Human Rights
Maged
HollyerRosendorff.pdf
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
6 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 36 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 1 24-APR-12 10:00
DO HUMAN RIGHTS AGREEMENTS PROLONG THE TENURE OF AUTOCRATIC RATIFIERS?
JAMES R. HOLLYER* AND B. PETER ROSENDORFF**
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791 R I. THE WORST OFFENDERS ACCEDE MORE
FREQUENTLY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 793 R II. ACCESSION BY AUTOCRACIES HAS LITTLE EFFECT
ON HUMAN RIGHTS ABUSE IN THOSE COUNTRIES . 800 R III. IS REGIME CHANGE DELAYED? . . . . . . . . . . . . . . . . . . . . 802 R IV. ARE THE HUMAN RIGHTS TREATIES TO BLAME FOR
DELAYING REGIME CHANGE? . . . . . . . . . . . . . . . . . . . . . 806 R A. Commitment Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806 R B. Information Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807 R C. Selection Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809 R
V. IMPLICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810 R
INTRODUCTION
The international human rights regime has evolved into a complex and far-reaching set of international treaties that has transformed traditional notions of sovereignty. As Beth Sim- mons remarks in the introduction to her recent book, “it is no longer acceptable for a government to make sovereignty claims in defense of egregious rights abuses,” and the West- phalian defense appears to be severely compromised.1 States have voluntarily consented to these intrusions into their sover- eignty, and have built a variety of international institutions de- signed to monitor, report on, and even adjudicate cases of human rights abuses.
Simmons’s remarkable book demonstrates some real im- provements in the human rights conditions in some countries, following their accession to certain international human rights treaties. She restricts attention to transitional democracies— states with the preliminary trappings of democratic institu-
* MacMillan Center for International and Area Studies, Yale Univer- sity.
** Wilf Family Department of Politics, New York University. 1. BETH SIMMONS, MOBILIZING FOR HUMAN RIGHTS: INTERNATIONAL LAW
IN DOMESTIC POLITICS 3 (2009).
791
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
6 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 36 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 2 24-APR-12 10:00
792 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
tions; institutions that are young, unconsolidated and unsta- ble—rather than established autocracies or democracies. She finds statistically significant (if relatively small in magnitude) improvements in the human rights conditions in such transi- tional countries in the years following accession to some of the human rights treaties, relative to states that do not accede.
One can argue whether Simmons’ claim is causal—after all the counterfactuals are difficult to identify, and these states may have behaved this way absent the agreement.2 States that are democratizing, she argues, permit the political space for domestic political interests to be activated; to demand more accountability, to focus attention, and to mobilize support in favor of more human rights. This induces the states to accede where they would not have done so otherwise; and conditional on accession, improve their human rights performance. The mechanism is plausible and she marshals a fair amount of compelling evidence about some of the treaties. One is re- minded however, that at no point is Simmons declaring the victory of the human rights regime over tyranny. She acknowl- edges that the improvements she identifies are very small in magnitude, and she is modest in her claims.
We have no argument with Simmons when it comes to the young democracies. In this and other related work, we have focused our attention instead on autocracies.3 In autocratic regimes there are either no elections, or elections are not con- tested by rival political parties in which the opposition stands any chance of actually winning. This definition of “autocracy” is discussed at greater length by Cheibub, Gandhi, and Vree- land.4 We use their coding of regime-type in all of our subse- quent statistical analysis.
Within these toughest cases, we present here a number of disturbing observable regularities. First, some of these hard- core autocracies accede to some of these human rights trea- ties, and others do not. Moreover, statistical analyses reveal
2. States that sign these agreements could simply be the well-behaved states, or alternatively sign because they expect to be rewarded for doing so.
3. See, e.g., James R. Hollyer & B. Peter Rosendorff, Why Do Authoritarian Regimes Sign the Convention Against Torture? Signaling, Domestic Politics and Non-Compliance, 6 Q. J. POL. SCI. 275 (2011).
4. Jose Antonio Cheibub, Jennifer Gandhi & James Raymond Vree- land, Democracy and Dictatorship Revisited, 143 PUB. CHOICE 67, 75–76, 76 n.12 (2010).
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
7 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 37 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 3 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 793
that those regimes that have violated human rights in the past have a greater propensity to enter into (both sign and ratify) human rights agreements.5 Second, signing and ratification of these treaties has—at best—a small effect on human rights performance in those autocratic countries. These findings alone are unfortunate, if not perhaps unexpected. The en- forcement powers of the international human rights treaties and institutions are relatively weak, and states rarely punish each other for such violations. But there is a third, and poten- tially more worrying statistical regularity that we highlight. Those autocratic leaders who accede to human rights treaties survive longer in office than those that do not. The interna- tional human rights treaties are associated with delaying the change of leadership in these autocracies. And, since these are the most severe abusers of human rights, there exists an association between accession to the international human rights regime and the long-term survival of these worst abus- ers. Overall, the international human rights regime—de- signed to reduce torture, repression, and other violations of physical integrity—may be implicated in prolonging the ten- ure of the worst abusers of human rights.
In what follows, we offer some empirical evidence of those regularities and offer some interpretations of these findings. We then offer a possible explanation of these observations based on strategic decision making by autocrats in the midst of domestic political conflict over their potential survival.
