This research is supported by the Social Sciences and Humanities Research Council (SSHRC) of Canada grant number 435-2022-0082. We would like to thank Jacob Buchan for his valuable research assistance.

Introduction

2To mitigate against the negative effects of automation, IOs and GMCFs have advised policymakers to shift education and training across the workforce (Deloitte, 2019; ILO, 2018; KPMG International, 2018; McKinsey, 2017; OECD, 2018; WEF, 2016), and to adopt “a human-centred agenda for the future of work” (ILO, 2019, p. 11). Governments have also been advised to ensure workers can adapt through publicly provided skills programs that emphasize what have been labeled as “future skills” or “human centred skills” such as creativity, collaboration, and problem solving, along with digital/technical skills (OECD, 2019), robust and effective technical and vocational training (WEF, 2017), and lifelong learning programs (ILO, 2019; WEF, 2017).

3This article aims to answer the question of whether and to what extent domestic policy makers utilize the advice and knowledge of IOs and GMCFs in this crucial area of economic and labor market policy. That is, are IOs’ and GMCFs’ ideas regarded as authoritative and do they inform domestic policy making? Literature in International Relations and comparative public policy on policy diffusion and learning points to the role of IOs and GMCFs as key sources of information and authority in transnational and international policy and governance (Acharya, 2004; Barnett & Finnemore, 2004; Cortell & Davis, 1996, 2000; Finnemore, 1993; Risse, 2016; Risse & Sikkink, 1999; Risse-Kappen, 1994) as well as policy diffusion outside of government-to-government transfer/learning (Dolowitz & Marsh, 2000). IOs in particular, have been found to play an important role in influencing domestic policymaking in a range of sectors and especially in countries in the Global South through mechanisms of coercion, competition, social construction, and learning (Dobbin et al., 2007).

4Less research exists, however, on whether and how IOs and GMCFs fit into domestic policy advisory systems in the Global North (Craft & Howlett, 2012; Fraussen & Halpin, 2017), where coercion mechanisms are largely absent (Linos, 2013; Simmons, 2009; White, 2017). This study thus contributes to the literature that examines international-domestic policy transfer (Acharya, 2004; Blatter et al., 2022; Cortell & Davis, 1996, 2000; Dobbin et al., 2007; Linos, 2013; Simmons, 2009; Simmons et al., 2008), as well as the literature on domestic policy learning and policy uptake of IO and GMCF ideas (Orenstein, 2008; Risse-Kappen, 1994, 2016; Weyland, 2007). As scholars such as Jacobs (2014) and Campbell (2002) note, to examine the effect of ideas on domestic policy decision making, researchers need to track not just from where ideas originate – that is, from what sets of actors and sources of information – but also how they travel – that is, who borrows ideas and whether and how domestic policy actors utilize this policy advice in domestic policy making.

5We examine the case of Canada’s “future skills” policy and the role of IOs and GMCFs as sources of policy advice. We pose the following questions:

1) do policy makers read the reports produced by these organizations and take up their ideas in domestic decision-making processes;

2) how authoritative or influential is work produced by these varied international actors among domestic policy actors;

3) are domestic policy makers more likely to disproportionately learn from IOs and GMCFs or do policy makers utilize them as just one of many sources of knowledge;

4) are some IOs or GMCFs more influential than others; and

5) does the organizational role of the domestic policy maker matter in terms of how closely they utilize IO or GMCF knowledge.

6In order to examine the relationship between IO and GMCF ideas and influence on domestic policy making, we employed qualitative methods of elite interviews and government document analysis. We conducted a citation analysis of the Canadian Federal Department of Finance’s Economic Advisory Council’s reports to examine the relative influence of IO and GMCF reports on the Advisory Council’s work. We also conducted an inductively generated thematic analysis of 26 interviews with members of the Canadian future skills policy community (see Appendix 1) to examine the sources of policy ideas interview participants reported using, and the perceived authoritativeness and agenda setting role of those sources (see Appendix 2 for the list of interview questions).

7 We find that the relative authority of IOs and GMCFs as policy advisors varied depending on the articulated goals of the sector and the forms of learning in evidence among domestic policy actors. We find that a majority of interview participants reported reading IO and GMCF reports and that those reports informed policy making. We find mixed evidence of the agenda-setting power of those reports, however, and GMCF reports, in particular. Interview evidence confirms that the federal government directly sought the advice of one GMCF in particular – McKinsey – in formulating a response to anticipated labor market disruptions. Interview participants reported that McKinsey was actively involved in the policy process and in informing the federal Department of Finance’s Economic Advisory Council’s work in ways that reveal its influence in framing the agenda. However, we also find evidence of more complex learning processes. Interview participants noted the utility of IO and GMCF reports but argued that policy decision making reflected wider sources of evidence and more comprehensively rational decision-making processes that focused on “what works” in the sector. In this case, IOs and GMCFs’ reports were useful as policy blueprints in the face of uncertainty (Blyth, 2002), but others were circumspect about the utility of their reports and their relevance to the Canadian policy making.

8The next section reviews the extant literature on the role of IOs and GMCFs as sources of policy advice as a first step in establishing them as contributors to domestic policy advisory systems. It then outlines our theoretical expectations of how those ideas may be utilized by domestic policy actors before presenting the findings from our study.

IOs’ and GMCFs’ Ideational Authority and Influence

9The International Relations literature documents the influence of IOs and GMCFs as key sources of policy advice globally, including in the area of education and skills (Jackobi, 2009). In the case of IOs, research tracks their role as knowledge hubs that provide and supply information and expertise on policy problems and solutions (Hass, 1992). IOs such as UNESCO act as “teachers” that guide states to develop bureaucracies and comply with new international norms (Finnemore, 1993). They can also provide information that advocates can turn to while advocating for particular policies domestically (White, 2017). They construct and spread ideas in a number of sectors, including the economy (Mahon & McBride, 2009), inclusive growth and social investment (Mahon, 2019), and even public management reforms (Pal, 2012).

