Balancing may be everywhere, but the proportionality test is not

Abstract

The relationship between balancing and proportionality has not always been clear. Because part of the literature falls short of adequately differentiating between the two tools, many people have become conditioned to see an instance of proportionality whenever the word ‘balancing’ is dropped. As a consequence, the ubiquity of balancing brought about the feeling that proportionality is equally ubiquitous. In this article, I show that the proportionality test is necessarily linked to judicial review and how this link is key to understanding why not every instance of balancing is part of the proportionality test and that proportionality cannot be as ubiquitous as many have claimed. This has not only analytical relevance, but also institutional consequences.

I. Introduction

Mary and John claim they have a constitutional right to do x and y. However, Mary’s exercise of the right to x is (at least partially) incompatible with John’s exercise of the right to y. This controversy reaches the judiciary. The court recognizes that there is a collision between the rights to x and y and, by resorting to the proportionality test, settles the matter in favour of Mary. John is therefore prevented from exercising the right to y, and the proportionality test is to be blamed.

Following at least part of the literature on proportionality, ¹ this seems to be a recurring situation. The problem is only that, even though the court might have resorted to balancing to decide the case, the proportionality test could have barely been applied in such a situation. Because part of the literature on proportionality and balancing falls short of adequately differentiating both tools, people have become conditioned to see an instance of the proportionality test whenever the word ‘balancing’ is dropped. The aim of this article is to provide an alternative point of view on the matter.

There are at least two ways of misunderstanding the relationship between balancing and the proportionality test. The first is to suggest that balancing is all that matters in the proportionality test. The second is to fail to recognize that balancing in legal argumentation performs other tasks, which have nothing to do with the proportionality test. Settling the first misunderstanding demands demonstrating that the proportionality test is more than just balancing, among other things because the test encompasses other (previous) steps, such as the suitability and necessity tests. Settling the second misunderstanding demands showing that balancing is more than the proportionality test because it is used in situations in which the proportionality test has no application whatsoever.

This article is the third in a series dedicated to balancing and the proportionality test. The first misunderstanding mentioned above was addressed in a previous article. ² Part of the second misunderstanding – related to the different roles played by balancing in legal argumentation – has also already been addressed elsewhere. ³ I have argued that balancing plays at least three different roles in the application of legal rules: (1) it may be part of the interpretive process of legal provisions (interpretation); (2) it may be employed as a justification for teleological reduction or analogical extension (correction); and (3) it may be part of a broader procedure – the proportionality test – aimed at reviewing the compatibility of an ordinary rule with a constitutional provision (judicial review). ⁴ The link between proportionality test and judicial review is key to understanding (1) why not every instance of balancing is part of the proportionality test, and (2) that the proportionality test cannot be as ubiquitous as many have claimed (because it is limited to cases of judicial review). ⁵

To substantiate these claims, this article is organized as follows. In Part II, I present and analyse three judicial decisions that have been used as examples of balancing and proportionality; my aim is to highlight that balancing plays different roles in those decisions and that proportionality is actually employed in only one of them. The analysis of those concrete examples will pave the way towards Part III, which is the core of the article. I will show that there is a necessary link between the proportionality test and judicial review because all prongs of the test presuppose the existence of a ‘means’ (or ‘measure’ or ‘act’), namely of a state action that is being reviewed. Conversely, judicial decisions that do not involve judicial review cannot be based on the proportionality test. In Part IV, I show that even though it has become common to speak of different versions of the proportionality test, this variety does not affect the conclusions reached in Part III. In Part V, I argue that the conclusions of the article have not only analytical and methodological relevance, but also important practical and institutional consequences.

II. Balancing may be everywhere, but the proportionality test is not

There are countless examples of judicial decisions based on balancing and on the proportionality test, but some have been used more frequently to show the formal structure of those tools and, more importantly, to highlight their strengths or weaknesses. I begin this article with a very short analysis of three decisions that have been used as examples of balancing and proportionality, both by supporters and by critics. I will return to these examples again in this article.