I. THE WORST OFFENDERS ACCEDE MORE FREQUENTLY
Theories of accession to the human rights regime are rare. Moravcsik suggests that unstable democracies can “lock- in” human rights norms by treaty accession, which might con- strain future governments.6 This tying-of-the-hands argument is more fully explored by Simmons and Danner in their analy-
5. Compare for example, Chad and Congo, both signatories of the Con- vention Against Torture with Zimbabwe and Angola, which are not. Multilat- eral Treaties Deposited with the Secretary General, ch. IV(9), U.N. Doc. ST/LEG/ SER.E/26 (2009), available at http://treaties.un.org/doc/Publication/ MTDSG/Volume%20I/Chapter%20IV/IV-9.en.pdf. More systematic evi- dence for this claim is presented in the statistical analyses below.
6. Andrew Moravcsik, The Origins of Human Rights Regimes: Democratic Delegation in Postwar Europe, 54 INT’L ORG. 217, 220 (2000).
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
7 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 37 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 4 24-APR-12 10:00
794 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
sis of membership in the Treaty of Rome and the formation of the International Criminal Court.7 They argue that newly de- mocratizing governments find it difficult to credibly promise to reduce human rights abuse to their domestic opponents. Signing a treaty with external enforcement provisions en- hances the credibility of this promise. Hafner-Burton, Mans- field, and Pevehouse suggest that a state transitioning between an autocratic and democratic system will need to find ways to “commit” itself to democratic values, including human rights.8 Those institutions that impose harsh penalties for violations of human rights—those that involve high “sovereignty costs”— serve as a commitment mechanism that ensures newly-democ- ratizing states adhere to their obligations to their citizens. States that desire democratization, therefore, seek out such treaties to promote their efforts at reform. Many other schol- ars argue that international agreements may create openings for non-governmental actors to engage in information-gather- ing, political action, and legal maneuvering in order to influ- ence state behavior.9
None of these theories, however, address autocracies. Au- tocrats who abuse human rights vary in their propensity to rat- ify human rights treaties. By 2004 (the last year in our dataset), the average autocracy had committed to 57% of the top twenty human rights treaties, optional protocols, and dec- larations.10 In the case of the United Nations Convention
7. Beth A. Simmons & Allison Danner, Credible Commitments and the Inter- national Criminal Court, 64 INT’L ORG. 225 (2010).
8. Emilie Hafner-Burton, Edward D. Mansfield & Jon C. Pevehouse, Human Rights Institutions, Sovereignty Costs and Democratization (2011) (unpublished manuscript) (on file with authors).
9. See, e.g., Eric Neumayer, Do International Human Rights Treaties Improve Respect for Human Rights?, 49 J. CONFLICT RESOL. 925 (2005) (demonstrating correlation between strength of civil society and compliance with human rights treaties); SIMMONS, supra note 1, at 14–15 (arguing that treaties can effect change at the national level by defining the political agenda, through litigation, and through the mobilization of political forces).
10. The six major human rights treaties are the Convention Against Tor- ture and Other Cruel, Inhuman or Degrading Treatment or Punishment, Dec. 10, 1984, 1465 U.N.T.S. 85 [hereinafter CAT]; International Conven- tion on the Elimination of All Forms of Racial Discrimination, Dec. 21, 1965, 660 U.N.T.S. 195 [hereinafter CERD]; the International Covenant on Eco- nomic, Social, and Cultural Rights, Dec. 16, 1966, 993 U.N.T.S. 3 [hereinaf- ter ICESCR]; the International Covenant on Civil and Political Rights, Dec. 16, 1966, 999 U.N.T.S. 171 [hereinafter ICCPR]; the Convention on the
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
8 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 38 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 5 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 795
Against Torture (CAT) specifically, 74 of 129 autocratic re- gimes in our sample (between 1985 and 1996) do not sign the CAT.
The autocrats that ratify human rights treaties (many of whom continue to torture and otherwise abuse and repress their populations after accession) have been described as “false positives.” Simmons suggests that these insincere ra- tifiers do so out of short-sightedness, errors in expectations, or mistakes in managing future uncertainty.11
Elimination of All Forms of Discrimination Against Women, Dec. 18, 1979, 1249 U.N.T.S. 13 [hereinafter CEDAW]; and the Convention on the Rights of the Child, Nov. 20, 1989, 1577 U.N.T.S. 3 [hereinafter CRC]. In addition to these six, the remaining instruments include: the European Convention for the Prevention of Torture and Inhuman or Degrading Treatment or Punishment, Nov. 26, 1987, E.T.S. No. 126 (text amended by Protocol 1, Mar. 1, 2002, E.T.S. No. 151, and Protocol 2, Mar. 1, 2002, E.T.S. No. 152); the Inter-American Convention to Prevent and Punish Torture, Dec. 9, 1985, O.A.S.T.S. No. 67; the African Charter on Human and Peoples’ Rights, June 27, 1981, 21 I.L.M. 58; the Minimum Age Convention, June 26, 1973, Inter- national Labor Organization Convention No. C138, available at http://www. ilo.org/ilolex/cgi-lex/convde.pl?C138; the Private Employment Agencies Convention, June 16, 1997, International Labor Organization Convention No. C181, available at http://www.ilo.org/ilolex/cgi-lex/convde.pl?C181; Convention Against Discrimination in Education, Dec. 14, 1960, 429 U.N.T.S. 93; the Optional Protocol to the ICCPR, Dec. 16, 1966, 999 U.N.T.S. 171 (establishing individual complaint mechanism under the ICCPR); the Second Optional Protocol to the ICCPR, Dec. 15, 1989, 1642 U.N.T.S. 414 (aiming at the abolition of the death penalty); the Optional Protocol to the Convention on the Rights of the Child on the Involvement of Children in Armed Conflict, May 25, 2000, T.I.A.S. No. 13,094, 2173 U.N.T.S. 222; the Optional Protocol to the Convention on the Rights of the Child on the Sale of Children, Child Prostitution, and Child Pornography, May 25, 2000, T.I.A.S. No. 13,095, 2171 U.N.T.S. 227; the Optional Protocol to the Convention on the Elimination of All Forms of Discrimination Against Women, Oct. 6, 1999, 2131 U.N.T.S. 83 (establishing reporting mechanism for individual complaints against States Parties to the Committee on the Elimination of All Forms of Discrimination Against Women); and whether a state has submitted a declaration allowing complaints by other states that it is not meeting its obligations under article 41 of the ICCPR, article 14 of the CERD, or article 11 of the CAT.
