Correlations and Codebook
Paula
9/5/2021
Codebook
| Variable | Description |
|---|---|
| country | country name variable |
| year | year variable |
| region_6 | Geopolitical regions: Post-Soviet, MENA, Latin America, Africa, Asia, Western countries |
| budget | Total PD budget per country (US State Dept.) |
| budget_pc | PD budget per capita (US State Dept.) |
| budget_const | Total PD budget per country at constant 2014 prices |
| budget_pc_const | PD budget per capita at constant 2014 prices |
| ppp_budget_div | Total PD budget divided by price level ratio (World Bank) |
| ppp_budget_div_pc | PD budget per capita divided by price level ratio (World Bank) |
| budget_const_ppp | Total PD budget at constant 2014 prices divided by price level ratio |
| budget_pc_const_ppp | PD budget per capita at constant 2014 prices divided by price level ratio |
| budget_proportion | Proportion of total budget to each country |
| pop | Population (World Bank) |
| gdp_const | Total country GDP 2010 constant $US (World Bank) |
| gdp_pc_const | GDP per capita 2010 constant $US (World Bank) |
| electoral_demo | V-DEM Electoral democracy index |
| civ_lib_fh | Freedom House civil liberties score from 1 (most free) and 7 (least free): freedoms of expression and belief, associational and organizational rights |
| rule_law_fh | Freedom House rule of law score from 1 (worst) to 16 (best): independence of the judiciary; the extent to which rule of law prevails in civil and criminal matters; the existence of direct civil control over the police, the protection from political terror, unjustified imprisonment, exile and torture; absence of war and insurgencies; and the extent to which laws, policies and practices guarantee equal treatment |
| pol_right_fh | Freedom House political rights score from 1 to : people to participate freely in the political process, including the right to vote freely for distinct alternatives in legitimate elections, compete for public office, join political parties and organizations, and elect representatives who have a decisive impact on public policies and are accountable to the electorate. |
| fh_index | 1 (Free) 2 (Partly Free) 3 (Not Free) |
| polity_index | Scale ranges from +10 (strongly democratic) to -10 (strongly autocratic). |
| bureau | US State Dept. bureau regions: African Affairs (AF), East Asia and Pacific (EAP), Europe (EUR), Near East (NEA), South and Central Asia (SCA), Western Hemisphere (WHA) |
| senate_maj | Whether the senate majority in US Congress is R or D that year |
| house_maj | Whether the house majority in US Congress is R or D that year |
| president_party | Whether the party of the US President is R or D that year |
| president | US president (Obama or Trump) |
| colonizer | Whether the country was colonized by a European country |
| language | Main official language of the country |
| dist_to_moscow | Distance from country’s capital to Moscow in km |
| dist_to_beijing | Distance from country’s capital to Beijing in km |
| dist_to_wash_dc | Distance from country’s capital to Washington DC in km |
| min_distance_to_us | Distance from country’s capital to nearest US city |
| muslim_population | Estimated total Muslim population |
| percent_muslim_pop | Estimated percentage of Muslim population |
| exports_per_gdp | Exports as a percentage of country’s GDP (World Bank) |
| imports_per_gdp | Imports as a percentage of country’s GDP (World Bank) |
| exports_pc | Exports per capita |
| imports_pc | Imports per capita |
| trade_openess | Total imports and exports divided by total GDP |
| exports_bop | Total exports of goods and services (World Bank) |
| imports_bop | Total imports of goods and services (World Bank) |
| infant_mortality | Number of infant deaths per 1000 live births per year (World Bank) |
| armed_forces_per_labour | Armed forces as a percentage of total work force (World Bank) |
| us_exports_to_country | Total US exports to country (US Foreign Trade Census) |
| us_imports_to_country | Total US imports from country (US Foreign Trade Census) |
| us_trade_dependence | Imports and exports with US divided by total imports and exports |
| us_exports_pc | Exports from US to country per capita |
| us_imports_pc | Imports to US from country per capita |
| china_exports_pc | Exports to country from China per capita (Comtrade) |
| china_trade_dep | Imports and exports with China divided by total imports and exports |
| russia_exports_pc | Exports to country from Russia per capita (Comtrade) |
| russia_trade_dep | Imports and exports with Russia divided by total imports and exports |
| un_vote_distance | Distance from US-led liberal order votes in the UNGA |
| us_same_vote_percent | Percentage of voting similarity with US on important US votes, as determined by US State Dept |
| china_same_vote_percent | percentage of voting similarity with China on important US votes, as determined by US State Dept |
| russia_same_vote_percent | Percentage of voting similarity with Russia on important US votes, as determined by US State Dept |
| defense_pact_us | Whether country has defence pact with US (Correlates of War) |
| rugged_terrain | Extent to which country is rugged |
| bases | Number of US main bases in the country |
| lily_pad | Number of small US military installation in the country |
| total_bases | Number of US bases and lily pad bases |
| sum_suicides | Number of terrorist suicide attacks per year |
| sum_attacks | Number of terrorist attacks per year |
| suicides_pc | Number of terrorist suicide attacks per year |
| troops | Total number of US troops in the country |
| troops_pc | Number of US troops per capita |
| us_igos | Number of IGOs country shares membership with US |
| china_igos | Number of IGOs country shares membership with China |
| russis_igos | Number of IGOs country shares membership with Russia |
Abstract
The United States (US) is the birthplace of public diplomacy as we understand it today. US’ public diplomacy is also the most researched in the literature, which is dominated by qualitative or normative studies. Virtually no attention has been paid to the determinants of US public diplomacy spending abroad. Why does the US spend more in certain countries, but not in others? What informs these decisions and how can we make sense of this theoretically? To answer these questions, in this paper, we analyze the determinants of US public diplomacy spending in over 150 countries in 2019, looking at variables that cover demographics, political, military, trade or cultural relations with the US, US strategic competition with China or Russia among others. Employing a cross-sectional data set, we employ OLS analysis to analyze the determinants of US’ public diplomacy spending in US’ overseas missions. Our findings from OLS analysis indicate a segmented allocation of PD spending that has different trends in Western and non-Western countries, with much emphasis on counter-terrorism efforts as well countering Chinese and Russian influence.
