the average polity score in each sub-region.

International organizations (see Pevehouse 2002a,b; Gleditsch and Ward 2006) may impact democratization and consolidation by supporting domestic forces and creating incentives for reform in, or to gain membership in, an organization.

Therefore we will control for the number of international organizations that each country is a member of. This can highlight another potential pacifying part of the Kantian triangle: that international organization membership can inculcate democracy norms and socialisation.

Additionally, some scholars argue that transnational networks may promote democratic reforms in authoritarian societies (see Keck and Sikkink 1998). For instance, according to Schmitz (2004:408), “transnational activist networks diffuse democratic principles, support domestic allies, and exert pressure on authoritarian regimes.”

library(scales)

palette <- c("#5f0f40","#9a031e","#fb8b24","#e36414","#0f4c5c", "#386641")


pdf %>% 
  filter(region_political6 != 5) %>% 
  group_by(region_name, year) %>% 
  filter(country_wb != "Iraq") %>% 
  mutate(sum_ned = sum(ned_aid, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = sum_ned, group = region_name )) + 
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 0) +
  scale_y_continuous(labels = scales::comma) + 
  # bbplot::bbc_style() + 
  scale_fill_manual(values = palette) + 
  ggtitle("Total NED aid per region 2001 - 2019") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust=1, size = 12),
        axis.text.y = element_text(size = 14),
        legend.title = element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>% 
  filter(region_political6 != 5) %>% 
  group_by(region_name, year) %>% 
  mutate(sum_ned = sum(peace_corps_aid, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = sum_ned, group = region_name )) + 
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 0) +
  scale_y_continuous(labels = scales::comma) + 
  # bbplot::bbc_style() + 
  scale_fill_manual(values = palette) + 
  ggtitle("Total PEACE CORPS aid per region 2001 - 2019") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust=1, size = 12),
        axis.text.y = element_text(size = 14),
        legend.title=element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>% 
filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  filter(country_wb != "Iraq") %>% 
  mutate(sum_ned = sum(civil_engagement_aid, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = sum_ned, group = region_name )) + 
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 0) +
  scale_y_continuous(labels = scales::comma) + 
  # bbplot::bbc_style() + 
  scale_fill_manual(values = palette) + 
  ggtitle("Total Civil Society engagement aid per region 2001 - 2019") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust=1, size = 12),
        axis.text.y = element_text(size = 14), legend.title=element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>% 
filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  filter(country_wb != "Iraq") %>% 
  mutate(sum_ned = median(cso_repress, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = sum_ned, group = region_name )) +
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 0) +
  scale_y_continuous(labels = scales::comma) + 
  scale_fill_manual(values = palette) + 
  ggtitle("Median CSO freedom per region 2001 - 2019") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust = 1, size = 12),
        axis.text.y = element_text(size = 14), legend.title=element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>%
  filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  mutate(sum_iso = median(sum_igo_anytype, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = sum_iso, group = region_name )) +
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  # expand_limits( y = 0) +
  scale_y_continuous(labels = scales::comma) + 
  scale_fill_manual(values = palette) + 
  ggtitle("M/edian ISO membership per region") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust = 1, size = 12),
        axis.text.y = element_text(size = 14),legend.title=element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>%
  filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  mutate(sum_iso = median(trade_per_gdp, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = sum_iso, group = region_name )) +
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 20) +
  scale_y_continuous(labels = scales::comma) + 
  scale_fill_manual(values = palette) + 
  ggtitle("Median trade as percentage of GDP per region") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust = 1, size = 12),
        axis.text.y = element_text(size = 14),legend.title=element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>%
  filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  mutate(total_us_export = sum(us_exports_to_country, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = total_us_export, group = region_name )) +
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 20) +
  scale_y_continuous(labels = scales::comma) + 
  scale_fill_manual(values = palette) + 
  ggtitle("Total US exports to region") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust = 1, size = 12),
        axis.text.y = element_text(size = 14), legend.title=element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>%
  filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  mutate(median_us_export = median(us_exports_to_country, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = median_us_export, group = region_name )) +
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 20) +
  scale_y_continuous(labels = scales::comma) + 
  scale_fill_manual(values = palette) + 
  ggtitle("Median US exports to region") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust = 1, size = 12),
        axis.text.y = element_text(size = 14), 
        legend.title = element_blank(),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

