Level I: Foreign Policy Decision-Making
This week drops below the level of the state-as-actor. The main claim is that to explain real decisions, such as why the missiles were deployed, why the quarantine was chosen, why Crimea was annexed, or why the balloon was sent, we must open the black box and look at the organizations that implement policy and the leaders who choose it. The level sits mostly at I (individual leaders and their beliefs) but spills into II (domestic institutions and bureaucratic politics).
Reading list
- Allison. 1969. “Conceptual Models and the Cuban Missile Crisis.” American Political Science Review.
- Halperin, Clapp, and Kanter. 2006. Bureaucratic Politics and Foreign Policy, Ch. 3.
- Saunders. 2009. “Transformative Choices: Leaders and the Origins of Intervention Strategy.” International Security.
- McFaul. 2020. “Putin, Putinism, and the Domestic Determinants of Russian Foreign Policy.” International Security.
- Jost. 2022. “Leaders, Bureaucracy, and Miscalculation in International Crises.” Working Paper.
- Application: Jost. 2023. “The Bad Advice Plaguing Beijing’s Foreign Policy.” Foreign Affairs.
Allison (1969): Conceptual Models & the Cuban Missile Crisis
American Political Science Review · three models of decision-making
Allison (and his book Essence of Decision) offers three lenses. Model I, the Rational Actor Model, treats the state as a unitary, rational chooser with a single set of objectives that perceives options and maximizes utility; it reads the Soviet deployment as a rational move to improve the nuclear balance or to bargain, and Kennedy’s blockade as the option that maximizes security while minimizing the risk of war. Model II, the Organizational Process Model, sees government not as one actor but as large organizations running on routines and standard operating procedures (SOPs); it explains the delay in finding the missiles (CIA–Air Force disputes over U-2 flights), why the Soviets used detectable launchers (routine deployment practice), and why the Air Force reached for pre-set strike plans. Model III, the Bureaucratic Politics Model, makes action the outcome of bargaining among players who each have their own interests and power, “where you stand depends on where you sit”; it explains the fight inside the Kennedy administration (military chiefs wanting air strikes, State pushing diplomacy, the President balancing them) and why the final quarantine was a compromise. Model I is simple, intuitive, and widely used; it fits cost–benefit analysis and much of IR theory, and works where state action looks coherent and goal-directed, but it ignores internal routines and political bargaining, so it is weak for complex, multi-actor decisions. Model II usefully depersonalizes the state and exposes inertia, routine, and implementation gaps; its contribution is showing that policy is often the output of complex organizational machinery rather than a unified rational actor. But it leans too deterministic, underrating leaders’ room to override SOPs; it under-explains how cross-organizational conflict actually gets bargained into compromise; it is too static, missing how organizations innovate and learn in a crisis; its empirical reach is limited where decisions are too fast for SOPs to bite; and it brackets the power and politics that shape the routines themselves, which is exactly why Model III is needed. Model III captures real politics, power struggles, and the weight of individuals, and fits pluralistic systems with competitive elites. Its weaknesses: it is largely descriptive with little predictive power (it tells us the outcome was a compromise but not who will win); it can underrate the president’s decisive coordinating power and overstate bureaucratic fragmentation; it neglects institutional, constitutional, and international-structural constraints; its scope is narrow (strong on crises, weaker on routine or non-security policy, and even in crises leaders sometimes bypass the bargaining); and it is hard to test, since tracing the internal interactions demands inside material that is rarely available. He cites Morgenthau explaining WWI through the balance of power, Stanley Hoffmann reading U.S. policy as “imaginative reconstruction,” and Thomas Schelling building deterrence theory on rational calculation (the discussion grows out of “How do analysts account for the coming of the First World War?”). In my experience the claim is often true: analysts and policymakers reach for rationalist explanations because they are simple and tidy, even though real decisions are far messier.
