This year we introduce two new metrics. The first is the neighborhood risk score. As the name suggests, the score is an attempt to gauge the regional level of risk around a country. There are several reasons to do this. Experience indicates that regional context enables or amplifies many of the negative consequences from arms sales. The likelihood of negative consequences from a weapons transfer depends not only on the recipient nation but also on what is happening next door. Nations that are near unstable regimes, conflict, and active black markets for weapons are more likely to contribute to diffusion, dispersion, and other negative outcomes—even if their own risk scores are relatively low—than if they had stable neighbors.

The second new metric is the average customer risk score. As the name suggests, the average customer risk score is an easy way to determine how risky the recipient of the “typical” arms sale is in a given year. Calculating this metric allowed us to confirm in this year’s report, for example, that the American arms sales portfolio has gotten riskier since the George W. Bush administration. It also allowed us to illustrate that despite U.S. leadership in arms exports, the average customer of American arms is less risky than the customer of most of the other leading arms exporters. China’s and Russia’s average customers, for example, are significantly riskier than the average U.S. client.

We invite scholars and policymakers to read the report and download the data for further analysis.

After another booming year of sales, the United States remains the world’s dominant exporter of major conventional weapons. In 2019, the Trump administration notified Congress of at least $85 billion in sales. From 2002 to 2019, the Bush, Obama, and Trump administrations sold over $640 billion in arms to 167 different nations.¹

Arms sales can be a useful tool in America’s foreign policy, but every sale involves risks as well as potential benefits. The Arms Export Control Act of 1976 requires the executive branch to produce an assessment to ensure that the risks do not outweigh the potential benefits of selling major conventional weapons. Unfortunately, as our research indicates, there is little evidence that the United States weighs these risks fully during the approval process for weapons sales.²

The United States has sold weapons to almost any nation seeking them. Examples of the harmful side effects of U.S. arms sales are plentiful. Over the past decade, American weapons have wound up in the hands of the Islamic State and other terrorist groups, landed on the black market in Yemen and elsewhere, been used by oppressive governments to kill their own people, and enabled nations to engage in bloody military conflicts. American arms sales have helped prop up authoritarian regimes, encouraged military adventurism, spurred arms races, and amplified existing conflicts. We created the Arms Sales Risk Index in 2018 and continue to update it to give U.S. policymakers and researchers a way to measure these risks.³

The Trump administration’s arms sales record has been even more questionable than that of its predecessors. As we illustrate in this year’s report, the Trump administration has sold more weapons to a riskier portfolio of clients than either the Bush or the Obama administrations did. The most visible sign of this trend has been the Trump administration’s continued sales of weapons to Saudi Arabia despite the tragic consequences of its military intervention in Yemen. But sales to the Philippines and Nigeria, for example, also come with significant concerns about the prospects of weapons winding up in the wrong hands or being misused by those governments.

Over the past two years, Congress has tried to exert more influence to limit and shape U.S. arms sales. In April 2019, the Senate joined the House in passing a war powers resolution requiring the United States to stop supporting Saudi Arabia’s war in Yemen. In the wake of Trump’s veto of that resolution, a bipartisan group of senators introduced the Enhancing Human Rights Protections in Arms Sales Act, designed to ensure that “U.S. manufactured weapons are not used in the commission of heinous war crimes, the repression of human rights, or by terrorists who seek to do harm to Americans and innocent civilians abroad.”⁴ In February 2020, Rep. Ilhan Omar (D-MN) introduced the Stop Arming Human Rights Abusers Act, which would prevent the United States from selling weapons or providing security assistance to any government that has committed gross violations of human rights.⁵

The world was slightly riskier in 2019 (average risk score 39.1) than in 2018 (38.7) but still a bit less risky than things were in 2017 (41.7). Figure 1 shows the level of risk throughout the world; high‐​risk zones are somewhat predictable year to year.

