Why Ratify?

Reservations, Institutional Changes, and Commitment to Human Rights Treaties

By Moonhawk Kim, Yvonne M. Dutton, and Cody Eldredge moonhawk.kim@colorado.edu ydutton@iupui.edu cody.eldredge@colorado.edu

Revised: September 4, 2012

Abstract Scholars have long analyzed the causes of states ratification of human rights treaties. Existing arguments point to various factors that influence state ratification decisions, but they tend to take the characteristics of the institutions as a given at a particular point in time and test how states behave based on those static characteristics. We propose and test a contrasting argument that emphasizes how states change their behavior over time depending on how the institutions actually function. We argue that while many states may believe they could benefit from the constraints imposed by human rights treaties, only those with little risk of noncompliance will commit initially when treaty constraints may appear on paper to be strict and inflexible. States with poorer prospects for compliance will likely avoid the sovereignty costs associated with joining an institution that may punish them before they are able to improve their domestic practices. These states may wish to be constrained, but will wait to ratify until they have observed increased flexibility in how the institution actually functions. To test this theory about the effects of institutional flexibility on state behavior, we use data on any reservations states made to the treaty at the time of commitment. Our analysis examining state behavior as relates to the Convention against Torture provides support for our theory. We find that states with poorer initial prospects for compliance tend to join the CAT when flexibility is high and thereafter take advantage of the institutions constraints to improve their human rights practices.

I.

Introduction Since World War II, the world community has increasingly turned to international

institutions as a way to monitor and improve states domestic legal practices relating to the protection of individual human rights. Indeed, the international human rights regime now boasts six primary treaties to which the great majority of states have committed. But, are states making meaningful commitments to these treaties? Does the fact of their ratification signal an intention to better their domestic practices? Although scholars have long analyzed the causes of states ratification of human rights treaties, the existing literature does not provide a precise answer to the questions posed above. Downs, Rock, and Barsoom (1996), for example, argue that a states decision to ratify a particular treaty is endogenous to its expectations about future compliance. In other words, states will avoid risks to their sovereignty and will tend to join treaties that require them to do only what they would have done in the absence of the treaty. On the other hand, a significant body of literature suggests that one reason states join international institutionsincluding international human rights treatiesis precisely so that they can tie their hands and commit to better future practices (Simmons and Danner 2010; Simmons 2000; Moravcsik 2000). Both of these contrasting explanations help us understand state ratification behavior, but both also assume a static institution when analyzing state commitment decisions. In particular, they tend to take the characteristics of the institution as given at a particular point in time and test how states behave based on those static characteristics. Because institutions do change over time and because states join institutions at various points during the institutions evolution, however, we suggest that the existing theories may not account for all state ratification behavior. This

paper thus proposes and tests an argument that situates state ratification behavior in the context of the institutions change and functioning over time in an effort to aid in a more complete understanding of why states join international human rights treaties. In particular, we argue that while many states may believe they could benefit from the hand-tying constraints imposed by human rights treaties, only those with little or no risk of being punished for bad and noncompliant behavior will commit initially when treaty constraints may appear on paper to be relatively strict and inflexible. States with weaker domestic protection for human rights and with poorer prospects for compliance, on the other hand, are likely to avoid the sovereignty costs associated with joining an institution that may punish them before they are able to improve their domestic practices. These states may wish to be constrained, but will be more likely to ratify only after they have observed increased flexibility in how the institution actually functions. We use novel data on one aspect of how institutions actually function over time to empirically test the effects of institutional flexibility on state behavior. Specifically, states may commit to a given treaty in a qualified manner by filing reservations which purport to exempt them from certain obligations spelled out in the text of the treaty. To the extent that a significant number of states file such reservations, they limit the realm of the institutions influence and also signal to non-member states that the original treaty terms can be interpreted more flexibly to allow behavior that does not strictly comply with treaty terms. The results of our analysis examining state behavior as relates to the Convention against Torture provide support for the idea that the ratification behavior of states with poorer initial prospects for treaty compliance will be influenced by increased institutional flexibility. The

results indicate that as institutional flexibility increases, states that would most benefit from such flexibility become more likely to ratify, but that they also use the institutions constraints to improve their human rights practices. These results generate implications not only for understanding the international politics of human rights in particular, but also for analyzing the effects of international institutions more generally. We proceed by briefly situating our argument within the contemporary literature addressing the design of international institutions and why states commit to those institutions. We then more fully elaborate on our argument that in the international human rights context, observed increased flexibility in how the institution functions over time can induce some states with poorer prospects for compliance to join and also to use the institution to constrain their behavior and improve their domestic protection of human rights. Next, we present the results of the empirical analyses testing the hypotheses generated from our argument in the context of state ratification of the CAT. The conclusion discusses the implications from our findings and directions for future research. II. Treaty Design and State Ratification Decisions What causes states to join an international human rights treaty? One strand of the literature suggests that states design international treaties so that the states joining will not be required to cooperate in any deep and meaningful manner (Downs, Rocke, and Barsoom 1996). According to this literature, states will want to avoid sovereignty costs and will prefer to join treaties that will only require them to do what they otherwise would have done in the absence of the treaty. Jana Von Stein (2005) makes a similar point about treaty selection effects, noting that institutional design is at least in part endogenous because states will not likely invest their time

and resources in creating agreements unless they also have some interest in complying with those agreements. Thus, as Von Stein (2005) argues, we may find that treaties will actually screen out potentially bad and noncompliant members, rather than inducing them to join and thereafter alter their behavior to conform to treaty terms. Another strand of the literature, though, points to the constraining effects of treaties and how states might use costly commitment to their benefit and to send a credible signal about their intention to act in a particular way in the future under conditions of incomplete information and uncertainty (Simmons 2000). Simmons and Danner (2010) argue that states with poor prospects for compliance will ratify the treaty creating the International Criminal Court (ICC) in order to send a costly signal to their domestic audiences to better protect against human rights abuses in the future. Moravcsik (2000) also posits that new, transitioning democracies can overcome the sovereignty costs associated with joining international human rights treaties by locking in the treatys democratic principles and thereby constraining the activities of future governments that may seek to subvert democracy. Accordingly, some states may wish to ratify human rights treaties because the potential for incurring costs can actually prove beneficial in that commitment enables them to signal to the world and their domestic audience that they are sincerely committed to a particular policy change. However, some states might have a higher risk of being immediately punished for bad and noncompliant behavior even if their aspiration and intention is to eventually have policies that comport with treaty terms. To be enticed to commit to a treaty and bind themselves to future policy change, some states may require evidence that the institution is at least somewhat flexible so that the state will not be punished before it has time to implement the domestic changes necessary to permit it to fully comply. Indeed, Koremenos (2005) finds that flexibility in the 5