11. See SIMMONS, supra note 1, at 77–80. After all, they cannot be signing in order to accrue benefits from the international community, for as Beth Simmons argues, there aren’t any. See Richard Nielsen & Beth A. Simmons, Rewards for Rights Ratification? Testing for Tangible and Intangible Bene- fits of Human Rights Treaty Ratification 2 (Jan. 11, 2012) (unpublished manuscript) (on file with authors) (“We find almost no evidence that states can expect increased tangible or intangible rewards after ratification.”). But
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
8 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 38 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 6 24-APR-12 10:00
796 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
Given this variation among autocracies in the propensity to join the international human rights regime, we first explore the relationship between accession and levels of political re- pression and torture within the potential autocratic signatory states. We assess this relationship by relying on two commonly used indexes of human rights violations: the Political Terror Scale (PTS) and the CIRI index of torture.12 Both measures are ordinal indexes of the level of government repression. The PTS measure ranges from 1 to 5, while the CIRI varies between 0 and 2. Higher scores on the PTS index indicate increased levels of repression, such that a score of 1 indicates that a country is “under a secure rule of law, . . . [and] torture is rare or exceptional,” while a score of 5 indicates that the leader “place[s] no limits on the means or thoroughness with which [she] pursue[s her] personal or ideological goals.”13 We have inverted the CIRI measure so that higher scores indi- cate higher levels of repression.14 Since both indexes are ordi- nal, we assess the association between treaty ratification and repression levels using ordered probit regressions.
Our question of interest is whether autocratic leaders with a past history of abuse are more prone to enter into human rights treaties than autocrats without such a past history. We therefore examine the behavior of any autocrat during all of the years the autocrat was in power in a given country prior to that country’s ratification of the given treaty.
The treaties under consideration are the Genocide Con- vention, the ICCPR, the ICCPR Optional Protocol, and the
see Hollyer & Rosendorff, supra note 3, for a signaling game with a rational- ity-based explanation for autocratic accession to the CAT.
12. For the PTS, see Mark Gibney et al., Political Terror Scale Ratings 1976–2010, POLITICAL TERROR SCALE (Nov. 12, 2011), http://www.politi- calterrorscale.org/. For the CIRI index, see David L. Cingranelli & David L. Richards, The Cingranelli-Richards (CIRI) Human Rights Dataset, CIRI HUMAN RIGHTS DATA PROJECT (Jan. 14, 2012), http://www.humanrightsdata.org/.
13. Gibney et al., supra note 14. 14. In this paper, the CIRI index takes the value 2 when torture is fre-
quently practiced, the value 1 when occasionally practiced, and the value 0 when never practiced. See DAVID L. CINGRANELLI & DAVID L. RICHARDS, SHORT VARIABLE DESCRIPTIONS FOR INDICATORS IN THE CINGRANELLI-RICHARDS (CIRI) HUMAN RIGHTS DATASET 4 (2008), available at http:// ciri.binghamton.edu/documentation/ciri_variables_short_descriptions.pdf (describing the CIRI index coding).
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
9 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 39 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 7 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 797
CAT.15 We regress the repression measures discussed above on an indicator variable that is equal to one if a given auto- cratic leader enters into a human rights treaty in the future. Our estimates thus reflect whether or not leaders who go on to sign a given human rights treaty behave in a manner that is systematically different from those that do not. We addition- ally control for GDP per capita in constant 2005 US dollars mea- sured at purchasing power parity, growth in per capita GDP, economic openness, the presence or absence of opposition parties from the legislature, and a cubic time trend.16 The co- efficient on the ratification variable is predicted to be positive when regressed on both the PTS score and the CIRI score. That is, autocratic leaders who go on to ratify human rights treaties are also the ones that engage in more abusive behavior (prior to signing) than those that do not ratify.17
Results of these regressions are reported in Tables 1 and 2. Nearly all the coefficients are positive (with the exception of when the PTS score is regressed on the ratification of the ICCPR Optional Protocol), and these results are often statisti- cally significant. Authoritarian states parties to the Genocide Convention, ICCPR, and CAT on average practice significantly more torture than non-states parties prior to the signing of the treaties according to the CIRI measure. Regressions using the PTS produce similar, though less precisely estimated results. Our results are strongest with respect to the Genocide Conven-
15. We focus on these specific human rights treaties, because these are the most important of the universal human rights treaties; they are also the treaties that require strictest adherence to human rights behaviors, and these are the treaties that have the most severe enforcement provisions.
16. Cubic time trends are a statistical technique for controlling for the passage of time, which can affect the amount of torture or the decision to accede to the treaty. For a discussion of cubic time trends, see Nathaniel Beck et al., Taking Time Seriously: Time-Series-Cross-Section Analysis with a Binary Dependent Variable, 42 AM. J. POL. SCI. 1260 (1998); David B. Carter & Curtis S. Signorino, Back to the Future: Modeling Time Dependence in Binary Data, 18 POL. ANALYSIS 271 (2010).