We constructed the Arms Sales Risk Index (ASRI) by identifying four different causal pathways by which arms sales might lead to negative outcomes. These four “risk vectors” include a country’s level of corruption, its stability, its treatment of its people, and the level of conflict, both internal and external, in which it is engaged. To measure these and to calculate the overall risk score for each nation, presented in Table 1, we average their scores on six independent indices (see Appendix A for a complete methodology):

• Corruption: Corruption makes it more likely that weapons will be stolen or sold to unintended customers. To assess this factor, we rely on Transparency International’s Corruption Perceptions Index, which ranks 180 countries and territories by their perceived levels of corruption.
• Instability: Instability also correlates with a much higher likelihood of weapons being stolen or misused by governments and other groups. We assess instability using the Fragile States Index, produced by the Fund for Peace.
• Domestic human rights: States that have a poor record of human rights and/​or regularly use violence against their own citizens pose a greater risk of misusing weapons. We rely on two indices to measure these factors: Freedom House’s Freedom in the World index and the U.S. Department of State’s Political Terror Scale.
• Conflict: States engaged in conflict are inherently riskier when it comes to factors such as dispersion, blowback, entanglement, and human rights abuses. To assess these risks, we rely on the Global Terrorism Index and the Uppsala Conflict Data Program/​Peace Research Institute Oslo Armed Conflict Dataset.

Table 2 shows the biggest increases and decreases in country risk scores between 2018 and 2019. Each year, the maximum shift in risk score is roughly 14 to 18 points in either direction—getting riskier or less risky. That variance holds over the 2017–2019 time frame as well. But although there is plenty of movement around the world each year, there is not always much correlation year to year. For example, Syria got 16 points riskier between 2017 and 2018 but 1 point less risky in 2019. The 15 countries with the biggest 2018 to 2019 risk increases averaged an eight‐​point increase in 2019 but only a one‐​point increase the prior year—with several moving in the opposite direction.

Overall, 61 countries have gotten riskier (defined by a more than one‐​point increase in risk score) since 2018. Thirty of those countries have gotten significantly riskier (defined by a more than five‐​point increase in risk score). Of those 30 countries that got significantly riskier, 12 countries had double‐​digit increases to their risk scores since 2018.

Only five countries scored the same each year from 2017 to 2019. Of the other countries, 117 have gotten less risky (defined by a more than one‐​point decrease); 67 have gotten significantly less risky (defined by a more than a 5‐​point decrease), with 40 countries decreasing their risk scores by over 10 points.

A closer examination of the trends indicates that state violence drives the most change from year to year. Countries with troubled human rights histories have seen their risk scores rise after states tried to suppress civil liberties or attacked civilians. These changes validate the risk index by illustrating how and why political violence and human rights records are related to the downstream consequences of arms sales. It also illustrates a fundamental dilemma with arms sales: that risk assessments are a moving target. Given the decades‐​long lifespan of many weapons systems, understanding and tracking how arms sales risks change over time is critical.

The second most influential factor in annual variance is terrorism, primarily thanks to ebbs and flows in the impact of terrorism in the Middle East. While this factor could lead to more dispersion of stockpiled weapons in a recipient nation or more active black markets of goods, it does not speak to a state‐​based balance of power. Logically, increases and decreases in terrorism would also have far more impact on the small arms and light weapons trade that could be more easily diverted from national control than major conventional systems such as fighter jets.

The 2020 ASRI data set includes a new way to assess a potential vector for arms sales risk: risky neighbors. Many of the risk vectors that the ASRI measures are amplified when they cross borders. Interstate wars are the clearest cross‐​border danger, but even risks such as the diversion of weapons from a national stockpile to terrorist groups or black markets have serious implications for neighboring nations with porous borders. To assess this risk, we calculated a neighborhood risk score for each country.

The ASRI determines a country’s neighborhood risk as the total of all its neighbors’ risk scores, rather than as an average (see Appendix A for details). Each neighbor is a potential pathway in and out of a nation for illicit weapons—and a neighbor to race against for arms or conflict with. The average global neighborhood risk score is 249; Figure 2 shows which countries have the highest overall risk score. China, surrounded by many neighbors, tops the chart with a neighborhood risk score of 899. Fiji has the least risky neighborhood (thanks to having only two neighbors) with a neighborhood risk score of six.

