What historical event gave rise to the greater cooperation between levels of government?

Introduction

The comparative federalism literature commonly sorts federations into two broad camps, distinguished by the dominant distribution of legislative powers across levels of government (Palermo & Kössler, 2017, p. 146). The dualist camp are those federations where the constitution predominantly assigns exclusive legislative competences to the levels of government (e.g., Canada, United States, Belgium, Ethiopia).Footnote 1 When defining dual federalism, the literature relies heavily on the imagery of governments legislating and administering policies autonomously within their own “water-tight compartments.” Courts in dualist systems protect government autonomy chiefly by interpreting where the boundaries of these compartments lie. The second camp are those federations where governments largely share legislative competences. In other words, legislative concurrency is the dominant constitutional rule (e.g., Germany, South Africa, Austria). Courts in concurrency-based systems protect government autonomy through constitutional principles like subsidiarity (Schütze, 2009, p. 347). The comparative federalism literature has a name for this second camp: cooperative federalism. Embedded within this nomenclature is the claim that intergovernmental behaviour is endogenous to legal rules about legislative competences: a concurrency rule systematically induces intergovernmental cooperation, where an exclusivity rule systematically impedes it. This presents a normative challenge to those seeking to justify dualism, because it is generally assumed that intergovernmental cooperation is by nature a good outcome.

So if one accepts that dualism leads to the underprovision of a cooperative good, but constitutional reform to purge dualism’s misguided exclusivity principle is fantastical, what is to be done? Courts with dualist architectures have used judicial doctrine to soften the commitment to legislative exclusivity. One important way courts can do this is by interpreting the compartments’ boundaries so that they overlap. By allowing overlap between exclusive legislative competences, courts can use legal doctrine to fashion zones of de facto legislative concurrency (Dziedzic & Saunders, 2017). By creating a legal rule of concurrency in this way, courts argue that they can refashion dualism from within, to achieve an important measure of “cooperative federalism. ”

In this paper, we employ game theory to explore the logical validity of that central claim. While courts assert that “this rule is good because this government behaviour follows”, that mere assertion does not specify why the behaviour follows from the rule. Courts need not specify underlying assumptions about governments’ strategic environments, but scholars do. What choices do governments have, other than to cooperate? What are their policy preferences? How clear is the legal framework? Our formal model explores the conditions under which various legislative outcomes (status quo, unilateralism, or cooperation) are sustained, given uncertainty over the applicable legal rule (federal exclusivity, provincial exclusivity, or de facto concurrency). We incorporate costs associated with policy development, with policy confrontation, and litigation. We then explore the conditions under which the introduction of a de facto concurrency rule is consistent with the emergence of intergovernmental cooperation in equilibrium.

Our analysis complicates the expectation that importing a concurrency rule into a dualist architecture is reliably consistent with increased intergovernmental cooperation. In our model, a key assumption is that the concurrency rule is paired with a federal paramountcy rule, and this combination is key to our findings.Footnote 2 Under our assumptions, cooperation may be supported in equilibrium, but only under quite restrictive conditions. We also show that the rule may have no effect at all; or it may at times undermine cooperation achieved under an exclusivity regime. These findings stand in stark contrast to the prevailing judicial orthodoxy. When the concurrency rule induces cooperation, it does so chiefly by incentivizing federal unilateralism. Cooperation may arise when an unhappy province seeks to fend off the threat of federal policy, and a federal government would prefer to share policy costs rather than go it alone.

We combine our game theoretic approach with a real-world case study. A key feature of our model, variation in the probabilities that a court in a dualist system will apply exclusivity versus concurrency rules, drives case selection. The dominant dualist system in the federalism literature, the United States, shows little contemporary variation on those probabilities. The United States Supreme Court made a dramatic break from a jurisprudence of exclusive legislative competences in the New Deal era. (Corwin, 1950) Whatever “worrisome” flirtations the US Supreme Court has made since, they do not amount to a revival (Young, 2014). In contrast, in the 1980s, the Supreme Court of Canada (SCC) began a gradual shift away from “water-tight compartment” federalism towards a de facto concurrency rule, what it calls the double aspect rule. However, that shift has been uneven, and the SCC has not abandoned exclusivity rules to the degree and consistency we see in the United States.Footnote 3 This may be because the exclusivity principle is explicitly entrenched in the Canadian constitutional text. The Constitution Act, 1867 enumerates a list of exclusive competences for Parliament (s.91) and a list for the provincial legislatures (s.92). Formal concurrency is limited to immigration and agriculture (s.95), and old age pensions and supplementary benefits (s.94A).

We begin the paper with a reading of the Canadian jurisprudence, identifying the reasoning the SCC offers for its decisions on division of powers cases. Motivated by this jurisprudence, we present our formal model, and analyse its implications. Specifically, we examine equilibrium outcomes in an Old world where there are only two legal rules: either provincial exclusivity, or federal exclusivity. We then analyse what our model predicts when we introduce a third rule: concurrency, or as the Canadians say, the double aspect rule. We close the paper with a particular policy case. Canadian securities regulation has become deeply associated with the SCC’s tug-of-war with dualism, intergovernmental cooperation, and the double aspect rule (Poirier, 2020). For our purposes, a focus on Canadian securities regulation is useful because there has been important variation in the jurisdictional rules over time. There has also been variation in the policy responses of governments, including a cooperative legislative proposal. It allows us to show how government policy choices are conditioned by both shifting legal rules and a complex intergovernmental reality, where provinces vary in both political importance and preferences for regulation. In this case, the shifting likelihood and scope of a double aspect rule do not map easily onto a cooperative outcome.