Introduction
The present paper hopes to contribute to the study of public diplomacy by theorising what could motivate a country - in this case the US - to dedicate finite public diplomacy resources to particular places over others. PD budget choices can be seen as a proxy for how seriously the US government considers the need to improve US favourability and relations with particular foreign audiences.
The task of quantifying public diplomacy efforts is not a straightforward endeavour. Like many other constructs in political science, it is difficult to capture and operationalise a concept that acts as an umbrella term for diverse actions and goals.
This effort is more difficult when we consider the plurality of competing definitions in both the literature and among policymakers / PD practitioners. That is not to discuss those that relegate PD to a mere re-branding of propaganda (see Noya 2006; Arndt 2009).
The present study aims to focus on a relatively narrow facet of public diplomacy - US budgetary spending - and explore its correlates across all countries. Where can we see flows of public diplomacy, and furthermore, can we account for these patterns of variation as parsimoniously as possible in our model?
The present paper does not argue that these budgetary figures account for the entirety of all public diplomacy spending by the US. That is an impossible task. Often, many forms of aid, foreign investment and international interactions impact how foreign audiences react to the US, even if they are not the primary purpose. Plus, domestic actions that the US takes can often be observed by foreign audiences and have public diplomacy-type effects that we cannot operationalise at a country-level. Public Diplomacy that occurs at these multiple levels cannot all be captured in a few budget lines.
The aim of the present paper is to focus on a relatively narrow but well-defined area; we wish to examine the figures that the US government has declared themselves as being explicitly public diplomacy spending and, as such, may provide a window into the priorities of the US as they see them. The paper considers these budgets can offer a snapshot into the minds of the administration at the US State Department. The declared budgetary commitments made by the US government can highlight the priorities and preferences that they explicitly wish to pursue under the banner of primarily PD.
Do the activities that the US describe as Public Diplomacy related correspond to the definitions and conceptualizations that are present in academic literature on PD?
Theoretical Framework
There are various main theoretical distinctions which may explain the motivations of the US when it is deciding how to allocate finite PD budgetary figures.
The theoretical underpinnig for the present paper comes from the large corpus of research on determinants of donor allocations of ODA in recipient countries.
These motivations range from the security-oriented motivations on one end of the spectrum to more ideologically-orieted motivations. For example, the ‘societal’ model of foreign assistance derived from research on official development aid (ODA) views aid as a tool for promoting donor economic interests. As such, aid flows to countries that have strong economic ties with the donor. Alternatively, a ‘statist’ model of spending allocations argues that foreign aid flows to advance ideological interests, such as the promotion of democracy and human rights.
There is a large corpus of evidence that official development aid (ODA) is utilised as a tool to further the goals of US foreign policy and national security objectives.
When examining ODA allocations by the US, Meernik, Krueger & Poe (1998) found that national security goals, derived primarily from realism, could provide powerful explanation of US foreign assistance. Meernik, Krueger & Poe (1998) also test the societal and statist perspectives. Their findings suggests that the societal model fares poorly as an explainer of aid allocations, but the statist model performs quite well and its indicators seem to be increasing in explanatory power with the end of the Cold War.
A large corpus of literature has focused on the case of U.S. foreign aid allocation and its relationshop with political and civil rights. See (Cingranelli and Pasquarello, 1985; Carleton and Stohl, 1987; Poe, 1992; Abrams and Lewis, 1993; Poe and Sirirangsi, 1994; Poe, Pilatovsky, Miller, and Ogundele, 1994; Apodaca and Stohl, 1999).2 These studies differ of course in their results from each other, and sometimes substantially so, due to different data sets, different time periods looked at, different estimation techniques used. Nevertheless, most of these studies come to the conclusion that more respect for political freedom and, albeit less clearly so, respect for personal integrity rights is rewarded with a higher probability of receiving any U.S. aid as well as possibly with a higher level of aid allocated.
Smith (1990) indicates that US-funded NGOs are often prohibited from distributing US food aid even in states with a significant number of malnourished children if the state is on the US political blacklist.
This is also the case regarding Public Diplomacy. Even if the US hopes to reach the “hearts and minds” of the general publics in countries hostile to the US - such as North Korea, Syria or Cuba - it is next to impossible to carry out public diplomacy activities on the ground due to the lack of formal diplomatic relations with these countries and the absense of any consulary personnel on the ground to reach out the the publics in these countries. So even if there is a need for US public diplomacy in these hostile countries, it is not aways feasible to meet that need.
Kuziemko and Werker (2006) found that temporary members of the United National Security Council receive a substantial increase in foreign aid commitments from the United States in years they serve as temporary members of the UNSC. They hypothesised that this aid was thus given for US political reasons. The US State Department very publicly takes note of how every country in the United Nations General Assembly votes and tallies up how they vote on issues that the US explicity label as ’important". Therefore, the present study will add variables that measure the extent to which each countries votes similar to the US on important votes.