The risk score data set is very useful for tracking trends and highlighting the dangers of imprudent arms sales. Large‐​N data, however, cannot substitute for closer analysis of specific cases. This year we have written brief risk profiles to provide an illustration of the sort of analysis that governments should be conducting when considering selling weapons around the world.

Figures 3.1–3.4 present the data for Turkey’s risk profile.

Since the foiled coup d’état in 2016, Turkish president Recep Tayyip Erdoğan has cracked down on political opponents and freedom of speech while centralizing power in his office.⁶ Turkey’s unsettled domestic political situation has driven its arms sales risk score from 63 in 2018 to 77 in 2020, which lends weight to recent complaints about Turkey’s participation in the North Atlantic Treaty Organization (NATO) F-35 Lightning II aircraft development project.⁷ Turkey’s location only amplifies those concerns, making it a sobering case study of the risks involved with arms sales, even to allies.

Turkey has the second most risky neighborhood with a neighborhood risk score of 886. Turkey became a NATO member in 1952 and enjoyed the decreased U.S. congressional oversight required for exporting American‐​made weapons to trusted NATO countries. Since joining NATO, Turkey has bought more than $20 billion worth of American arms—from aircraft to artillery—including F‐​35s.⁸

Turkey’s inclusion in the F-35 program became problematic several years ago when the country began entering into negotiations with its neighbor, Russia, to purchase the S-400 surface‐​to‐​air missile system.⁹ Turkey placing the S-400 system near F‐​35s raises legitimate security concerns; the air defense system could gather important data on the F-35’s systems and highly classified features.

By procuring the S-400, Turkey endangered its role in the F-35 program: Ankara had planned to purchase 100 F-35A aircraft and had negotiated offsets with the deal that allowed Turkish defense companies to build some of the parts for the aircraft.

U.S. Congress members and U.S. Department of Defense officials issued consistent statements before, during, and after Turkey’s S-400 acquisition that the country would be cut off from the F-35 program as a consequence of the sale and even threatened to impose sanctions on the NATO country.¹⁰

But even as Washington diplomats attempted to exert leverage on Turkey to stop the sale by threatening sanctions and nullification of the existing F-35 deal, Ankara continued negotiations with Moscow and eventually bought the S-400 system.¹¹

This example shows that although the United States deemed it acceptable for Turkey, a NATO member with codified security alliance ties to the United States, to purchase F‐​35s several years ago, changing security concerns and arms imports from other countries have rendered Turkey a high‐​risk recipient for this technology.

Saudi Arabia is another nation whose risk score has risen recently, jumping from 60 in 2018 to 71 in both 2019 and 2020, as illustrated in Figure 4.2. Figures 4.1–4.4 present the data for Saudi Arabia’s risk profile. Saudi Arabia also claims the third‐​highest neighborhood risk score with a score of 816. Since 2001, Saudi Arabia has purchased over $35 billion of American weapons, which have fueled instability thanks to the kingdom’s regional ambitions.

Not only is the kingdom engaged in an ongoing battle with Iran for regional supremacy, but it is also waging a deadly war against its neighbor to the south, Yemen. The war in Yemen has made headlines for years due to Saudi Arabia’s military tactics that frequently lead to civilian deaths. According to Human Rights Watch, since the conflict began in 2015, over 17,500 civilians have been killed or injured.¹² More than 20 million people in the country experience food insecurity, with 10 million at risk of famine. And yet, in 2017, on his first international press tour, President Trump announced over $110 billion worth of forthcoming arms sales to the kingdom.¹³

