Two-Level Games and the Policy Process: Assessing Domestic–Foreign Policy Linkage Theory

Literature Review

In pursuit of causal explanations of foreign policy behavior, the study of international relations has largely been conducted at one of three levels (or “images”), proffered by Kenneth Waltz in the 1950s: the systemic level, where analysts seek policy explanations given a state’s standing relative to the international system; the state or domestic level, where analysts consider the social, cultural, and political drivers of a given state’s behavior; and the individual level, where a single statesman’s personal and psychological motivations are scrutinized (Waltz 1959). Waltz (1959, 229–30) recognized the shortcomings of focusing on a single level, noting that “the partial quality of each image sets up a tension that drives one toward inclusion of others . . . One is led to search for the inclusive nexus of causes.” An example of this single-level theoretical analysis is arguably the single most dominant school in international relations, the realist tradition. Realism holds domestic factors constant and relies heavily upon a system level, top-down approach to illustrate the constraints of state behavior based upon the relative distribution of international power (Gourevitch 2002; Waltz 1959; Wivel 2017).

In the 1960s and 1970s, James Rosenau was an important international relations theorist who argued that the isolation of domestic-level factors hindered the growth and overall explanatory power of the field. His concept of linkage politics sought to address this deficiency by explicitly making the connection between the state and system levels (Rosenau 1969b). Rosenau’s work successfully brought domestic-level concerns onto the research agenda. Yet, despite this potential, domestic concerns seemed for a time to lack the influence to strongly drive the direction of methodological development, and the process of including the domestic variable in foreign policy analysis remained somewhat stagnant (Lentner 2006; Putnam 1988). ²

In the late 1980s, however, a breakthrough was made when Robert Putnam presented the Two-Level Game as an enhanced method of foreign policy analysis. Building on Rosenau’s work, Putnam (1988) argued for the necessity of focusing on the interplay between domestic and foreign politics. As Putnam’s analysis does much to fill in the gaps of international relations theory, it is a significant milestone in overcoming the domestic and foreign policy bifurcation. ³ The Two-Level Game is essentially a game theoretical international bargaining model where two or more nations are engaged in negotiations at a particular point in time over a particular issue. At the center of the Two-Level Game is the head-of-state—eventually referred to as Chief of Government (COG)—who represents their state at the bargaining table. However, the Two-Level Game weaves together Waltz’s three levels in its assumption that the COG operates simultaneously at the system level (Level I) and at the domestic state level (Level II) (Moravcsik 1993). The COG seeks to gain advantage vis-à-vis other states while placating the domestic political interests to whom his presence in office is beholden and avoiding political self-harm. The analytical necessity of investigating both levels, based upon the realistic assumption of the COG’s own need to do so, is the model’s prime innovation. The Two-Level Game allows for a more complex and nuanced analysis than previously existed.

Beyond the COG, the key link between the two levels is the concept of ratification (Moravcsik 1993; Putnam 1988). At Level II, there is a range of foreign policy outcomes acceptable to domestic constituents (the “win-set”) that are said to be ratifiable. Ratification can be a formal process (such as the constitutional requirement of a two-thirds Senate vote to approve a treaty) or an informal one based on bureaucratic, interest based, or public opinion constraints. When the win-sets of each negotiating state overlap and an understanding is reached at Level I, ratification occurs at both levels and an agreement is made. Larger win-sets make Level I agreements more likely, while smaller win-sets make successful agreement at Level I less likely. Win-sets are essential, and they are determined by three fundamental factors: (1) domestic preferences and coalitions, (2) domestic political institutions, and (3) the COG’s bargaining strategy at the systems level.

