Award sizes and fiscal capacity to comply

Beginning with state fiscal capacity to comply, we examine the considerable variation in the size of monetary awards. In our dataset, pecuniary awards range between $555 and $28 million USD, while non-pecuniary awards range from $2,000 to $14.9 million.Footnote ¹³ While the range for pecuniary damages is much higher, the mean pecuniary damage award is $264,040, compared to the mean non-pecuniary damage award of $757,900. Because awards at the IACtHR are compensatory, and not punitive, the award totals represent the Court’s best calculation of the harm done to the victim and next of kin.

The size of the total monetary award in a case might create significant resource constraints on the part of the state. Evidence indicates that international courts consider the role of domestic political and economic circumstances in limiting or facilitating state ability and willingness to comply with reparations orders (Altwick-Hámori, Altwicker, and Peters Reference Altwicker-Hámori, Altwicker and Peters2016; Sandholtz and Padilla Reference Sandholtz and Padilla2020). With respect to state capacity to pay monetary damages, Crawford (Reference Crawford2013, 481) highlights that concerns over the inability of states to pay monetary damages was an issue discussed by the drafters of the International Law Commission’s Articles of Responsibility of States for Internationally Wrongful Acts because of the fear that the amount of the claims would negatively impact the subsistence of certain populations. States have also directly indicated the difficulty they face in making large payments. In the 2006 judgment of Miguel Castro-Castro Prison v. Peru, Peru noted that “if the Court’s standards were to be applied, a compensation of approximately $17,052,000 (USD) would be set between the deceased and the injured, without taking into account those submitted to cruel treatments which would be 322 people” and this would generate damages awarded that “would be a state obligation difficult to handle” (para. 412).

Thus, the value of pecuniary and non-pecuniary damages awarded by the IACtHR may affect time to compliance. We expect that higher awards are more likely to raise fiscal capacity concerns, resulting in lower probabilities of compliance in a given year. More formally, we posit:

However, we recognize that not all states face equal capacity limitations to pay out monetary damages. Although many Latin American states face resource limitations, states have varying levels of capacity to raise additional revenue to pay out monetary damages awarded by the IACtHR. As a result, some Latin American states face fewer budgetary challenges in paying out monetary damages and thereby complying with this form of financial reparation. Although we expect a negative relationship between the value of monetary damages awarded in a judgment and time to compliance with the payment of monetary damages, we expect this relationship to be less negative for states with fewer capacity limitations, as in those with greater fiscal capacity. Thus, we expect:

Number of victims and bureaucratic capacity to comply

In addition to fiscal capacity, we also argue that bureaucratic challenges can hinder compliance with the payment of monetary damages. The number of victims in a single judgment can vary quite substantially, with some judgments involving only a single victim and others involving hundreds. Several IACtHR judgements involve massacres of entire communities and groups of people. For example, in the 2004 Plan de Sanchez Massacre v. Guatemala judgment, members of the military and other state actors perpetrated massive human rights abuses against hundreds of members of the Mayan indigenous community, including killing over 250 people.Footnote ¹⁴ Plan de Sanchez Massacre v. Guatemala was a landmark judgment in that it represented the first time reparations were ordered for the survivors and next of kin in a full scale massacre (Antkowiak Reference Antkowiak2008). The Court adopted a flexible approach in determining who could qualify as a victim, “considering that those unable to prove their identity could also classify as victims if certain requirements were met" (Sandoval Reference Sandoval2017, 2). The state was ordered to pay $5,000 for pecuniary damages and $20,000 for non-pecuniary damages per victim, meaning the state had to pay $7.9 million to provide redress as a result of the Plan de Sanchez judgment (Sandoval Reference Sandoval2017). Following Plan de Sanchez, four more massacre cases were heard by the IACtHR over two years (Antkowiak Reference Antkowiak2008), involving hundreds of victims, both direct victims and their next of kin.

