Introduction
Prime ministers in parliamentary systems are leaders of a collective executive, delegating policy responsibilities to their ministers. In doing so, they face the problem of agency loss: misalignment of interests or effort between prime ministers and ministers as their agents (Strøm, Reference Strøm2000; Berlinski et al., Reference Berlinski, Dewan and Dowding2010). Prime ministers have an array of tools to mitigate agency loss, including oversight by central agencies, control over cabinet process, coalition arrangements in multiparty executive contexts, and power to select, deselect or shuffle ministers. In recent decades, Canadian prime ministers have implemented a novel method of aligning their interests with ministerial activity: the mandate letter. These letters are documents provided to each minister, reinforcing government-wide priorities, explaining the minister's responsibilities and stating specific policy expectations for the minister's performance, creating essentially a contract for ministers to fulfill (Zussman, Reference Zussman2013: 22). They have been adopted at the subnational level in almost all Canadian provinces; Australian governments at both federal and state levels have also implemented a variant called charter letters (Weller, Reference Weller2007).
In 2015, Justin Trudeau became the first Canadian prime minister to publicly release these letters under the guise of accountability. The government itself, for example, maintained a mandate tracker on its website, which concluded that 68 per cent of mandate letter commitments had been fully met in the Liberal government's first term (Privy Council Office, 2019). Although mandate letters are well established and important to cabinet government in Canada, scholarly analysis of these letters is notably absent. This article seeks to fill the gap by examining Trudeau's mandate letters from 2015 to 2021, from the perspective of policy co-ordination. Our core research question is: how do mandate letters induce interministerial co-ordination on policy tasks? Using automated quantitative content analysis (topic models) and social network analysis methods, I demonstrate that Trudeau's mandate letters have increasingly sought to strengthen horizontal co-ordination of ministerial activity: expectations for interministerial collaboration and engagement on policy tasks have grown over time. Correspondingly, ministerial performance increasingly depends on a minister's ability to work productively with cabinet colleagues rather than individual or departmental achievement. The pattern of increasing ministerial co-ordination also has implications for broader debates over centralization of power in Canada.
The article proceeds as follows. First, I describe mandate letters in further detail and the relevant literature. Second, I provide a conceptual and theoretical framing of mandate letters in terms of the principal-agent problem and policy co-ordination. This section also introduces the context of Justin Trudeau's government and approach to cabinet decision making. Third, I discuss the study's data and methodological approaches. The fourth section relates the article's findings. Finally, I discuss these findings and directions for future research.
Mandate Letters
Mandate letters are documents issued by Canadian prime ministers or provincial premiers to each cabinet minister outlining overall government priorities, advice on ministerial behaviour, and specific tasks and responsibilities in each portfolio (Zussman, Reference Zussman2013: 93). They have been issued federally since the 1970s (Savoie, Reference Savoie1999a: 343), and most provincial governments have also adopted the practice. Prime Minister Justin Trudeau has issued full sets of letters four times as of this writing: after the 2015, 2019 and 2021 elections, and in January 2021 to address the COVID-19 pandemic. Letters were typically issued annually under Trudeau's predecessor, Stephen Harper (Brodie, Reference Brodie2018: 119). However, despite their long-standing existence, mandate letters have not been examined by scholars except in passing or as relevant to specific policy issues. Public availability of the letters since 2015 has led to media commentary (for example, McParland, Reference McParland2019; Wells, Reference Wells2020), but scholarly analysis has been absent.
What we do know about mandate letters is somewhat partial given the fragmented literature, much of which is now not recent. Often, development of mandate letters begins with the transition team preparing a party leader for government or with the political staff in the Prime Minister's Office (Zussman, Reference Zussman2013: 93; Marland, Reference Marland2016: 214). Savoie (Reference Savoie1999a) found that civil servants in the Privy Council Office write and finalize the letters in the post-election period, but they are distributed to ministers in the name of the prime minister. As Zussman (Reference Zussman2013: 177–78) reports, mandate letters contain partisan commitments, policy directives, and guidance on how to be an effective minister. The mixture of these aspects varies by prime minister, but the trend has been toward greater elaboration. Zussman (Reference Zussman2013: 93–94, 178) also reports that Jean Chrétien's (1993–2003) letters were more strategically oriented, while Stephen Harper's (2006–2015) letters took a more detailed, directive approach, featuring a “call to arms” activist style focused on a small number of core priorities. Marland (Reference Marland2016: xix, 210) argues that Trudeau's mandate letters mostly frame ministers as “brand ambassadors,” with little independent influence on the policy agenda.
