III.3 Addressing Conceptual Challenges: Compliance and Impact

This chapter argues that time is a fundamental consideration to understand how States implement the orders of the Inter-American Human Rights System (IAHRS). Time is relevant not only to assess delays in legal outcomes but also to conceptualize variation in the causes of compliance. Given this premise, we propose a new set of criteria to assess levels of compliance and illustrate the use of those criteria with extensive evidence from the Inter-American Court of Human Rights (IACtHR). The proposed approach shows that compliance is sometimes at odds with broader transformative impacts, a point underscored toward the end of the chapter.

Our focus on the IACtHR allows us to place growing concerns about a crisis of compliance in proper perspective. The Court expects full compliance with its rulings for the sake of the victims of human rights abuses.¹ Reparations for victims may include State recognition of human rights violations, financial compensation, the prosecution of perpetrators, or institutional reforms to prevent abuses from recurring. Yet the Inter-American Court has few mechanisms to enforce such orders.² Although the Court issues annual reports and, in extreme cases, can invoke Article 65 of the American Convention of Human Rights (ACHR), the General Assembly of the Organization of American States rarely addresses compliance issues.³ Member States also face a variety of compliance hurdles, including a lack of political will and institutional capacity.⁴

Observers have lamented the ongoing crisis of compliance in the Inter-American System, which continues to cast doubt on its effectiveness.⁵ As César Rodríguez Garavito and Celeste Kauffmann point out, though it is undeniable that the Court has made progress in promoting human rights, “it is equally evident that the implementation of reparation and non-repetition measures ordered by the Commission and the Court is scant.”⁶ Indeed, recent research suggests that noncompliance is widespread, particularly for reparations demanding institutional change. Damián A. González-Salzberg finds that implementation rates range between 3 percent and 31 percent for measures requiring prosecution or legislative changes.⁷ Darren Hawkins and Wade Jacoby report compliance rates between 7 percent and 19 percent for similar measures. In recent years the Court itself implemented a strategic plan to overcome widespread “practices of impunity.”⁸

In this context of perceived crisis, we focus on a technical issue with significant implications: the definition and measurement of compliance. We show that existing metrics cannot give a full picture of aggregate levels of compliance within the Inter-American System. Most reports measure compliance by assessing the percentage of reparations implemented within a particular period of time. However, such measures cannot account for the time it takes for States to comply. Because the Court’s caseload has increased in recent years,⁹ it is difficult to discern whether rates of compliance have decreased over time, or whether more cases are now at the supervision stage.

We advocate an alternative approach, one that considers not only whether a State complies with a given reparation measure but also how long it takes them to do so. We describe this analytic perspective as a discrete-time approach, for reasons explained in the next section. Although a discrete-time approach can help scholars and practitioners evaluate levels of compliance more accurately, it has rarely been applied to an analysis of the Inter-American Court.¹⁰

The chapter proceeds through three sections. In section 2, we introduce the discrete-time approach for assessing rates of compliance then discuss the relevance of time as a crucial dimension of implementation before comparing two traditional (static) metrics of compliance against two discrete-time metrics. We introduce the concepts of a yearly probability of compliance and an expected time for compliance (ETC) and document their objective equivalence. Section 3 illustrates these concepts with data from all cases decided by the IACtHR until 2018. In addition to comparing Latin American States, this section shows that the implementation of Court orders follows a distinctive life cycle, as the yearly probability of compliance varies over time. There is a window of opportunity in which States tend to comply, but compliance becomes less likely the longer a reparation remains under supervision. The final concluding section 4 addresses the distinction between compliance and impact. Though it is true that the effectiveness of the Inter-American System rests “to a large measure on compliance with the decisions of its organs,”¹¹ we identify four distinct patterns of alignment between compliance and impact: direct transformative impact, indirect transformative impact, State resistance, and backlash.

Compliance with international court rulings necessarily involves a temporal dimension. States must adapt their behavior in order to conform to a norm or ruling,¹² and because any change in behavior is necessarily never immediate, time is a crucial dimension to consider when conceptualizing and measuring compliance. In this section, we compare two approaches to quantify compliance. The first, traditional approach calculates rates of compliance across cases (or reparation measures) at a particular point in time, offering a static “snapshot” of the situation. The second approach introduced in this chapter conceptualizes compliance as an event that takes place within discrete-time units (years), and thus allows for a dynamic analysis of the process.