10Transnational policy actors such as international think tanks also promote the “globalization of ideas,” where they play a major role in the diffusion of ideas by actively engaging in media and professional networks (Stone, 2021). During the Eurozone crisis, economists from various EU member states were brought together by Brussels think-tanks to produce and circulate technical knowledge and solutions to solve the Euro crisis (Coman, 2019). These transnational actors produced solutions that diffused across member states and linked together transnational, national, and local level actors (Coman, 2019).

11IOs can also be regarded as monitoring agents where they act as peer reviewers of their member states’ activities. Peer pressure is a method of soft persuasion that can encourage a government to adhere to international standards (Pagani, 2002). Peer review can be performed through a series of performance indicators such as the OECD’s performance rankings in the economy, education, and social policy (Pagani, 2002; McBride, McNutt & Williams, 2008; Niemann & Martens, 2018). Generating ideas through ranking and ratings can serve as a useful tool in encouraging domestic policy makers to adhere to international rules and standards, such as human rights (Goodman & Jinks, 2013; Simmons, 2009).

12IOs also influence domestic policymaking through “best practices.” “Best practices” have been a recurring model of global governance that aims to move domestic actors towards certain goals by recommending procedures and offering incentives (material and non-material to encourage compliance) (Bernstein & van der Ven, 2017). This “best practice” governance could encourage policymakers to adopt certain goals or standards to achieve legitimacy in their respective community.

13The domestic influence of international norms and ideas can be contingent on several factors, however. Cortell and Davis (1996) argue that domestic policy actors adopt international norms and rules because domestic actors are receptive to accepting new norms. Receptivity is contingent on domestic policy makers’ calculations that these international norms can further their own interests in their community or gain legitimacy and reputation (Cortell & Davis, 1996). In addition, the domestic spread of international norms also depends on the existence of domestic and transnational networks that can influence other states (Risse & Sikkink, 1999). Power also matters. In the case of IOs such as the IMF and World Bank, countries, especially in the Global South, may feel coerced into adopting stringent rules in exchange for financing (Simmons et al., 2008). Finally, states may respond to IO advice for reasons of economic competition.

14A growing body of research reveals that GMCFs are similarly influential as ideas generators and domestic policy influencers. Indeed, a growing popular literature documents – and critiques – the use of private or external consultancies (Howlett & Migone, 2013; Hurl & Vogelpohl, 2021; Marciano, 2022, 2023; Momani, 2013, 2017; Stone et al., 2021) and the perceived outsized influence of those organizations in domestic policy formulation and government decision making (Bogdanich & Forsythe, 2022; Mazzucato & Collington, 2023). That influence is deliberately cultivated by the consulting firms themselves. As Kipping (2021, p. 37) points out, while consulting firms operate on a for-profit basis through contracts with specific clients, they also have their own think tanks – such as the McKinsey Global Institute – which produce their own reports that are free and widely available. These reports, Kipping (2021, p. 37) argues, “increasingly touch upon issues of relevance to policymakers in a wide range of domains.” And GMCFs are motivated to produce these knowledge products: 1) because it makes good business sense; 2) it confirms and extends their authority; and 3) it contributes to their hegemonic influence (Kipping, 2021, pp. 50-51).

15A growing empirical literature tracks consulting firms’ growing role in a number of countries including Canada over the past several decades (van den Berg et al., 2019) for several reasons. Consultant firms can provide innovative and creative ideas that differ from the public sector (Momani, 2013; Momani, 2017). For example, consulting firms often provide expertise to clients that is unavailable elsewhere, because of their advantage in accessing data and big data (Momani, 2017). Consulting firms also can “challenge governance regimes, disrupt/create jurisdictions, and transform identities, practices and systems of regulation in the professions themselves.” (Faulconbridge & Muzio, 2017, p. 220). In synthesizing the literature on the diffusion of consultant firms’ ideas, and examining the case of McKinsey and Co. specifically, O’Mahoney & Stury (2016) categorize the power of consultancy firms in spreading ideas in three categories: 1) power of resources; where firms have financial and network resources to increase their influence in diffusing ideas; 2) power of process; where firms can influence decision-making process; and 3) power of meaning; where firms can directly shape and formulate solutions to policymakers.

16In the area of future skills, the global arms of management consulting firms have produced dozens of reports on the topic. GMCFs, like IOs, can thus be seen as high-capacity policy advisors in general and in this area specifically. However, few studies have traced the impact of consultancies on domestic policy making and outcomes in the area of education, training, and future skills, the goal of this project.

17Given these findings (summarized in table 1 below) from the International Relations and Comparative Public Policy literature on the authority and agenda setting influence of IOs and GMCFs, we derive the following expectations regarding the scope and extent of domestic policy uptake of these ideas. First, we expect that domestic policymakers will look to IOs and GMCFs as sources of (authoritative) knowledge and information; that is, policymakers will read products produced by these organizations and those reports will provide sources of information for the sector. While some domestic policy actors may regard policy ideas from international and transnational organizations in general as irrelevant or illegitimate (Linos, 2013) or be distrustful of specific IOs and GMCFs, ample empirical research demonstrates that domestic policy actors do take up their ideas, even if they are selective in the sources of information they find relevant. We expect the same to be true in the future skills policy sector (Question 1), although we expect that some IOs and GMCFs will be more influential than others based on their perceived authority in the policy area – such as the number and quality of the reports they have written – and their perceived dominance in the sector (Kipping 2021) (Question 2).

