Introduction

Contracting war-related functions raises serious concerns regarding regulation, especially when private military and security companies (PMSC) take a greater part in the execution of security policies across the globe. War-making on the part of standing armies has been traditionally curbed and regimented by international laws and conventions as well as domestic checks and balances (Bongard and Somer 2011; Gray 2018; Lebovic and Voeten 2006; Swed 2018, Coletta and Feaver 2006). Even if not perfect (Bussmann and Schneider 2016), these regulations contributed to the restriction of state-led violence and its relative transparency. Conversely, PMSCs are not as well-regulated as standing armies. Furthermore, they recurrently operate in the margins of international restrictions and traditional checks and balances (Avant 2005, Singer 2003). Those regulatory gaps have been discussed extensively in academic and professional scholarship, highlighting gaps in transparency, accountability, PMSC impunity (Chapman 2010, Dickinson 2011, Jordan 2009, Leander 2010, White and MacLeod 2008), and underscoring a principal-agent problem (McCoy 2010). Scholarship has been focused on the regulation of PMSCs and singled out the companies themselves as the main concern, identifying them as rogue actors that require special regulation (Cullen 2000). In this paper, we would like to refocus this discussion towards the contracts used for PMSCs’ employment. We assert that the type of contract, and therefore the setting where it occurs, is more likely to predict contractor misconduct than whether the contractor is a PMSC or non-PMSC; consequently, the contracts and how they are regulated should be the main concern of any discussion. Contracts to provide military services in conflict areas come with severe implications for the prospect of inspecting and auditing any contractor’s activities. The government has weak oversight due to the oversight barriers implicit in conflict areas, and can consequently be considered an incapable guardian against crime (Cohen and Felson 1979), which cultivates an opportunity structure for white-collar crime (Benson and Simpson 2009). Per our argument, white-collar crimes are more likely to occur in conflict areas where the oversight is faint, regardless of whether the company involved is generally considered a PMSC or not. Or in other words, a hard to oversee setting, such as conflict areas, can dramatically increase the cost for the principal.

To examine PMSCs’ oversight, we follow Dickinson’s (2003) focus on contract as the tool for regulating PMSCs. We examined 2380 instances of misconduct tracked by the Federal Contractor Misconduct Database (FCMD) as of January 10, 2018. When referring to our data, we specify if we are referring to violations, which is one of our variables, or misconduct, as a general term, as they have different implications for our analysis. We use the term misconduct when referring either to the FCMD data specifically or the overall industry trends and scholarship on PMSCs’ transgressions and accountability to remain conversant with the FCMD dataset and industry rhetoric. While the FCMD tracks general misconduct, which can account for a range of undesirable behavior, we derived two measures from the misconduct data: the first is a contract or criminal violations—any activities which were either explicitly illegal or contrary to contractual requirements. These, we refer to as violations throughout the paper, are the first dependent variable in this study.

We define military contracting as a contract to perform any action which was historically or is contemporarily performed by the military, and which could be and would have to be performed by the military if a contractor was not forthcoming. Our analysis compares the differences in violations and legal repercussions between military contracting and non-military contracting, and later between cases where the Peace, War, and Social Conflict Laboratory at Texas Tech University (PWSCL-TTU) identified a contractor as a PMSC or not.

Concentrating on the level of violation is used as a proxy to determine accountability. It is important to note that the FCMD often considers an event as being a single case of misconduct even when there are multiple constituent acts of misconduct detailed in the description of the event; for instance, if a whistleblower reports inappropriate inventory practices, harassment, retaliation, and criminal conduct, the FCMD would aggregate those constituent acts of misconduct under a single instance. Because our analysis attempted to identify and quantify individual violations detailed in the primary sources, our methodology focused on enumerating the constituent violations. More numerous violations mean a company is less accountable.

The FCMD also reported the disposition of complaints of misconduct which we further categorized based on the severity of legal repercussion, combining substantively similar dispositions (such as settlements, restitutions, and other pre-prosecution agreements). This is our second dependent variable. A legal repercussion is a consequence imposed on the contractor as a result of its misconduct; it can take the form of civil proceedings such as suits related to monetary damages resulting from control deficiencies, criminal proceedings for violations such as racketeering or arms smuggling, or administrative actions such as debarring a contractor from being awarded more contracts to name only a few. In this way, civil, criminal, and administrative repercussions are all captured under the umbrella of “legal repercussion.” The level of legal repercussion is used as a proxy to measure impunity. In this case, lower levels of legal repercussions (i.e., investigative findings, settlements or non-prosecution agreements, or fines) indicate impunity while higher level repercussions (i.e., guilt findings, monetary sanctions, or disbarment) indicate little or no impunity.

The contribution of this study is twofold. First, it addresses a gap in our understanding of PMSCs’ misconduct. Volumes of academic articles and books underscore the problems and challenges of PMSC regulation, emphasizing the lawless nature of the industry and its impunity issues (Blyth 2007; Leander 2010; Liu 2015; Rosemann 2005), yet this important discussion remains for the most part on the theoretical, conceptual, and general plane, with very limited empirical research that explores the PMSC relations between PMSCs and violations. This is due to significant difficulties regarding data collection on this industry, and, particularly, on PMSC criminal and contract violations. Our study tackles this gap, empirically examining the relations between PMSCs and contract or criminal violations as reported in the FCMD. Second, it employs the PMSC contracting case study to better understand the Principal Agent Problem, a problem that illustrates the challenges related to outsourcing and in this specific case of monitoring PMSCs. We suggest that the setting of the relations, in this case, conflict areas, is a determining factor in the ability of the principal to monitor the agent and of the agent to deceive the principal. A setting that challenges the principal’s capacity to monitor the relations and at the same time cultivates opportunities for white-collar crime (Benson and Simpson 2009) is in fact increasing the cost of monitoring and can exacerbate the principal-agent problem.

