The dictator’s dilemma: Why communist regimes oppress their citizens while military regimes torture and kill

Abstract

What makes some authoritarian regimes more willing to employ extrajudicial violence (torture and killings), as opposed to more conventional forms of repression (restrictions on speech and association)? A voluminous literature addresses the causes and dynamics of state repression. Whereas large-N studies explain repressive activities as proportional responses to the challenges governments face, historical work reveals instances of disproportionate repression. This literature, moreover, is inconclusive regarding the effects of communist and military regimes on violations of physical integrity rights. Another shortcoming of current work is that different types of repression are modeled separately. I distinguish between oppression (restrictions on speech), repression (the use of beatings, arrests, and trials to restrain the rights of assembly and association), and state terrorism (when governments intimidate political opponents using extrajudicial violence). I examine the relationships among them in a multivariate regression framework from 1952 to 2010. My analysis reveals that communist dictatorships repress the freedoms of expression, travel, and association, whereas military dictatorships engage in extrajudicial violence. My study contributes to the literature by providing an institutional account of why tactics of repression differ between these two political systems, and by considering the effects of temporal lags, endogeneity, and diffusion processes on state repression.

View correction statement:

Correction

Acknowledgments

I would like to acknowledge Fordham University’s support for this research through a Research Expense Program award and a Faculty Fellowship award. I would also like to thank the editors, an anonymous reviewer, Rodwan Abouharb, William Akoto, Courteney Conrad, Anjali Dayal, John Entelis, Erica Frantz, Boris Heersink, José Kaire, Graig Klein, Carl Knutsen, Enrico Antonio La Viña, Dong Wook Lee, J. Patrice McSherry, Chaitra Nagaraja, Olena Nikolayenko, Melissa Patel, Jan H. Pierskalla, Kurt Weyland, Tim Wilson, Dwayne Woods, Joe Wright, and Joe Young for their comments or help with various aspects of researching and writing this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article was originally published with errors, which have been corrected in the online version. Please see Correction: https://doi.org/10.1080/14754835.2023.2232239

Notes

1 In the literature, “repression” sometimes refers to the totality of tactic authorities employ to control their subjects, usually preceded by the term “state.” I use it in this sense, but also more narrowly to distinguish it from two other forms of repression—oppression, and state terrorism (Lopez, 1984, p. 59). Repression in the latter sense refers more narrowly to efforts to limit the rights of assembly and association using tactics such as beatings, destruction of property, arrests, and trials.

2 According to Jorge Videla, one of the architects of the dirty war and head of the first military junta (March 29, 1976—March 29, 1981), the number of dead and disappeared during the “National Reorganization Process” (the name the Argentine military gave its last dictatorship) is between 7,500 and 8,875. (Reato, 2012, p. 20). Pion-Berlin (1991, p. 136) spoke of 15,000. Most put the true number at 20,000 to 30,000 (Finchelstein, 2014, p. 1; Pereira, 2005, p. 212; Pion-Berlin, 1988; Zaretsky, 2018, pp. 285–287).

3 I use “communist” interchangeably with “state-socialist” to refer to a society “controlled by a dominant communist party which seeks, on the basis of Marxism–Leninism and through the agency of the state, to mobilize the population to reach a classless society” (Lane, 1996, p. 1, quoted in Armstrong, 2003, p. 2).

4 In my data, the People’s Democratic Republic of Ethiopia (1975–1990), the People’s Republic of the Congo (1969–1991), and Somalia (1969–1990) are the only three communist regimes that were simultaneously military.

5 “Personal integrity abuse refers to any violent government action that violates rules that have developed in political theory and international law concerning acceptable uses of physical force by governments” (Hill, 2016, p. 824). It refers to torture, killing, political imprisonment, and disappearance (Davenport, 2007c, p. 487).