Table : Summary of IO and GMCF Influences on Domestic Policymaking

Type of influences	Means of influences
Teachers/knowledge hubs	Supplying information and expertise on policy problems/solutions Producing technical knowledge on policy issues Constructing innovative ideas and norms Guiding states to comply with new norms/ideas
Monitoring agents	Peer pressure Peer review through performance ranking and rating
Best practices	Creating models of governance Developing rules/standards Offering incentives

Source: the Authors

Factors that Affect Domestic Policy Uptake of International Ideas

18In contrast to literature that focuses on the perceived authoritativeness of IOs and GMCFs that would lead us to expect an automatic or disproportionate uptake of their ideas domestically, other literature focuses more on the complex nature of domestic policy agendas and learning processes that suggest a range of pathways that can be observed in ideational diffusion and domestic learning processes (Blatter et al., 2022; Dobbin et al., 2007; Dolowitz & Marsh, 2000; Linos, 2013; Orenstein, 2008; Weyland, 2007).

19We expect, given the high perceived authoritativeness of IOs and GMCFs in this sector, that the agenda setting influence of IOs and GMCFs will be high. However, the policy impact (that is, the utilization of the policy advice) will vary depending on the recipients of the policy advice – that is, the organizations and the policy sectors to which IO and GMCF advice is directed – as well as the articulated goals and the learning processes established within and between the organizations receiving the advice (Questions 3 and 4). We define agenda setting, as Zahariadis (2016, p. 7) does, as comprising four elements of that power: to persuade; to highlight what issues are perceived as important; to reveal the potency of issues, for example, the severity of consequences for non-action; and the importance of the issue in terms of its proximity to people’s lives. We define learning as “the updating of beliefs based on lived or witnessed experiences, analysis or social interaction” (Dunlop & Radaelli, 2013, p. 599).

20While many accounts of actors’ learning conceive of it as a comprehensively rational process of Bayesian updating “in which individual actors add new information to prior knowledge and beliefs and revise their behavior accordingly,” Blatter et al. (2022, p. 808) point to the importance of intersubjectively held (causal) or scientific beliefs and values that inform policy (Blyth, 2013; Haas, 1992) and that can become paradigmatic (Hall, 1993) and limit the search for new information. Furthermore, as Dunlop and Radaelli (2013) argue, the nature of a policy challenge – such as whether there is a high degree of certainty or uncertainty – affects the extent to which policy makers rely on particular forms of policy advice.

21Given the high level of uncertainty around the scope and depth of anticipated market disruptions in the fourth industrial revolution, we would expect to find some evidence of selective learning or selective uptake of information by domestic policy actors – using more boundedly rational learning processes – that selectively imitated policies from elsewhere (Jones & Baumgartner, 2005; Kahneman, 2011; Weyland, 2007), or that relied on shortcuts such as reputationally authoritative sources such as IO and GMCF reports. We also expect that domestic policy makers, faced with a policy environment that is characterized by a high degree of technical complexity but also a high degree of knowledge about “what works” generally, would tend to rely on expert authority (including governments, IOs, academic economists, and the like) and task them with providing advice to provide technical policy solutions (Dunlop and Radaelli, 2013, 603). In the area of future skills, given the rather technical nature of the sector as well as the wide range of economically oriented actors and organizations involved, we thus expect selective uptake of the advice of authoritative actors such as IOs and GMCFs.

22In contrast, though, Blatter et al. (2022) examine more values-driven processes and diffusion frameworks that assign principled beliefs a crucial role in underpinning the exchange of ideas and goals and may rely more on information derived from a range of policy experts grounded in more diverse forms of authority or knowledge of best practices, or popular attention to an issue. While the policy arena may still be grounded in technical knowledge and epistemic learning, it may in fact entail the gathering of a wide range of information and evidence from a variety of sources, and a weighing and assessment of the quality of evidence in the process of decision making, that yields a more comprehensively rational decision process overall.

23Similarly, an interest-driven diffusion process, for example, would place “the exchange of information among state governments at the heart of the diffusion process … triggered by some external problem or pressure where governments in one jurisdiction will adjust policies in reaction to others (Blatter et al., 2022, p. 816). We would expect, then, that when policymakers’ policy goals are based on strategic assessments of comparative economic competitiveness, they will be less likely to rely on external expertise such as that given by IOs and GMCFs.

24Finally, building on research by Koga et al. (2023) about the relationship between bureaucrats’ organizational role and the sources of evidence they use, we expect that the authority and influence attributed to these reports will vary depending on the domestic policy actors’ roles in an organization and the type of organization (Question 5). That is, the composition of the policy community – and the sources of information to which it attends – can affect the kind of learning undertaken. We expect that where actors are situated in the policy community and their organization affects their form of learning. For example, members of think tanks are more likely to be influenced by other think tank reports. Those at the head of an organization, given their limited attention and competing policy agendas (Jones & Baumgartner 2005), may be more likely to engage in selective uptake of reports that are perceived as more authoritative. We empirically examine these expectations below.

Study Design and Methods

25This study seeks to examine whether and to what extent we could find evidence of IO and GMCF authority and agenda setting power in Canadian policy making in the area of future skills. We focus on the case of Canada because we observed some initial evidence that the federal government was very attentive to developing a “future skills” agenda, most visibly in the establishment of a Future Skills Lab (2017a) announced in the 2017 federal budget. The Canadian federal government committed $225 million starting in 2018-2019 for four years, and $75 million per year thereafter “to establish a new organization to support skills development and measurement in Canada” focused on providing information on the future of work, bridging employers and workers on specific needed skills, and researching new approaches to education and skills training (Department of Finance Canada, 2017, p. 57).