form of agreements of more finite duration help induce states to willingly tie their hands and meaningfully commit to an international agreement under conditions of high uncertainty. Thus, flexibility may not only provide the kind of insurance that will convince states to cooperate and commit to future policy change, but it may also ensure that international institutions achieve greater breadth and variety in their membership. Each of these theories about the interaction of institutional design and treaty membership has merit and is helpful in explaining how international agreements can have both screening and constraining effects. Each, however, examines international agreements from a static perspective and looks at the resulting state behavior based only on the initial text of the agreement. As such, each of the theories discussed likely fails to explain all of the reasons why states join international agreements and whether states are likely to change their behavior as a result of their membership. Taking a more dynamic approach by observing how international institutions change over time as a function of the behavior of extant members can lead to a new and fuller understanding of how an institution may both screen and constrain members as it becomes more flexible. III. State Commitment in Response to Increased Institutional Flexibility We advance a theory of flexibility to explain why some states with poorer initial prospects for complying with international human rights treaties might nevertheless commit and use the institutions constraints to improve their domestic practices. In doing so, we make an assumption about the type of states to which our argument applies. States that currently engage in human rights violations themselves are unlikely to respond to flexibility. By contrast, states that have engaged in violations in the past but currently do not and states that do not currently

engage in violations but may do so in the future are likely to attempt to tie their hands and to seek an opportunity to do so. Our flexibility theory also proceeds from the assumption that all international human rights treaties place constraints on their membership even if the body in charge of treaty compliance has no formal powers to issue binding sanctions against states that fail to comply with treaty terms. Indeed, we assume that at the very least, commitment places constraints on member states that the states would not have confronted had they refused to join the treaty. Thus, the basic calculation of interest for states with weak domestic protection for human rights is whether it is better for them to be inside a human rights institution or outside. That many weak human rights protectors do not ratify human rights treaties immediately even when offered large material incentives by liberal states to do so suggests that states believe being inside the institution can have costly adverse consequences. In short, human rights institutions provide sufficient constraints for states to consider. This renders increased flexibility a significant aspect of human rights institutions for states to consider when making commitment decisions. Unlike Koremenos (2005), however, we do not focus on the flexibility that can be built into agreements by, for example, making them of limited duration so as to expose states to less future uncertainty. Analyzing the flexibility that can be built into agreements certainly increases our knowledge about how to structure treaties so that states most concerned about uncertainty feel compelled to join. Instead, we focus on a different and understudied aspect of institutional flexibility: the flexibility that states may obtain, not from the terms of the agreement itself, but rather by observing how the institution actually functions and changes over time. After all, institutions do change over time, and some states join those institutions later than others.

Furthermore, if states are in fact wary of committing under conditions of uncertainty as Koremenos (2005) suggests, some might conclude that one way to reduce that uncertainty is to observe how the institution functions. At the time of drafting, states may not have contemplated the many different types of state behavior that may or may not satisfy particular treaty terms. If, over time, the institution demonstrates that it allows some conduct that violates the treaty provisions, then states obtain helpful information about the institutions flexibility. Similarly, member states can signal to non-members the institutions flexibility when they fail to strictly and fully comply with treaty terms, but nevertheless continue to remain members without meaningful consequences. Of course, some states may join international agreements later because they choose to wait until their domestic policies are more consistent with provisions of the institutions. However, the timing of state ratifications of international human rights treaties shows that a significant number of states ratify at a time when their human rights ratings are relatively low an indication that they are not likely already in complete compliance with treaty terms. As Table 1 below shows, more than half of the states that ratified the Convention against Torture did so when their human rights ratings were moderate or poor (0 represents the weakest protection for human rights and 8 the strongest protection). Accordingly, while some states may ratify because they also can comply, the evidence suggests that even some late joiners ratify notwithstanding indications that immediate compliance is unlikely. Given this pattern, we focus on one particular way in which the actions by the extant membership of an international human rights treaty can alter the de facto treaty terms and render them more flexible than they are according to the institutions text. Specifically, states joining a treaty can file reservations at the time of their commitment to limit their treaty obligations. 8

According to the Vienna Convention on the Law of Treaties, a reservation is a unilateral statement, however phrased or named, made by a State, when signing, ratifying, accepting, approving or acceding to a treaty, whereby it purports to exclude or to modify the legal effect of certain provisions of the treaty in their application to that State.1 Human Rights Rating 0 1 2 3 4 5 6 7 8 Frequency 5 4 8 18 18 15 15 18 23 Percentage 4.03 3.23 6.45 14.52 14.52 12.10 12.10 14.52 18.55 Cumulative 4.03 7.26 13.71 28.23 42.74 54.84 66.94 81.45 100.00

Table 1: Human Rights Rating of States during Year Ratified CAT Treaty In the human rights context, such reservations are somewhat common. Neumayer (2007), for example, suggests that states file more reservations to human rights treaties than to any other type of international treaty. Whether the existence of a high number of reservations is desirable or not is the subject of some debate, however. Permitting reservations may encourage more widespread commitment to international human rights treaties because reservations allow states to commit to treaties in a more qualified fashion. Furthermore, some scholars argue that although reservations allow states to limit their obligations to fully protect individual human rights, the reservations are nevertheless helpful in providing information about the precise extent to which a state intends to abide by a particular treatys terms. In fact, some scholars suggest that liberal
1

See Article 2(1) of the Vienna Convention on the Law of Treaties. States can also file declarations and understandings as a method by which to indicate their interpretation of treaty terms or how and whether they intend to comply with particular provisions. As discussed more fully in the section describing reservations as an independent variable of interest, we include declarations and understandings within our definition of reservations to the extent a state used those devices to similarly limit its treaty participation or exempt itself from some or all of the treatys obligations.