17. For example, the Chadian dictator Idriss Déby faced repeated at- tempted military coups between 1990 and 1995. He retaliated frequently with large-scale torture and repression of the domestic opposition. In 1995 Déby signed the Convention Against Torture. Uganda during the 1980s is also a major human rights abuser. The Museveni regime acceded to the CAT in 1986. Multilateral Treaties Deposited with the Secretary General, supra note 5.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 3
9 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 39 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 8 24-APR-12 10:00
798 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
tion, and we also find significant results with the CAT and the ICCPR.
TABLE 1: TORTURE AND TREATY RATIFICATION: CIRI Genocide ICCPR Opt.
ICCPR CAT Convention Prot.
Ratify 0.471*** 0.311* 0.091 0.344** [0.175,0.767] [−0.022,0.645] [−0.257,0.439] [0.049,0.640]
−0.162* −0.102 −0.142 −0.160 GDP per capita [−0.336,0.012] [−0.316,0.113] [−0.357,0.074] [−0.353,0.033] Ec. Openness −0.004*** −0.005*** −0.005*** −0.004*** Growth [−0.007,−0.002] [−0.007,−0.002] [−0.007,−0.002] [−0.007,−0.002] Opp. Party −0.572 −0.420 −0.526 −0.510
[−1.449,0.305] [−1.249,0.410] [−1.385,0.333] [−1.386,0.366] 0.093 0.133 0.141 0.094
[−0.172,0.358] [−0.124,0.391] [−0.126,0.407] [−0.162,0.350] Cubic Time 0.453* 0.427* 0.423* 0.412* Polynomial [−0.013,0.920] [−0.045,0.899] [−0.041,0.887] [−0.051,0.875] # of Subjects 1922 1922 1922 1922 # of Countries 113 113 113 113
Coefficient estimates from an ordered probit regression of CIRI torture scores on human rights treaty ratifications. Ninety-five percent confidence intervals are presented in brackets. * denotes significance at the 90 percent level, ** denotes sig- nificance at the 95 percent level, and *** denotes significance at the 99 percent level. All standard errors are clustered by country.
TABLE 2: REPRESSION AND TREATY RATIFICATION: PTS Genocide ICCPR Opt.
ICCPR CAT Convention Prot.
Ratify 0.576*** 0.204 −0.129 0.122 [0.279,0.874] [−0.106,0.515] [−0.455,0.197] [−0.167,0.411] −0.296*** −0.264*** −0.291*** −0.290***
GDP per capita [−0.418,−0.174] [−0.416,−0.112] [−0.454,−0.129] [−0.437,−0.143] Ec. Openness −0.005** −0.005** −0.005*** −0.005** Growth [−0.008,−0.001] [−0.009,−0.001] [−0.009,−0.001] [−0.009,−0.001] Opp. Party −1.013*** −0.992*** −1.030*** −1.019***
[−1.692,−0.333] [−1.606,−0.377] [−1.665,−0.396] [−1.642,−0.395] −0.150 −0.111 −0.102 −0.117
[−0.436,0.135] [−0.394,0.173] [−0.382,0.178] [−0.399,0.166] Cubic Time Polynomial X X X X
# of Subjects 2575 2575 2575 2575 # of Countries 113 113 113 113
Coefficient estimates from an ordered probit regression of PTS repression scores on human rights treaty ratifications. Ninty-five percent confidence intervals are presented in brackets. * denotes significance at the 90% level, ** denotes significance at the 95% level, and *** denotes significance at the 99% level. All standard errors are clustered by country.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
0 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 40 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 9 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 799
These observations are consistent with several extant stud- ies that find that autocratic governments that torture more heavily in a given year are more likely to sign the UN Conven- tion Against Torture (CAT) in the subsequent year than are non-torturing governments.18 Vreeland argues that the pres- ence of domestic opposition parties both causes autocrats to torture more heavily and forces these governments to sign human rights treaties.19 He argues that when opposition par- ties exist, there must be some freedom to engage in speech and activities that have the potential to contravene the bound- aries of permitted speech and action. In such a situation, op- position activists are likely to “cross the line” in their criticisms, leading the government to employ torture to maintain its con- trol. These opposition parties will also pressure the govern- ment to enter into human rights agreements. Finally, in our previous work, we show that eventual signatories of the CAT are roughly fifteen percent more likely to score a two on the CIRI measure (the highest score, indicating frequent use of torture) than non-signatories.20 This difference is significant at the ninety-five percent percent level.21 In short, there is ro- bust evidence that authoritarian signatories of human rights treaties practice substantially greater levels of repression than non-signatories in the period of autocratic rule prior to sign- ing and ratification.
At first glance, the concerned reader might interpret these findings with some optimism—after all, it is the worst offenders who appear to sign these human rights treaties more
18. See, e.g., Oona A. Hathaway, Why Do Countries Commit to Human Rights Treaties?, 51 J. CONFLICT RESOL. 588 (2007); James Raymond Vreeland, Politi- cal Institutions and Human Rights: Why Dictatorships Enter into the United Nations Convention Against Torture, 62 INT’L ORG. 65 (2008).
19. Vreeland, supra note Error: Reference source not found, at 84, 87. 20. Hollyer & Rosendorff, supra note 3, at 298–99. 21. Some authors—applying different indexes of human rights perform-
ance—have found no difference between the human rights records of join- ers of the relevant treaties and non-joiners. See Jay Goodliffe & Darren G. Hawkins, Explaining Commitment: States and the Convention Against Torture, 68 J. POL. 358, 365 (2006) (showing that past use of torture does not affect likelihood of joining the CAT); Emilie M. Hafner-Burton & Kiyoteru Tsutsui, Justice Lost! The Failure of International Human Rights Law to Matter Where Needed Most, 44 J. PEACE RES. 407, 410–11 (2007) (showing no difference in accession rates between high human rights abusers and protectors of human rights).