There are numerous studies that have studied the relationship of economic growth with foreign aid, economic growth along related to foreign direct investment (FDI) and economic growth related to with economic freedom (openness), by the use of time series or panel framework of analysis. However, these studies got conflicting results, both in terms of significance and in terms of the direction of impact (reference)
For that reason, the present study will examine the extent to which PD spending correlates with levels of US exports, level of trade dependence by countries on the US.
Zero-sum conceptions of great power dynamics contend that any gain for powerful rivals can translate into a loss of power for their adversaries. Therefore, it is rational to take actions that thwart any threatening adversary’s path to gaining power.
As one of the dominant theories of international relations, realism hypothesizes that states seek to maximize their power in an international system characterized by anarchy (Waltz, 1979). In more recent years, however, neorealists have amended the theory to claim that states not only seek power but also national economic interests, which are important for security (Gilpin, 1987).
Therefore, we add into the model variables that capture Chinese and Russian power. Do we see evidence that the US is spending more PD budget in countries that are voting similar to the other great powers or those countries that are more dependent on CHina or Russia for trade?
Securitization is a term used by constructivist security scholars to refer to the process of social construction of threats (Buzan ad Waever). Different patterns of relations are characterised on a spectrum from enmity-based regions (a more Hobbesian reality with relationships based on suspicion and lack of institutional infrastructure constraints ) to amnity-based regions (a more Kantian reality with relationships based on mutual trust and norms). The present study will examine patterns of PD budget allocation at the regional level to examine whether there are distinct patterns of spending across the global regions.
Promoting culture as a tool for soft power and therefore increasing US favourability may have a “higher return on investment” in foreign countries that already have a baseline familiarity with US cultural traditions. Countries that are a priori distinct from US in terms of culture, language, social traditions et cetera, may need far more PD spending to see any real increase in US favorability in the country’s populations.
Examining many donor countries, Neumayer (2003a) finds that some donors give more aid to countries that are geographically close and most donors give more aid to those countries that import a higher share of the donor country’s exports. The US is a global power, and some go as far to argue it is the unipole in our post-Cold War world. Do regions and proxmity matter to the US or does it have the ability to project its power equally around the world?
A 2016 paper by Lebovic and Sauders examined the determinants of high-level US diplomatic visits throughout the 20th century following the end of World War II. They explored to what extent strategic influence affected choices of diplomatic visits. With regression analysis, they investigated strategic variables such as military spending, US military aid, US war alliances and US Defense Pacts.
Hypotheses
First we will examine whether the PD budget is spent to maintain friends or to win over foes?
With foreign aid, researchers have found that the United States is likely to give more aid to nations aligned against it than are other OECD nations. These countries are more likely to choose countries that are more aligned.
Bueno and Smith (2009) report “the United States is likely to give more to nations aligned against it than are other OECD nations.” Bueno and Smith argue that the United States tends to give more to larger nations compared to other OECD donors. This perhaps reflects greater resources of the United States. Contrary to the findings for other OECD nations, the United States gives less aid as its resources increase.
H1 0 : that the US is equally likely to spend money to maintain allies than it is to spend money to win “hearts and minds” in countries that are divergent from US national interests.
H1 a1 : that the US spends more among allies / ideologically similar countries
H1 a2: that the US spends more in countries antagonistic to US national / ideological interests
Appealing to traders
The US could spend more money speaking to countries with whom it is more connected economically - through trade or FDI - or it could spend more in countries with whom it has yet to form dense economic ties.
H2 0 : the US is equally as likely to spend money IN countries that it has economically dense ties as it is to countries that are not economically tied to the US
The alternative hypothesis argue that economic ties are related.
H2 a1: the US spends more PD budget in economically-connected countries
H2 a2: the US spends more PD budget in countries that it doesn’t already have economic ties.
Military concerns / guilt money?
Does the US spend more PD budget in countries with whom it has a history of military conflict?
In the foreign aid literature, Kisangani and Pickering (2015) found that aid flows from traditional DAC donors rise significantly when one or more of their members dispatch soldiers in support of a target government, but gradually recede after troops depart
The year after the US invasion of Iraq in 2003, US aid to that country totaled $7.6 billion. This amount is nearly double the total sum of aid Iraq received from the USA from 1962 to 2003, $3.89 billion (USAID, 2008). It represents a 195% increase on the average annual level of aid given to Iraq prior to 2003.
The USA is not alone in sending economic aid to target countries during and after military missions abroad. A more obscure example comes from the 1978 French and Belgian intervention to shore up Mobutu Sese Seko in Zaire (now Congo) when the autocrat encountered a challenge from rebels based in Angola. Foreign aid to Zaire jumped roughly 25% in the wake of the French and Belgian military action (OECD, 2007)
Meernik et al. (1998) theorise that the US may allocate foreign aid to states that house US troops in order to influence or maintain good relations with them.
H3 0: that there is no relationship between the presence of US troops in the country and levels of public diplomacy spending.
The null hypothesis indicates that there is no relationship between the presence of US troops in the country and the need to invest in public diplomacy efforts. Alternative hypotheses ask whether there is a positive relationship with levels of US troops and spending
H3 a1: the US spends more PD budget in countries with US troops.
The alternative hypothesis indicates that as troop levels increase in the country, PD budget also increases.
War on Terror
The emergence of the US-led ‘Global War on Terror’ has been viewed as another potential watershed in foreign aid patterns. While some analysts feared that the US and potentially other traditional donors would increasingly focus aid on narrowly defined strategic concerns after the events of 11 September 2001 (9/11), the evidence for this has been limited (Clist, 2011; Hoeffler & Outram, 2011).