Balancing may be everywhere …: the cigarette and the officer

Robert Alexy is probably the most influential contemporary advocate of balancing and proportionality in constitutional adjudication. Indeed, balancing lies at the heart of his seminal work A Theory of Constitutional Rights. ⁶ Moreover, since the publication of the book, he has dedicated a considerable part of his work to further developing and refining his concept of balancing. ⁷

One of his central claims is that balancing is always based on the same formal structure, which is summarized by his law of balancing and by his weight formula. The law of balancing was already in the original version of his theory of constitutional rights and states the following: ‘The greater the degree of nonsatisfaction of, or detriment to, one right or principle, the greater must be the importance of satisfying the other.’ ⁸ The weight formula was a later development and is an attempt to use arithmetic elements to increase the rationality of balancing. ⁹

My argument here is agnostic to his claim that balancing always has the same formal structure. I argue that, irrespective of whether the structure of balancing is constant, the roles that different instances of balancing play may be very different. To make my point clear, I will start with the very same cases which Alexy has used to illustrate the application of both his law of balancing and the weight formula: two decisions from the case law of the German Constitutional Court. The first is the decision concerning the duty of placing health warnings on tobacco products. ¹⁰ The second is the trial against a German satirical magazine (Titanic), which had called a paraplegic reserve officer of the German armed forces a ‘born murderer’ and a ‘cripple’. ¹¹ Alexy has used both decisions to demonstrate that balancing is a rational procedure. ¹² Here I will use them with another aim: to make explicit the different roles that different instances of balancing play. I will attempt to show that it is inadequate to treat them as if they were alike and that failing to take their differences seriously may have far-reaching consequences.

When the German Constitutional Court decided on the duty of placing health warnings on tobacco products, the court was clearly performing judicial review. The plaintiffs challenged the constitutionality of some provisions of the German regulation on the labelling of tobacco products. If their claim had been successful, the challenged provisions would have been declared unconstitutional. The German Constitutional Court engaged in balancing as part of a broader procedure: the proportionality test. Thus, the court first assessed whether the legal rule that established the duty of placing health warnings on tobacco products was suitable and necessary (steps 1 and 2 of the proportionality test). ¹³ Since the provision was considered both suitable and necessary for fostering the aim of protecting the health of actual and potential smokers, the third and last step of the proportionality test – balancing – took place. The court decided that the obligation to display warning notices did not exceed the threshold of what was reasonable. ¹⁴ Had the court decided otherwise, the provisions of the German regulation on the labelling of tobacco products would have been declared unconstitutional. The proportionality test – and therefore the balancing that is part of it – was used as a tool for judicial review.

The balancing that took place within the second decision, concerning the trial against the satirical magazine Titanic, was a tool aiming at a completely different goal. At stake was the imposition of financial compensation for pain and suffering, granted by a regional court, to be paid by the magazine to the reserve officer. The German Constitutional Court did not perform judicial review. Therefore, in that case, balancing was not the third step of the proportionality test (indeed, the proportionality test is not even mentioned in that decision). The constitutionality of the legal rule that provides for the possibility of financial compensation for pain and suffering was not challenged at all. ¹⁵ Balancing was used for a different purpose: to calibrate the scope of the legal provision to avoid an undue restriction to the freedom of expression and of the press. ¹⁶

… but the proportionality test is not: Otto-Preminger-Institut v Austria

At least since Tsakyrakis raised charges against the proportionality test, ¹⁷ the decision of the European Court of Human Rights (ECtHR) in Otto-Preminger-Institut v. Austria ¹⁸ has been considered a paradigmatic decision and used both by critics and supporters of the proportionality test to justify and illustrate their claims. ¹⁹ Therefore, it seems to be a good idea to analyse this decision because – contrary to what has been assumed by many, and for reasons that will be fleshed out in the course of this article – it is highly questionable whether the proportionality test was applied in that decision at all.