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
0 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 40 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 10 24-APR-12 10:00
800 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
frequently. If the treaties are effective, then the treaties would appear to be helping the countries that are most in need of improvement. Unfortunately, this optimism is not rewarded. In the next section we show that the effect of these treaties is minimal on the human rights abuse in these signing (and rati- fying) autocracies.
II. ACCESSION BY AUTOCRACIES HAS LITTLE EFFECT ON HUMAN RIGHTS ABUSE IN THOSE COUNTRIES
Hafner-Burton and Tsutsui confirm that signing human rights treaties has little or no effect on the behavior of the world’s worst repressors.22 As they put it, there is a “rising gap between states’ propensities to join the international human rights regime and to bring their human rights practice into compliance,” and this gap brings the efficacy of international law into fundamental question.23
Gilligan and Nesbitt contend that norm-based arguments for the adoption of the CAT have not had any noticeable effect on torture levels.24 Elsewhere, we offer evidence that torture falls weakly in autocratic signatory states relative to non-signa- tories.25 It may be that signatories are torturing less after sign- ing; or it may simply be that non-signatories torture more after they decline the opportunity to sign. Both could even be true.
Figure 1 provides a graphical presentation of the finding that signing CAT has a weak effect on actual torture levels. We regress changes in the CIRI and Hathaway torture indexes on signatory status, using ordered probit regressions. In the first panel, signing the CAT is associated with a predicted lowering of the Hataway torture score – signatory governments have a higher probability of seeing a reduction (a score of -1) than non-signatories. This effect is much less pronounced in the second panel, where reductions in torture (as measured by CIRI) are not as strongly related to CAT signing.
22. Emilie M. Hafner-Burton & Kiyoteru Tsutsui, Human Rights in a Globalizing World: The Paradox of Empty Promises, 110 AM. J. SOC. 1373, 1395 (2005).
23. Id. at 1374. 24. Michael J. Gilligan & Nathaniel H. Nesbitt, Do Norms Reduce Torture?,
38 J. LEGAL STUD. 445, 445 (2009). 25. Hollyer & Rosendorff, supra note 3, at 315.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
1 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 41 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 11 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 801
FIGURE 1: TORTURE’S RESPONSE TO SIGNING THE CAT
Results from a regression of changes in torture levels as a function of changes in CAT signatory status. Torture scales have been recoded such that a score of 1 is associated with a rise in torture, a score of -1 denotes a decline, and a score of 0 indicates no change. Additional controls include the change in per capita GDP, the change in military capabilities, changes in the growth rate, and changes in economic open- ness, plus a cubic polynomial of time. Diamonds depict the mean difference in the predicted probability of receiving a given change in torture score, comparing those that have signed the CAT in the past year to those that experience no change in signatory status. Whiskers depict ninety-five percent confidence intervals around this difference in predicted probabilities.
The news is disheartening. The worst offenders sign/rat- ify the human rights treaties more frequently, and physical in- tegrity rights show little or no improvement in these autocratic states.
In evaluating the welfare implications, the astute analyst must consider the set of possible counterfactuals. How would these autocratic repressors have behaved absent the opportu- nities to accede to these treaties? Have these treaties influ- enced behavior in any meaningful way? The absence of a meaningful counterfactual is complicated by the fact that states are voluntarily choosing to be bound by the treaty obli- gations. To address this issue, Hill matches signatories with non-signatories along a number of observable characteristics (such as size, income, human rights histories, among other cri-
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
1 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 41 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 12 24-APR-12 10:00
802 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
teria), and only admits such matched pairs into his dataset.26 He then explores differences in performance between states that have signed (the CAT and the ICCPR) and those that have not. The news here is even worse: Hill shows that ratifica- tion of the CAT and the ICCPR is actually associated with re- duced respect for physical integrity rights.
These findings are consistent with earlier studies. Hathaway concludes that “ratification of regional treaties ap- pears more likely to worsen human rights practices than to im- prove them.”27 Similarly, Keith finds no relationship between ratification of the ICCPR and subsequent human rights pro- tections.28 The accumulated evidence suggests that signing and/or ratifying major human rights treaties has little effect in autocracies, and may actually worsen abuses in some autocratic states.29
III. IS REGIME CHANGE DELAYED?
In this section the news gets even worse. If autocratic rights-abusers are signing these treaties more frequently than the lesser abusers, there are two possibilities. The first is that these leaders believe that these treaties are irrelevant to their prospects for surviving in office and serve no purpose—either useful or harmful—and they believe that they are so strong
26. Daniel W. Hill, Estimating the Effects of Human Rights Treaties on State Behavior, 72 J. POL. 1161, 1169 (2010).
27. Oona Hathaway, Do Human Rights Treaties Make a Difference?, 111 YALE L.J. 1935, 2004 (2002).
28. See Linda Camp Keith, The United Nations International Covenant on Civil and Political Rights: Does it Make a Difference in Human Rights Behavior?, 36 J. PEACE RES. 95, 112 (1999) (“[I]t may be overly optimistic to expect that being a party to [the ICCPR will] . . . effect observable change in actual human rights behavior.”).