democracies that embrace the rule of law are actually more likely to file reservations because they want to be clear about their obligations so as not to run afoul of them (Simmons 2009). Others are more skeptical and view reservations as undermining the legitimacy of human rights treaties, since by making reservations states are essentially opting out of what were supposed to be universal obligations to protect (Neumayer 2007). For example, some autocracies or other states with weak protection of human rights may file significant reservations not to clarify their obligations, but rather to allow them to commit without actually having to alter their behavior to conform to treaty terms. For our argument, though, we need not resolve this debate about whether permitting states to file reservations is desirable or not as a general matter because our focus is on how reservations by members can change non-members understanding of how constraining or flexible a human rights treaty is. First, the very fact that a treaty allows for reservations is some indication that the treaty terms may be more flexible than they appear on paper. Furthermore, the extent to which states actually file reservations can provide even more information to nonmember states about treaty flexibility because by filing reservations states actually indicate their intent to reduce, limit, or modify the obligations they would otherwise have to accept according to treaty terms. As a result of observing this increased flexibility in practice, even states with records less likely to make them able to immediately comply with the terms of international human rights treaties will choose to commit. This evidence of increased flexibility can provide the insurance against uncertainty that some states need in order to make a commitment they would otherwise be wary of making before they have actually instituted all of the policies and changes necessary to fully comply with treaty terms. 10

Not all reservations are created equal, however. Depending on the type of state that files reservations, the type of flexibility the reservations imply may be different. The difference is mainly a function of how likely a reservation-filing state will be able to comply with the provisions of a human rights treaty. Reservations by democracieswhich are more likely to respect and protect human rightswill provide information about how precisely the treaty will function and merely narrow the scope and applicability of provisions. By contrast, reservations by autocracieswhich are less likely to respect and protect human rightswill provide information about how states can undermine and weaken the institution. Some scholars argue that states with weak protection of human rights will simply take advantage of this flexibilityinterpreted at its extreme as weak enforcementwithout any intention of complying. States may obtain some tangible or intangible benefit, such as aid or trade, in exchange for ratifying, but fail to subsequently comply with treaty terms (Simmons 2009). We, however, suggest that some states with weaker domestic protection of human rights may actually use evidence of flexibility as a mechanism to sincerely commit to meaningful change. At the time the various international human rights treaties were negotiated, states recognized that the codified norms had not yet been universally adopted by all states. Yet, part of the purpose of allowing reservations was to make the treaties more flexible so as to encourage universal participation (Swaine 2006). Furthermore, states likely recognized that if they were encouraging all states to join, some may need more time than others to make the necessary changes to domestic law and practice. There would seem little reason to move forward with negotiating a series of international human rights treaties if none of the states with worse practices indicated any interest in improving their practices.

11

Hypotheses This implies that different regime types not only generate different information through the reservations they file, but also that they respond differently to the information generated by the reservations. First, in general, states with better practices should be likely to commit to international human rights treaties because ratification will not pose a significant risk to their sovereignty. With respect to reservations, states with strong protection of human rights democraciesmight not need signs of increased flexibility to facilitate ratification. On the other hand, signs of a weakened institution as a result of reservations filed by weak protectors of human rights can make democracies more hesitant to ratify. Democracies prefer to not join institutions that are inconsistent with their preferences and valuesones significantly weakened by autocracies. For states with strong protection of human rights, compliance is something of a foregone conclusion because treaty commitment will require no changes or only minimal changes to the states already compliant domestic policies and practices. Accordingly, these states commitment to a treaty will have little influence on their future behavior. States with moderate or weaker protection of domestic human rights practices, by contrast, should respond to the lure of increased institutional flexibility. These states should initially be wary of committing to an institution that could punish them for bad and noncompliant behavior. With evidence that the institution is at least somewhat flexible, however, the costs of committing to the institution become lower. In terms of reservations, the strongest indication of an institutions weakening comes from autocracies reservations. When such weak

protectors of human rights file reservations with a human rights treaty, other states with weak protection will be more likely to ratify the treaty. Moreover, once they commit, these weak protectors can use the institutions constraintseven with increased flexibilityto improve their future behavior. Consistent with the literature suggesting that states with poorer prospects for treaty compliance will seek out international mechanisms for domestic constraint (Simmons 2000; Simmons and Danner 2009; Moravcsik 2000), we hypothesize that states with weaker domestic protection of human rights will use evidence of increased flexibility first as an opportunity to join the treaty and subsequently as a mechanism to constrain and improve their practices. IV. Empirical Analyses In this section we carry out an analysis of the hypotheses we advanced in the previous section using novel data on reservations states made at the time they ratified a human rights treaty. The human rights institution against which we test our argument is the Convention against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment, frequently referred to simply as the Convention against Torture or the CAT. While there are five other major international human rights treaties to which we expect to expand our research in the future, the CAT provides an attractive institution against which to initially test our hypotheses. First, the target behavior the CAT seeks to regulate is quite clearly defined and essentially more easily measurable than that of other human rights treaties. For example, a states level of compliance with treaties protecting social and cultural rights (International Covenant on Economic, Social and Cultural Rights, ICESCR), womens rights (Convention on the Elimination of Discrimination against Women, CEDAW), childrens rights

(Convention on the Rights of the Child, CRC), and against racial discrimination (Convention on the Elimination of Racial Discrimination, CERD), are more difficult to measure systematically. Second, the time period covered by the CAT (1984 to present) facilitates analysis by making more data available. The two main covenantsthe International Covenant on Civil and Political Rights (ICCPR) along with ICESCRin contrast, were available for ratification in the 1960s. Lastly, for these and other reasons, the CAT has received a lot of attention from scholars of empirical research in recent years (Goodliffe and Hawkins 2006; Vreeland 2008; Hathaway 2004; Hollyer and Rosendorff 2011). This paper contributes to that growing body of literature. Research Design and Data Dependent Variables The analysis proceeds in two main steps. First, we examine the determinants of the timing of states ratification of/accession to the CAT. Second, we analyze states level of human rights protection subsequent to their ratification compared to the level prior to their ratification and also to those states that do not ratify the CAT. Given these two steps, we model two separate dependent variables. For the timing analysis, the outcome of interest is how long states wait before ratifying the human rights treaty, and the variable is each states time until ratification. This variable codes the number of years until states ratify the CAT, starting with the year of their independence or the year that the CAT became available for ratification (1984), whichever is later. In the sample, the minimum of this variable is one and the maximum is 27, with a mean of 19.8 years.