According to Smith (2011), upon taking office in January 2009, President Obama quickly established Afghanistan and Pakistan as one of his highest foreign policy priorities.
H4 0: there is no relationship between number of suicide attacks in a country and level of public diplomacy spending.
The null hypothesis indicates that there is no relationship between the presense of US troops in the country and the need to invest in public diplomacy efforts. Alternative hypotheses ask whether there is a positive relationship with levels of US troops and spending
Culturally close or culturally distinct?
Does the US spend more PD budget in countries with whom it has similar cultural characteristics - similar religious make-up or similar legal / political traditions - or does it endeavour to engage with publics from culturally distinct countries via larger budgetary allocations?
Fight against misinformation / preserve US liberal world order against Russian and Chinese influence.
The US spends more money in countries that are in the Russian sphere of influence / Chinese sphere of influence.
Modern Issues:
countering violent extremism (CVE);
countering negative Russian influence;
promoting landmark trade agreements like Trans-Pacific Partner-ship Agreement (TPP) and Transatlantic Trade and Investment Partnership (TTIP) Agreement; and
enduring efforts to support the advancement of democracy, human rights and civil society and protect the global environment.
Dataset and methodology
Endogeneity between aid and other variables are present in the literature and no consensus as to which of the many papers contributing to the debate convincingly address the identification problem (see Dreher and Langlotz (2015)). These same concerns are present in this paper so correlation relationship analysis is more tenable. The present paper does not claim to determine causal relationships; the time horizon is short and there are numerous factors - international and domestic - that can influence public diplomacy in both directions. Rather, we aim to examine the patterns and correlates across countries and regions. We aim to see to what extent we can associate various country or regional characteristics with levels of budgetary spending.
Dependent Variable
Taking the term from economics, opportunity cost is the idea that when one allocates resources to one area, there is an additional cost of the missed opportunity to not allocate funds to another area. Applying this to PD spending, can we understand the decisions that the US government made spending their finite resources in certain countries - at the expense of others - to achieve the maximum favourability in foreign publics. Do they reflect the strategic goals of the US?
We will examine proportion of total annual US PD budget allocated to each country. The PD budget pie is relatively small (when we consider the size of the overall US state department spending). How they cut it up may offer insights.
Future research that has access to country-level breakdown of US PD budgets beyond 2019 may be able to extrapolate more robust findings concerning the long-term or even causal relationships. Similar to FDI spending, it is likely that there is a sunk-cost aspect to PD spending; it may be that the fruits of PD labour do not appear for years after “initial investment”. This would be particularly evident with spending on education programmes for the youth. For these reasons, a lag of one to two years will be added to the models.
The source for this data comes from Annual Report on Public Diplomacy and International Broadcasting, published each year by the US Advisory Commission on Public Diplomacy (ACPD) since 2013.
ACPD has represented the public interest by overseeing the United States government’s international information, media, cultural, and educational exchange programs.
The Commission is a bipartisan body created by Congress, but independent from its operations. It recommends policies and programs in support of US government efforts to inform and influence foreign publics. It is mandated by law to assess the work of the State Department and to report its findings and recommendations to the President, the Congress, the Secretary of State, and the American people.
The report was compiled with the support of the U.S. Department of State’s Public Diplomacy (PD) and Broadcasting US Agency for Global Media (US-AGM) leaders who opened their databases for the ACPD to compile and sort through budget data and program descriptions from Washington and the field.
Independent Variables
One characteristic of “successful” public diplomacy campaigns is that they are long-term actions (Melissen, 2005). That is to say, they are not reactionary to short-term trends. A long time horizon can facilitate nurturing broad-base relationships with foreign publics beyond the elite level.
There may be spending inertia and a steadiness to the PD budgets to maintain long-term projects. Regarding ODA, the case study literature suggests that aid officials may develop long-term commitments to specific recipients, resulting in budget incrementalism (Hirata, 2002).
As such we add a lagged variable of previous year’s budget to account for spending that reflects long term spending.
In keeping with previous research on ODA determinants, we add GDP per capita (in constant 2010 US dollars) and population figures for each country, both from the World Bank Indicators. This is similar to ODA studies (X Y Z)
However, diverging from the ODA spending findings, we do not expect GDP per capita - a proxy for a country’s level of economic development - to be a significantly large coefficient.
Typically, ODA spending occurs in countries that are struggling with economic development. However, PD spending is evident across all regime types and all stages of economic development. While it would be regarded with surprise if Germany became a recipient of US foreign aid, Germany is one of the biggest recipients of total annual PD spending.
The reasons for increasing US favorability are not contingent on whether the recipient country is a democracy or not. Nor is it contingent on whether the country is economically developed or not. In short, we should see a divergence from ODA spending patterns. It would be absurd to argue that the US needs to invest in public diplomacy in foreign countries due to the humanitarian needs of the recipient countries for opinion change. Rather, we can focus on US PD budget spending as a tool for furthering US foreign policy objectives.
The logic is similar for a country’s level of democracy. The present study will add POLITY IV (in line with studies such as X Y Z)
To assess whether US political interests impact US public diplomacy spending.
