Safety Regulations and the Charter

I wrote earlier this week about the decision of the Court of Appeal for Ontario R. v. Michaud, 2015 ONCA 585, which upheld the constitutionality of regulations requiring trucks to be equipped with a speed limiter that prevents them going faster than 105 km/h. The Court found that the regulations could put some truck drivers in danger by leaving them unable to accelerate their way out of trouble, and thus infringed s. 7 of the Canadian Charter of Rights and Freedoms, but that they were justified under s. 1 of the Charter. This is a most unusual outcome ― I’m not sure there a s. 7 violation had ever before been upheld under s. 1 ― and the Court itself suggested that the s. 7 analytical framework set out by the Supreme Court in Canada (Attorney General) v. Bedford, 2013 SCC 72, [2013] 3 S.C.R. 1101 is not well-suited to cases where the constitutionality of “safety regulations” is at issue. In this post, I would like to comment on the role of s. 7 in this and similar cases, and the role of courts in applying the constitution in such circumstances.

* * *

The Court may well be right that the current s. 7 framework is not adequate to deal with “safety regulations” ― at least if it has interpreted that framework correctly. Referring to Bedford, the Court took the position that any negative effect on a person’s security is enough to engage the “security of the person” right protected by s. 7. But I’m not sure that this is really what the Supreme Court meant when it said that “[a]t this stage, the question is whether the impugned laws negatively impact or limit the applicants’ security of the person.” [Bedford 58] Is there no threshold, at least a more-than-deminimis one, for a court to find a s. 7 infringement? Such thresholds exist in jurisprudence on other provisions of the Charter, for example on s. 2(a), where a “trivial or insubstantial interference” with religious freedom, one “that does not threaten actual religious beliefs or conduct,” does not engage the Charter. Admittedly, Bedford says nothing about such a threshold in s. 7, but then neither it nor the other s. 7 cases that come to mind involved situations where the interference with security interests was of a potentially trivial magnitude.

As the Court of Appeal suggests in Michaud, “safety regulations” are likely to create precisely this sort of interference with the security interests, or even the right to life, of people who engage in the regulated activity. The Court explains that it is always possible to say that a more stringent regulation would have prevented a few more injuries or even deaths, so one could argue that the increase in each person’s likelihood of being injured or dying as a result of a somewhat laxer rule is a s. 7 violation. The Court is concerned that acceptance of such arguments will trivialize s. 7, and I agree that this would indeed be disturbing.

But it seems to me that the best response to this problem is to say that a purely statistical increase in the odds of being injured should not count as sufficient to establish the violation of any given person’s rights. In Bedford itself, the courts were able to show how the prostitution-related provisions of the Criminal Code directly and substantially interfered with the security of sex-workers who had to comply with them. The negative impact on their safety was not just statistical; one did not need an actuarial table to see it ― though statistical evidence was used to show the extent of the problems beyond the stories of the claimants themselves.

The Court of Appeal suggests a different approach, which is to treat safety regulations differently from other (perhaps especially criminal) laws, and to take their beneficial effects into account at the s. 7 stage of the analysis, and not only at the s. 1 justification stage as is usually done. In my view, there are two problems with this solution. First, it is inconsistent with the Supreme Court’s longstanding aversion for introducing balancing into the substantive provisions of the Charter. This aversion is justified, not only by the pursuit of coherence, but also by the desirability of putting the onus of proving social benefits on the government.

The other reason I find the creation of a special category of “safety regulations” problematic is that its contours would be uncertain, and would generate unnecessary yet difficult debate. The rules requiring speed limiters in trucks under pain of relatively limited penalties are obvious safety regulations. But it seems like a safe bet that the government would try to bring other rules within the scope of that category if doing this made defending their constitutionality easier, including for example the prostitution provisions enacted, in response to Bedford, as the Protection of Communities and Exploited Persons Act. Of course, the parties challenging these laws would fight just as hard to show that such rules are not really about safety. The uncertainty and the costs of litigation would increase, while the benefits to be gained from this approach are not obvious.