He faults Model I for being overly simplistic, for ignoring organizational and political realities, and for failing to explain inefficiencies, errors, and internal conflict (the model assumes important events have important, purposive causes, which he says must be balanced). I think the criticism is fair: Model I is useful but incomplete. Most real decisions involve bureaucratic inertia and political bargaining; governments are not single calculators but complex institutions. Even so, the rational model still earns its keep for big-picture strategy. On the bureaucratic-politics view, an official’s policy position reflects their bureaucratic role and institutional interest: a general favors military solutions, a diplomat favors negotiation, a Treasury official favors economic tools. I largely agree, because your post shapes the information you see and what you value, but individuals still matter, through their personal beliefs and experiences. As for the models, Allison presents them as complementary lenses, each illuminating a different part of the decision; the Cuban crisis makes most sense when you run all three together, rational (the broad strategic goals), organizational (why certain options looked feasible), and bureaucratic-political (why the compromise was chosen). The piece reaches beyond Cuba too: Model III explains Yamamoto’s decision at Pearl Harbor despite his strategic doubts, Model II explains Soviet weapons procurement driven by organizational interest rather than pure strategy, and Vietnam appears briefly through the logic of surrender.
Halperin, Clapp & Kanter (2006): Bureaucratic Politics & Foreign Policy
Ch. 3 · organizational “essences” and turf
The chapter makes government look like a “makeshift theatre troupe” (草台班子): huge, serious questions, which weapons to build, whether to intervene abroad, often emerge from organizational turf wars rather than pure national-interest calculation. Three anecdotes stuck with me: the Air Force resisting ICBMs in the 1950s because bombers were its “essence” and it had to be forced to accept missiles; the Navy dismissing Polaris submarines as a “national program, not a Navy program,” because subs didn’t fit its carrier-centered identity; and the Army resisting Kennedy’s push to make the Green Berets central, because special forces clashed with its identity as a conventional ground-combat force. The lesson is how much policy flows from protecting organizational identity rather than rational national strategy. Army: essence is ground combat with conventional divisions; it is wary of special forces, air defense, or nuclear roles unless they protect that primacy. Navy: essence is sea control via ships; it is internally split (carrier “brown shoes,” surface “black shoes,” submariners), with carriers dominant and often resisting subs or land-based air. Air Force: essence is delivering weapons by air, originally nuclear bombing; it favored bombers over missiles, resisted transport/airlift, and later shifted toward conventional precision strike. CIA: divided among three essences, clandestine collection, covert operations, and intelligence analysis, each faction defending its own role. Foreign Service: essence is diplomacy, reporting, representation, and negotiation; it resists being pulled into operational programs outside classic diplomacy. The patterns: every organization protects its essence and resists roles that dilute it, fights to control budgets, missions, and promotion paths, and prefers policies that expand its influence and justify its existence. The chapter only touches Congress (it returns later), but the logic extends naturally. Legislators’ interests run to protecting constituencies, securing pork-barrel defense contracts, signaling responsiveness to voters, and partisan positioning. Compared with executive officials, their interests are even more tightly bound to domestic politics, reelection and party battles, so in foreign policy they often frame decisions around local jobs, defense contracts, or electoral cycles. Pork-barrel and partisan maneuvering can outweigh the national interest.
Generalizable: organizations everywhere have “essences,” defend turf, fight over roles and budgets, and see national interest through an organizational lens, dynamics that travel to other states. U.S.-specific: the exact services (Army, Navy, Air Force, CIA, Foreign Service), the role of Congress, and the degree of transparency. In other contexts, especially China, secrecy and patronage networks (红二代/红三代, the “red” second and third generations) add layers of family, faction, and loyalty that shape military and bureaucratic decisions differently. In China, research shows the Central Military Commission and Xi’s appointment of loyalists play a similar essence-protection role, but corruption, family ties, and factional networks add a dimension beyond organizational identity (see Tai Ming Cheung on PLA modernization; Andrew Scobell on PLA politics). Beyond the chapter, USAID is a good case: formally developmental, yet entangled with U.S. strategic and political goals, so conflicts arise when local missions clash with Washington’s priorities. Other candidates: Homeland Security, NGOs that depend on U.S. funding but carry local agendas, and international organizations like the UN or World Bank that must balance member states’ competing bureaucracies. To test whether standpoint really follows seat, I would combine research designs: process tracing memos and speeches before and after an official changes posts (does a general’s view shift when he becomes Secretary of State?); comparative case studies tracking how different agencies responded to the same crisis (CIA vs. Air Force during the Cuban Missile Crisis); elite surveys and interviews asking officials how their organizational role shaped their position; and content analysis coding whether statements map onto institutional turf. Together these could reveal systematic patterns where individuals’ views shift to match organizational identity.