There are numerous examples of the Two-Level Game case studies, which speaks to the benefit of conducting foreign policy analysis across multiple theoretical levels (da Conceição-Heldt and Mello 2017). As a comprehensive view of the literature is well beyond the scope of this article, only a few applications will be discussed to illustrate the basics of the model in action. Putnam’s original article discussed the case of the 1978 Bonn economic summit led by the United States, Germany, and Japan. Putnam (1988, 428) stresses the importance of endogenous politicking and argued that an agreement was only possible “because a powerful minority within each government actually favored on domestic grounds the policy being demanded internationally.” These powerful minorities were able to overcome domestic opposition through international pressure to achieve ratification of the Bonn agreement, which they saw as being in their national interest. Later, a collection of Two-Level case studies called Double Edged Diplomacy broadened the scope of the model by analyzing, for example, the U.S./USSR negotiations over German reunification, the failure of the 1933 World Economic Conference, and weapons and security arrangements among North Atlantic Treaty Organization (NATO) allies (Evans, Jacobson, and Putnam 1993). The voluminous applications of the model reflect its widespread acceptance, its significant advancement of the field of linkage politics, ⁴ as well its versatility in approaching a wide variety of foreign policy issues (da Conceição-Heldt and Mello 2017). But the Two-Level Game should be further scrutinized in comparison with the frameworks from the policy process field and their applications in foreign policy analysis to explore and assess its theoretical assumptions.

Policy Process Frameworks and Foreign Policy

The policy studies field has primarily focused on domestic policy processes, and policy process frameworks have been overwhelmingly applied to explain domestic policy outcomes (Lentner 2006). In addition, policy process analysts have, until recently, mostly ignored their roots in civil defense and national security—subsystems which often overlap with foreign policy (Ripberger 2011). Regardless of their potential for enhancing widely accepted models such as the Two-Level Game and identifying causal factors that are unseen or underemphasized, they have been only marginally utilized in the field of foreign policy analysis (Hirschi and Widmer 2010). Despite this, the foreign policy applications of three such frameworks are presented here to assess their utility. There are other applications of these frameworks relevant to international relations scholarship, but these have been specifically chosen for their concentration on domestic and foreign policy linkage.

The ACF is essentially about the consequences of beliefs in policy making. It approaches foreign policy making with the aim of explaining the strategic impact of the formation and maintenance of advocacy coalitions, which are defined as groups of people (and usually their organizational affiliations) who create informal alliances on policy issues in pursuit of shared policy goals based on their deep core beliefs, which are comprised of normative and ideological assumptions (Weible and Ingold 2018). The ACF aims to connect advocacy coalitions to the potential for policy learning by actors and the occurrence of major and minor policy change based on their proximity to power. The assumption is that individuals who join advocacy coalitions are motivated primarily by their beliefs and that these coalitions will remain relatively stable over a long time period (Jenkins-Smith et al. 2017).

Jonathan Pierce (2011) used the ACF to look at the development of U.S. foreign policy toward the creation of Israel from 1922 to 1944 to identify and place important actors into advocacy coalitions and examine the ACF’s assumption of coalitional stability through time. He analyzed Congressional hearings on the question of who should have sovereignty over Palestine, which centered on the issue of whose right to self-determination mattered more: the Jews or the Palestinians. He identified three main coalitions: pro-Zion, anti-Zion, and pro-Arab. This issue was (and, indeed, remains) one with immense emotional charge, with both Jews and Palestinians seeing failure to achieve their policy goals as a threat to their survival as nations. The findings supported the ACF’s expectation of coalitional stability through time and showed that advocacy coalitions can converge to include heterogeneous deep core beliefs while still maintaining overall stability. Pierce argues that the highly contested normative nature of foreign policy making makes the ACF a useful tool in foreign policy analysis.

Another relevant application of the ACF was done by Su-Mi Lee (2015), who used the framework to observe the effects of the competition between two advocacy coalitions, the Yalta axioms and the Riga axioms, on early-Cold War U.S. foreign policy making (Lee 2015). The Yalta axioms were based on three beliefs: that the Soviet Union acted not out of ideological motivation, but its own national interest; that it was possible, and even worthwhile, for the United States to engage the USSR in diplomatic talks; and that the United States should accept the USSR as one of the great powers in the international system. The Riga axioms were based on the policy of containment but can be distinguished by two approaches. First, the militant orientation was comprised of several high-level members of the State Department who were opposed to engaging the Soviets, based on their beliefs that the Soviets were driven fundamentally by Marxist ideology, and that they were committed to world revolution and global conquest. The pragmatic orientation, developed by influential diplomats George Kennan and Charles Bohlen, reflected a middle ground between the moderate Yaltas and the more militant Rigas in their view that the main motivation of Soviet action was their psychological need to maintain internal security and ensure the survival of the Soviet state. While the question of which coalition was most influential on Cold War policy remains a controversial one, Lee (2015) argues that the Yalta axioms were initially dominant (from 1944 to 1947) until the Riga axioms became dominant as the Cold War more fully materialized.

