Skip to main content Accessibility help

When Do (Imposed) Economic Sanctions Work?

  • Jon Hovi (a1), Robert Huseby (a2) and Detlef F. Sprinz (a3)


Previous research has documented only a modest success rate for imposed sanctions. By contrast, the success rate is higher in cases that are settled at the threat stage. In this article, the authors provide new insights about the circumstances under which sanctions cause behavioral change only after being imposed. First, the target must initially underestimate the impact of sanctions, miscalculate the sender's determination to impose them, or wrongly believe that sanctions will be imposed and maintained whether it yields or not. Second, the target's misperceptions must be corrected after sanctions are imposed. A game-theoretical model with incomplete information is used to develop and clarify the argument.



Hide All

1 Elliott, Kimberly Ann, “The Sanctions Glass: Half Full or Completely Empty?” International Security 23, no. 1 (1998); Hufbauer, Gary Clyde, Schott, Jeffrey J., and Elliott, Kimberly Ann, Economic Sanctions Reconsidered: History and Current Policy, 2nd ed. (Washington, D.C.: Institute for International Economics, 1990).

2 Galtung, Johan, “On the Effects of International Economic Sanctions, with Examples from the Case of Rhodesia,” WorldPolitics 19 (April 1967); Pape, Robert A., “Why Economic Sanctions Do Not Work,” International Security 22, no. 2 (1997); idem, “Why Economic Sanctions Still Do Not Work,” International Security 23, no. 1 (1998).

3 Drezner, Daniel W., “The Hidden Hand of Economic Coercion,” International Organization 57 (Summer 2003); Lacy, Dean and Niou, Emerson, “A Theory of Economic Sanctions and Issue Linkage: The Roles of Preferences, Information and Threats,” Journal of Politics 66, no. 1 (2004); Parker, Richard W., “The Problem with Scorecards: How (and How Not) to Measure the Cost-Effectiveness of Economic Sanctions,” MichiganJournal of InternationalLaw 21, no. 235 (2000).

4 We adopt the standard terminology whereby the country (or countries) that threatens or imposes sanctions is called “sender,” while the country threatened with or suffering sanctions is called “target.”

5 Of course, there are also other concerns of interest to policymakers in this situation, for example, the sender country's credibility.

6 A divergent view is found in Baldwin, David A., Economic Statecraft (Princeton: Princeton University Press, 1985). It is worth noting that Baldwin takes a broader view than most writers on what constitutes success for imposed sanctions.

7 Galtung (fn. 2); Margaret P. Doxey, International Sanctions in Contemporary Perspective (Basing-stoke: McMillan Press, 1987). For an extensive list of negative assessments of economic sanctions, see Baldwin (fn. 6).

8 Baldwin (fn. 6), 57.

9 An extreme example is the American grain embargo against the Soviet Union in 1979. Whereas the Soviet Union lost only U.S.$225 million due to higher prices, the United States lost exports worth an estimated U.S.12.3 billion. See O'Quinn, Robert P., “A User's Guide to Economic Sanctions,” Roe Backgrounder no. 1126 (Washington, D.C.: Heritage Foundation, 1997), 9; available at http://www.

10 A former British ambassador to South Africa once claimed that the sanctions against Rhodesia probably caused more damage to its neighbors than to Rhodesia itself. See Berridge, G. R., Internatioanl Politics: States, Power and Conflict since 1945 (Hemel Hampstead: Harvester Wheatsheaf, 1997), 126.

11 Galtung (fn.2).

12 For example, Castro's popularity does not appear to have been notably weakened by decades of sanctions.

13 Hufbauer, Schott, and Elliott (fn. 1).

14 Pape(fn.2,1997,1998).

15 Hufbauer, Gary Clyde, Schott, Jeffrey J., and Elliott, Kimberly Ann, Economic Sanctions Reconsidered: History and Current Policy (Washington, D.C.: Institute for International Economics, 1985), 8191.