To understand the difference between the two approaches, imagine a hypothetical case in which the IACtHR orders a State to comply with two reparation measures. Three years later, the Court issues a supervision resolution documenting that the State complied with the first order within two years of the decision but has yet to comply with the second order. The conventional procedure estimates the rate of compliance across orders at the time of the resolution. This “snapshot” of the situation would show that by the end of the third year, the State has complied with 50 percent of the orders (one out of two). In contrast, the discrete-time procedure records every year until an order meets compliance. In the previous example, the first order met compliance after two years, thus the annual rate of compliance is 1/2, that is, an event of compliance over a two-year period. The second order has not yet been met with compliance by the end of the third year, thus the annual rate for the second order is 0/3. We can easily aggregate this information across reparation measures. Overall, the yearly probability of compliance for the State is 1/5.

Why is the second approach necessary? The conceptual implications of the two approaches become clear if we imagine that the Court issues a new supervision resolution a decade after the decision. The new resolution reminds us that the State complied with the first order within two years but warns that the State has not complied with the second order ten years after the ruling. A decade after the ruling, the snapshot analysis would reiterate the initial conclusion: the rate of compliance remains at 50 percent. In contrast, the discrete-time estimate would penalize the State for the long delay in compliance. The annual rate of compliance for the first order is still 1/2, but the annual rate for the second order is now 0/10. Overall, the yearly probability of compliance for the State is now 1/12. That is, one event of compliance, on average, every twelve years.

Time is a relevant dimension of the concept of compliance for two reasons. First, as the previous hypothetical example illustrates, delays are relevant to characterize legal outcomes. Even if States conform to the orders of the IACtHR, they may display considerable divergence in how long they take to do so. Delays with compliance ultimately matter for the victims and for the Court’s reputation. Second, time is relevant to understand the causes of compliance. Contextual variables that influence State behavior normally fluctuate over time. In the following sections, we discuss the reasons for this fluctuation and explain how the discrete-time approach allows us to improve our understanding of those issues.

A good measure of compliance must take into account not only whether a State complied with a ruling but also how long it took to do so. States are unlikely to respond to Court rulings right away, and a variety of factors can impose delays. To treat equally cases in which a State complied after fifteen years with cases in which a State complied after fifteen months, for instance, would draw a false equivalence between two very different patterns of State behavior.

Consider, for example, the Garrido y Baigorria v. Argentina case. In response to the illegal detention and disappearance of Adolfo Garrido and Raúl Baigorria in 1990, the Court ruled that Argentina needed to compensate the families of both victims, pay the lawyers’ fees for their work on the case, identify two extramarital children of Raúl Baigorria—in order to pay them reparations—and investigate and sanction the authorities complicit in the disappearances. Although these orders were issued simultaneously in 1998, Argentina’s compliance record varied according to the reparation measure. A snapshot of this case in 2017 indicated that Argentina had complied with three-quarters of the reparation measures ordered by the Court. However, Argentina took nine years to comply with the first reparation, five years to comply with the second, and nineteen years to comply with the third. Because of this variance, the aggregate rate of compliance observed in 2017 (3/4) masks important information about Argentina’s overall record and variation by type of reparation.

Conversely, time also matters for assessing noncompliance. A snapshot treats a lack of compliance at the end of the observation period (say, by 2017) as a negative outcome, irrespective of the time elapsed. Yet the hypothetical example introduced at the start of this section illustrates why this metric can be misleading. In Garrido y Baigorria, Argentina failed to comply with only one of four orders, but its lack of compliance with the fourth order deserved very different interpretations nineteen years after the ruling compared with two years after the ruling. Delays represent an important feature of a State’s compliance record that scholars must consider when measuring levels of compliance.

The second reason to incorporate a temporal element is that compliance is not a static phenomenon. The contextual factors that influence a State’s propensity to comply with a ruling evolve over time. For instance, changes in governments or regimes often affect the likelihood that leaders will recognize State culpability in past human rights abuses. Guatemala came into rapid compliance with a variety of historical obligations following the election of Óscar Berger in 2004.¹³ A snapshot measure that encompasses this period would report a higher level of compliance for Guatemala but fail to account for the sudden increase associated with political change. Other contextual variables influence a State’s propensity to comply over time. These include public opinion, the electoral calendar, economic conditions, and the political ideology of incoming governments.¹⁴

Even if these variables remain stable for several years, we may observe temporal fluctuations when we analyze the probability of compliance over time. As we discuss in section 6, compliance follows a distinctive life cycle. Compliance is unlikely in the wake of a ruling, becomes more likely after States have had time to implement the required measures, and it becomes unlikely again as reluctant States drag their feet. A good definition of compliance should allow us to document this life cycle.