* * *

Now, this whole issue of statistical increases in risk being treated as a violation of s. 7 of the Charter is actually irrelevant to Michaud. That’s not to say the Court should not have brought it up ― I think it did us a favour by flagging it, and we should take up its invitation to think about this problem. Still, the issue in that case is different: it’s not that a safety regulation does not allegedly go far enough, but that it allegedly goes too far. The two possibilities are, of course, two sides of the same coin; they are both possible consequences of the regulator’s preference for a bright-line rule over a standard. The Court is right to observe that there are good reasons to prefer rules to standards (some of the time anyway). And surely the Charter wasn’t supposed to eliminate bright-line rules from our legislation.

However, to speak of the speed limiter requirement as a “bright-line rule” is to miss what is really distinctive about it. Those who challenge the requirement aren’t seeking it be replaced by a standard. They are content with a bright-line speed limit ― provided that they are able to infringe it on occasion (and, one suspects, that they are not prosecuted for doing so). Unlike a speed limit enforced, sporadically and ex-post, by the police, which can be broken if need be, a speed limit enforced permanently and ex-ante by an electronic device cannot be broken at all. In other words, the issue is not simply one of rules versus standards; it’s one of rules whose nature as rules can on occasion be ignored (put another way, rules that can be treated as if they were standards) versus rules that stay rules.

This creates a difficulty for constitutional law. Can a court acknowledge that a rule sometimes needs to be broken? Can a court go even further than that, and say that a rule is constitutionally defective if it doesn’t allow a mechanism for being broken? To say that the legislator is entitled to choose rules over standards does not really answer these questions. As thoughtful and sophisticated as the Michaud opinion is, I don’t think that it really addresses this issue.

That’s too bad, because this problem will arise again, and ever more urgently, with the development of technology that takes the need, and the ability, to make decisions away from humans. Self-driving cars are, of course, the obvious example. As it happens, the New York Times published an interesting story yesterday about the difficulties that Google’s autonomous vehicles run into because they are “programmed to follow the letter of the law” ― and the drivers of other cars on the road are not. Google’s cars come to a halt at four-way stops ― and cannot move away, because the other cars never do, and the robots let them by. Google’s cars keep a safe distance behind the next vehicle on a highway ― and other cars get right into the gap. The former situation might be merely inconvenient, although in a big way. The latter is outright dangerous. What happens if regulators mandate that self-driving cars be programmed so as never to break the rules and it can be shown that this will increase the danger of some specific situations on the road? What happens, for that matter, in a tort claim against the manufacturer (or rather the programmer) of such a vehicle? Michaud gives us some clues for thinking about the former question, though I’m not sure it fully settles it.

* * *

Thinking about constitutional questions that challenges to safety regulations give rise to also means thinking about the courts’ role when these regulations are challenged. In Michaud, the Court took a strongly deferential position, drawing a parallel with administrative law, where courts are required to defer to decisions that are based on the exercise of expert judgment. It noted, however, that “situations, in which a legislature or regulator uses a safety regulation for an improper collateral purpose, or where the regulator makes a gross error, are imaginable.” [152] In these situations, the courts should step in.

I think this is exactly right. Courts must be alert to the possibility that rules that ostensibly aim at health and safety are actually enacted for less benign purposes, and in particular as a result of various public choice problems. Safety rules are attractive to those who want to limit competition precisely because they look so obviously well-intentioned and are difficult to criticize. That said, when ― as in Michaud itself ― there seems to be no dispute that the rule at issue is genuinely meant to pursue safety objectives, courts should indeed adopt a hands-off approach. Michaud illustrates the difficulties they have in dealing with conflicting expert reports based on complex and uncertain science. And the Court is right to suggest that governments should be entitled to err on the side of caution if they so wish ― though, by the same token, I think they should also not be required to do so (and the Court does not say otherwise). This is fundamentally a policy choice, and the courts should not be interfering with it.