The Supreme Court of Canada and its problem with dualism

Dualism protects government autonomy by defining and enforcing exclusive legislative boundaries. Within its assigned competence, a legislature is free to legislate when its policy preferences lead it to change the status quo. A legislature may find it politically prudent to take other legislatures’ preferences into account when designing and enacting its own policies, and in this way dualism does not preclude the emergence of intergovernmental cooperation. But, dualism does require governments to coordinate while staying in their own lanes. Importantly, as there is no legally enforceable duty to legislate, a legislature is free to enact nothing at all. In other words, dualism promises that governments have the autonomy to stand still, or to go their separate ways should they not get along. And while dualism allows certain cooperative schemes to emerge, dualism does not privilege such schemes as inherently more attractive than simple unilateral action. What dualism does not abide is a government enacting legislation outside of its prescribed lane. No matter the offending legislation’s wisdom, being ultra vires renders that legislation invalid. The judicial punishment for transgressive policy behaviour is stopping policy action in its tracks.

Formal concurrency differs from dualism in that there is no jurisdictional boundary drawn between the two governments to police. There is one lane, and both levels of government have the authority to legislate within it. Judges still devise doctrines to either enable or limit government action within this one lane, but that is a drawing of limits of a different sort (Bulman-Pozen & Gerken, 2009). What is interesting is that judges can fashion a kind of concurrency while working within a dualist framework. They can do so by reasoning that the distinct and exclusive lanes can overlap. In the zone where their exclusive legislative powers overlap, both governments can legislate in respect of their distinct purposes. In Canadian jurisprudence, the judicial doctrine that creates this zone of joint exclusivity is called the double aspect doctrine. In the following extract, Beetz J sets out how the double aspect doctrine works within the constraints of dualism’s exclusivity principle:

The double aspect theory is neither an exception nor even a qualification to the rule of exclusive legislative jurisdiction. Its effect must not be to create concurrent fields of jurisdiction, such as agriculture, immigration and old age pensions and supplementary benefits, in which Parliament and the legislatures may legislate on the same aspect. On the contrary, the double aspect theory can only be invoked when it gives effect to the rule of exclusive fields of jurisdiction. As its name indicates, it can only be applied in clear cases where the multiplicity of aspects is real and not merely nominal.Footnote 4

The core problem is whether one can objectively determine when the “multiplicity of aspect is real and not merely nominal.” Some would deny Beetz’s assertion, on the grounds that dualism’s jurisdictional boundary drawing enterprise is inherently arbitrary given the logical overlaps that constitutional texts allow.Footnote 5 The accusation that dualist federalism review is arbitrary is an existential challenge to a legal official, as law is antithetical to the exercise of arbitrary power. Given judges in dualist systems are stuck with having to interpret jurisdictional boundaries whatever the grumbling from legal theorists off-stage, one way judges can respond is to allow jurisdictional boundaries to overlap, thereby allowing the law to reflect more of a text’s actual logical ambiguities. However, judges interested in softening dualism’s embrace through the double aspect doctrine do not tend to highlight on the page the inherent arbitrariness of their interpretative task.

Instead, judges write much about policy outcomes and the legal rules that facilitate some outcomes over others. There are two interesting narratives that intertwine. One narrative is about the temporal distribution of policy complexity. Basically, the past was simple and uncomplicated, while modernity is a complicated mess. As a result, past governments did not have to do much, while modern governments do. The more governments have to do, the more likely they will have to cooperate. The other narrative links the appropriateness of federal design to policy complexity. Basically, “water-tight compartment federalism” is too rigid and (by implication) too often strikes down legislative action. It follows that strict dualism is inappropriate in modern times. The double aspect rule can be employed to achieve a cooperative federalism of sorts.

We illustrate the narratives set out above using extracts from Canadian court judgements. Too rigid an adherence to the exclusivity principle, through either a formalistic “pith and substance” analysis, inflexible aversion to minimal transgressions of the division of powers (“overflow”),Footnote 6 or the application of the interjurisdictional immunity doctrine, would dissuade government cooperation. The SCC writes:

The Supreme Court of Canada, as final arbiter of constitutional disputes since 1949, moved toward a more flexible view of federalism that accommodates overlapping jurisdiction and encourages intergovernmental cooperation – an approach that can be described as the “dominant tide" of modern federalism...If there was any doubt that this Court had rejected rigid formalism in favour of accommodating cooperative intergovernmental efforts, it has been dispelled by several decisions of this Court over the past decade.Footnote 7

The SCC values intergovernmental cooperation, but it is also identifies a second policy concern: legislative inaction. The spectre of statutory silence looms in the SCC’s writing across a number of policy areas: the creation of legislative no-go zones in banking,Footnote 8 a legislative “hiatus” in interprovincial trade,Footnote 9 a legislative vacuum with respect to controversial medical procedures,Footnote 10 the “patchwork” and hence ineffective regulation of forests,Footnote 11 and the (relative) dearth of legislation pertaining to the Metis community.Footnote 12

The SCC argues that a move toward a joint exclusivity rule will deliver on two fronts, by inducing intergovernmental cooperation and preventing legislative inaction. Nonetheless, the SCC signals that it cannot simply invoke a joint exclusivity rule in all future cases. Some “water-tightness” can still be expected into the future. The majority in Firearms Registry took pains to reiterate the following:

While flexibility and cooperation are important to federalism, they cannot override or modify the separation of powers...[N]otwithstanding the Court’s promotion of cooperative and flexible federalism, the constitutional boundaries that underlie the division of powers must be respected. The “dominant tide” of flexible federalism, however strong its pull may be, cannot sweep designated powers out to sea...Footnote 13

The SCC recognizes that unilateralism has an important role to play:

Neither this Court’s jurisprudence nor the text of the Constitution Act, 1867 supports using [the] principle [of cooperative federalism] to limit the scope of legislative authority or to impose a positive obligation to facilitate cooperation where the constitutional division of powers authorizes unilateral action.Footnote 14