We can now compare four different ways of conceptualizing and measuring compliance according to their capacity to address the two problems discussed previously. First, a static rate of compliance reflects the percentage of closed cases—or the percentage of implemented reparation measures—at the time of the snapshot, without acknowledging changing conditions. Scholars in this tradition look at a set of orders within a given period and simply calculate the proportion of reparations that were met with compliance.¹⁵

A second, less common approach reports the average number of years States take to comply. This measure tackles the first challenge discussed previously (delays) by reporting the average time to compliance. However, because the units of analysis are cases or reparation measures rather than discrete-time units, this measure cannot tackle the second problem (changing conditions over time). Moreover, the analyst is able to measure the time to compliance only if compliance has taken place by the end of the observation period. In the previous example of Garrido y Baigorria v. Argentina, an analyst taking a snapshot of the case by year nineteen would observe an average time to compliance of eleven years ((9 + 5 + 19)/3) without accounting for the fourth, pending measure. Thus, this approach presents the problem of selection bias, given that States are likely to comply with lenient measures first. To overcome the limitations of the snapshot approach, we advocate for the discrete-time approach introduced earlier. There are two possible discrete-time measures of compliance, one expressed as a yearly probability and a second expressed in terms of duration. Although they are expressed in different metrics, these expressions are mathematically equivalent.

The yearly probability of compliance, illustrated in section 2, reflects the likelihood that a State will comply with a given reparation measure at a given point in time. Because this third metric can vary from year to year, it is sensitive to changes in explanatory factors. For instance, the probability of compliance may be low in year t but increase substantially after a new government enters office in year t + 1. We show in section 6 that when we compare a large number of reparation measures this metric allows us to reconstruct the life cycle of compliance. Moreover, the yearly probability of compliance contains the necessary information to assess duration—a low probability of compliance in a given year suggests that the State will take long time to comply—but it is not a very intuitive metric to assess delays. Therefore, we need an alternative metric to convey this information.

For ease of interpretation, we propose a fourth measure: the expected time for compliance (ETC). The ETC represents the expected number of years until the State implements an order. We calculate the ETC in three steps. First, we record the number of discrete-time units (years) until we observe compliance. Returning to Garrido y Baigorria, for instance, there are nine time units for the first reparation measure, five for the second, nineteen for the third, and nineteen and counting for the fourth. Second, we estimate the yearly probability of compliance—the third metric discussed in the previous paragraph. The average probability of compliance per annum in Garrido y Baigorria is 3/52: three events of compliance in 9 + 5 + 19 + 19 time units. Third, we retrieve the ETC by taking the inverse of that probability. If the average probability of compliance is 3/52, the inverse of this figure provides the expected number of years (17, or 52/3) until the State honors an order.

Because the ETC is derived from the yearly probability of compliance, the discrete-time approach allows us to report the ETC and the estimated probability of compliance interchangeably. These two statistics are conceptually equivalent: an ETC of two years reflects a compliance probability of 0.50, while an ETC of ten years reflects a compliance probability of 0.10. We often prefer the ETC because of its intuitive interpretation: a high ETC means that the State will likely take many years to comply, while a low ETC indicates that a State is likely to comply promptly.

Before discussing our findings, it is important to note some caveats for the interpretation of our fourth metric. The ETC already accounts for the possibility that a State will not comply with a given order. The measure penalizes cases of noncompliance by reporting longer expected compliance horizons. Therefore, a very long ETC should not be interpreted as a specific prediction about the number of years until compliance but rather as an indication of unlikely compliance over the long run. For instance, an ETC of one hundred years does not imply that a State will wait a century to comply with a ruling but that noncompliance is likely over the long run—the yearly probability is just 0.01. In addition, because we normally report ETCs that summarize information for several years, this figure may mask important information about the implementation life cycle. Two States may have similar ETCs but vary in their propensity to comply at specific points in time following a ruling. Given this limitation, in the following section we report numerical information about ETCs to compare States, but also present graphical information about cycles of compliance.