The controversy that led to this decision may be summarized as follows. After a complaint by the Catholic Church, the public prosecutor in Innsbruck, Austria requested the seizure of the film Das Liebeskonzil, which was programmed for exhibition by the Otto-Preminger-Institut (OPI). The OPI advertised the film as follows: ‘Trivial imagery and absurdities of the Christian creed are targeted in a caricatural mode and the relationship between religious beliefs and worldly mechanisms of oppression is investigated.’ ²⁰ The prosecutor argued that showing the film would be a violation of section 188 of the Austrian Criminal Code (‘disparaging religious doctrines’) and requested its seizure based on section 36 of the Austrian Media Act. The application was granted by the Innsbruck Regional Court. As a result, the OPI could not exhibit the film. Appeals to other courts were dismissed.

The case reached the ECtHR, which then asked: (1) whether there was interference with freedom of expression; (2) whether the interference was prescribed by law; (3) whether the interference had a legitimate aim; and (4) whether the seizure and the forfeiture were necessary in a democratic society. ²¹ It is questionable whether these questions – which are standard in decisions of the ECtHR ²² – amount to an actual application of the proportionality test. The ECtHR did check whether the ascertained interference with freedom of expression complied with Article 10, paragraph 2 of the Convention. However, to do so the court did not assess whether section 188 of the Austrian Criminal Code or section 36 of the Austrian Media Act themselves were compatible with the Convention but rather whether these domestic provisions were rightly or wrongly applied by the Austrian courts.

Balancing may have been useful to assess whether these domestic statutes were wrongly applied, and therefore whether the seizure or the forfeiture of the film violated the Convention. Indeed, the ECtHR argued that, ‘The Innsbruck courts had to strike a balance between the right to artistic freedom and the right to respect for religious beliefs’ and that this balancing was not wrongly performed. ²³

Not surprisingly, the proportionality test is not even mentioned by the ECtHR. It is true that the court occasionally refers to the general ‘idea of proportionality’ between means and ends. However, it is the Austrian Media Act itself that establishes such a demand, when it provides that the court ‘may order the seizure of the copies intended for distribution to the public of a work published through the media … if the adverse consequences of such seizure are not disproportionate to the legitimate interests served thereby. Seizure may not be effected in any case if such legitimate interests can also be served by publication of a notice concerning the criminal proceedings instituted.’ ²⁴ Assessing whether this provision was correctly interpreted and applied may surely involve balancing (among other interpretive tools), but not an application of the proportionality test. The proportionality test would have been applied if this provision itself had been challenged before the judiciary. However, such a judicial review was not at stake. What was at stake was how to interpret Austrian domestic legislation in harmony with the European Convention and whether the Austrian courts were successful in this task.

Therefore, using the Otto-Preminger-Institut v. Austria decision as a (good or bad) example of the application of the proportionality test is futile. There is no need to defend the proportionality test against objections that are actually directed against balancing as an interpretive tool, not as the last stage of the proportionality test. One could, of course, argue that my conclusion is based merely on terminological hair-splitting and that Tsakyrakis’s charges against proportionality (as well as Khosla’s, Klatt and Meister’s, and Möller’s attempts to defend it) are not affected by it. However, this is not true.

There is no doubt that it would be possible to argue that the Austrian courts and the ECtHR struck the wrong balance when they interpreted both Austrian domestic legislation and the European Convention. One could further argue that this wrongful balancing resulted in an undue restriction on the plaintiff’s freedom of expression. One could even conclude – using Tsakyrakis’s provocative words — that this result was an ‘assault on human rights’. But the proportionality test had nothing to do with it. ²⁵

On the contrary: the proportionality test could actually have been used to avoid undue restrictions in the plaintiff’s freedom of expression. Tsakyrakis argued that, ‘Religious sentiment and freedom of expression can never be put on the scale, whatever we take that scale to be.’ ²⁶ If this is true, then there is only one way out: the declaration of the unconstitutionality of section 188 of the Austrian Criminal Code (or its incompatibility with the Convention) because it is this provision that chose to put religious sentiment and freedom of expression on a scale in the first place by criminalizing the disparagement of religious doctrines. The tool for checking whether this provision of the Austrian Criminal Code is unconstitutional or contrary to the Convention could have been the proportionality test. ²⁷ But the ECtHR decided differently.