6 See Weyland’s (2019, p. 115) discussion of the disproportionate violence used by the Chilean armed forces following the coup of September 11, 1973. Dinges (2004) made a similar point about Operation Condor countries. Operation Condor was “a United States-backed campaign of political repression and state terror involving intelligence operations and assassination of opponents, officially and formally implemented in November 1975 by the right-wing dictatorships of the Southern Cone of South America.” Described by the CIA as “a cooperative effort by the intelligence/security services of several South American countries to combat terrorism and subversion,” its operations crossed national borders as dissidents and leftists “who ‘had gone into exile’ … were kidnapped, tortured and killed in allied countries or illegally transferred to their home countries to be executed.” https://en.wikipedia.org/wiki/Operation_Condor (Retrieved July 29, 2019). The founding members consisted of Chile, Argentina, Uruguay, Paraguay, and Bolivia. “Brazil, Ecuador, and Peru joined later” (Dinges, 2004, p. 122). Some “50,000 persons were murdered in the frame of Condor, 9,000–30,000 disappeared … and 400,000 incarcerated.” See https://en.wikipedia.org/wiki/Dirty_War (Retrieved July 29, 2019). In Brazil and Uruguay, however, repression was more limited in scope. (Feierstein, 2010, p. 506). Condor ended in 1983 (McPherson, 2019, p. 45).

7 According to Hill, Poe and Tate (1994, p. 854) defined personal integrity abuse as equivalent to state terrorism as discussed by Gurr (1986). However, Gurr used the term “state terror” to refer to … lethal or life-threatening violence that targets opponents of the government with the intention of intimidating those who are not victims of the violence (rather than only eliminating the victims). State terror defined this way is a special category of repression (Hill, 2016, p. 823).

8 See also Davenport and Inman (2012, pp. 627–629).

9 The following provides a good illustration: “People weren’t killed in the streets, making it clear to families that they were gone. They weren’t officially executed. They were arrested and then disappeared in the middle of the night. Loved ones often had no idea if their relatives were still alive, making them even more paralyzed with fear” (Bevins, 2020, pp. 144–145; emphasis in the original).

10 Interestingly, whereas the Chilean military organized the National Intelligence Directorate (DINA) to manage repression soon after taking power (November 1973) (Policzer, 2009, p. 73), the armed forces themselves conducted torture and disappearances during Argentina's dirty war. As Peter Smith wrote, “To wage its ‘dirty war,’ the junta decentralized power, creating many separate units within the military and police … these units operated with nearly total autonomy, selecting victims according to their own criteria. … The decentralization, secretiveness, and competition among the military services meant that friends and relatives of the victims, no matter how influential they were, could not take advantage of their social contacts” (cited in Sloan, 1984, p. 87). These precautions notwithstanding, officers who did not share the leadership’s ideological beliefs did not conduct the dirty war with the same zeal as those who did (Scharpf, 2018).

11 According to Richards et al.’s (2015, p. 293) comprehensive review of the scholarship spun by Poe and Tate’s (1994) workhorse model of state repression of physical integrity rights, the literature is inconclusive on the effects of “leftist” government and “military control” on physical integrity rights violations.

12 “Preventive repression … involves restrictions of speech, prohibitions of assembly, controls of travel and selective application of laws.” (Slantchev & Matush, 2020, pp. 219–220; emphasis in the original). See also Mason and Krane (1989, p. 176).

13 The dirty war, for example, was “plainly illegal under the law of Argentina as it stood during … military rule” (Argentine National Commission, 1986, p. xviii). The same is true of the terrorist actions conducted by government forces in Uruguay during the country’s last military dictatorship (1973–1985; see Servicio Paz y Justicia-Uruguay, 1992, pp. 79–80). Not only that, but in the directives the Argentine military produced internally as part of its planning, “[t]here is no direct reference to the use of illegal methods.” (Moyano, 1995, p. 85).

14 Sometimes the constitution, or parts of it, are not in effect because it has been abrogated by emergency measures such as a state of siege, martial law, or a state of emergency. One common tactic is to establish special security courts (Moustafa & Ginsburg, 2008, p. 4). Another is to use the military penal code. In Chile during the first five years of the military regime, for example, “most prosecutions that did take place occurred in ‘wartime’ military courts, insulated from the civilian judiciary” (Pereira, 2005, p. 4). In addition, because the regime “felt under threat from subversion throughout its rule,” it preserved some form of military tribunals “right up until the transition to democracy” (Pereira, 2005, p. 52).

15 For this reason, when coding torture, Hathaway (2002, p. 1970) “disregarded punishments carried out pursuant to a country's legal system, even if that system may be considered by some to sanction torture.” Her dataset, however, contains too few observations to base my analysis on it.

16 In Eastern Europe (except the Balkans) and North Korea, however, communist movements rode to power in 1945 on the back of Soviet tanks. I thank Tim Wilson for bringing this point to my attention.