We illustrate the four metrics discussed previously using evidence from the IACtHR. Between 1989 and 2018, the Inter-American Court ruled against States in 238 cases, ordering some 1,783 reparation measures. We compiled an original data set for these cases, documenting the year of each ruling and the year of the resolutions in which the IACtHR determined that the State had complied with the reparation measures. Because the Court’s supervision resolutions identify two possible levels of compliance (“partial” or “full”), we calculate measures of compliance for two events: the first acknowledgment of any form of compliance whether partial or full, and the acknowledgment of full compliance, that is, the end of the supervision process for a particular order.

Table III.3.1 summarizes this information, comparing States along the two snapshot measures of compliance. The first four columns in the table identify the country, the number of cases in which the IACtHR ruled against the State, the number of cases that the Court has archived due to full compliance, and the number of reparation measures ordered in total. The following columns present conventional measures of compliance based on a snapshot at the end of 2018. The two columns under “Compliance (%)” compare rates of implementation across countries, using the conventional estimate for the percentage of reparations. The last two columns compare the average number of years elapsed between the Court’s ruling and the moment when the Court acknowledges compliance.

The initial portrait presented in Table III.3.1 is admittedly dim, with only 33 out of 238 cases archived. This means the IACtHR has closed only 14 percent of the cases due to full compliance, while 86 percent of the cases still burden its supervision efforts. At the country level, it is also disappointing that no State has closed more than half of its cases. This evidence has played into the hands of critics who highlight the limited effectiveness of the Inter-American Human Rights System.¹⁶

The first three columns of the table also illustrate some problems with an analysis based on overall cases, which does not disaggregate rulings into specific orders. The column reporting the total number of cases makes it evident that we risk placing very different situations in the same category when comparing the rate of archived cases. For example, Uruguay, Colombia, and Venezuela had closed no cases by the end of 2018, yet Uruguay had only two cases pending, while Colombia and Venezuela had some twenty pending cases each. Moreover, the compliance rate for specific orders shows that the political will in Colombia and Venezuela has been quite different.

The remaining columns in Table III.3.1 compare levels of compliance based on individual reparation measures. To overcome some limitations of the analysis based on cases, legal scholars opted to break down cases into individual reparation.¹⁷ The focus on individual reparations represented a considerable advance. States such as Bolivia, Ecuador, or Panama, which appear as noncompliant in most cases, are implementing most of the reparation measures ordered in the context of those cases even though the cases remain open. Perhaps most importantly, the literature analyzing compliance with specific reparation measures documented which type of remedies is more likely to be implemented. The evidence consistently indicates that States are more likely to honor monetary compensation measures and less likely to implement nonrepetition measures and orders addressing the State’s obligation to prosecute perpetrators.¹⁸

The different rates of compliance across different types of reparation measures underscore the importance of treating compliance as a gradual rather than a discrete outcome. A gradual approach to compliance is especially important when it comes to orders that involve long-term processes and several domestic actors, such as guarantees of nonrepetition that demand changes in legislation. Specialists have argued for a flexible understanding of compliance, given that the Inter-American Court has a relatively expansive and maximalist jurisprudence.¹⁹

Fortunately, the IACtHR reports partial compliance—that is, demonstrated progress toward implementation—in its monitoring resolutions. Table III.3.1 illustrates the contrast between a strict definition of compliance, acknowledging only full implementation (with an average rate of 34%) and a broad definition including partial or full implementation (with an average rate of 42%). For complex orders that involve, for example, investigating, judging, and sanctioning perpetrators, specialists argue for an even more nuanced classification that goes beyond the two categories of partial and full compliance.

The central columns in Table III.3.1 report rates of compliance as a percentage of reparations with any level of implementation (partial or full) or strictly in full compliance. We consider all reparations ordered by the Court from 1989 to 2018. The picture emerging from this analysis, based on individual reparation measures, is far more promising than the one based on individual cases. More than 40 percent of the reparations ordered since 1989 met with some degree of compliance, and over a third have been fully complied with. This might be a reasonable number for a Court credited with ordering high-bar reparation measures and sometimes at the cutting edge of human rights jurisprudence—something that other tribunals, like the European Court of Human Rights (ECtHR), do not aim for. As a point of reference, the ECtHR obtained a 55 percent implementation rate for its leading cases between 2009 and 2018.²⁰

The last two columns in Table III.3.1 report the observed time to compliance for the average reparation measure by country. The figures are somewhat surprising, with just three and a half years on average between the date of the ruling and the date when the IACtHR acknowledges compliance. However, these estimates exclude all reparation measures without implementation and thus present an overly optimistic picture. Countries with extremely low rates of compliance, such as the Dominican Republic (7.9%), may also display a prompt (three-year) execution of the few measures they actually choose to implement. Only a dynamic duration analysis is able to overcome this inferential problem.