pdf %>%
  filter(region_name != "Eurasia") %>% 
  group_by(region_name, year) %>% 
  mutate(median_democracy = mean(polity, na.rm = TRUE)) %>% 
  ungroup() %>% 
  ggplot(aes(x = year, y = median_democracy, group = region_name )) +
  geom_line(aes(color = as.factor(region_name)), size = 2) + 
  geom_point(aes(color = as.factor(region_name)) ) +
  expand_limits( y = 20) +
  scale_y_continuous(labels = scales::comma) + 
  scale_fill_manual(values = palette) + 
  ggtitle("Median polity score per region") + 
  theme(axis.text.x = element_text(angle = 40, vjust = 0.5, hjust = 1, size = 12),
        axis.text.y = element_text(size = 14),
        title = element_text(size = 10)) + 
  scale_fill_manual(values = palette) +
  scale_color_manual(values = palette) + bbplot::bbc_style()

library(GGally)

my_palette <- c("#005D8F", "#F2A202", "#ae2012")

corr_matrix <- pdf %>%
  filter(year == "2015") %>% 
  filter(!is.na(regime_cat)) %>% 
   dplyr::mutate(
    ned_ln = log(ned_aid + 1),
    mil_ln = log(military_aid + 1),
    econ_ln = log(economic_development + 1),
    demo_ln = log(demo_aid + 1),
    human_ln  = log(humanitarian_assistance + 1), 
    civil_ln = log(civil_engagement_aid + 1)) %>% 
  dplyr::select(regime_cat, ned_ln, civil_ln, mil_ln, hdi, polity) %>% 
  ggpairs(
    mapping = ggplot2::aes(color = as.factor(regime_cat)), alpha = 0.6)   
  # scale_fill_manual(values = my_palette) +
  # scale_color_manual(values = my_palette)

corr_matrix + bbplot::bbc_style()

DV AND CONTROLS

dv <- "lead(polity, 2)"

#### Controls ####

controls <- c(
              "log(population)",
               "log(gdp_pc)",
              # "polity", 
              # "KOFGI", 
              # "KOFTrGI",
              # "as.factor(war_ongoing)",
               "avg_neighbourhood_democracy",
               "log(trade_per_gdp)",
               "log(sum_igo_full)"                 
              # "vote_percent_same_us_all"
              )

#### TOTAL AID ####

iv_ned_all <- c("log(ned_aid + 1) + ")
iv_total_all <- c("log(total_aid + 1)  + ")

iv_pc_all <- c("log(peace_corps_aid + 1) +  ")
iv_civ_eng_all <- c("log(civil_engagement_aid + 1) + ")
iv_demo_all <- c("log(demo_aid + 1) + ")
iv_econ_all <- c("log(economic_development + 1) + ")
iv_mil_all <- c("log(military_aid + 1) + ")
iv_hum_all <- c("log(humanitarian_assistance + 1) + ")



lm_controls <- as.formula(paste(dv, paste(controls, collapse = " + "), sep = " ~ "))
lm_total <- as.formula(paste(dv, paste(iv_total_all, controls, collapse = " + "), sep = " ~ "))

lm_formula_ned <- as.formula(paste(dv, paste(iv_ned_all, controls, collapse = " + "), sep = " ~ "))
lm_formula_peace_corps <- as.formula(paste(dv, paste(iv_pc_all, controls, collapse = " + "), sep = " ~ "))
lm_formula_civ_eng <- as.formula(paste(dv, paste(iv_civ_eng_all, controls, collapse = " + "), sep = " ~ "))
lm_formula_demo <- as.formula(paste(dv, paste(iv_demo_all, controls, collapse = " + "), sep = " ~ "))
lm_formula_econ <- as.formula(paste(dv, paste(iv_econ_all, controls, collapse = " + "), sep = " ~ "))
lm_formula_mil <- as.formula(paste(dv, paste(iv_mil_all, controls, collapse = " + "), sep = " ~ "))
lm_formula_hum <- as.formula(paste(dv, paste(iv_hum_all, controls, collapse = " + "), sep = " ~ "))