Regardless of style and structure, the letters have become important internal accountability mechanisms for prime ministers as they delegate responsibilities (Wernick, Reference Wernick2021; Zussman, Reference Zussman2013). Trudeau's public release of the letters added an element of public accountability, supplemented by the mandate letter tracker on the Privy Council Office website (Privy Council Office, 2019), which kept a record of commitments met in the Liberal government's first term (2015–2019).Footnote 1 This accountability has largely been internalized, such that most ministers place great importance on their instructions and the expectations of the prime minister. As Wernick (Reference Wernick2021: 121) advises, one of the first things ministers should do is plan the implementation of their mandate letter with their political and civil service advisers. Thus, mandate letters are also a way of keeping ministerial policy ambitions in check (Savoie, Reference Savoie1999a: 324; Wernick, Reference Wernick2021). Because of this, deputy ministers aim to shape the contents of letters to protect departmental interests against deviation from ministers (Savoie, Reference Savoie1999a: 343; Zussman, Reference Zussman2013: 179–80). As mandate letters have become more detailed, ministers have less time and space to develop policy initiatives of their own (Lewis, Reference Lewis, Lewis and Everitt2017: 281). Indeed, this constraint on ministerial innovation and autonomy is the basis for Lang's (Reference Lang2022) argument that these letters undermine “cabinet government.” However, Lewis (Reference Lewis2013: 812) reports that ministers tend to believe they have “a lot” of policy influence both within and outside of their mandate letters.
While mandate letters have become established tools of prime ministerial leadership, it is important to recognize that the act of publicly releasing the letters makes Trudeau's mandate letters different in purpose to previous letters, to some extent: they serve a public relations function that previous letters did not. Thus, we cannot and do not directly compare the private, pre-2015 letters to those issued by Trudeau. Still, this additional public-oriented function complements, rather than detracts from, the fundamental use of letters as tools of centralized control; they have substantive effects on how ministers behave and reflect how prime ministers construct their policy priorities and implementation. We turn to these roles in the next section.
Prime Ministers, Ministers and Policy Co-ordination
The issuing of mandate letters by Canadian prime ministers is firmly established practice. Prime ministers find it useful to issue marching orders to ministers as a mechanism of accountability and benchmarking; they instruct ministers on what they need to accomplish and how they should do so (Wernick, Reference Wernick2021). But mandate letters may also reflect, and indeed structure, co-ordination of policy effort among ministers. Why would prime ministers use mandate letters to induce policy co-ordination among ministers? I frame the answer within the principal-agent perspective. The structuring of co-ordination among ministers is a mechanism to counteract agency loss between the prime minister as principal and ministers as agents. This characterization is a structural feature of Canada's parliamentary system, which has a chain of delegation from voters to the legislature, from the legislature to the prime minister, and from the prime minister to ministers (Strøm, Reference Strøm2000). By appointing ministers, prime ministers delegate governmental responsibilities. In so doing, they face the moral hazard problem—motivating ministers to act in the prime minister's interest once they are appointed, given imperfect capacity to monitor and/or sanction. Many prime ministerial powers, such as appointment and portfolio allocation, shuffles, centralized pressure, and cabinet committees, are framed as mechanisms to mitigate these problems (Berlinski et al., Reference Berlinski, Dewan and Dowding2010; Dewan and Hortala-Vallve, Reference Dewan and Hortala-Vallve2011; Indridason and Kam, Reference Indridason and Kam2008).
Mandate letters are one solution to the prime minister's dilemma of motivating ministers to act in the governmental, or prime ministerial, interest. This dilemma arises not so much because ministers have fixed preferences that are misaligned with the prime minister's and will try to move policy closer to their ideal point. The typical cabinet minister in Canada is a loyalist, with broad ideological commitments but rarely specific policy goals (Alexiadou, Reference Alexiadou2016). The strictness of party discipline, cabinet solidarity, and centralization of power in Canada mean that policy concerns are outweighed by desires to be a loyal team player. As Savoie (Reference Savoie1999b) argues, very few ministers will risk their careers and status to defend personal ideological or policy beliefs. He also argues that there are very few, if any, “big beasts” in modern-day Canadian cabinets: that is, ministers with independent power bases and clearly defined, strongly held beliefs (Reference Savoie2019: 198). Moreover, the relatively high turnover and shuffling rate of ministers in Canada means that ministers have little opportunity to develop policy expertise or be subject to bureaucratic capture. Indeed, policy disagreement is extremely rare: only about 10 ministers have ever resigned from federal cabinet for this reason (Heard, Reference Heard2014: 108).