For all the policy good that greater use of the double aspect rule will purportedly induce, judges do recognize risks to this approach. Given a concurrency rule allows for both levels of government to legislate at the same time, there needs to be a legal rule to regulate that scenario. Dubbed a paramountcy rule in Canada, this rule can be either textually entrenched or established through judicial doctrine. In Canada, the SCC applies a federal paramountcy doctrine.Footnote 15 A risk to applying the double aspect doctrine indiscriminately is that it will lead, via the federal paramountcy rule, to a centralization of power (Bulman-Pozen & Gerken, 2009; Weiser, 2000):

...the risk [is] that...two fields of exclusive powers will be combined into a single more or less concurrent field of powers governed solely by the rule of paramountcy of federal legislation. Nothing could be more directly contrary to the principle of federalism underlying the Canadian ConstitutionFootnote 16

The SCC’s answer to this centralization fear has been to develop a federal paramountcy doctrine that renders valid provincial legislation inoperative if it is incompatible with federal legislation, or if it frustrates a federal purpose (Wood, 2016). The SCC views its paramountcy rule as addressing its underlying policy concerns:

Increasingly, as our society becomes more complex, effective regulation requires cooperation between interlocking federal and provincial schemes. The two levels of government possess differing tools, capacities, and expertise, and the more flexible double aspect and paramountcy doctrines are alive to this reality: under these doctrines, jurisdictional cooperation is encouraged up until the point when actual conflict arises and must be resolved.Footnote 17

To broadly summarize this section of the paper, we have set out how the SCC justifies its increasing use of the double aspect doctrine in policy terms. It sketches out some basic propositions for how legal rules induce legislative behaviour. We note that the SCC does not provide a reasonably specified model of legislative behaviour. A jurist may be satisfied with the SCC’s sketch, but a political scientist should not be. Many variables factor into a government’s decision to legislate or not, including but not limited to its own policy preferences, uncertainty over a proposed policy’s effects once implemented, and the degree to which it has support of other governments. Given this, why should we think that the choice of a legal regime has a direct and simple effect on legislative behaviour? Given a world where governments care about more than just legislative jurisdiction, under what conditions is the SCC justified in thinking that “double aspect federalism” is meaningfully synonymous with “cooperative federalism”?

We are not alone in raising questions about whether intergovernmental cooperation easily follows from the SCC’s premises. As Wade K. Wright notes, clear jurisdictional limits could support intergovernmental bargaining under some conditions, but not in others (Wright, 2016, p. 450). While some scholarship has addressed the question (Schertzer, 2018), and Wright is clear that conditions other than federalism doctrine will be determinative, this scholarship basically stops there. There is little in the way of serious investigation of what those extra-legal conditions are, and very little work beyond the occasional case study to answer the question through an empirical analysis.Footnote 18

Our contribution here is to address the theoretical validity of the SCC’s central claim: that flexible/double aspect federalism is better than inflexible/dualist federalism in producing what the SCC holds out as better legislative outcomes, while preventing the worst outcomes. The next section of the paper sets out a formal model of legislative supply that is endogenous to the selection of a legal rule, while considering both government policy preferences and the costs to both legislating and litigating.

The formal game

Despite considerable interest in cooperative federalism in both political science and law, there is very little game theoretic modelling that is directly relevant to our enterprise here. The most relevant published work is that of Gemma Sala (2014). Sala explicitly models legislative supply and intergovernmental cooperation under conditions of judicial uncertainty. In her model, there are two governments, both of whom are uncertain over whether a court would veto legislation. If one government chooses to enact legislation, the other government chooses whether to initiate judicial review. If litigation is chosen, the first government can offer to negotiate over the policy, an overture which the second government can choose to accept or not. Cooperation (a negotiated outcome) can therefore be induced, given certain policy payoffs and probabilities of a judicial veto. It generates a number of possible equilibria, the most interesting of which is one where governments cooperate to transgress the division of powers when they are certain that a policy would not survive judicial review. This finding problematizes intergovernmental cooperation as one courts should automatically prefer.

We wish to add some complexity to Sala’s model. One complexity is over what the judicial choices are. The SCC signals an important liking for the double aspect rule, while still being prepared to invalidate legislation on exclusivity grounds.Footnote 19 We therefore model the court as having three choices, rather than simply two. These are: whether a legislative competence is (1) solely exclusive to the federal government, (2) solely exclusive to the provincial government, or (3) jointly exclusive. The other layer of complexity we wish to add is that both governments can be legislators, and both can bring a legal challenge to an enacting government’s legislation. These additional complexities make for a more realistic game, but at the price of some tractability. The benefit of the added complications are the model’s ability to generate all of the outcomes that the SCC worries about in its cooperative federalism decisions: policy inaction, where neither government legislates; legislative unilateralism, and whether the enacting government is met with a response by the non-enacting government; and an explicitly cooperative outcome, where both governments agree on a shared or coordinated policy. We ask: what conditions support these possible outcomes in equilibrium?

Fig. 1

The extensive form of the game

Full size image

Where we added complexity by allowing for three legal rules, we simplify by limiting the number of strategic actors. The court is not a strategic player. We are interested in modelling what governments will do given uncertainty over the legal rule; we are not concerned with modelling a court’s choice of a legal rule given what governments might do. We assume the actors in the game share the same knowledge of the court’s probabilities of selecting among its legal rules. There is also no third party litigant. There are two strategic actors: a subnational (provincial) and a central (federal) government. The provincial government is the first mover. This is an important modelling decision, one we consider arguable but reasonable. One could think about it in at least two ways: (1) in actual federations with a number of subnational governments, it is likely that at least one of these subnational governments is acting proactively in a policy space, or (2) that subnational governments are the work-horses of a federation, with subnational statutory regimes forming the core of governance (Atkinson et al., 2013; Cairns, 1977) The provincial government starts the game facing three choices, and the federal government is always able to respond to the province’s choice. In other words, the province cannot shut down the game on the first move. During the course of the game, both the provincial and federal governments can find themselves as legislators or litigants. If a government enacts a policy, and the other government accepts that policy, the operative legal rule will remain probabilistic, because neither government asks the court to reveal which legal rule would apply in that particular case. Any newly enacted policy accepted in equilibrium may well be a transgression of the division of powers.