United Nation voting affinity, measures the similarity of UN General Assembly votes between the US and states to determine if a relationship between political support and aid allocations exist. This variable has been used as a proxy for political interests by a number of studies (Dreher et al. 2010; Dreher et al. 2008; Neumayer 2005). The UN affinity variable is taken from Gartzke (2006)
Caveats
This phenomenon of changing budget categories - and evolving the definition of what is and what is not PD spending - is similar to OECD DAC categories of aid spending. The criteria for what is and is not considered aid has fluctuated over the years. The OECD/DAC has typically restrained attempts by donors to broaden the definition of official development assistance. Nevertheless, important changes have occurred. According to the OECD (1998, p. 114), these ‘changes in interpretation have tended to broaden the scope of the concept’, and as a result have had an important influence on ODA performance indicators, in particular in relation to ODA volume. In 1991, assistance provided to refugees from recipient countries during the first year after arrival in the donor country was introduced as an ODA-eligible item. This followed the addition of administrative costs (1979) and the cost of students from the South (1984) (OECD 1998, p. 114). Over time, emergency and disaster relief, debt forgiveness, changes regarding capital subscriptions, and other related themes have been critical components of ODA calculations. Most recently, and clearly linked to the current security paradigm, the DAC announced in April 2004 adjustments and clarifications on the definition of ODA relating to ‘preventing the recruitment of child soldiers, enhancing civil society’s role in the security system, and promoting civilian oversight and democratic control of the management of security expenditure’ (OECD 2004).
Purchase Power Parity
Source: https://data.worldbank.org/indicator/PA.NUS.PPPC.RF
Price level ratio is the ratio of a purchasing power parity (PPP) conversion factor to an exchange rate. It provides a measure of the differences in price levels between countries by indicating the number of units of the common currency needed to buy the same volume of the aggregation level in each country.
Indicator source: International Comparison Program, World Bank
Russian sphere
soviet_iron_curtain_vector <- c("Albania",
"Bulgaria",
"Czech Republic",
"Poland",
"Kosovo",
"Romania",
"Hungary",
"Slovakia",
"Bosnia and Herzegovina",
"Croatia",
"Macedonia",
"Montenegro",
"Serbia")
the_df %<>%
dplyr::mutate(soviet_iron_curtain = ifelse(country %in% soviet_iron_curtain_vector, 1, 0))
plyr::count(the_df$soviet_east_sphere)[1] freq <0 rows> (or 0-length row.names)
soviet_republics <- c("Lithuania",
"Georgia" ,
"Estonia" ,
"Latvia" ,
"Ukraine" ,
"Belarus" ,
"Moldova" ,
"Kyrgyzstan",
"Uzbekistan",
"Tajikistan",
"Armenia" ,
"Azerbaijan",
"Turkmenistan",
"Russia" ,
"Kazakhstan")
the_df %<>%
dplyr::mutate(soviet_republic = ifelse(country %in% soviet_republics, 1, 0))
the_df %<>% mutate(russian_sphere = ifelse(level_contig_rus == 1, 1,
ifelse(level_contig_rus == 2, 1,
ifelse(soviet_republics == 1, 1,
ifelse(soviet_iron_curtain == 1, 1, 0)))))China Sphere
library(haven)
aiib <- read_dta("C:/Users/Paula/Desktop/AIIB-ForResubmission.dta")
aiib %<>%
select(ccode:adbnonregionalmember, chinaalignment:usmilitaryalliance)
the_df <- merge(the_df, aiib, by.x = c("cow_code"), by.y = c("ccode"))
the_df %<>%
mutate(china_sphere = ifelse(cow_code == 710, 1,
ifelse(level_contig_china == 1, 1,
ifelse(chinaalignment > 8800, 1, 0))))
the_df %>%
filter(year == 2019) %>%
select(country, chinaalignment) %>%
arrange(desc(chinaalignment)) %>%
filter(chinaalignment > 8700) %>%
kbl() | country | chinaalignment |
|---|---|
| Vietnam | 9407.000 |
| Burma | 9340.000 |
| Laos | 9223.000 |
| Cuba | 9175.999 |
| Zimbabwe | 9109.000 |
| Libya | 9086.000 |
| Yemen | 9070.000 |
| Sudan | 9061.000 |
| Oman | 9027.000 |
| Egypt | 9019.000 |
| Qatar | 9016.000 |
| Iraq | 9008.000 |
| Chad | 8978.001 |
| Pakistan | 8954.000 |
| Tunisia | 8943.000 |
| Djibouti | 8941.000 |
| Saudi Arabia | 8932.000 |
| Indonesia | 8931.001 |
| Mauritania | 8908.000 |
| Niger | 8906.001 |
| Malaysia | 8904.000 |
| Bangladesh | 8892.000 |
| Bahrain | 8891.000 |
| Sri Lanka | 8856.000 |
| Kuwait | 8854.000 |
| United Arab Emirates | 8854.000 |
| Morocco | 8847.000 |
| Algeria | 8846.000 |
| Gabon | 8825.000 |
| Gambia, The | 8823.000 |
| Jordan | 8821.000 |
| Guinea | 8807.000 |
| Afghanistan | 8807.000 |
