Wicked problem definition and gradual institutional change: federalism and immigration in Canada and Australia

Abstract This article examines the impacts of problem definition, defined as a social mechanism, in bringing about gradual institutional change. Focusing on a similar process of gradual institutional change in Canada and Australia, it shows that problem definition is one pathway by which actors’ interests and behaviors are redefined inside an institutional regime. By tracing the process of federalization of Canada and Australia’s immigration regime since the 1990, it demonstrates that problem definition contributed to the rise of subnational governments as legitimate actors in the management of immigration. In these two countries, the specificities of the operation of this mechanism, including the actors mobilized for change, and the content of the policy problem being put forward generated different processes of federalization that nonetheless resulted in inclusive immigration federalism. In dialogue with historical institutionalism, this points to the potential of the mechanismic approach for theory building regarding the consequences of the dynamics of problem definition.

translations of the balance of power between actors and of the needs of a society, at a given point in time.
Despite cross-case differences, the postwar immigration institutional regimes of Canada and Australia converged toward liberal principles and policies. At the same time, these regimes did not translate an important characteristic of both countries: their federal nature. Indeed, the two countries are federal systems with different, yet highly comparable, Westminster-inspired institutions. Interestingly, this federal dimension had been, up until recently, absent from Canada and Australia's immigration institutional regimes. Instead, both of these regimes featured the federal government as the dominant or as the sole legitimate government actor. This paper considers, via institutionalism and problem definition theories, the process by which federalism has gradually made an entry into these regimes since the mid-1990s. As an outcome, the rise of immigration federalism in Canada and Australia represents a considerable break in the internal practices of these institutional immigration regimes. The two countries depart from other federations where immigration federalism also occurs: it is not associated with seemingly 'unwanted' or 'uncontrollable' migration movement, nor is it focused on rights restriction or exclusion. In addition, the process of federalization of national immigration regimes (Paquet, 2014a) has occurred in the absence of immigration-related critical junctures and without an overhaul of the general principles infusing Canada's and Australia's approaches to immigration. For these reasons, dynamics endogenous to both countries' institutional regimes must be explored to account for change and to develop new theoretical insights on immigration federalism.
Comparing these most similar cases (Boucher, 2013), this article makes two claims. First, I propose that problem definition -considered as a social mechanism -should be included in the toolbox of the variant of historical institutionalism that focuses on gradual processes of institutional change. Based on the work of Streeck, Thelen, and Mahoney, I show that problem definition has impacts on power distribution and on actors' behaviors within institutional regimes, two elements identified as the internal sources of gradual change. As such, both for policy scholars and for institutionalists, considering problem definition is crucial to identify the pathways linking policy outcomes and processes of change to the formal characteristics of institutions.
Second, I argue that the similar rise of immigration federalism in Canada and Australia has been supported by different conceptions of the policy problem and, more broadly, that different actors have activated the mechanism of problem definition. These differences, leading to comparable outcomes, are partially the fact of the divergent nature of each country's federal system. Canada and Australia are mature federations of continental proportions characterized by a large degree of executive federalism. In both countries, federal and subnational governments have legislative and executive authority on their areas of jurisdiction (Watts, 2015). Despite these similarities, Canada is generally more decentralized that Australia (Hueglin & Fenna, 2015). In addition, whereas immigration is a shared jurisdiction in Canada, it is a commonwealth jurisdiction in Australia. The process of federalization, that gave rise to immigration federalism as an outcome, does reflect these specificities, as each institutional configuration provide actors with different capacities to define policy problems. In Canada, the provincial leadership in defining the problem associated with immigration federalism has resulted in a federalized regime where subnational governments are highly active. In contrast, the commonwealth-led definition of the problem in Australia has resulted in a federalization from the top. As consequences, some Australian states have remained somewhat resistant to a larger role in immigration.
Yet, this paper shows that considering this outcome as solely the result of formal institutions is limiting. Precisely, it provides limited indications as to where change might originate from and how change might have unfolded. This analysis addresses these limits in two ways. First, by comparing two federations with concordant outcomes and comparable institutional features, it becomes possible to move away from idiographic and descriptive analyses of immigration federalism. Instead, a comparative outlook on Canada and Australia makes it possible to highlight empirical similarities and to move the analytical focus toward processes of change and the production of outcomes. Second, this article contributes to a better understanding of immigration federalism by a focus on mechanisms of gradual institutional change and, in this specific case, by considering the operation of problem definition. This is better done by identifying, in empirical settings, the leading conceptions of the policy problem, and the operation of the mechanism of problem definition, lead by specific actors. Here, comparing highly comparable cases with a mechanismic approach produces insights on the operation of a particular mechanism, in relation to different national and institutional contexts. This is a first step toward theory development regarding the role of problem definition in gradual processes of institutional change. As this paper demonstrates, comparing the federalization of immigration in Canada and Australia better retrace processes of change and to gain relevant information, to reject institutional determinism, and to understand current policy dynamics.