Saunders (2009): Transformative Choices
International Security · leaders and the origins of intervention strategy
Most great-power interventions in smaller states are “wars of choice.” They do not spring from a direct, existential threat, so leaders genuinely choose where and how to respond to indirect threats. Theories built on slow-changing or stable factors (the structure of the international system, regime type) can’t account for how a state’s intervention choices vary over time; shifting the focus to individual leaders captures that variation. Leaders matter because they hold different causal beliefs about where threats come from, internally, from a target state’s domestic institutions, or externally, from its foreign-policy behavior, and those beliefs decide whether they pick transformative (nation-building) or nontransformative (limited, conventional) intervention. This sits inside a wider move to restore agents after long structural phases: Byman and Pollack’s call to “bring the statesman back in” (International Security, 2001) parallels comparative politics’ “bringing the state back in.” The focus is especially salient in presidential systems like the U.S., where executive power is concentrated and discretion is wide; parliamentary leaders may be more constrained by coalitions. Saunders’ concepts are meant to generalize, but the U.S. is a useful laboratory, rich archives and frequent, well-documented interventions, and it sets a deliberately hard test for a leaders-matter claim. A transformative strategy aims to reach deep into and reshape the target state’s domestic institutions, political, economic, social, through nation-building, counterinsurgency with civic action, or institutional reform. A nontransformative strategy resolves a conflict or rolls back aggression without trying to change those institutions: repelling an invasion, enforcing a ceasefire, a limited humanitarian intervention. Saunders argues the beliefs are pre-existing causal beliefs about the origin of threats, formed before a leader takes office: “externally focused” leaders locate threats in other states’ foreign policies, while “internally focused” leaders trace them to the target’s domestic order and so favor transformation. My worry is that the line can blur in practice, but she treats the distinction as ideal-typical, a heuristic, and codes the intended strategy at the outset of intervention rather than the messy reality that follows. She defends the dichotomy for analytic clarity: actively re-making institutions is categorically different from conventional coercion that doesn’t aim to remake them. To isolate leaders’ beliefs she holds the country (the U.S.) and the system (the Cold War) constant and compares two presidents who both agreed Vietnam merited intervention but read the threat differently, same party, inherited advisors, stressed continuity, so divergent strategies most likely flow from their own beliefs. She calls this a hard test because we usually expect leaders to matter less in democracies and under the Cold War’s strong “threat consensus”; if they still chose differently, the leaders-matter claim is strengthened. To avoid crisis-time rhetoric she codes beliefs from the pre-presidential period, speeches, writings, congressional records, policy positions on foreign aid and counterinsurgency, and early policy investments like budget priorities and bureaucratic reforms. Replicating this is hard: leaders may say one thing and do another, memoirs are unreliable and self-serving, and conflicting accounts must be triangulated against archives. You would need reliable historical records (archives, declassified documents), careful content analysis, and a clear coding scheme for classifying beliefs as internally or externally focused.
