Introduction: the evolution of the idea of a policy portfolio
The concept of policy mixes originates with Mundell’s (Reference Mundell1962) discussion about the intersection of fiscal and monetary policy but only began to migrate to the policy sciences about a quarter century later around the discussion of how policy instruments and policy designs operated in the then-emerging environmental (Gunningham, Grabosky, and Sinclair Reference Gunningham, Grabosky and Sinclair1998) and innovation policy spheres (Smith Reference Smith1994). Observers noted that in these realms, and others, policies tended to appear as bundles or “mixes” of tools rather than as single, discrete entities (Howlett Reference Howlett, Berg-Schlosser, Badie and Morlino1991). Later work extended this finding to older sectors such as health and to others like welfare and education policy, arguing that these policy areas, too, existed as portfolios of measures and featured significant interaction effects, both positive and negative, as the relationship between portfolio components changed over time and contributed to increased policy complexity, success, or failure (Lecuyer and Bibas Reference Lecuyer and Bibas2012; del Río González Reference del Río González2007).
This “portfolio” approach to studying policy components is now commonplace (Howlett and del Río González Reference Howlett and del Río González2015) and a “new generation” of policy mix studies (Howlett and Rayner Reference Howlett, Rayner and Howlett2022) has examined important questions about how these bundles of policy tools and goals develop and evolve over time, how they can feature conflicting or complementary relationships among their key elements, and precisely how this contributes to policy (in)effectiveness (Bali, Howlett and Ramesh Reference Bali, Howlett and Ramesh2022). This research in the policy sciences has connected with concepts developed in earlier work on social and political institutions and the descriptions of their development through processes such as conversion, drift, layering, and replacement (Béland Reference Béland, Stoney and Bruce2013; Thelen and Steinmo Reference Thelen, Steinmo, Steinmo, Thelen and Longstreth1992) to form a new orthodoxy in the study of comparative policy analysis (Howlett and Rayner Reference Howlett, Rayner and Howlett2022; van der Heijden Reference Van der Heijden2011).
These considerations around the emergence and interrelationship of elements of policy mixes have become increasingly important in the policy sciences and especially in studies of policy design (Howlett Reference Howlett2024). Largely replacing an earlier orientation on the study and evaluation of single tools, scholars interested in policy optimality have increasingly turned their gaze toward topics such as policy integration and better understanding of the manner in which the components of policy interact and how these interactions can be oriented and managed (Howlett Reference Howlett2004; Howlett, Mukherjee, and Rayner Reference Howlett, Mukherjee and Rayner2014). This is true even as many questions remain about the topic, such as how precisely to measure the complex interconnections among policy objectives, means, and instruments within a policy mix to better analyze and design public policies. Such interactions can often create non-trivial issues in terms of design and evaluation (Bouma et al. Reference Bouma, Verbraak, Dietz and Brouwer2019) and require clearer analysis and conceptualization.
Three aspects of the issue stand out. First, focussing on mixes as a design subject raises the complexity of what is being designed and why from earlier work looking at single tools (Trebilcock and Hartle Reference Trebilcock and Hartle1982; Tinbergen Reference Tinbergen1952; Knudson Reference Knudson2009). Policy mixes are typically more complex, sensitive to initial conditions and subsequent developments in their specific policy area(s) and final outcomes regarding policy success and failure depend on multiple variables that vary over time, not all of which can be controlled by policy-makers (Maor and Howlett Reference Maor, Howlett and Howlett2022). Hence, with more elements to reconcile and coordinate in a design, policy-makers must strive not just to choose a single tool but to harmonize each of the individual elements within a policy mix to maximize the likelihood of policy success. Considerations related to how to enhance the consistency of the tools used, the coherence of policy goals, and the congruence or fit between goals and tools (Kern and Howlett Reference Kern and Howlett2009; Howlett Reference Howlett2019a; Gunningham, Grabosky, and Sinclair Reference Gunningham, Grabosky and Sinclair1998) thus have risen to analytical prominence but so far remain understudied. The integration between vertical (between levels of government) and horizontal (within the same level of government) policy dimensions, for example, is critically relevant to the success of policy mixes (Howlett, Vince, and del Río González Reference Howlett, Vince and del Río González2017).
Second, it is also the case that many different types of policy mixes can exist depending on factors including the complexity of the policy-making environment (from single to multi-level governance) and the number and type of problems being addressed. These “horizontal” and “vertical” aspects directly affect topics such as the number and type of actors involved in policy formulation, decision-making, and implementation, their tool and goal preferences, and how they are reconciled (or not) in the design process (Howlett & del Río González, Reference Howlett and del Río González2015).