Christopher Hirschi and Thomas Widmer (2010) used the ACF to investigate the differences between two cases in the foreign policy history of Switzerland: Swiss defiance of the international community in their refusal to sanction the regime of apartheid South Africa from 1968 until the regime’s collapse in 1994, and their policy of implementing economic sanctions (for the first time in Swiss history) against Saddam Hussein’s Iraq after its invasion of Kuwait in 1990. Swiss foreign policy has traditionally been defined by isolationism, noninterference, and liberal trade policies, and the deep core beliefs of the Swiss were defined by their economic policy as well as their respect for human rights and democratic traditions. The authors argue that Switzerland’s policy toward South Africa was the nation’s most important postwar foreign policy issue given their extensive economic ties, and that its foreign policy toward South Africa represented a deeply embedded subsystem in the Swiss national political system. This subsystem was dominated by a stable coalition which predisposed the Swiss state toward policy stasis. However, despite the same beliefs being acted upon by mostly the same actors, in the Iraqi case, there was only a nascent subsystem and the external shock of the swiftness and unanimity with which the international community moved against Iraq was enough to galvanize the Swiss to apply sanctions.

The use of the MSF shows the messiness and ambiguity inherent to foreign policy making. The MSF seeks to answer the following question: “What makes people in and around government attend, at any given time, to some subjects and not others?” (Smith and Larimer 2017, 108–09). It focuses on the “coupling” of three streams: the problem stream (where problems are identified), the policy stream (where solutions are proposed), and the political stream (where policy makers create actual policy). When this coupling occurs, a policy window is opened where an actor enjoys widespread attention given to their issue, making it more likely that they will achieve their policy goals. Policy entrepreneurs (advocates who work to promote particular problems and/or solutions) are the key actors of the MSF and drive the policy process. Their successes and failures largely determine policy outputs (Herweg, Zahariadis, and Zohlnhöfer 2017).

Nikolaos Zahariadis (2015) applied the MSF to explain the role of emotion in a salient foreign policy issue in a case study on Greece’s efforts to block international recognition of a burgeoning state, the former Yugoslav Republic of Macedonia, from 1990 to 1993. Given the prominent Slavic minority population in the northern Greek region also called Macedonia, Greek leaders feared the recognition of a separate Macedonian state would undermine their sovereignty over “Greek Macedonia.” Greek Prime Minister Constantine Mitsotakis strategically agitated the short-term emotional mood of the Greek public (the “national mood”) into a deeper and more prolonged opposition of recognizing Macedonian statehood using fear and negative emotional endowment. He successfully used emotional persuasion to impose his preferred policy solution onto the problem and political streams. The irony is that Mitsotakis’s short-term success in arousing the national mood prevented him from reversing the course of his steadfast opposition over the long run. He persisted in this strategy despite the fact that the costs of blocking recognition increasingly outweighed the diminishing benefits of doing so. Policy change did not occur, despite “the game being lost.”

Michael J. Mazarr (2007) explained the buildup to the U.S. decision to go to war against Iraq in 2003 by employing the MSF concepts of agenda and attention. Mazarr shows how Iraq had long been on the “radar screen” of the U.S. government, particularly among high-level members of what became the George W. Bush administration in 2001. The dominant policy entrepreneurs all held an ideological predisposition to remove Saddam Hussein from power. The focusing event here was the 9/11 terrorist attacks, and Bush administration actors’ determination to find a link between Hussein’s regime and the perpetrators of 9/11 essentially fits the expectations of the MSF: that actors wait for a problem to come along to open a policy window that allows them to successfully attach their favored solution to. When the policy window opened after 9/11, Bush administration officials immediately began discussing the Iraq problem and of course eventually seized the opportunity to conduct the invasion.