| Togo | 8783.000 |
| Burkina Faso | 8779.000 |
| Lebanon | 8777.000 |
| Tanzania | 8775.000 |
| Ghana | 8771.000 |
| Senegal | 8765.000 |
| Namibia | 8762.000 |
| Nigeria | 8759.000 |
| Mali | 8729.000 |
| Madagascar | 8715.000 |
| Guinea-Bissau | 8714.000 |
| Democratic Republic of the Congo | 8706.000 |
# %>% kable_paper("striped",full_width = F)library(WDI)
ppp = WDI(indicator='PA.NUS.PPPC.RF', country="all", start=2013, end=2019)
ppp$ppp <- ppp$PA.NUS.PPPC.RF
ppp$PA.NUS.PPPC.RF <- NULL
ppp$cow_code <- countrycode(ppp$iso2c, "iso2c", "cown")
ppp %<>%
dplyr::mutate(cow_code = ifelse(country == "West Bank and Gaza", 6666,
ifelse(country == "Serbia", 345,
ifelse(country == "Hong Kong SAR, China", 997, cow_code)))) %>%
filter(!is.na(cow_code))
ppp$country <- NULL
the_df <- merge(the_df, ppp, by = c("cow_code", "year"), all.x = TRUE)
the_df$X <- NULL
the_df$country.y <- NULL
the_df %<>%
mutate(us_exports_pc = us_exports_to_country / pop ) %>%
mutate(ppp_budget_div = budget / ppp) %>%
mutate(ppp_div_pc = ppp_budget_div / pop )country <- "United States"
countries_dataframe <- show_countries()Generating URL to request all 299 results
inflation_dataframe <- retrieve_inflation_data(country, countries_dataframe)Retrieving inflation data for US Generating URL to request all 61 results
# Provide a World Bank API URL and `url_all_results` will convert it into one with all results for that indicator
original_url <- "http://api.worldbank.org/v2/country"
# "http://api.worldbank.org/v2/country?format=json&per_page=304"
url_all_results(original_url)Generating URL to request all 299 results [1] “http://api.worldbank.org/v2/country?format=json&per_page=299”
the_df$budget_const <- adjust_for_inflation(the_df$budget,
the_df$year,
country,
to_date = 2014,
inflation_dataframe = inflation_dataframe,
countries_dataframe = countries_dataframe)
the_df$budget_pc_const <- adjust_for_inflation(the_df$budget_pc,
the_df$year,
country,
to_date = 2014,
inflation_dataframe = inflation_dataframe,
countries_dataframe = countries_dataframe)
the_df$budget_const_ppp <- adjust_for_inflation(the_df$ppp_budget_div,
the_df$year,
country,
to_date = 2014,
inflation_dataframe = inflation_dataframe,
countries_dataframe = countries_dataframe)
the_df$budget_pc_const_ppp <- adjust_for_inflation(the_df$ppp_div_pc,
the_df$year,
country,
to_date = 2014,
inflation_dataframe = inflation_dataframe,
countries_dataframe = countries_dataframe)US troops and military bases data
library(troopdata)
# the_df <- read.csv("df6.csv")
us_troops <- get_troopdata(branch = FALSE,
startyear = 2013,
endyear = 2019)
the_df <- merge(the_df, us_troops, by.x = c("cow_code", "year"), by.y = c("ccode", "year"), all.x = TRUE)
the_df$troops[is.na(the_df$troops)] <- 0
the_df$X <- NULLus_bases <- get_basedata(country_count = TRUE,
groupvar = "ccode")
map <- ggplot2::map_data("world")
basepoints <- get_basedata(host = NA)
basemap <- ggplot() +
geom_polygon(data = map, aes(x = long, y = lat, group = group), fill = "gray80", color = "white", size = 0.1) +
geom_point(data = basepoints, aes(x = lon, y = lat), color = "purple", alpha = 0.6) +
coord_equal(ratio = 1.3) +
theme_void() +
theme(plot.title = element_text(face = "bold", size = 15)) +
labs(title = "Locations of U.S. military facilities, 1950-2018")
the_df <- merge(the_df, us_bases, by.x = "cow_code", by.y = "ccode", all.x = TRUE)
basemap
the_df %<>%
mutate(troops_pc = troops / pop) %>%
mutate(suicides_pc = sum_suicides / pop) %>%
mutate(total_bases = basecount + lilypadcount)
the_df$total_bases <- the_df$total_bases[is.na(the_df$total_bases)] <- 0
the_df$bases <- the_df$basecount[is.na(the_df$basecount)] <- 0
the_df$lily_pads <- the_df$lilypadcount[is.na(the_df$lilypadcount)] <- 0the_df %<>%
group_by(year) %>%
mutate(total_budget = sum(budget)) %>%
mutate(total_budget_const = sum(budget_const)) %>%
mutate(total_budget_ppp = sum(ppp_budget_div, na.rm = TRUE)) %>%
mutate(total_budget_const_ppp = sum(budget_const_ppp, na.rm = TRUE)) %>%
ungroup() %>%
group_by(country, year) %>%
mutate(budget_proportion = budget / total_budget, na.rm = TRUE) %>%
mutate(budget_proportion_const = budget_const / total_budget_const, na.rm = TRUE) %>%
mutate(budget_proportion_ppp = ppp_budget_div / total_budget_ppp, na.rm = TRUE) %>%
mutate(budget_proportion_const_ppp = budget_const_ppp / total_budget_const_ppp, na.rm = TRUE) %>%
ungroup()Regression Analysis
dv <- "budget_proportion"
dv1 <- "budget_proportion_const"
dv2 <- "budget_proportion_ppp"
dv3 <- "budget_proportion_const_ppp"
iv <- c(
"log(pop)",
"log(gdp_pc_const)",