Third, there is a distinct temporal dimension to mixes: that is the age of the mix is also a concern since successive processes of policy change such as layering, patching and packaging of various measures can heavily impact its coherency and ability to achieve intended goals (Howlett Reference Howlett2019b; Reference Howlett2024). That is, crafting a more effective alternative after some period of time through wholesale “replacement” of an existing mix may not always be possible and it is fully possible that even an originally superbly crafted and well-integrated policy will degenerate through successive rounds of layering or incremental adjustment. Thus, not unexpectedly, we commonly find in practice not only that different policy mixes operate in different sectors and issue areas but also that in many of those mixes tools and goals are incongruent, incoherent, and inconsistent with one another and that much policy work involves trying to hold them together in the face of increasing pressure on their various parts (Howlett Reference Howlett2018).
Better understanding how mixes are initially composed and how they evolve are therefore key questions for analysts and practitioners alike.
While evidently these questions require study, existing methods used to describe or diagnose a policy mix only partially address the needs of this essential work. Many studies, for example, focus on “density” or the number of tools found in a mix (Knill, Schulze, and Tosun Reference Knill, Schulze and Tosun2012; Lindberg, Markard and Dahl Andersen Reference Lindberg, Markard and Dahl Andersen2019; Schaffrin, Sewerin and Seubert Reference Schaffrin, Sewerin and Seubert2014; Reference Schaffrin, Sewerin and Seubert2015) either singly or alongside other measures designed to assess the “intensity” of tools use such as the extent to which they deploy coercive methods to compel compliance with government intentions (Schaffrin, Sewerin and Seubert Reference Schaffrin, Sewerin and Seubert2014; Reference Schaffrin, Sewerin and Seubert2015; Maor and Howlett Reference Maor, Howlett and Howlett2022; Attwell and Navins Reference Attwell and Navin2019).
These kinds of approaches and methods inherently assume that the differences in the number of tools found in a mix and the number of them that are compulsory are significant, although it is difficult a priori to assess whether this is correct and also to predict precisely what impact on policy such differences in numbers might have. With respect to density, for example, the suggestion is often that a denser mix is more effective. But consider for example, the fallout from the recent Dobbs v. Jackson Women’s Health Organization (2022) decision whereby the US Supreme Court reversed Roe v. Wade: the abortion bans that followed in many US States were quite simple but, have had a critical policy impact (Davis Reference Davis2022). While adding the second intensity measure tries to correct for the problems with a purely quantitative density measure by adding a qualitative one qualifying any pure increase in numbers, precisely how to operationalize and measure “intensity” is also problematic (Maor and Howlett Reference Maor, Howlett and Howlett2022; Virani et al. Reference Virani, Singh Bali, Cashore, Howlett and Ramesh2024). While “intensity” is often linked to degrees of “coercion” (Schaffrin, Sewerin, and Seubert Reference Schaffrin, Sewerin and Seubert2014), for example, it is difficult to assess whether a potential jail sentence is more coercive than a significant fine (Baxter-Moore Reference Baxter-Moore, Jackson and Jackson1987) or how such a measure helps us understand a topic such as the effectiveness of welfare service delivery (Bode Reference Bode2006).
The policy mix literature needs to better define and analyze mix creation and evolution if it is to aid better portfolio analysis, design and practice (Maor and Howlett Reference Maor, Howlett and Howlett2022) and this paper develops a new technique for so doing.
Here we propose an additional way of assessing a policy mix which is different from simply counting the number of tools, policies, or levels of government in a mix (Howlett and del Río Gonzáles Reference Howlett and del Río González2015; Howlett, Vince and del Río González Reference Howlett, Vince and del Río González2017; Schaffrin, Sewerin and Seubert Reference Schaffrin, Sewerin and Seubert2014; Reference Schaffrin, Sewerin and Seubert2015) and also differs from trying to quantify tool “effort” (Maor and Howlett Reference Maor, Howlett and Howlett2022). Rather we argue that what is most important about a mix is its original purpose. In particular, we argue that using a matrix linking the dual continua of the extent of flexibility/standardization expected of a mix and of its intended maintaining/innovating behavior provides a superior method of assessing the logic and differences and similarities found to exist between mixes. This technique also helps better describe and explain why mixes evolved the way they did, aiding better design and analysis of such complex entities over time.
Of course, this is not to say that policy mixes have an intent of their own, but rather that decision-makers and administrators are purposeful in designing policy and structuring its components so that either innovation or maintenance ambitions shape the design alongside the desire for a mix to be responsive to change or resistant to it. How exactly this is done depends on a variety of variables including the external environment decision-makers face, their available resources, and the salience of the issue, among others, but these concerns always exist within a purposeful framework. Due to the multidimensional and complex nature of most policy fields, it is almost impossible to “precisely” direct a policy mix over any significant span of time, so decisions are rather taken among broad dimensions and orientations of policy packages such as innovation or maintenance, standardization or flexibility. In this sense, policy-makers “set the stage” for the outcomes from a policy mix as best as they can depending on their preference to either change or maintain current processes and outcomes.