Rick Travis and Nikolaos Zahariadis (2002) used the MSF to examine the intersection of the domestic political agenda and U.S. foreign aid policy. They note that this is done in specific response to prior work in foreign aid decision making being focused on external, systemic-level events. They wanted to show that aid decisions are also made as a result of the political ideologies of the executive branch and Senate interacting with preexisting funding levels. They distinguished two types of aid: economic assistance (long-term developmental support, for example, for medical care and infrastructure) and economic support funds (or ESF) that are used tactically to promote short-term economic goals and political stability derived from realist geopolitical concerns. Using a quantitative approach, they found that, for economic assistance, economic aid decisions were not affected by whether the White House and Senate were controlled by Democrats or Republicans, while for ESF decisions, a Democratic-led Senate tended to fight against high levels of funding. For our purposes, it is useful to make clear that while most observers had previously focused on the problem and policy streams, Travis and Zahariadis (2002) showed that the domestic political stream is an important factor in some foreign aid decisions.

Finally, the application of PET to foreign policy problems allows observers to explore continuous periods of policy stasis as well as periods of intense, rapid policy change (Baumgartner, Jones, and Mortensen 2017). PET is concerned with macro-level political systems and assumes that such large systems have a tendency toward inertia. Policy change is usually incremental, but when enough momentum is generated, the system can be overwhelmed to the point of causing major policy changes (or punctuations). Like the MSF, PET uses the concepts of agenda and attention as measurement tools. A study relevant to foreign policy conducted by Heather Larsen-Price (2012) employed this concept by looking at the use of presidential communication “policy tools” across several decades. Although the quantity of PET applications relevant to this study is not as robust as the ACF and MSF, it does contribute to our understanding of the way problems are perceived through time and how such perceptions often impede policy learning and policy change.

This contribution has primarily been made within the context of enduring rivalries, or “long-standing militarized competitions between the same pairs of states” marked by “the persistent, fundamental, and long-term incompatibility of goals between two states” (Ishtiaq and Ebert 2015). Enduring rivalries arise from states’ unfavorable psychological perceptions of one another and result in potential or actual violent clashes over a long period of time (Cioffi-Rivella 1998; Ishtiaq and Ebert 2015). PET shows that the characteristics of enduring rivalries are mostly determined at the outset of competition during a quick “lock-in” phase and that little change occurs beyond this phase (Diehl and Hensel 1998). Domestic politics are included in the discussion of enduring rivalries because such rivalries will persist despite the numerous domestic political changes that inevitably occur over time (Cioffi-Rivella 1998). Enduring rivalries are framed within the PET jargon: they are said to be in stable equilibrium until they are uprooted by factors within both the domestic and systemic levels (Ishtiaq and Ebert 2015). This linkage makes it relevant to this discussion.

Two notable contemporary examples of enduring rivalries have been examined through the lens of PET. Ahmad Ishtiaq and Hannes Ebert (2015) studied the rivalry between India and Pakistan, two states whose foreign policy goals have been in fundamental opposition since 1947 and who have engaged in four wars and numerous disputes. They cited the election of new governments in each state in 2013 and 2014, whose leaders initially seemed agreeable to friendlier relations, to examine the possibility for rapprochement. However, they find “rivalry de-escalation and eventual termination” were quickly prohibited by “structural impediments caused by domestic politics” (Ishtiaq and Ebert 2015, 56). Chief among these impediments has been the dominance of Pakistani foreign policy by its military class, which is deeply suspicious of India and highly resistant to change. This has prevented democratic consolidation from becoming a truly influential variable in the cessation of the rivalry, which PET argues is one of the drivers of policy change in an enduring rivalry such as this.

Another seemingly intractable rivalry of note is the decades-long hostility between Israel and the Arab states. Ben D. Mor (2004) uses PET in his examination of early Israeli strategic beliefs related to national security. He finds that after the creation of the Jewish state in 1948, Jews felt very early on that the potential for lasting peace had an illusory quality that obscured the security threats all around them, and that they could not afford to rely on the international community but only on themselves for their safety and survival. Mor (2004) also finds that this self-fulfilling belief became entrenched in a “vicious cycle” of conflict with the Arab states, which virtually drowned out accommodationist political views and empowered security hardliners. Although it only approaches one side of the rivalry, the study confirms the expectations of PET in the context of enduring rivalries: Israeli strategic beliefs, especially their extremely defensive nature, quickly locked-in and became entrenched to such a degree that they have remained a significant barrier to rivalry termination to this day.