Table 1 Game legend

Full size table

There are therefore three main branches of the game tree, which is illustrated in Fig. 1. The legend to the game is Table 1 and lists the players, actions and payoffs shown in the game tree. Note that the decision nodes of the tree are indicated numerically, and we will refer to these nodes throughout the paper.

The top branch of the tree is where the province opens by choosing to do nothing, N. At node 2, the federal government has two options. The federal government can also choose to do nothing. This ends the game, and the result is the status quo, SQ. Alternatively, the federal government can enact its own policy, F. If the federal government chooses F, then the province can respond. At node 5, the province can either accept F, or challenge F’s validity in court.

The second branch of the game tree has the province open the game by extending an offer to the federal government to cooperate, denoted “let’s talk”, or LT. At node 3, the federal government decides how to respond to this offer. The federal government can accept the offer. If it does, then the governments enact a cooperative policy, C. This is the only way to reach an explicitly cooperative vertical policy scheme in the game. Because there is no third party litigant, C never faces the possibility of judicial review.Footnote 20 While the province’s move “let’s talk” at node 1 opens the possibility of a cooperative policy outcome, there are other possible outcomes lurking down this branch of the game tree. The federal government can spurn the offer to deliberate, and by doing so, the game ends with the status quo. The province therefore knows that offering the “let’s talk” option at node 1 forecloses its future ability to legislate unilaterally, and that the province’s offer to talk may result in policy stasis. Allowing this branch of the tree to conclude with the status quo accomplishes two things. Firstly, it is simplifying, as the province has no follow-up move to make. Secondly, this increases the seriousness of the province’s deliberation offer at node 1, because if it wasn’t, the province should simply start the game with either its own unilateralism, or inaction. The structure of the game suggests that a province’s offer to deliberate is a serious offer. Should the federal government decide at node 3 to spurn that deliberation offer and respond with legislative unilateralism of its own, enacting F, the province at node 6 can respond to this unilateralism with either acceptance, or a challenge in court.

The bottom branch of the game tree starts when the province chooses to enact its own legislation, P, at node 1. At node 4, the federal government has three choices in the face of P. First, it can accept P, and the game ends. Second, the federal government can challenge P in court. Third, the federal government can choose to legislate, confrontationally enacting F in the face of P. If the federal government enacts F, the province has a choice to make at node 7. If the province accepts F, then F overrides P. This game allows the federal government to be confrontational in a way that the provincial government cannot be. If the federal government imposes its policy and the province accepts this imposition, the federal government’s confrontation is successful. The province can also choose to challenge an imposed F in court, although the challenge will only succeed if the court chooses provincial exclusivity as the legal rule.

Here are the assumptions as to costs. Whenever a government enacts a policy unilaterally, it incurs a policy development cost, d. No policy is enacted for free. For the sake of simplicity, we assume that policy development costs are symmetrical. One way to think about this is that the two governments have the same policy development cost per capita. We make no assumption that a provincial government will deliver a policy more efficiently than a central government, or in the reverse that there are economies of scale if it is developed centrally. If the game’s outcome is C, then the governments split the policy development cost equally. There is a tradeoff to cooperation: C is at the midpoint of their policy preferences, but they are able to share the cost burden.

There is an imposition cost, i. The federal government incurs i if it enacts F after the province has already legislated. The imposition cost is there to parameterize the idea that a central government will certainly face a backlash when legislating directly in the face of provincial action. By selecting P at node 1, the province has clearly mobilized itself, and likely a political constituency, in favour of P. A federal choice to impose F in the face of this provincial mobilization bears a particular cost. It is true that a federal government might incur a political backlash of some sort no matter what its policy choices are, but for simplification of the model we consider those generalized costs to be built into the relative preferences over F or SQ.Footnote 21

This is what happens in the event of litigation. Any time a government challenges a policy in court, both parties incur a litigation cost, l, with the stipulation that l is less than d. Litigation results in the court choosing one of three legal rules: federal exclusivity, provincial exclusivity, or the double aspect rule. If the legal rule is provincial exclusivity, then a P would be upheld, and an F would be struck down (resulting in SQ). If the rule is federal exclusivity, then an F would be upheld, and a P struck down (resulting in SQ). If the rule is the double aspect rule, then either P or F is valid and will be allowed to stand. However, a federal paramountcy rule upholds a federal government’s imposition of F over P. We therefore model a robust federal paramountcy rule, where F simply pre-empts P.Footnote 22

We note that there is no cost parameter related to either making a deliberation offer, or of accepting it. In this sense, negotiating itself is costless. This would not be a reasonable assumption in the real world, but it has value in a formal modelling context. First, it addresses tractability concerns. Second, it arguably skews the modelling towards a cooperative outcome. Even in this cooperation-optimistic scenario, is it really hard to get to C?

The game gives us a conceptual framework to think through certain intuitions. It lets us think more critically about how to interpret the possible outcomes of the game. A simplistic view that “cooperation is good” and “non-cooperation is bad” breaks down upon fairly quick reflection. If the cooperative outcome C is upheld in equilibrium by a central government’s threat to impose its own policy off the equilibrium path, then C looks less cooperative and more coercive. However, we may also be right in thinking that one government acquiescing to another’s policy, and not litigating it, is a different way of coordinating. This is possible in a situation where that policy is more attractive to both governments relative to the status quo, but where the development costs are not so large for the enacting government to seek cooperation over proceeding unilaterally. This is possible even if the enacting legislation would likely fail judicial review. In this instance, a unilateral policy enactment could be both mutually acceptable and a transgression of the division of powers (Scholtz & Polataiko, 2019).