As is argued below, a model of policy portfolios focusing on how a mix furthers these options provides greater insight into their formation and dynamics than do existing quantitative or modified quantitative models. The article is organized into four sections: after this introduction, we discuss our framework within the scope of the policy mix idea, then provide two illustrative cases of policy mixes in high-profile sectors in Canada – banking and pensions – to establish the benefits of this approach. Finally, some conclusions and future research directions are provided.
Better defining and operationalizing the concept and purpose of a policy mix
The existing literature on mixes in the policy studies field has faced criticisms for being overly generic and therefore reducing the utility of the analysis (Rogge and Reichardt Reference Rogge and Reichardt2016): and for improperly confusing key terms and phenomena – such as “policy mixes” with “instrument mixes” – that is to say, for having strongly prioritized the analysis of policy tool combinations whereas both goals and tools need to be included in any analysis of a portfolio of policies (Capano and Howlett Reference Capano and Howlett2020).
Calls for more sophisticated and detailed definition and analysis of mixes can thus be seen in such areas as innovation (Flanagan, Uyarra, and Laranja Reference Flanagan, Uyarra and Laranja2011): sustainability (Rogge and Reichardt Reference Rogge and Reichardt2016): and resource governance (Howlett and Rayner Reference Howlett, Rayner and Howlett2022): with the general concern being to better describe and model a mix in order to be able to better inform how to align policy designs to the achievement of policy goals (Howlett and Rayner Reference Howlett and Rayner2007). This is especially relevant in fields where complex mixes are common and both observers and practitioners alike need more clarity about policy mix design, impacts, and outcomes both when a mix is originally created and as it changes over time.
Such work requires a more accurate and parsimonious method of assessing the nature of a policy mix, how one mix differs from another, and why any such difference is significant, than currently exists (del Río González and Howlett Reference del Río González and Howlett2013; Maor and Howlett Reference Maor, Howlett and Howlett2022).
As noted above, much existing effort has focused on distinguishing between different kinds of mixes using quantitative measures such as “density”, or the number of governments, policies, instruments, or tools found in a mix (Knill, Schulze, and Tosun Reference Knill, Schulze and Tosun2012; Lindberg, Markard and Dahl Andersen Reference Lindberg, Markard and Dahl Andersen2019; Schaffrin, Sewerin and Seubert Reference Schaffrin, Sewerin and Seubert2014 and Reference Schaffrin, Sewerin and Seubert2015), and the “intensity” of tools use (Schaffrin, Sewerin and Seubert Reference Schaffrin, Sewerin and Seubert2014; Reference Schaffrin, Sewerin and Seubert2015). While this approach has been useful as a first cut at the subject, as was pointed out above, it is not clear, besides serving as a rough measure of “complexity,” what are the implications of having more or fewer governments involved in a mix or what is the impact of a difference between a mix with a large number of elements and a simpler one or between a more, or less, “severe” or coercive one. Important design notions such as “parsimony” and “elegance,” of course, do rely on a better appreciation of the match between the number of elements and the attainment of the aims and ambitions of the policy but also emphasize that “bigger is not always better” and that policies with fewer elements if carefully designed and implemented, may be as, if not more, effective than those featuring more tools or more coercive ones (Tinbergen Reference Tinbergen1952; Howlett Reference Howlett2024).
The literature in general thus retains a distinct tendency to discuss mixes as a collection of policy instruments but, rather than only count the number of elements involved in it, one must also examine why a mix is created and especially the diachronic aspects of this question. Neglecting its objectives is especially problemmatic (Rogge and Reichardt Reference Rogge and Reichardt2016; Maor and Howlett Reference Maor, Howlett and Howlett2022). That is, analysts need to harken back to the earliest definitions of policy mixes in the policy literature which describe them as “complex arrangements of multiple goals and means which, often, have developed incrementally over many years” (Kern and Howlett Reference Kern and Howlett2009, 395).
As is argued below, policy goals are key, and understanding how much change and flexibility was anticipated when a policy mix was first created and how have those two parameters changed as a policy evolves provides an indication both of the purpose of a portfolio and the ability to judge whether or not that purpose has been retained as the policy has changed. It provides a baseline template against which additions and deletion of tools and objectives can be assessed, and, of course, a set of parameters informing overall policy design.
Key goals for policy mixes: flexibility and innovation vs rigidity and standardization
In attempting to achieve anything whatsoever, governments must decide whether to more or less retain the status quo or to innovate and they can do so in either a flexible or rigid way. Policy mix designs reflect these initial decisions and subsequent adjustments or changes are made to them as time passes. But in all cases, mixes can be classified according to these two criteria and so doing improves greatly on existing models of mixes and government mix behavior in terms of providing instruction to policy advisors, formulators, and decision-makers regarding the point and purpose of the mix.