* * *

The questions that the Michaud case raises are many and complex. The Court of Appeal’s opinion is thoughtful and interesting, though as I explained above and in my previous post, I’m not sure that its approach to the existing constitutional framework and to the evidence is the correct one. But that opinion does not address of all these questions. Eventually ― though not necessarily in this case, even if there is an appeal ― the Supreme Court will need to step in and start answering them.

Safety, First

Yesterday, the Ontario Court of Appeal issued an interesting decision in R. v. Michaud, 2015 ONCA 585, a test case challenging the constitutionality of regulations requiring trucks to be equipped with a speed limiter that prevents them going faster than 105 km/h. The Court found that the regulations infringed the truckers’ right to the security of the person, and were overbroad and thus contrary to section of the Canadian Charter of Rights and Freedoms. However, the Court held that this violation of section 7 is saved by section 1 of the Charter.

This is a most unusual result, and the Court itself is well aware that it is an anomaly. Indeed, it Justice Lauwers, the author of the Court’s unanimous opinion, offers some observations for why the s. 7 framework he felt bound to apply might not have been suited to the case. While I am not sure that the Court’s conclusion under s. 7 is correct, its reasons deserve careful consideration, because they engage thoughtfully with a number of issues that are likely to be important on the years and perhaps decades to come.

* * *

The main argument for the unconstitutionality of the speed limiter requirement was that in some situations it may be necessary for a truck driver to accelerate in order to get out of a dangerous situation, and that insofar as the limiter prevents him from doing so, it compromises his safety, and thus his constitutionally protected “security of the person.” In addition, it was said that forcing trucks to move more slowly than other vehicles on the road was a source of inherent danger. The government, for its part, argued that speed limiters serve “to reduce greenhouse gas emissions, to reduce the severity of collisions, and to prevent accidents.” [7] Both the defence and the prosecution also submitted expert reports detailing the speed limiters’ alleged dangers and benefits.

In applying section 7 of the Charter, the Court of Appeal stressed that, under the framework set out by the Supreme Court in  Canada (Attorney General) v. Bedford, 2013 SCC 72, [2013] 3 S.C.R. 1101, the concern at this stage of the analysis is with “the relationship between the individual claimant and the law,” not “the relationship between the private impact and the public benefit of the law.” [62] If the law has an impermissible effect on the claimant, then it infringes s. 7, and its “public benefit” can only be considered at the s. 1 stage of the analysis. The Court considered itself bound, “[o]n a strict and literal reading of Bedford,” to conclude that the regulations do indeed endanger truck drivers in an overbroad way. They prevent a truck driver “from accelerating beyond 105 km/h in all situations where it is needed to avoid collisions,” [73] compromising his security. “For those in such a situation,” the Court holds, “the law contradicts its own purpose of improving highway safety; for them the legislation is overly broad and operates in an arbitrary manner.” [74]

The Court then turns to s. 1. It begins by observing that, in Carter v. Canada (Attorney General), 2015 SCC 5, [2015] 1 S.C.R. 331, the Supreme Court recognized that it might be possible to justify violations of s. 7 by reference to the “public good,” which is not considered in an analysis under that provision. “This,” the Court says, “is one such situation. More are predictable” [83] due to the exclusion of the beneficial effects of the impugned law from consideration under s. 7.