Immigration federalism in Canada and Australia
Immigration federalism, a US-born concept had been traditionally used to refer to the debates and tensions over the attribution of responsibilities and legislative capacities between the federal government and the states. 1 The concept has more recently been mobilized to describe the increasing activism of subnational government in immigration policy-making, as well as the growing '[…] involvement of multiple levels of government in immigration matters' (Baglay & Nakache, 2013, p. 335). Indeed, in the last 20 years constituent units in immigrant-receiving federations have become more active and mobilized over immigration-related issues, resulting in the emergence of both conflicts and modes of coordination between governments. As a relatively new phenomenon that challenges core assumptions about the role of the state in relation to immigration, most of the work on immigration federalism has considered single countries or has explored policy outcomes and impacts for immigrants and society (eg Baglay & Nakache, 2014;de Graauw & de Chantal, 2009;Filindra & Kovács, 2012;Joppke & Seidle, 2012;Reich & Barth, 2012;Thangasamy, 2015;Varsanyi et al., 2012). In relation to the empirical reality of immigration federalism, powerful explanations have been developed to account for anti-immigrant outcomes but less is available to explain pro-immigrant federalization. At this point, the rise of immigration federalism is attributed idiographically to several causes, including: the failure of national policies to address local needs; the mobilization of political or administrative entrepreneurs at the local level; partisan politics and nationalism; policy diffusion; and the recognition by 1 For an overview of the origins, content and applications of the concept in the american literature, see: (Varsanyi, lewis, Provine, & decker, 2012).
substate governments of the direct and indirect costs of federally controlled immigration on their own jurisdictions (eg education or health) (eg Gulasekaram & Ramakrishnan, 2015;Hepburn & Zapata-Barrero, 2014;Newton, 2012;Paquet, 2015Paquet, , 2016Suro, 2015). This paper builds on these accounts but proposes instead a process-focused institutional explanation of the rise of immigration federalism, an outcome that is qualitatively similar in Canada and Australia. While most visible in the United States, because of the contentious and exclusive nature of state ordinance and policies (eg Arizona's), immigration federalism in Canada and in Australia is quite alive and is becoming an important feature of governance. The rise of immigration federalism in these two countries has come as a relative break with the highly centralized and federally dominated management of these issues (Hugo, 2008;Paquet, 2014a). In Canada, the jurisdiction over immigration is formally shared between provinces and the federal government (article 95 of the BNAA), while in Australia the powers over immigration are the purview of the Commonwealth Government (articles 27 and 51 of the Constitution Act). Since immigrant settlement and integration -outside of land allocationwere not considered to be the responsibility of the state nor a clear policy objective, they were not an assigned jurisdiction in both countries at the time of constitutional drafting (Lake & Reynolds, 2008). The federal government, both in Canada and Australia, led to the establishment of institutional regimes that supported the gradual establishment of the contemporary national immigration after the WWII. These national models were characterized by immigrant recruitment and selection policies that allowed economic migrants to settle permanently and work at any location in the country. This resulted in a skewed distribution of immigrants in Canada and Australia; most of newcomers settling in provinces and states with urban centers, existing immigrant communities as well as vibrant economies. Up until the mid-1990s, subnational governments did not challenge federal leadership, nor the corollary federal dominance inside the institutional regime of each country.
In Canada, the French-speaking province of Quebec presented an exception to this pattern. Indeed, starting in the 1960s, the province mobilized toward gaining more power in immigrant selection and immigrant integration. In 1991, following a succession of immigration intergovernmental agreements and in the context of a constitutional crisis, most of immigrant selection as well as all of the settlement and integration services were devolved to Quebec (Paquet, 2016). Still in force, the 1991 Canada-Quebec Accord also includes an ascending financial compensation for the province (Becklumb, 2008). Before the 1990s, Quebec's mobilization toward immigration and federalism aligned with its nationalist movement and the desire to protect the French-speaking nature of its society. As such, while the demands and actions of Quebec resulted in a first break in federal immigration dominance in Canada, they were qualitatively different than the ones described in this paper. They nonetheless paved the way to problem definition by adding to some provinces' agenda the issue of cross-provincial fiscal equity with regard to immigration (Paquet, 2014a).
Beyond Quebec, the break from this pattern of federal dominance and subnational avoidance shared similar features in both countries. Contrary to other federations, immigration federalism is results in pro-immigrant policies in Canada and Australia. In particular, immigration federalism in Canada and Australia is characterized by a growing interdependence of both orders of government and demands for the extension of the benefits associated to immigration for subnational units as well as the recognition of the inefficacy of a 'one government' management of the challenges associated in with immigration. In practice, immigration federalism in the two countries displays three comparable features.