In short, Saudi Arabia’s misuse of American weapons illustrates a wide range of the potential pathways for the negative side effects of arms sales. Not only have the sales increased tensions between the United States and Iran as Saudi Arabia and Iran fight their proxy war in Yemen, they have also enabled civilian casualties and prolonged that war.¹⁴ American weapons have also been stolen and sold from the battlefield in Yemen through black market networks that operate in nearby Somalia, further exacerbating problems of violence and terrorism across the region.¹⁵

As Venezuela’s political situation has descended into chaos and civil conflict, its ASRI score has risen from 46 in 2018 to 60 in 2020, as Figure 5.2 shows. Figures 5.1–5.4 present the data for Venezuela’s risk profile. Venezuela’s immediate neighborhood is relatively stable—its neighborhood risk score is 276. Given its domestic troubles, Venezuela has been on and off the embargoed list of nations deemed unsuitable for American weapons exports, but its recent history illustrates the relevance of considering risk factors—at both the state and regional levels—when thinking about the impacts of arms sales.

Though Venezuela has not bought much in the way of major conventional weaponry from the United States, it has purchased a great deal from Russia over the past two decades. According to a Foreign Policy article by Ryan C. Berg and Andrés Martínez‐​Fernández, “Between 1999 and 2019, billions of dollars’ worth of Russian arms, financed through Russian loans, poured into the country.”¹⁶

Thanks to high levels of corruption and economic distress, the security of Venezuela’s abundant weapons stockpile is increasingly in doubt.¹⁷ With leadership in dispute and disarray, disgruntled or desperate generals could look to sell off the country’s arsenal for profit. The Venezuelan military consistently ranks high on corruption indexes, and multiple top‐​tier military officials have ties to regional guerrilla and criminal organizations—many of which already traffic drugs and weapons throughout the region.¹⁸

Venezuela’s increasing instability has led, in turn, to increased American arms sales to at least one of its neighbors. U.S. Navy Adm. Craig Faller, commander of U.S. Southern Command, told reporters in 2019, “The conditions in Venezuela, the same chronic conditions that have just created abject suffering for the Venezuelan people, have created a lawless zone.” As a result, American diplomats offered 15 F-16 Fighting Falcon jets to Colombia “to prevent a crisis from spilling over into neighboring Colombia, an important ally, by beefing up its defenses.”¹⁹

The United States brokered arms deals worth $85.1 billion in 2019 alone, the Trump administration’s highest annual total to date.²⁰ Over the past 10 years, the United States has sold over $183.8 billion worth of weapons. The heat map in Figure 6 shows the geographic distribution of all U.S. weapons sales around the world since 2002. As Table 3 shows, since 2002, America’s biggest customers have purchased an average of $14 billion in major conventional weapons.

From a total dollar value perspective, military aircraft and their associated parts make up the majority of American exports over the past two years, with bombs and missiles coming in second and ground vehicles third. (See Figure 7 for a detailed breakdown of arms sales categories for 2018 and 2019.) What does this mean from a risk perspective?

The total dollar value of an arms sale is not necessarily the best metric by which to measure either its risks or its benefits. The balance of military power is tricky to measure, and different weapons systems carry different risks. Though military aircraft could be used to engage in an interstate war, a civil war, or an air campaign against terrorist groups, for example, aircraft are also at the lowest risk for consequences such as diversion.

On the other hand, the “firearms, guns, and ammunition” category, which acts as a catch‐​all for small arms and light weapons, includes equipment that is frequently at high risk of diversion and often used to carry out human rights abuses. Thus, although the overall dollar figures involved are often orders of magnitude lower, the risks with respect to increased violence and black marketeering are likely much higher in most cases. Many researchers in this field are still trying to analyze the pathways and indicators surrounding the abuse or misuse of small arms and light weapons. In future work we hope to more carefully map weapon types to specific risks and benefits to assess the potential consequences of the international flow of weapons and American arms sales in particular.

Since the George W. Bush administration, there have been two notable trends indicating that the United States is taking increasing levels of risk when it sells weapons to other states. First, the United States is now selling more weapons on an annual basis than it did before, a trend that began under the Obama administration and has continued under Trump.²¹ In 2019, the Trump administration notified Congress of $85 billion in sales.