For the analysis of human rights protection within each state, we employ the widely used physical integrity rights index from the Cingranelli-Richards (CIRI) Human Rights Dataset.2 The index consists of four sub-indices on torture, extrajudicial killing, political imprisonment and disappearance. This index provides a good measure of the rights protected by the CAT because the index considers the very types of human rights violations prohibited according to the terms of the treaty.3 Each of these sub-indices ranges from zero to two, with higher numbers indicating stronger and better protection of human rights at home. Accordingly, the physical integrity index ranges from zero to eight. In the sample, the variable varies across the entire range, with a mean of 4.9. In the robustness check of the analysis, we also examine the four sub-indices. Explanatory Variable We model these two dependent variables as a function of an explanatory variable. The variable is the number of states filing reservations, understandings, or declarations along with their ratification. Article 2(1)(d) of the Vienna Convention on the Law of Treaties defines a reservation as a unilateral statement, however phrased or named, made by a State, when signing, ratifying, accepting, approving or acceding to a treaty, whereby it purports to exclude or to modify the legal effect of certain provisions of the treaty in their application to that State. In short a reservation limits the legal obligation for the state filing it. By international law, unless the treaty provides otherwise, states may file reservations to the treaties they ratify. While the Vienna Convention limits the extent of a reservation to that which is not incompatible with a
2 David L. Cingranelli and David L. Richards, The Cingranelli-Richards Human Rights Dataset Version 2008:03.12, available at http://www.humanrightsdata.org. 3 According to Article I of the CAT, torture is defined as any act by which severe pain or suffering, whether physical or mental, is intentionally inflicted on a person for such purposes as obtaining from him or a third person information or a confession, punishing him for an act he or a third person has committed or is suspected of having committed, or intimidating or coercing him or a third person, or for any reason based on discrimination of any kind, when such pain or suffering is inflicted by or at the instigation of or with the consent or acquiescence of a public official or other person acting in an official capacity . . ..

treatys object and purpose (Article 19(c)), states frequently file significant reservations with their ratifications.4 Following Neumayer (2007), we group together understandings and declarations with reservations, as their practical effects on states legal obligation are the same. To facilitate the reference, we refer to the three legal measuresreservations, understandings and declarationssimply as reservations in the remainder of the paper.5 We code several aspects surrounding these reservations. The first two measures code new reservations that states file at the time of their ratification. One variable counts the total number of reservations that states file in a year. The other variable counts the number of states that file reservations in a year. In years that no state ratified the convention, we take the value from the preceding year with a non-missing value, as that captures the latest number of ratifying states that filed reservations. The number of reservation-filing states ranges between zero and six with a mean of 1.8 states per year. In some years, as many as six ratifying states file reservations. In nine years, no states file reservations. The number of reservations ranges between zero and twelve with a mean of 3.3 reservations per year. In 1988, states filed a total of 12 reservations. In nine years, zero reservations are filed. The second set of measures codes information about all the reservations that are in place in a year. The first of these measures captures the level of autocracy of states filing reservations. Using the Polity measure of regime type, we capture the average level of autocracy of all states that have reservations on file. The higher measure indicates that the reservations on record in a year have been filed by states that are more autocratic than democratic. The second measure combines this first measure with a measure of the severity of reservations states filed. Each
4

For example, at least one state filed a reservation to the CAT stating that it reserved the right to apply Sharia law over the law outlined in the CAT. 5 Empirically, there are no declarations filed for the CAT.

reservation is coded on a scale between zero and three on how much they undermine a ratifying states obligations under the treaty. This combined variable therefore weights more severe reservations heavier than less severe ones and reservations by autocracies heavier than the ones by democracies. The average regime type of states that have filed reservations ranges between 4.6 and 11.7 with a mean of 8.0. The same measure weighted by the severity of reservations ranges between 8.7 and 23.3 with a mean of 14.3. Control Variables In addition to this main explanatory variable, we include a set of control variables in the analysis. Two economic variableslogged GDP per capita and logged total GDPcapture states propensity to join human rights institutions. States that are economically more developed are more likely to have better protection of human rights at home, which means they may be more likely to ratify human rights treaties because compliance is a foregone conclusion. On the other hand, decreasing the likelihood that these states will ratify is the fact that they may not need an external commitment mechanism. Moreover, economically larger states may be more capable of resisting international pressures for joining a human rights treaties or improving human rights protection at home. We also incorporate two additional variables that even more directly capture states protection of human rights. We use the CIRI Physical Integrity measure to control for states extant level of human rights protection. The other variable is the Polity measure of regime type. Protection of human rights is consistent with democratic values, and democracies are better at protecting them than autocracies. Similar to the level of development, this relationship may either increase or decrease the likelihood of ratifying the CAT.

Lastly, we include a measure of the proportion of states in a states region that has ratified the CAT. This measure captures the competition or diffusion effect from neighboring states, which can lead states to ratify a treaty that many of their neighbors have ratified or to increase their compliance with the treaty once they ratify it (Goodliffe and Hawkins 2006; Wotipka and Tsutsui 2008). Analysis and Results To test the timing-of-ratification hypothesis, we carry out an event history analysis using this data. The analysis examines how the covariates contribute to the survival of each state until they ratify the CAT. Each state becomes at risk of ratification either once the CAT becomes available for ratification or when the state gains independence, whichever event is later. Then, the analysis tracks how long the state survives before ratifying the treaty. To analyze the dynamics, we use Cox proportional hazard model, whichin addition to being a less restrictive model than the parametric ones requiring assumptions about the underlying distribution of the error termcan accommodate time-varying covariates (Allison 1984). The sample for this analysis contains 141 states, for which the relevant data exists. 109 of those states ratified the CAT at some point and the analysis time period ending in 2008, with the remaining 32 states not having ratified the convention by that time. Figure 1 plots the KaplanMeier survival estimates of states ratification dynamics over time. States slowly ratify the convention over time, and by the end of the analysis period, only 25% of the states in the sample remain not having ratified it. A test of the Schoenfeld residuals shows that two variables violate the proportional hazard assumption under some specifications. Accordingly, we specify these variableslogged

total GDP and logged GDP per capitaas time-varying covariates, which then interacts them with the analysis time. However, because the results are substantively the same as the model without the variables, we present the results from the models without the variables specified as time-varying.