First, changes in governance practices that include a marked increase in intergovernmental collaboration and coordination over immigration-related matters (Schertzer, 2015). Second, autonomous design, funding, and implementation of immigration-related public policies by subnational governments. Third, the creation of policy and programmatic innovations providing subnational governments with the capacity to select permanent immigrants. In Canada, the Provincial Nominee Program (PNP) allows provinces to directly select immigrants for a settlement on their territory, based on provincially defined criteria since 1998. In 2015, Canadian provinces outside of Quebec nominated successfully 44,534 applicants for permanent immigration, which account for about 16% of the overall immigration intake of the country and 26% of the economic stream of the national immigration program (Canada, 2015(Canada, , 2016. Similarly, since 1995, different iterations of State-specific Migration Mechanisms (SsMM) are in place in Australia. All Australian states (and two territories) now have such programs, and some programs also target regions independently from state governments. They allow states to recruit skilled workers, potential entrepreneurs, and other categories of permanent immigrants (Australia, 2001). In 2015, Australian states have issued about 24,650 visas, which represents about 13% of the overall immigration intake of the country and 19% of the skilled migration stream (Australia, 2016).
These similar features indicate that immigration federalism is a reality in both Canada and Australia. In contrast with other federations, this breed of immigration federalism focuses on increasing the number of immigrants, ensuring their integration into the labor market, and on modifying the distribution of international newcomers across the territory. These features cannot easily be explained with new accounts of the rise of immigration federalism, which has had more success in conveying the sources of immigration federalism characterized by conflicts and immigrant exclusion. These features make Australia and Canada important cases to compare as a way to build better explanations of the processes by which immigration federalism becomes a feature of institutional regimes.

Problem definition as a mechanism of gradual institutional change
To account for the federalization of Canada and Australia's immigration regimes, this article builds on historical neo-institutionalism to consider the role of problem definition in bringing about gradual change (Radaelli, Dente, & Dossi, 2012). Despite its inherent fecundity, historical institutionalism has been highly criticized for producing research overly focused on stability and continuity (Thelen, 1999). The most promising conceptual innovation in response to these criticisms has been the relational and dynamic representation of institutions proposed by Streeck and Thelen (2005) and mobilized in Mahoney and Thelen (2010).
These authors propose that a focus on interactional dynamism is a first step toward overcoming the natural tendency of institutional studies to focus on stability. To do so, they propose to analyze political phenomena via the larger prism of institutional regimes instead of simply focusing on formal political institutions. The concept of regime includes: general institutions (institutional venues, public administrations, etc.); public policies; implicit normative and cognitive elements; and, most importantly, actor with various degrees of legitimacy (Streeck & Thelen, 2005, pp. 12-13). This definition puts forward formal elements (rules), interactional elements (actors and their relationships) and is based on a strong foundation that considers institutions as primarily a tool for resource distribution (power, legitimacy, money, etc.) (Mahoney & Thelen, 2010, pp. 7-8). All of these elements are the result of certain agreements, implicit or explicit, between within-institution actors and of practices reproduced constantly. These agreements, however, may also be the result of compromises produced by long-disappeared actors. As a consequence, the current practices of actors may challenge, deliberately or because of a mismatch with a new context, the basis of any given institutional regime. There is, as a consequence, an immanent dynamism to institutional regimes.
Two consequences have stemmed from the work of authors using this variant of institutionalism. First, efforts toward the identification of instances of changes unfolding in the absence of the usual exogenous forces leading to institutional evolution. Most of these are gradual; changes unfold over mid-range duration (10-30 years) and are not the result of forces originating from outside of institutions (Falleti, 2010). Second, this scholarship has shown that the identification of processes and mechanisms by which change operates gradually inside institutions is the key to counter the deterministic bias of historical institutionalism. That is to say, it demonstrates that theory development is better served if the focus of inquiry is placed on the 'how' of change, instead of on the 'why' or the 'what' of change.
A focus on mechanisms and processes allows for the explanation of internal dynamism, even in the absence of external indicators of institutional change. As such, this article proposes a mechanismic analysis of institutional change as a way to account for an analogous process generating change in highly comparable contexts. Distinct from the 'variable' approach to explanation, this approach aims to identify the mechanism(s) active in a specific causal chain (a process). Social mechanisms can be defined as a '[t] heorized system that produces outcomes through the interaction of a series of parts that transmit causal forces from X to Y' (Beach & Pedersen, 2012, p. 176). By nature, mechanisms are relational and non-deterministic (Falleti & Lynch, 2009); the outcomes they bring about are idiosyncratic, and are functions of the context of their operation (timing, politics, institutions, etc.) (Grzymala-Busse, 2011). The goal of a mechanism-based explanation is thus to document instances of the operation of a mechanism and to contribute to developing knowledge and theory about how mechanisms function. This goal well served by qualitative, small-n comparisons.