29. One possible concern relating to measurement emerges when com- paring human rights performance before and after accession to an interna- tional treaty. It is possible that after ratification the international commu- nity, NGOs, and other bodies may subject the member state to closer scru- tiny, making the measurements of behavior before and after joining not directly comparable. This may bias the results away from finding improve- ments in human rights outcomes—for closer scrutiny may turn up cases of abuse that might have been missed with less detailed inspection so that the data misleadingly shows more human rights violations after ratification than before ratification. We have no basis or evidence for believing this bias to be present, but we mention it as a possible caveat to our empirical findings.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
2 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 42 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 13 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 803
that their security is unaffected by their membership. The sec- ond possibility is that these autocrats, always concerned about their survival in office,30 believe that acceding to the interna- tional human rights regime in some form may actually strengthen their hold on power.
While we leave further theoretical exploration of these possibilities for the following section, we start by answering the empirical question: is regime change delayed within the auto- cratic states that accede to the human rights treaties? We find in the affirmative.31
We examine the association between the ratification of human rights treaties by autocratic governments and the sur- vival of these governments in office. We assess this relation- ship using a Cox proportional hazards regression, which as- sesses the relationship between governments’ hazard rates (the probability that the government fails in time t conditional on having survived until that point) and covariate values. We draw on the Archigos dataset on Political Leaders (version 2.9) for the timing of leaders’ entry into and removal from office.32 Our principal explanatory variables of interest are the propor- tion of all human rights treaties ratified. We rely on informa- tion provided by the UN Office of the High Commissioner for Human Rights (OHCHR) to code these variables.33
We additionally control for leader age (from Archigos); GDP per capita measured in constant 2005 US dollars mea- sured at purchasing power parity, growth in real GDP per cap- ita, and economic openness all drawn from the Penn World
30. See BRUCE BUENO DE MESQUITA, ALASTAIR SMITH, RANDOLPH M. SIVER- SON & JAMES D. MORROW, THE LOGIC OF POLITICAL SURVIVAL 9 (2003) (“We take it as axiomatic that everyone in a position of authority wants to keep that authority . . . .).
31. Regime change here means that the autocratic leader is removed from office. The leader may be replaced with another autocrat, a demo- cratic regime, or some other political structure; for our purposes, regime change means the replacement of the leader.
32. H.E. GOEMANS, KRISTIAN SKREDE GLEDITSCH & GIACOMO CHIOZZA, ARCHIGOS: A DATABASE ON POLITICAL LEADERS 1875–2004: VERSION 2.9 (2006), available at http://www.rochester.edu/college/faculty/hgoemans/ Archigos.2.9-August.pdf.
33. This data was kindly provided to us and compiled by Beth Simmons from OHCHR sources. See the U.N. Treaty Collection, UNITED NATIONS, http://treaties.un.org (last visited Feb. 28, 2012).
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
2 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 42 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 14 24-APR-12 10:00
804 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
Table version 6.3;34 an indicator of whether multiple parties served in the legislature;35 and measures of political repression drawn from the Political Terror Scale.36 A final control is ad- ded for whether a given autocratic leader ‘inherits’ a human rights treaty—i.e., the treaty was signed by one of his predeces- sors. This control adjusts for the possibility that treaties affect the survival of accessor governments differently than they af- fect successor governments. All regressions are run only for autocracies, as coded by Cheibub, Ghandi, and Vreeland.37 We additionally control for whether a given autocrat is a mon- arch or member of the military.
We have not reported the regression results here because the coefficient estimates from a Cox regression are difficult to interpret directly. We instead use the estimated regression co- efficients to compute estimates of the hazard function for au- tocratic governments that have not ratified any of the twenty human rights treaties listed by the OHCHR and for those that have ratified half of these treaties in Figure 2. Hazard rates— the probability that a given leader is removed in time t condi- tional on having survived until time t—are presented on the y- axis. Time, measured in years in office, is presented on the x- axis. The solid line refers to authoritarian regimes that have not ratified any of the treaties reported by the OHCHR, the dashed line refers to authoritarian regimes that have ratified half of these treaties.
34. ALAN HESTON, ROBERT SUMMERS & BETTINA ATEN, CTR. INT’L COMPAR- ISONS, PENN WORLD TABLES VERSION 6.3 (2009), available at http:// pwt.econ.upenn.edu/php_site/pwt_index.php.
35. See Cheibub et al, supra note Error: Reference source not found, at 69 (listing multi-party elections as one of four requirements for classifying a regime as a democracy).
36. Gibney et al, supra note Error: Reference source not found. 37. See Cheibub et al, supra note Error: Reference source not found, at
74–76 (explaining their coding rules for determining whether a country can be categorized as a democracy).
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
3 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 43 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 15 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 805
FIGURE 2: HAZARD RATES: FREQUENT AND INFREQUENT AUTHORITARIAN RATIFIERS
Hazard function estimates from a Cox regression of leader survival on the proportion of 20 human rights treaties reported by the OHCHR as ratified.
As is evident from the figure, autocratic governments that ratify a large number of human rights treaties suffer a substan- tially lower risk of removal than non-signatories for each year that they are in office. Roughly 15 in 100 autocratic non-signa- tories are predicted to be removed in their fifth year in office, while only 9 in 100 frequent signatories are predicted to suffer a similar fate.
Figure 2 is generated using fitted predictions from the model. The estimated hazard rate is defined for all times dur- ing which there is an observed exit from office, and smoothed lines are fit between these discrete points. Since the Cox model is a proportional hazards model, the hazard rate of non-ratifiers is always assumed to be equal to a fixed constant times the hazard rate of leaders who have ratified half of all available treaties.38 However, at the boundary points—i.e., af- ter forty years in office—the smoothing function used to con-
38. Conventional tests of this assumption fail to reject the hypothesis that the hazard rates are proportional. Thus, our results are not driven by the unusual behavior of leaders who survive for particularly long or short peri- ods in office.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
3 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 43 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 16 24-APR-12 10:00
806 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
nect the discrete points for which hazard estimates are identi- fied somewhat distorts the shape of the graph.