The Rational Actor Model

To put the Two-Level Game in better theoretical perspective, one must recognize caveats to its success. While it accounts for greater complexity inherent in foreign policy modeling than previously existed, mainly by refuting the notion of the state as a unified actor, its simplifying assumptions limit its validity. Its reliance on the rational actor model in its construction of the COG is problematic (Moravcsik 1993). The COG has clear goals and clear means of achieving them, and the COG’s rationality is woven across levels. As Putnam notes, “Moves that are rational at one (level) may be impolitic at another,” and the assumption is apparent that the COG knows which moves are impolitic and will know how to avoid making them (Putnam 1988, 434). Yet, each of the policy process frameworks shown here assume bounded rationality and thus account for even greater complexity in how decisions are made in foreign policy contexts. They emphasize the cognitive and informational limitations foreign policy makers face (Baumgartner, Jones, and Mortensen 2017, 64–67; Herweg, Zahariadis, and Zohlnhöfer 2017, 18–20; Jenkins-Smith et al. 2017, 139–40). In addition, the policy process frameworks assume that foreign policy can arise out of divergent causes, beyond the friction between the COG and the relevant domestic interest groups.

If the ACF were to utilize the individual foreign policy actor in its analysis more fully, it would place them into an ideological context and it would be less concerned with political self-interest as a motivating factor. However, the ACF does not really have an individual role analogous to the COG: it is more easily comparable to the domestic Level II. It approaches domestic concerns with a different lens, as it construes domestic politics as less of pluralistic mish-mash of pressure groups that simply seek to maximize their self-interest. Actors are instead motivated by their beliefs, which are often rife with passion and emotion (Nohrstedt 2008). ⁵ As they form coalitions, it is quite possible that behavior motivated by deeply held ideological beliefs can result in faulty and illogical outcomes, especially when actors holding those beliefs are challenged in an adversarial context (as the political arena often is) and become aggressive or defensive. ⁶ Coalitions then become tribes that engage in motivated reasoning (Weible and Sabatier 2017). Contrast this with the Two-Level Game: there you will not find a discussion on the beliefs, passions, and emotions involved in policy making that get in the way of clear-headed thinking. Despite his or her centrality, the COG is a type of highly visible phlegmatic drone who mostly aims to reach international agreements to please his or her domestic base, a group of citizens who are equally unburdened by emotion. Some argue, however, that in its reliance on the individual as the building block unit of analysis of the framework, the ACF does not distance itself far enough from the rational actor model, but in any case it is certainly a departure from the Two-Level Game (Jorgensen 2017).

Emotion and passion preclude rational assessment: as Morgenthau (1963, 7) noted, “the need to marshal popular emotions to the support of foreign policy cannot fail to impair the rationality of foreign policy itself.” Zahariadis’s (2015) work on Greek foreign policy toward the nascent Macedonian republic is a persuasive confirmation of this truism and exposes deeper faults in the Two-Level Game. He notes that the MSF is the only framework to pay attention to concept of national mood. The effect of emotion and fear on the rationality of Greek foreign policy makers was evident in their failure to take corrective action (as the rational actor model expects that they should) once a policy becomes unsustainable and losses begin to mount. Zahariadis (2015, 477) notes further that “the process of seeking information that supports the national mood eclipses the process of calculating costs or benefits.” It is as if, instead of simply making the wrong calculation, they were not making any formal cost-benefit calculations at all. They relied upon simple emotional intuition, which can be a powerful decision-making mechanism, yet one that often leads people to make harmful and regrettable choices.