FULL DATASET - ALL YEARS


	Dependent variable:

	lead(polity, 2)
	CONTROLS	TOTAL	NED	PEACE CORPS	CIVIC ENGAGE	DEMOCRACY	ECONOMIC	MILITARY	HUMANITARIAN
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)

log(total_aid + 1)		0.550^***
		(0.056)

log(ned_aid + 1)			-0.108^***
			(0.021)

log(peace_corps_aid + 1)				0.040^**
				(0.018)

log(civil_engagement_aid + 1)					-0.022
					(0.026)

log(demo_aid + 1)						0.004
						(0.024)

log(economic_development + 1)							0.181^***
							(0.021)

log(military_aid + 1)								0.294^***
								(0.027)

log(humanitarian_assistance + 1)									0.021^**
									(0.009)

log(population)	0.632	-0.799^***	-0.274^***	-0.446^***	-0.437^***	-0.470^***	-0.686^***	-0.665^***	0.535
	(0.460)	(0.102)	(0.105)	(0.099)	(0.104)	(0.104)	(0.100)	(0.098)	(0.462)

log(gdp_pc)	-0.315	0.319^***	-0.296^***	-0.043	-0.190^*	-0.139	0.237^**	-0.009	-0.337
	(0.278)	(0.108)	(0.103)	(0.110)	(0.110)	(0.118)	(0.107)	(0.097)	(0.278)

avg_neighbourhood_democracy	0.431^***	0.870^***	0.819^***	0.832^***	0.841^***	0.844^***	0.828^***	0.853^***	0.424^***
	(0.071)	(0.035)	(0.036)	(0.037)	(0.036)	(0.036)	(0.036)	(0.035)	(0.071)

log(trade_per_gdp)	1.079^***	0.362^*	0.751^***	0.605^***	0.706^***	0.679^***	0.240	0.188	1.080^***
	(0.146)	(0.212)	(0.214)	(0.217)	(0.217)	(0.216)	(0.217)	(0.213)	(0.146)

log(sum_igo_full)	0.263	6.528^***	5.819^***	6.038^***	6.078^***	6.166^***	6.492^***	6.050^***	0.336
	(0.621)	(0.592)	(0.605)	(0.607)	(0.612)	(0.609)	(0.595)	(0.587)	(0.621)


Observations	1,922	1,922	1,922	1,922	1,922	1,922	1,922	1,922	1,922
R²	0.065	0.364	0.342	0.334	0.333	0.332	0.357	0.372	0.068
Adjusted R²	-0.016	0.358	0.335	0.328	0.326	0.326	0.350	0.366	-0.014
F Statistic	24.618^*** (df = 5; 1767)	181.744^*** (df = 6; 1902)	164.531^*** (df = 6; 1902)	159.137^*** (df = 6; 1902)	158.051^*** (df = 6; 1902)	157.880^*** (df = 6; 1902)	175.914^*** (df = 6; 1902)	187.917^*** (df = 6; 1902)	21.445^*** (df = 6; 1766)

Note:	p<0.1; p<0.05; p<0.01

ASIA SUB-DATASET


	Dependent variable:

	lead(polity, 2)
	CONTROLS	TOTAL	NED	PEACE CORPS	CIVIC ENGAGE	DEMOCRACY	ECONOMIC	MILITARY	HUMANITARIAN
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)

log(total_aid + 1)		0.080
		(0.096)

log(ned_aid + 1)			0.056^**
			(0.027)

log(peace_corps_aid + 1)				-0.018
				(0.029)

log(civil_engagement_aid + 1)					-0.002
					(0.032)

log(demo_aid + 1)						0.014
						(0.037)

log(economic_development + 1)							0.074^***
							(0.026)

log(military_aid + 1)								0.014
								(0.032)

log(humanitarian_assistance + 1)									0.031
									(0.023)

log(population)	-0.428	-0.397	-0.461	-0.395	-0.416	-0.421	-0.584	-0.475	-0.425
	(0.795)	(0.796)	(0.792)	(0.797)	(0.811)	(0.795)	(0.791)	(0.803)	(0.794)

log(gdp_pc)	0.577	0.510	0.382	0.603	0.579	0.564	0.516	0.531	0.521
	(0.461)	(0.468)	(0.469)	(0.463)	(0.462)	(0.463)	(0.458)	(0.473)	(0.462)