Most of the pathways to change identified within the gradual institutional change scholarship have focused on the role of actors' behaviors and of their changing interests. These acts like the gears activating gradual institutional evolution, by instigating chain reactions between interdependent actors and modifying feedback effects. Up to this point, Streeck and Thelen (2005) have summarized the major mechanisms animating processes of change as: defection, differential growth, deliberate neglect, redirection, and depletion. In addition to these, in this paper, I show that 'problem definition' may be a mechanism underlying processes of gradual change inside institutional regimes. To do so, however, it is more useful to conceptualize the mechanism of problem definition from the stand point of the public policy literature, as opposed to the scholarship on 'ideas' and 'discourses' . 2 Building on the seminal work of Rochefort and Cobb (1993), I propose that problem definition is a general mechanism that allows for the emergence of an issue to come to be seen as a problem that may warrant some attention. Taking no stance on the actual 2 as opposed to the approached used by Shpaizman (2014) to account for gradual policy change to israël's immigration policies. existence of problems in reality or about the degree of social construction involved, I propose that the mechanism rests on the actions of political actors in bringing about the problem to the attention of a broad group of actors inside the institutional regime. These actions are strategic and involve the priming of specific characteristics of the phenomena, in relation to the institutional regime, as well as the use of other discursive or cognitive devices to ensure success. As a mechanism, problem definition may lead to various outcomes depending on the context and of the actors involved. Yet, as most of the literature on the topic demonstrates, it is safe to assert that it generally has a structuring effect on policy and politics over time.
Moving away from the policy cycle and agenda-setting, I propose that the mechanism of problem definition is a pathway by which institutions gradually come to internally evolve. Translated to the context of an institutional regime, problem definition becomes one strategic behavior in which actors may engage over time. Because institutions are distributional devices, this strategic behavior will serve to support the interests of actors in effecting a new arrangement of resources and powers inside institutions. This is not to say that this perspective considers a shared conception of the problem as a precondition to the establishment of an institutional regime, as some variants of sociological institutionalism would. On the contrary, in most cases, institutional regimes include actors working together with different ideas about the meaning of an institution and of its objectives, in a state of 'contentious contention' (Mahoney & Thelen, 2010, p. 219).
Casting problem definition as a social mechanism that can potentially to contribute change is different from the current claims about discourse and ideas in institutional theory. The approach is distinct from discursive institutionalism since the dynamics of persuasion are not a central empirical focus of the analysis and because it rejects Schmidt's dichotomy between ideas and material interests (Schmidt, 2010). On the contrary, I argue that problems and, generally ideas, matter only insofar as they have material consequences for actors. Because it is centered on actions and actors, instead of on the characteristics and contents of cognitive elements, this conceptualization of policy problems diverges from the concepts historical institutionalists have used to speak of problems, from Hall's 'paradigm' (1993) to Béland's different iterations of 'ideas' (eg 2007). This perspective is also different, yet inspired by the role of problem definition in the punctuated equilibrium model (Baumgartner & Jones, 2009). Yet, it favors a much more porous and dynamic model of institutions, even in periods of stability and supposes that a problem/policy image may change without generating the sudden disruption of an entire system (eg punctuation).
The comparison of two countries with highlight similar institutional features and convergent immigration federalism is especially amendable to the development of knowledge about problem definition as a social mechanism contributing to gradual institutional change. In practice, doing so implies two analytical tasks. First, the leading conception of the policy problem, if it exists, must be identified in a given empirical setting. This will provide an indication of the operation of the mechanism, as well as a general idea of the potential distributional consequences attached to this problem. Second, the working of the mechanism, as part of the process of change, should be considered. In particular, because problem definition is actor-driven, efforts should be made to identify what group, individual, or institution has engaged in problem definition. This will lead to some information regarding the process of gradual institutional change and its consequences.

The 'problem' of immigration federalism and gradual change
To explore the role of the mechanism of problem definition in bringing about the federalization of immigration in Canada and Australia, I combine multiple sources of qualitative data produced between 1990 and 2010. In both countries, elements considered included: newspaper articles, federal and legislative assembly debates transcripts, other parliamentary literature, policy documents (including program evaluations and enquiries) as well as secondary literature. This allows me to proceed to a content analysis as a way to uncover the way immigration federalism is defined as a policy problem in each national setting. Using this policy problem as a starting point, I then compare the contribution of the mechanism of problem definition to gradual institutional change in both countries, via an event-explaining process-tracing methodology (Beach & Pedersen, 2012).

Immigration federalism in Canada
In Canada, federalization occurred in large part in as a reaction the rise of a new policy problem brought up by subnational governments. As a consequence, the issues attached to institutional change were always related to the needs of provinces, which were presented as being different from the interests of the federal government. The provincial activation of the mechanism of problem definition partially explains the qualitative richness of immigration federalism in Canada.

The problem
Immigration federalism in Canada is rooted in the wicked problem of societal development in the heterogeneous context of the Canadian federation. Indeed, the survival of subnational societies and economies was central to the mobilization of provinces toward immigration in Canada. This survival imperative was and continues to be framed as a matter of fairness, justice, and equality for the federation's constituent units. After Quebec's mobilization, this conception of the problem has been pioneered by one specific province -Manitoba -and has over time been picked up by all subnational governments.