"electoral_demo",
"log(bases + 1)",
"log(troops + 1)",
"log(sum_attacks + 1)",
"us_trade_dependence",
"china_trade_dep",
"russia_trade_dep"
)full_model <- plm(log(budget_proportion) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = the_df,index = c("cow_code", "year"), model = "within", effect = "time")
full_model2 <- plm(log(budget_proportion_const) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = the_df,index = c("cow_code", "year"), model = "within", effect = "time")
full_model3 <- plm(log(budget_const_ppp) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = the_df,index = c("cow_code", "year"), model = "within", effect = "time")
full_model3 <- plm(log(budget_pc_const) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = the_df,index = c("cow_code", "year"), model = "within", effect = "time")
full_model4 <- plm(log(budget_proportion_const_ppp) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = the_df,index = c("cow_code", "year"), model = "within", effect = "time")
stargazer(full_model, full_model2,
# full_model3,
full_model4,
model.numbers = FALSE,
type = "html")| Dependent variable: | |||
| log(budget_proportion) | log(budget_proportion_const) | log(budget_proportion_const_ppp) | |
| log(pop) | 0.421*** | 0.421*** | 0.428*** |
| (0.016) | (0.016) | (0.018) | |
| log(gdp_pc_const) | 0.045** | 0.045** | -0.122*** |
| (0.020) | (0.020) | (0.022) | |
| electoral_demo | 0.199* | 0.199* | -0.246** |
| (0.103) | (0.103) | (0.112) | |
| log(troops + 1) | 0.075*** | 0.075*** | 0.065*** |
| (0.011) | (0.011) | (0.012) | |
| log(sum_suicides + 1) | 0.221*** | 0.221*** | 0.267*** |
| (0.024) | (0.024) | (0.027) | |
| us_trade_dependence | 0.357* | 0.357* | 0.284 |
| (0.191) | (0.191) | (0.207) | |
| china_trade_dep | -0.202 | -0.202 | -0.323** |
| (0.143) | (0.143) | (0.154) | |
| russia_trade_dep | 3.100*** | 3.100*** | 4.466*** |
| (0.407) | (0.407) | (0.440) | |
| Observations | 971 | 971 | 966 |
| R2 | 0.614 | 0.614 | 0.634 |
| Adjusted R2 | 0.608 | 0.608 | 0.629 |
| F Statistic | 189.947*** (df = 8; 956) | 189.947*** (df = 8; 956) | 206.221*** (df = 8; 951) |
| Note: | p<0.1; p<0.05; p<0.01 | ||
lm_formula <- as.formula(paste(dv, paste(iv, collapse = " + "), sep = " ~ "))
lm_formula1 <- as.formula(paste(dv1, paste(iv, collapse = " + "), sep = " ~ "))
lm_formula2 <- as.formula(paste(dv2, paste(iv, collapse = " + "), sep = " ~ "))
lm_formula3 <- as.formula(paste(dv3, paste(iv, collapse = " + "), sep = " ~ "))corr_vector <- c ("budget_proportion_const_ppp", "polity_index", "gdp_pc_const", "pop", "basecount", "troops", "suicides_pc", "china_exports_pc", "us_exports_pc", "ideal_point_distance")
corr_vector_no_un <- c( "budget_pc_const_ppp", "polity_index", "gdp_pc_const", "pop", "basecount", "troops", "suicides_pc", "china_exports_pc", "us_exports_pc")
the_df %>%
select(corr_vector) -> try_matrix
kor_cov_no_na <- na.omit(try_matrix)
conf_test2 = cor.mtest(kor_cov_no_na, conf.level = 0.95)
corrplot(cor(kor_cov_no_na, method = "pearson"),
p.mat = conf_test2$p,
method = 'circle',
type = 'lower',
insig ='blank',
addCoef.col ='black',
number.cex = 0.8,
# order = 'AOE',
diag = FALSE,
tl.srt = 45)
Correlates in Asia
Correlates in Latin America
Correlates in MENA
Correlates in Post-Soviet countries
Correlates in Africa
Correlates in the West
Clustering in UN voting patterns - similar and dissimilar voting clusters
the_df %>%
ggplot(aes( x = ideal_point_distance, y = log(budget_proportion_const_ppp))) +
geom_point(aes(color = region_6)) +
facet_wrap(~year) +
ggthemes::theme_pander() +
theme(legend.position = "bottom") +
geom_vline(xintercept = 2.4, linetype = "dashed",
color = "#a4161a", size = 1)
HIGH UN voting similarity
LOW UN VOTING SIMILARITY
low_df %>%
mutate(us_exports_pc = us_exports_to_country / pop ) %>%
select(corr_vector_no_un) -> low_matrix
kor_cov_no_na <- na.omit(low_matrix)
conf_test2 = cor.mtest(kor_cov_no_na, conf.level = 0.95)
corrplot(cor(kor_cov_no_na, method = "pearson"),
method = 'circle',
type = 'lower',
insig ='blank',
addCoef.col ='black',
number.cex = 0.8,
# order = 'AOE',
diag = FALSE,
tl.srt = 45)
high_model <- plm(log(budget_proportion) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = high_df,
index = c("cow_code", "year"), model = "within", effect = "time")
low_model <- plm(log(budget_proportion) ~
log(pop) +
log(gdp_pc_const) +
electoral_demo +
log(bases + 1) +
log(troops + 1) +
log(sum_suicides + 1) +
us_trade_dependence +
china_trade_dep +
russia_trade_dep,
data = low_df,
index = c("cow_code", "year"), model = "within", effect = "time")