avg_neighbourhood_democracy	0.340^***	0.341^***	0.336^***	0.328^***	0.340^***	0.338^***	0.352^***	0.341^***	0.334^***
	(0.108)	(0.108)	(0.108)	(0.110)	(0.108)	(0.108)	(0.107)	(0.108)	(0.108)

log(trade_per_gdp)	1.136^***	1.122^***	1.150^***	1.134^***	1.137^***	1.135^***	1.033^***	1.135^***	1.127^***
	(0.215)	(0.216)	(0.214)	(0.215)	(0.215)	(0.215)	(0.216)	(0.215)	(0.215)

log(sum_igo_full)	0.570	0.533	0.507	0.471	0.570	0.573	0.489	0.598	0.654
	(0.952)	(0.953)	(0.949)	(0.966)	(0.953)	(0.953)	(0.945)	(0.955)	(0.953)


Observations	530	530	530	530	530	530	530	530	530
R²	0.093	0.095	0.101	0.094	0.093	0.093	0.108	0.094	0.097
Adjusted R²	0.011	0.010	0.018	0.010	0.009	0.009	0.025	0.009	0.013
F Statistic	9.971^*** (df = 5; 485)	8.420^*** (df = 6; 484)	9.097^*** (df = 6; 484)	8.364^*** (df = 6; 484)	8.293^*** (df = 6; 484)	8.317^*** (df = 6; 484)	9.772^*** (df = 6; 484)	8.326^*** (df = 6; 484)	8.617^*** (df = 6; 484)

Note:	p<0.1; p<0.05; p<0.01

AFRICA SUB-DATASET


	Dependent variable:

	lead(polity, 2)
	CONTROLS	TOTAL	NED	PEACE CORPS	CIVIC ENGAGE	DEMOCRACY	ECONOMIC	MILITARY	HUMANITARIAN
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)

log(total_aid + 1)		0.213^**
		(0.083)

log(ned_aid + 1)			0.056^***
			(0.017)

log(peace_corps_aid + 1)				-0.047^**
				(0.023)

log(civil_engagement_aid + 1)					0.084^***
					(0.026)

log(demo_aid + 1)						0.077^***
						(0.026)

log(economic_development + 1)							-0.003
							(0.018)

log(military_aid + 1)								0.035
								(0.022)

log(humanitarian_assistance + 1)									0.013
									(0.016)

log(population)	0.525	0.517	-0.087	0.187	-0.030	-0.099	0.509	0.528	0.502
	(1.323)	(1.316)	(1.324)	(1.330)	(1.324)	(1.330)	(1.327)	(1.321)	(1.323)

log(gdp_pc)	-2.255^***	-2.346^***	-2.063^***	-1.868^***	-2.185^***	-2.174^***	-2.239^***	-2.238^***	-2.220^***
	(0.678)	(0.675)	(0.674)	(0.702)	(0.673)	(0.674)	(0.684)	(0.677)	(0.679)

avg_neighbourhood_democracy	1.177^***	1.004^***	1.178^***	1.160^***	1.201^***	1.224^***	1.176^***	1.135^***	1.141^***
	(0.259)	(0.266)	(0.257)	(0.259)	(0.257)	(0.258)	(0.259)	(0.260)	(0.263)

log(trade_per_gdp)	1.140^***	1.081^***	1.014^***	1.162^***	1.082^***	1.108^***	1.153^***	1.023^***	1.161^***
	(0.363)	(0.362)	(0.362)	(0.362)	(0.360)	(0.360)	(0.370)	(0.369)	(0.364)

log(sum_igo_full)	-0.343	-0.518	-0.721	0.345	-0.057	-0.122	-0.332	-0.324	-0.300
	(1.769)	(1.762)	(1.758)	(1.797)	(1.758)	(1.759)	(1.772)	(1.767)	(1.771)


Observations	644	644	644	644	644	644	644	644	644
R²	0.122	0.132	0.138	0.128	0.137	0.135	0.122	0.126	0.123
Adjusted R²	0.045	0.054	0.060	0.050	0.059	0.057	0.043	0.047	0.044
F Statistic	16.399^*** (df = 5; 591)	14.905^*** (df = 6; 590)	15.734^*** (df = 6; 590)	14.422^*** (df = 6; 590)	15.558^*** (df = 6; 590)	15.332^*** (df = 6; 590)	13.649^*** (df = 6; 590)	14.141^*** (df = 6; 590)	13.762^*** (df = 6; 590)