III. The necessary link between the proportionality test and judicial review

In works from both advocates and critics of proportionality, there is a widespread consensus on the structure of the test. According to Rivers, the proportionality test entails the following steps and questions:

(1) Legitimacy: does the act (decision, rule, policy, etc.) under review pursue a legitimate general aim in the context of the right in question? (2) Suitability: is the act capable of achieving that aim? (3) Necessity: is the act the least intrusive means of achieving the desired level of realisation of the aim? (4) Fair balance, or proportionality in the narrow sense: does the act represent a net gain, when the reduction in enjoyment of rights is weighed against the level of realisation of the aim? ²⁸

According to Urbina, the stages may be characterized as follows:

a. Legitimate aim: the measure interfering with the right has to have an objective of sufficient importance; b. Suitability: the measure interfering with the right has to be rationally connected to the legitimate aim; c. Necessity: the measure should impair as little as possible the right in question; d. Proportionality stricto sensu: there must be a proportionality between the effects of the measures which are responsible for limiting the right, and the objective which has been identified as of sufficient importance. ²⁹

Similar definitions may be found in almost every work on proportionality. ³⁰ Sometimes the first stage (legitimate aim) is not included in the definition ³¹ because it may be considered a result rather than part of the test. However, this is a minor variation that is not relevant for the present discussion.

In all descriptions of the stages of the proportionality test, there seems to be no doubt on what is being tested: it is an ‘act’, a ‘means’ or a ‘measure’. In the definitions quoted above, we find questions and assertions such as ‘does the act …?’, ‘is the act …?’, ‘the measure … has to’ or ‘the measure should …’ One may find similar expressions everywhere, such as ‘the means must …’, ³² ‘whether the act is capable of …,’ ‘whether the act impairs …’ ³³

However, what exactly are these ‘acts’, ‘means’ or ‘measures’? They are not individual acts but state actions. ³⁴ More specifically, they are usually a piece of legislation – sometimes an executive act. ³⁵ In this sense, in actual applications of the proportionality test, one has to replace those general terms (‘means’, ‘measures’, ‘acts’) with specific terms such as ‘legal provision p’ ‘article n of statute s’ ‘section x of statute y’, among others. Legal provisions, articles or sections of a statute (the means) are subjected to the proportionality test for a very specific reason: because they allegedly restrict a constitutional right. To ask whether those means are suitable, necessary and well balanced is tantamount to testing their compatibility with the constitution. The proportionality test is therefore judicial review.

However, since legal debates are frequently quite court-centred and since judicial decisions are undoubtedly also state actions, it is surely legitimate to ask why judicial decisions cannot be considered state actions in the context of the proportionality test. If the proportionality test aims at controlling acts that restrict fundamental rights and if fundamental rights may supposedly be restricted not only by legislative and executive acts but also by judicial decisions, why could these very decisions not be subjected to the proportionality test? This question has actually not been asked. The literature on proportionality seems to assume that there is no doubt that judicial decisions of lower courts are acts whose proportionality may be tested by higher courts (or, similarly, that the proportionality of decisions of domestic courts may be tested by supranational courts). In the next sections, I attempt to show why this assumption is unsound.

Judicial decision as a means?

The idea that a judicial decision could be considered a ‘means’ in the sense of the proportionality test presupposes the following scenario: someone claims that a judicial decision restricted one of her or his constitutional rights and therefore lodges an appeal to a higher court (an appellate court, a supreme court, a constitutional court or even a supranational court); the higher court thus resorts to the proportionality test by asking the set of questions that this test entails. I will call this set of questions SQ1 to distinguish it from a further set of questions to which I will refer later.

(SQ1) Is the decision of the lower court (assumed as ‘the means’) apt to foster the aim it wants to foster? Is the decision of the lower court the least intrusive means of achieving the desired level of realization of the aim? Does the decision of the lower court represent a net gain when the reduction in enjoyment of rights is weighed against the level of realization of the aim?