16 In the 1985 edition, Hufbauer, Schott, Elliott (fn. 15) do report the results of a multivariate regression analysis. Only two of eighteen coefficients obtain t-values larger than 2. Curiously, neither of these two variables is included on the authors’ list of the main determinants of successful sanctions.

17 Drury, A. Cooper, “Revisiting Economic Sanctions Reconsidered,” Journal of Peace Research 35, no. 4 (1998); Dashti-Gibson, Jaleh, Davis, Patricia, and RadclifFe, Benjamin, “On the Determinants of the Success of Economic Sanctions: An Empirical Analysis,” American Journal of Political Science 41, no. 2 (1997).

18 Blanchard, Jean-Marc F. and Ripsman, Norrin, “Asking the Right Questions: When Do Economic Sanctions Work Best?” Security Studies 9, no. 1 (1999).

19 Ibid., 225.

20 Dorussen, Han and Mo, jongryn, “Ending Economic Sanctions: Audience Costs and Rent-Seeking as Commitment Strategies,” Journal of Conflict Resolution 45, no. 4 (2001); Lindsay, James M., “Trade Sanctions as Policy Instruments: A Re-examination,” International Studies Quarterly 30, no. 2 (1986).

21 Lacy and Niou (fn. 3); Parker (fn. 3).

22 A possible objection to this argument is that, when a threat of sanctions fails, the sender will update the probability that imposed sanctions will work. An instrumentally motivated sender will impose sanctions only if the updated likelihood of success is sufficiently high. Thus, we should not expect sanctions to be imposed in all cases where the threat of sanctions fails. In other words, the subset of cases in which sanctions are actually imposed is not a random sample from the pool of threat failures. Rather, we should expect cases in which sanctions are imposed to have a higher than average probability of success. This selection effect is largely ignored in this article.

23 Drezner (fn. 3) reanalyzes data previously considered by Bayard, Thomas O. and Elliott, Kimberly Ann, Reciprocity and Retaliation in U.S. Trade Policy (Washington, D.C.: Institute for International Economics, 1994); Elliott, Kimberly Ann and Richardson, J. David, “Determinants and Effectiveness of Aggressively Unilateral Trade Actions,” in Feenstra, Robert C., ed., The Effects of U.S. Trade Protection and Promotion Policies (Chicago: University of Chicago Press, 1997); Elliott, Kimberly Ann, “Preference for Workers? Worker Rights and the Generalized System of Preferences,” Working Paper (Washington, D.C.: Institute for International Economics, 2000), available at paper.cfm?ResearchID=313; DeSombre, Elizabeth R., Domestic Sources of International Environmental Policy: Industry, Environmentalists, and U.S. Power (Cambridge: MIT Press, 2000).

24 It is not yet clear whether such a generalization can be made, because there might be qualitative differences between the unilateral sanctions in Drezner's material and broad multilateral sanctions. However, there is nothing in our argument that depends on whether such generalization is possible.

25 To be more specific, the first of the two conditions identifies three different scenarios. Only the first two possibilities are analyzed formally. The third possibilityrequiresa different model and is thus only considered informally in this article (see Section III).

26 Cf. Pape (fn. 2,1997).

27 Lopez, George A. and Cortright, David, “Containing Iraq: Sanctions Worked,” Foreign Affairs (July-August 2004).

28 See the Comprehensive Report of the Special Advisor to the DCI on Iraq's WMD (September 30, 2004), available at

29 Lopez and Cortright (fn. 27), 91.

30 Ibid., 92.

31 The crux of this case is that the weapons inspectors were not certain about the amount of WMD still to be destroyed prior to the onset of the second Iraq war.

32 Our argument presupposes throughout that the parties behave rationally. Hence, it does not capture nonrational reasons that a state may have for yielding or not yielding to sanctions.