A new world: enter the double aspect rule

We want to ask how government behaviour changes from an initial environment in which a court can produce two possible legal rules (solely federal exclusive jurisdiction or solely provincial exclusive jurisdiction) to a new environment in which the court allows for the overlap in jurisdictions, via the double aspect rule.Footnote 23 Let us call the initial legal environment “Old” and the new legal environment “New”. We assume that, under Old, the probabilities of the provincial exclusive and federal exclusive rules are \(\pi ^{\text {Old}}_{Px}\) and \(\pi ^{\text {Old}}_{Fx}\), respectively. Under New, the double aspect rule is introduced with some probability, \(\pi _{D}\). \(\pi _{D}\) is carved out of either \(\pi ^{\text {Old}}_{Px}\), \(\pi ^{\text {Old}}_{Fx}\), or both. If the double aspect rule is introduced at the expense of the provincial exclusivity rule, then \(\pi ^{\text {Old}}_{Px}=\pi ^{\text {New}}_{Px}+\pi _D\) and \(\pi ^{\text {New}}_{Fx}=\pi ^{\text {Old}}_{Fx}\). If the double aspect rule is introduced at the expense of the federal exclusivity rule, then \(\pi ^{\text {Old}}_{Fx}=\pi ^{\text {New}}_{Fx}+\pi _D\) and \(\pi ^{\text {New}}_{Px}=\pi ^{\text {Old}}_{Px}\).

The introduction of a double aspect rule does not change the underlying cost structure. In other words, d, l, and i remain constant from the Old World into the New. But \(\pi _{D}\) does effect the expected utility of going to court, for the province (at nodes 5, 6, and 7) and the federal government (at node 4). This in turn may be significant enough to change the federal government’s first move (at nodes 2, 3, and 4), which may reverberate so far as to change the province’s opening move (at node 1), and the federal government’s response. We note that any terminal node can represent an equilibrium under certain parameter restrictions. Whether and how \(\pi _{D}\) might shift equilibria from Old to New depends on: its magnitude, whether it erodes \(\pi ^{\text {Old}}_{Px}\) in particular, relative to the underlying costs, and the actor’s policy preferences relative to the status quo.

Sustaining the cooperative equilibrium

Given the judiciary’s core interest in intergovernmental cooperation, we will focus our analysis on the game’s cooperative outcome, C. Although it is generally complicated to fully characterize this equilibrium, intuitively we can summarize the conditions that lead to C in equilibrium in the following way. The conditions for a cooperative equilibrium are, first, relatively high development costs, such that no player wants to pay it in full, but not so high as to make a policy alternative unfeasible. Second, both players prefer the cooperative outcome to the status quo.Footnote 24 Third, federal and provincial interests are not too misaligned, meaning the distance between the payoff from one’s own legislation and the other player’s legislation is not too large. Otherwise, the gains from deciding to legislate yourself are high enough that the cooperative equilibrium collapses. However, government interests are misaligned enough (at least relative to the development costs), that no player is willing to accept and free ride on the other’s legislation simply to forego those costs. Fourth, there is sufficient uncertainty about the legal outcome such that neither of the players prefer going to court to the cooperative outcome, and no player can use going to court as a “slam-dunk” threat in order to force their counterpart to accept its legislation. At the outset, let us note that these conditions describe a highly contingent equilibrium. Let us explore which scenarios under Old are (or are not) amenable to shifting to this tricky cooperative outcome under New.

The first scenario to explore is one where you would think the introduction of a double aspect rule in New would be the most helpful in getting both governments to C. This is when, in Old, the province chooses LT, but the federal government spurns the offer (ie. at node 3 choosing N or F). In a colloquial sense, the Old equilibrium is “half way there”. But according to our model, this intuition is misleading. Under Old, we know that the federal government in equilibrium preferred either SQ or F to LT (formally, max(SQ,F) > LT). The mere introduction of the double aspect rule cannot change the payoff for SQ. In order for the federal government to prefer LT under New, it must be that the federal government’s payoff for F decreases. That would only happen if the province was more likely to prevail in court (and choose to litigate at node 6). However, it would not make sense for the province to choose litigation at node 6 under New. This is because, given the paramountcy assumption, the province’s payoff (in expectation) from going to court under New is either the same, or worse. Therefore, from the standpoint of the federal government at node 3, it only makes sense to either stick with SQ, or to double down on F. Formally, it cannot be the case under New that LT > max(SQ,F). To draw out the point, getting from the Old “half way there” scenario to the cooperative outcome in equilibrium is about shifting political preferences and costs, rather than the introduction of the double aspect rule alone.

What about an Old equilibrium where the province legislates at node 1? We know that whatever lurks down the tree at nodes 4 and 7 is more attractive to the province than doing nothing. To choose P at node 1, the province must be very confident about either the strength of its jurisdictional claim in court, its ability to impose a political cost on the federal government, or so dislike both F and the status quo that it is willing to risk a confrontation with the federal government. So, when is it plausible that a New double aspect rule would push the province to LT at node 1 instead? If the province’s decision is backed up by its political strength outside the courtroom, then the double aspect rule should induce no change in the province’s behaviour. However, if the double aspect rule takes a bite out of provincial jurisdiction, hence eroding the province’s case in court, a politically weak province may be induced to pick LT at node 1, in order to get a policy change it can live with. However, in order for the federal government to accept that deliberation offer at node 3 despite the federal government’s appreciably improved chances in court, the federal government must not think that going to court is now so easy as to be a “slam-dunk”. It must also be happier to deal with the province than either carry all of the policy costs itself, or to have no policy change at all. To draw out this point, the cooperation that a double aspect rule might induce is one between a strengthened federal government and a legally weakened already politically weak province. By contrast, the politically strong province does not need the court’s help.