In Manitoba, changes in the economy and fears of population decline have motivated the political elite to consider immigration as a potential resource for the development of the province (Paquet, 2014b). After immigration started internally to be presented as a solution to the province's challenges, actors projected a new definition of the problem in relation to the institutional regime. Central to this new conception of the problem was the importance of immigration for the provincial society and the limited capacity of subnational governments to influence national policy. In the early 1990s, Manitoba's conservative Premier, Gary Filmon presented to his colleagues of the legislative assembly a summary of the general policy problem: […] the member opposite talked earlier in the preamble to our discussions about our population not growing as rapidly as it did in the early '80s, and one of the major differences, of course, is in the proportionate share that we have gotten of Canadian immigrants coming to Manitoba which has dropped to a third of what it was in the early '80s. We are not going to have an opportunity to get a greater share of immigrants. We are not going to have an opportunity to get a greater share of independent immigrants who may have the skills that we need for particular growth and development areas in our economy or entrepreneurial immigrants who come here with business skills and the capital with which to start a business, if we do not have this agreement. With all of the work that has been done to try and influence the federal system, we are absolutely convinced that we need to have that direct influence by way of a federal-provincial agreement. (Filmon, 1993(Filmon, , pp. 2020(Filmon, -2040 Central to this conception of the problem is a diagnostic about the inability -or lack of willof the federal immigration policy to meet demographic and economic needs of provinces, in the context of regional economic disparities. This feeling of being ill served by federal policies and programs reinforced fears about the demographic decline of the province and, in particular, the decline of rural communities.
In addition to the need to respond to the specific challenges of the province via immigration, the policy problem was also presented as one of equality and justice across the federation. Indeed, Manitoba pioneered an unequivocal discourse about unfairness when it came to the distribution of immigrants between Canadian provinces. In 1993, Manitoba's Minister of Culture, Heritage and Citizenship, Bonnie Mitchelson, summarized her government's position for the Assembly: Mr. Speaker what we want to do in Manitoba is to get our fair share of immigrants, and we all know by the numbers that Manitoba's numbers of immigrants have been decreasing and that, proportionately, our fair share should be around 4 percent of those who immigrate to Canada. (Mitchelson, 1993(Mitchelson, , pp. 1355(Mitchelson, -1400 Quickly, this conception of the problem spread across provinces. Beyond Manitoba, it was also maintained after the 1990s and was reinforced by changes in provincial economies, as well as by gradual modifications in the institutional regime. As an example, the strong link between the role of immigrants as a resource for provincial economies and the necessity of a provincial role in immigration was still reiterated by Alberta's government in 2007. In the context of an economic boom associated with the growing demand for oil, the government explained in the Speech from the throne: Despite so many people moving to Alberta each year, our economy is in dire need of people to answer the calls for 'help wanted' across the province. To help meet this demand, your government will focus on better co-ordination of economic development, immigration, and labour force planning. It will craft a made-in-Alberta solution to labour needs. (Kwong, 2007) Similarly, the issue of fairness remained an important part of the policy problem. Over time, it expanded to also include the resources associated to immigrants and not only the actual number of immigrants. In 2005, the Premier of Ontario -the province that attracts most of the country's immigrant -presented his government's own conception of the issue of lack of fairness to the legislative assembly: We understand that immigration strengthens our society and our economy, but yesterday's [federal] budget did nothing to correct the unfairness that sees $3,800 invested in immigrants who land in Montreal and just $800 in those who land in Toronto. Some have argued that we should not be complaining that we are receiving our per capita share of the new immigration money. But that, too, is unfair: When we receive over half of the country's immigrants, it is unfair that we receive only 40% of the new federal immigration funding and 34% of the old federal immigration funding. (McGuinty, 2005(McGuinty, , p. 1350 Thus, the identification of immigration as a vital resource for provincial societies and economies was central to the rise of immigration federalism in Canada. The process of federalization rested on a definition of the problem as being the limited access to this resource for provinces, based on lack of fairness, and poor management by the federal government. This conception implies a very important role for subnational governments, who are in a better position to design and implement interventions that will respond to regional needs. At the same time, discourses about unfairness in the distribution of immigrants and the funding associated to immigrants, indicates the need to rethink the redistribution of resources in the institutional regime.

The mechanism of problem definition in Canada
In Canada, the problem definition mechanism has been activated and maintained in movement by provinces between 1990 and 2010. Over time, the federal government has had to contend with the conception of the issue brought forward by provinces and has modified its interests and behaviors as to maintain partially its dominance in the institutional regime.