Note:	p<0.1; p<0.05; p<0.01

MIDDLE EAST AND NORTH AFRICA SUB-DATASET


	Dependent variable:

	lead(polity, 2)
	CONTROLS	TOTAL	NED	PEACE CORPS	CIVIC ENGAGE	DEMOCRACY	ECONOMIC	MILITARY	HUMANITARIAN
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)

log(total_aid + 1)		0.174
		(0.111)

log(ned_aid + 1)			0.097^**
			(0.043)

log(peace_corps_aid + 1)				0.851^***
				(0.134)

log(civil_engagement_aid + 1)					0.086^**
					(0.042)

log(demo_aid + 1)						0.081^**
						(0.040)

log(economic_development + 1)							0.045
							(0.034)

log(military_aid + 1)								0.091^*
								(0.051)

log(humanitarian_assistance + 1)									0.079^**
									(0.031)

log(population)	-1.499	-1.174	-1.506	-1.189	-1.779	-1.435	-1.473	-1.691	-1.341
	(1.136)	(1.150)	(1.124)	(1.037)	(1.135)	(1.127)	(1.133)	(1.134)	(1.121)

log(gdp_pc)	-0.540	-0.440	-0.878	-1.339	-0.798	-0.830	-0.569	-0.396	-0.314
	(1.157)	(1.155)	(1.155)	(1.063)	(1.155)	(1.158)	(1.155)	(1.154)	(1.144)

avg_neighbourhood_democracy	0.688^**	0.646^**	0.656^**	0.390	0.675^**	0.646^**	0.725^***	0.676^**	0.574^**
	(0.270)	(0.270)	(0.267)	(0.251)	(0.268)	(0.268)	(0.270)	(0.268)	(0.269)

log(trade_per_gdp)	2.359^*	1.994	1.878	1.746	1.852	2.104^*	2.057	2.023	2.391^*
	(1.230)	(1.247)	(1.235)	(1.126)	(1.245)	(1.227)	(1.248)	(1.237)	(1.212)

log(sum_igo_full)	-4.737	-5.467	-3.207	-3.068	-5.467	-4.155	-5.543	-4.290	-4.465
	(5.147)	(5.149)	(5.137)	(4.703)	(5.119)	(5.116)	(5.171)	(5.124)	(5.074)


Observations	216	216	216	216	216	216	216	216	216
R²	0.046	0.058	0.071	0.210	0.066	0.066	0.055	0.062	0.078
Adjusted R²	-0.057	-0.049	-0.035	0.120	-0.040	-0.041	-0.053	-0.045	-0.027
F Statistic	1.882^* (df = 5; 194)	1.990^* (df = 6; 193)	2.466^** (df = 6; 193)	8.564^*** (df = 6; 193)	2.285^** (df = 6; 193)	2.263^** (df = 6; 193)	1.877^* (df = 6; 193)	2.125^* (df = 6; 193)	2.732^** (df = 6; 193)

Note:	p<0.1; p<0.05; p<0.01

LATIN AMERICA SUB-DATASET


	Dependent variable:

	lead(polity, 2)
	CONTROLS	TOTAL	NED	PEACE CORPS	CIVIC ENGAGE	DEMOCRACY	ECONOMIC	MILITARY	HUMANITARIAN
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)

log(total_aid + 1)		0.049
		(0.155)

log(ned_aid + 1)			0.016
			(0.022)

log(peace_corps_aid + 1)				0.003
				(0.023)

log(civil_engagement_aid + 1)					0.017
					(0.031)

log(demo_aid + 1)						0.006
						(0.031)

log(economic_development + 1)							-0.034
							(0.029)

log(military_aid + 1)								0.018
								(0.071)

log(humanitarian_assistance + 1)									0.020
									(0.015)

log(population)	-1.518	-1.520	-2.000	-1.481	-1.537	-1.533	-1.817	-1.486	-1.636
	(1.786)	(1.789)	(1.909)	(1.813)	(1.789)	(1.791)	(1.803)	(1.794)	(1.785)

log(gdp_pc)	-1.066	-1.045	-0.977	-1.072	-1.134	-1.076	-1.105	-1.044	-1.097
	(0.725)	(0.729)	(0.736)	(0.727)	(0.736)	(0.728)	(0.725)	(0.731)	(0.724)