Although ‘the means’ in SQ1 is identified (it is ‘the judicial decision of the lower court’), this is done in a very general way. There is no hint of its content. SQ1 more closely resembles a textbook description of the test (Is the means suitable? Is the means necessary? Is the means well balanced?) than a concrete proportionality analysis. Once we leave this level of methodological generality and attempt to apply the proportionality test to a concrete judicial appeal, problems arise. Suppose that a constitution provides for a right to freedom of information; suppose further that a judicial decision denied the plaintiff access to some public documents. The court based its decision on a statute that regulates access to public documents, and some provisions of this statute allow the government, under very specific circumstances, to classify documents as secret if their publication puts national security in danger. The requirements for classifying a document as secret are very strict. The court understood that these requirements were met, and therefore, the classification as secret was justified.

The plaintiff then appeals to a higher court. It looks like a perfect context for the application of the proportionality test: the decision of the lower court (assumed to be ‘the means’) brought about a restriction in the constitutional right to freedom of information (the impaired right) to foster national security (the aim). If the court tries to apply the proportionality test, it could ask the following questions:

(SQ2) Is the classification of documents as secret a suitable means to foster national security? Is there an alternative means that is less restrictive to the freedom of information and equally effective in fostering national security? Is the degree of restriction in freedom of information justified by the intensity in the promotion of national security?

Although the set of questions SQ2 may be considered a prime example of the application of the proportionality test, these questions are not a concrete application of the more general questions presented before (SQ1). It is easy to notice that in the second set of questions (SQ2), the judicial decision of the lower court does not truly play a relevant role. In fact, the judicial decision of the lower court is not even mentioned. One could object that this is simply a question of wording and that references to the judicial decision of the lower court are implicit in SQ2: a question such as ‘is the classification of documents as secret a suitable means to foster national security?’ should actually be read as ‘is the judicial decision of the lower court that accepted the classification of documents as secret a suitable means to foster national security?’ However, this is unwarranted.

The lower court did one straightforward thing: it assessed whether the requirements for the classification of a given set of documents as secret were met. If an appeal is lodged, the higher court will do exactly the same thing: it will review whether those requirements were truly met. If yes, the decision of the lower court will be upheld; if not, the decision will be overruled. There is no room for the proportionality test in this context. The exception is if the higher court considers that even though the interpretation of the statute and the assessment of the circumstances made by the lower court were correct, restricting the right to access the documents should nevertheless be considered a disproportionate means to foster national security. In this case, there is no doubt that the decision of the higher court could (or even should) be based on the proportionality test. However, what the court would be testing in this case would not be the decision of the lower court (remember, the lower court did everything just right), but the statute itself. This is, not surprisingly, judicial review. This implies – and this is of paramount importance here – that the higher court can engage in such proportionality analysis if it has the authority to perform judicial review of legislation. If it does not, there is no room for the proportionality test. In other words, depending on the institutional design of the justice system, a controversy may be assessed by means of the proportionality test in one country, but it might not in another.

Suppose that the controversy takes place in a country in which only the constitutional court has the power to perform judicial review of legislation. In this country, whenever judges of other courts consider that a given piece of legislation restricts or violates a constitutional right, they must refer the case to the constitutional court. Thus, it is the constitutional court – not a trial judge or a local, regional or appellate court – that may ask whether a given provision of ordinary legislation is suitable, necessary and well balanced – that is, engage in proportionality analysis – simply because a negative answer to any of these questions would entail that the legislative provision is incompatible with the constitution. And only the constitutional court has the power to issue such a declaration of unconstitutionality.

In summary, (a) if neither the trial judge nor any court other than the constitutional court has the power to declare a provision of ordinary legislation incompatible with the constitution, and (b) if it is a feature of the proportionality test that a provision of ordinary legislation might be considered disproportionate and therefore incompatible with the constitution, then (c) trial judges and courts other than the constitutional court cannot resort to the proportionality test.

Judge-made law: The difference between choosing and testing

Are judges never allowed to freely ask which is the best way to deal with a controversy brought before the court? May a judge never ask, ‘Which is the best way to balance freedom of information and national security?’ If yes, why could not this judge use the proportionality test to answer this question?