The second sign that the United States is embracing greater risks is that the riskiness of the average American arms customer, weighted by dollar value, has also been increasing (see Appendix A for details). Under the Bush administration, the average risk score of American customers was 37; that score rose to 39 under the Obama administration (also the average risk score across all nations in 2020), and it has risen to 41 under the Trump administration. In short, not only is the United States selling more weapons than before, but it is also selling more of them to riskier customers.

America’s average customer risk score can rise without any intention on the part of the United States if its customers’ risk scores change from year to year. What is worrisome, however, is that recent policy changes have made it more likely that the United States will export weapons to risky customers.

In 2018, the Trump administration rolled out a controversial update to the Conventional Arms Transfer Policy, the White House’s primary policy guidance on arms sales. The most concerning change is the explicit inclusion of benefits to the American economy as a factor in whether to sell weapons abroad. The previous presidential directive on arms sales, Obama’s Presidential Policy Directive 27 from January 2014, identifies a host of criteria to be included in risk assessments and declares the following:

The 2018 policy guidance from the Trump administration strikes a markedly different tone:

The Trump administration’s transactional approach and emphasis on the defense industrial implications of arms sales to create American jobs elevates supply‐​side benefits of the arms trade and creates a system that doesn’t adequately value the long‐​term implications of selling weapons abroad, especially as it pertains to potential human rights abuses.²⁴

In 2020, the administration changed the arms sales process and oversight functions to cut red tape associated with the foreign military sales process. The bureaucracy surrounding foreign military sales is often critiqued as inefficient and slow to the point of having a detrimental effect on potential sales. The Trump administration adopted a policy that the Obama administration originally proposed—transitioning firearms (both nonautomatic and semiautomatic) and their associated ammunition from the State Department’s U.S. Munitions List (USML) to the purview of the Commerce Department’s Commerce Control List.²⁵

Changing the export controls around this category of weapons system may speed the arms sale approval process, but it is also likely to reduce the efficacy of the risk assessment process and remove critical oversight functions regarding these sales. Because the Commerce Department is not required to notify congressional committees before finalizing sales from the Commerce Control List, Congress will lose the ability to provide any meaningful oversight.²⁶

This policy change also presents problems for transparency. While the Foreign Military Sales and Foreign Military Financing programs both have fairly robust transparency and reporting measures to facilitate congressional oversight, sales conducted through the Commerce Control List have much less stringent reporting requirements. It is already difficult to collect and publish data on weapons sold from the list, which represents a significant problem as small arms and light weapons are some of the riskiest exports when it comes to humanitarian concerns.²⁷

These policies create a system wherein some of the riskiest weapons are exported more freely worldwide with less emphasis on potential impacts, without any oversight, and with a complete lack of transparency.

The United States is not the only nation engaged in the international arms trade. Nor does it have the riskiest customer base when compared with other leading arms‐​exporting nations. For each of the top 10 arms exporters, we used arms sales data from the Stockholm International Peace Research Institute and our index to calculate the risk score of the typical recipient nation (see Appendix A for details). As Figure 8 shows, China sells to the riskiest customer base; its average customer has a risk score of almost 70. At the low end of the chart is the Netherlands, whose more responsible approach to arms sales is reflected in an average customer risk score of 16. The United States, which accounts for 36 percent of the arms trade over the past five years, has an average customer risk score of 40, putting it in the middle of the pack.²⁸

The Trump administration has pursued the same basic approach to arms sales as previous administrations. With a few exceptions, the United States is ready to sell weapons to any nation with the money to buy them. The risk of arms transfers to bad actors and other negative outcomes continue to make little discernable impact on American decisions about where to sell weapons. Nations involved in disastrous conflicts, nations with ugly human rights records, and countries struggling to manage crime, corruption, and terrorism have all purchased millions of dollars of weapons from the United States over the past two decades. History suggests that most of these arms sales will do little to improve American security and that some of the weapons will end up in the wrong hands and be used for the wrong reasons.