Framing the topic this way with a focus on the general nature of the goals or objectives of a mix qua mix complements earlier more quantitative studies and helps to better understand why and how these combinations may have altered over time. It enables observers to better assess the current status of a policy and how its environment and target populations and objectives may have changed.
In other words, simply saying that a policy mix is complex or “dense” or has become more complex or denser is not as helpful or as instructive towards understanding how the proposed level or nature of the mix matches, or not, its impact and intention on the ground. That is, unlike earlier measures, the examination and identification of the stability and flexibility of an initial policy design, can provide a superior measure or indicator of important aspects of a design and expectations surrounding it and its subsequent alteration, such as its degree and type of robustness over time (Howlett and Ramesh Reference Howlett and Ramesh2022).
Operationalizing policy mix intent: flexibility and attitude toward the status quo
In better understanding, the purpose of a mix, the dimensions of government activity identified by the Observatory for Public Service Innovation and its analysis of organizational modeling around innovative practices (OPSI 2018) are helpful in characterizing the general intention of a mix and the cumulative impact of each of its component elements.
The first dimension of a mix, following the logic identified by OPSI, concerns whether its purpose is to maintain an existing policy position or to promote innovative alternatives. The second dimension concerns the aggregate level of standardization/flexibility expected of it, meaning whether it is intended to change and respond over time to different situations and contexts (robustness) or if it promotes a standard response regardless of such alterations.
Operationalizations of these two dimensions can be derived from the model proposed by Denyer (Reference Denyer2017) with regard to organizational resilience, which is structured around the dimensions of progressive-defensive and consistency-flexibility of effort through which organizations manage their core businesses.
Figure 1 below brings together these two key continua.
The lower quadrants in Figure 1 represents the realm of relatively standard policy responses to a problem. While the precise content of these varies by country/jurisdiction and policy style, they are fairly predictable and often use the same “package” of policy solutions, such as a penchant for the use of regulation or any other policy tool commonly deployed in that jurisdiction (Kagan Reference Kagan2001; Howlett and Tosun Reference Howlett, Tosun, Howlett and Tosun2021; Tosun and Howlett Reference Tosun and Howlett2022).
The bottom-left quadrant, where Standardized Maintenance policies are located, focuses on the creation and maintenance of a mix featuring standardized action, highly repetitive actions that can be standardized and optimized such as bureaucratic routines. Many activities in public service operate more or less exactly so, especially for service delivery and finance where policy mixes are designed specifically with the aim of providing predictable, standard responses (Lægreid, Reference Lægreid, Howlett and Mukherjee2018).
If a slightly more flexible response is desired, a mix devoted to Flexible Maintenance emerges, which operates on the premises of mechanical resilience. This is well represented by the semi-automated responses that can be planned to emerge when a known disruption of the system occurs, such as an up or down movement in a commodity price cycle. Thus, for example, large snowfalls in a city like Vancouver represent a disruption, but policy-makers expect and deal with them effectively and efficiently by adding some flexibility to otherwise standardized responses to winter weather allowing some contingency for multiple successive snowfalls over short periods of time, or abnormally cold or long winters. The response is generally well understood and additional resources are put in place to secure resilience, hence the system prefers to lean on maintaining existing approaches. Often, these responses come from experience and expertise accumulated over many years and key performance indicators are used to monitor service delivery and manage implementation in a flexible way.
The top quadrants represent the extent of the desire for more innovative responses (i.e. they are more exploratory/anticipatory efforts). This varies in terms of its static or dynamic aspects: whether it refers to changes in policy means while goals are held constant, or if a design also allows goals to also change over time (Howlett and Ramesh Reference Howlett and Ramesh2022).
These upper quadrants by definition involve more “complex” mixes since they introduce further goals of innovativeness into a policy and thus into the policy mix space. The desire to promote innovation rather than standard responses involves more open-endedness in policy responses, allowing decision-makers more agility in tackling unexpected problems or problems not expected to respond well to more standardized measures.
In the case of the upper left-hand quadrant (Howlett and Ramesh Reference Howlett and Ramesh2022): a mix is expected to meaningfully alter tools and calibrations within it, while generally holding goals constant – resulting in a kind of standardized innovation. This is the situation with policy mixes put into place in many jurisdictions in order to deal with climate change impacts. Decision makers, in this case, need to have a response plan and policy towards events such as sea level rises, forest fires, and more severe storms and weather and realize that the impacts of these events may be largely unpredictable and will require not only agility in applying existing rules and procedures for housing and energy grids, for example, but also more innovative actions developing alternative energy sources moving well beyond just enhanced disaster relief provisions.