26That policy initiative was traced to a recommendation made by the 14-member Advisory Council on Economic Growth. That Advisory Council, announced in March 2016, was one of the major influential policymaking bodies created by the federal Department of Finance under then-Finance Minister Bill Morneau (Department of Finance Canada, 2016). The Council was chaired by Dominic Barton who at the time was the Global Managing Director of McKinsey and Company, and consisted of members from both public and private sectors. It was explicitly set up to help advise the government of Canada on policies to support economic growth. One of the 12 reports the Advisory Council on Economic Growth (2017a) issued explicitly recommended the establishment of a Future Skills Lab, which the federal government implemented in the 2017 federal budget.

27Given that the Future Skills Centre represented the federal government’s largest effort to inform policy in this area, we were interested in examining what sources of information informed policy making on this initiative and “future skills” more broadly. Drawing on Saleem et al. (2023), we first conducted a citation analysis of the Advisory Council’s reports to examine the relative influence of IO and GMCF reports on the Advisory Council’s work. This method has been used by researchers to examine the policy impact of expert groups in other policy areas and country contexts (e.g. Christensen, 2023; Christensen & Hesstvedt, 2024). An analysis of the 12 reports produced by the Advisory Council reveals that the Council’s work was highly influenced by IOs and GMCFs, especially McKinsey. Table 1 reports the number of citations by different organizations in the 12 Advisory Council’s reports. We collected 571 citations across the 12 reports produced by the Council. Our results show that 98 citations (17%) are attributed to IOs and GMCFs. IOs make up about 9% of the total citations whereas the sources from GMCFs make up about 5% of the total citations. Among IO sources, OECD sources are the most frequent citations in the advisory reports. Most importantly, 25 citations are directly attributed to McKinsey, followed by only two citations attributed to PwC. The context of these citations is similar for both IO and GMCF sources. The majority of the citations attributed to IOs and GMCFs are used for statistical information, background information and other general information. The Advisory Council cited these organizations to showcase the strength (or lack thereof) of Canadian economic performance compared to other economies. It is also worth mentioning that statistical analyses provided in the reports used a combination of data sources from IOs, GMCFs, and local policy think tanks. One report featured a statistical analysis performed by the Advisory Council’s members. We also found two reports that featured a statistical analysis performed by the McKinsey team.

Table : IOs and GMCFs Citations in the Federal Advisory Council Reports

Sources	Number of Citations	Percentage of Citations
OECD	31	5.4%
WEF	6	1%
United Nations	6	1%
World Bank	6	1%
IMF	2	0.3%
IOs total sources	51	9%
McKinsey	25	4.4%
PwC	2	0.35%
GMCFs total sources	27	4.7%

Source: the Authors

28We also conducted 26 semi-structured interviews among key members within the Canadian future skills policy community to understand more about the processes of decision formation, actors’ beliefs, and key sources of information for this policy area (see Appendix 2 for the list of interview questions). Our sample of participants consisted of key members within the future skills policy community (Skogstad, 2008) from public and private sectors across Canada. We considered the policy community to include the universe of public and private actors – government agencies, domestic interest groups and think tanks, media, and individuals, including academics – who have an interest in the future skills field and attempt to influence it. We classified the participants’ employment background into seven categories to reflect the range of organizational perspectives: 1) policy institute/think tank; 2) research and evaluation agency/institute; 3) government; 4) quasi-government; 5) academia; 6) non-profit organization/foundation; and 7) organizational/individual consultancy (see Appendix 1). We also classified participants’ roles (e.g. researcher/analyst; manager or head of organization). Among those members in the future skills policy community, 13 of our interview participants have been directly involved in the policymaking process related to future skills. In addition, 10 participants have provided advice to the government. Nevertheless, all of our interviewees have been involved in policymaking, either through direct participation or through research and advocacy.

29It is challenging to delineate boundaries between various organizations, as some organizations have multiple goals and responsibilities; as such, we coded the organizations based on their primary mission. We differentiated between research agencies and policy institutes/think tanks by whether their primary goal is research or research plus advocacy, while acknowledging that some researchers (e.g. Stone, 2017, 149; see also Stone, 2021) argue that not all think tanks engage in advocacy. Furthermore, we classified organizations as consultancies if their primary role was providing research and advice for public or private sector clients under contract (Momani, 2013) even while recognizing that think tanks may also play a similar contractual role.

30We began our recruitment of members of the policy community by first conducting a single in-depth interview with a key informant who had deep experience in the future skills policy area. That interview generated an initial list of 24 potential participants across Canada. These potential participants were largely Ontario provincial and some federal civil servants, representatives of interest groups and think tanks, and academics. Of those, 13 were identified to contact in the first round based on our key informant’s assessment of their willingness to participate in an interview, and their knowledge and expertise. We augmented that initial list by conducting a search on LinkedIn Professional to triangulate that information with employment ties with organizations identified as part of the future skills policy community.

31After we began the interview process, we utilized snowball sampling to recruit further participants. At the end of each interview, we asked participants to list others whom they thought would be useful to interview and whose names might not be obvious. These techniques generated a list of 88 names. Of those, we contacted 52 participants in total via email and received positive responses from 26 participants.

32All interviews were conducted via Zoom and recorded. All participants were asked at the beginning of the interview whether they wished their comments to be made with attribution. We asked all participants a structured sequence of questions that focused on our key topics, but we conducted the interview in a semi-structured manner to allow for more conversation and participant freedom in structuring responses, to highlight what they felt were important themes, drawing on their own understandings of the topics (Aberbach & Rockman, 2002; Herzog, Handke & Hitters, 2019).