The Rise and Rise of the Outsourcing of War

In August 2009, the number of US troops in Iraq and Afghanistan was estimated as high as 200,000. At the end of 2020 in November, the number of private contractors in the field exceeded this number (Schwartz 2010). This was not a one-time event, part of the surge (Biddle et al., 2012, Marsh 2012). It was part of an organized policy of increasing the reliance on contractors in the field at the expense of the armed forces (Avant 2013a, b; Stanley 2015). The trend became very clear in the following years as contractors significantly outnumbered soldiers. Privatization of war continues in full force in Afghanistan with a ratio of 1 soldier to 2.3 contractors, more than double, in 2016 (Peters et al., 2017).

While all of the contractors mentioned in government and the Department of Defense (DOD) reports are deployed in the field, only a handful are employed in physical security missions. Private military and security contractors provide a host of services traditionally performed by the military in conflict areas. Those services include communication, logistics, translation, maintenance, intelligence, infrastructure, and training as well as numerous other roles and functions (Schwartz 2011). Those services are not negligible to the war effort and often involve severe security threats (Swed et al. 2018). Truckers delivering fuel and munitions to frontline outposts in Afghanistan face the same threats as the soldiers occupying the outposts or traveling the road regardless of the trucker’s title as a contractor and his functions in the logistical chain. Modern warfare includes a complex array of participants, technologies, and specializations that require highly synchronized effort to support the combat units—those in the front lines engaging with the enemy (Moore 2017). Today, a great portion of this array is outsourced.

Some issues with the outsourcing of war and security functions in Iraq and Afghanistan were noticed by the media and general public when Blackwater, a PMSC providing security services in Iraq, was involved in a shooting incident in Nisour Square in 2007. During a security escort mission of the US embassy convoy in Bagdad, Blackwater operatives shot at Iraqi civilians, killing 17 and injuring 20. This incident triggered scholarly research on PMSCs’ accountability and impunity (Liu 2015; Whitten 2012, Thurnher 2008), and the regulation of their operations (Chen 2009; Hurst 2007; Cockayne 2008). Percy (2007) states that dealing with mercenary forces rests mostly on strong norms rather than on strong legislation. Consequently, when the norms are changed, a legal cleavage emerges which allows mercenaries/private military security contractors to operate outside of legal boundaries with no risk of real repercussions. This lack of coverage means that effectively the PMSCs are in fact operating outside of the existing legal or administrative framework, underlining a principal-agent problem where the state (principal) cannot effectively control the outcome of working with PMSCs (agent) (McCoy 2010). These gaps became apparent when PMSCs began to be sued in the mid-to-late 2000s, illustrating the complexity of first legally defining those actors (Scoville 2005; Cameron 2006), and assessing which laws they operate under (Finkelman 2008, Lehnardt 2008, Lindemann 2007, Staino 2010, Williams 2010, White and MacLeod 2008). The legal gap opened the door to a host of other concerns, with questions about PMSCs’ accountability and regulation assuming the forefront of scholarly investigation of the industry. Comparing the responsibility of the state versus those of contractors, Hoppe (2008) identified a regulatory gap. Hoppe noted that, unlike the state, all things being equal, PMSCs were not held accountable in multiple cases. These regulatory and legal gaps brought scholars to recognize that the existing accountability system for PMSCs is not fit for purpose (Hedahl 2012). Leander (2010) describes this state of affairs as paradoxical impunity, where PMSCs have expansive authority yet almost no legal accountability or repercussions.

Concerns over the impunity and weak regulation have been frequently raised due to PMSCs’ involvement in human rights violations. Notable examples are the Nisour Square shooting (Liu 2015; Snukal and Gilbert 2015; Tiefer 2009) or the torture at Abu Ghraib (Bina 2004; Schooner 2005; Carney 2005). Additionally, criminal behavior of PMSCs is also an issue, encompassing a variety of violations, including sexual assault (Vrdoljak 2010), child abuse (Bakker and Greijer 2010), worker abuse (Higate 2012), sex trafficking (Maffai 2008), smuggling (Rothe and Ross 2010), illegal arms sales (Makki et al. 2001), and fraud (Grasso 2010).

Incidents such as the Nisour Square shooting carried with them much negative media coverage of the US operations (Associated Press 2007,von Zielbauer 2007), placing significant pressure on the industry, by provoking demands for regulations from states or international organizations and professional associations. The industry’s solution was self-regulation. PMSCs joined professional associations with codes of conduct such as the International Stability Operations Association (ISOA),Footnote 1 the American Society for Industrial Security International Association (ASIS), the International Code of Conduct for Private Security Service Providers’ Association (ICoCA), and the British Association of Private Security Companies (BAPSC). PMSCs also took part in the writing and signing of non-binding and non-legal documents, such as the Montreux Document or the International Code of Conduct for Private Security Service Providers (ICoC). Without an effective alternative, this solution became the main accountability answer to the regulatory gap.