As a macro-level theory, PET assumes that governments are hindered by cognitive deficiencies, just as individuals are. Governmental organizations are potentially able to transcend the limitations of the individual by their sheer size and engage in parallel processing, or the consideration of a multitude of problems at a given time. If parallel processing fails, serial processing (where only a very small number of high-profile problems are broached, fought over, and decided at the same time) takes place (Baumgartner, Jones, and Mortensen 2017). This is essentially a problem of attention scarcity. By measuring presidential communication tools, Heather Larsen-Price (2012) argued that, since presidents are constrained by narrow information inputs and are only able to process some of the information they receive, they cannot possibly be considered rational decision makers since they do not have access to complete information. Larsen-Price argues instead that presidents are cybernetic decision makers. In addition, attention given to one problem necessarily decreases the attention given to another, which compounds the problem of being a boundedly rational actor. We should extend this attention deficit problem by considering not only the president but also its effects on the entire executive branch, which is mostly responsible for making foreign policy in the United States.

The Dynamics of Agenda Setting

Another weakness of the Two-Level Game is its static, cross-sectional nature. Although some studies have attempted to add a temporal element into the model, this has yet to be properly developed (da Conceição-Heldt and Mello 2017). The Two-Level Game is concerned with specific episodes of bargaining at particular times and is successful in describing why states succeed or fail to come to an agreement. However, it probably will not tell us very much about why they chose to arrive at the bargaining table in the first place, nor will it say a lot about states that chose not to come to the bargaining table at all. This is a significant blind spot of the model, one which the policy process frameworks supplement. These frameworks can see things that develop over a longer period of time that the Two-Level Game is likely to miss.

Each of the policy process frameworks presented here contains a temporal element that enhances the analysis of foreign policy outcomes. They can each be applied to show why governments look at certain problems and solutions (and not others) over time. The ACF asks, “how stable are coalitions over time and which ones have the most influence on the policy process?” For the MSF the question is “do policy entrepreneurs go around selling policy solutions, waiting for the right moment to get their solutions adopted as actual policy?” PET asks us to consider “what is the pace at which policy changes, and how does the attention of government become fixed onto a single issue?” It is possible that the importance of dynamic agenda setting has been overstated by these frameworks (Smith and Larimer 2017, 113–14). Regardless, many of the applications were specifically concerned with foreign policy problems as they were perceived in over many decades, and even the entire lifespan of nation-states in the case of the PET literature on Pakistan, India, and Israel.

In the ACF, it is explicitly suggested that researchers should measure the role of advocacy coalitions over an extended period of time (Weible and Sabatier 2017). Consider Hirschi and Widmer’s (2010) placement of Swiss sanctions against Iraq into the wider context of Swiss foreign policy fundamentals since World War II. Much analytic depth and power would be lost if we chose only to look at that one period in 1991 and construed that as a single bargaining episode. Similarly, the MSF depicts the policy entrepreneur as waiting for the right time to pounce. As Mazarr (2007) showed, the roots of the decision to invade Iraq in 2003 can be found in a 1992 Defense Policy Guidance paper that outlined a post-Cold War vision of expanding American dominance globally that included the removal of Saddam Hussein. Key Bush administration players were either involved in writing paper or became proponents of its axioms upon reading it, most notably Dick Cheney. What would we lose if we chose just to focus on, say, the failed American negotiations with the United Nations in the buildup to the invasion? The PET applications show that states will sometimes be too distracted to come to the bargaining table, or else their perceptions of their counterpart will be colored by such hostility that it is too costly, psychologically and politically, to attempt to enter into negotiations.

Strategic Behavior

The treatment of the strategic behavior of relevant political actors, including domestic pressure groups as well as the foreign policy maker, is an important factor in all of the frameworks discussed here. The Two-Level Game shows statesmen to be the pivot points of foreign policy, and the COGs are the main strategic players. Putnam (1988) notes that shrewd COGs can, through clever maneuvering, trigger moves at other levels to their advantage. The COG operates within domestic constraints and seeks to realize personal goals and build domestic support insofar as possible. The COG is cognizant of which domestic groups are likely to be most important in ratification (Moravcsik 1993). If a COG is especially constrained by Level II political needs, he can use his small win-set as a way to strengthen his bargaining stance at Level I (Putnam 1988). Although Putnam arguably overemphasizes the role of domestic groups, he shows that they also have the opportunity to advance their interests through transnational cooperation with domestic groups in the other side’s Level II (Moravcsik 1993). In addition, the COG may increase the chances of Level I agreement by helping his or her counterpart save face politically and achieve Level II ratification with their domestic base. One salient example is the humble diplomacy of George H. W. Bush and James Baker during negotiations with the crumbling USSR over German reunification (Sebenius 2013).