33We conducted a thematic analysis of the interview transcripts using NVivo to identify patterns and find common themes among our responses (Braun & Clarke, 2006; Herzeg, Handke & Hitters, 2019). Using manually coded thematic analysis in NVivo allowed us to inductively generate the themes uncovered from the interviews, rather than explicitly “testing” for “alternative explanations” generated in advance of our interviews. However, our thematic coding was broadly informed by the theoretical expectations we derived from our review of the policy diffusion and learning literature.

34In the thematic analysis, we first coded the interview participants’ perceptions of the core goals underpinning Canada’s future skills policy making. Those that emerged inductively from the interviews were: 1) strategic goals of economic competitiveness; and 2) other non-material benefits/values. Interview participants’ goals were coded as characterizing economic competitiveness if they referred to concerns about Canada’s position in the global economy or how Canada is economically behind or catching up with other countries. Goals were coded as reflecting other values if interview participants used language such as future skill policy making is something that Canada “ought” to do, for example, to create opportunities for children and workers or as a way of reducing inequality.

35Second, we classified the different types of information sources identified. We noted whether the sources were domestic (Canada) or international. The domestic sources of information included domestic think tanks, academic research, data repositories, government related sources, and other experts. The international sources included think tanks, IOs, GMCFs and other governments. Each mention of an information source in an interview was coded as 1; if it was not mentioned, it was coded as 0.

36Third, we coded the data based on participants’ statements about the perceived influence and authority of IOs and GMCFs, on a scale of low, moderate and high. A low level of influence meant the participant did not view IOs or GMCFs as useful. A moderate level of influence meant the participant found the reports useful but held some skepticism or reservations about a report’s quality or utility. A high level of influence meant that participants expressed the view that IOs’ or GMCFs’ work was perceived to be, for example, important, methodologically sound, or influential.

37To probe the authoritativeness and agenda setting power of the IOs and GMCFs, we further categorized the participants’ reported perceived influence based on whether they stated the reports were useful sources of information in and of themselves; and separately whether they were influential in setting or shaping the domestic policy agenda. We coded as agenda-setting any references by the participants to the importance of the reports in: 1) framing the topic; or 2) catalyzing the discussion; or 3) helping to see the topic in a different light based on new information.

38Lastly, our coding strategy focused on the participants’ assessment of both perceptions of the federal government’s learning and our assessment of the participants’ own reported learning process. We coded the participants’ perceptions of government learning processes as comprehensively rational if they mentioned that government decision making was perceived to draw from a variety of sources of information to inform policy making, or if the government engaged in trial-and-error assessments of what works, or rigorous review of their own data. We coded the participants’ perceptions of government learning processes as selective if they reported that they perceived government to favor one or few sources of information disproportionately in policymaking.

39Likewise, in our assessment of participants’ own learning processes, to be coded as a comprehensive rational learner we examined the number and variety of sources of information they mentioned as informing their understanding of the sector. We coded the participant as a selectively rational actor if they reported to only draw from a single source of information or disproportionately favored one or few sources over other types of information sources or perceived the authority of the source based on reputation rather than rigorous review and did not engage in reviewing the source information themselves. We coded the participant as a comprehensive learner if they reported drawing on a wide range of evidence. We separately coded any observations they made about the quality of the sources reported. We also coded the participants’ assessment of the credibility or reputational authority of the sources.

Findings

IOs and GMCFs as Sources of Information

40Analysis of the 26 semi-structured interviews captures key dynamics of ideational diffusion and domestic policy uptake in our case. We find evidence to support our expectation that members of the future skills policy community perceive IOs and GMCFs as key sources of knowledge and information. A majority of the participants stated that they read and were familiar with knowledge products produced by these organizations and to a varying extent had read them.

41Our findings reveal two ways that members of the policy community became familiar with IO and GMCF reports. First, they turned to IOs and GMCFs for background information that served as foundational knowledge about the topic of future skills. Second, they turned to IOs and GMCFs sources for comparative analysis and learning from government practices in other jurisdictions. Four participants additionally reported instances where policymakers looked to IOs reports and paid attention to rankings and indexes to determine where Canada stands against other jurisdictions in terms of skills-related fields such as education. One participant noted that the OECD reports in particular are important, as “everyone” likes the rankings in determining how Canada is doing relative to other countries. We present the number of interviewees who have read the different types of sources in Table 3.

42Of the IOs, the OECD was mentioned most frequently as a source of information: 15 out of 26 participants were aware of and had read OECD reports related to future skills. Those participants noted that the OECD often produces reports and other knowledge products that are useful for understanding the policy context. Beyond the OECD, the ILO, WEF, World Bank, and UNESCO were mentioned by 13 participants. However, there were fewer references to these sources. Three participants mentioned that they regarded the WEF as one of the main international forums that produces original findings or information regarding future skills. One participant noted that “the WEF pioneered the term industry 4.0”; another participant stated that the WEF’s report on the future of work is widely cited and was used as background information to inform the policy conversation in Canada. For those with a policy interest in education, the UNESCO reports were regarded as relevant, including their reports on lifelong learning and micro-credentialing programs.

43A majority of interviewees (21) mentioned they had read and were familiar with GMCF reports but their perceived relevance was much lower than IOs reports where 25 of 26 reported reading IOs reports. McKinsey was the most mentioned GMCF source. Ten participants reported that McKinsey was the most influential policy consulting firm because of its perceived expertise and innovative research. One participant noted they perceived the McKinsey reports as more methodologically sound than the other “big four” consultancies (Deloitte, Ernst and Young, KPMG, and PwC). However, another noted McKinsey reports were widely used as sources because of the firm’s reputation and branding rather than their quality. They noted that “McKinsey reports are influential because McKinsey wrote it, not because they are good.” Another participant noted that GMCFs are proficient in leveraging public anxiety about certain issues and therefore their reports are often highly regarded as influential.