Moving forward, the United States should incorporate the risks of arms sales more fully in the arms sales process. An obvious first step is to halt sales to the riskiest nations—any nation scoring at the bottom of any of the risk index metrics. Banning sales to serial violators of human rights, to nations at war, or to extremely fragile or corrupt states, for example, would prevent many of the most predictable and most disastrous harms from occurring. A second useful step would be to ensure that Congress approves all arms sales, rather than forcing Congress to vote down arms sales as the process works today. Arms sales can be an important tool of U.S. foreign policy, but history shows that the benefits do not automatically outweigh the costs. Requiring a vote in Congress would help ensure debate on the merits of each case.²⁹

For a more detailed discussion of the risks from arms sales, please see A. Trevor Thrall and Caroline Dorminey, “Risky Business: The Role of Arms Sales in U.S. Foreign Policy,” Cato Policy Analysis no. 836, March 13, 2018.

The potential negative side effects of arms transfers take many forms. One extreme example is blowback—when American weapons end up being used against American interests. After the Iranian Revolution in 1979, the revolutionary government took possession of billions of dollars’ worth of American fighter jets and other weapons, an arsenal that Iran has used to assert itself ever since. A more common example is when American troops end up fighting other forces armed with American‐​made weapons that the United States had willingly provided, as happened in Somalia in 1991 with weapons sold during the Cold War.

Arms sales and transfers can also harm the regions into which American weapons flow. This frequently occurs via dispersion—when weapons sold to a foreign government end up in the hands of criminal groups or adversaries. This risk is highest with sales or transfers to fragile states that are unprepared, unwilling, or too corrupt to adequately protect their stockpiles. For instance, despite America’s efforts to train and equip the Iraqi army, Islamic State fighters in 2014 captured three Iraqi army divisions’ worth of American equipment, including tanks, armored vehicles, and infantry weapons.

U.S. arms sales can also prolong and intensify interstate conflicts. Although the goal might be to alter the military balance of a conflict to facilitate a speedy end, sending weapons can also encourage the recipients to continue fighting even with no chance of success, leading to more casualties.

Finally, U.S. weapons sales in the name of battling terrorism and insurgency undermine U.S. national security when they are made to corrupt regimes and to nations with a history of human rights violations. American firepower can enhance regime security and enable oppressive governments to mistreat minority groups and wage inhumane actions against insurgents or terrorist groups. Currently, Saudi Arabia is waging war in Yemen primarily using American weapons, which the United States has continued to provide even though the Saudis have been cited repeatedly for human rights violations and targeting civilian populations. In countries where serious corruption is endemic, American weapons can be diverted from their intended recipients and wind up in the wrong hands. For example, as a result of military and police corruption, the small arms and light weapons that the United States sends to Mexico and several other Latin American countries in support of the war on drugs often facilitate the very crimes they were meant to stop.

Unfortunately, no historical data exist at present that would allow analysts to measure how often these various outcomes occur, or under what circumstances they are more or less likely. These data should be collected. For now we must assess risk by relying on historical and analytical work on arms sales and their impacts to identify factors that are likely candidates for increasing the probability of negative consequences.

Given the limits of our current knowledge, we take what we believe to be a conservative approach based on straightforward assumptions, using available data, about the correlations between risk factors and negative outcomes. We do not claim to make precise predictions about the probability of specific outcomes according to the Arms Sales Risk Index (ASRI). The ASRI’s immediate usefulness lies in helping policymakers incorporate the consideration of risks in decisions about selling and transferring American weapons abroad. In the longer run, the ASRI is an important step in developing a more empirically grounded assessment of the consequences, both positive and negative, of arms transfers.

The ASRI contains four risk vectors. Each vector, in essence, represents a causal mechanism or an explanation of how arms sales can lead to negative consequences. We measure the strength of each of these vectors by assessing six components. The risk vectors include a state’s level of corruption, the state’s stability, the state’s treatment of its people, and the level of conflict, both internal and external, in which the state is engaged.