For the state to fully comply, the state must possess the appropriate personal information for all victims in order to make payments, which can require substantial coordination across domestic institutions. Moreover, victims eligible to receive payments awarded by the IACtHR must register with the appropriate domestic body in order to receive compensation, which may be challenging if victims are not aware of the process. The process is further complicated by specific orders of the IACtHR, which can include designating payments through a trust, rather than a bank account because “trusts are institutions designed to increase the real value of assets” (Pasqualucci Reference Pasqualucci1996, 52). Moreover, Pasqualucci (Reference Pasqualucci1996) notes that the IACtHR sometimes even specifies the operational details of the trust. For example, in the 1993 reparations judgment of Aloeboetoe et al. v. Suriname, the Court stipulated that the trust fund be set up in dollars “under the most favorable conditions consistent with banking practice.”Footnote ¹⁵ The IACtHR even recognizes some of the sizeable obstacles faced by victims in claiming reparations, particularly indigenous and tribal peoples (Antkowiak Reference Antkowiak2014). For example, in the Moiwana Village case, the IACtHR provided petitioners with “more lattitude…with respect to acceptable means of proving identity” because “many Maroons do not possess formal identity documents, and were never inscribed in the national registry.”Footnote ¹⁶

The large number of victims in many IACtHR judgments requires utilizing the existing state bureaucratic apparatus to disseminate payments. As the number of victims awarded damages in a judgment grows, the bureaucratic burden placed on the state grows as well. Formally, we expect:

However, some states experience a greater bureaucratic burden than others when it comes to disseminating damages awarded by the IACtHR. Bureaucratic capacity, including capacity that is administrative in nature, contributes to the capacity of the state to “implement logistically political decisions” (Mann Reference Mann1984). Cole (Reference Cole2015, 414) notes that bureaucratic institutions empower states to effectively engage in rights protection and finds that “bureaucratic capacity will give states the administrative and logistical abilities to implement their human rights treaty commitments.”

We argue that the presence of a domestic reparations program (DRP) represents a direct indicator of bureaucratic capacity to comply with IACtHR orders to pay monetary damages. Many states under the IACtHR’s jurisdiction have set up DRPs, which are transitional justice mechanisms designed to provide the bureaucratic infrastructure necessary for providing reparations to victims of human rights abuses. DRPs are “domestic executive or legislative initiatives which identify a group of victims as beneficiaries of reparation and then provide them with redress, usually through similar reparations measures; a summary and fast process; and without the legal costs and high burden and standard of proof required in a judicial setting” (Sandoval Reference Sandoval2017, 2). DRPs exist in many countries in Latin America. For example, Colombia established a DRP through the enactment of the 2011 Victims and Land Restitution Law, providing reparations to individual and collective victims of killing, internal displacement, torture, kidnapping, and other abuses on or after January 1, 1985. Victims can claim reparation by registering with the National Registry of Victims and providing documentation on how they were victimized during the time frame established by the law (Sandoval Reference Sandoval2017, 2).

When the IACtHR has evidence of the effectiveness of state DRPs, the IACtHR can work through this domestic institution to provide redress to victims. For example, the case of Operation Genesis v. Colombia involved the displacement of around 3,500 Afro-descendants living in communities in the northeast of Colombia as a result of an anti-guerilla operation (Operation Genesis). During the operation, one of the members of the displaced communities, Marino Lopez, was killed.Footnote ¹⁷ In total, 372 victims were identified by the IACtHR (Sandoval Reference Sandoval2017, 8). The IACtHR ordered Colombia to provide reparation, including compensation, through its DRP within a year of notification of judgment. Additionally, in the case of Rio Negro Massacres v. Guatemala, which involved five massacres in the early 1980s in Guatemala, the state claimed that victims had received reparations through the DRP and that the state was able to pay reparations to all victims in the same manner. The IACtHR acknowledged the DRP, finding that “the amounts that have already been awarded to the victims in this case at the domestic level under the PNR [National Reparations Program] must be recognized as part of the reparation due to them and subtracted from the amounts established by the Court.”Footnote ¹⁸

These examples suggest that DRPs provide the bureaucratic infrastructure necessary to disseminate damages awarded by the IACtHR. Thus, we expect:

Victim identity and willingness to comply

Moving beyond capacity, we argue that compliance is also driven by the identity of victims involved in IACtHR cases. Victims of human rights violations vary from marginalized individuals and groups (e.g., members of indigenous groups), to political dissidents, to even those convicted of contravening legal statute and serving prison sentences. We know relatively little about whether and to what extent victim identity impacts the behavior of the IACtHR and state compliance with remedial orders of the IACtHR (Stiansen, Naurin, and Bøyum Reference Stiansen, Naurin and Bøyum2020). However, Greenstein (Reference Greenstein2020) suggests that victim identity matters for securing reparations. In her in-depth case study on a West German reparations policy, Greenstein (Reference Greenstein2020) shows that prejudice and the lack of an advocacy organization prevented many Romani Germans from accessing reparations promised by West German law. Victims lacking public support and organizational structure can inhibit government willingness to pay out reparations.