In sum, Table III.3.1 illustrates the advantages and the limitations of snapshot measures of compliance. By moving from an analysis of overall cases to an analysis of specific orders (reparation measures), conventional statistics provide important insights. At the same time, however, they fail to effectively account for the role of time. A specific example will help convey this point: analyzing compliance in 2012, Cecilia Baillet noted that Mexico, which had at that time a zero percent compliance rate at the case level, behaved particularly well with regard to orders of investigation and punishment, complying with a remarkable 67 percent of those challenging orders.²¹ Thus, the analysis of specific measures provides more nuanced information than the analysis of overall cases. However, nuance gained from comparing orders does not translate into nuance over time. The Court decided on four additional cases involving Mexico within five years of Baillet’s study, issuing three of the four rulings in 2018. If we had conducted a similar analysis of decisions involving Mexico by the end of 2018, compliance rates at the reparation level would have dropped considerably because the State did not have enough time to implement the orders within a few months.

To overcome these limitations, Table III.3.2 displays discrete-time measures for the same cases. The central columns report the yearly probability of compliance, and the last two columns report the ETC for each member State. This ensures that countries with notable delays are brought to the forefront.

The States most likely to comply with pending Court orders have been Ecuador, with an average yearly probability of partial or full compliance of 0.169, or 16.9 percent; Chile, with 16.3 percent; and Bolivia, with 14.9 percent. At the other end of the spectrum we find Haiti and Trinidad and Tobago, with no compliance events to date; and Venezuela, with a yearly probability of 0.007, or just 0.7 percent. Trinidad and Tobago and Venezuela denounced the American Convention in 1998 and 2012, respectively. As a result, the probability of Ecuador honoring a Court order has been twenty-four times greater than the probability of Venezuela doing so.

It is worth noting that the number of reparation measures pending is unrelated to the probability of compliance. Some States, like Haiti and Barbados, are confronted with only a few orders, but they are unlikely to comply with them. Countries like Ecuador and Mexico, however, confront a large number of orders but they display annual rates of compliance well above the mean. It follows that backlog is not the main explanation for annual rates of compliance. Causality could in fact flow in the opposite direction, as unresponsive States may discourage victims from appealing to the Inter-American System.

For a more intuitive metric, the last two columns of Table III.3.2 display the ETC. Because the ETC figures incorporate information about noncompliance, the contrast with Table III.3.1 can be shocking. While the observed time for compliance for measures honored by the Dominican Republic is three years, the expected time for compliance for the country is eighty-three years.

To place those States in perspective, Figure III.3.1 plots the expected time until the first manifestation of partial or full compliance for all countries in Table III.3.2. The figure allows us to distinguish between two qualitatively distinct groups: noncompliers—the Dominican Republic, Haiti, Trinidad and Tobago, and Venezuela—and the rest. Noncompliers have zero probability of compliance in any given year or display unrealistic ETCs that indicate a probability effectively approaching zero. The remaining States present ETCs that range continuously between six and twenty-one years, as in the cases of Ecuador and Paraguay. Such a continuum suggests that States in this second group belong in the same category: their differences, although very significant, are a matter of degree. The figure shows that eight countries in this group are likely to comply with their reparation orders within a decade. These country averages, however, hide a considerable amount of variance across types of reparation measures and over the life cycle of reparations, as we discuss in the following section.

Based on Table III.3.2. ETCs undefined for Trinidad and Tobago and Haiti.

The most important advantage of discrete-time measures is their capacity to track levels of compliance over time. Although Table III.3.2 reports the average probability of compliance for each State in a typical year, a State’s propensity to comply naturally varies over the years. This variation in part reflects idiosyncratic conditions, for example, government changes, but it also reflects the nature of the implementation process. It is unlikely that States will comply with reparation measures immediately after a ruling because it takes time to address the Court’s requests.

