|
|
|
|
How And Whom the U.S. President Sanctions: A Time-Series Cross-Section Analysis of U.S. Sanction Decisions and Characteristics
Arizona State University,
Department of Political Science
International Studies Association
March 1998
Abstract
Does the U.S. President levy economic sanctions against some nations more than others? When sanctioning a nation, does the president tend to make that sanctioning effort more or less severe, or longer or shorter, depending on the target nation? Do the ways international and domestic factors influence the president’s decision to use and alter sanctions vary by the type of target nation? To answer these questions, I apply three decision-making models, international, domestic, and cognitive, to the president’s decision to use and alter economic sanctions. The data are analyzed as a time-series—cross-section of 22 nations sanctioned by the U.S. with the following results. The president does not tend to sanction either communist or Latin American nations more or less frequently, independently from the predictions by the three models. However, the type of nation does influence the reason the President uses sanctions. The type of sanction does affect the President’s decision to alter economic sanctions such that he tends to make sanctions against Latin American targets shorter and more intense. Finally, the type of nation affects the President’s decision to alter economic sanctions differently from the decision to use them.
Introduction
Past studies of economic sanctions focused on their effectiveness. Very few scholars directly addressed against whom economic sanctions are used and why. 1 This primary focus on effectiveness leaves several questions unanswered. Considering the U.S. as the sender of the sanctions, which nations tend to be sanctioned the most? Are there any biases affecting which nations the President decides to sanction? Why does the President sanction different nations? Does he use sanctions only for international reasons attempting to alter the target nation’s policies like the majority of economic sanctions scholars assume, or do domestic imperatives affect the decision?
In a previous analysis, Drury (1996) showed that the President’s decision to use economic sanctions was based on three different types of incentives and imperatives: dyadic, domestic, and cognitive. These three aspects of the President’s decision to use and alter sanctions were modeled as (1) coercive diplomacy—the President’s attempt to influence the target; (2) domestic imperatives—the President’s attempt to please his public or divert their attention; and (3) procedural rationality—the influence the President’s advisory group had on his decisions. All three of these models influenced the President’s decision to use economic sanctions, while the first two models also influenced the President’s decisions to alter the sanctions.
This paper is an extension of that work addressing two main questions. First, does the U.S. President levy economic sanctions against some types of nations more than others? Second, do the reasons the President uses economic sanctions vary by the type of nation he is sanctioning? Both of these questions are applied to not only the President’s decisions to use sanctions but also to his decisions to alter (lift, decrease, maintain, or increase) those sanctions. To answer these questions, I apply the three decision-making models, coercive diplomacy, domestic imperatives, and procedural rationality to the President’s decision to use and alter economic sanctions. The data are analyzed as a time-series—cross-section of 22 nations sanctioned by the U.S. between 1948 and 1978.
The paper unfolds in two main stages. In the first stage, I begin by reviewing the conditions that lead the President to use sanctions. I analyze these results to determine if the President is more likely to sanction a specific type of nation, or whether the reasons he sanctions are affected by the type of nation. In the second stage, the conditions that lead the President to alter the sanctions are reviewed. I then analyze these results to see if the President is more likely to lift or decrease sanctions against a specific type of target. These results are also analyzed to see if the type of nation affects the reasons the President alters the sanction policy. Finally, these results provide the information needed to characterize how and whom the President sanctions.
The Decision to Use Economic Sanctions
In the previous analysis, I laid forth seven hypothesis to test three models of political rationality: coercive diplomacy, domestic imperatives, and procedural rationality. These models and their hypotheses are presented in Table 1 below.
Table 1:
Hypotheses for the Decisions to Use and Alter Economic Sanctions
| Hypotheses | Independent Variable | Hypothesized Relationship | |
| Model I: Coercive Diplomacy | When the magnitude of the tension between the U.S. and the target is high, the President is more likely to enact economic sanctions. | magnitude | Positive |
| The greater the rate of increase in the tension level (denoting increased escalation), the more likely is the President to enact economic sanctions. | escalate | Positive | |
| If the target acts provocatively, the President is more likely to enact economic sanctions. | provoke | Positive | |
| Model II: Domestic Imperatives | The lower the President's public approval rating, the more likely he is to enact economic sanctions. | approval | Negative |
| During election periods, the President will be more likely to enact economic sanctions. | election year | Positive | |
| A low approval rating during an election year raises the likelihood that the President will enact economic sanctions. | electoral approval | Negative | |
| Model III: Procedural Rationality | Presidents with informal advisory systems will be more likely to enact economic sanctions. | advise | Positive |
These hypothesis were then tested using a time-series—cross-section data set encompassing 22 nations sanctioned by the U.S. from 1948 to 1978. The dependent variable is whether there were sanctions against the given target for a given month. Because the dependent variable is binary, logistic regression is used. Further, because the data are arrayed into a BTSCS (binary time-series—cross-section), I use Beck, Katz, and Tucker’s (1997) method of accounting for temporal dependence. Thus, the variable timepast is the number of months since the last sanction for that target, and the variables spline1, spline2, and spline3 are the three knots estimated in the cubic spline. Finally, the estimation is a fixed-effects model; the results appear below in Table 2.