Therefore, the prime minister's dilemma is less reining in policy-ambitious ministers and more co-ordinating the actions of uncertain, imperfectly informed ministers. Even if ministers share the goals of the prime minister, their work as agents will be ineffective if those goals are not clear to them. These ideas reflect work by Calvert (Reference Calvert1992), Dewan and Myatt (Reference Dewan and Myatt2008) and Landa and Tyson (Reference Landa and Tyson2017), among others, which considers the importance of political leaders in solving co-ordination problems by communicating information to followers about desired states of the world and/or introducing costs of deviation from the leader's policy goals. Mandate letters provide information about the goals and priorities of the prime minister, and they increase costs of deviation because they increase the capacity of the leader to measure and monitor compliance.
The dilemma of co-ordination also arises because most modern policy problems are crosscutting “wicked” problems that require interactive and collaborative processes across government departments and agencies (Pollitt, Reference Pollitt2003: 46). This problem has been called the “holy grail” of government and public administration because most agree on its desirability, but both its meaning and attainment have been elusive (Peters, Reference Peters2013: 569). One definition refers to “instruments and mechanisms that aim to enhance . . . the alignment of tasks and efforts” to achieve “greater coherence, reduce redundancy, lacunae, and contradictions” in decision making and policy implementation (Bouckaert et al., Reference Bouckaert, Guy Peters and Verhoest2010: 16). Similarly, Bakvis and Juillet (Reference Bakvis and Juillet2004: 8) refer to co-ordination as “the practice of aligning structures and activities to improve or facilitate the likelihood of achieving horizontal objectives, to reduce overlap and duplication, and, at a minimum, to ensure that horizontal objectives are not impeded by the actions of one or more units.” Successful policy co-ordination entails multiple actors working together to achieve shared objectives, while poor or absent co-ordination involves multiple actors working against each other, undermining shared goals. Even if ministers are well intentioned, the policy literature tells us that many modern policy issues are complex and involve crosscutting responsibilities of many public sector actors, not just one agency or department (Peters, Reference Peters2018: 2). Problems such as climate change simply cannot be addressed by the efforts of one minister or department. However, the involvement of multiple actors raises the likelihood that actors’ efforts will suffer from inconsistency, contradiction, duplication, or displacement (one actor's decision creating problems for other actors) (Peters, Reference Peters2018: 3–4). While problems of policy co-ordination are typically considered as issues of public administration, they are equally compelling for the political executive. Prime ministers must delegate tasks to ministerial agents, but these tasks are commonly in the domain of wicked policy problems, for which individual ministerial effort is insufficient and could be counterproductive when it impedes or undermines the efforts of others.
Our empirical expectations are informed by the context of Justin Trudeau's tenure as prime minister (2015–2021). The Liberal victory in 2015 was partly driven by the successful framing of Trudeau and the party as reforming across a host of arenas—the House of Commons, the Senate, the electoral system, the Supreme Court—in which Trudeau promised “real” change. Cabinet was no different, with Trudeau stating that “government by cabinet is back” (Van Dusen, Reference Van Dusen2015). Drawing a sharp contrast to Harper's centralized approach to managing ministers, the implication was that Trudeau would allow ministers greater autonomy from the centre, both in communications and in policy direction. He would allow ministers to speak and act more freely and to initiate and develop policy without interference from the centre of government. However, Savoie's skepticism (CBC News, 2015) that such reform would be enduring seems to have been justified. For many reasons—inexperienced ministers, political imperatives of message control and branding, decision-making efficiency, and policy co-ordination and implementation goals—Trudeau's “cabinet government” arguably looked much like his predecessors (Allen, Reference Allen2017; Ivison, Reference Ivison2019; Marland, Reference Marland2020). As Savoie (Reference Savoie2019: 196) concluded at the end of the first term, “if anything, we have moved further away” from cabinet government. This reading of cabinet government under Trudeau thus suggests a shift over time from emphasis on ministerial autonomy toward ministerial control. This speaks to a learning process in which both the need for centralized direction and crosscutting policy complexity became more apparent. Empirically, this leads us to posit an over-time trend in the extent to which mandate letters induce policy co-ordination. I posit that from 2015 through 2019 and the COVID-focused letters of 2021, the prime minister will instruct ministers to work increasingly collaboratively on increasingly shared policy priorities. Specifically, this means that (1) policy priorities will be increasingly shared among ministers, (2) ministers’ scope of policy activity—the tasks for which they are responsible—will broaden and (3) explicit co-ordination instructions will increase over time. In short, the mandate letters should demonstrate stronger recognition in Canada of the problems of agency loss and the necessity of policy co-ordination: problems that are of larger significance for prime ministers and political executives generally.