17 Geddes et al. (2014a, p. 148) counted only four cases in which military-led dictatorships instituted redistributive policies.

18 Keeping the military out of power has proved exceedingly difficult in some countries. “Thus, between 1946 and 2008, soldiers participated in the removal or installation of roughly two of every three Latin American leaders” (Svolik, 2012, p. 123). Another extreme case is Thailand, which has experienced more than a dozen military coups since 1932 (Ginsburg & Hug, 2018, p. 50). In 1973, the country witnessed an “anti-communist extermination program” resulting in the death of 3,000 leftists or accused leftists (Bevins, 2020, p. 266).

19 According to Abouharb et al. (2013, p. 371), from 1981 to 2010 two-thirds of authoritarian regimes had courts that were somewhat or completely independent. For example, the Park Chung-hee and Chun Doo-hwan military regimes in South Korea “kept an appearance of formal constitutional legality. Courts were relatively autonomous, but the scope of their activity was carefully circumscribed” (Moustafa & Ginsburg, 2008, p. 6).

20 Shen-Bayh (2018) made the same point.

21 Barros (2002, p. 25) echoed this claim when he spoke of “power holders” directly applying “extrajudicial force upon political adversaries, while they allow the rule of law to operate in less conflictive areas, such as … repression of moderate opponents.”

22 Mason and Krane (1989) offered a similar interpretation of state terrorism in Salvador in the 1980s. That regimes that paid lip service to the writ of habeas corpus engaged in mass disappearances is tragic. As Keith et al. (2009, p. 649) pointed out, “Observance of habeas corpus rights would preclude disappearances.”

23 To be clear, “when opposition to a regime is mass-based, organized, and potentially violent, the military is the only force capable of defeating it” (Svolik, 2012, p. 127).

24 The CIRI (Cingranelli and Richards) Human Rights dataset provides three physical integrity rights indicators that match my empirical domain (killings, disappearances, and torture). I did not use them in my analysis, however, because CIRI data are only available since 1981, greatly hampering coverage, given that death squads emerged in the 1960s and 1970s “in at least ten states in Latin America alone” and “disappearances were first documented in the mid-1970s” (Sluka, 2000, p. 4). Second, Cingranelli and Richards distinguished between killings and disappearances, but they defined killings in the same way as V-Dem. Third, in defining disappearances, it is not clear if they referred only to those perpetrated by governments and/or their agents. See the CIRI (Cingranelli and Richards) Human Rights Data Project Short Variable Descriptions, Version 5.21.14, p. 3, available at https://drive.google.com/file/d/0BxDpF6GQ-6fbY25CYVRIOTJ2MHM/edit (Retrieved May 21, 2020). Finally, torture was not limited in their coding to torture that takes place during incarceration (Cope et al., 2019, p. 3).

25 A Bayesian measurement model aggregates responses from expert surveys and factual-based coding (Pemstein et al., 2018). This model accounts for noise in the data, providing boundaries for country-year values similar to a confidence interval.

26 As McCormick and Mitchell (1997, p. 512) wrote, “There is likely considerable circularity between the measurement of civil war and the measure of government ‘repression of human rights to personal integrity’ that occur as a result of these wars.” I use version 2017 of the Polity data (Marshall et al., 2018) to delineate my population.

27 See Appendix A for detailed variable definitions and operationalizations.

28 See the Autocracies of the World codebook, pages 1–2, located at https://fsi-live.s3.us-west-1.amazonaws.com/s3fs-public/res/Codebook.pdf (Retrieved August 7, 2022).

29 I rely on Svolik (2012) for a list of communist countries, which I edited with the help of Wikipedia's list of communist states (see https://en.wikipedia.org/wiki/Communist_state). See Appendix B for a list of discrepancies between the two sources and how they were resolved.

30 Using Magaloni’s et al.’s (2013) regime classification, communist regimes constitute 16.26% and military ones 26.52% of autocratic regime observations. Using Geddes et al.’s (2014b) regime classification and Kim and Kroeger’s (2018) coding of military regimes, the equivalent figures are 18.97% and 27.4%, respectively.

31 Davenport (2007c, p. 81) assumed that this is why “authorities prefer to increase political restrictions before escalating violence.” This fails to explain, however, why military regimes engage in forms of repression that are highly polemical.