While the industry seems convinced self-regulation is adequate, some studies raise concerns over the efficacy of this mechanism. De Nevers (2009) asserts that the PMSC industry does not exhibit the capacity to adopt and implement effective self-regulation on its own. Looking at the IPOAs and BAPSC’s record, De Nevers (2010) argues that PMSCs’ ideas of self-regulation are not working. In both cases, the PMSCs’ self-regulatory mechanisms, designed by the organizations, were not able to monitor or sanction member companies’ behavior, rendering the self-regulation mechanism irrelevant. Krahmann (2016) identifies another hindrance in this system of self-regulation, challenging the validity of self-regulation as a control mechanism. The working assumption behind self-regulation is that the consumers of PMSC services will facilitate the development and enforcement of professional standards. This process arises as consumers shift their contracts to companies that have signed up to self-regulation codes of conduct. Krahmann underscores the obstacles behind this assumption, focusing on the limitations of consumer interest, restrictions set by countries on operations, and, finally, the influence PMSCs exercise on the choices of potential and existing clients. To put it simply, the assumption of market enforcement based on companies’ self-regulation is not as simple and straightforward as some may argue.

Focusing on the Contract

With self-regulation, the industry claims that it achieves a level of accountability but does not provide any evidence that their practices are effective or successful. On the other hand, scholars and experts disagree with industry claims about the efficacy of self-regulation, arguing that there are structural limitations and problems with the self-regulation assertion while providing minimal proof. Even knowing that regulation is essential, the lack of support for claims on either side of the argument means we cannot be sure whether the current self-regulation scheme offers an adequate solution or not, and we cannot determine how proponents for increased regulation should focus their efforts.

To address this conundrum, we suggest forwarding the general discussion and conducting an empirical investigation to determine if the industry needs further regulations, or if self-regulation is on par with general contracting standards and accountability. Regarding the questions of accountability (H1 and H3) and impunity (H2 and H4), we explore whether PMSCs are more likely to engage in misconduct than any other contractor working under private or other government contracts and if they are penalized less than their counterparts. This is not merely a policy debate but also a theoretical question. The principal-agent problem presents four stages of relationship: screening, negotiating, monitoring, and sanctioning. Focusing on accountability and impunity corresponds with the last two. Monitoring is what allows effective accountability and sanctioning is what coerces the agent to comply (Drutschmann 2007; Mahoney 2017).

Building on Dickinson’s (2003) suggestion of using the contract as the tool to regulate PMSCs, we use the contract as the organizing concept of our analysis. Contracts are legal agreements that determine the presiding principles of relations between employers and PMSCs. They establish the boundaries of engagement, the compensation measurements, and the evaluation parameters. They are an essential part of the monitoring aspect of the principal-agent problem and determine the nature of the sanctions warrantied in case of a breach of contract. The contract clauses are tailored to fit the specific tasking and are influenced by industry standards and best practices. Dickerson explains that contracts do more than this; they are the vehicle that carries the values and norms of public international law into the private sector if done properly. Contracts are powerful policy instruments that can increase contractors’ accountability and efficacy (Romzek and Johnston 2005). Dickerson is not alone. In the field of accounting, contracting can mitigate the principal-agent problem (Miller 2005). A contract design can address issues of interest congruity between the principal and the agent (Qin et al. 2019), yet if done without proper oversight practices or with flaws in contract requirements, the outcome can lead to misconduct and waste (Berrios 2006). In the field of PMSCs, the significance of good contracting was illustrated in Tkach (2019) study, on the association between the contract structure of PMSCs in Iraq and the levels of violence. Tkach shows that PMSCs’ contracts that did not have performance incentives were associated with a higher likelihood of violence. Continuing this discussion, we hypothesize the following:

  • H1: Companies operating under PMSC contracts commit more violations than those operating under non-PMSC contracts.

  • H2: Companies operating under PMSC contracts suffer fewer legal repercussions in response to misconduct than those operating under non-PMSC contracts.

  • H3: PMSCs commit more violations than non-PMSCs

  • H4: PMSCs suffer fewer legal repercussions in response to misconduct than non-PMSCs

Not all PMSC contracts are executed by PMSCs. The wars in Iraq and Afghanistan brought with them a booming market for government contracts into which numerous companies quickly expanded (Beelman et al. 2012). Among those companies, we can see companies that are clearly not PMSCs, such as AT&T, FedEx, and Dell, which translate their services to support the military operation. The US soldiers used AT&T plans on their cell phones, received FedEx to their bases across Iraq, and had Dell computers in their offices. Companies like Chemonics, a company specialized in development, the Corrections Corporation of America, which specializes in correction facilities, and the Chugach Alaska Corporation, a for-profit cooperation created by Alaskan natives that focusses on investment in minority and economically disadvantaged business enterprises, all show up in our sample. These companies are not PMSCs, yet they pursued and received PMSC contracts.