avg_neighbourhood_democracy	0.135	0.163	0.127	0.139	0.131	0.133	0.155	0.150	0.162
	(0.282)	(0.295)	(0.282)	(0.284)	(0.282)	(0.283)	(0.282)	(0.288)	(0.282)

log(trade_per_gdp)	0.416	0.420	0.373	0.415	0.396	0.405	0.429	0.404	0.399
	(0.468)	(0.469)	(0.473)	(0.469)	(0.470)	(0.473)	(0.468)	(0.471)	(0.468)

log(sum_igo_full)	5.025^*	4.852^*	5.036^*	4.952^*	5.010^*	5.024^*	5.673^**	4.860^*	4.919^*
	(2.600)	(2.660)	(2.602)	(2.668)	(2.604)	(2.605)	(2.657)	(2.686)	(2.597)


Observations	304	304	304	304	304	304	304	304	304
R²	0.026	0.027	0.028	0.026	0.027	0.026	0.031	0.027	0.033
Adjusted R²	-0.069	-0.072	-0.071	-0.073	-0.072	-0.073	-0.068	-0.073	-0.066
F Statistic	1.493 (df = 5; 276)	1.257 (df = 6; 275)	1.328 (df = 6; 275)	1.243 (df = 6; 275)	1.291 (df = 6; 275)	1.246 (df = 6; 275)	1.473 (df = 6; 275)	1.251 (df = 6; 275)	1.552 (df = 6; 275)

Note:	p<0.1; p<0.05; p<0.01

NOT ASIA SUB-DATASET = ALL COUNTRIES BUT ASIA


	Dependent variable:

	lead(polity, 2)
	CONTROLS	TOTAL	NED	PEACE CORPS	CIVIC ENGAGE	DEMOCRACY	ECONOMIC	MILITARY	HUMANITARIAN
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)

log(total_aid + 1)		0.088^***
		(0.033)

log(ned_aid + 1)			0.043^***
			(0.009)

log(peace_corps_aid + 1)				-0.016
				(0.012)

log(civil_engagement_aid + 1)					0.029^**
					(0.011)

log(demo_aid + 1)						0.034^***
						(0.012)

log(economic_development + 1)							0.003
							(0.009)

log(military_aid + 1)								0.035^***
								(0.013)

log(humanitarian_assistance + 1)									0.015^*
									(0.008)

log(population)	-0.132	-0.223	-0.349	-0.105	-0.361	-0.296	-0.143	-0.283	-0.178
	(0.446)	(0.447)	(0.446)	(0.447)	(0.455)	(0.449)	(0.447)	(0.449)	(0.447)

log(gdp_pc)	-0.369	-0.392	-0.453	-0.350	-0.362	-0.388	-0.370	-0.367	-0.375
	(0.281)	(0.281)	(0.280)	(0.281)	(0.281)	(0.281)	(0.281)	(0.281)	(0.281)

avg_neighbourhood_democracy	0.670^***	0.645^***	0.662^***	0.656^***	0.685^***	0.675^***	0.673^***	0.668^***	0.656^***
	(0.110)	(0.110)	(0.110)	(0.111)	(0.110)	(0.110)	(0.111)	(0.110)	(0.110)

log(trade_per_gdp)	0.877^***	0.874^***	0.826^***	0.879^***	0.876^***	0.890^***	0.874^***	0.820^***	0.878^***
	(0.209)	(0.209)	(0.208)	(0.209)	(0.209)	(0.209)	(0.209)	(0.210)	(0.209)

log(sum_igo_full)	-1.364	-1.264	-1.214	-1.370	-1.215	-0.924	-1.357	-1.406	-1.264
	(0.877)	(0.876)	(0.872)	(0.877)	(0.877)	(0.889)	(0.878)	(0.875)	(0.878)


Observations	1,638	1,638	1,638	1,638	1,638	1,638	1,638	1,638	1,638
R²	0.050	0.054	0.063	0.051	0.054	0.055	0.050	0.055	0.052
Adjusted R²	-0.034	-0.029	-0.020	-0.033	-0.030	-0.029	-0.034	-0.029	-0.032
F Statistic	15.786^*** (df = 5; 1505)	14.400^*** (df = 6; 1504)	16.767^*** (df = 6; 1504)	13.458^*** (df = 6; 1504)	14.273^*** (df = 6; 1504)	14.530^*** (df = 6; 1504)	13.171^*** (df = 6; 1504)	14.478^*** (df = 6; 1504)	13.729^*** (df = 6; 1504)