The definition of the problem volunteered first by Manitoba had clear implications for the distributions of resources inside the regime. Central to the problem were the issues associated with the incapacity of the federal government to address the specific needs of regions with a national policy. As a result, Manitoba used its conception of the problem to demand an arrangement inside the regime; the province wanted the capacity to directly act in immigration, via the selection of immigrants (Paquet, 2014b). These demands were met with growing and a significant resistance from the federal government, who maintained that provinces, except Québec, should not be active in immigration selection. To compound Manitoba's discontent, the federal government agreed to allow the province to experiment with temporary and highly limited selection schemes. The success of this period of experimentation partially appeased federal fears about the actual capacity of a province such as Manitoba to manage an immigration selection program. In 1998, the Provincial Nominee Program (PNP) was finally established. From this point on, Manitoba became able to select, almost independently, a growing number of newcomers wishing to settle on its territory, and to increase migration flows to meet the needs of its economy.
This new instrument gave a large credence to Manitoba's conception of the problem and soon, other provinces became active in the mechanism of problem definition. This led to important behavioral changes in all provinces, even in provinces that did not have a tradition of attracting immigrants. In the space of about 7 years, all provinces demanded selection powers from the federal government, based on the idea that immigration was a resource for the maintenance of their societies and economies and using a discourse depicting Ottawa as incapable of responding to the needs of individual provinces. Growing discontent with the performance and economic integration of immigrants selected using the federal 'human capital' also contributed credence of this conception of the problem (Reitz, 2005). In 2006, the Premier of Alberta, Ed Stelmach, demonstrated the growing importance of immigration for provinces by stating the following in the province's legislative assembly.
Another important area, Mr. Speaker, is gaining control of the tools to manage immigration policy. It could be as fundamental to Alberta's future prosperity as the affirmation in 1929 of constitutional jurisdiction over natural resources has been to our present prosperity. (Alberta, 2006, p. 72) Problem definition as a mechanism is the pathway by which subnational governments modified their interests and behavior inside the institutional regime. This change challenged the dominance of the federal government as the dominant actor inside the regime and Ottawa did not witness this evolution with passivity. In addition to resisting problem definition at the onset, it then worked to redefine its role in relation to the emerging policy problem. For example, following the publication of the 2001 census, the federal government attempted to prime the skewed distribution of immigrants across the country as a policy problem (Abu-Laban & Garber, 2005), as a clear continuation of the provinces-defined policy problem.
Federal resistance was mostly performed, however, through a reassertion of its steering and redistributive roles in the federation. As early as 2002, the federal government started to investigate the implementation of Provincial Nominee programs via criteria such as program integrity, equality, and regional redistribution (Canada, 2003). These efforts unfolded with a discourse similar to the one put forward by provinces: subnational immigration policy was needed to respond to the diversity and specificity of regional needs. At the same time, federal involvement in the mechanism brought forward the national dimension of the problem. While provinces have increasingly demanded more resources for immigration management, a demand that is hard to discount because of the current conception of the policy issues, the federal government has started to volunteer a new message. In 2012, in a speech to policy-makers and researchers, then Immigration and Citizenship Minister Jason Kenney exemplified the emerging conception of the problem still and now favored by Ottawa: They're [the provinces] just saying, 'Give us more-double, triple-PNP intake. ' And they're really saying, 'Who cares about the Federal Skilled Worker Program?' I'm here to say, and I hope that most of you agree with me, that I believe there is a critically important ongoing role for federal selection of immigrants. Immigration is about nation building. It's not just about addressing regional labour needs. We will not vacate the role of the federal government from selecting economic immigrants. (Canada, 2012) The province-led nature of the problem definition mechanism in Canada has supported, in large part, to a process of federalization from the bottom up. This process created new pathways for subnational influence in the immigration institutional regime, following the redefinition of the provinces' interests. As such, Canada's regime has changed because of the modification of the passive behavior of constituent units, which have succeeded in identifying the consequences in terms of resources of the institutions as a problem. This mechanism created pressures on the historical interests and behaviors of the national government. To regain balance, Ottawa embraced the conception of the problem favored by the provinces while incrementally reinserted a 'federal' dimension to the issues of fairness and equity.
As a consequence, in Canada, the mechanism of problem definition has been instrumental to a slow process of gradual institutional change by which some passively subordinated actors have gained influence as active participants in the institutional regime. Canada's immigration regime is now a regime were both orders of government recognize interdependence, while also continually challenging the current distribution of power and resources, in part via attempts to redefine the problem.

Immigration federalism in Australia
In contrast with Canada, where provinces were successful in bringing about a new conception of the issue, it is the Commonwealth Government that was the central actor in activating the mechanism of problem definition in Australia. This difference highlights the importance of also considering formal institutions as part of an institutional regime -here, in particular the centralized nature of Australian federalism and the central government's jurisdiction over immigration -when accounting for policy problem definition. The central government favored a particular definition of the policy problem: federalism and immigration were to be considered together because of the tensions associated with the skewed distribution of immigration across Australian states. This instance of the operation of the mechanism maintained the federal government in a central position and limited the distributional consequences of the changes to the institutional regime in the short and medium term.