In comparison, strategy in the policy process frameworks is a far messier practice and develops in a nonlinear fashion (Mazarr 2007; Smith and Larimer 2017, 99; Weible and Sabatier 2017). In some ways, the domestic constraints utilized in all three frameworks are analogous to those in the Two-Level Game. They all claim that, in short, creating significant policy change is difficult, and there are many impediments. One of the core assumptions of the ACF is that strategy is determined by the competitive interaction of advocacy coalitions and that the presence of a dominant coalition will tend to bend policy outcomes to their will, while minority coalitions attempt to gain greater attention and eventual adoption of their policy proposals. Mostly, the position of the dominant coalition and the solidity of their core beliefs will prohibit major policy change, with only relatively marginal change. ACF actors tend to employ “devil shifting” where the strength and maliciousness of their opponents is emphasized and exaggerated (Weible and Sabatier 2017).

Su-Mi Lee’s (2015) exploration of American policy makers’ perceptions of the USSR during the early-Cold War period offers a good illustration. The beliefs of each coalition did not simply reflect their worldview but a strategic outlook. Mostly, the beliefs of the most influential actors did not change, but it was their positions relative to political power that changed. The proximity of each coalition to the presidency, and to the tremendous legal authority which it possesses in foreign policy making, was a strong determining factor in policy output. Initially, the Yalta axioms were in the dominant position, but when the Riga axioms became dominant, they were able to push through a more belligerent stance. They used their leadership positions to garner widespread support for aggressive policies like the Truman Doctrine, the invasion of Korea, and the National Security Council policy paper, NSC 68. Yet, the ACF’s conceptualization of individual actors is problematic, despite its alleged reliance on them. It is less able to show individuals acting out of their political self-interest, whereas the Two-Level Game is guided by the notion of political self-interest and self-preservation of the COG. Unless they are identifiably part of an advocacy coalition, the ACF is unlikely to be concerned with their actions.

The MSF is similar to the Two-Level Game in that it is centered on the actions of an individual. The difference is that while the COG is assumed to be the head of state, the main actor in the MSF, the policy entrepreneur, can either operate inside or outside of government. They are less defined by their formal position than they are by their pursuit of a “pet project.” They must be extremely determined in this pursuit, and they must also be skilled in coupling, behaving somewhat similarly to a businessman seeking a buyer (Herweg, Zahariadis, and Zohlnhöfer 2017). They must also be lucky: some event happens that presents an opportunity for the entrepreneur to achieve his goal. But a good policy entrepreneur can increase the chances of coupling. Of the applications mentioned here, Zahariadis (2015) made good use of Mitsotakis’s role in Greek foreign policy, as did Mazarr (2007) in his description of the anti-Saddam entrepreneurs in the Bush administration. One gets a vibrant look at the way these actors operate and how fallible and fiasco-prone they really can be. That policy entrepreneurs look to attach their solutions to problems is really not an ideal process, and this does not necessarily make wise foreign policy (Zahariadis 2015). In addition, the MSF allows varying levels of import in its modeling of domestic politics: Zahariadis (2015) showed how foreign policy was driven by domestic sentiment, while Mazarr (2007) showed that the decision to invade Iraq was essentially immune from public opinion. The Two-Level Game assumes that domestic political groups are influential in foreign policy outcomes, though this is not always the case. One suspects that Putnam indeed wishes this were the case, based on his image of the COG being held somewhat captive by domestic factors.