Finally, Flexible Innovation, represents the most potentially non-linear quadrant, underpinned by the goal of dynamic robustness where both policy aims and activities are allowed to “float” and provisions are made in policy mixes not only to alter policy means but also policy goals in the face of future circumstances (Howlett and Ramesh Reference Howlett and Ramesh2022). This represents a policy response needed to plan against very open-ended issues, wicked problems, and highly interconnected policy areas such as financial crises and social problems from homelessness to unemployment. This kind of policy mix can be found in banking regulation legislation, for example, whereby periodic severe crises are to be expected but whose precise contours and correctives are difficult or impossible to predict. This last type of policy mix highlights the use of various kinds of procedural tools allowing policies to self-adjust as circumstances unfold, such as ongoing review committees and decision-making bodies – such as independent regulatory agencies - which can meet and adjust faster than can traditional legislative and executive actors (Bali et al. Reference Bali, Howlett, Lewis and Ramesh2021).
All four types of mixes feature multi-instrument and often multi-policy and multi-level elements (Howlett and del Río González Reference Howlett and del Río González2015): but the number and specific severity of tools, and the number and level of governments involved in a mix are less significant to the subsequent evolution of the elements of a mix than the general purposes and expectations concerning a policy design. That is, the initial expectations about the current and future conditions existing when a mix is first put into place, and how those original elements are expected to, and actually do, change as the environment and mix evolve, are key indicators of its character and ability to change over time.
Altering policy mix type
Policy mixes by definition contain a number of distinct types of policies and policy instruments with different characteristics. However, some regularity in terms of policy changes and dynamics can be designed into mixes through the use of particular kinds of tools. Where a stationary kind of resilience is a key design goal it can be expected that a mix will be composed mainly of standardized policy elements whose contours, problems, and expected deliverables are well known. Thus a purely standard intervention where change is expected to be constant and linear exists in many welfare policy regimes, for example, which feature relatively standardized welfare payments. Where a policy needs to respond to modest change with resilience and flexibility to changing circumstances, on the other hand, the trajectory of the mix may instead approximate a regular sine wave as changes are implemented to restore the status quo ante in repetitive and cyclical ways, such as when pension, welfare or unemployment payments are designed to retain their value in inflationary times through periodic planned adjustments linked to prevailing inflation rates. This is what Salamon (Reference Salamon1989) referred to as “automaticity.”
In the case of mixes that are expected to be more innovative, changes would be expected to be less linear and more irregular as, for example, occurs through provision of additional welfare payments made to individuals who reach certain age, income, or family limits or other such demographic criteria which often vary in fairly predictable ways but nevertheless place a great deal of stress on existing payment systems. These mixes require some procedural mechanism to allow for changes in payment types beyond simply the automatic adjustment of existing ones, such as the creation of periodic reviews empowered to collect information and make recommendations on policy changes (Lang Reference Lang2019; Reference Lang and Howlett2022).
In the ultimate case of mixes that are expected to be highly flexible, provisions can also be made to deploy alternative sets of tools that can, for example, shift welfare payments completely away from standardized models by adding new aspects, such as workfare requirements or social impact bonds which build in changes not only in instruments but in policy goals (Chiapello and Knoll, Reference Chiapello and Knoll2020; Tiikkainen, Pihlajamaa, and Åkerman Reference Tiikkainen, Pihlajamaa and Åkerman2022).
Using this approach, we can see how the dynamics of each kind of mix will vary depending on the goals embodied within a mix. In all cases, however, the complexity of the mix can be seen to depend on the configuration of elements present within it rather than simply upon the number of items, governments, or policies involved (Knill, Schulze, and Tosun Reference Knill, Schulze and Tosun2012; Bauer and Knill Reference Bauer and Knill2014; Schaub et al. Reference Schaub, Tosun, Jordan and Enguer2022). The superiority of the former approach to the study, assessment and design of policy portfolios is illustrated below in the presentation of two cases based on the Canadian experience with the creation and evolution of financial and social welfare mixes.
Two Illustrative cases: Canadian banking and pension policy mix designs and their evolution
The enhanced utility of this approach to the study and design of policy mixes can be illustrated through a comparative examination of existing mixes using older quantitative tool-oriented and newer more qualitative goal-oriented techniques.
For this purpose, the policy mixes found in two prominent Canadian policy spaces of the types discussed above are examined using both more traditional counting methods and the model of flexibility and innovativeness set out above. These are the regulation of the banking system, which is designed to respond to crisis and is undertaken through periodic reviews of banking legislation, and old age pensions, which are designed for stability and routinization. We show how in both cases if traditional quantitative methods are utilized, incorrect sets of conclusions will be drawn concerning the nature of the existing mix and its dynamics, obfuscating its lessons for policy design.
More specifically, it is shown that using purely quantitative density and intensity measures, the Canadian banking policy mix appears to be neither very dynamic nor robust. Banking policy does not employ many policy tools, is one of the few areas in Canada where only a single level of government is involved, and its main goal is a straightforward and relatively set one: bank stability. However, such an analysis masks a very rich policy mix in which various procedures have been implemented for well over a century in order to ensure dynamic system robustness in the face of both expected and unexpected, but anticipated, change. The impact and nature of these procedural tools in particular is missed by simply tallying up measures found in various enabling Acts and legislation.