To be included in the ASRI, a risk factor had to meet the following criteria:

• It had to be mentioned in the literature, or logic had to strongly suggest its inclusion.
• It had to be something people can measure in a reliable fashion.
• The data source it came from had to be one that is regularly updated.
• The data source had to be a credible one that used a transparent methodology.
• The component had to complement and not overlap too much with other components already in the index.

The first risk vector we consider is the corruption of the recipient nation’s regime. We assume that states that prevent their citizens from gaining income, have fake elections, and have questionable business practices pose a greater risk of misusing weapons, weapon dispersion, and human rights abuses. To assess this factor, we rely on Transparency International’s Corruption Perceptions Index, which ranks 180 countries and territories by their perceived levels of corruption. It includes 16 different surveys from 12 different institutions to create a composite “corruption perception score.”

The second risk vector we consider is the stability of the recipient nation. We assume that fragile states with tenuous legitimacy that also lack the ability to deliver services and cannot manage conflict within their own borders pose a greater risk for dispersion and misuse of weapons. The Fragile States Index, produced by the Fund for Peace, examines international conditions that lead to state fragility. It works by examining when these pressures outweigh a state’s capability to maintain stability. It uses 12 indicators (security apparatus, factionalized elites, group grievance, economic decline and poverty, uneven development, human flight, state legitimacy, public services, human rights, demographic pressures, refugees, and external intervention) to create a composite score for “state fragility.”

The third risk vector we examine is a state’s behavior toward its citizens. States that have a poor record of human rights and/​or regularly use violence against their own citizens pose a greater risk of misusing weapons. We rely on two indices to measure these factors: Freedom House’s Freedom in the World index and the U.S. Department of State’s Political Terror Scale. The Freedom House index includes measurements for electoral process, political pluralism and participation, functioning of government, freedom of expression, associational rights, rule of law, and personal autonomy.

The U.S. Department of State’s Political Terror Scale is a yearly report measuring physical integrity rights violations worldwide. It looks at political violence and terror that a country experiences in a particular year using a five‐​level terror scale.

Finally, we also consider conflict as a critical vector for negative consequences. States currently engaged in conflict are inherently riskier when it comes to factors such as dispersion, blowback, entanglement, and human rights abuses. To assess these risks, we rely on the Global Terrorism Index and the Uppsala Conflict Data Program/​Peace Research Institute Oslo (UCDP/PRIO) Armed Conflict Dataset. The Institute for Economics and Peace annually publishes the Global Terrorism Index, which provides a comprehensive summary of global trends and patterns in terrorism since 2000. It looks at total number of terrorist incidents in a given year, total number of fatalities caused by terrorism in a given year, total number of injuries caused by terrorism in a given year, and the approximate level of total property damage caused by terrorism in a given year. It combines these factors to create a composite score that ranks the amount of terrorism facing 163 countries. The UCDP/PRIO Armed Conflict Dataset is used to track the severity of armed conflict. It uses three measures (high level, low level, and no conflict) to assess the strength of a conflict in a given location. Larger conflicts are inherently riskier because of the longer duration and greater damage.

There is some overlap across indices. For example, the Fragile States Index includes variables that measure the presence of conflict, as do the Global Terrorism Index and UCDP/PRIO Armed Conflict Dataset. But as the correlation matrix in Table 4 indicates, even though the variables are positively correlated with one another, all of them appear to bring something different to the ASRI.

To combine each of the risk factors, themselves constructed using a variety of different scales, we began by normalizing the scores for each risk factor. We normalized the variables using Min‐​Max methods. Unless otherwise noted, normalization follows the following formula: , where sk′ represents the normalized score of the measurement variable s for observation k.