The built in status-quo bias in PET makes it a less fertile ground for analyzing strategic behavior. However, during the lock-in phase of enduring rivalries, individual strategic actors can play an important role. It was during the lock-in phase of Israel’s tense standoff with the Arab states that Mor effectively showed how Israel’s first national leader, David Ben-Gurion, was able to almost unilaterally shape Israel’s strategic national security perceptions (Mor 2004). His actions are somewhat analogous to the policy entrepreneur, and he was obviously very successful in achieving his goals. After lock-in, Israel’s hard-line security actors were able to effectively and continually use the logic of “an actor who does x is a Y; Y actors are also expected to do z” to convince their domestic base that their Arab neighbors could not be trusted, and therefore Israel’s security was constantly in danger (Mor 2004). However, the strategic actions of the state as a whole, as well as the structural-level impediments to policy change, are more relevant to PET than the actions of individuals. As a macro-level framework, one does not get the sense that the inclusion of such vivid strategic actors is a recurring theme in PET. It therefore loses some descriptive power in foreign policy, since it loses sight of the individual-level decision maker more than it should.

Conclusion

A fundamental maxim to foreign policy analysis is that analysts, whether they are academics or practicing professionals, must be cognizant of the influence of theory. Graham Allison’s (1999) classic work Essence of Decision showed that a single episode could be explained in numerous ways, and the origin of these different causal outcomes is the theoretical models employed and their assumptions, which place emphasis on various contrasting factors. In short, different questions produce different answers (Allison and Zelikow 1999, 379–88). Returning to the example of the Iranian nuclear negotiations, we might now have a better view of the types of approaches one might take in pursuit of causal explanation of the deal based on the frameworks presented here. How did Presidents Obama and Rouhani operate in the face of domestic constraints and factionalism, as well as with compounded difficulty of their counterpart’s also having to do so, in order to increase the chances of ratification? How did American beliefs toward the Iranians form, and how has the ideological struggle between liberal engagement and hard-line realism impacted the United States’s foreign policy toward Iran? Was President Obama merely a policy entrepreneur searching for a problem to attach his preferred policy solution of engaging the Muslim world to? How were the two sides able to penetrate decades of antagonism and psychological hostility to transcend decades of rivalry and policy stasis? Each of these questions prioritizes a certain set of causal factors and rejects others, and this fact will lead the analyst in different directions.

The study of international relations, including the subfield of foreign policy analysis, has suffered from the compartmentalization of research traditions. Albert Hirschman (1981) noted the potential for enhanced theoretical development and refinement that would result from “trespassing,” or the deliberate choice of researchers to borrow from the innovations and creativity of other models outside their dominant purview. This is not simply an academic concern: the original perspectives likely to arise from this process may offer scholars more effective ways to look at problems of real global importance (Katzenstein and Sil 2008). The need for states to interact with each other will never cease, and the seriousness of the issues which arise from this interaction should cause scholars to aim for their elucidation as well as their resolution. The potential for cross-fertilization of domestic politics and foreign policy in the policy sciences clearly exists and should be encouraged (Lentner 2006; Ripberger 2011).

The policy process frameworks presented here have much to offer the subfield of domestic–foreign policy linkage, despite their underutilization thus far. When the assumptions of the Two-Level Game clash against those of the policy process frameworks, it is apparent that the policy process frameworks offer a more realistic, if messier, depiction of human behavior, given their assumption of boundedly rational individuals and organizations. They also allow for a more dynamic and expansive view of events across time and for greater complexity in exploring the strategic behavior of actors relevant to the foreign policy process. This does not nullify the Two-Level Game as a useful model of foreign policy analysis since it has proven to be very influential, even inspirational. It is effective when it is used as designed, but perhaps it has been employed in ways not originally intended (Gourevitch 2002). Given this, one must recognize the limitations of the Two-Level Game and seek to redress its simplifications and shortcomings.

The ACF, MSF, and PET allow for deeper exploration into the nexus of causes of state behavior internationally, which Waltz recognized as crucial for creating insightful analytical models. Even when breakthroughs connect various levels of analysis, the search for causal factors need not halt. Although the modeling of foreign policy in these frameworks is a positive step in the inclusion of deeper complexity and dynamism, the frameworks are not without fault. Bounded rationality is an increasingly accepted assumption across several disciplines, but policy process theorists have not yet integrated bounded rationality into the policy-making context in a fully systematic way (Smith and Larimer 2017, 187–89). Further application of these frameworks on the domestic–foreign interface is necessary to refine their shortcomings and to more fully assess their contribution to the field at large.