She considers structural/material conditions (changing circumstances in Vietnam), bureaucratic/organizational preferences (the military’s dislike of nation-building), and domestic political competition (electoral pressure). She debunks them by showing Kennedy and Johnson chose differently under similar conditions, that both at times overruled their advisors and bureaucracies, and that electoral pressure weighed on both yet produced different strategies. Other alternatives deserve a look, media and public opinion, alliance pressures from NATO or regional partners, economic interests like oil and trade, plus groupthink, and she doesn’t foreground them, though her claim is narrower: leaders’ causal beliefs systematically shape whether strategies are transformative. On quality, the case is strong. The design is hard: the country and system stay constant, leaders vary, beliefs are measured before action, and the observable implications are coded clearly. The evidence also draws on presidential archives, memos, and speeches, while pre-presidential coding reduces circular reasoning. The limits: it leans on interpretation of historical materials, which is always somewhat subjective, and Vietnam is a single case, broader comparative work would strengthen the theory. Beliefs may not drive strategy when domestic constraints or coalition politics sharply narrow the options, when time pressure or an intelligence shock compresses the choice set, or when bureaucratic SOPs dominate implementation, a Model II/III world, to borrow from Allison. And other causal beliefs could layer onto Saunders’ framework: beliefs about military effectiveness (counterinsurgency vs. conventional force), about international law and norms (whether the UN or international community legitimizes intervention), about the feasibility of nation-building and state capacity, and about domestic audience costs or alliance credibility. These would help predict how far a leader will push once they have chosen a transformative or nontransformative path.
McFaul (2020): Putin, Putinism & Russian Foreign Policy
International Security · the domestic determinants of Russian foreign policy
He sets up two common structural stories. First, great-power inevitability: Russia’s clash with the U.S. is just “what great powers do” as polarity shifts, with rising and established powers bound to collide. Second, historical/cultural continuity: Russia has always behaved this way, tsars, communists, KGB men alike, so today’s confrontation is simply Russia “being Russia.” His reaction is that both leave no room for leaders and ideas, which lets him offer an alternative centered on Putin and “Putinism.” The framing is a bit straw-man-ish in that it flattens the structural accounts to set up his own, but the two stories are real positions in the debate, and his point, that they can’t explain change driven by a particular leader, is a fair one. His alternative is that relations soured mainly because of Putin’s ideas, “Putinism”, and the domestic politics around him. Borrowing Saunders’ language, McFaul gives Putin a few core causal beliefs: that the U.S. orchestrates democratic uprisings near Russia and in the Middle East, which threaten Russia and orthodox/civilizational values; and that international politics is fundamentally ideological, so Russia should back illiberal forces and resist “liberal projects.” These beliefs translate into policy: in Ukraine (2014) he acted to crush a revolution he framed as a Western-backed “coup,” despite a weak material payoff; in Syria (2015) he intervened when Assad’s fall looked likely, to block another regime change. McFaul stresses these choices often did not maximize Russia’s material security or power but did serve Putin’s ideological agenda and regime-security narrative. McFaul lists predictions that diverge from structural theories. We should observe Russian behavior driven by Putin and his ideas rather than by power balances or “Russian tradition”; a focus on the internal politics of other states, supporting illiberal leaders and movements, opposing liberal change; and enduring U.S.–Russia tension as long as the U.S. is the leading liberal power and Putin rules. His argument would be undercut if Russia behaved like any great power regardless of leader, or like past Russian rulers, or like a captive of the siloviki and oligarchs (the leader being irrelevant); if it were indifferent to other states’ regime type or ideology; or if Russian policy promoted liberal democracy or deep cooperation with the U.S. without a change of leadership.