33 A fourth possibility is that the threat is irrelevant because the target is incapable of yielding despite its desire to do so. In this case, neither threatened nor imposed sanctions will have any chance of being effective. For a discussion of conditions for a threat to be effective, see Hovi, Jon, Games, Threats and Treaties: Understanding Commitments in International Relations (London: Pinter, 1998), chap. 2.

34 For example, this might explain Iraq's intransigence with the sanctions regime and the weapons inspectors. A second illustration—the sanctions against Libya—is discussed later in this section.

35 Please note that the illustrations provided here are meant to exemplify, rather than prove, our theoretical findings. Further research is needed to test these findings empirically.

36 The reason is simply that it takes time to impose sanctions. If the sender really is determined to impose sanctions, then sooner or later this is likely to become clear to the target, even if it has initially misjudged the senders intentions. The target is then likely to yield, provided that the threat of sanctions is also sufficiently potent and contingent. If the target yields, then sanctions will usually not be imposed.

37 Beloff, Max, The Foreign Policy of Soviet Russia: 1929-1941 (London: Oxford University Press for Royal Institute of International Affairs, 1947), 35; Strang, William, Home and Abroad (London: A. Deutsch, 1956), 8189, 117.

38 Strang(fn. 37), 87.

39 Blanchard and Ripsman (fn. 18), 232.

40 Ibid., 233.

41 Strang (fn. 37), 110; Beloff (fh. 37), 35.

42 Strang (fn. 37), 35; Beloff (fn. 37), 110-14. Hufbauer, Schott, and Elliott (fn. 1) count these sanctions as successful (p. 32).

43 There are also cases where a country violates an international norm until potent sanctions are credibly authorized, because this enables it to reap certain benefits in the meantime. For example, the George W. Bush government introduced protective measures for the U.S. steel sector during the 2004 reelection campaign only to remove them just before the November election because the World Trade Organization authorized the EU to impose potent sanctions. While the WTO ruling could be reasonably anticipated, the Bush government essentially reaped an electoral benefit in “battleground” states of the Midwest before it yielded.

44 Doxey (fn. 7), 40.

45 Ibid., 41.

46 Hufbauer, Schott, and Elliott (fn. 15), 416.

47 Robin Renwick, Economic Sanctions (Cambridge: Harvard University Center for International Affairs, 1981), 54; Harry R. Strack, Sanctions: The Case of Rhodesia (Syracuse, N.Y.: Syracuse University Press, 1978), 238. Both are quoted in Baldwin (fn. 6), 197.

48 Baldwin (fn. 6), 196-201.

49 Ibid.

50 Doxey (fn. 7), 46.

51 Baldwin (fn. 6), 193-96.

52 Ibid., 197.

53 Plachta, Michael, “The Lockerbie Case: The Role of the Security Council in Enforcing the Principle Aut Dedere Aut Judkare,” European Journal of International Law 12, no. 1 (2001), 125-29.

54 Zoubir, Yahia H., “Libya in U.S. Foreign Policy: From Rogue State to Good Fellow?” Third World Quarterly 23, no. 1 (2002), 35.

55 The UN sanctions were not, however, permanently lifted until 2003 (Security Council Resolution 1506).

56 Zoubir (fn. 54), 35—36; see also O'Sullivan, Meghan, Shrewd Sanctions: Statecraft and State Sponsors of Terrorism (Washington, D.C.: Brookings Institution Press, 2003), 184.

57 To be more precise, the lifting of sanctions requires that at least nine of the fifteen members vote in favor and that no permanent member votes against lifting them.

58 Zoubir (fn. 54), 43.

59 Ibid., 41.

60 O'Sullivan (fn. 56), 184.

61 An exception is the model proposed by Lacy and Niou (fn. 3), where under certain conditions imposed sanctions can induce the target to yield. However, this is only possible in a mixed strategy equilibrium (p. 37). In contrast, our model admits this outcome in a pure strategy equilibrium.