After normalization, for each measure every country has a score between 1 (the lowest risk) and 100 (the highest risk). We then weighted each of the six components equally, averaging them to find the composite risk score for each country. We choose to weigh the six components equally rather than the four vectors because we do not yet have any empirical reason to consider different weighting systems and because although they are similar, the pairs of factors that measure state behavior and conflicts measure very different types of specific data. It is worth noting that while two countries may have the same overall score, they may be risky for different reasons. Thus, it is important to pay attention to the individual components as well as their risk index scores.

The result of this process is a common‐​sense index that provides a starting point for policymakers to rethink the risk assessment process, although it unavoidably sacrifices the nuance of the individual components. But because our approach measures only significant differences in state freedom, stability, and so forth rather than attempting to claim undue precision, there is good reason to believe that nations scoring higher on this index are indeed riskier customers even though we cannot be certain about the precise weighting of different components (an area for future research).

Neighborhood risk uses the Correlates of War Direct Contiguity data set that codes a state as a “neighbor if it shares a land border or if they are separated by 400 miles of water” (the maximum distance at which two 200‐​mile exclusive economic zones can intersect).

We then compile the total risk of every country’s neighbor to create a “neighborhood risk score.” This allows us to check the risk of countries that are in very risky neighborhoods but otherwise domestically stable. Examples of this include India (58th most risky, 8th riskiest neighborhood, over $2.6 billion in arms sales since 2009); Greece (123rd most risky, 30th riskiest neighborhood, over $2.4 billion in arms sales since 2009); Qatar (126th most risky, 39th riskiest neighborhood, over $2.2 billion in arms sales since 2009); Oman (106th most risky, 14th riskiest neighborhood, over $1.5 billion in arms sales since 2009); and Indonesia (55th most risky, 20th riskiest neighborhood, over $944 million in arms sales since 2009).

Additionally, the neighborhood risk metric allows us to track what neighborhoods are the riskiest. Countries in the Persian Gulf and Middle East tend to receive a substantial amount of government‐​to‐​government weapons sales, but the neighborhood itself amplifies the risk of these sales. In Central America, while the individual countries see significant individual risk, they also see an equal amount of neighborhood risk—an unfortunate trend given that the region receives a substantial amount of direct commercial sales.

As its name suggests, the average customer risk score is a common‐​sense measure of how risky the average arms sales customer of any given arms‐​exporting nation is in a given year. The more weapons a nation sells to riskier nations, the higher the score will be. If a nation mostly sells weapons to low‐​risk clients, the average customer risk score will also be low. We calculate the score by multiplying each recipient country’s risk score by the dollar amount of sales it purchased and then dividing by the total sales in dollar amounts made that year.

For example, if the United States sold $1 million of weapons to a nation with a risk score of 100, the U.S. average customer risk score would be (100 x 1,000,000)/1,000,000 = 100. If the United States sold $1 million of weapons to a nation with a risk score of 100 and $1 million to a nation with a risk score of 0, the average customer’s risk score would be (100 x 1,000,000) + (100 x 0)/(1,000,000 + 1,000,000) = 50. Where it gets more interesting is when the dollar amounts vary, as they do in the real world. For example, if the United States sells $2.5 million of weapons to a nation with a risk score of 0 but sells $7.5 million to a nation with a risk score of 100, the average customer risk score would be (100 x 7,500,000) + (0 x 2,500,000)/(7,500,000 + 2,500,000) = 75.

Security Assistance Monitor

Freedom House, Freedom in the World 2018

Transparency International, Corruption Perceptions Index 2018

The Fund for Peace, Fragile States Index 2018

The Political Terror Scale, U.S. State Department Reports on Human Rights Practices

Uppsala University and Peace Research Institute Oslo, Armed Conflict Dataset

Institute for Economics and Peace, Global Terrorism Index 2018

Correlates of War Project. Direct Contiguity Data, 1816–2016. Version 3.2.

The views expressed in this paper are those of the author(s) and should not be attributed to the Cato Institute, its trustees, its Sponsors, or any other person or organization. Nothing in this paper should be construed as an attempt to aid or hinder the passage of any bill before Congress.