Moreover, the market of war is limited and is tied to the fluctuation of foreign policy decisions and geopolitical reality. Therefore, companies which want more opportunity to sell their services must look to other more stable markets. As domestic service providers find customers in the war market and martial service providers diversify into the domestic market the distinctions between the two fade, making identifying military contractors a difficult if not untenable proposition. The expansion has been recognized in the literature as some of the PMSCs became very versatile (Prem 2018). For example, KBR, one of the major military service providers is a diverse company that also provides construction and engineering services in the USA and across the globe. In our sample, PMSCs such as KBR, DynCorp International, and CACI International are commissioned to provide services that are not military, or conflict-related. In other words, not all government contracts with PMSCs are for conflict-related purposes and many non-PMSC companies can have a PMSC contract for particular endeavors.

Because of that, we assert that the type of contract (PMSC or non-PMSC) and not the company’s identity (PMSCs or non-PMSC) is the determining influence on the behavior of a company and its likelihood of engaging in violation, contractual, or criminal. Furthermore, we assert that the type of contract, rather than the contractor, predicts the severity of legal repercussions contractors face for any transgression. Consequently, we expect H1 and H2, which follow Dickinson’s model, to be accurate. Namely, that we will see an association between higher levels of violation by contractors working under PMSC contracts. Furthermore, we will see an association between low-level legal repercussions while operating under a PMSC contract. Conversely, we would expect H3 and H4 to be rejected. This means that companies identified by our research as PMSCs are not more likely to commit violations nor are they subject to lower levels of legal repercussions.

We contend that violations proliferate in PMSC contracting environment due to organizational limitations that are functions of overseeing companies operating in legal grey-areas, in distant lands, often under the cloak of secrecy. The setting of the principal-agent relationship can drastically affect the cost for the principal. Our assertions correspond with existing arguments about the organizational realities and limitations of auditing a company operating in a conflict area, where security risks hinder effective inspection (Berrios 2006; George-Nichol 2019, Sopko 2016). The obstacles in such situations are persistent and recognized as a well-known hindrance to mission oversight in distant or hard-to-reach locations. Insecurity hinders operational effectiveness at almost every level, and accountability efforts are often disproportionately affected due to their perceived auxiliary relationship to mission execution. We suggest that this environment of weak oversight couples with ample opportunities to overstep the legal boundaries creating opportunities for crime. Drawing from the Routine Activity Theory’s (Cohen and Felson 1979) emphasis on the environment and crime, we identify conflict areas as environments with “the absence of a capable guardian against crime.” The serious limitations on oversight in isolated settings plagued with corruption and black-market economies constitute an opportunity structure for white-collar crime (Benson and Simpson 2009).

To illustrate this oversight gap, we can look at the challenges of auditing operations and contracts in Afghanistan. Since 2002, over $126.30 billion has been appropriated for Afghanistan relief and reconstruction missions. These funds are used to build the Afghan National Security Forces, promote good governance, conduct development assistance, and engage in counter-narcotics and anti-corruption efforts (SIGAR website). Afghanistan is a large country (647,500 sq. km) with challenging terrain and a limited road system that does not cover most of the country well. Even without conflict and strips of land controlled by hostile forces, it is a hard country to run oversight missions. The large investment, along with the challenges presented, paved the way for the installment of the Special Inspector General for Afghanistan Reconstruction (SIGAR). In a 2016 report, SIGAR enumerates its achievements, such as the recovery of an estimated $36 million and identification of some $950 million that could be put to better use, for a combined impact approaching $1 billion (Sopko 2016). Despite these impressive successes, SIGAR recognizes the hurdles to oversight it must contend with; in the report it acknowledges the challenges in auditing and inspecting areas they cannot reach. With a team of only thirty employees based at the US Embassy Kabul and two more at Bagram Airfield, it is difficult to monitor every contract in a large and complicated arena such as Afghanistan. In an interview posted on the SIGAR website, the Deputy Assistant Inspector General for Audits and Inspections describes how security challenges inhibit effective oversight. “….even in the time, back in 2012, when we could travel to the maximum extent that we could ever travel, we would never have been able to hit all the provinces… nor could we stay on-venue. Even if we do go out to a site, typically, we can only be on the venue for about an hour or two and then we can only do it once a week.” (George-Nichol 2019).

Similarly, the conflict setting is also prone to corruption and a flourishing black-market economy. In various studies, Reno underscores the endemic nature of the illicit economy in war settings (Reno 1996, 2000, 2009). Examining conflicts across Africa, Reno shows that the black-market economy is an integral part of the political and economic system during the conflict. Moreover, both violent non-state actors (insurgents, rebels, and terrorists) and the local government take part in it. Building on Reno’s work, James (2012) stresses the importance of organized crime in stabilizing those settings. Returning to the Afghanistan example, we can identify similar processes of a flourishing black-market economy and high levels of corruption. Afghanistan is the leading opium producer in the world, accounting for over 90% of world trade since 2001. Moreover, it suffers from endemic corruption, which has affected elections (Callen and Long 2015), aid, and development efforts (Marquette 2011), and involved networks of government officials and insurgents (Bojicic-Dzelilovic et al. 2015).

This example illustrates the very real and endemic obstacles to effective accountability operations for contracting in conflict settings, explaining the rationale behind our argument vis-à-vis criminal opportunities and limited oversight. Conflict settings where contractors are commissioned to operate suffer from weak oversight mechanisms. As such, they pose a challenge for principal in monitoring the agent. The guardian against crime is incapable or weak. Furthermore, the setting itself is full of opportunities for crime, with endemic corruption and a flourishing illicit economy. This environment means that companies, regardless of their specialization (PMSC or non-PMSC), would be more likely to commit violations and less likely to suffer from repercussions in those settings.