We argue that victims considered “blameless” are viewed by the public as more deserving of awards than victims perceived to be at fault or having done something to deserve the human rights abuses perpetrated against them. Christie (Reference Christie and Fattah1986, 18) differentiates between “ideal” and “non-ideal” victims, in which an “ideal” victim is “a person or category of individuals who - when hit by crime - most readily are given the complete and legitimate status of being a victim.” For example, the ideal victim might include elderly victims of robberies or burglaries, or children who are sexually abused (Christie Reference Christie and Fattah1986). These “blameless” victims garner sympathy and support from the public, including the expectation of state actions like publicly funded compensation schemes as a way to repair their suffering to the extent possible (Miers Reference Miers2019). Ideal victims meet several criteria, including weakness and vulnerability (Schwöbel-Patel Reference Schwöbel-Patel2018); as such, women and children are often considered ideal victims (Kapur Reference Kapur2002). Haglund and Parente (Reference Haglund and Parente2023) argue that women and children victims are perceived by the public to be of privileged status and the public perceives their suffering to be more legitimate and worthy of compensation. We expect that because the public is more likely to support the payment of damages in judgments involving women and children victims, governments are more willing to comply.Footnote ¹⁹

On the other hand, victims not considered blameless garner less sympathy and support from the public. McEvoy and McConnachie (Reference McEvoy, McConnachie, Spencer and Walklate2016, 115) claim that “when victims are not faultless […] they become more problematic, both as an object of public empathy but also in terms of their entitlement to formal compensation on the part of the state.” Miers (Reference Miers2019, 8) highlights this notion with respect to the United Kingdom’s publicly funded compensation scheme for injuries caused to victims of violent crime in the UK, noting “the distinction between ideal (‘deserving’) and non-ideal (‘undeserving’) victims continues to be central to its political legitimacy…the government considered it inappropriate for those with significant criminal records, whose own conduct before, during or after the incident led to their being injured, or who were otherwise of bad character, to receive compensation from public funds.” Government and public backlash is evident in cases involving non-ideal victims. For example, statements from public officials in Peru reflect the unwillingness of the state to pay reparations to victims the state views to be terrorists (Sandoval Reference Sandoval2017, 4). In response to an impending 2015 IACtHR judgment, Peruvian President Humala stated “What I can tell the court and the country is that I am not going to give them a single sol, I am not going to give up a single sol, to the terrorists or their relatives; The Peruvian State is not going to pay a single sol, no matter what the court orders, I am not going to pay money to those who murdered our families and left the country in mourning” (RPP Noticias 2015).Footnote ²⁰ Compliance with the payment of monetary damages to “non-ideal” victims is more likely to generate public backlash. As a result, we expect the state to be less willing to comply with the payment of monetary damages when judgments involve “non-ideal victims.”

For example, consider two judgments against Ecuador. The case of Herrera Espinoza et al. v. Ecuador involved the prosecution of members of a drug trafficking cartel. Specifically, four foreign individuals were arrested during a drug enforcement operation in 1994, denied legal counsel, sustained injuries consistent with torture, and provided incriminating statements to the police. In 2016, the IACtHR found that the state had violated several rights, including the victims’ rights to be free from torture and cruel, inhumane, and degrading treatment; their rights to personal liberty and security; and their right to a fair trial. The Court ordered Ecuador to undertake many reparations, including awarding non-pecuniary damages of $10,000 each to three of the victims and $80,000 in non-pecuniary damages to one victim for torture and unjustified detention.Footnote ²¹ In a 2019 compliance monitoring report, the IACtHR noted the state had only partially complied with the monetary reparation, as restitution payment for some of the victims was pending.Footnote ²²

In a second judgment, Guzmán Albarracín et al. v. Ecuador, the IACtHR found that the state was responsible for the sexual violence suffered by fourteen year old Paola del Rosairo Guzmán Albarracín by the Assistant Principal of the public school she attended, which was tolerated by the school and related to the victim’s suicide. The Court found that Ecuador had violated the right to personal integrity, judicial guarantees, and judicial protection, among other articles of the American Convention.Footnote ²³ Among the various reparations ordered, Ecuador was ordered to pay compensation for pecuniary and non-pecuniary damages to the victim’s mother and sister. Specifically, the Court ordered the state to pay $200,000 to the victim’s family as well as $56,502 for expected lost income of the victim. Additionally, the Court also ordered the state to pay $210,000 in non-pecuniary damages for the suffering experienced by the victim, her mother, and her sister. Importantly, despite the larger award in this case, within a year of the judgment, Ecuador had paid the damages.Footnote ²⁴