It does not seem to be contentious to state that it is not the task of judges to unconstrainedly decide which is the best way to deal with all cases brought before them. Nevertheless, there may be cases for which either there is no directly applicable legal rule or there is an applicable legal rule that leads to unjust outcomes. How judges should deal with such situations is a controversial question that has occupied legal scholars for a long time. It is not necessary to take sides in this debate here. For the sake of this article, it is enough to show that even if judges may exceptionally resort to balancing to deal with unforeseen situations or with unjust outcomes, ³⁶ the proportionality test is nevertheless still not an option. ³⁷

The proportionality test is not an option because it is not a tool for finding the solution for a problem but a tool for testing whether one given solution (the means!) is proportional. The proportionality test is not able to point to one single (and supposedly best) solution ³⁸ because, as a tool for assessing whether a given solution is proportional, it delivers a binary outcome: either yes (the means is proportional) or no (the means is not proportional). However, ‘yes/no’ answers are neither equivalent to nor sufficient reasons for reaching ‘the best/the rest’ judgments.

One could still insist that a judge facing an unforeseen controversy for whose solution there is no applicable law could freely come up with ten possible solutions (S1 to S10), test the proportionality of all of them, conclude that four of those solutions (say, S1, S4, S7 and S9) are proportional, and choose one of these four. However, the question remains of how this judge could or should choose one among those four proportional solutions. One thing is certain: the proportionality test cannot be the criterion. The judge could at the most say, ‘Don’t worry, I know that my solution is proportional; I pretested it,’ but she or he could not say, ‘My decision is based on the proportionality test’ or, ‘The proportionality test led me to choose solution S1 instead of solutions S4, S7, and S9.’ ³⁹

The duty to be proportional (or, the ‘idea of proportionality’)

The idea of proportionality is pervasive in law, both in legislation and in adjudication. However, even though the proportionality test is surely grounded in this idea, the two are not synonymous. One reason why the proportionality test is perceived as more ubiquitous than it really is exactly the pervasiveness of the idea of proportionality. The realm of punishment may be the one in which such a general idea of proportionality is most salient. The law should provide for punishments that are proportional to the gravity of the crimes. And courts should do the same when deciding a concrete case: legislation usually defines a range between minimum and maximum penalties, which enables an even more precise realization of the idea of proportionality.

There are similar examples to be found in the realm of private law. Financial compensation for pain and suffering is a good example. In my assessment of the decision of the German Constitutional Court in the Titanic case, ⁴⁰ I argue that the court resorted to balancing but not to the proportionality test. To be sure, there is no doubt that the ‘idea of proportionality’ guided the judges in their judgment: the amount of money to be paid must be proportional to the gravity of the offence. In such decisions, judges do not test the suitability and necessity of the rule that provides for financial compensation for pain and suffering. The suitability and necessity of the means (financial compensation) is presupposed, and the judges simply apply the rule to the concrete case. Because it is the court, and not the legislation, that defines the precise amount of money in each case, judges may resort to balancing to fine-tune their decision. Unless the court performs judicial review of legislation, it does not make sense to ask whether the compensation for pain and suffering is suitable or necessary. To ask whether financial compensation for pain and suffering is necessary, for instance, the court would have to compare this type of compensation with alternative means – after all, this is what the necessity test is all about ⁴¹ – and this is not what judges do when they apply the legal rule that provides for compensation for pain and suffering.