Data

The ever-shifting and expanding nature of the industry makes it a real challenge to define what PMSCs are (Prem 2018). While we typically think about contractors as armed mercenaries, the reality on the ground presents a more nuanced and complicated picture. The broad outsourcing of military functions includes multiple duties and positions that can easily pass as civilian rather than military jobs. Those functions consist of menial jobs, such as cleaning military bases in Iraq, emergency services such as firefighting and emergency medical services, and online or logistical services located at a distance from the battlefield to name just a few. Consequently, in this study, we define a PMSC as “any company which offers an array of services which were historically or are contemporarily performed by the military to support and conduct combat operations, and which could be and would have to be performed by the military if a contractor was not forthcoming.” This definition captures both combat and combat support functions. It also differentiates the occasional service provider from the professional contractor, focusing on services. Working in collaboration with the PWSCL-TTU, we compiled a list of over 1515 known PMSCs around the world.

Violation and legal repercussion data were collected from the Federal Contractor Misconduct Database (FCMD), which is maintained by The Project on Government Oversight (POGO). The dataset is a compilation of misconduct and alleged misconduct committed by federal government contractors from 1995 to the present. At the time of analysis (10 January 2018), it tracked 2380 instances of misconduct across 224 companies. We analyzed each “instance” of misconduct tracked by the FCMD, going through the primary source documentation in PDF format uploaded to the website to identify relevant information for our analysis. Building on these primary sources, we constructed a new dataset of 2218 instances of differentiated PMSCs and non-PMSCs contract misconduct. Our new dataset accounted for and removed duplicated instances discovered in the FCMD dataset.Footnote 2

To define a private military security contract, we used similar logic to our PMSC definition, focusing on the services contracted. We considered a PMSC contract “a contract to perform any service which was historically or is contemporarily performed by the military, and which could be and would have to be performed by the military if a contractor was not forthcoming.” The focus is on whether a company accepts a contract which requires the performance of PMSC functions regardless of other factors including how they identify themselves. It is important to clarify that companies that identify as PMSCs may have non-PMSC contracts with the government and “non-PMSC” companies may have PMSC contracts. Out of the 2218 instances of misconduct, we identified 94 instances which were related to PMSC contracts (4.2%). Of those 94 instances, 62 (65.9%) were related to PMSCs identified by PWSCL-TTU.

We use two dependent variables: (1) the number of violations and (2) the severity of legal repercussions, both per instance per company. Those two variables echo the scholarly debate of whether PMSCs are indeed actors that are prone to greater violations (Chapman 2010, Rothe and Ross 2010) and if PMSCs are less accountable than other violators (Leander 2010; Liu 2015). The violation variable is an ordinal variable that captures the number of violations committed by the company contracted as written in the primary sources (Table 1). The primary sources do not provide an accurate number of violations across all documents. Moreover, in some cases, the number of violations is more than a million. Therefore, we created an ordinal measure of fourteen levels of violations from one, through hundreds, and thousands, to millions. An example for a violation could be the firearms violations committed by Xe Services LLC, which was reported by the State Department. The settlement report mentions 288 alleged violations of the Arms Export Control Act and the International Traffic in Arms Regulations between 2003 and 2009. Violations are not limited to illicit arms sales. They include a variety of activities, including fraud, wrongful death, false claims, payment disputes with subcontractors, sexual assaults, and bribery.

Table 1 Coding
Full size table

The disposition stage variable is an ordinal variable that acts as a proxy for the severity of legal repercussions for violations (Table 1). It captures five stages of the administrative or legal process with each stage representing a higher-level of legal repercussions for the company: (1) the conclusion of an investigation with investigative findings regarding the alleged misconduct, (2) having a pending legal case on misconduct at the appropriate court, (3) the company has settled with the government or is following an administrative, deferred prosecution, or non-prosecution agreement, (4) the company plead guilty or there is a judgment against the company in court, and (5) the company is fined, suspended, or debarred. The categorization is based on the wording within the FCMD reports. Our ordination process accounted for substantially similar dispositions which could be considered to fall into one of five steps in the legal process: step 1 encompasses the identification of an issue; step 2 shows that the issue was severe enough and the evidence clear enough to warrant pursuing legal action; step 3 indicates that the company acknowledges the issue, and its culpability, and acquiesces to punitive action; step 4 represents a judge or arbiter’s decision that the company is culpable in the case that the parties could not agree on culpability or sanctions in step 3; and step 5 indicates that a judge or arbiter ordered sanctions as a result of culpability. In this way, an investigative finding or pending case, no matter how damning the evidence or how egregious the transgression, is less severe dispositions than a much less impactful transgression further along in the process by merit of the former being merely an allegation, and the latter being an agreed upon or adjudicated transgression. An example for a disposition at step 4 is a Halliburton employee guilty plea in a kickback scheme related to the supply of semi-tractors and trailers in and Iraq and Kuwait for the American military in 2003. Other dispositions include settlement cases, restitution, investigative findings, fines, debarment, and others (see Table 1 for a detailed review).