Even if most factors driving compliance remain stable, on average we observed temporal fluctuations when we analyzed compliance in time. Willing States will be unlikely to comply immediately, but they will do so within a few years. After willing States have complied within a reasonable period, only orders issued to reluctant States will remain in the analysis. Thus, the average probability of compliance should be low immediately after a decision (as willing States prepare to comply), will increase within few years, and then drop again when only reluctant States remain under supervision. While conventional measures of compliance (calculated for cases or reparation measures) are unable to track changes in the probability of implementation over time, discrete-time measures (calculated annually) allow us to document the life cycle of compliance with precise accuracy.

Figure III.3.2 documents the life cycle using data from the IACtHR. The horizontal axis reflects the number of years a measure has remained under supervision; the vertical axis reflects the probability of compliance by the end of the year. The series tracks the yearly probability of compliance for two outcomes: the first indication of compliance, whether partial or full, and indicated by the dotted line, and full compliance, indicated by the solid line. Annual probabilities are calculated for pending orders, that is, those without any implementation (dotted line) or those without full compliance (solid line). Thus, while the solid line in Year 1 reflects 69 episodes of full compliance for 1,607 pending orders, with a probability of 0.043, or 4.3 percent, a similar rate in Year 6 reflects 32 episodes for 734 pending measures, with a probability of 4.4 percent. Only 11 orders remain under supervision by year 20.

Although the average ETC reported in Table III.3.2 is more than fourteen years, the figure shows that this average hides an uneven historical trajectory: the probability of compliance increases consistently within the first three years of a ruling, as willing States prepare to implement the required measures. By the third year the probability of any form of compliance is about 16 percent, and the probability of full compliance is close to 11 percent. The likelihood of compliance declines in the following years, hitting a nadir by the end of the first decade.

In practice, this life cycle means that the cumulative probability of compliance, whether partial or full, approximates 50 percent within the first decade. The number of reparation measures monitored by the IACtHR therefore drops considerably after ten years. This pattern is hard to grasp from Table III.3.2, since the average ETC is prolonged by reluctant States and by a small percentage of measures without implementation. Figure III.3.2 therefore suggests that there is more room for optimism than commonly assumed. Moreover, the data tends to overestimate the time to compliance. Actual implementation takes place a year or two before the Court acknowledges State behavior. Most studies, including the one contained in this chapter, employ the date-of-supervision resolutions as the official time to compliance, but on average State actions precede those resolutions by at least eighteen months. Figure III.3.2 also suggests that compliance with lagging reparation measures appears to improve about two decades after a ruling. However, because very few measures remain open at this stage, this “surge” reflects the experience of only a very few cases (Castillo Petruzzi v. Perú; Garrido y Baigorria v. Argentina and Suárez Rosero v. Ecuador) and thus it is uncertain.

The study of life cycles introduces a dynamic perspective to the analysis. It provides a more encouraging outlook than the static comparison of compliance rates (as in Table III.3.1) or the comparison of ETCs across States (as in Table III.3.2). It also allows scholars and practitioners to identify the best window of opportunity to elicit State compliance. Given the large number of cases decided by the Court in recent years, it is hard to anticipate whether the observed life cycle will remain stable in the future.

This chapter has shown that a dynamic analysis of compliance is able to sustain more reliable (and perhaps more optimistic) conclusions regarding how the Inter-American Human Rights System influences outcomes in Latin America. However, the longitudinal perspective also calls for a long-term distinction between compliance and impact. Compliance narrowly defines whether State actions align with the orders of the Inter-American System, while impact refers to the broader legal and social consequences of those orders.

An extensive literature has acknowledged that legal decisions have implications that transcend State behavior. For instance, Yuval Shany (2014) develops the idea of international court effectiveness to analyze whether tribunals are able to “attain, within a predefined amount of time, the goals set for them by their relevant constituencies.”²²  Karen J. Alter, Laurence R. Helfer, and Mikael Rask Madsen (2018) conceptualize international court authority to understand “how the audiences that interact with international courts embrace or reject international court rulings.”²³ We build on those distinctions to emphasize that over the long run compliance and transformative impacts may not coincide when it comes to expected outcomes. In the ideal-typical cases, State compliance leads to positive impacts, and a lack of compliance leads to negative human rights outcomes. However, observers can also identify “misaligned” instances in which a lack of compliance is followed by unexpected positive transformations or, by contrast, situations in which compliance triggers a backlash against the courts. We therefore close our discussion by identifying four potential patterns that link compliance and impact: direct transformative impact, indirect transformative impact, resistance, and compliance backlash.