In contrast, when assessed using existing measures, the Canadian old-age pension policy mix appears to be rather complex: since it involves multiple jurisdictional layers and a host of different policy tools and policy goals that vary across these different jurisdictions. However, when utilizing the new method outlined above it becomes clear that the processes at work here are highly standardized and the system is very stable and routinized, if not completely static.
The cases were also chosen because they developed broadly over the same period of time, therefore that possible problem of comparing different governments’ policy goals at two very different points in time is minimized. It is worth mentioning here that the two cases are also useful because they carry different “specific weights” in terms of political impact: pensions are an important issue and failure there can have grave repercussions on individuals, but this policy field is not nearly as sensitive as the banking one, where policy failures can potentially have devastating consequences for the whole economy.
When we apply our framework to these policy mixes, we can better highlight their specific characteristics and provide a more nuanced analysis of these policy mixes than is currently the case, illustrating their usefulness in helping to better understand and design such arrangements in these and every other sector.
The Canadian bank act: dynamic robustness in a complex global landscape
Banking regulation in Canada centers on the Bank Act (1991, c. 46): the last incarnation of a legislative framework dating back to 1871 and which mandates a full-scale review and revision of the act every ten years.
The policy network in Canada that supervises the financial system is comprised, besides financial institutions like banks, of five public entities: the Bank of Canada; Finance Canada, the federal Department in charge of the sector’s legislative and policy strategy; the Office of the Superintendent of Financial Institutions, the agency in charge of monitoring the “financial health” of banks, pension plans and insurance companies; the Canada Deposit Insurance Corporation, where responsibility for managing risk in the banking system is allocated; and finally the Financial Consumer Agency of Canada that is in charge of ensuring compliance by financial institutions towards clients. Each of these agencies has an enabling Act that gives it regulatory authority over aspects of the banking system and overall comprises a relatively small number of measures for a policy mix regulating trillions of dollars in deposits, loans, and other financial instruments.
While the system is comprised of multiple actors, it is quite small by the standards of some countries, like the US, as it is a branch banking system based on the presence of fewer than a dozen large federally regulated banks and a number of small provincially regulated credit unions and other similar actors. Even this small number has drawn criticism and charges of fragmentation (Williams Reference Williams2012) and starting in the 1980s the Canadian government began to allow mergers between banks and non-bank actors such as trust companies and stock brokerages, in the process strengthening its oversight over the industry but also increasing the oligopolistic nature of the sector. These regulatory changes stabilized and rationalized the sector, in the name of heightened resilience and resulted in less dramatic swings in bank failures than in many other countries during major financial crises such as that which followed the US housing crisis of 2007–2008 (Williams Reference Williams, Lindquist, Howlett, Skogstad, Tellier and Hart2022).
These most recent set of reforms of the Canadian financial system moved through various phases: the first – between 1987 and 1997 – beginning with the creation of Office of the Superintendent of Financial Institutions (OSFI) and a renewal of the Bank Act. These changes hinged on “de-pillarizing” the sector, effectively reversing previous policy dating from the 1930s which prevented banks from owning trust companies and brokerages, allowing the Canadian Big Five banks to enter the securities sector, but stopped mergers among them so as not to create excessive concentration, and also retained a separate insurance industry (Williams Reference Williams, Lindquist, Howlett, Skogstad, Tellier and Hart2022). This policy shift was mainly motivated by the global pressure exerted on the international financial system by increasing interconnection and cross-border developments which featured the emergence of very large multi-national banks which threatened the commercial viability of the relatively smaller Canadian financial institutions. Ultimately, the de-pillarization period refocused the sector from a banking policy to financial services policy writ large (Williams Reference Williams, Lindquist, Howlett, Skogstad, Tellier and Hart2022: 251): even if the system remained complex and in certain senses still fragmented by the exclusion of some sectors, such as insurance companies (Roberge et al., Reference Roberge, Dunea and Alan Williams2015).
In the late 1990s, a second phase began with concern centered around risk management of what had become a highly concentrated sector in the face of possible future financial shocks. The main tool deployed for this was a Task Force on the Future of the Canadian Financial Services Sector (the MacKay Task Force) which discussed multiple contrasting future visions of the banking sector and prepared the system for(un)expected shocks, which did ultimately occur in the form of the 2007–2008 Global Financial Crisis. The changes brought in at the time eliminated regulatory gray areas among banks, insurance companies, and securities and favored the development of a Canadian financial sector that, while more centralized and concentrated, was also more robust, and better prepared to face global financial challenges (Porter and Coleman Reference Porter, Coleman, Clement and Leah2003).