0.00
0

0.25

Proportion of States 0.50 0.75

1.00

10 15 Time until Ratification (years)

20

25

Figure 1: Time until Ratification Because we expect states with different levels of human rights protection to respond differently to the perceived level of flexibility in the institution, we divide the sample into two groups of statesthose with weak protection of human rights and those with strong protection of human rightsusing states physical integrity score at the beginning of the analysis period. 6 Because states with the highest levels of protection tend to be substantively different from those
6 We also do not use stratification to divide the sample because our argument implies that the three groups of states will differ not only on their baseline hazard rate, but also that the explanatory factors have different effects on the hazard rates.

with lower levels, the strong protection group consists of states with the highest three scores (8, 7 or 6) and the weak protection group consists of the rest with scores of six and below.

Weak Protectors

Strong Protectors

0.00
0

Proportion of States 0.25 0.50 0.75

1.00

10 15 Time until Ratification (years)

20

25

Figure 2: Time until Ratification, by Level of Human Rights Protection

Figure 2 replicates Figure 1 with the sample divided into the two groups. Although no notable differences between the groups exist during the first ten years or so of the time that states have an opportunity to ratify the treaty, the strong protectors rate of ratification slows down relative to that of weaker protectors starting in year 12. Variation in Human Rights Protection and CAT Ratification Using the distinction between the two groups, we model states choice of timing in when they ratify the CAT. In the first pair of models (1a and 1b), we use the count of countries

reservations as the main explanatory variable. In the second pair of models (2a and 2b), we use the count of countries filing reservations. Table 2 summarizes the results. The reported estimates are hazard ratios, which are interpreted as percentage increase and decrease in the likelihood of ratification compared to the baseline likelihood. The gray shading highlights the results of interest. The results from all four models are consistent with the hypotheses generated above. For weak protectors of human rights, each unit increase in the count of reservations increases the risk of ratification by 20% (hazard ratio of 1.2). By contrast, for strong protectors of human rights, each unit increase decreases the risk of ratification by ratification (hazard ratio of 0.8). Both estimates are statistically significant at the 0.01 level. The effects are even larger in Models 2a and 2b. Each unit increase in the count of states filing reservations increases the risk of ratification by 30% (hazard ratio of 1.3) for the weak protector group but decrease the risk by 40% (hazard ratio of 0.6) for the strong protector group. The results from the models also show some systematic differences between the two groups. Per capita GDP only affects the weak protectors positively, i.e. shortens the time until ratification. Total GDP, by contrast, only affects the strong protectors positively. The Polity score measuring a states level of democracy only affects strong protectors, and states existing level of human rights protection only affects weak protectors. Regional ratification is the only variable that affects both groups, although the effect is much larger for the weak protectors. That states human rights ratings have significant effects on CAT ratification, albeit limited to the weak protector group, is at odds with some findings in the literature. Goodliffe and Hawkins (2006) find that a states physical integrity score does not systematically affect CAT

10

ratifications. More generally, Dutton (2012) shows that states ratification of human rights institutions with relatively weak enforcement mechanisms, such as the CAT, bears no relationship to their level of human rights practices. Cole (2005) finds the same with respect to the ICCPR and ICESCR. Our contrasting result arises from the separation of states by their level of human rights protection. In the sample combining the two groups, the effect of states existing physical integrity score no longer exists. Weak HR Strong HR Protector Protector Model 1a Model 1b 1.2*** 0.8*** (0.06) (0.04) 1.4*** (0.1) 1.1 (0.1) 1.0 (0.02) 19.4** (1.2) 1.2** (0.07) 719 67 52 0.8 (0.2) 1.3** (0.1) 1.1* (0.03) 4.9** (0.8) 1.0 (0.1) 642 52 39 Weak HR Protector Model 2a 1.3*** (0.10) 1.4*** (0.1) 1.1 (0.1) 1.0 (0.02) 20.2** (1.2) 1.2** (0.07) 719 67 52 Strong HR Protector Model 2b 0.6*** (0.09) 0.8 (0.2) 1.3** (0.1) 1.1* (0.03) 4.8** (0.8) 1.0 (0.1) 642 52 39

Reservation Count Reservation Country Count Per Capita GDP (logged) Total GDP (logged) Polity Regional Ratification Physical Integrity N States Failures

Standard errors clustered on states in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 2: Models of CAT Ratification Timing, by Levels of Human Rights In the next models, we look at the effects of the characteristics of reservations, going beyond the simple counts. In Models 3a and 3b, we capture the regime type of states making reservations. As discussed above, the measure is increasing in autocracy. In Models 4a and 4b, we weight this measure of reserving state regime type by the severity of the reservations. Table 3 summarizes the results. 11

Reserving Country Regime Type Severity-Weighted Regime Type Per Capita GDP (logged) Total GDP (logged) Polity Regional Ratification Physical Integrity N States Failures

Weak HR Protector Model 3a 0.9 (0.06) 1.4*** (0.1) 1.1 (0.1) 1.0 (0.02) 18.3** (1.2) 1.2** (0.07) 660 67 52

Strong HR Protector Model 3b 0.8*** (0.06) 0.8 (0.2) 1.3** (0.1) 1.1* (0.03) 5.0** (0.8) 1.0 (0.1) 592 52 39

Weak HR Protector Model 4a 1.0 (0.03) 1.4*** (0.1) 1.1 (0.1) 1.0 (0.02) 18.2** (1.2) 1.2** (0.07) 660 67 52

Strong HR Protector Model 4b 0.9*** (0.03) 0.8 (0.2) 1.3** (0.1) 1.1* (0.03) 5.0** (0.8) 1.0 (0.1) 592 52 39

Standard errors clustered on states in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 3: Models of CAT Ratification Timing, by Levels of Human Rights The control variables have similar effects as in the previous models. In Model 3a, the main explanatory variable does not have a significant effect. In Model 3b, however, the estimate is statistically significant. Each unit increase in the reserving country regime type leads to a 20% reduction in the risk of ratification by the strong protector group. In other words, when more autocratic states file reservations to the CAT than democratic states, the ratification behavior of weak protectors is not affected. However, the presence of more reservations by autocratic states does make the strong protectors more hesitant to ratify the treaty and delays their ratification. Although the substantive effect is smaller, the results from Models 4a and 4b are consistent with those from Models 3a and 3b. The main explanatory variablethe regime type measure weighted by the severitydoes not systematically affect the ratification behavior by