Data and Methodology
I use mandate letter data, topic models, and social network analysis to uncover structures of induced ministerial co-ordination. There have been four sets of cabinet-wide mandate letters released by the Trudeau government: one set each after the 2015 and 2019 elections; one set of “supplementary” letters in January 2021 (which I label “2021a”), focused on the federal COVID-19 pandemic response; and one set in December 2021, after the September election (which I label “2021b”). In total, N = 140 (30 in 2015; 36 in 2019; 36 in January 2021; and 38 in December 2021).Footnote 2 Texts of the 2019 and 2021 letters were gathered from the prime minister's official website, while older letters were collected from archived versions of that site. Mandate letters are written using a standard template in which specific tasks for each minister are sandwiched between shared instructions and advice; we extracted the unique section of each letter for analysis.
Our empirical strategy for assessing ministerial policy co-ordination is twofold. First, we uncover the policy priorities in mandate letters and measure the extent to which they are shared across ministers and, conversely, the degree of ministerial involvement in multiple policy areas. This allows us to assess the first two hypotheses: shared policy priorities and ministerial scope. Second, I examine explicit co-ordination instructions to ministers in mandate letters, in the form of directing ministers to “work with,” “support,” or otherwise pursue joint efforts. This addresses the third hypothesis of increasing co-ordination instructions over time. To determine the policy priorities embedded in the letters, I employ topic modelling. Topic models are algorithms for identifying a latent set of topics or themes from a set of texts (Isoaho et al., Reference Isoaho, Gritsenko and Mäkelä2019). Rather than predetermining categories and employing classification to allocate texts to categories, topic models simultaneously estimate latent topics and the proportions of each topic found in each document. The increasing number of applications in political science include measuring legislative agendas (Grimmer, Reference Grimmer2010; Quinn et al., Reference Quinn, Monroe, Colaresi, Crespin and Radev2010) and parliamentary speech (Curran et al., Reference Curran, Higham, Ortiz and Filho2018). I employ the most widely used algorithm for topic modelling: Latent Dirichlet Allocation (LDA) (Grimmer and Stewart, Reference Grimmer and Stewart2013: 284). LDA assumes that each document (mandate letter here) contains a mixture of underlying topics that are theoretically common to all documents; each of these topics is a distribution over words. The algorithm estimates a posterior distribution of the latent variables of interest—the topics, overall topic proportions, and per-document topic proportions—as a function of the observed word distributions per document (Curran et al., Reference Curran, Higham, Ortiz and Filho2018; Grimmer and Stewart, Reference Grimmer and Stewart2013; for a technical description, see Blei et al., Reference Blei, Ng and Jordan2003). The first step in constructing the topic models was to preprocess each document.Footnote 3 I then proceeded to estimate the topic models using the topicmodels R package.Footnote 4 The chosen topic labels and each topic's highest probability words are included in the appendix as Table A1.Footnote 5
To examine explicit co-ordination instructions found in mandate letters, I employ social network analysis techniques, which have gained currency in political science (for overviews, see Ward et al., Reference Ward, Stovel and Sacks2011; Victor et al., Reference Victor, Montgomery and Lubell2017). The underlying ideas of social network analysis—that many phenomena in politics are relational and that the features of interest are often interactions among actors, rather than individual actors—are profound. However, network analysis concepts have yet to be applied widely in studies of executive politics. This is surprising because cabinets in Canada have developed network-like structures and purposes as they have become more institutionalized, with complex, multilayered ministerial interactions through such features as collegial budgeting processes and cabinet committees (Bernier et al., Reference Bernier, Brownsey and Howlett2005). The specific measures and methods I use are described in more detail below; here, I explain the overall application of network analysis to the mandate letter data.
The basic concepts of network analysis are the node and the tie (or edge). A node is simply an actor, such as a country or legislator; a tie is a theorized relationship between actors, such as trade flows or cosponsorship of bills. In the ministerial context, I take advantage of the fact that Trudeau's mandate letters specifically instruct ministers to work with certain other ministers. For example, the 2019 Crown–Indigenous Relations letter states that the minister should “deepen work with the Minister of Finance, working with the Minister of Indigenous Services, to establish a new fiscal relationship with Indigenous Peoples” (Trudeau, Reference Trudeau2019). Such statements explicitly induce structures of co-ordination among ministers. They communicate to ministers that many of the tasks the prime minister has set for them are not, and cannot be, accomplished individually but rather in partnerships and teams that the prime minister directs to form.