44While GMCFs might not be involved in creating public anxiety about the effect of automation, their reports often provide information in forming solutions to tackle policy issues and that directly target their audiences. One participant stated “GMCFs are good at taking the polls and understanding the anxiety policy makers are feeling and then they can effectively position themselves to be solutions for those anxieties.”

45Beyond these two types of sources, however, our interviewees also mentioned a range of other sources. Academic sources were one of the most popular sources of information: 21 participants mentioned they were aware of and had read or used academic resources in their work and many described that they paid greater attention to academic sources relevant to the Canadian context. Six participants mentioned using academic sources as the major sources of information; one participant noted that they only use academic sources to dig deeper into the issue beyond the grey literature. The participants who engaged in policy research mentioned that academic sources were quite respected in their organization and were used to support their research formulation and policy advocacy. However, two interview participants emphasized that academic sources had zero impact in policy making tables.

46As evidence of more comprehensive learning, 15 participants stated that they drew from data repositories as their main source of knowledge to produce reports for their organization. Interview participants noted that knowledge products generated by domestic think tanks were also useful; 14 participants stated domestic think tanks reports contributed to greater understanding and shed light on the Canadian context. In contrast, only eight interview participants mentioned that they look to international think tanks such as the Brookings Institution.

47In reference to government reports or knowledge products, only three participants directly mentioned using government related documents as a source of information. However, participants did mention using government ad-hoc committees reports, particularly those produced by the Advisory Council on Economic Growth, as well as bank economists and other expert roundtables or conferences. In terms of other sources, nine participants stated reading reports from other experts was insightful and useful. Lastly, reports or inspiration from other governments were useful in serving as background information for 11 interview participants, with three participants noting much of their future skills policy often embedded references to jurisdictions such as the UK, Singapore, and Australia.

48Based on the range of sources of information and interview participants’ own assessment of quality and utility, we argue that the policy community demonstrated a high level of epistemic learning that was not selective. We note that interview participants emphasized that they did not simply look to IOs and GMCFs as sources of information, but that their sources and processes of information gathering were extensive and diversified, including academic sources (some international but mainly the US and Canada), other experts, international and domestic think tanks, other governments, and data repositories.

Table : Summary of Interviewees’ Mention of Reports by IOs and GMCFs

Type of sources	Number of interviewees who read reports	Major organizations/reports mentioned
IOs	25	OCED, WEF, ILO, World Bank, UNESCO
GMCFs	21	McKinsey
Academics	21	Articles on Canadian context
Data repositories	15	Statistics Canada
Domestic thinktanks	14	The Dais (formerly Brookfield Institute)
International thinktanks	8	Brooking Institution
Government of Canada	3	Advisory Council on Economic Growth
Foreign governments	11	UK, Singapore, Australia

Source: the Authors

The Agenda Setting Power of IOs and GMCFs

49Our second expectation was that IOs and GMCFs are influential not only in providing information but also in setting the agenda and providing the framework for policy ideas and domestic policy decision making. That expectation is derived from the policy diffusion literature and research that predicts that economic competition drives the (rational) pursuit of countries’ adoption of policy ideas from elsewhere. We thus expected the authoritativeness and influence of those IO and GMCF reports would be most influential when the reported goals were economic competitiveness, as opposed to or in addition to other values. That is, pursuit of other values would lead actors to seek more sources of evidence.

50Instead, based on our coding, we found no relationship between policy goals and the learning processes. Rational learning processes of information gathering informed the interview participants regardless of whether they viewed the primary policy goals as economic competitiveness or other values such as equality or anti-poverty.

51Another finding contrary to the literature relates to the perceived disproportionality in the influence of IOs or GMCFs, with the expectation that disproportionate reliance on a single source or type of information can contribute to selective learning. Other researchers (e.g. Drezner, 2017; Fourcade et al., 2015) have tracked the disproportionate influence of economists in policy making; others the disproportionate influence of GMCFs (e.g. Mazzucato & Collington, 2023). The interview data reflected somewhat of a consensus among participants that IOs and GMCFs had a relatively high degree of influence, serving as credible and authoritative sources of information. Eighteen participants stated that they perceived IO reports as authoritative and 11 stated GMCFs have the ability to produce influential reports.

52We coded authoritativeness thematically. For example, one participant noted that IOs such as the OECD have influence in terms of measurement and standards assessments. They added that “whether you like it or not, [you have] to take it seriously, to try to improve, you want to see progress.” One participant noted that they read OECD reports “backward and forward”. Regarding GMCFs, one participant stated that “McKinsey will drop a report and everyone would stop and listen to see what they have put out. I think they command a lot of attention.” Another participant stated “McKinsey is exceptional” because they have mastered the art of influence in providing briefings that are attractive and informational to Ministers’ offices. Furthermore, six participants emphasized that IO reports are also influential in agenda-setting, whereas eight participants noted GMCFs are also influential in agenda setting. One participant noted that big IOs “tend to set the tone and the frame of the discussion when they come out with a report. It is usually a cool big splashy thing.” Another asserted that “many of those organizations have a lot of credibility inside government circles.” They further emphasized that IOs are more authoritative than GMCFs because the OECD, World Bank, and IMF usually have the same agenda. Thus, issues are brought to the agenda table of G20, G8, and G7 meetings that “put pressure on Canada to do something.” This signifies that IOs and GMCFs can influence the domestic agenda-setting process.