Ecuador’s capacity to comply with the orders to pay monetary damages in the case of Herrera Espinoza et al. v. Ecuador and Guzmán Albarracín et al. v. Ecuador is arguably similar. The judgments occurred only one year apart (2019 and 2020), and state-level capacity largely remained unchanged. Both judgments involved only a few victims, four direct victims in the Herrera Espinoza et al. case and a single direct victim (deceased) and two beneficiaries in the Guzmán Albarracín et al. case. The damages awarded were substantially larger in the Guzmán Albarracín et al. case, and capacity concerns would suggest lower levels of compliance as a result. Yet compliance with the payment of monetary damages has been delayed (and only partial) in the Herrera Espinoza et al. case and almost immediate in the Guzmán Albarracín et al. case. The difference? In Herrera Espinoza et al. v. Ecuador, the victims were “non-ideal,” specifically foreign members of a drug trafficking cartel. Being situated between Colombia (the world’s largest cocaine refiner) and Peru (the world’s largest coca leaf grower), Ecuador has experienced a long battle with narcotics smuggling. At the time of the arrests of the victims in this case, the public was supportive of the police carrying out drug raids to help maintain peace in the country (Lusk Reference Lusk2019, 1204). In contrast, the victim in Guzmán Albarracín et al. v. Ecuador was a female minor who had suffered sexual abuse; in other words, an “ideal” blameless victim. In the latter case, there was likely greater public support for compliance, and subsequently greater state willingness to comply, despite capacity challenges.

The theoretical logic outlined above suggests the following observable implication:

Measuring compliance

To test these hypotheses, we begin by identifying all orders for payments of pecuniary and non-pecuniary damages in judgments against the state from 1989 through 2019 (228 judgments in total). Pecuniary damages are material, compensating the actual losses suffered as a direct result of a human rights violation. This can include lost profits, loss of property which has been taken, and costs for medical treatment following a physical injury, among other types of loss (Altwicker-Hámori, Altwicker, and Peters Reference Altwicker-Hámori, Altwicker and Peters2016, 6). Non-pecuniary damages (e.g., moral damages) include “the harm suffered by victims as a result of the human rights violation that are not immediately financial in nature" (Gonzalez-Salzberg Reference Gonzalez-Salzberg2021, 83); this includes “the trauma, anxiety, anger, etc. coming with the attack on human dignity, the loss of trust in state institutions, the loss of beloved persons, the mental and physical pain lasting after torture, imprisonment, censorship, separation from family members, and so on” (Altwicker-Hámori, Altwicker, and Peters Reference Altwicker-Hámori, Altwicker and Peters2016, 6). Both pecuniary and non-pecuniary damages can be awarded in the same case, so the 228 cases yield 416 payment orders.

To measure time to compliance with each of these orders, we looked at evidence from the Inter-American Court’s monitoring reports. In each monitoring report, the Court reviews the orders given to the state and decides whether the state has fully complied, partially complied, or not yet complied based on information from all parties. Claims by both the victim and the state are verified by each other; the Court generally does not accept a payment order as having been fulfilled unless and until both sides agree that it has been paid. Through December 2023, 407 of the 416 pecuniary and non-pecuniary payment orders had been monitored at least once; we exclude the nine unmonitored orders as we do not have any information about the state’s payment status for them. If the Court declares full compliance with the payment order in a particular monitoring report, we then search the report to find the year in which payment was made. In about 15% of instances of full compliance, the payment was made in the same year as the monitoring report was issued (although not necessarily in the same month). On average, monitoring reports lag behind actual payments by 1.74 years. Thus, using the actual payment year gives us the most accurate possible measure of time to compliance. Of the 407 monitored orders, 62.7% eventually result in full compliance, 23.8% result in non-compliance, and the remaining 13.5% end in partial compliance.

However, we are interested in time to compliance, not just the outcome of compliance. Thus, we construct a time series version of the dataset in which the unit of analysis is a payment-order-year. The outcome (full compliance) is coded as 1 if the state fully complied with the order in that year and 0 otherwise. Payment orders remain in the dataset while they are pending compliance. Thus, the 407 monitored payment orders generate 2,770 payment-order-year observations.