Someone could insist and claim that the judge must ask whether compensation of, say, ten golden coins is suitable and, above all, whether it is necessary. This would entail asking whether there is no alternative amount that could be considered less restrictive and at least equally effective (as mentioned above, this is the rationale that underlies the necessity test). The judge could, for instance, ask whether a compensation of five golden coins would not be just as effective as ten coins and clearly less restrictive. However, such reasoning would be ill-founded because the question whether five golden coins is as effective as ten golden coins demands a complement: effective for what? I argue that this question is out of place within retributive reasoning. It certainly makes sense to ask whether a punishment is more or less effective than others in relation to aims such as deterrence – that is, it makes sense to compare two penalties to define which is more effective in discouraging people from committing an illegal act. However, when deciding a concrete case, a judge does not ask this question for the simple fact that the illicit act has already been committed. When deciding a concrete case, the judge can only compare different measures in relation to retribution. ⁴² And this is what judges usually do. But in this case, the concept of effectiveness, which is at the core of the necessity test, is completely absent. Retributions may be more or less just, but not more or less effective. If a court sentences John to one thousand golden coins in damages for calling Mary a fool, such a decision would certainly be disproportionate and unjust, but it is neither effective nor ineffective as a means of retribution. Attempting to introduce a comparative element in this debate would also be out of place: to ask whether compensation of one thousand golden coins is more or less effective than compensation of, say, one golden coin simply brings us to our starting point: effective for what?

In summary, whenever judges ponder whether the compensation should be x or y golden coins, they are not truly comparing different means. The means has already been defined by the legislation: it is financial compensation. Judges would only fine-tune the amount of money. Deciding which is the right amount of money is an operation grounded in the ‘idea of proportionality’ (the compensation must be proportional to the gravity of the offence). However, the tool for this decision is not the proportionality test but simply balancing.

A last attempt for those who see a full proportionality test (suitability, necessity, balancing) instead of an autonomous balancing in such decisions could start with the following question: what if legislation provides for two or more alternative forms of punishment or compensation? Criminal laws sometimes provide for different penalties for the same crime, such as imprisonment, fines and restitution or community service. In this case, wouldn’t a judge have to compare among different means in order to determine which one is suitable and necessary? In other words, couldn’t a judge ask whether in that specific case, a fine would be just as effective as imprisonment but less restrictive to fundamental rights? I argue that the judge could not do so because if the criminal code provides for alternative penalties, either the criteria for the decision are laid down by the criminal code itself (for instance, if conditions a, b and c are fulfilled, the penalty should be imprisonment; if not, the penalty should be a fine), or the criminal code does not provide any criteria whatsoever, and the judge may choose. In the first case, judges are bound by the criteria laid down by the law, and they do not compare anything; they simply assess whether conditions a, b, and c are met or not. In the second case, one could argue that if judges may choose the penalty, they could compare the alternatives using the proportionality test. However, this possibility has already been rebutted above: the proportionality test is not a tool for choosing among many alternatives but for testing their proportionality. As a test, it is not impossible for all alternatives to be approved. ⁴³

IV. Proportionality test^®, AOC?

Thus far, this article has attempted to show that neither balancing and proportionality test are synonymous nor that the former always implies applying the latter. Nevertheless, one could agree with everything that has been claimed in this article but at the same time maintain that my reasoning is correct in relation to only one of the many versions of the proportionality test. In relation to other possible versions – one could further assert – my efforts to clearly distinguish the proportionality test from autonomous balancing might turn out to be rather irrelevant.

To be sure, it is not possible to control how legal ideas arise, develop, migrate and are received in different places and in different moments. The proportionality test is neither a registered trademark nor submitted to a certification of appellation d’origine contrôlée (AOC). There may be different tools for assessing whether the relation of a means to an end is well-balanced. Some of them are more structured, others less. From this perspective, the three-pronged test assumed here could be considered only one among several other tools called a ‘proportionality test’.

In this sense, one could state, for instance, that I am not able to see an application of the proportionality test in Otto-Preminger-Institut v Austria because I am looking for one type of proportionality test, while the ECtHR employs a different one. Furthermore, one could also claim that I am not able to recognize the problematic aspects that several authors have imputed to the proportionality test because I am fixated on a very specific version of the test, within which these problems may be less salient. In other words: these problems are out there, and one needs only to widen one’s perspective to see them.

Although there is hardly any work on the proportionality test that does not start with exactly the same canonical description of the proportionality test as a three- or four-pronged test, ⁴⁴ let us suppose that there may be six, seven or ten different tools called ‘proportionality test’ out there. Nevertheless, this does not mean that we are allowed to use outcomes of one version of the test to criticize a different version of it.