Thus, I coded all instances of such statements, which I call co-ordination ties, for all ministers and use these ties as our basis for assessing levels of induced ministerial co-ordination. Depending on the measure, I treat co-ordination ties as either directed or undirected and either binary or valued. Directed ties specify a direction—for example, “Health → International Development” specifies that the first minister should co-ordinate with the second. Undirected ties specify only the presence of a link between two nodes. In fact, the letters are not entirely symmetric: some co-ordination ties are not reciprocated. A binary tie indicates only presence or absence of one or more co-ordination ties, while a valued tie denotes strength: the number of co-ordination ties between two ministers. For example, the 2015 National Defence letter instructs the minister to work with the foreign affairs minister three separate times. Examining the co-ordination ties among ministers given in mandate letters enables us to assess how Trudeau has structured policy co-ordination at the ministerial level and, especially, whether changes over time are evident.
Mandate Letters and Ministerial Co-ordination: Analysis
This section tackles our primary research question: to what extent have Justin Trudeau's mandate letters structured ministerial policy co-ordination and how has that changed over time? To reiterate, my empirical strategy is twofold: (1) to analyze the policy priorities in mandate letters, how they are shared across ministers, and the extent to which ministerial tasks engage multiple priorities; and (2) to examine co-ordination ties in mandate letters. Figure 1 shows the results of the topic models, which uncover the latent policy priorities in the mandate letters. Specifically, the figure shows, in descending order, the overall topic proportions—the estimated proportion of words in the letters that are assigned to each topic—and thus display a relative ordering of policy priorities. In 2015, upon entering office, Trudeau's mandate letters most frequently mention climate change, trade and investment, the public service, and sport and culture as policy priorities. In 2019, the most frequent topic mention is social services, at almost 10 per cent, followed by seven policy areas with essentially equal proportions, including climate change, immigration, and the economy and taxes. Unsurprisingly, several of the most frequent topics in the January 2021 mandate letters relate to the COVID-19 pandemic, including vaccines (12 per cent), economic issues (9 per cent) and pandemic-related foreign policy issues (7 per cent). However, because this set of mandate letters was comprehensive, even if motivated by the pandemic specifically, other policy priorities are evident, including diversity in the public service and Indigenous services. Finally, in the 2021 post-election mandate letters, regional policy (essentially, federal government funding for local and regional economic development, infrastructure, and housing) comes to the fore, while employment and health care are also frequently mentioned priorities. While these results are suggestive about the relative priorities in Trudeau's delegations of responsibilities to ministers, they should not necessarily be interpreted as indicators of the overall policy priorities of the Trudeau government. A topic may be more frequently mentioned because it is of higher priority; alternatively, it may be because there are more specific, concrete tasks associated with that policy area for ministers to work on, or even that the policy area implicates ministers who require more guidance in the prime minister's view. These competing characterizations cannot be readily evaluated here, but merit future consideration.
Figure 1 Policy priorities in ministerial mandate letters, 2015–2021. Bars indicate estimated topic proportions.
More importantly, our uncovering of the relative policy priorities in ministerial mandate letters allows us to assess the extent to which policy priorities are shared among ministers and, conversely, the extent to which each mandate letter focuses on multiple policy priorities. I consider both indicators of how Trudeau's mandate letters induce co-ordination on policy priorities across ministers, though from different angles. The first is more directly a measure of policy co-ordination, while the second is a measure of ministerial scope of activity; these are related but distinct. For example, if each minister's mandate letter only concerned one policy priority but many policy priorities were shared among multiple ministers, policy co-ordination on those priorities is more likely but each minister has limited policy scope. If each minister's mandate letter concerned all policy priorities and every policy priority was the responsibility of every minister, the level of co-ordination would be maximal. These two measures are extracted from the per-document, per-topic scores: the proportion of each mandate letter assigned to each policy priority. These are visualized as heatmaps in Figures 2 and 3 for the four sets of mandate letters. Proportions are indicated by shading: a black-shaded cell indicates a proportion close to 1 (that is, the letter concerns only one policy priority), a white cell indicates proportions close to 0, and a greyscale gradient is used for intermediate values.
Figure 2 Per-document, per-topic proportions: 2015 and 2019 mandate letters.
Figure 3 Per-document, per-topic proportions: January and December 2021 mandate letters.