12

weak protectors. It does, however, reduce strong protectors risk of ratifying the Convention by 10%. Implications for Reservations Filing The proposed argument and the analysis so far suggest a testable implication. States that are weak protectors of human rights and benefit from increased flexibility at the institution level are also likely to benefit from having flexibility measures that are specific to themselves. Accordingly, these states should 1) themselves have a higher likelihood of filing reservations than states that are strong protectors of human rights and 2) when they do, file more reservations than strong protectors of human rights. We test these implications using the same underlying data as in the models above. Models 5a and 5b examine whether or not states file any reservations at the time that they ratify the CAT. We use the count of reservations states file in a yearthe variable used in Models 1a and 1bas the measure tracking reservations at the institution level. Table 4 summarizes the results from Models 5a and 5b, which use logit to estimate the factors influencing states likelihood of filing reservations. As the results show, reservation count is significant for both groups of states, weak and strong protectors, although the estimate is larger and statistically stronger for the weak protectors. The coefficient estimates alone, however, are difficult to interpret substantively. The difference between the two groups in the substantive effects of the reservation count in terms of how it affects states probability of ratification is more striking. Among the strong protectors, the difference in the predicted probabilities of filing reservations resulting from the count of institution-wide reservations increasing from its 25th-percentile value to its 75th-percentile value 13

is statistically indistinguishable.7 By contrast, among the weak protectors, the difference in predicted probabilities resulting from the same interquartile change is large (0.31) and statistically significant. When the reservation count is at 25th-percentile, or 0, weak protectors probability of filing reservations is 0.07. When the reservation count is at 75th-percentile, or 4, weak protectors probability of filing reservations increases to 0.38. In short, not only are weaker protectors of human rights more likely to ratify the CAT when flexibility at the institution is high, they are more likely to piggyback on those opportunities by becoming more likely to file reservations themselves when they ratify. Weak HR Protectors Model 5a 0.5*** (0.2) -0.2 (0.5) 1.1*** (0.4) -0.2** (0.08) 8.8*** (3.2) 0.08 (0.2) -31.4*** (9.7) 53 Strong HR Protectors Model 5b 0.3* (0.2) -0.3 (0.8) 0.1 (0.4) 0.005 (0.1) 3.1 (2.1) 0.3 (0.5) -6.9 (7.0) 38

Reservations Count t Per Capita GDPt-1 Total GDP t-1 Polity t-1 Regional Ratification t-1 Physical Integrity t-1 Constant N

Standard errors clustered on states in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 4: Models of Reservations Filing, by Levels of Human Rights Models 6a and 6b test the natural implication that the same dynamics should influence not only states likelihood of filing reservations but also the number of reservations they file. We

7 The predicted probabilities are generated using Clarify, holding all the other variables at their mean. See King, Tomz, and Wittenberg 2000).

14

use negative binomial regression to model these countsvarying from zero to fivebecause they exhibit overdispersion. The standard errors are clustered on states. Table 5 summarizes the results. Weak HR Protectors Model 6a 0.1** (0.06) 0.2 (0.2) 0.5*** (0.2) -0.08*** (0.03) 2.9** (1.2) -0.03 (0.1) -16.1*** (4.1) 53 Strong HR Protectors Model 6b 0.2* (0.1) -0.3 (0.7) 0.3 (0.3) -0.03 (0.09) 3.3** (1.6) 0.3 (0.4) -8.9* (5.0) 38

Reservations Count t Per Capita GDP t-1 Total GDP t-1 Polity t-1 Regional Ratification t-1 Physical Integrity t-1 Constant N

Standard errors clustered on states in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 5: Models of CAT Reservation Counts, by Levels of Human Rights Similar to the analysis above, we generate the first difference of expected values of the count of reservations for moving the count of reservations from its 25th-percentile value to its 75th-percentile value. These first differences are statistically insignificant for states with strong protection of human rights. By contrast, the difference is statistically significant for states with weak human rights protection at the 95% level. When the number of states in the CAT filing reservations increases from its 25th-percentile value (0) to its 75th-percentile value (2), the expected value of reservations filed by weak protectors increases by .17, from .16 to .33. Weak protectors of human rights file more reservations upon ratification when the number of reservations other states file is high. 15

CAT Ratification and Human Rights Protection The analysis so far demonstrates that evolving institutional flexibility increases states hazard of ratifying the CAT and that such flexibility affects weak human rights states more than strong human rights states. A key premise of our argument is that some subset of states with poor human rights protection seeks to strengthen their human rights protection for various domestic political reasons. These states, once in the institution with the level of flexibility optimal for themselves, will use the institution to constrain their behavior and improve their domestic protection of human rights. We test this hypothesis in this section. The dependent variable is now the physical integrity score. Given that the variable is an ordinal index, we use ordered probit to analyze the dynamics. In the following analysis, we divide the sample into three groupsstrong, moderate and weak protectors of human rights whereas the analysis so far has combined the moderate and weak protectors into a single group. Analysis reveals that the weak and the moderate protectors behave substantially differently, and combining the groups masks the effects of the CAT as well as the difference between the groups. A new variable added to the analysis is an indicator marking the period when the states in the sample are under the CAT. The variable is zero until the year that a state ratifies the treaty and then is one thereafter. The hypothesis expects that this variable will be positive and significant for states with weak human rights protection. The standard errors are clustered on states. Table 6 summarizes the results. The indicator measure for states CAT commitment is statistically insignificant in Models 7b and 7c. Moderate and strong protectors of human rights do not behave any differently when they are under the CAT commitment and when they are not. By contrast, the physical integrity 16

score of poor human rights protectors improves after these states have ratified the CAT and are under the commitment. For this group of states, the predicted probability of having a low physical integrity score (0, 1 or 2) falls once they are in the institution. However, the predicted probability of having a moderate physical integrity score (3, 4 or 5) increases once these poor protectors of human rights join the institution. The increase in the probabilities is by .031, .044 and .030, respectively. While the predicted probability of having a high physical integrity score (6, 7 or 8) increases as well, the magnitude of the increases is much smaller. In short, states that start out as poor human rights protectors at the beginning of the analysis period increase the probability of becoming moderate human rights protectors after they ratify the CAT. Weak Moderate Strong HR Protector HR Protector HR Protector Model 7a Model 7b Model 7c 0.3*** 0.03 0.01 (0.1) (0.09) (0.2) 0.2*** 0.2*** 0.4*** (0.06) (0.06) (0.05) 0.2*** 0.2*** 0.2*** (0.04) (0.03) (0.03) 0.004 0.02*** 0.03*** (0.009) (0.007) (0.008) 0.6*** (0.04) 0.2 (0.2) 513 0.5*** (0.04) 0.1 (0.1) 1102 0.6*** (0.04) 0.03 (0.2) 1280