McFaul argues that annexing Crimea did not actually advance Russia’s security, power, or economic interests, it brought sanctions, NATO reassurance, and costs, so the driver must have been ideological and regime security: stopping a “Western-backed” revolution next door and rallying domestic legitimacy. I’m not convinced ideology is the core. Deterring NATO’s eastward moves and securing a great-power position can be sufficient security motives. Russia’s earlier restraint in Ukrainian crises doesn’t prove Putin could have stayed his hand in 2014: conditions had changed, with Ukraine’s pro-Western, pro-EU orientation strengthening while Russia’s ability to mobilize long-term power weakened amid economic stagnation and energy dependence, so from a realist view it was rational to act sooner rather than later, and to signal a willingness to use force. NATO’s post–Cold War enlargement also can’t be reduced to ideology: once the Warsaw Pact dissolved, NATO could either become an all-European structure including Russia or keep expanding as an alliance, and admitting Russia would have upended its internal balance and diluted U.S. dominance, so enlargement inevitably cast Russia as the “other.” Inheriting the Soviet arsenal, Russia expected to be treated as an equal in European security, as a nuclear superpower and a contributor to ending the Cold War, but the West treated it as the loser with no right to equal status, and without the economic and political resources to back its demands it kept retreating as NATO expanded. In that light, blocking NATO enlargement was itself a sufficient strategic goal, and reading Putin as acting mainly on ideology misses this larger realist logic. The framework, leaders’ ideas plus domestic institutions, is general, and it draws on leader-focused IR and FPA work, but it is most useful where the leader is dominant, domestic checks are weak, and ideology clearly structures threat perception. Where those conditions hold, a leader-and-ideas account adds real explanatory power; where power is more dispersed or threat perception is driven by material structure, the framework will explain less. This is the standard trade-off between complexity and parsimony. McFaul gains explanatory richness for the Putin case, but the theory becomes less portable.
Jost (2022): Leaders, Bureaucracy & Miscalculation
Working Paper · national-security institutions and crisis failure
A good typology should be mutually exclusive, collectively exhaustive, and theoretically meaningful, capturing variation that helps explain the outcomes we care about. Jost builds his on two dimensions. The first is leaders’ information-search capacity: structures are “inclusive” (low transaction costs for a leader to reach bureaucratic information) or “insular” (high costs, so leaders rely mostly on their own information). The second is inter-bureaucratic information sharing: structures are “open” (bureaucrats can access each other’s information) or “closed” (limited sharing). Crossing them yields four types, Integrated (inclusive + open), Siloed (inclusive + closed), Fragmented (insular + closed), and Dictatorial (insular + open). The typology is conceptually clear and seems to capture meaningful variation; one weakness is that the “dictatorial” cell is underdeveloped, Jost concedes it should be observed only rarely, since a state has few incentives to build inter-bureaucratic sharing if the leader stays insulated from it. Jost identifies two pathways to miscalculation: incomplete information, where leaders lack critical information that bureaucrats hold, and low-quality information, where leaders receive inaccurate or partial information. Integrated institutions perform best because open structures let bureaucrats “police each other’s information” and spot interdependent information, pieces whose meaning changes when evaluated together, while inclusive structures ensure that high-quality information actually reaches the leader. Siloed institutions suffer because bureaucrats supply lower-quality information without inter-bureaucratic scrutiny; fragmented ones suffer because leaders can’t reach critical expertise. I find the logic generally compelling, especially the argument about information policing and interdependence, but the mechanism could be stronger, it isn’t fully clear why bureaucrats would always be motivated to provide higher-quality information just because they can see each other’s work. The unit of analysis is the directed dyad-year (country A toward country B in year i). The models are logistic regressions, per the table note. The dependent variable is international crisis failure, binary, coded 1 if A initiated a crisis against B that failed to achieve its objectives, 0 otherwise. The independent variables are the national-security institution types (categorical, with “integrated” as the base category). The stars mark statistical significance (* p<0.05; ** p<0.01; *** p<0.001). The primary takeaway: all three non-integrated types, siloed, fragmented, dictatorial, are associated with significantly higher rates of crisis failure than integrated institutions, supporting Jost’s hypotheses. Country fixed effects (in two of the models) control for unobserved, time-invariant country characteristics, isolating the effect of institutional variation within countries rather than between them.