We also control for the type of services provided in the contract, assuming that it may affect a company’s behavior (Mahoney 2017). The type of contract is represented by three dummy variables: the first is operations contracts, which refers to contracts that affect the operational capacity of the military (i.e., security, training, translation, intelligence, and logistics); the second is infrastructure contracts, which address contracts for construction, facilities maintenance, and the production and disposal of materials; and the third variable for the type of contract is professional contracts, which focus on professional services such as IT, medical, financial, and administrative.

We also address the number of control deficiencies per instance of misconduct as an ordinal variable (0–3). Control deficiencies refer to a design or operation that does not allow management or employees, in the normal course of performing their assigned functions, to prevent or detect noncompliance on a timely basis. Simply put, this variable refers to the way the company plans (or does not plan) to prevent violations. Similar to violations, control deficiencies are considered misconduct; nonetheless, violations are more acute transgressions, often characterized by intent and culpability, whereas control deficiencies may be the result of bad planning or incompetent management. Additionally, control deficiencies are more difficult to detect, often being identified either as a result of investigations into violations or routine audits. The assumption here is that the higher the level of the control deficiencies, the higher the likelihood for a violation to occur and the lower the likelihood that a contractor can detect or address those violations.

Finally, we control for the mechanism that led to the exposure of the violation, meaning how willing or able a company was to be forthcoming in revealing misconduct by classifying which party identified the misconduct. For this, we use three different mechanisms all represented as dummy variables: (1) unknown; (2) internal exposure, which includes internal accountability processes or employee whistle blowers; and (3) external exposure, which includes government or external accountability processes or events which were immediately apparent based on the nature of the event (explosions, bridge collapses, IT system failures, etc.).

Table 2 presents the descriptive statistics of the variables. It shows that there are 89 PMSC cases and 322 PMSC companies, which present 5% and 18% of the sample, respectively. The table also shows that the violations are clustered at the lower end of the spectrum, with codes 2 and 4 of the violations accounting for about 70% of the sample. The disposition variable is grouped around coding 3, which captures the cases of restitution, settlement, administrative agreement, and deferred prosecution agreement. A review of the distribution of types of services contracted shows that infrastructure and operations related services were more prominent in the sample than professional ones. The control deficiency variable is also clustered in the lower codes, with about 76% of the sample showing no control deficiencies and codes 1 and 2 accounting for 20% of the sample. The misconduct identifier shows that most violations were identified by externally and not from internal processes. Lastly, it shows that most violations are non-contractual violation, such as selling arms, which a contract would not anticipate or explicitly prohibit, but which would be prohibited by statute.

Table 2 Descriptive statistics
Full size table

Method

To explore whether PMSCs are in fact chronic violators that escape the legal or administrative consequences of their actions or merely a different type of government contractors, we conduct an Ordered Logit regression in a two-stage four-model analyses.Footnote 3 Ordered Logit regression best fits our models given that the dependent variables are ordinal. The first stage of the analysis explores the relations between contract and non-contract violations, as well as legal repercussions for those violations, for companies executing PMSC contracts versus contractors PSCL-TTU identified as PMSCs. In the first model, we examine the association between having a PMSC contract and the number of violations conducted. The second model examines the association between identified PMSCs and the level of violations conducted. The third model examines the association between having a PMSC contract and the level of legal implications suffered for violations. The fourth model explores the association between identified PMSCs and the level of legal implications suffered for violations. To ease the interpretation of the Ordered Logit results, we present margin effects of models I and III. The second stage of the analysis adds a variable to differentiate between contract violations (i.e., not meeting with the contract expectations) and criminal violations (e.g., illicit arms sell, trafficking, violence against the local population) (1 = contract-related violation; 0 = criminal-related violation). The four models’ structure at the second stage is parallel to the first stage.

Analysis

This paper’s two main arguments are that companies with a PMSC contract would be more likely to commit violations than other companies and to suffer a lower level of legal repercussion. Similarly, companies identified as PMSCs would be more likely to commit violations than other companies and to suffer a lower level of legal repercussions. Table 3 presents the level of violations and the legal repercussions of the violations associated with the PMSC contract and the PMSCs. For our models, we use infrastructure contracts as the reference for the type of contract and the unknown identifier of the misconduct as the reference for the misconduct identifier.

Table 3 Association between PMSC contract and PMSCs with contract and criminal violations and legal repercussions, between 2000 and 2016
Full size table

Table 3 illustrates that companies with a PMSC contract have a significantly higher level of violations than companies without such a contract (p < 0.01). That is the ordered log-odds estimate for violation is 0.66 higher for companies with a PMSC contract than other violators in the sample when all other variables are held constant. In other words, companies with a PMSC contract, whether they are PMSC or not, are greater violators than companies without this type of contract. The model also shows that operations and professional types of contracts are associated with higher levels of violations than are infrastructure types of contracts (p < 0.05). In comparison to the unknown violation identifier, external misconduct identifier is associated with higher rates of violations (p < 0.01). While having a PMSC contract is significantly associated with violations, model II finds no significant association between companies identified as PMSCs and higher levels of violation. At the same time, other associations observed between contract type and violation identifier seen in the model I are replicated in model II.