stargazer(high_model, low_model,
column.labels = c("Highy similar UN voting", "Dissimilar UN voting"),
type = "html")| Dependent variable: | ||
| log(budget_proportion) | ||
| Highy similar UN voting | Dissimilar UN voting | |
| (1) | (2) | |
| log(pop) | 0.457*** | 0.469*** |
| (0.034) | (0.019) | |
| log(gdp_pc_const) | -0.189*** | 0.086*** |
| (0.056) | (0.025) | |
| electoral_demo | -0.041 | 0.640*** |
| (0.340) | (0.127) | |
| log(troops + 1) | 0.023 | 0.144*** |
| (0.017) | (0.015) | |
| log(sum_suicides + 1) | -0.076 | 0.184*** |
| (0.101) | (0.026) | |
| us_trade_dependence | 0.826** | -0.036 |
| (0.360) | (0.228) | |
| china_trade_dep | -0.740*** | 0.020 |
| (0.225) | (0.169) | |
| russia_trade_dep | 0.801 | 3.502*** |
| (0.746) | (0.472) | |
| Observations | 300 | 657 |
| R2 | 0.701 | 0.657 |
| Adjusted R2 | 0.687 | 0.650 |
| F Statistic | 83.631*** (df = 8; 285) | 153.899*** (df = 8; 642) |
| Note: | p<0.1; p<0.05; p<0.01 | |
Time Series to see the relationship between budget and lagged budget
budget.ts <- ts(the_df$budget_proportion)
pacf(budget.ts)
acf(budget.ts)
lag1price <- lag(budget.ts, -1)
plot(log(budget.ts) ~ log(lag1price), xy.labels = F)
lagdata <- ts.intersect(budget.ts, lag1price, dframe = T)
summary(lm(log(budget.ts) ~ log(lag1price), data = lagdata))Call: lm(formula = log(budget.ts) ~ log(lag1price), data = lagdata)
Residuals: Min 1Q Median 3Q Max -3.1137 -0.2519 0.0116 0.2660 4.3795
Coefficients: Estimate Std. Error t value Pr(>|t|)
(Intercept) -1.09370 0.10355 -10.56 <2e-16 log(lag1price) 0.80986 0.01771 45.72 <2e-16 — Signif. codes: 0 ‘’ 0.001 ’’ 0.01 ’’ 0.05 ‘.’ 0.1 ’ ’ 1
Residual standard error: 0.6634 on 1108 degrees of freedom Multiple R-squared: 0.6536, Adjusted R-squared: 0.6533 F-statistic: 2091 on 1 and 1108 DF, p-value: < 2.2e-16
if(!require(devtools)) install.packages("devtools")
devtools::install_github("kassambara/ggpubr")
summary(the_df$ideal_point_distance)Min. 1st Qu. Median Mean 3rd Qu. Max. NA’s 0.1073 1.9416 3.0864 2.7979 3.3865 4.6349 22
the_df %>%
mutate(un_ally = ifelse(ideal_point_distance < 2.8, "Similar", "Dissimilar")) %>%
mutate(un_ally = ifelse(country == "Japan", "Similar",
ifelse(country == "Belgium", "Similar", un_ally ))) %>%
filter(!is.na(un_ally)) -> ally_df
ally_df %>%
group_by(un_ally) %>%
summarise(
count = n(),
mean_budget = mean(budget_const_ppp, na.rm = TRUE),
median_budget = median(budget_const_ppp, na.rm = TRUE),
sd_budget = sd(budget_const_ppp, na.rm = TRUE),
mean_pc = mean(budget_pc_const_ppp, na.rm = TRUE),
median_pc = median(budget_pc_const_ppp, na.rm = TRUE),
sd_pc = sd(budget_pc_const_ppp, na.rm = TRUE)) %>%
kbl %>%
kable_paper()| un_ally | count | mean_budget | median_budget | sd_budget | mean_pc | median_pc | sd_pc |
|---|---|---|---|---|---|---|---|
| Dissimilar | 681 | 8835700 | 3006220 | 22488243 | 0.5314034 | 0.2807506 | 0.9031925 |
| Similar | 415 | 4506843 | 2292558 | 7741195 | 0.5817149 | 0.2378150 | 1.0128602 |
library("ggpubr")
ally_df$region_short <- recode_factor(ally_df$region_6, "Asia and Pacific" = "Asia",
"Latin America Caribbean" = "Latin",
"Sub-Sahara" = "Africa")
ggboxplot(ally_df, x = "region_short", y = "budget_const_ppp", utlier.shape = NA,
color = "un_ally", palette = c("#00AFBB", "#E7B800"),
ylab = "Budget per capita", xlab = "Similar versus Dissimilar", repel = TRUE )
# Shapiro-Wilk normality test for Budgets
with(ally_df, shapiro.test(budget_pc_const_ppp[un_ally == "Similar"]))Shapiro-Wilk normality testdata: budget_pc_const_ppp[un_ally == “Similar”] W = 0.50785, p-value < 2.2e-16
with(ally_df, shapiro.test(budget_pc_const_ppp[un_ally == "Dissimilar"]))Shapiro-Wilk normality testdata: budget_pc_const_ppp[un_ally == “Dissimilar”] W = 0.46708, p-value < 2.2e-16
# F-test to test for homogeneity in variances.
res.ftest <- var.test(budget_pc_const_ppp ~ un_ally, data = ally_df)
library(broom)
res.ftest F test to compare two variancesdata: budget_pc_const_ppp by un_ally F = 0.79517, num df = 654, denom df = 414, p-value = 0.009176 alternative hypothesis: true ratio of variances is not equal to 1 95 percent confidence interval: 0.6668352 0.9449365 sample estimates: ratio of variances 0.795173
# T-test
t.test(budget ~ un_ally, data = ally_df, var.equal = TRUE, alternative = "greater") %>%
tidy() %>% kbl() %>% kable_paper()| estimate | estimate1 | estimate2 | statistic | p.value | parameter | conf.low | conf.high | method | alternative |
|---|---|---|---|---|---|---|---|---|---|
| 712483.8 | 3264134 | 2551650 | 2.023731 | 0.0216203 | 1094 | 132898.4 | Inf | Two Sample t-test | greater |