Independent variables

Our main independent variable for hypothesis 1 is Monetary award, which is the size of the monetary award in US dollars from Haglund and Parente (Reference Haglund and Parente2023).Footnote ²⁵ To test hypothesis 1a, we interact Monetary award with Revenue, which is total tax revenue as a percentage of GDP.Footnote ²⁶ Because of the large range and heavy right skew in monetary awards, we log the awards in all cases.Footnote ²⁷

To test hypothesis 2, we use the total number of direct and indirect victims in each case from Haglund and Parente (Reference Haglund and Parente2023). As the total number of victims can range from one to over 10,000 depending on the case, we also log this number.Footnote ²⁸ For hypothesis 2, we interact the logged number of victims with the binary variable Reparations program from Greenstein (Reference Greenstein2018). The reparations program variable is coded as 1 if Greenstein indicates the state has a DRP that has paid at least one victim, and 0 otherwise. States that have DRPs have the bureaucratic infrastructure in place to compensate victims and fulfill orders to pay monetary reparations from the IACtHR, even if the domestic reparations apparatus does not exist to compensate victims in Inter-American cases specifically.Footnote ²⁹

To test hypothesis 3, we create binary variables Female victims only and Minor victims only, indicating whether all the victims in the case are females or minors (18 years of age and under), respectively. We also interact these variables (Female × minor victims only) to account for the possibility that female children are given additional special consideration as being members of both vulnerable classes. Since we are interested in assessing the effect of women or child victims compared to non-ideal victims, we exclude the three instances in which the victim is a female criminal and the two instances in which the victim is a criminal (male) minor.Footnote ³⁰ This is admittedly the easiest possible test of the argument. However, if we do not find that women and children (the most sympathetic victims) are paid faster than the least sympathetic victims, we should have little confidence that victim identity affects compliance at all. Additionally, as a more robust test, we compare time to compliance with awards paid to female victims, minor victims, alleged criminals, and convicted criminals compared to all other victims.

Models and control variables

We conduct a discrete time event history analysis with dummy variables for each year post-judgment to capture duration dependence, following Parente (Reference Parente2025) and Pérez-Liñan, Schenoni, and Morrison (Reference Pérez-Liñan, Schenoni and Morrison2023). The dependent variable is the hazard, or probability of full compliance in a given year, conditional on covariates and non-compliance in all previous years. Any order not fully complied with by December 2023 is considered right-censored.Footnote ³¹

We include several control variables to account for alternative explanations for compliance. First, we include three variables to capture other indicators of state willingness. The first is Preliminary objections, which is the number of preliminary objections filed by the state. Higher levels of preliminary objections may be indicative of less willingness to comply since the state was trying hard to get the case thrown out prior to the merits ruling. We also include Accept responsibility, which is an ordinal variable measuring whether the state fully accepted, partially accepted, or did not accept responsibility for the violation prior to the ruling on the merits. Accepting responsibility may indicate greater willingness to comply because the state has already conceded the violation. The third willingness variable is Impunity, which captures the number of years that have passed since the violation occurred. We are agnostic as to the direction of this variable: more impunity might indicate less willingness to comply (since the state has allowed the violation to go unremedied for so long), or it could indicate greater willingness to comply (since the perpetrators may no longer be in power).

Second, we include variables to control for state characteristics that may contribute to compliance. We include Government effectiveness from Worldwide Governance Indicators, capturing perceptions about the quality of state services. We expect higher levels of government effectiveness to be correlated with greater levels of compliance. We also include Rule of law, expecting that states with greater rule of law will be more likely to comply (Simmons Reference Simmons2009).

Third, we include variables for case-level factors that may affect compliance. We include binary variables for Article 4 and 5 violations, which indicate cases in which the right to life and right to physical integrity, respectively, were violated. We are again agnostic as to the direction of these variables. On the one hand, Article 4 and 5 violations are serious violations, and would likely be recognized as such by the public; therefore, states might be more likely to comply in these cases, as the public perceives those cases to involve severe or legitimate violations worthy of compensation. On the other hand, Haglund and Parente (Reference Haglund and Parente2023) show that the size of monetary awards is larger in cases involving Article 4 and 5 violations, which means compliance might be less likely, as the awards would be higher. Next, we include a variable for Cumulative orders which is the total number of monetary orders the state has received to date. States that have received more orders overall might face more financial or bureaucratic constraints on compliance.

Finally, we include an indicator variable for whether the payment order is for non-pecuniary damages and country fixed effects to capture other differences between states not already captured by our existing battery of controls.Footnote ³² Where appropriate, we also include controls for our other hypotheses to isolate the effect of the variable of interest.