The results are somewhat mixed but generally support the hypothesis that the level of induced ministerial co-ordination on policy priorities has increased over time. Figure 2 shows a strikingly low level of induced policy co-ordination on both measures. On average, only 2.3 ministers are assigned to a policy priority. Two priorities, housing and energy and resources (the two least mentioned), have a single minister assigned tasks (Families, Children & Social Development and Natural Resources, respectively). The most shared policy areas are shared by only three ministers. As well, almost all 2015 mandate letters—28 of 30 (93 per cent)—focus on a single policy priority. Put another way, only 7 per cent of ministers are estimated to have tasks assigned in multiple policy priorities. In 2019, the average number of ministers assigned to a policy priority increases slightly to 2.7. More tellingly, no priorities are the domain of a single minister, and more are shared among clusters of three or more ministers (8 versus 6 in 2015). The most shared policy area, social services, is shared among four ministers. However, 35 of 36 ministerial mandate letters (97 per cent) focus on only one policy area; only one minister, Diversity, Inclusion & Youth, has tasks in multiple policy areas. These results suggest that the level of direct co-ordination over policy increased from 2015 to 2019 but that the policy scope of ministers—how wide-ranging their tasks are—did not change.
The January 2021 mandate letters, however, demonstrate significant increases in induced policy co-ordination on both measures. The effect of the COVID-19 pandemic on increasing demands for policy co-ordination is clear, but the changes are not limited to pandemic-specific policy priorities. Figure 3 shows visually that the average number of ministers assigned to a policy priority is significantly higher (3.5) and that there are larger clusters of ministers assigned to several priorities. For example, the most frequent topic, Vaccines, is shared by five ministers: Health, Public Services and Procurement, Digital Government, Seniors, and International Development. Policy priorities such as Indigenous Services and Workforce Diversity are also shared by five ministers. Ministers are also more likely to be assigned tasks in multiple policy areas, with 10 of 36 now assigned to multiple policy priorities (28 per cent), significantly up from 2015 and 2019.
Finally, the December 2021 mandate letters demonstrate stronger policy co-ordination on the measure of shared policy priorities but less so as regards ministerial scope. The average number of ministers assigned to a policy priority increases from 3.5 to 3.9. The most frequently mentioned and shared policy priority—regional policy—is shared across eight ministers with varying portfolios: Rural Economic Development, Sport, Northern Affairs, Official Languages, Tourism, Housing, Diversity & Inclusion, Government House Leader, and the minister for the Federal Economic Development Agency for Southern Ontario. There are also large ministerial clusters in policy priorities such as health care, mitigation of climate change, climate change infrastructure, Indigenous services, and employment. However, we do not see the same pattern for ministerial scope: the percentage of ministers assigned to multiple policy priorities declines from 28 to 16 per cent. Thus, I conclude that our first expectation of increasingly shared policy priorities among cabinet ministers is supported, while the picture for our expectation of broadening policy scope for individual ministers, encompassing more policy priorities, is less clear. There is a significant increase in ministerial policy scope from 2015 and 2019 to the COVID-19 supplemental mandate letters in January 2021, but there is a decline in December 2021, although the latter level is still significantly higher than in 2015 and 2019.
Our second approach to assessing policy co-ordination in ministerial mandate letters is examination of the letters’ explicit co-ordination instructions: their “coordination ties.” Table 1 displays four measures for the four sets of mandate letters. “Density” is the number of observed ties as a proportion of all possible ties, a basic indicator of the connectedness of a network. This is calculated on the binary version of the networks, where only presence or absence of a co-ordination tie, rather than its strength, is denoted.Footnote 6 The “Avg. degree (binary)” entries indicate the mean number of co-ordination ties across ministers if considering only binary ties, while the “valued” column averages the strength of co-ordination ties across ministers. Finally, the “Clustering coefficient” is an average across ministers of the local clustering coefficient: the proportion of ties between ministers with whom a minister has ties. For example, if a minister has co-ordination ties to three ministers A, B and C, but only A and B have co-ordination ties between them, the local clustering coefficient is 0.33 (there are three possible nonreflexive ties: AB, AC and BC). Intuitively, this is a measure of how connected each minister's local network is; averaged over the network, it indicates how strongly the network structures groups of connected ministers.