53However, participants in academia were more likely to view IO reports as having a moderate to low level of influence intellectually. As one participant stated: “OECD reports indeed form part of the overall context that eventually evolves and shapes the consciousness and agenda over time, but OECD reports tend to be less important than Ministers participating in OECD meetings.” Two participants who work in a research and evaluation agency or institute also noted that IO reports tend to have a moderate to low level of influence. One stated “IOs reports have varying quality and relevance for sure. They are indeed useful and inform our work but they have pretty superficial analysis….there is work we can do to improve the quality and relevance of them.”

54With regard to GMCFs, we observed a strong disagreement among some participants regarding their perceived authority; seven participants noted GMCF reports have a low level of influence and are not authoritative. One participant strongly suggested that “reports conducted by GMCFs lack methodological credibility that have zero ethics oversight or involvement”. That made them skeptical of their authoritativeness. Another participant similarly stated that GMCFs are largely based in the US and thus “many of their findings are either completely irrelevant or do not translate well for Canada.”

55Overall, the perceived authority/influence of IOs and GMCFs among participants was inconsistent; however, we observed a consensus that IOs and GMCFs are useful and credible in providing context, background information, and policy recommendations in the future skills policy area. Those who rejected the authority of IOs and GMCFs did so because they noted that many of their reports cannot effectively translate to a Canadian context.

56Our interview script also prompted interview participants to reflect on a key moment in Canadian federal policy making around future skills when the Government of Canada invested over $200 million dollars to create the Future Skills Centre. That decision stemmed from an independent economic advisory council report that recommended its establishment (Advisory Council of Economic Growth (2017a), suggesting epistemic learning at the heart of this policy decision. The interviews instead suggest a more complex decision-making process leading to the Centre’s establishment. Eight participants stated that the federal government’s decision to create the Centre is reflective of selective learning (policy makers disproportionately relying on a single source of information), in this case McKinsey. One participant stated McKinsey directly provided “a telling rationale for the creation of the Future Skills Centre, …[and]…was directly influential in creating the recommendation and they wrote the Learning Nation report” also produced by the Advisory Council of Economic Growth (2017b). Another participant stated that there is a sense that McKinsey was very involved in the drafting and curating of the recommendation to create the Future Skills Centre. One participant mentioned McKinsey (along with the federal Department of Finance staff) played an important role in providing information to the Advisory Council and curating resources and briefings to support the Advisory Council members’ discussions and report writing.

57The use of McKinsey statistical analysis in the advisory council’s reports is particularly interesting for our case because it reflects a comment made by one of our interviewees about the advisory council being highly influenced by the work of McKinsey team along with the Department of Finance team. An interviewee noted that many members of the advisory council had full-time jobs and therefore did not have adequate time to dig in for data or further information. So, they relied heavily on the support provided by outsiders, especially the analyses of the Department of Finance and McKinsey staff.

58Additionally, two participants specifically linked the influence of McKinsey to Dominic Barton, the former chair of the Advisory Council. They stated that McKinsey was very influential, not just in the advisory council process but also in future skill decision-making in general. One interviewee stated, “I think every now and then, McKinsey will drop a report and sort of everyone stops and listens to see what they’ve put on. I think they command a lot of attention.” This also signifies the importance of McKinsey in particular in the future skill policy community.

59Some interview participants observed IO and GMCFs’ agenda setting influence and expressed concern that decision making in the sector reflected selective uptake of ideas in an emulative fashion akin to “fast policy” making (Peck, 2011; Peck & Theodore, 2015). In particular, the interview evidence and citation analysis suggest decision-making around the establishment of the Future Skills Centre has some markers of fast policy making (White et al., 2022) fueled by selective learning where McKinsey played a major role in feeding information and curating resources to support the Advisory Council’s recommendations to government.

60Overall, though, and contrary to our expectation that selective learning would dominate, we find much more evidence of comprehensive epistemic learning, with variation observed based on the participants’ roles within their organizations. One participant noted that the federal Liberal government began in 2015 to hear from a broader range of academics who could inform Canadian economic and social policy making. Some participants noted the dearth of information in Canada regarding labor market interventions specific to the Canadian education and employment context. One goal for the Future Skills Centre was piloting programs, experimentation, and evaluation that emerged from a civil service reform ethos around “what works” in Canada. One interviewee noted the creation of Future Skills Centre rested on an idea of “let’s run some experiments and let’s inform future programs.” Another concurred that the idea of a “what works” Centre was around since 2013-2014 where the government expressed interest in understanding what programs are and are not working. Another participant stated that the creation of a Future Skills Centre might have been externally driven, but there was also a need for Canada to take the issue seriously by creating an organization to systematically look at these issues and generate evidence and research to help prepare for the future world of work. In total, 13 participants referenced the idea of “what works” or “experimental approach”. Interview participants, including those who worked in government, expressed the view that governments need more and better evidence to inform future programs and policies.

61Overall, our key take aways from our analysis is that IOs and GMCFs are perceived as authoritative in their informational capacity and somewhat so in their agenda setting power. Contrary to expectations, we found no relationship between what the participants articulated as goals of policy in their organization or in Canadian policy making in general and their own learning processes. Second, our evidence points to more epistemic learning in the future skills policy community than expected (Saleem et al., 2023; White et al., 2022) as well as news reports about McKinsey’s disproportionate influence in federal policy making (Boynton and Gilmore, 2023; Front Burner, 2023; Philipupillai, 2023; Schué & Gerbet, 2023). Evidence from interviews suggests a disproportionate influence of McKinsey in setting a policy agenda but also a great deal of comprehensively rational decision-making processes grounded in epistemic learning. It was surprising how much participants noted assessing the quality and relevance of the evidence they consumed.