What about an Old scenario where the federal government legislates at node 2, which the province either accepts, or litigates? This Old scenario is sustained by either (1) the strength of federal jurisdiction, so that a province would likely have to live with F in any case, (2) a strong federal jurisdiction but a province unsatisfied enough with F to take its slim shot in court, (3) a province happy enough to free ride on F, so it has no reason to get in the federal government’s way (at node 5). The introduction of double aspect rule should have no impact at all on the happy free-riding province just described. Given that a double aspect rule either maintains or improves the federal government’s chances in court, a federal government willing to legislate on its own in Old would presumably still be so inclined under New. However, a province unhappy with both F and SQ might be willing to shift its choice at node 1 and extend a deliberation offer to the policy-hungry federal government. With high development costs, the federal government might accept that offer. This possible cooperation outcome is predicated on the federal government having both a strong legal hand and a strong interest in expensive policy change, with the court’s intervention bringing an unhappy (but not necessarily politically weak) province to the policy bargaining table, to at least shape the expensive new policy. However, a strong unhappy province willing to litigate F (at node 5) under Old may also, under New, be so unhappy with both F and SQ that it will preemptively legislate a cheaper policy at node 1. This makes sense because if the federal government chooses to litigate at node 4, a double aspect opinion by the court will uphold P. This points to the possibility under New of more unilateralism and more litigation between polarized and strong governments.

The next Old scenario we will engage with is where the status quo is the equilibrium outcome, when both governments choose N.Footnote 25 How would a double aspect rule induce both those governments to work together on C? Intuitively, getting from policy stasis to cooperation through only a double aspect rule seems a leap too far. And, this intuition is mostly borne out. This Old SQ equilibrium could be sustained in a number of ways. Governments could have a shared preference over the status quo, such that neither government is terribly interested in a policy alternative, even at some reasonable cost. If so, a double aspect rule in New would have no effect at all. It could also be sustained by some shared level of dissatisfaction with the status quo, but where a policy alternative is too costly for governments to implement, even if those costs are shared. In this case, having a stronger hand in court under New also should have no effect on government behaviour. Lastly, this equilibrium can be sustained given provincial satisfaction with the status quo and a strong likelihood that a court would uphold exclusive provincial jurisdiction, thereby dissuading a federal government from legislating and facing the province in court. In this case, if \(\pi _D\) is introduced at the expense of exclusive provincial jurisdiction, the equilibrium could shift. Knowing the province would be less likely to prevail in court, the federal government faces a higher payoff from legislating unilaterally, so either F or C could emerge, depending (again) on the development costs, and how much the federal government hates the status quo.

Next, consider a situation where C is the equilibrium outcome in Old. In other words, a very tricky set of legal and political conditions has already been met—in the absence of a double aspect rule.Footnote 26 Given the analysis above, we know that the legal framework supporting that equilibrium is one that keeps the province out of court while keeping the federal government from thinking it has a free policy rein. The double aspect rule in New may disrupt that balance. Adding a rule that erodes provincial jurisdiction would most likely destabilize the equilibrium in the direction of federal unilateralism. However a rule at the expense of federal jurisdiction will only favour a province if the federal government doesn’t legislate. The dynamic should now be familiar. In the case of high imposition costs, a province dissatisfied with the status quo might be induced to legislate pre-emptively. Given the high imposition cost, the federal government might accept P at node 4, but it might also choose to litigate.Footnote 27

To summarize, the Canadian judiciary justifies its doctrinal move towards a double aspect rule in policy terms. It argues that this doctrinal innovation will yield intergovernmental cooperation, where before there was none. Our formal model allows us to analyze this claim, under what we consider some reasonable assumptions. The exercise allows us to counter the judiciary’s simplistic claim in this way. Even when assuming deliberation is costless, intergovernmental cooperation is a highly conditional outcome, and if a new double aspect rule has any effect, it is not straightforward. The effect (if any) depends on political preferences and costs. Given federal paramountcy, the introduction of the double aspect rule is favorable to the federal government in all cases in which it decides to legislate. This has the chief effect of encouraging federal unilateralism. Only when the federal government is unwilling to legislate, and the province is willing to legislate, does the new regime benefit a province. Given high imposition costs but reasonable policy development costs, a province may legislate preemptively. Given high policy development costs and/or low imposition costs, cooperation that might arise from this rule change, but it would most likely involve a weak province unhappy to be there.

The case of Canadian securities regulation

This section applies the modelling exercise to the unique case of Canadian securities regulation. Canada is the only country with a developed securities market that does not have a national securities regulator. Instead, there are 10 provincial and 3 territorial securities regulators. This is a policy field where reform proposals have run the gamut from maintaining a decentralized and competitive regulatory structure, to a national securities regulator, to an explicitly cooperative intergovernmental scheme. This is a case with variation on model parameters: shifting jurisdictional rules (and expectations of those rules) over the last two decades; provinces with varied preferences for securities regulation; varied political strength to defend provincial preferences; and a catalyst (the 2008 financial crisis) that shifted underlying federal preferences for a national regulator. We also see variation in federal government choices. The federal government has across various points in time resisted any role in securities regulation, proposed legislation for a national regulator, proposed yet never implemented a cooperative scheme, and chosen to litigate jurisdictional questions. We argue that the case confirms some central intuitions derived from the model, while also pointing to some interesting empirical dynamics that the model does not address.

The judiciary’s expansive reading of the property and civil rights power is the constitutional basis for provincial regulation of securities in Canada. The constitutional power of the provinces to do so has not been in doubt. However, as economies have become more complex and integrated, expert interest in a national level response to securities regulation began to grow in the 1960s. Arguments for a national regulator rely on the premise that a uniform set of national laws, with uniform enforcement, will achieve market efficiency and investor protection, over and above that provided for by a decentralized provincial framework.Footnote 28 Those in favour of federal regulation argue that the power over trade and commerce could provide the constitutional basis for a national regulator. Though not then directly tested, in the 1980s that legal argument became more likely for two reasons. First, the SCC had used the double aspect doctrine to uphold federal power in a securities-related domain (insider trading).Footnote 29 Second, in a market competition case, the SCC expanded the scope of the federal government’s trade and commerce power.Footnote 30 In short, through the 1990s and into the 2000s, there was a growing perception that courts were increasingly likely to use the double aspect doctrine to the detriment of sole exclusive provincial jurisdiction, rendering a legal threat of a national regulator increasingly credible (Fraiberg, 2009).