Table 2:
The Decision to Use Economic Sanctions
| Variable | Coefficient | SE | P-Valu e* |
| magnitude | 0.522 | 0.074 | 0.000 |
| escalate | -1.488 | 0.359 | 0.000 |
| provocation | -0.181 | 0.082 | 0.014 |
| approval | -0.012 | 0.014 | 0.199 |
| election year | -2.528 | 0.930 | 0.004 |
| electoral approval | 0.029 | 0.016 | 0.039 |
| advise | 0.669 | 0.239 | 0.003 | < /tr>
| timepast | -0.510 | 0.036 | 0.000 |
| spline1 | 0.000 | 0.000 | 0.000 |
| spline2 | 0.000 | 0.000 | 0.000 |
| spline3 | 0.000 | 0.000 | 0.000 |
| Constant | 0.040 | 1.557 | 0.490 |
% Correct = 96.8
* 1-tailed Test timepast and the three splines represent the temporal variables.
While statistically significant, the relationships in Table 2 do not unconditionally support the causal stories specified by the three models and their hypotheses. For the coercive diplomacy model, the results support Hypothesis 1 that predicts higher U.S.—target tensions raise the probability that the President will initiate sanctions.
Hypotheses 2 and 3 are not supported. escalate is significant but shows that slower rates of escalation lead to sanction use (p < 0.000). The provocation level is also negative, indicating that increased provocation by the target decreases the probability that the President will initiate economic sanctions (p < 0.014). These results raise the question of whether a rapidly escalating situation or very provocative stimuli lead decision-makers to skip sanctions as an option and either (a) resort to more intense options or (b) capitulate.
A careful inspection of the data shows that the former explanation (a) does not hold. In the 80 months where escalation rates were 100% or higher, the President never decided to use force or covert action, although a war may have already been in play. It seems more probable that the President perceives the target’s belligerency as a show of resolve and decides on (b), an alternative, less intense approach to the crisis (George, 1991:77). Another possible explanation is that the target, by acting provocatively or in such a manner that would create a rapidly escalating situation, may be successfully deterring the U.S. That is, the President may see the target’s aggressive behavior as a sign of resolve and political integration within the target (Galtung, 1967; Eland, 1996). The President’s perception that the target will not succumb to the sanctions leads him not to use them in the first place (Smith, 1996).
While two of the three variables representing domestic imperatives are significant, the significant variables do not act as hypothesized. approval is related in the hypothesized direction but is insignificant (p < 0.199). Elections do not have the hypothesized effect on the President’ s decision, and their effect is significant (p < 0.004). Instead of increasing the probability that the President will use or maintain economic sanctions, elections decrease the probability. For example, during a non-election year and holding all variables at their means, a president will use sanctions when the tension level rises above 11. During an election year, the President must have a approval rating above 60% and the tension between the U.S. and the target must exceed 14. 2
The interaction term (electoral approval, p < 0.020) shows that during an election year, presidents with a high approval rating are more likely to use sanctions. This finding indicates that during election periods, a strong president (one with high approval) is slightly more likely to use sanctions than a weak president. Perhaps only strong presidents are willing to risk using economic sanctions knowing that they have the political strength to withstand the possible backlash from campaign supporters or the public. Weaker presidents realize that they do not have the political strength either to endure attacks from businesses hurt by the sanctions or to risk a public backlash at a possibly ineffective policy.
A statistically significant (advise, p < 0.003) and positive sign supports the hypothesis that predicts presidents with informal systems will be more likely to use economic sanctions. However, unlike the rather strong election effect, presidential information processing only affects the decision at the margin. 3 Who the President Sanctions
The estimation in Table 2 is a fixed-effects model. This technique inserts dummy variables for each of the targets thus controlling for cross-target heteroskedasticity. 4 That is, the President may decide to use sanctions against each of the targets for slightly different reasons. While the three models explain the President’s use of economic sanctions, the individual nations he sanctions also affect his decision. It is possible that instead of individual targets affecting the President’s decisions, types of targets affect his decision. If this is the case, then dummy variables identifying the type of target nation would replace the dummy variables identifying the individual targets in the fixed-effects model. The advantage would be to explain the President’s decisions as not simply contingent on individual nations but on specific types of nations.
Following this possibility, there are two basic classifications for the dyads. First, it seems entirely possible that the President would be more likely to use sanctions against Latin American nations because of the Monroe Doctrine and American historical interest and involvement in western hemisphere politics. Because the U.S. has a higher level of influence and closer relations with Latin American nations relative to the rest of the world, the President will be more likely to use economic sanctions. To identify these nations, I created a dummy variable (latin) that equals 1 for all Latin American or Caribbean nations and 0 for all other nations. 5
The second classification concerns communist regimes. I expect that the President is more likely to sanction a communist country during the 1948-1978 period because of the cold war (Nincic and Wallensteen 1983). I identify only those nations that were openly communist, coding them as 1 and all other nations as 0. 6 While any nations with a leftist orientation may be more likely to incur economic sanctions from the President, this relationship is captured by the magnitude of the dispute. The tension between the U.S. and a leftist target will be higher than between the U.S. and a non-leftist target. I am not assessing whether a leftist oriented nation is more likely to be sanctioned. Instead, I am testing if the President is, in the face of other variables measuring the relations between the U.S. and target, still more likely to use sanctions on a communist target nation.
To test these two propositions, I must determine if the differences found in the fixed-effects model are actually an artifact of the target’s being either a Latin American and/or communist country. The best way to assess this possibility is through the use of the fixed-effects model. Both the two variables indicating Latin American and communist countries need to be analyzed with the individual dyad effects to determine if they (the group effect) or the individual effects are significant. If the two variables are significant, then it is not the case that the President has independent decision rules for each target nation; instead, those differences are accounted for by the target’s location (Latin America) and/or ideology (communist).
There is a statistical problem with this testing procedure: latin and communist and those dummy variables identifying the individual targets over-specify the equation. For example, dummy variables identify the USSR twice: the dummy variable identifying the USSR and communist. The over-specification inflates the dummy variable’s coefficients and therefore inflates their significance to such a degree that the significance measures are no longer reliable.
To solve this problem, I follow an unconventional procedure. First, I estimate two fixed-effects models. In the first of these estimations, only the variables from the three models are entered. This estimation I call Model A. In the second equation, I add all of the model variables and the two group variables, latin and communist. I call this estimation Model B.
Second, I compare the coefficients of the dyad dummy variables from the two estimations. If at least one of the two group variables has a significant and independent effect on the President’s decision, then the coefficients for the dyad dummy variables in Model B would be significantly different from those in Model A. That is, if the differences caused by the individual dyads are a function of the targets in those dyads being either a Latin American and/or communist nations, then the individual dyad dummy variable’s coefficients in Model B would be effectively reduced to zero. For example, if latin is significant, then the effects of those variables identifying nations such as Argentina and Brazil should be zero. Therefore, the dyad dummy variable coefficients in Model B would significantly differ from those in Model A. The end result is, if the dyad dummy variable coefficients from the two models are significantly different, then at least one of the group variables is significant.
Third, if the test above indicates that one of the two variables has an independent effect, I will then estimate the model with latin and communist. This estimation will allow for interpretation of the effects and also show if one or both is significant. 7
To perform this analysis, I generate two estimations: Model A and Model B. The dependent variable is simply the President’s dichotomous decision to use economic sanctions. Instead of reporting the two estimations, I report the differences between the dyad dummy variable coefficients in Table 3 below.
To test if the two sets of coefficients are significantly different, I take their means and perform a matched difference of means test. The test reports the differences are not statistically significant (p < 0.502). 8 The fact that nations are Latin American and/or communist does not have an independent significant effect on the President’s decision to use economic sanctions. Instead, the individual dyads have independent effects on the President’s decision to levy economic sanctions.
Table 3:
Dummy Variable Coefficient Differences
The Decision to Use Variable:-
| Dummy Variables Only | Dummy Variables with latin and communist | |
| Argentina | -2.034 | -1.044 |
| Brazil | -1.139 | -0.139 |
| Ceylon | -0.450 | 18.794 |
| Chile | -0.913 | 0.028 |
| China | -1.448 | -18.575 |
| Cuba | 0.366 | -34.458 |
| Dominican Republic | -3.204 | -2.232 |
| Egypt | -1.047 | 18.206 |
| Ethiopia | -0.757 | -17.661 |
| India | -1.173 | 18.084 |
| Indonesia | -1.084 | 18.305 |
| Iran | -1.471 | 17.794 |
| Kampuchea | -1.447 | 17.541 |
| Laos | -0.961 | 17.419 |
| North Korea | -0.400 | -17.739 |
| Pakistan | -1.740 | 17.493 |
| Peru | 0.591 | 1.526 |
| Uganda | -1.135 | 18.066 |
| USSR | -0.468 | -17.826 |
| Constant | 1.122 | -18.335 |
P < 0.502 (2-tailed Test)
While it seems surprising that the President is not independently affected by a target’s location or regime, the mean tension levels between the U.S. and these two types of nations partly explain this result. The mean tension level between the U.S. and a communist nation is 9.04, while it is 6.94 for all other nations, a statistically significant difference (p < 0.000). The greater tension between the U.S. and communist countries translated into more sanctions. The U.S. had economic sanctions in place against communist nations 81.2% of the time and only had sanctions in place against non-communist targets 16.8% of the time, a significant difference (p < 0.000). 9 While the President sanctioned communist nations more, that difference appears in the magnitude of the disputes between the U.S. and target, not through an independent communist effect. 10
Like communist nations, the President does not have a bias toward sanctioning Latin American targets. Contrary to the hypothesis that the U.S. will sanction Latin American nations more, the data show that the President tends to sanction Latin America less. The U.S. held economic sanctions against Latin American nations only 20.6% of the time compared to 35.9% of the time for all other nations (significantly different at p < 0.000). 11 While this finding is the result of the three decision-making models, it means that the U.S. has closer more friendly relations with its Latin American neighbors. When put into a world perspective, this finding becomes quite logical. During the 1948-1978 period, the U.S. had much closer relations with Latin America compared to the other nations in the data set such as Indonesia, India, China, and the Soviet Union, to name just a few.
A further explanation of this finding, economic sanctions tend to decrease the sender’s influence on the target once they are in place. Because long term influence in Latin America is important for U.S. interests, the President would be more likely to use less obvious forms of coercive diplomacy than economic sanctions. In addition, the closer ties between the U.S. and Latin America would allow the President more avenues for resolution of a dispute without having to resort to economic sanctions.
While the President does tend to sanction Latin American countries less and communist countries more, these tendencies are a function of the President’s decision calculus, not a bias toward one of those types of counties. I now turn to the possibility that the type of nation affects the reasons the President uses sanctions.
Why the President Sanctions a Specific Type of Nation
While the type of country does not lead the President to sanction more or less, the type of country may affect why he sanctions them. That is, the analysis above shows that the President is not more or less likely to sanction Latin American or communist nations, but he may sanction them for different reasons. For example, the President may sanction communist countries for primarily international reasons and not domestic ones.
To asses whether the President sanctions different types of nations for different reasons, I use latin and communist to select the cases included in the estimation. That is, I run the analysis for Latin American, communist, and all other nations in three separate estimations. Because all of the signs for each of these estimations are a reflection of the larger overall model, the interesting aspect of the results are which of the variables are statistically significant; Table 4 displays the results from all three estimations. In all non-Latin American and non-communist nations, the President bases his decision to use sanctions on simply the magnitude and escalation of the tension. 12 This finding suggests that for these nations, the President is influenced purely by the dispute between the U.S. and the target. Only international imperatives affect his decision, not domestic ones or his advisory system.
Table 4:
The Decision to Use for latin, communist, and neither.
| Neither | Latin American Nations | Communist Nations | ||||
| Variable | Coefficient | SE | Coeffi cient | SE | Coefficient | SE |
| magnitude | 0.398 | 0.084 | 0.651 | 0.176 | 0.687 | 0.196 |
| escalate | -0.999 | 0.451 | -2.635 | 0.734 | -0.263 | 1.280 |
| provocation | -0.119 | 0.101 | -0.16 9 | 0.189 | -0.467 | 0.197 |
| approval | 0.009 | 0.018 | -0.157 | 0.043 | -0.022 | 0.030 |
| election year | -0.813 | 1.189 | -11.103 | 2.953 | - 3.709 | 2.099 |
| electoral approval | 0.004 | 0.022 | 0.154 | 0.046 | 0.045 | 0.035 |
| advise | 0.177 | 0.276 | 0.416 | < td>0.68210.263 | 3.505 | |
| timepast | -0.450 | 0.046 | -1.021 | 0.179 | -0.540 | 0.071 |
| spline1 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| spline2 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| spline3 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| Constant | -1.151 | 1.125 | 9.172 | 2.848 | -0.852 | 2.434 |
Bolded Italics = Significant p < 0.05
For Latin American nations, both international (magnitude and escalation) and domestic (approval, electoral approval, and elections) variables affect the President’s decision to use economic sanctions. These results suggest that there is a strong domestic component to the President’s decision to use sanctions against a Latin American nation. This result can be explained by America’s greater attention to those nations with which it shares a hemisphere. U.S. companies invested a great deal in Latin America and thus, were far more interested in its politics than the politics of more distant nations. The fact that three of the four expropriation disputes were with Latin nations is evidence of the American investments and interest in the region. 13 If U.S. companies are more interested in their investments, which tend to be in Latin America, it follows that the President will be more influenced by domestic politics relative to other nations.
Finally, the President decides to sanction communist nations for a variety of reasons. First, the level of tension and provocation affect the President’s decision, but a highly escalatory dispute does not. Thus, the President does not flinch at a rapidly intensifying dispute with a communist nation like he does with other nations, but he does tend to back down in the face of belligerent provocations. As always, the tension level affects the President’s decision. Domestically, elections decrease the President’s propensity to use sanctions.
Finally, the only nations against which the President’s advisory system affects his decision are communist nations. Thus, only when involved in a dispute with a communist nation do president’s with informal advisory systems tend to sanction more than their formal counterparts. Another way of looking at this finding is that the President’s advisory system only affects decisions about communist nations. Because the cold war encompasses the 1948-1978 period, it is quite probable that the President only relied heavily on his advisors when the target was with a communist nation. That is, the decision to sanction a non-communist nation (including Latin American nations) was routine enough that a great deal of consulting with advisors was unneeded. Because the President did not consult with his advisors as much, there was less opportunity for his decision to be affected by the advisory system. Therefore, the greater the consultations with advisors over sanctioning communist countries lead to an advisory system effect for communist nations.
These results indicate that the President’s decision calculus does vary by the type of target, i.e., the type influences the reasons he sanctions. I now assess whether the type of nation affects the decisions to alter economic sanctions.
The Decision to Alter Economic Sanctions
Once the President enacted sanctions, he must decide what to do with them in the following month. In the next stage of the analysis the dependent variable is whether the President, in a given month against a given target, lifted, decreased, maintained, or increased economic sanctions against that target. The hypotheses appear in Table 5 below.
Table 5:
Hypotheses for the Decisions to Alter Economic Sanctions
| Hypotheses | Independent Variable | Hypothesized Relationship | ||
| Model I: Coercive Diplomacy | When the magnitude of the tension between the U.S. and the target is high, the President is more likely to increase or maintain economic sanctions. | magnitude | Positive | |
| The greater the rate of increase in the tension level (denoting increased escalation), the more the more likely is the President to maintain, or increase economic sanctions. | escalate | Positive | ||
| If the target acts provocatively, the President will maintain, or increase economic sanctions. | provoke | Positive | ||
| Model II: Domestic Imperatives | The lower the President's public approval, the more likely he is to maintain, or increase economic sanctions. | approval | Negative | |
| During election periods, the President will be more likely to maintain, or increase economic sanctions and less likely to decrease or lift them. | election year | Positive | ||
| A low approval rating during an election year raises the likelihood that the President will maintain, or increase economic sanctions and lowers the likelihood that he will decrease or lift them. | electoral approval | Negative | ||
| Model III: Procedural Rationality | Presidents with informal advisory systems are more likely to maintain or increase sanctions and less likely to decrease or lift them. | advise | Positive |
Because the dependent variable is ordinal, I use ordered logit to estimate the model. Another consequence of the ordinal data is that Beck et. al.’s solution to temporal dependence does not apply. However, the number of actual sanctions against a target nation is coded into the data set. In order to capture the trend flowing through time, I take the percent change in the number of sanctions applied against a nation from one month to the next and then lag the variable one month. The result is a variable that shows the past sanctioning trend thereby tapping the time dependency. Table 6 shows the results of the fixed-effects model when applied to the decision to lift, decrease, maintain, or increase economic sanctions.
Table 6:
The Decision to Alter Economic Sanctions
| Variable | Coefficient | SE | P-Valu e* |
| magnitude | 0.431 | 0.129 | 0.001 |
| escalate | -1.503 | 0.718 | 0.018 |
| provocation | -0.527 | 0.142 | 0.000 |
| approval | -0.030 | 0.016 | 0.030 |
| election year | -2.542 | 1.404 | 0.035 |
| electoral approval | 0.038 | 0.025 | 0.064 |
| advise | 0.633 | 0.424 | 0.068 | < /tr>
| trend | 0.115 | 0.010 | 0.000 |
Pseudo R2 = 0.455
* 1-tailed Test
Three differences in the President’s decision calculus are apparent. First, approval has a significant effect on the President’s decision to alter sanctions, indicating that the greater a president’s approval, the more likely he will lift or decrease the sanctions. Second, electoral approval is insignificant. Third, the President’s advisory system has no effect on the decisions to alter economic sanctions. advise’s insignificance most likely results because the President and his advisors have developed a set of contingencies to handle the target’s different reactions, such as acquiescence, belligerence, and so forth. Once the sanctions are in place, the President already has decided what to do, given certain target reactions to the sanctions. Therefore, the President no longer needs to consult, or at least consults less with his advisors, washing out advisory system effects.
How the President Alters Sanctions Against Different Nations
Like the decision to use economic sanctions, the fixed-effects model is significant, indicating that the decision to alter sanctions is not constant across all targets. To determine if these differences are individual or group effects, I follow the same procedure used above. The differences between the coefficients from the two fixed-effects models appear in Table 7 below. Unlike the decision to use economic sanctions, latin and/or communist do significantly alter the coefficients (p < 0.0349). The results show that the differences between the fixed-effects models can be attributed, at least in part, to either the target nation’s location and/or its ideology. That is, either latin or communist (or both) replaces the individual dyad effects on the president’s decision to lift, decrease, maintain, or increase sanctions.
Table 7:
Dummy Variable Coefficient Differences.
The Decision to Alter:
| Variable | Dummy Variables Only | Dummy Variables with latin and communist |
| Argentina | 0.339 | -0.338 |
| Brazil | -2.433 | -3.079 |
| Ceylon | -4.112 | -3.945 |
| Chile | -0.996 | -1.657 |
| China | -1.417 | -4.042 |
| Cuba | -0.877 | -4.338 |
| Dominican Republic | -2.607 | -3.338 |
| Egypt | -1.492 | -1.313 |
| Ethiopia | -1.452 | -4.026 |
| India | -0.663 | -0.465 |
| Indonesia | -0.245 | -0.033 |
| Iran | -5.385 | -5.248 |
| Kampuchea | -2.411 | -2.845 |
| Laos | -4.078 | -4.564 |
| North Korea | -1.853 | -4.467 |
| North Vietnam | -1.940 | -4.564 |
| Pakistan | -2.134 | -1.884 |
| Peru | 0.199 | -0.439 |
| Uganda | -2.016 | -1.851 |
| USSR | -0.532 | -3.133 |
P < 0.003 (2-tailed Test)
Because Table 7 indicates that at least one of the variables has an effect, I now estimate the model with both latin and communist to interpret their significance and coefficients. The results appear in Table 8. The estimation shows that latin is significant, but communist is not. Like the standard fixed-effects model (Table 6), both advise and electoral approval become insignificant, and approval becomes significant. The fact that those effects appear in both the fixed-effects model as well as the model with latin is more evidence that latin specifies the differences found in the standard fixed-effects model.
Table 8:
The Decision to Alter Economic Sanctions with Country Type Variables
| Variable | Coefficient | SE | P-Valu e* |
| magnitude | 0.320 | 0.108 | 0.002 |
| escalate | -1.179 | 0.666 | 0.039 |
| provocation | -0.451 | 0.133 | 0.001 |
| approval | -0.028 | 0.015 | 0.029 |
| election year | -2.370 | 1.293 | 0.034 |
| electoral approval | 0.037 | 0.023 | 0.056 |
| advise | 0.486 | 0.340 | 0.076 | < /tr>
| communist | 0.278 | 0.377 | 0.231 |
| latin | 0.654 | 0.370 | 0.039 |
| trend | 0.109 | 0.009 | 0.000 |
Pseudo R2 = 0.406
* 1-tailed Test
All of the coefficients remain the same as the previous analyses. The one interesting and unexpected discovery is the positive effect latin has on the president’s decision. The positive sign means that once the president levies sanctions against a Latin American country (something he is less likely to do in the first place), he is more likely to maintain or increase those sanctions. But which is it? Does the president tend to maintain or increase the sanctions more often?
The mean length for sanctions levied against Latin American nations is 14.5 months, a significantly lower mean (p < 0.000) than the 36.5 months for other nations. Further, once a Latin American nation was sanctioned, the president added, on average, 2.7 more sanctions. This mean is significantly (p < 0.000) greater than the 1.2 sanctions the president added to other nations. 14 Thus, the positive sign in Table 8 suggests not that the president is maintaining sanctions against Latin American nations, but that he is increasing their intensity.
The story these findings tell seems to have face validity. The president tends to sanction non-Latin American nations more often and for longer periods of time. However, once he decides to sanction a Latin American nation, he does so with relatively greater intensity. This tendency to increase the intensity is probably the President’s attempts to compensate for the brevity of the sanctions. Thus, U.S. sanctions directed at Latin America can be characterized as slightly less frequent, short, but intense.
In sum, the target’s location affects the President’s decision to lift, decrease, maintain, or increase sanctions. Once sanctions are in place, the only additional effect outside the three models (coercive diplomacy, domestic imperatives, and procedural rationality) is the greater likelihood that the President will make sanctions against a Latin American target short and intense.
Why the President Alters Economic Sanctions Against a Specific Nation
I now turn to the question of whether the location and/or ideology affect the reasons that the President alters economic sanctions. Like in the analysis above, I use latin and communist to select the cases for the three different estimations. These three estimations will reveal the reasons the President sanctions different types of countries. The results appear in Table 9 below.
Table 9:
The Decision to Alter for latin, communist, and neither.
| Neither | Latin American Nations | Communist Nations | ||||
| Variable | Coefficient | SE | Coeffi cient | SE | Coefficient | SE |
| magnitude | 0.377 | 0.177 | 0.554 | 0.258 | 0.389 | 0.237 |
| escalate | -0.554 | 1.013 | -3.031 | 1.058 | 0.635 | 2.084 |
| provocation | -0.256 | 0.187 | -0.47 0 | 0.287 | -1.481 | 0.417 |
| approval | -0.004 | 0.030 | -0.050 | 0.036 | -0.080 | 0.033 |
| election year | -1.622 | 2.192 | -2.445 | 2.795 | -5 .432 | 2.718 |
| electoral approval | 0.016 | 0.039 | 0.043 | 0.050 | 0.102 | 0.048 |
| advise | 0.540 | 0.510 | -0.541 | 0.811 | 1.431 | 0.784 |
| trend | 0.054 | 0.010 | 0.204 | 0.187 | 0.024 | |
Bolded Italics = Significant p < 0.05
The President’s decision calculus changes from his decision to use economic sanctions for all three types of targets (neither Latin American or communist, Latin American, and communist). For the non-Latin American and non-communist countries, the President’s reasons for altering sanctions is very similar to his decision to use them. The only change is that he no longer considers escalation and makes all of his decisions on the tension between the U.S. and the target. This finding suggests that domestic political motives never enter the President’s calculus for this type of nation. Interestingly, the President can only be deterred when first using sanctions. Once the sanctions are in play, the target’s actions cannot compel the President to decrease or lift the sanctions.
When only Latin American nations are run in the analysis of the President’s decisions to alter sanctions, only the international variables (magnitude, escalate, and provoke) are significant. The President does not consider domestic imperatives once economic sanctions are in place against a Latin American nation. In his initial decision use, the President considers the domestic arena as an important part of his decision because U.S. interests in the region are so great. However, once those sanctions are in place, the President only considers the relations with the target. Because sanctions against Latin American countries tend to be shorter, there is little time for the domestic political agenda to change, making the President’s decision to alter the sanctions should be unaffected by domestic politics simply because the domestic sentiments have not changed since the initiation of the sanctions. Further, in such a short time period, the President has little time to read any changes in the domestic political arena. Therefore, he bases all alternations to his sanctions policies against Latin American nations on the relations with the specific target. When the tensions between the target and the U.S. increase, the President is more likely to increase or at least maintain the sanctions. When the Latin American nation becomes provocative or the dispute begins to escalate rapidly, the President will tend to decrease or lift the sanctions.
The President’s concern for the domestic arena increases once he has sanctioned a communist nation. Unlike the brevity of Latin American sanctions, the long lasting sanction against communist countries provide the President ample time to read the many possible changes in the domestic arena. Further, communist nations provide the President with a popular opponent to sanction when his presidency is in need of bolstering. The model shows that the lower the President’s approval the more likely he is to increase or maintain the economic sanctions. This condition only holds during non-election periods. During the eleven months leading up to an election, the President is not only less likely to maintain or increase sanctions. Finally, the President’s advisory system still affects only sanction decisions against communist countries. This finding supports the idea that the President consults his advisors primarily when deciding to use and alter sanctions against a communist nation but not when alter sanctions against other types of nations.
Conclusion
This paper shows that the U.S. President’s sanction decisions vary by the type of nation he is sanctioning. The three types of nations (Latin American, communist, and neither Latin American or communist) all have different decision calculi, and these calculi change from the decision to use to the decision to alter sanctions.
Specifically, the President sanctions non-Latin American non-communist nations for international reasons. Only the coercive diplomacy variables magnitude and escalate are significant when the President decides to use economic sanctions, indicating that he is only concerned with the relations with the target nation and not domestic politics; nor is the affected by his advisory system when sanctioning these nations. Once he sanctions, the President cannot be compelled by the target to decrease or lift the sanctions. It is encouraging to find that the President decides to use economic sanctions on the basis of the relations with the target and not domestic political considerations. That is, while contemplating economic sanctions against one of these nations, the President seems to concern himself with the dispute in which the U.S. and the target are embroiled and not what the sanctions would mean for his domestic political standing.
On average, the U.S. has had fewer sanctions against Latin American nations, but this difference is explained by the decision models rather than a independent bias against using sanctions on Latin American nations. When deciding to sanction these countries, the President tends to consider both international and domestic factors. This finding is contrary to the trend of sanctioning other nations for only international reasons. The greater U.S. interests in the region clearly lead the President to be more concerned with domestic considerations.
This tendency to consider domestic politics ends once the sanctions are in place. The President no longer concerns himself with the domestic sentiment and proceeds to base any changes in his sanctions policy solely on the relations with that Latin American nation. Considering only relations with the target, the President is significantly more likely to increase the severity of sanctions aimed at a Latin American nation. These Latin American sanctions tend to be shorter and more intense than other countries, indicating a positive independent bias toward increasing sanctions against Latin American nations.
In essence, the President begins sanctions against Latin American nations by relying relatively heavily on domestic considerations. Once these sanctions are in place, the President begins increasing the sanctions against the Latin American target while only considering the relations with that target in his decision calculus. While these sanctions do tend to be more intense, they also tend to be shorter, indicating that the President is unwilling to engage in a drawn out economic sanctions campaign because the sanctions themselves will erode American influence in the region as a whole as well as for the sanctioned country.
Finally, the President sanctions communist nations for a variety of reasons. While there is no independent bias toward sanctioning communist nations, the President does sanction communist nations for reasons different from those for other nations. When deciding to use sanctions against a communist nation, the President considers international factors and election periods. Further, it is only when sanctioning communist countries that the President’s advisory system affects his decision. This result is probably due to the heightened importance of relations with communist nations during the cold war. The threat of possible escalation with the Soviet Union or China, or the possibility of drawing them into a conflict with a smaller communist state, would make the decision to use coercive diplomacy against one of these nations a more weighty decision. As such, the President would be more likely to confer with his advisors, making the type of advisory system more likely to affect the decisions to use and alter sanctions against a communist nation.
Once the President has levied sanctions against a communist country and must decide what to do next, his decision calculus changes slightly. He still reacts to the relations with the target and is still influenced by his advisory system, but the President becomes much more concerned with domestic politics. Instead of simply considering elections when deciding to use sanctions, the President’s decision to alter the sanctions is affected by both his approval and electoral approval. It seems quite possible that once the sanctions are in place against a communist nation, the President is willing to use more sanctions against that target to bolster his approval or increase his electoral position. That is, the President can use these already sanctioned communist countries as examples of his strong anticommunist commitment, a popular stance in the 1948-1978 period.
Overall, the findings brought forth in this paper are interesting and provocative. No other analyses have addressed the questions of who and why the President sanctions. These findings about the President’s decision-making process against communist nations exemplifies the need for (1) more data so that both the Reagan and Bush Presidencies can be included, and (2) archival research to learn if indeed the characteristics shown in the quantitative data hold up in actual decision making situations.
References
Barber, J. 1979. "Economic Sanctions as a Policy Instrument." International Affairs. 55: 367-384.
Beck, N., J. Katz, and R. Tucker. 1997. "Beyond Ordinary Logit: Taking Time Seriously in Binary Time-Series—Cross-Section Models." Paper presented at the 1997 annual American Political Science Association meeting, Washington, D.C.
Cortright, D. and G. Lopez. 1995. Economic Sanctions: Panacea of Peacebuilding in a Post-Cold War World? Boulder: Westview Press.
Dashti-Gibson, J., P. Davis, and B. Radcliff. 1997. "On the Determinants of the Success of Economic Sanctions: an Empirical Analysis." American Journal of Political Science. 41:608-618.
Deese D. 1983. "The Vulnerability of Modern Nations: Economic Diplomacy in East-West Relations." in Dilemmas of Economic Coercion: Sanctions in World Politics, edited by M. Nincic and P. Wallensteen, pp. 155-182. New York: Praeger.
Doxey, M. 1971. Economic Sanctions and International Enforcement. London: Oxford University Press.
Doxey, M. 1980. Economic Sanctions and International Enforcement. London: Oxford University Press.
Doxey, M. 1987. International Sanctions in Contemporary Perspective. London: Macmillan Press Ltd.
Drury, A. 1996. "Economic Decisions and Presidential Decisions: Contending Models of Political Rationality." Paper presented at the annual meeting of the International Studies Association, San Diego, CA.
Drury, A. 1997. Economic Decisions and Presidential Decisions: Models of Political Rationality. PhD dissertation. Arizona State University.
Drury A. Forthcoming 1998b. "Revisiting Economic Sanctions Reconsidered." Journal of Peace Research.
Eland, I. 1995. "Economic Sanctions as Tools of Foreign Policy." in Economic Sanctions: Panacea of Peacebuilding in a Post-Cold War World? edited by D. Cortright and G. Lopez. Boulder: Westview Press.
Galtung, J. 1967. On the Effects of International Economic Sanctions, with Examples from the Case of Rhodesia. World Politics. 19:378-416.
George, A. 1991. Forceful Persuasion: Coercive Diplomacy as an Alternative to War. Washington D.C.: United States Institute for Peace.
Green, J. 1983. "Strategies for Evading Economic Sanctions." in Dilemmas of Economic Coercion: Sanctions in World Politics, edited by M. Nincic and P. Wallensteen, pp. 61-83. New York: Praeger.
Hufbauer, G., J. Schott, and K. Elliott. 1990. Economic Sanctions Reconsidered: History and Current Policy. Washington, D.C.: Institute for International Economics.
Kaempfer, W. and A. Lowenberg. 1992. International Economic Sanctions: A Public Choice Perspective. Boulder: Westview Press.
Lindsey, J. 1986. "Trade Sanctions as Policy Instruments: A Re-examination." International Studies Quarterly. 30:153-173.
Morgan, C. 1995. "Clinton’s Chinese Puzzle: Domestic Politics and the Effectiveness of Economic Sanctions." Issues and Studies. 31:19-45.
Morgan, C. and V. Schwebach. 1996. "Economic Sanctions as an Instrument of Foreign Policy: The Role of Domestic Politics." International Interactions 36:25-52.
Morgan, C. and V. Schwebach. 1997. "Fools Suffer Gladly: The Use of Economic Sanctions in International Crises." International Studies Quarterly. 41: 27-50.
Nincic, M., and P. Wallensteen. editors. 1983. Dilemmas of Economic Coercion: Sanctions in World Politics. New York: Praeger.
Renwick, R. 1981. Economic Sanctions. Rensselaer: Hamilton Printing Company.
Simon, H. 1985. "Human Nature in Politics: The Dialogue of Psychology with Political Science." American Political Science Review, 79: 293-304.
Simon, M. 1996. "When Sanctions Can Work: Economic Sanctions and the Theory of Moves." International Interactions. 21: 203-228.
Smith, A. 1996. "The Success and Use of Economic Sanctions." International Interactions. 21: 229-245.
StataCorp. 1997. Stata Statistical Software: Release 5.0. College Station, TX: Stata Corporation.
Wallensteen, P. 1968. "Characteristics of Economic Sanctions." Journal of Peace Research. 3: 248-267.
Notes:
Note 1: For scholars studying effectiveness see Drury 1998, Dashti-Gibson, Davis, and Radcliff 1997, Deese 1983, Doxey 1971, 1980, 1987, Eland 1995, Galtung 1967, Green 1983, Hufbauer, Schott, and Elliott 1990, Morgan 1995, Morgan and Schwebach 1996, 1997, Nincic, and Wallensteen 1983, Renwick 1981, Simon 1996, Smith 1996, and Wallensteen 1968. For scholars studying the reasons sanctions are used see Barber 1979, Lindsey 1985, and Kaempfer and Lowenberg 1992. For a complete review see Drury 1997. Back.
Note 2: According to the model, regardless of the tension level, no president with approval below 45% during an election period will use economic sanctions. Back.
Note 3: Holding all other variables at their mean, presidents with an informal advisory system will use economic sanctions when the magnitude of the tension level reaches 11.2, whereas a formal president requires a level of 12.5. Back.
Note 4: These dummy variables are not shown in Table 2 for presentation reasons. Back.
Note 5: The Latin American nations are as follows: Argentina, Brazil, Chile, Cuba, Dominican Republic, Nicaragua, and Peru. Back.
Note 6: The communist nations are as follows: China, Cuba (1959-1978), Ethiopia (1974-1978) , Kampuchea (1975-1978), Laos (1975-1978), North Korea, North Vietnam, and the USSR. Back.
Note 7: This final significance test is only valid once I determine that the variables may have an effect independent of the individual dyads. Back.
Note 8: The probability reported in Tables 3 is based on a two-tailed Student’s t distribution. Independent significance tests were run for Latin American targets and communist targets. In both cases, the results were insignificant. Unfortunately, because the collinearity problem persisted, some dyads had to be removed during estimation. Back.
Note 9: These percentages are derived from the mean number of months sanctions were in place against the two types of targets. Back.
Note 10: This finding supports the decision to code only openly communist regimes. Because the dispute’s magnitude accounts for any effects communist may have, it also accounts for any effects had the variable been coded for leftist leaning regimes. Back.
Note 11: The difference remains significant with or without communist nations as part of the non-Latin American data. Back.
Note 12: Non-Latin American and non-communist nations include: Ceylon, Cuba (1948-1958), Egypt, Ethiopia (1948-1973), Kampuchea (1948-1974), Laos (1948-1974), Uganda, India, Indonesia, Iran, Pakistan, and South Africa. Back.
Note 13: The four nations sanctioned for expropriation issues are: Brazil, Chile, Ethiopia, and Peru. Back.
Note 14: If Cuba is excluded from the analysis, the mean length of U.S. sanctions is only 6.4 months and the number of sanctions added is 1.5. Both of these differences remain statistically significant (p <0.000). Back.
Email this citation