Critical to this series of events and alterations to policy mixes in the Canadian financial sector was the practice of entrusting to Parliament the periodic thorough review of the Bank Act with a mandate to recommend and implement changes needed to face new circumstances.
This review is an important procedural tool in the banking policy mix which allows Parliament to call upon various organizations, like OSFI and consumer advocates that have a watchdog role, expanding the number of voices with a stake and impact in the policy sector and allowing them a space for their concerns and future expectations to be heard. These reviews also help de-politicize the sector, which means that partisan agreement on the goals and tools for financial policy is possible in Canada unlike, for example, in the United States where regulators are often opposed by companies and opposition parties. As a result, it is possible for Canadian governments to agree on policy goals and implement instruments that increase the system’s resilience (Williams Reference Williams, Lindquist, Howlett, Skogstad, Tellier and Hart2022; Bordo, Redish, and Rockoff Reference Bordo, Redish and Rockoff2015).
This approach has allowed the system to resist demands that have undermined the resilience of banking sectors in other countries, like loss of competition via mergers or unregulated entry of foreign actors (Booth Reference Booth2009): and features a set of tools that allow some innovations and flexibility in the system.
Generally speaking, this strategy has succeeded in the sense that the Canadian banking sector has proven to be more resilient than that found in many other countries, effectively navigating both the 2008 financial crisis and the COVID-19 economic downturn (Roberge Reference Roberge, Jesuit and Alan Williams2017; Baron Reference Baron2013; Dostie Reference Dostie2020). This story, however, would be lost by simply counting the number and/or coerciveness of the policy tools found in the banking policy mix, which actually declined during this period of retrenchment. The presence of important procedural tools – like the decennial Bank Act review – would be ignored and the analysis of change in the sector would be considered to be one moving towards simplicity – fewer larger actors – with some modest changes to regulatory activity. The entire point and purpose of the new mix – to continue to enhance resilience and stability in the face of global challenges – would be missed.
Canadian pensions: standardized responses and maintenance in a layered policy mix
While the banking policy mix is relatively small in terms of actors and tools and is focussed on a single - federal - level, the Canadian pensions regime exists as multilayered multi-governance system reflecting the constitutional division of powers found in the country and is negotiated and operated in a very de-centralized fashion. Pensions are a relatively new area and were created specifically as one of the few areas of constitutionally entrenched joint jurisdiction in the country, allowing provinces to operate their own systems, which Quebec currently does (Simeon Reference Simeon1972).
The current Canada Pension Plan (CPP)/Québec Pension Plan (QPP) regime was created in 1966 when Ottawa finally went from an early earnings-related contributory model that had been in place since the 1920s (Banting Reference Banting1987) to a new system where provincial means-tested payments were replaced by a new universal federal-provincial system. The older pension payments were supplemented by the federally administered flat-rate Old Age Security (OAS) pension (which had been created in the early 1950s): and an income-tested Guaranteed Income Supplement (GIS) to support low-income seniors created in 1967.
Effectively, the Canadian system is quite “dense” having been constructed of three layers of pension: the first is the most basic CPP/QPP payments, with age and income-tested supports for lower-income individuals added through a second tier of OAS and GIS payments. The third is a private layer that includes personal savings schemes (RRSP/TFSA and others) and workplace pension savings plans and payments (World Bank 2017).
Significantly for the discussion here, however, despite this multi-level and multi-tool complexity, which includes more recent changes in how pension plan investments are invested and managed such as the creation of the Canadian Pension Plan Investment Board in 1997, Canadian pension policy has highly favored routinization and policy stability, if not complete inertia (Béland Reference Béland, Stoney and Bruce2013; Béland and Weaver Reference Béland and Kent Weaver2019).
Only two major shifts have occurred since the 1980s and both occurred due to the recognition of impending financial and demographic pressures that threatened to undermine the financial viability of the existing contributory public system (Béland and Myles Reference Béland, Myles, Bonoli and Shinkawa2005). In the mid-1980s as the likelihood of future demographic shifts associated with an ageing working population became clear, Ontario kickstarted a country-wide move to shift the management of pension assets from government to investment boards and corporations operating much more independently and in private sector-like fashion, in the effort to secure a higher return on pension savings (World Bank 2017; Little Reference Little2008). In the mid-1990s Ottawa and the provinces agreed to reform of both payments and payouts while creating a new investment branch (CPPIB) in the same vein. But, given the opposition of Québec to cutting benefits, the only possible solution to long-term issues was to increase payroll taxes to fund future pension liabilities, which duly occurred.
As this brief illustration shows, pensions in Canada have come under pressure from demographic and economic variables, but the policy dynamics of the whole policy area are heavily affected by the desire for stability and long-term solvency of the existing plans (Béland and Weaver Reference Béland and Kent Weaver2019). This means that policy innovation is discouraged in this sphere, and policy shocks like crises tend to be the prime movers for changes, which are taken in a fashion congruent with the original design and intent of the funds to promote long-term plan solvency while reducing elderly poverty. Incremental modifications such as altering the calibrations of existing instruments – for example increasing contribution rates – thus tend to be the norm.