IV. ARE THE HUMAN RIGHTS TREATIES TO BLAME FOR DELAYING REGIME CHANGE?
The evidence presented thus far suggests a powerful and significant correlation between human rights accession and autocratic leader-survival. As we know, however, correlation does not imply causality.
It is possible that this is a causal effect: that accession and/ or ratification of human rights treaties causes delays in auto- cratic leader turnover or delays in democratization. Alterna- tively, the observed relationship may simply be a correlation driven by a host of possible differences between ratifiers and non-ratifiers that are not controlled for in regression specifica- tions. One likely source of such differences is selection into ‘treatment’—i.e., those unobserved factors that drive ratifiers to accede to a given treaty may also increase the duration of their tenure in office.
A. Commitment Effect
If the treaties truly have no consequences for human rights abusers, then we would not expect the treaties to have any noticeable effects on the behavior of abusive autocrats. On the other hand, if there are enhanced legal consequences for violating the terms of the treaties, then we would expect leader behavior to adjust to take account of these conse- quences.
The international human rights regime has seen a pro- gressive strengthening of enforcement provisions. The Geno- cide Convention first merely required states to make genocide a crime under domestic law. The Treaty of Rome now permits the UN Security Council to refer a violator to the International Criminal Court for prosecution, clearly raising the potential costs of committing such crimes in the instance that the auto- crat is deposed. Similarly, the CAT, with its provisions employ- ing the principle of “universal jurisdiction,” makes the use of torture and subsequent attempts to run into exile much more costly and unlikely to succeed.
The rational response of any autocratic leader who has ratified these treaties, recognizing the worsened conditions he
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
4 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 44 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 17 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 807
will likely face on losing power, will be to fight harder to hold onto office than he would have otherwise. The effect of the treaties may be to discourage the voluntary abdication of dicta- tors, and to increase violence and conflict.
We call this incentive to fight to hold onto power, in or- der to avoid more serious punishments if the leader were to fall, the commitment effect. By committing to the treaty, the costs of failure are so much larger that more effort is placed on survival. Leaders may thus take more drastic actions—in- cluding repressive behavior—to cement their hold on office. In this way the agreements can be seen to enhance survival in office.39
B. Information Effect
Sometimes, a leader engaged in a conflict needs to con- vince his opposition that he is prepared to fight to the end, to incur any cost, in order to survive. A recent example of this can be found in the case of Syria under Bashar al-Assad, who has been engaged in brutal oppression of his political oppo- nents. He has been using tanks and snipers to eliminate protestors—according to the New York Times, more that 150 protestors were killed in Syria in the last week of 2011 alone. These attacks coincided with the visit of the Arab League to Syria to monitor the situation. Syrian government troops, in full view of the monitors, were observed to open fire on protes- tors in the town of Hama. According to a leader of the opposi- tion militia, Colonel Riad as-As’aad, “[t]he killing has in- creased with the presence of the observers.40
The key element is the self-infliction of additional costs, which sends a signal of the strength of the commitment to
39. For example, many argued that the indictment by the International Criminal Court of the Libyan dictator, Muammar Khaddafi made his volun- tary departure from Libya impossible. Without the indictment and with guarantees of a safe haven, Khaddafi and his family might have left earlier, preventing the loss of life and significant military expense incurred by the US and NATO during the months of conflict. In contrast, the amnesties granted by the successor government to Augusto Pinochet in Chile facili- tated the negotiated democratic transition and Pinochet’s exit from govern- ment.
40. Kareem Fahim, Syria Killings Continue; Group to Add Observers, N.Y. TIMES, Jan. 3, 2012, http://www.nytimes.com/2012/01/04/world/mid- dleeast/syria-killings-continue-despite-observers.html.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
4 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 44 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 18 24-APR-12 10:00
808 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
fight till the end. The current al-Assad regime in Syria is incur- ring international reputational and other costs by being ob- served to brutally crack down and massacre its opponents, and its leaders will be subject to international indictment should the government fall. This leads to a credible belief that al-As- sad is prepared to do anything, including being seen to kill protestors, in order to survive in office.
In the human rights arena, when an autocrat accedes to the Genocide Convention or the CAT, the autocrat is running the additional risk of prosecution on eviction from office. There is no longer the option of retreat into comfortable ex- ile. Since the costs of backing down are now higher, the do- mestic opposition learns the toughness of the autocrat, and the strength of his commitment to hold onto office in the face of opposition. The opposition, on learning that the autocrat is stronger than expected, retreats—perhaps to garner their forces and fight another day. The reduced effort of the oppo- sition reduces the risk the government will actually fall, and the information conveyed improves the probability of regime survival.
We call this effect the information effect—where the do- mestic opposition learns something about the strength of the autocrat—and responds optimally and accordingly. The net effect may be a reduction in the probability of regime failure, since the opposition backs down.41 This effect is distinct from the commitment effect above: the commitment effect de- scribes the autocrat’s behavior. If there are penalties to sign- ing and then abusing human rights, an autocrat will abuse even more in order to avoid losing office and then being pe- nalized. The information effect involves instead, the rational response of the domestic opposition—who on seeing the sign- ing (and abuse) rationally respond by fighting less hard. Both effects act to strengthen the autocrat’s hold on office.
Now there are many ways a venal autocrat could signal his strength to the domestic opposition—simply shooting some
41. A good example of this is the case of Sierra Leone. Siaka Stevens signed the CAT in 1985, and promptly retired. His carefully handpicked successor was Joseph Saidu Momoh. The signing was clearly a signal to the opposition of the strength of the regime, and Momoh (though corrupt and subject to much international criticism) survived in office until a military coup in 1992. See Hollyer & Rosendorff, supra note 3, at 307–10, for statisti- cal evidence that opposition effort declines after CAT signing.