The Court then proceeds to discuss safety regulations and risk management at some length. It notes that “[s]afety regulation often sets bright line rules, rather than standards,” [88] which makes for greater legal certainty. The trouble is that such rules will both allow behaviour that contradicts their purpose to happen, and penalize behaviour that isn’t actually inconsistent with their aims. The substance of each rule is also subject to a cost-benefit analysis. The more stringent a safety rule, the more accidents and deaths it will prevent; but the more onerous compliance with it will be. Designing an optimal rule is necessarily a trade-off between safety and efficiency, complicated by the uncertainty of the relevant science and lack of experience. And however the balance is ultimately struck, it will always be the case that “the regulator countenances the possibility that someone participating in the regulated activity will be put at risk of injury or even death” [98] by not making the regulation stricter than it is. As a result, the Court cautions, “much safety regulation, if it falls to be assessed under the singular approach required by Bedford, would be seen to be inconsistent with security of the person under s. 7 of the Charter.” [99] An additional complication results from the choice between ex-post regulation of conduct by means of imposing penalties for infringing a rule and ex-ante regulation, as in this case, that makes infringement impossible. Ex-ante regulations are an application of the “precautionary principle,” which the Supreme Court has endorsed in the context of environmental law, but they are more likely to be over-inclusive and thus overbroad in a way contrary to s. 7. The takeaway from this discussion is an insistence that, in view of the complexities of regulatory design and the expertise that goes into it, and also considering that what is at issue is a regulatory rather than a criminal offence, courts ought, as in administrative law, to defer to legislative decisions.

Applying at last the s. 1 framework to the speed limiter requirement, the Court concludes that its safety objectives are pressing and substantial, considering “[t]he daily carnage on our roads.” [115] The requirement is connected to these objectives, because of “the link between speed and the severity of collisions.” [119] It is also minimally impairing ― in sense of being well within the realm of reasonable regulatory choices to which courts ought to defer. Even though the ban on speeding by trucks is “absolute,” the legislature was entitled to conclude that its safety objectives demanded no less. The choice of the figure at which the limiter must be set is also something on which courts ought to defer to the regulators. Finally, the positive effects of the limiter outweigh the negative ones, which only arise in a very small fraction of cases.

Before concluding, the Court offers its “reflections” on what it sees as the defects in the analytical framework it saw itself bound to apply. In its opinion, neither the trade-off between maximum safety and efficiency nor the choice of (occasionally over-inclusive) rules over standards “truly engage either deprivation of security of the person or the constitutional principles of fundamental justice; the idea that they do risks trivializing these concepts.” [149; footnote omitted] The Court suggests that the Supreme Court’s definition of overbreadth might itself be overbroad, and that treating any negative impact on a person’s security interest as a “deprivation” within the meaning of s. 7 allows violations to be made out too easily. Its proposed solution “is to recognize them as a distinct category of legislation,” [151] for which societal effects would be taken into account at the s. 7 stage.

* * *

As I have suggested above, I’m not entirely sure that the Court’s s. 7 analysis is right. Mostly I wonder whether the Court is right to conclude that security of the person is actually engaged. The way it describes the evidence, there doesn’t seem to be much if any proof that any situations where we know that accelerating past 105 km/h actually occurred. The Court is content to infer that from the finding by the court below on the basis of a study, “acceleration was used in fewer than two per cent of traffic conflicts to avoid potential crashes” [23] ― which the court then recasts a conclusion that “acceleration to avoid collisions is needed in about 2% of traffic conflicts.” [73] But, for one thing, that an “evasive manoeuvre” was used doesn’t show, it seems to me, that it was needed and there were no alternatives. For another, as the Court itself points out, we don’t know the actual speeds at which these incidents occurred.

In short, unlike in cases like Bedford, where social scientific evidence was used in addition to stories of actual people affected by the impugned legislation, here, the case seems to be based purely on statistical guesswork. To be sure, there was, apparently, some “anecdotal” evidence from the accused himself, but the Court does not even mention it in its analysis. Whose security was affected? To what extent? Was there a more than de minimis negative impact, much less a deprivation? I’m not sure the Court has answers to these questions.

This case is most interesting, however, for the broader issues it raises. Is the Court right that the approach to section 7 set out in Bedford is not suited to analyzing the constitutionality of safety regulations? Is its proposed solution to this problem the right one? What role should the courts play in such cases? I will address these questions, and others, in a separate post later this week. In any event, unless the Supreme Court takes up this case, if Ontario regulators want to put safety first, they’ll have to rely on the Charter’s section 1.