“Threats to inference” are alternative explanations that could produce the observed relationships without supporting the theory. Jost discusses three. Reverse causation: crisis failures might cause institutional change rather than the reverse, but existing research suggests failures typically prompt reform toward better institutions. Leader characteristics: maybe certain leader types both choose bad institutions and decide badly, so he controls for observable leader characteristics and shows most leaders inherit rather than design their institutions. Selection effects: perhaps the results reflect which crises states initiate rather than how they perform, and here the design does real work. Because the dependent variable is an initiated crisis that failed, he ties choosing-to-act and final-outcome together; robustness checks then separate initiation from outcome and find that institution type has no significant relationship with initiation but a significant one with failure after initiation. Figure 4 splits self-initiated from passively-entered crises and shows non-integrated institutions are significantly more failure-prone only when the state itself initiates, evidence that the difference comes from miscalculation at the point of choosing which crises to enter, not from worse battlefield performance. My remaining caveat: measuring “miscalculation” through crisis failure is indirect, since we don’t observe the actual decision-making process or leaders’ beliefs. Jost challenges a key assumption of the canon. Works like Allison and Zelikow (1999), and figures like Kissinger (1979), held that “no one organizational structure is best” and that bureaucracy generally undermines good judgment. Jost’s findings push back: institutional design matters, since some structures systematically outperform others, and bureaucratic participation can improve decisions, contrary to the assumption that bureaucrats always supply worse information than a leader could gather alone. That is a real shift, from treating bureaucracy as inevitably problematic to recognizing that how bureaucracy is organized determines its effects. He flags new directions: integrating other bureaucratic features (merit-based appointment, professionalism, competence), exploring the informal side of advisory systems, and understanding how national-security institutions change over time. I’d add cross-national variation in how similar formal institutions are actually implemented, the role of bureaucratic culture and norms beyond formal structure, how these dynamics play out in non-crisis foreign policy, and the interaction between domestic political constraints and bureaucratic organization.
Application · Theory in current events
Jost. 2023. “The Bad Advice Plaguing Beijing’s Foreign Policy.” Foreign Affairs.
Jost (2023) takes the recent miscalculation to be the January 2023 Chinese spy balloon that entered U.S. airspace and derailed Secretary Blinken’s planned visit. He argues Xi likely approved the balloon program in general but didn’t know the particulars of that flight or how it would collide with his near-term diplomatic goal, and that the deeper cause lies in China’s national-security system: incentives that make subordinates hesitant to correct the leader, and siloing between military and civilian organs that degrades the information reaching Xi. This maps onto his 2022 typology, a siloed system (PLA vs. civilian), where deference-and-distortion (officials tell the boss what he wants to hear) and stovepiping (organs don’t share well, so each pushes narrow, un-cross-checked inputs) flow from autocratic survival incentives: centralizing control and appointing loyalists keeps elites weak but degrades information quality, which is why even a strong leader can blunder.
I’d weigh some alternatives that could weaken the bureaucratic-error story: the balloon’s path might have been an unanticipated meteorological drift rather than a decision failure; the cancellation of Blinken’s trip might reflect Washington’s domestic politics more than Beijing’s expectations; and parts of the PLA could run low-visibility collection programs with limited oversight, an autonomy story more than a pure information-quality one. Jost also insists the pathology isn’t unique to autocracies: bureaucracies everywhere commit “sins of commission and omission,” but authoritarian politics make the risks more acute. His other Chinese cases run the same way, Mao’s 1969 “escalate to de-escalate” move against the Soviets inside a Cultural-Revolution echo chamber; the costly 1979 war with Vietnam amid institutional decay and over-optimistic assessments; and the 2001 EP-3 crisis, where the PLA gave Jiang a skewed narrative that siloed civilian agencies couldn’t correct. Each is mostly a bureaucratic-politics and information-pathology story, with individual leaders making the final call under bad information, not simply “bad leaders” or “bad groups,” but bad information environments producing bad choices. It connects directly to this week’s broader question: do major miscalculations come from individual-leader biases or bureaucratic pathologies, and how would we ever know?
← Back to the genealogy tree · ← Prev: Week 2, Classical Realism · Next: Week 4, Structural Realism & Anarchy →