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
5 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 45 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 19 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 809
opponents might convey the same message. But any tinpot dictator can do that, and this could as easily be interpreted as a sign of weakness rather than strength. If a leader voluntarily binds his own hands, personally incurring potential extra costs of engaging in human rights abuses—this is a credible signal of commitment. Killing opponents isn’t very costly (to the au- tocrat), and as a result conveys little information. But offering to incur additional personal costs—such as going to prison if deposed—suggests to the opposition that the leader has no intention of going anywhere. The opposition either must fight harder, or back down.
These two effects—the information and the commit- ment—suggest that the human rights treaties are causing the enhanced leader survival. Absent the treaties, these effects would not be operative, and the leaders’ survival would not be enhanced.
There is a third possible effect, which is not causal.
C. Selection Effect
In a world absent a set of human rights treaties, it is the toughest (as opposed to the weakest) autocrats that would fight the hardest to secure their survival. And the toughest of these autocrats would be more likely to be the ones abusing human rights. Hence the human rights abuse and the signing of the treaties are both caused by a third variable—the under- lying toughness of the autocrat. This omitted variable in- troduces “selection bias” into the inference process. This se- lection effect means that the human rights regime does not cause increased survival. Rather, the regimes most likely to survive are the ones that self-select into ratification.
In an ideal universe (for the social scientist, that is), in order to test the claim that ratification causes leader survival, we would take a large number of countries, and subject some of them randomly to “ratification” and then watch to see if the treated leaders survive longer. But the world is of course a more complicated place, and the ratification “treatment” is not randomly assigned. Most importantly, states are self-select- ing into treatment—they are choosing whether to sign the treaties. If the underlying characteristics of states that sign the treaties are fundamentally different from the characteristics of non-signatories, then comparing the survival of leaders in sig-
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
5 S
id e B
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 45 Side B 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 20 24-APR-12 10:00
810 INTERNATIONAL LAW AND POLITICS [Vol. 44:791
natory vs. non-signatory countries misattributes the differences in survival rates to the (signing of the) treaty. It is instead re- lated to underlying differences between countries that led some to sign and others to not sign. This is the problem of selection bias, due to omitted variables. In this case, since it is the strong autocrats that sign and torture more, it could be that it is not the treaty causing survival, but instead that it is the strong leaders that sign, and these are the types that survive in office longer anyway.
All three effects work in the same direction when it comes to prolonging autocratic survival in office. Empirically, we see that survival is enhanced by treaty accession. This could be due to any one, or any combination, of these effects. But we have not as yet definitively established that the human rights regime is responsible for this unanticipated outcome. Future research is likely to explore which of these three effects—com- mitment, information, and selection—are operative, and which are most dominant.
V. IMPLICATIONS
Nevertheless, the human rights regime potentially has a severe and unanticipated consequence—of securing a greater degree of survival in office for the worst human rights abusing dictators. This has some important implications.
Should the process of strengthening the international human rights regime continue, and should the prospect of more severe post-tenure punishment become a reality, some human rights abusers may be deterred from violating their citi- zens’ physical integrity. The evidence suggests, however, that the penalties for human rights abuse are used to make the commitment not to leave office more credible—if a leader risks jail or worse on leaving office after signing a treaty, then signing the treaty suggests that the leader has no intention of leaving office. If we increase those penalties, and the leaders continue to sign, they are simply communicating that they re- ally are never going to leave office.
If tougher penalties for human rights abuse are imple- mented as part of the voluntary human rights regime, this would further enhance the survival in office of those autocrats that accede. Recall that the worst torturers are more likely to sign the CAT than the less tough autocrats. If penalties for
3 1 7 2 5 _ n yi_
4 4 -3
S h e
e t N
o . 4
6 S
id e A
0 4
/2 5 /2
0 1
2 0
9 :5
4 :2
5 31725_nyi_44-3 Sheet No. 46 Side A 04/25/2012 09:54:25
\\jciprod01\productn\N\NYI\44-3\NYI303.txt unknown Seq: 21 24-APR-12 10:00
2012] HUMAN RIGHTS AGREEMENTS 811
human rights violations are moderate, it is the tougher and the moderate autocrats that sign the treaties. As the penalties increase, however, only the tougher types join, sending a stronger signal, and signatories fight harder to stay in office, preserving the worst offenders in office.
The effect of the human rights treaties on global abuse and repression can be reconsidered in the light of these find- ings. We have shown that human rights abuse stays constant or falls slightly among signatories. But these autocrats are sur- viving in office significantly longer. The net effect may be more or less overall torture in those acceding states—torture may decline on a year-to-year basis, but those most prone to torture may be left in a position to practice abuse for a longer period. The overall level of human rights abuse may indeed be worse—because it extends over a longer period of time.
There is also an unexpected benefit that we should point to. The survival in office of the autocratic signatories is com- pared to the survival in office of autocratic non-signatories. The presence of the treaties may shorten the tenure in office of the non-signers. Since these autocrats are also abusing human rights, their shorter reign is welfare-improving. The problem of course is that the non-signers were not the big abusers to start with—the big abusers are more likely to sign, as we showed above.
The lesson here is that the earnest attempts by the devel- oped world to bring the power of international law to bear against the worst violators of civil and physical rights may have unanticipated consequences that make the underlying prob- lem worse. Human rights treaties may not (if either the infor- mation or the commitment effect is operative) reduce human rights abuse world-wide.