Compliance with human rights rulings often creates lasting consequences for society. In the domestic realm, iconic rulings, such as Brown v. Board of Education (1954) in the United States, have contributed to profound social transformations, even though compliance was achieved after considerable resistance. In the Inter-American System moreover, some decisions have transformative impacts beyond the original case and country. For example, when the Argentine Supreme Court nullified the 1987 amnesty law in 2005, it relied on the Barrios Altos case, an IACtHR decision referring to Peru (2001). This pattern of recursive interaction between domestic law and the Inter-American System led Armin von Bogdandy et al. to conceptualize “an original Latin American path of transformative constitutionalism,” described as the emergence of an Ius Constitutionale Commune in Latin America.²⁴ This development “builds, far more than on neo-constitutionalism, on the Inter-American system of human rights, whose influence in the region the authors of the 1990s could not foresee.”²⁵

This type of pattern refers to surprising instances in which court rulings induce positive outcomes despite the lack of direct compliance. For example, although the two central measures ordered by the IACtHR in the 2006 Almonacid Arellano y otros v. Chile case—involving the State’s obligation to investigate and sanction human rights violations—remain without compliance to this day, the Criminal Chamber of Chile’s Supreme Court cited the decision within a few months in the Hugo Vásquez Martínez and Mario Superby Jeldres case to assert that crimes against humanity are not subject to statutes of limitations. This was not the first time that Chilean courts built on international law, but while “before Almonacid international law was mostly mobilized by parts of the Chilean judiciary as an interpretative tool, following the IACtHR ruling, international legal norms have also been deployed as distinctive legal criteria.”²⁶

The evidence presented in previous sections shows that States too often resist the implementation of reparation measures. In some cases, however, passive resistance escalates into active defiance. Wayne Sandholtz et al. note that “non-compliance with, and even criticism of, the decisions of international human rights courts are normal forms of resistance to adverse rulings. But sometimes States strike at international human rights courts with more far-reaching forms of resistance.”²⁷ States may cease to cooperate with the court, narrow the court’s jurisdiction, limit access (standing) to the court, withdraw from the court’s jurisdiction, and even—as in the case of the Southern Africa Development Community Tribunal—collectively terminate the court.

As mentioned before, Trinidad and Tobago (1998) and Venezuela (2012) have denounced the American Convention of Human Rights and withdrawn from the IACtHR’s jurisdiction. The Dominican Republic has not taken this step, but its Constitutional Tribunal ruled in 2014 that the IACtHR’s decisions are nonbinding. Sandholtz et al. discuss these cases as instances of backlash. However, we want to emphasize that those reactions were part of a deliberate strategy to avoid compliance. These preemptive forms of backlash are analytically distinct from the backlash triggered by compliance efforts discussed in the next section.

We employ this term to refer to episodes in which actual or anticipated compliance with controversial rulings triggers unexpected negative consequences. For instance, in late 2017 the IACtHR asserted equal rights for same-sex couples in a consultative opinion (24/17) requested by Costa Rica. The Constitutional Chamber of the Costa Rican Supreme Court acknowledged the opinion and ultimately ruled against the Family Code in August 2018. However, the Inter-American Court’s position triggered a political storm in the context of the 2018 presidential election campaign. A conservative public backlash against the decision bolstered mass support for presidential candidate Fabricio Alvarado, who railed against the Court and won the first round of the presidential election, though he was defeated in the runoff.

The Costa Rican experience illustrates a critical fact: compliance backlash is led by political entrepreneurs who exploit social reactions against unpopular rulings. We distinguish this pattern from instances of preemptive backlash discussed previously, in which State agents undermine human rights tribunals as part of a deliberate strategy to avoid compliance. Although the boundaries between the two categories are sometimes ambiguous, the distinction can help us differentiate between qualitatively different situations. For example, in the context of the European Court of Human Rights, the 2014 Yukos case resembles an example of preemptive backlash by Russia, while the 2005 Hirst case resembles an example of compliance backlash from the United Kingdom. Nevertheless, the distinction can be fluid: compliance backlash easily turns into a preemptive strategy when political actors leading the charge against human rights tribunals gain control of the national government or domestic courts. The complex relationship between compliance and transformative impacts underscores the importance of adopting a diachronic perspective when assessing State compliance with the orders of the Inter-American System. The discrete-time approach introduced in this chapter offers an effective strategy to address some of the major conceptual challenges created by such a diachronic perspective. Further development of this approach will therefore be crucial to advance consistent standards within the region’s multilevel legal system.