In the case of securities regulation in Canada, the federal government sat on the legislative sidelines even as its chances in court became viewed as strong into the 2000s (Expert Panel on Securities Regulation, 2009, p. 62). Federal governments have historically declined to impose a federal regulator in the face of the pre-existing provincial framework, despite increasing expert and industry calls for reform. Observers identify the source of federal reluctance to be largely political, specifically intergovernmental, rather than electoral (Anand and Green, 2010, p. 686). A national securities regulator has traditionally faced varied support from among the provinces with the largest securities markets. Ontario has been a significant proponent, Alberta and Québec significant opponents, and British Columbia flip-flopping to some degree. In Canada’s intergovernmental reality, the entrenched opposition of Quebec in particular is politically meaningful.

We’ll pause in the storytelling here in the late 1990s, to reframe it in terms of the model. What we have described so far is an Old equilibrium (pre-1980s) where the provinces set up a regulatory framework (P at node 1), which the federal government accepts (at node 4). This is consistent with two conditions, both of which have empirical support: high imposition costs, and a legal regime that strongly supports exclusive provincial jurisdiction over securities regulation. In New, with the SCC increasingly favorable towards a double access rule, \(\pi _{D}\) is increasing at the expense of provincial jurisdiction. Yet, we see no corresponding shift in federal behaviour. This dynamic of federal inaction we describe here is consistent with our modelling, which predicts that in a world where the federal government accepts provincial legislation chiefly because imposition costs are high (e.g., at node 4), the introduction of the double aspect rule to the detriment of provincial jurisdiction should on its own have little effect on federal behaviour.

Let’s rejoin the story. It is interesting to contrast federal inaction with provincial behaviour in the early 2000s. While legal arguments for federal jurisdiction were strong, and expert opinion in favour of a national regulator was gaining “urgency” (Johnston et al., 2014, p. 646), the provinces, except Ontario, engaged in a process of horizontal cooperation. In 2004, the participating provinces, led by Quebec, signed a memorandum of understanding to establish the so-called “passport system”. In this system, market participants meeting the prospectus clearance, discretionary exemption, continuous disclosure, and registration requirements of their principal regulator, would have that regulator’s approval be respected by the other participating provinces. The Ontario Securities Commission set out its own and the provincial government’s position this way:

The Ontario Government has indicated that it is not prepared to participate in the passport system without a roadmap, with reasonable timelines, to get to a common securities regulator. The passport proposal is based on rule-making powers that the passport members have or expect to receive through statutory amendments. These statutory amendments have not been enacted in Ontario and we understand that there are no plans to introduce them...The OSC supports a common securities regulator that would interpret, apply and enforce securities laws consistently for all market participants in Canada (Ontario Securities Commission, 2007).

However, because the 12 other regulators have agreed to respect the decisions of the Ontario Securities Commission, Ontario passively participates in the passport system by default. The passport system has increased coordination and decreased regulatory duplication for market participants, but it is not designed to achieve a uniformity of other securities regulations or enforcement (Anand & Klein, 2005).

This example of horizontal cooperation is interesting to us. Our model sets out a way of thinking about the conditions sustaining vertical intergovernmental cooperation. It is not a model of interprovincial cooperation in a federal system. But what we can note is that in this real world example, interprovincial coordination (ie. the passport system) was put into place during the period when courts appeared increasingly receptive to the federal jurisdiction over securities via the trade and commerce clause, and when industry actors were making loud calls for a national regulator (Johnston et al., 2014, p. 646). This suggests that the growing likelihood of the double aspect rule at provincial expense had some role in inducing Quebec and other provinces to proceed with the passport system, as a way of fending off the growing attractiveness of a federal regulator. Whether provinces actually sought horizontal cooperation to preempt a perceived threat of federal action remains a matter for further analysis. However, if confirmed, it would bolster the SCC’s general claim that double aspect federalism can help induce intergovernmental cooperation. However, the example also demonstrates that it would have done so through the mechanism of eroding the likelihood of sole exclusive provincial jurisdiction.

The reluctance of the federal government to move towards imposing a national regulator changed in 2008. When imposition costs remain high and federal government behaviour does change, our model encourages the analyst to explore whether there are significant changes in either underlying preferences, or policy development costs. Policy experts and the business press point to the financial crisis as the exogenous event that shifted the federal government’s preference for F further away from P (Ibbitson, 2010). With federal interest in F now increasing, in July 2009 the federal government established the Canadian Securities Transition Office (CSTO), with the mandate “to assist in the establishment of a Canadian securities regulatory authority”.Footnote 31 In 2010, the CSTO proposed a Securities Act to implement a federal securities regulator. The proposed framework set out a single comprehensive national regulator framework, where some provisions would apply throughout the whole of Canada, but where the majority of the provisions would apply once provinces opted in. This part-voluntary design might have taken the edge off accusations of federal unilateralism, but few viewed it as a serious inoculation of entrenched provincial opposition (Anand & Green, 2011). In 2010, Quebec, Alberta, and the federal government each sought judicial opinions on the constitutionality of the federal government’s proposed legislation. In the final analysis, the federal government’s response was not to actually impose its legislation on top of the existing provincial framework. It chose to litigate the constitutionality question instead. In terms of our model, in 2011 we see a shift from the Old equilibrium, where the federal government accepts P at node 4, to a New equilibrium, where the federal government litigates at node 4.