Table 1 Network Statistics for Mandate Letter Networks, 2015–2021
These overall network measures tell a consistent story: over the period 2015 to 2021, through four sets of mandate letters, Trudeau has sought to induce increasing policy co-ordination among cabinet ministers. In 2015, 20 per cent of all possible ties between ministers were observed. The 2019 density is the same, but both 2021 mandate letters demonstrate higher density scores, increasing to 23 per cent of possible co-ordination ties in January 2021 and 27 per cent after the 2021 election. While density is only significantly different in 2021, relative to 2015, it should be noted that the number of possible ties in a network typically increases faster than the number of actual ties, since n + 1 nodes increases the possible edges by n: density will decrease unless each additional tie increases the average number of ties by n/(n − 1). Thus, densities are “almost always lower in large networks than in small networks” (Borgatti et al., Reference Borgatti, Everett and Johnson2018: 175). The density value for 2019, then, should not be interpreted as indicating a less connected network compared to 2015. Indeed, the trend of increasing connectedness is clear if we consider the average number and strength of co-ordination ties (degree) of individual ministers. For the binary measure indicating only presence or absence of a tie, ministers in 2015 are instructed to work with an average of 5.77 other ministers; this rises to 6.69 in 2019, 7.94 in January 2021, and 9.84 in December 2021. This difference over time is statistically significant under a one-way ANOVA test (F(3,136) = 6.97, p = 0.00). Put another way: in 2015, ministers were only instructed to co-ordinate on policy efforts with 20 per cent of cabinet; by 2021, this rises to 27 per cent. We see the same increasing trend when considering the strength of co-ordination ties, not just their presence or absence (“avg. degree (valued)” in Table 1). In 2015, the average minister was mandated by their mandate letter to co-ordinate policy efforts with other ministers 8.53 times. This increases to 9.19 times in 2019, 11.44 times in January 2021, and 14.11 times in December 2021.Footnote 7 This change is also statistically significant (F(3,136) = 4.76, p = 0.00). As Trudeau's tenure continues, the mandate letters demonstrate that ministers are increasingly being instructed explicitly to work with and support more of their colleagues and in more of their delegated tasks.
The clustering coefficient measure also supports the narrative of increasing co-ordination over time in Trudeau's mandate letters. This measure begins at 0.20, indicating that in 2015, on average, a minister's local network (the ministers with whom they have ties) has only one-fifth of the possible ties among them. In 2019, the clustering coefficient increases to 0.28, then to 0.38 in January 2021, and finally to 0.40 in December 2021. There are twice as many ties in the average minister's local network at the end of 2021, relative to Trudeau's first mandate letters in 2015. Over time, then, Canadian cabinet ministers under Trudeau are more likely to have local networks that are more strongly connected; clusters of ministers who are all directly connected to each other (called cliques) are more likely to arise over time. Indeed, this can be measured directly by counting the number of cliques for all node groups of two or greater (called a clique census). In 2015, there are only 40 ministerial clusters with maximal direct connections.Footnote 8 The 2019 mandate letters construct 54 such clusters, jumping to 63 in January 2021, including two 6-minister cliques for the first time. Finally, I find 79 cliques in the December 2021 letters, a twofold increase over 2015. This observation is reinforced by examining a related measure, the triad census, which counts occurrences of all possible configurations of three nodes in a network (Borgatti et al., Reference Borgatti, Everett and Johnson2018: 182). We focus on two configurations that most demonstrate clusters of co-ordinated ministers: the triad A↔ B ↔ C, meaning that there are mutual ties between A and B and B and C, suggesting B as a “lead” minister, and A ↔ B ↔ C, A ↔ C, three ministers with all mutual ties. Occurrences of the first configuration number 183 in 2015, 322 in 2019, 753 in January 2021, and 945 in December 2021. Occurrences of the second configuration number 29 in 2015, 54 in 2019, 153 in January 2021, and 210 in December 2021. These measures strongly support the hypothesis that explicit inducements of ministers to co-ordinate policy effort increase during Justin Trudeau's tenure.
Discussion and Conclusion
Mandate letters are important documents provided to Canadian ministers by the prime minister and subnational executives but have yet to receive systematic scholarly investigation. This article begins to fill that gap by examining mandate letters as mechanisms of co-ordination among ministers. I hypothesized that analysis of Justin Trudeau's mandate letters from 2015 to 2021 would show that policy priorities are increasingly shared across ministers, that ministers’ scope of policy activity would broaden, and that explicit instructions to co-ordinate would increase over time. Empirical analysis revealed reasonably consistent evidence that policy priorities in mandate letters have become increasingly shared among ministers over time, demonstrated by growth in the average number of ministers assigned to policy priorities and larger clusters of ministers for the most frequently discussed priorities. Evidence for broadening ministerial scope was more mixed: the COVID-focused January 2021 mandate letters showed a significant increase in scope from 2015 and 2019, measured as the share of ministers with tasks in multiple priorities, but this share declined in the December 2021 letters, if still higher than previously. Finally, the hypothesis that explicit co-ordination instructions to ministers would increase was consistently supported on several distinct measures. Ministers had more and stronger co-ordination ties and were more likely to be members of locally connected clusters of ministers, such as cliques and triads with mutual co-ordination ties.