Note:	p<0.1; p<0.05; p<0.01

OUTCOME VARIABLE: ONSET OF WAR

 dv_conflict <- "lead(as.factor(war_ongoing), 2)"


iv_ned_pc <- c("log(ned_aid_pc + 1) + ")
iv_total_pc <- c("log(total_aid_pc + 1)  + ")
iv_pc_pc <- c("log(peace_corps_aid_pc + 1) +  ")
iv_civ_eng_pc <- c("log(civil_engagement_aid_pc + 1) + ")
iv_demo_pc <- c("log(demo_aid_pc + 1) + ")
iv_econ_pc <- c("log(econ_aid_pc + 1) + ")
iv_mil_pc <- c("log(mil_aid_pc + 1) + ")
iv_hum_pc <- c("log(human_aid_pc + 1) + ")

lm_controls <- as.formula(paste(dv, paste(controls, collapse = " + "), sep = " ~ "))
lm_total <- as.formula(paste(dv, paste(iv_total_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_ned <- as.formula(paste(dv, paste(iv_ned_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_peace_corps <- as.formula(paste(dv, paste(iv_pc_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_civ_eng <- as.formula(paste(dv, paste(iv_civ_eng_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_demo <- as.formula(paste(dv, paste(iv_demo_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_econ <- as.formula(paste(dv, paste(iv_econ_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_mil <- as.formula(paste(dv, paste(iv_mil_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_hum <- as.formula(paste(dv, paste(iv_hum_pc, controls, collapse = " + "), sep = " ~ "))

 
lm_controls <- as.formula(paste(dv_conflict, paste(controls, collapse = " + "), sep = " ~ "))
lm_total <- as.formula(paste(dv_conflict, paste(iv_total_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_ned <- as.formula(paste(dv_conflict, paste(iv_ned_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_peace_corps <- as.formula(paste(dv_conflict, paste(iv_pc_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_civ_eng <- as.formula(paste(dv_conflict, paste(iv_civ_eng_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_demo <- as.formula(paste(dv_conflict, paste(iv_demo_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_econ <- as.formula(paste(dv_conflict, paste(iv_econ_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_mil <- as.formula(paste(dv_conflict, paste(iv_mil_pc, controls, collapse = " + "), sep = " ~ "))
lm_formula_hum <- as.formula(paste(dv_conflict, paste(iv_hum_pc, controls, collapse = " + "), sep = " ~ "))

Maximum Likelihood estimation Newton-Raphson maximisation, 7 iterations Return code 8: successive function values within relative tolerance limit (reltol) Log-Likelihood: -394.9104 7 free parameters Estimates: Estimate Std. error t value Pr(> t) (Intercept) -14.50894 3.49543 -4.151 3.31e-05 *log(population) 2.00047 0.23148 8.642 < 2e-16* log(gdp_pc) -0.77577 0.14250 -5.444 5.21e-08 *avg_neighbourhood_democracy -0.39289 0.06058 -6.486 8.82e-11* log(trade_per_gdp) -0.37718 0.32195 -1.172 0.241367 log(sum_igo_full) -3.14593 0.81484 -3.861 0.000113 *sigma 4.85229 0.50875 9.538 < 2e-16*
Signif. codes: 0 ‘’ 0.001 ’’ 0.01 ’’ 0.05 ‘.’ 0.1 ’ ’ 1

Maximum Likelihood estimation Newton-Raphson maximisation, 7 iterations Return code 8: successive function values within relative tolerance limit (reltol) Log-Likelihood: -394.9104 7 free parameters Estimates: Estimate Std. error t value Pr(> t) (Intercept) -14.50894 3.49543 -4.151 3.31e-05 log(population) 2.00047 0.23148 8.642 < 2e-16 log(gdp_pc) -0.77577 0.14250 -5.444 5.21e-08 avg_neighbourhood_democracy -0.39289 0.06058 -6.486 8.82e-11 log(trade_per_gdp) -0.37718 0.32195 -1.172 0.241367 log(sum_igo_full) -3.14593 0.81484 -3.861 0.000113 sigma 4.85229 0.50875 9.538 < 2e-16