The problem
Regional survival and interstate equality were central to the problem of immigration federalism in Australia. As with most federations, Australia combines natural poles of attraction for international migrants (eg Sydney and Melbourne) with low populated regions (eg Northern Territory). In addition to the economic and social dynamics that attract newcomers, these spaces tend to be popular destination because of the presence of already existing diverse populations. Always an Australian feature, the skewed distribution of immigration has become more apparent after the gradual increase of immigration intakes Australia witnessed starting in the 1990s (Taylor & Bell, 2012).
Faced with states' complaints about lagged economic development and the challenges to maintain their current level of prosperity, the Commonwealth Government started to posit a novel version of the problem. Generally, it associated the dire situation of the states with the uneven presence of immigrants: large receiving poles were suffering from this population's presence while less popular destinations were not accessing the population they needed (AFP, 2012). During a parliamentary enquiry into SsMM, John Raymond Hared, Director of Business and Skilled Migration for South Australia summed up the overall discourse about the immigrant population and states: [South Australia] is a regional low growth, regional low populated area of Australia.
[…] The state is a designated area under the Migration Act for these particular programs. The reason it is designated is because we have such a low incidence of skilled migrants settling here in South Australia. We are currently round about four per cent of the migration intake compared to eight per cent of our population of Australia. (Australia, 2000a, p. 404) As part of this conception of the problem, the Commonwealth Government insisted in the importance of a proactive role for the states. Canberra itself was in no position to force immigrants to settle in non-traditional regions (Australia, 2000c). Moreover, only subnational governments could secure the collaboration of local stakeholders -employers or local administrations -in implementing interventions tackling the problem.
This version of the problem became visible in the mid-1990s, when the central government started to make efforts to orient skilled and business migrants to less populated states as part of a broader reform that increased the intake of skilled migrants in Australia (Wright, 2015). After some policy advances, however, the problem became defined not only as one of distribution of immigrants but also as resulting from the passive role of the states. In 1998, Australia's Immigration Minister Philip Ruddock stated in an interview that subnational governments were partially at fault for not reaping the benefits of the country's immigration increase: If you are serious about getting skilled migration, there are opportunities for the State and Territories who are keen to be involved. […] If the States cannot pick up those opportunities, it seems to me it's far more difficult to argue that there should be an overall increase in skilled migration. (Carruthers, 1998) Similar discourses about the problem were maintained and reinforced over time, especially as changes in the institutional regime started to unfold. Discussing Tasmania's limited use of the SSMMs, Ruddock stated that it was up to the subnational government to make the changes necessary to address the issues associated with its limited capacity to attract immigrants: For Tasmania, the job is for the Tasmanian Government to identify the [employment] area [for sponsorship] […] Up to date, they haven't used [this scheme], and all I am saying is that if they are serious about regional migration, there are mechanisms they are able to use. (Warner, 1999) At the same time, the federal government retained the responsibility to fix issues related to labor shortages and overall economic management of the federation as its responsibility (Australia, 1999). As such, the policy problem associated with immigration federalism in Australia is one of incorrect distribution of immigrants across subnational units, partially caused by the passivity of state governments. This conception of the problem implies a role for both orders of government in the management of immigration, with specific tasks that may only be performed effectively by state and territory governments.

The mechanism of problem definition
In Australia, while the policy problem is directly presented as being generated by states, the federal government has propelled the mechanism of problem definition. Indeed, the Commonwealth Government has started the process of change by creating a new institutional venue rooted in its conception of the problem. In 1996, the National Department of Immigration and Migration created a working party on regional migration. This stable intergovernmental forum gathered all levels of government to discuss and collaborate over the issue of immigration distribution across regions (Australia, 2007, p. 43;Hugo, 2008). In the following years, the Commonwealth government has maintained control over such coordination mechanisms. It also mandated the creation of region-specific working parties on immigration and conducted several studies about the issue of regional migration.
These institutions served as the venue for the mechanism of problem definition. In linking immigration with regional disparities in population and their consequences, the central government shielded itself from potential challenges from states. Instead, commonwealth control over the policy problem provided opportunities for a rearrangement of the institutional regime according to the central government's interest. The central change stemming from the new conception of the problem of immigration federalism has been the creation of SSMMs, presented as the main solution to state's population issues. By providing subnational governments with a tool to address their needs, the Australian Government effectively reiterates the division of responsibilities inherent to the problem definition it favored.