Under CAT Commitment t-1 Per capita GDP t-1 Total GDP t-1 Polity t-1 Physical Integrity t-1 Regional Ratification t-1 N

Standard errors clustered on states in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 6: Models of Human Rights Protection and CAT Commitment, by Levels of Human Rights Consistent with the results so far, the next tests show that reservationsflexibility measuresstates carve out for themselves help poor protectors improve their human rights protection. Specifically, Models 8a, 8b and 8c test whether having filed any reservations affects 17

states level of human rights for the three groups of states. The sample for each model is limited to states that have ratified the CAT. The main explanatory variable is an indicator measure of whether a state has filed any reservations to the convention. Table 7 summarizes the results. Weak Moderate Strong HR Protector HR Protector HR Protector Model 8a Model 8b Model 8c 0.3*** 0.03 0.01 (0.1) (0.09) (0.2) 0.2*** 0.2*** 0.4*** (0.06) (0.06) (0.05) 0.2*** 0.2*** 0.2*** (0.04) (0.03) (0.03) 0.004 0.02*** 0.03*** (0.009) (0.007) (0.008) 0.6*** 0.5*** 0.6*** (0.04) (0.04) (0.04) 0.2 0.1 0.03 (0.2) (0.1) (0.2) 281 615 628

Filed Reservations t-1 Per capita GDP t-1 Total GDP t-1 Polity t-1 Physical Integrity t-1 Regional Ratification t-1 N

Standard errors clustered on states in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 7: Models of Human Rights Protection and Reservations, by Levels of Human Rights Whereas having reservations does not affect human rights protection of strong and moderate protectors that are under a commitment to the CAT, it does positively affect poor protectors. Similar to the results above, having reservations reduces poor protectors probability of having low protection and increases the probability of having moderate protection. While we do not report the results due to space constraints, parallel tests using the count of reservations as the explanatory variable generates the same substantive results. Poor protector states with more reservations increase (decrease) their probability of improving (deteriorating) human rights protection more than those states with fewer reservations. By contrast, the count of reservations states file does not affect moderate or strong protectors of human rights.

18

These results are broadly consistent with what Simmons (2009) finds, although her argument is substantially different. She argues that transitioning countries that have ratified the CAT are more likely to improve their human rights practices than those that have not. While the results of Model 7a can be consistent with Simmons (2009) argument, those of Models 8a8c suggest that flexibility is, in fact, important for poor human rights protectors. These results increase the confidence in our argument that flexibility both at the institution level and at the state-specific level can facilitate compliance and improved human rights practices in weak protectors. It is true that any single set of results we present here do not explicitly demonstrate the independent effect of the human rights institution. Taken together, however, they suggest the simultaneous screening and constraining effects of the CAT, as anticipated by our argument. The strong protectors of human rights demonstrate the screening effect. These states already have strong protection of human rights and presumably care about maintaining and spreading strong human rights protection. These states become less likely to ratify the CAT when they notice a weakening of the institution through a large number of reservations. Other states filing reservations does not lead these states to file reservations as well. Lastly, these states do not improve their human rights protection while they are under the CAT, likely because they already have a high level of human rights protection. For related reasons, the reservations these states file do not systematically influence their behavior. On the other hand, the poor protectors of human rights demonstrate the constraining effect. The main critique against the results of Model 7a is likely that these states chose to improve their human rights protection and would have improved even in the absence of their commitment to the CAT. We agree that these states likely had exogenous motivation for seeking 19

to improve human rights protection at home. However, we disagree that they would have made the same improvements in the absence of the CAT. While the models of reservation filing and the effects of reservations could be consistent with the argument that the institution had no independent effect, the timing of poor protectors ratification is not. If the CAT did not impose any constraints on poor protectors motivated to improve, then these states should not have been affected by the number of states filing reservations with their ratification. Moreover, if the institution did not have an independent effect, then the reservations should in fact not have affected the level of poor protectors human rights. The Case of Guatemala In fact, an examination of Guatemalas CAT ratification behavior provides support for the idea that some states do respond to institutional flexibility and also use institutional constraints to improve domestic human rights practices. Historically, Guatemala regularly experienced civil guerilla warfare and was ruled by military dictatorships from about 1970 until the election of civilian President Cerezo in late 1985. Cerezo promised to instill democratic rule, and it was during the last year of his presidency in January 1990 that the country committed to the CATwhen the country had a human rights rating of 1. Change did not come immediately, and the country continued to be plagued by violence. But, successor presidents also promised a commitment to democratic reforms, and over time, Guatemalas human rights ratings improved. By 2003, it was scoring a 5 or 6 on the CIRI scale. First, as to the role that institutional flexibility can play in motivating states with poor initial prospects for compliance to commit to international human rights treaties, Guatemalas own ratification behavior is illustrative. It filed two reservations upon ratification limiting its treaty obligations. One reservation exempted it from committee competence to investigate with 20

state cooperation any reliable complaints that the country was systematically practicing torture. The other exempted it from a mechanism to resolve disputes between state parties to the CAT concerning the interpretation of treaty terms. Guatemala later removed those reservations on May 30, 1990. Nevertheless, by filing the reservations initially, Guatemala obtained greater state-specific flexibility concerning the application of CAT treaty terms to its behavior, at least for a time. The evidence also shows that although Guatemala evidently had exogenous reasons for seeking to improve its human rights practices, it also used the CATs constraintswhich require reports to a committeeto help bind the leadership to making those improvements. For example, in response to reiterated recommendations by the CAT Committee, Guatemala: (1) began instituting training programs for prosecutors and police ,8 (2) ratified CAT Article 21 in 2003, allowing individuals to bring complaints against it alleging their rights under the treaty have been violated;9 and (3) began amending its laws to criminalize torture consistent with the CAT definition. Indeed, Guatemalas own statements suggest it was using the reporting process to assist it improving its practices. It stated that it regards the submission of reports to committees as a valuable opportunity to reflect on progress in and obstacles and challenges to the proper exercise of human rights in Guatemala. V. Conclusion Human rights treaties constitute a major category of institutions at the international level. Since the signing of the Universal Declaration of Human Rights, idealistic actors around the world have sought to improve the protection of human rights by pursuing formal, legal instruments. Contrary to the large and growing number of these institutions at the multilateral
8 9