Overall, then, our analysis demonstrates that mandate letters are not simply instructions to individual ministers, working autonomously. Rather, they can and have been used to induce co-ordination of ministerial policy effort to address the fundamental agency problem facing prime ministers and other chief executives. Like many new leaders, Trudeau entered office promising greater ministerial autonomy and a “return” to cabinet government. Our results suggest that the imperatives of coherent, strategic governance and cross-government policy co-ordination make such promises difficult to keep; they may also suggest that Trudeau did not sufficiently account for the complexity of governing in earlier periods and has learned the importance of co-ordinated effort to meet that complexity over time. They also speak to broader debates about centralization of power in Canada and the erosion of cabinet government by an overly powerful prime minister and central agencies. While this article takes no position on the normative aspects of these debates, it suggests that prime ministers are increasingly adept at employing mechanisms of control that do not require constant engagement and oversight. Donald Savoie's (Reference Savoie1999a: 336) view that prime ministers have limited time and resources and thus typically should “govern by bolts of lightning”—that is, by focusing on and driving through priorities and well-timed interventions—remains true. But the explicit effort in mandate letters to dictate the substance and process of ministerial policy making suggests that prime ministers and the centre of government need not rely on active intervention to enforce centralized decisions; rather, they can construct ministerial roles from the outset and allow ministers the autonomy of being accountable for their performance relative to these roles.
Any comparison of Trudeau's cabinets and change over time encounters the problem of COVID-19, which reached political salience in early 2020 and continued to dominate governmental policy making at all levels for at least three years after. The federal government was called upon to produce dramatic, unprecedented economic supports, constrict international movement, disseminate public health information, frame the crisis for the Canadian public, and manage protective equipment and vaccine distribution in a complex federal system (Turnbull and Bernier, Reference Turnbull and Bernier2022; Paquet and Schertzer, Reference Paquet and Schertzer2020). The pandemic thus necessitated crisis and emergency modes of governance, which tend to reduce executive accountability to legislatures and encourage executive aggrandizement (Bolleyer and Salát, Reference Bolleyer and Salát2021; Brock, Reference Brock2022; Gidengil et al., Reference Gidengil, Stolle and Bergeron-Boutin2022; Turnbull and Bernier, Reference Turnbull and Bernier2022). As Turnbull and Bernier (Reference Turnbull and Bernier2022) argue, a virtue of the Canadian Westminster executive in periods of crisis governance is its flexibility and inherent centralization. The federal executive was able to quickly implement procedural changes within cabinet decision making, such as the COVID-19 cabinet committee, without obstruction. Similarly, the pandemic provided both opportunity and institutional adaptability to further entrench horizontal co-ordination of ministerial activity as a tool of centralized control, with Trudeau significantly reshaping and strengthening co-ordination among ministers in January and December 2021 to focus efforts on pandemic response. The enormity of the COVID-19 crisis, and the uncertainty and chaos of its policy-making environment, also generated a need for stronger co-ordination and for effective delegation and allocation of ministerial work (Turnbull, Reference Turnbull2020: 36). Thus, one clear implication of this study is that future crises should have a similar effect of enabling and incentivizing prime ministers to induce co-ordination within the political executive.
While these findings represent an initial assessment of mandate letters as key aspects of policy making, executive politics and prime ministerial leadership, they are promising for future research. For example, does executive-level co-ordination matter substantively? Are more strongly co-ordinated cabinets more (or less) effective than more weakly co-ordinated cabinets? Do they produce different outputs? Given that Canadian prime ministers seem constantly to effect cabinet and machinery of government change, it is plausible that such changes do matter. Measuring the impacts of co-ordination is difficult given that cabinet-level mandates are only one moving part out of many in government and the broader political context. Still, more qualitative analysis informed by firsthand participant accounts would be a step in answering these questions.
Other avenues suggest themselves. First, there is much more fine-grained, detailed work to be done investigating collaboration and co-ordination relationships in cabinets. Which ministers are more and less integrated into these networks, and why? How does mandate letter co-ordination map onto other mechanisms of cabinet management, such as cabinet committee membership? Second, comparing Justin Trudeau's mandate letters to those of previous prime ministers, provincial premiers, or the charter letters issued in Australia could greatly illuminate and contribute to a broader theorization of the role of such mechanisms. Finally, this article focuses on the text of mandate letters as input into co-ordination relationships among ministers. It does not go beyond that to examine how mandate letters are used in practice, in everyday cabinet decision making and relationships between first ministers and ministers. How often, and how, do prime ministers use mandate letters in interactions with ministers? Mandate letters are a distinctive form of executive management: students of executive politics and policy making in Canada and elsewhere could learn much from them.
Competing Interests
The author declares none.
Appendix
Table A1 Topic Labels and Most Frequent Words by Topic
Open access