32 A little over half “of all coup attempts succeed” (Geddes et al., 2018, p. 33).

33 I have changed values for 899 observations from missing to 0 because V-Dem leaves this index empty for cases in which legislatures were closed in the aftermath of successful military coups. The correct inference for these cases is not, however, that they lie outside the scope of the variable, but that legislative checks on the executive are absent.

34 Some indexed measures or scales of constraints on executive authority such as Polity’s XCONST indicator “lack a direct institutional interpretation” (Svolik, 2012, p. 37).

35 Svolik’s fourth dimension is military involvement in politics.

36 I consider only structural variables that are consistently used as explanatory factors in extant work (Hill & Jones, 2014, p. 661; Keith, 2012, chap. 3).

37 Although some scholars have examined more than one repressive strategy (e.g., Cingranelli & Richards, 1999), the latter did not “consider political and civil liberties, nor did they investigate the causal determinants of different repressive strategies” (Davenport, 2007c, p. 46).

38 A prominent exception is Ritter and Conrad, who claimed that governments and dissidents “choose their actions in anticipation of the other’s behavior” (Ritter & Conrad, 2016, p. 85).

39 In addition, generalized least squares is used “to account for the correlation structure in the disturbances across the equations.” See p. 6 in StataCorp’s “reg3” user’s manual, available at https://www.stata.com/manuals13/rreg3.pdf (Retrieved August 6, 2022).

40 See https://www.stata.com/manuals13/rreg3.pdf, pp. 12–18, for a similar approach.

41 Collinearity results in some countries being dropped from the analysis (45 out of 191). This, however, does not bias the findings because collinearity does not affect any regime type in particular. Of the communist countries, for example, Afghanistan, East Germany, North Korea, Somalia, and South Yemen are dropped, but 19 other regimes remain in the analysis.

42 A long-run equilibrium is plausible in light of differences among dictatorships in repressive profiles (Davenport, 2007c).

43 The lagged dependent variable also controls for autocorrelation.

44 Military regime was exogenous in the previous analysis because it is not significantly associated with all four forms of repression (i.e., repression) and the variable is impossible to instrument for in a model for continuous dependent variables.

45 According to Fariss and Schnakenberg (2014, p. 1021), “political imprisonment is the mode of enforcement for violations of rights to movement or freedom of association.”

46 The possible range of the dependent variable, delta torture, is −3.786 to 3.665.

47 See also the positive and statistically significant long-term effect of protest.

48 Habeas corpus also ceased to exist there in November of 1959 (Thomas, 1998, p. 1,253). Deadly purges also occurred in other communist countries at the end of World War II, to be followed by less violent, more “legal” repression. I thank Tim Wilson for bringing this point to my attention.

49 See Cuba Archive. (2018). How many political prisoners are there in Cuba? https://cubaarchive.org/wp-content/uploads/2020/07/How-many-Cuban-political-prisoners.pdf. (Retrieved March 31, 2021).

50 Slantchev and Matush (2020, p. 219) have referred to civil rights restrictions as “the bread and butter of autocratic regimes.”

51 Compare this with Davenport (2007c, p. 500), who claimed that “single-party governments … are less likely to restrict civil liberties.”

Additional information

Funding

This work was supported by Fordham University Research Expense Program award and Faculty Fellowship award.

Notes on contributors

José A. Alemán

José A. Alemán is a professor of political science at Fordham University. He has degrees from Cornell University (B.A.) and Princeton University (M.A., Ph.D.) and teaches courses on comparative politics and political economy. His work focuses on democratic institutions, social movements, regime change, human rights, and social science methodology. He is associate editor of the Journal of Contemporary East Asia and was the topic editor of “Comparative Political Science and Measurement Invariance: Basic Issues and Current Applications” for Frontiers in Political Science. His work has appeared in the Korea Observer; Methods, Data, Analyses; Peace Studies; International Studies Quarterly; Nations and Nationalism; Migration and Development; the Oxford Handbooks Online; the Oxford Handbook of Employment Relations; Review of European Studies; Social Science Quarterly; Comparative Political Studies; Industrial Relations Journal; International Political Science Review; Political Studies; and International Journal of Korean Studies. He has received fellowships from the Macarthur Foundation, the Mellon Foundation, and the Korea Foundation.