We next examine the association between legal implications of violations and PMSC contracts (model III, Table 3). We find that having a PMSC contract is significantly associated with the level of legal implications for the company’s violations (p < 0.01). Namely, the ordered log-odds estimate for legal implications for a violation is 0.64 lower for companies with a PMSC contract than other violators in the sample when all other variables are held constant. Companies with a PMSC contract experience weaker legal repercussions than their counterparts. Looking at the type of contract in model III shows that professional and operations types of contracts are associated with lower rates of legal implications than are infrastructure contracts (p < 0.001). At the same time, having an internal misconduct identifier is associated with lower rates of legal implications than an unknown one (p < 0.001). Similar to the results in model II, there is no significant association between companies identified as PMSCs and higher or lower levels of legal repercussion (Table 3, model IV). Other variables follow the trends presented in model III.

Figure 1 illustrates the results of models I and III in Table 3. Companies with a PMSC contract show a higher probability to have dozens of violations (outcomes 3 and 4) and a lower probability to have one or two violations (outcomes 1 and 2) than companies without a PMSC contract (Panel A). Higher outcomes, representing thousands and millions of violations, show the same trend with a higher probability for companies with a PMSC contract. Panel B shows that companies with a PMSC contract have lower probabilities to advance in the legal process than companies without one, and experience more substantial legal consequences. Namely, companies with PMSC contracts are less likely to have a judgment against the defendant (outcome 4), reach the settlement or restitution phase (outcome 3), or have a pending case against them (outcome 2). Similarly, they are more likely to be under an investigative finding (outcome 1), the first stage in the legal process.

Fig. 1
figure 1

Predicted probabilities and marginal effects of models I and III, Table 3

Full size image

Not all instances in the sample are contract-related. About 42% of the identified instances of misconduct in the examined sample are contract-related, meaning violations of contractual commitments, and about 57% are instead criminal (Table 2). The second group includes drug smuggling, arms trafficking, and other crimes. In the second part of our analysis, we address this difference in types of violations. In Table 4, we add to the models the type of violation as a dummy variable (contract-related incident vs criminal-related incident). Model I shows that while controlling for contract versus criminal violations, companies with a PMSC contract have a marginally significant association with higher levels of violations (p < 0. 1). This means that in this model, the ordered log-odds estimate for violation is 0.37 higher for PMSC contracts than other companies. In comparison to infrastructure contract, only operations contracts are marginally significantly associated with violation severity (p < 0.1), while external identifier is associated with higher rates of violations (p < 0.01). Lastly, the type of violation variable shows that contract-related violations are more common than criminal violations. Model II shows that there is no significant association between companies that were identified as PMSCs and the level of misconduct. Model III shows that companies with a PMSC contract have a significantly lower level of legal repercussions (p < 0. 05). That is the ordered log-odds estimate for legal repercussions is 0.48 lower for companies with a PMSC contract than other companies. Compared to infrastructure contracts, a professional type of contract is associated with fewer legal repercussions. For identifiers, both internal misconduct identifiers (p < 0.001) and external identifiers (p < 0.1) are associated with significantly lower levels of repercussion than an unknown identifier. In this model, criminal violations are associated with more severe legal repercussions than contract-related ones. Model IV shows that there is no association between whether or not a company is identified as a PMSC and the level of legal repercussion. The other results in this model are similar to model III.

Table 4 Association between PMSC contract and PMSCs with contract violations and legal repercussions, between 2000 and 2016
Full size table

To summarize the results of the analyses, the model I on Tables 3 and 4 suggests that companies with a PMSC contract have higher levels of violations than those without such a contract. Similarly, model II on the two tables indicates that there is no association between being identified as a PMSC and the level of violations. Focusing on the legal implications of the violations, model III on Tables 3 and 4 illustrates that companies with a PMSC contract have lower levels of legal implications than other violators in the sample. Lastly, model IV on the two tables shows that whether a contractor is an identified PMSCs had no significant association with this measure.

Discussion

This study tackles the scholarly and professional debate on PMSCs accountability and impunity (De Nevers 2010; Leander 2010; Liu 2015; Krahmann 2016). Considering the centrality of PMSCs in contemporary security and military policies (McFate 2017; Stanley 2015), clarifying these actors’ state of accountability and whether the impunity discussion is merited or not in this context are cardinal questions. The importance of this debate increases given the transparency chasm surrounding this industry, where we know very little about the actors involved and their actions (Swed and Crosbie 2017). This paper’s findings disclose an addendum to this discussion, offering empirically based answers to a critical debate. Building on data from the Federal Contractor Misconduct Database (FCMD), we compare the level of accountability and impunity of companies with PMSC contract vs companies identified as PMSCs by PWSCL-TTU. Our findings inform three points for discussion: (1) the lack of association between PMSC and accountability and impunity issues, (2) the association between the PMSC contract and issues of accountability and impunity, and (3) the broader implications of oversight in conflict areas.

First, on the policy level, the results indicate that PMSCs are not associated with higher levels of violations and lower levels of legal repercussion. It is a rejection of hypotheses 2 and 4. In other words, compared to other violators in the sample, PMSCs are not better or worse. According to our proxies and findings, PMSCs do not have a problem of accountability compared to other non-military contractors. Furthermore, we could not find a significant association between PMSCs and our proxy measure for impunity. According to our findings, the legal repercussions PMSCs pay for their transgressions are not different from other violators in the sample. It is important to say that those findings are not a vindication for the PMSC industry and can imply that there are generally poor standards in accountability and impunity for all government contractors. Those findings offer support for the industry’s argument that existing regulations are working and that there is not unique accountability or impunity issue with the sector, though this support is contingent on whether one only counts companies which are predominately PMSCs as “the industry.”