After 1996, SSMMs have grown: more visa categories have been created and immigrant intake has soared in some cases. Nonetheless, the process of change was slow and fraught with intergovernmental tensions. Illustrative of these were states' interventions in the 2000-2001 Australian's parliamentary inquiries about SsMM. During the recorded meeting with regional representatives, parliamentarians discussed several issues raised by states' submission to the commission. About 5 years after the implementation of the SSMMs, the problems created by the costs for state government were highlighted (Australia, 2000c). In some cases, state representatives resisted any suggestion to dedicate more than very minimal human or capital resources for the operation of the program (Australia, 2000c). Several states presented themselves as poorly equipped to deal with immigration, while Commonwealth representatives still maintained that it was their responsibility to engage in meaningful action regarding immigration. Discussing the limited information available to newcomers in Tasmania, a state parliamentarian summed up the central conception of the problem: The [Tasmanian] department gives information, but whether the information that is given is of value is an assessment for the other side. It is a bit bureaucratic just to say, 'Here is your information. ' Maybe the department is living in a different era as to how it operates as opposed to the new business-type migration concept, or the skilled migration concept […]. Seriously, it is not something that we are set up to do at the moment under our existing policy parameters. We are not resourced to do it. It falls outside of the existing procedural parameters that the department operates under.
[…] within our existing procedural parameters it is not something that we are resourced, required, or expected to do. And it is not something that we can do, given the sorts of priorities we have on our existing resource base. (Australia, 2000b, p. M344) Subnational resistance as part of the process of federalization is linked to the mechanism of problem definition in Australia. The country's immigration regime has come to change following the modification of the behaviors and interests of the actors already in a dominant position inside the regime: the Commonwealth Government. This 'federalization from the top' created new spaces for subnational actors in the regime: by identifying them as responsible for attracting newcomers and by supporting the creation of new tools allowing them to do so.
The definition of the policy problem and the fact that it was federally conceived, however, sustained a regime in which immigration remained first and foremost the responsibility of the central government. This has limited the redefinition of states' interests inside the regime and, as a consequence, has substantially slowed the growth of SsMMs and of other programs. Problem definition as also generated forms of political resistance to federal downloading in matters of immigration. As a consequence, in Australia, the mechanism of problem definition has been instrumental to a slow process of gradual institutional change by which some dominant actor has coerced others to become more active participants in the institutional regime.

Conclusion
Since the 1990s, Canada and Australia have seen their immigration institutional regimes included immigration federalism as a characteristic. In both countries, contrary to other federations, this has given rise to pro-immigrant policies. Despite the maintenance of their national approaches to immigration, this evolution represents an instance of considerable gradual institutional change. The historical dominance of the federal government in managing immigration has now been replaced by the agonistic interdependence of two orders of government. Since institutional regimes are relatively dynamic distributional arrangements between stakeholders, this process of change must be taken into account when considering the recent and future evolution of immigration governance in Canada and Australia. In the absence of exogenous forces leading to change, federalization can be explained by the operation of a mechanism of problem definition in Canada and in Australia. Based on recent advances in institutional theory, I proposed that problem definition might be one pattern of mobilization for actors inside a broad institutional regime. The mechanism of problem definition is actor-driven and this article has demonstrated that its influence is material, and not only ideational. Inside a given institutional regime, problem definition may be the source of the redefinition of actors' interests and behaviors toward the balance of power as well as the distribution of resource inherent to a given institution. Based on the comparison of these most comparable cases, problem definition as a non-deterministic social mechanism should be considered as contributing to institutional processes of gradual change. Using this perspective makes it possible to consider what triggers institutional change and how the process unfolds, something that is missed by solely considering formal institutions, jurisdictions or by comparing degrees of centralization.
The convergence toward immigration federalism is nonetheless the result of different definitions of immigration federalism as a policy problem. This difference, and the ensuing pathways of change in each national immigration regime, is largely the fact of the actor central to the activity of the mechanism of problem definition: provincial governments in the case of Canada and the Commonwealth Government in the case of Australia. Once put in motion, these processes of change gave rise to country-specific policy outputs, such as different subnational immigration selection programs that might, in the future, generate divergent policy outcomes. It remains to be seen whether the federalization of immigration governance in Canada and Australia will actually address the issues raised by the governments regarding immigrants' distribution and cross-country equity. Future work should consider systematically how institutional actors' stakeholders contribute to activate and maintain in motion the mechanism of problem definition. In this particular case, it remains an empirical question whether the mechanisms of problem definition also served as a mediator to employers' and economic actors' interests toward changes in economic migration regulations and programs. In addition, the explanatory contribution of the mechanism of problem definition should be considered where immigration federalism resulted in exclusive and restrictive immigration policies, such as the United States.
Identifying problem definition as a social mechanism creates links between public policy theories about policy problems and institutional analyses. In systematizing the link between institutional regimes and problems as having distributional consequences, it permits a more empirically grounded account of problems, beyond 'discourses' and 'ideas' . By focusing on the dynamics of problem definition and their impact on the content of the problem, as opposed to qualifying the problem, it contributes to current scholarship on the recognition of problems (Peters, 2005). In dialogue with the literature on social mechanisms, future research should aim to identify regularities in the operation of problem definition as a social mechanism, with the objective to build theory from the ground up.

Disclosure statement
No potential conflict of interest was reported by the author.