Contrary to expert expectations, in 2011 the SCC (for the most part) rejected the federal government’s case. The SCC concluded that federal legislation to regulate day-to-day securities issues was ultra vires its trade and commerce power.Footnote 32 This was a strong win for provincial jurisdiction. However, the SCC allowed that Parliament would have the power to regulate securities markets to the degree that such regulations applied to matters of specific national concern, such as mitigating systemic risk or allowing for national data collection. The SCC also stated that the federal government failed to present alternative constitutional arguments that may have been more successful in authorizing some scope for a national regulator.Footnote 33

The 2011 decision effectively took away any plausible federal threat to impose a day-to-day securities regulator, while still handing the federal government a restrained role in securities regulation (Rosseau, 2012). After this ruling, we see the federal government and, by 2015, a bare majority of provinces, sit down to talk about possible vertical cooperative scheme. It is possible to argue that this post-2011 vertical bargaining supports the SCC’s general claim that their use of the double aspect doctrine brought governments to the bargaining table. The challenge to that claim, however, is that the 2011 decision on the whole was a strong dualist defence of exclusive provincial jurisdiction over securities, with a side serving of double aspect (Poirier & Gaudreault-Desbiens, 2017; Poirier, 2020). But, even if one accepts the claim that the double aspect portion of the ruling brought provinces to the bargaining table, we note which provinces showed up: 1) Ontario, a strong province with a clear preference for F over P, and 2) all other provinces not strong enough to impose serious costs on the federal government. Alberta and Quebec remained immune to the "let’s talk" option.

In 2016, the federal government established the Capital Markets Authority Implementation Organization (CMAIO) to support the ongoing vertical cooperative process. What came out of that process was a proposed regulatory scheme. Unsurprisingly, Quebec challenged the proposed (not enacted) cooperative scheme in court. In 2018, the SCC ruled that the federal government did not exceed its legislative authority under the proposed scheme.Footnote 34 Despite this legal loss, Quebec signalled that it would nonetheless continue to reject what the SCC had clearly ruled as allowable federal action, framing it as encroachment instead: “Quebec Finance Minister Éric Girard insisted...that his government has no intention of facilitating what it sees as an intrusion by Ottawa into its jurisdiction over securities regulation” (Yakabuski, 2018). Quebec signalled that Ottawa would pay a political price for acting within the restrained role the Supreme Court has allowed. A law firm with a speciality in securities law identified the key barrier to implementing the cooperative scheme: “...it remains to be seen if sufficient political will exists to make a cooperative regime successful in Canada” (Davies Ward Phillips Vineberg LLC, 2019, p. 39).

Recent events have confirmed that the will is not there in the face of Quebec’s entrenched position. In March 2021, the CMAIO announced it was suspending its operations “...having completed all possible integration pending the completion of the [cooperative system] legislation and in consultation with the participating governments” (CMAIO, website). It seems that no one in the securities industry was surprised (Mezzetta, 2021). In June 2021, dedicated funding for the cooperative policy process did not survive budgetary politics in Canada’s minority Parliament (Carmichael, 2021). In the final analysis, the Old equilibirum, where the federal government accepts P in the face of high imposition costs, remains the equilibrium under New.

The Canadian securities case confirms a core intuition of our modelling exercise: the policy impact of a double aspect rule is highly conditional on the political strength of provinces to protect their jurisdiction outside of the court room, and their relative satisfaction with the status quo. In Canada’s complex intergovernmental environment, Ontario is a strong province whose preferences are for a strong national regulator. Ottawa could always count on Ontario’s support, regardless of how probable the double aspect rule would be applied. At the other end, Quebec has had a consistent preference to deny federal involvement, a position it has been willing to defend vigorously (Harris, 2003, p. 61). Ottawa would always face a political price to act within its jurisdiction, whatever that jurisdiction’s scope. This story reinforces yet again Quebec’s pivotal ability to influence federal policy choices, even in areas of clear federal jurisdiction.

We have provided here no quantitative calculations of imposition costs, probabilities of a legal rule, or relative values of policy options versus a shifting status quo. But we have told an empirical story of these parameters getting smaller or larger, and how our model confirms some government behaviours as consistent with the model’s predictions. The story at least confirms to us that we should resist simple expectations of how a legal rule will effect political behaviour.

Conclusion

Comparative federalism scholar Robert Schütze writes: "...is the philosophy of cooperative federalism the "better" federal arrangement?"? He answers: "cooperative federalism should be championed as an end in itself" (Schütze, 2009, p. 351). The normative case rests on the claim that legislative exclusivity works against the kind of cooperative intergovernmental practice that modern federations need. Dualist systems, built on foundations inherently unsuited to the challenges of modern governance, need all the judicial help they can get. Invest in a jurisprudence of overlapping competences, and cooperative dividends will follow. This is an intuitive claim, powerful in its simplicity.

We present an argument for why this claim should be met with more skepticism. According to the assumptions we have formalized here, all should be wary of the judicial pronouncement that intergovernmental cooperation unproblematically follows from increasing application of a de facto concurrency rule. If a double aspect regime has any effects at all, they are highly contingent on government policy preferences vis-a-vis the status quo, and on costs over which a court itself has little to no control. These costs are related either to the cost of policy alternatives, or to the political costs associated with intergovernmental confrontation. We paired our theoretical approach with a case study of Canadian securities regulation, as a way to link the mechanism of a changing concurrency rule with observed political behaviour, cooperative or not (Hédoin, 2020).

This paper does not mount a comprehensive defence of dualism over "cooperative" federalism, as an end in itself. This is not a full weighing of all relative costs and benefits between these two types of federal design. Our ambitions are admittedly more limited. We encourage federalism scholars to examine cooperative federalism’s core claim more closely; and perhaps, like us, to put "cooperative" in quotation marks.

What historical event gave rise to a greater cooperation between levels of government?

Which event is most responsible for changing to US from dual to cooperative federalism?

What event led to an increase in federal power and why has it continued to grow?

How did the federal government become much stronger over time?