Second, the findings denote that companies executing a PMSC contract are associated with higher levels of violations and lower levels of legal repercussion, supporting hypotheses 1 and 3 and contributing to our theoretical understanding of the principal-agent problem. On the policy level, this means that conversant with Dickinson’s suggestion to focus on the contract (Dickinson 2003), we see that having a PMSC contract can lead to gaps in accountability and impunity. Our research underscores that the main concern regarding PMSC activities should be governmental oversight and enforcement rather the focusing on identified PMSCs as high-risk partners. On the theory plain, those findings suggest that the location of the activity is fundamental in the understanding of the principal-agent problem. It is established that monitoring deficiencies incur costs on the principal. Those deficiencies are attributed to a variety of factors, mostly related to logistical and knowledge challenges, yet the case at hand underscores how the setting matters. In fact, it shows that companies that regularly exhibit standard compliance turn into violators when the setting changes. It is true that conflict areas represent the extreme end of the spectrum of settings, yet there is symmetry here given that the state is the most powerful principal, equal to none. It means that there are two general tenets that can be drawn from this case. First, the setting can pose real challenges to the principal to monitor and allow the agent extended freedom. Second, the setting’s challenges are relative to the principal’s capacities, a state will find conflict areas challenging, but a small business will discover the out of state outsourcing is beyond their ability to monitor effectively.

Third, the findings draw attention to the link between the oversight challenges for contracts that take place within conflict areas and white-collar crimes committed by the contractors. Per Cohen and Felson (1979) Routine Activity Theory, we show that an environment of weak oversight and copious opportunities to break or bend the legal boundaries creates opportunities for crime (Benson and Simpson 2009). Those opportunities, along with weak oversight, are the reason why we see higher deviance across companies with a PMSC contract. It also offers context for a broader discussion that focuses on the unintended consequences identified in aid and “hearts and mind” operations in conflict areas (Narang 2015; Wood and Sullivan 2015). Ineffective oversight can offer an explanation for the spillover of aid resources into the hand of insurgents and rebels. This, in turn, can lead to the counterproductive outcomes of prolonging conflict and increasing violence.

For a better interpretation of the results, policy-decisions, and future research, it is important to discuss the limitations of our study. First, our sample is comprised of companies that committed misconduct and alleged misconduct per the FCMD’s criteria. This means that our reference point misses the companies that did not commit misconduct. Additionally, our sample does not capture the entire universe of contract violators; we construct our dataset based on the data provided by the FCMD which stipulates that it does not claim to identify every instance of actual or alleged misconduct. Moreover, the FCMD stresses that the total amount of misconduct instances is probably understated due to the fact that the terms of settlements are often undisclosed.

Conclusion

The rise and rise of PMSCs’ role in security policies and war-making stirred a debate on the industry’s accountability. As private actors that are not regulated well, the fear of deviant behavior with impunity was recurrently raised by experts and academics (Chapman 2010, Dickinson 2011, Jordan 2009, Leander 2010, White and MacLeod 2008). The gravity of this debate became evident with several high-profile incidents that underscored PMSCs’ human rights violations and misconduct (Bina 2004; Carney 2005; Liu 2015; Schooner 2005; Snukal and Gilbert 2015; Tiefer 2009). The industry’s after-effect solution of self-regulation reform came under scrutiny by scholars that assert this solution is flawed and ineffective (De Nevers 2009; 2010; Krahmann 2016). Both PMSCs and scholars presented compelling arguments to back or attack the efficacy of the industry self-regulation solution yet offered minimal evidence to support their claims leaving this debate in a gridlock that could not effectively inform policies. In this paper, we address this gridlock, offering an empirical assessment of the association between PMSCs and misconduct and impunity. Examining an original dataset based on the FCMD files, we show that in contradiction of common assumption, known PMSCs are not more prone to either commit contractual or criminal violations or to enjoy impunity in comparison to all other contractors. We also indicate that companies with a PMSC contract, regardless of how they define themselves, are indeed associated with higher rates of violations and impunity in comparison to companies performing non-PMSC contracts, including known PMSCs.

These findings establish that the main challenge in regulating the industry is the establishment of an effective oversight mechanism in areas of conflict. It means that the challenge we have regarding PMSC accountability is part of a broader question of general oversight in areas with weak oversight and ample opportunities for crime. We believe that the outcomes of this study’s analysis are good news for policy makers and accountability advocates alike. Our data illustrates that PMSCs are not rogue actors that cannot be controlled. It also plots the path for two policy solutions. One is the development of a policy focused on smarter contracts and better oversight measures on the ground. Nonetheless, the path towards a stronger and more efficient oversight in conflict areas is not clear or easy to accomplish. The second is more direct, proposing that without an effective oversight mechanism, the government should reconsider outsourcing in conflict areas. How can we assure oversight in areas with limited access and high security risks? How can we address an environment that offers opportunities for crime? Which oversight policies and mechanisms work, and which fail? These answers will require future research that will plot the path forward and identify what is working and what is not.