62 For a review of the literature, see Drezner (fn. 3), 645-48.

63 For example, the relevant violation could be the invasion of a neighboring country or the illegal development of weapons of mass destruction.

64 In our model, an explicit threat is superfluous, because the threat of sanctions is implicit in the structure of the game, which is assumed to be common knowledge.

65 It should be emphasized that noncompliance with international norms is not always intentional. Violations can also be the result of incapacity problems; see, for example, Chayes, Abram and Chayes, Antonia Handler, “On Compliance,” International Organization 47, no. 2 (1993); idem, , The New Sovereignty: Compliance with International Regulatory Agreements (Cambridge: Harvard University Press, 1995). Conversely, countries comply with international norms for a variety of reasons, many of which have nothing to do with sanctions; see, e.g., Checkel, Jeffrey, “Why Comply? Social Learning and European Identity Change,” International Organization 55, no. 3 (2001). Thus, many cases of compliance with international norms can not be counted as successes for economic sanctions. To count as a success, compliance must be causedby an expectation on the part of the target that it would otherwise risk sanctions. Of course, this might be difficult to convincingly demonstrate in practice.

66 If sanctions are “potent,” Target (1) prefers not to violate the norm, rather than to violate the norm and suffer the sanctions. Also if sanctions are “potent,” Target (2) prefers to yield once sanctions are imposed. Similarly, if sanctions are “lenient,” Target (3) prefers to violate the norm and suffer the sanctions, rather than refrain from violating the norm. Finally, if sanctions are “lenient,” Target (4) prefers to stand firm even after sanctions are imposed.

67 One might think of a and fS as crude measures of how long it takes for Target to yield.

68 Dorussen and Mo (fn. 20), 403.

69 As usual in games of incomplete information, p and q are assumed to be common knowledge.

70 For example, suppose that Sender imposes potent sanctions. Then (using Bayes’ rule) Target is able to update the probability distribution of Sender's possible types so that the posterior probabilities become p = l, q = 0, 1 − p − q = 0.

71 The model also goes beyond the informal discussion in that it admits three options for Sender. At the same time, two of the scenarios discussed informally come out as special cases of the model's equilibria. In this sense, the model offers a more general treatment than the informal discussion.

72 A related suggestion is that Sender might initially impose lenient sanctions (or no sanctions at all) but shift to more potent sanctions ifTarget does not yield.

73 We do not consider the possibility that Sender might exaggerate its determination to impose (potent) sanctions. The reason is that such exaggeration would not increase the likelihood of successful imposed sanctions (even though it might successfully deter Target from violating the norm).

74 A third possibility, related to the discussion in Section III (but not covered by the model), is that Sender falsely signals a stern resolve in all contingencies, thereby deceiving Target to incorrectly assume that sanctions will be imposed even if it does not violate the norm. In such a case the target might yield after sanctions are imposed if Target then receives credible evidence that the sanctions will be lifted if and only if it yields. However, if Target is prepared to yield under such circumstances it would also have been deterred from violating the norm in the first place had it known the sender's true intentions at that stage. Therefore, deception is counterproductive, because it makes ineffective the threat of sanctions. Meanwhile, if sanctions are imposed as a result of such deception, both Sender and Target suffer.

75 There is, however, reason to doubt that smart sanctions are more potent than comprehensive sanctions. For an instructive discussion, see Tostensen, Arne and Bull, Beate, “Are Smart Sanctions Feasible?” World Politics 54 (April 2002).

* We are indebted to Geir B. Asheim, Jeffrey Checkel, Aanund Hylland, Ronald B. Mitchell, Havard Strand, and four anonymous reviewers for helpful comments. We also thank Frank Azevedo for excellent editorial assistance.

When Do (Imposed) Economic Sanctions Work?

  • Jon Hovi (a1), Robert Huseby (a2) and Detlef F. Sprinz (a3)


Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed