6. Data analysis: execution of EAW’s from other Member States
6.13 Quantitative country comparisons: automatic surrender
Surrender vs refusal ratio differences between the three country groups
One of the most interesting comparisons between the Member State groups is whether the ratio of surrenders versus refusals differs between more similar countries and more dissimilar countries. To be able to compute the data on surrenders in a meaningful way, it has been transformed into a categorical variable. Surrenders take on the value of 2, while refusals take on the value of 0. Partial refusals will be intermediately valued as 1. This transformation of the data leads to some reduction in its richness as some cases include partial refusals for instance for 1/3rd or 3/4th of the requests. Nevertheless, the
transformation was necessary to allow for the application of statistical tests for categorical variables in contingency tables. The data is summarized in table 6.3.
Similar countries Intermediate countries Dissimilar countries
Surrender completely allowed 26 30 23 Partial surrender allowed 5 8 7 No surrender allowed 2 6 9
Table 6.3: cross-tabulation of country-groups and surrender ratio
At first glance there does seem to be an increase in the amount of refusals for dissimilar countries, but the differences seem too small to be attributed to a structural relationship. Both the chi-squared and the Fisher’s exact test were applied to confirm this. For expected cell counts below 10, some methodologists have recommended to replace the chi-squared procedure with the safer Fisher’s exact test (see for instance Cochran, 1954, at p.420). As several cells do have lower expected values than 10, both the chi- squared test and a Fisher’s exact test have been applied in order to retain the option of choosing the safer score if the differences between both test results are substantial. The computation of neither test has resulted in a p value significant at the p<0,1 level. The Fisher’s exact reports a p value of 0,332 and the chi-squared reports a p value of 0,316.160 The results are roughly similar, indicating that the chi-squared
test was not adversely affected by the small sample size to a substantial degree. The results from either test therefore support the argument that although there were some differences in turnover speed, postponements of cases and the amount of additional information requests between the three groups of
94 Member States, there is no evidence to support the claim that perhaps the most important output of the Amsterdam court for the EAW system – its ratio of surrenders – also differs per Member State. This furthermore means that the other structural differences, for instance a relatively high amount of cases with insufficient information from some Member States, were mostly surmountable for the court. They did not lead to a structural problem with the amount of cases in which surrender is refused. As such, the finding nuances the earlier observed differences in performance between Member States. Combined with the observation that in a large majority of cases surrender is (partially) allowed, as earlier noted argument that most refusals that did occur seemed justifiable and well-motivated,161 these results support the claim
that the Amsterdam court generally performs well with regard to the automatic surrender element of the EAW.
A surprising result, furthermore, is that the purpose of a case seems to predict whether surrender will take place to reasonable extent. The sample data shows a relatively high amount of surrenders for prosecution cases, while execution of sentence cases are relatively often (partially) refused by the Amsterdam court. The data is summarized in table 6.4. It can be seen that although there are fewer execution of sentence cases present in the sample, these cases have generated the majority of refusals and partial refusals. A chi-square test approximating whether both variables are related confirms this statement at the p<0,05 level.162
Refusals Partial refusals Surrenders Total Purpose Prosecution 4 9 52 65 Execution of sentence 12 11 25 48 16 20 77 113
Table 6.4: cross-tabulation of the purpose of an EAW and surrender vs refusal ratio
One important factor in this issue seems to be the retrial guarantee. In the legal analysis of the sampled cases of the various Member States, several problems were already noted with regard to trials conducted
in absentia and the retrial guarantees required by the Dutch authorities. Early Bulgarian cases were shown to have experienced some translation problems, with the Dutch court not finding guarantees unambiguous enough. For some Polish cases, internal time-bars and court discretion resulted in several cases where Polish prosecutors were unable to provide adequate guarantees. A Belgian case was refused due to earlier experience under extradition law where the Belgian court eventually did not allow a retrial. Thus, a variety of differences in laws and domestic problems resulted in guarantees not being given or insufficient guarantees.
Conversely, the return guarantee seems to function slightly better due to it being partially standardized under the Convention on the Transfer of Sentenced Persons. While many return guarantees were necessary for both prosecution and execution of sentence cases, a large majority of these were seen as sufficient. Under the Convention, two elements are required. Firstly, the Convention requires an unambiguous guarantee that the person will be returned after sentencing. Secondly, an unambiguous
161 See the various country reports earlier in chapter 6.
162 χ² = 11,367, df=2, p=0,03, N=113 (3 cases were excluded due to the unusual presence of a combined execution
95 guarantee that the sentence can be adapted and executed in the Netherlands is required. Most guarantees seem to conform well to this structure and are thus accepted by the Amsterdam court. While the amount of retrial guarantees is unfortunately not high enough for a resilient statistical analysis, it is notable that out of 12 retrial guarantee cases 5 guarantees were not accepted and 1 case concerned the acceptance of a guarantee for one fact while the a guarantee for another fact was not given. For return guarantees the acceptance rate is higher, with 36 out of 38 guarantees being accepted.
Amsterdam court output: content analysis performed by the Amsterdam court
In addition to the direct output in terms of surrenders/refusals, chapter 4 also presented a methodology which utilized a measurement of the content analyses of the Amsterdam court in EAW cases. This resulted in four indicators. The first considers to which extent the Amsterdam court has considered the facts of a case (including facts pertaining the offence, subsequent occurrences and how authorities handled the procedure). The second indicator measures how many paragraphs the Amsterdam court has devoted to analyses of foreign criminal laws. The third measures to what extent an analysis has been devoted to the adequacy of an incoming EAW. Finally, the fourth accumulates these scores into a total amount of paragraphs devoted to analysis indicator. The results for each indicator, split up per group of country, have been summarized in table 6.5.
Facts of a case Laws of
issuing country Adequacy of a warrant Total analysis Similar countries (Sweden, Germany, Finland) N 33 33 33 33 Median 0 0 0 1 Mean 0,85 0,24 0,39 1,48 Intermediate countries (Spain, UK, Belgium N 44 44 44 44 Median 1 0 0 1 Mean 1,05 0,23 0,41 1,68 Dissimilar countries (Poland, Romania, Bulgaria N 39 39 39 39 Median 1 0 0 2 Mean 1,36 0,69 0,18 2,23
Table 6.5: Mean and median scores for the four in-depth analysis indicators
The table shows how most of the data clusters around the low values of 0 and 1, especially for the non- aggregated indicators. Furthermore, the fact that the mean is usually lower than the average score indicates that a relatively large proportion of cases tended towards a low score and that the average score was influenced by a smaller number of high values. These attributes of the data can be seen as an indication that, in general, the Amsterdam court does not apply a large amount of analysis of facts, laws or EAW’s and restricts itself to a position of mutual trust. The court recounts facts provided by issuing countries but does not often analyze them substantively. Furthermore, it is notable all of the three groups of countries seem to be approached relatively similarly by the Amsterdam court. The extremely slight differences between the facts, laws and adequacy indicators means that further statistical analysis would not provide additional insights.
Requests for additional information
In her relatively recent report, Weyembergh (2013, p.26-28) noted that executing authorities may express a degree of mistrust through requests for additional information to issuing authorities. In particular she
96 mentions the examples pertaining to EAW’s being handled by UK authorities, in which UK judges have made requests for additional information for seemingly small details. Thus, Weyembergh argues that such information requests may be harmful for the application of the principle of mutual trust. If applied appropriately, however, they may by contrast be seen as beneficial to the judicial protection of defendants. While the various country reports that were presented earlier in this chapter did not seem to hint at a structural problem with regard to the requests for additional information sent out by the Amsterdam court, it is interesting to see to what extent requests differ per country group. The cross- tabulation is presented in figure 6.6:
Similar countries Intermediate countries Dissimilar countries
No additional information request 32 38 30 Additional information request 1 6 9
Table 6.6: Cross-tabulation of additional information requests in cases and country-group
First of all it is necessary to re-emphasize that these numbers exclude information requests pertaining to additional guarantees on retrials and returns. Furthermore, it must be noted that the amount of cases in which extra information was requested is relatively small at 16 cases. As this also means that the expected value for cells in the cross-tabulation will be lower than 5 for the row listing cases that did require additional information, it is recommended to use the Fisher’s exact test instead of a chi-squared test based on the Pearson’s technique (see, Cochran, 1954, at p.420). Both tests, however, indicate with only a marginal difference between them that the Amsterdam court’s requests for additional information are related to the extent to which another Member State is dissimilar or similar. The Fisher’s Exact test returns a p value of 0,051, which is significant at the p<0,1 level, while the Chi-squared returns a p value of 0,049,163 which just crosses the boundary of significance at the p<0,05 level. Thus, both tests indicate that
the Amsterdam court more regularly confronts Bulgaria, Romania and Poland with requests for additional information than it does the other 6 sampled Member States. Conversely, the surrender relationship between Dutch authorities and Sweden, Finland and Germany seems to perform well on this point, as there is only one request for additional information between them. Explanations for the relatively large amounts of requests made towards dissimilar countries were already implicit in the various country reports. In particular Poland and Bulgaria have been noted to send incomplete warrants or warrants with seemingly ambiguous information. Accordingly, it was necessary for the court to request additional information, in order to be able to grant surrender. Therefore, in contrast to being an indication for mistrust as suggested by Weyembergh (2013) in relation to UK examples, it seems that in the case of the Amsterdam court an effort is being made to allow surrender despite the fact that an insufficient warrant is presented to it. Such requests were necessary to adequately weigh the goals of appropriate judicial protection and automatic surrender. This result can even be said to provide an indication that the requests are, perhaps counter-intuitively, an expression of trust in other authorities; while the information was incorrect or insufficient, the Amsterdam court assumes the underlying request is not unjustified. Instead of refusing surrender altogether it therefore seeks to resolve the conflict by asking for additional information. Simultaneously, the indication that warrants from some countries require additional information more often than others does illustrate a quality difference between the issuing authorities of
97 Member States, which should be resolved in order to achieve a higher degree of automatic surrender and reduce the amount of postponed cases.