CAT/C/74/Add.1, 27 May 2005, 8 and 28-30. CAT/C/74/Add.1, 27 May 2005, 8 and 33.

21

and regional levels, however, scholars have not consistently found patterns of ratification and compliance to support the hopes of idealists. In this paper, we have argued for a more dynamic understanding of human rights institutions. In particular, we argued that these institutions can change over time as a function of how members in them behave, which in turn can affect whether and when non-party states join the institutions. Empirically, we focused on the reservations that states filed as a measure of such evolving flexibility at the institutional level. Our analysis showed that states that are poor protectors of human rights ratified the CAT when more states filed reservations upon their ratification. Moreover, our analysis showed that these poor protectors of human rights improved their practices subsequent to ratifying the CAT, especially when they carved out some flexibility for themselves through reservations. The argument and the findings of this research have much broader implications for the study of international institutions beyond human rights treaties. In analyzing institutions, scholars have very much focused on the static and formal aspects of institutionsthe text and provisions of agreements and treaties. Although the institutionalist literature centered on cooperation theory has made a significant progress over the last three decades, the progress has come at the cost of a more complete understanding of how institutions develop and function over time. Consistent with how sociological institutionalists and constructivists view institutions, actors understanding of and expectations about institutions may change over time. Incorporating this perspective into other research on institutions will likely make a significant contribution to the scholarly debate.

22

Moreover, the argument and the findings demonstrate that the screening and the constraining effects of international institutions are not mutually exclusive. Conditional on how institutions dynamically change over time, states can balance between flexibility and constraint, thereby utilizing institutions to best serve their domestic and international political objectives. This implication moves the debate on institutional effects beyond the current one and warrants further analysis in the future.

23

REFERENCES Abbott, K. W. and D. Snidal, 1998, Why States Act through Formal International Organizations. The Journal of Conflict Resolution 42, 3-32. Abbott, K. W. and D. Snidal, 2000, Hard and Soft Law in International Governance. International Organization 54, 421-456. Allison, P., 1984, Event History Analysis: Regression for Longitudinal Event Data (Sage). Chayes, A. and A.H. Chayes, 1993, On Compliance, International Organization 47:175-205. Cingranelli, D.L. and D.L. Richards, 1999, Measuring the Level, Pattern, and Sequence of Government Respect for Physical Integrity Rights, International Studies Quarterly 43(2): 40718. Cole, W.M. 2005, Sovereignty Relinquished? Explaining Commitment to the International Human Rights Covenants 1966-1999. American Sociological Review 70 (3), 472-95. Downs, G. W., D. M. Rocke and P. M. Barsoom, 1996, Is the Good News about Compliance Good News about Cooperation? International Organization 50, 379-406. Dutton, Y., Commitment to International Human Rights Treaties: The Role of Enforcement Mechanisms, University of Pennsylvania Journal of International Law 34 (forthcoming Fall 2012). Goodliffe, J. and D.G. Hawkins, Explaining Commitment: States and the Convention against Torture, The Journal of Politics 68(2): 358-371. Haftner-Burton, E.M. and K. Tsutsui, 2004, Human Rights in a Globalizing World: The Paradox of Empty Promises, American Journal of Sociology 110(5):1373-1411. Hathaway, O.A., 2002, Do Human Rights Treaties Make a Difference, Yale Law Journal 111:1935-2042. _______________, 2003, The Cost of Commitment. Stanford Law Review 55, 1821-62. _______________, 2004, The Promise and Limits of the International Law of Torture, in Torture: A Collection, edited by S. Levison. Hillebrecht, C., 2011, The Legalization and Bureaucratization of the UN Human Rights System: Progress or Peril? Paper prepared for the 2011 Annual Meeting of the American Political Science Association, Seattle, WA. Hollyer, J.R. and B. P. Rosendorff, 2011, Why Do Authoritarian Regimes Sign the Convention Against Torture? Signaling, Domestic Politics and Non-Compliance, Unpublished Paper (Version June 2011). 24

King, G., Tomz, M. and J. Wittenberg, 2000, Making the Most of Statistical Analyses: Improving Interpretation and Presentation. American Journal of Political Science 44(2): 34761. Koremenos, B., 2005, Contracting around International Uncertainty. American Political Science Review 99(4), 549-65. Marshall, M.G., and Jaggers, K., 2007. Polity IV Project: Dataset Users Manual. Moravcsik, A., 2000, The Origins of Human Rights Regimes: Democratic Delegation in Postwar Europe. International Organization 54, 217-252. Neumayer, E., 2005, Do International Human Rights Treaties Improve Respect for Human Rights?, Journal of Conflict Resolution 49(6):925-953. ______________, 2007, Qualified Ratification: Explaining Reservations to International Human Rights Treaties. Journal of Legal Studies 36, 397-420. Simmons, B. A., 2000, International Law and State Behavior: Commitment and Compliance in International Monetary Affairs. The American Political Science Review 94, 819-835. ______________, 2009, Mobilizing for Human Rights: International Law in Domestic Politics, Cambridge University Press. Simmons, B.A., and Danner, A., 2010, Credible Commitments and the International Criminal Court. International Organization 64 (2):225-56. Swaine, E., 2006, Reserving, Yale Journal of International Law 307-365. Vreeland, R., 2008, Political Institutions and Human Rights: Why Dictatorships Enter into the United Nations Convention Against Torture, International Organization 62:65-101.

25