Actors’ Learning and the Relationship to Their Organizational Roles

62We uncovered one more finding from our data regarding the variation of interviewees’ perceptions of IO and GMCF authority and influence based on their organizational role. Participants who had worked in government (5) and participants outside government (3) who had sat at decision-making tables were more likely to report that they perceived the policy process to reflect the selective uptake of ideas and the disproportionate influence of consulting firms such as McKinsey. In contrast, participants (10) who had not worked in government were more likely to report that they perceived government policy making related to future skills to be rooted in rational learning, with the government looking at a variety of information sources and engaging in a rigorous review of information before determining outcomes.

63We further analyzed the interviews to assess whether the perceived authority and influence of IOs and GMCFs varied depending on the actors’ organizational affiliation. We expected, for example, that members of think tanks would be more likely to pay attention to think tank reports. However, we did not find evidence of that pattern. Instead, those at the head of an organization (director or above) were more likely to describe selective updating of ideas and bounded learning than those who hold research or policy analyst roles, regardless of the organization. Those in the directorial role (3) were coded as engaging in selective learning. One participant noted the limited attention that policy makers in multi-issue units can give to a single issue at any one time. Another participant noted they had to go out of their way as a civil service head to read more widely than is typical in such a high-level position. This reinforces how policymakers have limited attention and competing policy interests that may lead them to engage in selective uptake of information (Jones and Baumgartner, 2005). For example, one participant noted that they get summaries of sources from staff that combine grey and academic literature. Another stated that they did not have time to read journal articles and only reviewed the papers they have commissioned. One interview participant reported depending on peer-to-peer relationships to signal whether to read reports. They reflected needing to build a relationship with those who produce reports first before reading those reports, stating “it is like a cold call” if reports come across their desk without a relationship. These reports are likely to be ignored. The majority of participants did not demonstrate selective learning, however.

Conclusion

64As a country in the Global North and a G7 member, Canada provides a key country case to examine the authority and influence of IOs and GMCFs on domestic policy making. The future skills policy sector has recently been a major source of IO and GMCF attention (Saleem et al., 2023), and thus, we expected IO and GMCF reports would receive disproportionate attention. Our findings refute that expectation. We found that participants expressed two predominant sets of goals in the future skills policy area: 1. interest-driven motives based on securing Canada’s economic competitiveness; and 2. other values-driven motivations such as principled beliefs about what governments “ought” to do for students and workers. Despite the ubiquity of policy reports produced by IOs and GMCs that offer solutions to these challenges, we found members of the policy community drew on a variety of sources, including academic work and data repositories. Despite the reported agenda setting power of McKinsey in the specific policy process that established the Future Skills Centre, overall, we find much greater evidence of epistemic learning – with IO and GMCF reports as one set of a broad range of sources – and more comprehensively rational decision-making processes than expected (Saleem et al., 2023; White et al., 2022). We did find, however, variation of interviewees’ perceptions of IO and GMCF authority and influence based on their organizational role (inside or outside government) and their ranking in the organization.

65This article contributes to the literature in a number of ways. First, our research is part of an emerging wave in policy science literature examining the role of non-state actors such as IOs, GMCFs, and other private entities in setting policy agendas and shaping the dynamics and processes of domestic policy advisory systems (Barnett & Finnemore, 2004; Cortell & Davis, 1996, 2000; Drezner, 2017; Faulconbridge & Muzio, 2017; Fourcade et al., 2015; Grimshaw, 2020; Momani, 2013, 2017; Risse, 2016; Stone, 2021). In the area of GMCF influence in particular, there is a growing literature on the varying impact of consulting firms on domestic policy making (Howlett & Mignone, 2013; Kipping, 2021; Marciano, 2022, 2023; Stone et al., 2021; van den Berg et al., 2019; Ylönen & Kuusela, 2019), and their role as ideas brokers. Our research expands the literature on these actors’ influences on domestic policy making and outcomes. Second, our research contributes to the growing literature on global ideational diffusion more generally and IOs’ and GMCFs’ impact on domestic policy making (Acharya, 2004; Blyth, 2013; Boushey, 2010; Hall, 1993; Linos, 2013; Simmons et al., 2008; Weyland, 2007; White, 2017). It also contributes to the literature on policy learning. It combines analysis of particular policy advisors (IOs and GMCFs) with analysis of what happens after the advice is given by examining both the recipients and their learning processes. It examines how both the characteristics of policy advisors and the goals of policy makers and learning processes contribute to selective versus more epistemic forms of policy learning.

66One limitation of our study is that the evidence is drawn largely from participants’ own assessment of their learning and sources of information and their own assessment of governments’ policy learning. As well, some participants offered retrospective reflections from time in a previous role. A further limitation is the policy sector is technical in nature and engaged in data-driven forecasting, which may naturally lend itself to epistemic than selective learning and more comprehensively rational decision processes. The nature of the bureaucracies dealing with these files may differ from other related economic files in that they are smaller units with fewer possibilities of advancement directly than, for example, Finance or Treasury Board. Those working in government on these policy files may thus do so because they are drawn to the policy content.

67Nevertheless, we believe the results of our study are both theoretically and empirically informative. As one participant mentioned, the future skills policy sector is not high profile; individuals do not tend to advance to a larger role within their organization compared to a more expansive sector such as health. Thus, participants are less likely to be motivated to develop deep knowledge of future skills. The epistemic learning in evidence thus likely reveals a policy curiosity about “what works” in the sector, especially in the face of uncertainty about the appropriate course of policy action. This case study provides an example of how epistemic learning can occur even in the shadow of economic competitiveness.