In this sector, it is clear that strategic policy goals are quite constant, with a large amount of inertia/policy robustness being the default position. In this, our model matches the findings and conclusions of observers such as Béland and Weaver (Reference Béland and Kent Weaver2019) and generally the literature on the topic, but it provides an additional perspective on what would otherwise generally be considered a complex policy mix gaining additional complexity. While we see a host of policy instruments at play across multiple jurisdictions, the reality is that most change occurs within the general aim of “flexible maintenance” (retouching of contribution mechanisms) with the occasional timid forays into the “standardized innovation” quadrant (modest reform of fund management practices).
Hence the apparent complexity of the Canadian pension system when measured in terms of the number of policy instruments and layers of decision-making masks a relatively standardized situation that shies away from any form of dynamic robustness. This crucial dynamic is lost in analyses that focus only on the density or intensity of elements within a policy mix.
Conclusion: reassessing the dynamics of policy-making
We should note here that this call for a better and more goal-oriented approach to studying and characterizing policy mixes – strictly speaking – is not a novel call: Wildavsky (Reference Wildavsky1979) and Simon (Reference Simon1997) both argued convincingly that policy goals and policy means are in constant interaction and that it is best not to separate the two when assessing policies. Hence, this article proposes in part to return to this approach by advocating the use of a different approach to assessing the nature and purpose of policy portfolios than existing methods largely involving counts of policy measures (Howlett and del Río González Reference Howlett and del Río González2015; Howlett, Vince and del Río González Reference Howlett, Vince and del Río González2017; Schaffrin, Sewerin and Seubert Reference Schaffrin, Sewerin and Seubert2014; Reference Schaffrin, Sewerin and Seubert2015; Maor and Howlett Reference Maor, Howlett and Howlett2022).
Rather, we argue that what is most important in understanding the nature of a policy mix is its original purpose. And, in particular, that using a matrix linking the dual continua of the desired extent of flexibility/standardization expected of a policy and of its intended maintaining/innovating behavior provides a superior method of assessing the logic and differences and similarities existing between mixes and understanding and explaining why they have evolved how they did.
That is, we argue that decision-makers and administrators are purposeful in designing mixes and arranging their components so that either innovation or maintenance and flexibility or standardization are at the core of a design. In this sense, policy-makers “set the stage” for the outcomes of the policy mix as best as they can, premised on their desire to either change or maintain current processes and outcomes in the face of anticipated futures. Since the multidimensional and complex nature of most policy fields makes it almost impossible to “precisely” direct a policy mix, design decisions instead are typically taken along the broad dimensions of innovation/maintenance and flexibility/routinization set out above.
Regarding the two Canadian policy mixes analyzed above, applying our model enables greater granularity about the purpose of a portfolio, how the types of policy tools that are used are distributed, and how they interact together in a mix to reinforce or alter this purpose (Bali et al. Reference Bali, Howlett, Lewis and Ramesh2021). These two cases illustrate the benefits of this way of looking at mixes and help uncover critical aspects of the relationships among tools and between tools and goals that are lost in more quantitative analyses of tool and goal numbers.
Thus, for example, in the case of Canadian pensions, looking only at the policy mix’s density would lead to its categorization as a complex one, potentially involving many changes due to the many actors and levels of government involved in the sector. Whereas the distribution of the various instruments shows how the policy was designed to operate in a simple, continuous fashion.
The same is true, in reverse, for the banking system which features a much lower density in terms of the number of policy instruments and actors, which with a counting metric would suggest stability and routinization, whereas the system is in fact quite flexible and adaptable and is specifically designed to be that way.
Overall, the new model allows us to better assess the nature of policy mixes as they actually exist, and to consider in more detail and with more precision how specific instruments may fit within a policy design scheme, or not, and with what effect on policy dynamics (Virani et al. Reference Virani, Singh Bali, Cashore, Howlett and Ramesh2024). Not only does the model allow us to better take into consideration the differences and different effects of substantive and procedural tools in policy designs, but also helps us distinguish and determine how these tools reinforce each other in specific ways, such as in what Bali, Howlett, and Ramesh (Reference Bali, Howlett and Ramesh2022) described as “primary” or “secondary” roles given their contribution to achievement of goal expectations.
Future research should focus on applying this model to more cases to test its reliability under different conditions and explore to what extent the model holds with mixes belonging to different national jurisdictions and systems of government and when more sectors are examined. This information is key to successful policies and would help advance a better understanding of the nature and purpose of a mix, something which is needed to ensure both better policy designs and better policy outcomes.
Data availability statement
This study does not employ statistical methods and no replication materials are available.