If there are several versions of the proportionality test, which one is the target when it is argued that proportionality is ‘an assault on human rights’? ⁴⁵ The ECtHR decision in Otto-Preminger-Institut v. Austria, as well as Tsakyrakis’ charges against it, may be useful here again to illustrate how mixing up different versions produces confusing results. Tsakyrakis began his charges against proportionality by defining the test in the very same way as it was in this article – that is, as a rigidly structured, three-pronged test, which he described as follows:

the principle of proportionality consists in a three-prong test that assesses (a) whether a measure that interferes with a right is suitable for achieving its objective, (b) whether it is necessary for that purpose, and (c) whether it burdens the individual excessively compared with the benefits it aims to secure. ⁴⁶

However, all the cases to which he resorted to demonstrate the allegedly deleterious effects of the proportionality test on the protection of human rights come from the case law of the European Court of Human Rights. However, as I argued in the beginning of this article, the ECtHR employs – if at all – a very different version of the test. ⁴⁷ As a matter of fact, in many decisions the Court does not truly resort to the proportionality test, but simply to what I have called autonomous balancing. ⁴⁸ The test that the ECtHR has used entails rather different questions, such as whether the interference with a Convention right is prescribed by law, has a legitimate aim, corresponds to a pressing social need and is necessary in a democratic society. ⁴⁹

This diagnosis necessarily raises the question of whether it is justified to use the case law of the ECtHR to criticize a version of the proportionality test that the Court does not employ. It seems legitimate to argue that every critique and every defence of the proportionality test should comply with two basic requirements: (1) to clearly state which version of the proportionality test is at stake; and (2) to use examples that match that version. We would surely not accept a debate on football that starts by defining it as a ‘team sport played primarily with the feet’ and then refers only to examples of the US National Football League (as we know, American football – even though it is a team sport called ‘football’ – is not played primarily with the feet).

V. Conclusion

Proportionality test is not balancing. And vice versa. In this article, I attempted to highlight why not every instance of balancing in constitutional adjudication is an instance of the proportionality test. Furthermore, I have argued that autonomous balancing and proportionality test play different roles: the first is an interpretive tool; the latter is a tool for judicial review (which of course implies interpreting constitutional and legal provisions, but with a completely different aim).

This clear analytical differentiation paved the way to showing that several objections to the proportionality test – such as those advanced by Tsakyrakis – miss their target. Even though those critiques claim to assume their target is the standard version of the proportionality test as a three-pronged test aiming at assessing the compatibility of a legislative or executive measure with the constitution, they are actually directed at something completely different (in the case of Tsakyrakis, to the case law of the European Court of Human Rights, which does not employ the standard version of the proportionality test).

Thus, in addition to providing an analytical distinction between proportionality test and balancing, this article has also stressed the importance of conceptual coherence in the literature on proportionality. Conceptual coherence does not mean that there can be only one version of the proportionality test. It is a fact that different courts in different countries have applied different variants of the proportionality test. In contrast, it is hard to find scholarly works on the matter that do not assume the standard three- or four-pronged version of the test as their starting point. The consequence is usually a mismatch between conceptual assumptions and the assessment of judicial practice.

Although throughout this article I focused on analytical and conceptual issues, the clear differentiation between autonomous balancing and proportionality test has also important institutional and practical implications. In this article, these implications have remained in the background. I will explore them in a future article, in which I will argue that the link between proportionality test and judicial review – which was the core of this article – should necessarily lead (but has almost never led) to the following question of institutional design: who has the power to perform judicial review? Since any answer to this question must begin with ‘it depends on the jurisdiction’, many claims about the impact of the proportionality test can barely be grounded on purely methodological reasons. Institutional design plays an important role simply because answering the question ‘Who has the power to perform judicial review?’ is tantamount to answering the question ‘Which courts may apply the proportionality test?’ Depending on the jurisdiction, the answers may be: ‘All, many, some, a few, only one, or none.’ To be continued …