1.1 Temporal Variation in US Covert Activities

Identifying instances of covert FIRCs is a notoriously difficult task given the secretive nature of the subject matter. The task is also complicated by definitional disagreements regarding both covert action and regime change itself, with scholars disagreeing on what counts as sufficient involvement to warrant a case being considered an US-backed attempt at regime change. Setting aside these disagreements, we can identify common trends by examining patterns across different studies by Levin (Reference Levin2016), O’Rourke (Reference O’Rourke2018), Berger et al. (Reference Berger, Easterly, Nunn and Satyanath2013), and Johnson (Reference Johnson1989). Figure 1 depicts the patterns in the first two datasets by Levin and O’Rourke, illustrating that covert interventions peaked in the 1950s and 1960s and reached their nadir in the mid-to-late 1970s, before surging again in the Reagan administration. Importantly, in both datasets the sharp decrease in activity precedes the institutional reforms and oversight overhaul of 1973–75.

Figure 1 Covert electoral interference (Levin) and covert FIRCs (O’Rourke).

This general pattern is similar to the one found by Berger et al., who identify cases where the CIA removed “an existing leader and installed a new leader” (Berger et al., Reference Berger, Easterly, Nunn and Satyanath2013, 867) from 1945 to 1989. While the authors note a gradual decline in active influence operations starting in the late 1960s, their dataset only includes one instance of a new operation between 1969 and 1981. This pattern is also largely consistent with Loch Johnson’s estimates of the CIA’s operations budget from 1947 to 1986, represented in Figure 2, which pinpoints an apex of expenditures in the late 1960s, a sharp decline starting in 1968–69, and an uptick after 1979 (Johnson, Reference Johnson1989, 87).

Figure 2 % of CIA annual budget dedicated to covert action (approx.), reconstructed from Johnson (Reference Johnson1989, 103).

As we argue in the following sections, existing explanations focusing either on international factors, such as the rise and fall of detente, or domestic-political factors, such as partisan opposition or formal reform, struggle to account for this variation, especially these specific inflection points and the reversal in the 1980s. While most existing IR studies of US covert action acknowledge, even if in passing, the importance of the investigations and hearings in the early-to-mid 1970s and subsequent reforms, few attempt to more systematically theorize how changes in domestic politics affect the dynamics of covert FIRCs. Our theory suggests that rather than being a one-off chain of events or shocks that definitively and irrevocably changed the policy environment, effective constraints on presidential behavior are largely the product of sustained congressional attention and that this attention can wax and wane even after reforms are instituted, as formal rules and regulations can be undone or circumvented.Footnote ⁵

1.2 How Congressional Attention Constrains FIRCs

The battle between the Executive and Congress regarding control and authority over different policy instruments is well documented, and scholars have long noted that it is a perennial struggle that nevertheless varies greatly in intensity over time. This research has shown that there are inherent challenges for congressional assertiveness on foreign policy, but that this does not mean that Congress is powerless against the president even in matters of national security (Lindsay, Reference Lindsay1994; Howell and Pevehouse, Reference Howell and Pevehouse2007; Fowler, Reference Fowler2015). Much of the recent scholarship on democratic foreign policy has focused on the role of “domestic audience costs,” or the political costs that the general public, or at least the attentive public can impose on leaders in a democratic society (Fearon, Reference Fearon1994; Schultz, Reference Schultz2001; Davies and Johns, Reference Davies and Johns2013. Scholars such as Jessica Weeks and Elizabeth Saunders have shown that elites can constitute their own set of relevant audiences even when the public is inattentive or unable to hold leaders directly accountable (Weeks, Reference Weeks2008; Saunders, Reference Saunders2015). Others have argued that elites play an important role in mediating the effect of mass opinion, by shaping or “cueing” public beliefs (Page and Shapiro, Reference Page and Shapiro2010; Guisinger and Saunders, Reference Guisinger and Saunders2017), or by acting as “filters” through which public opinion is interpreted and acted upon (Foyle, Reference Foyle1997). Saunders explains that “[l]eaders can retain significant autonomy to conduct foreign policy – including foolish or losing policies – as long as an elite consensus holds, or elite dissent is contained” (Saunders, Reference Saunders2012, 3). One of our contributions in this volume is to argue how the influence of political elites, particularly members of the US Congress, is at least partly conditional on their paying attention to certain foreign policy issues. Attention is a scarce resource and, therefore, an important indicator of political priorities (Jones and Baumgartner, Reference Jones and Baumgartner2004). Legislators, like all political actors, must pick and choose how they spend their limited time and cognitive resources, and what issues they choose to focus on at any given time. We draw on multi-archival research to offer compelling causal-process evidence that legislators’ ability to influence foreign policy is partly determined by where their attention lies and that both legislators and policy-makers in the Executive are acutely aware of these dynamics.

Our research builds on scholarship from Howell and Pevehouse (Reference Howell and Pevehouse2007), Peck and Jenkins (Reference Peck, Jenkins, Carson and Lynch2020), Kriner (Reference Kriner2010, Reference Kriner2018), and Tama (Reference Tama2020, Reference Tama2024), among others, in showing how legislators can influence foreign policy through formal and, especially, informal mechanisms. As scholars of American politics have noted, legislators spend their valuable and scarce time not only on introducing and passing legislation but also on various other activities meant to shape policy and increase their influence and power, either individually or institutionally. As Josh Chafetz argues, “Congress has numerous powers other than the power to pass bills into law, powers that tend to receive scant treatment even in isolation; they are almost never grouped together and conceptualized as a coherent set of legislative tools. And yet these tools together are potent, giving Congress the ability to assert itself vigorously against the other branches” (Chafetz, Reference Chafetz2017, 2). This is also one of the central arguments in Mayhew’s seminal work on congressional behavior. He notes that “Members of Congress perform actions beyond just making laws [ … ] [they] take part in the public sphere in an impressive variety of other ways. [ … ] As important as anything, perhaps, they take stands” (Mayhew, Reference Mayhew2000, 227). James Lindsay, in his study of congressional influence over foreign policy, devotes an entire chapter to nonlegislative activities designed to influence public and elite opinion and consequently shape policy (Lindsay, Reference Lindsay1994). This broader range of actions can increase the costs to the executive after a policy is implemented as well as affect how they anticipate the costs associated with future actions (Howell and Pevehouse, Reference Howell and Pevehouse2007). For example, Kriner found experimental evidence that “Legislatures need not necessarily exercise their constitutional war powers to influence the conduct of military affairs” (Kriner, Reference Kriner2018, 64). In particular, he contends that congressional criticism can greatly affect popular support for military action and the Commander-in-Chief, thus constraining the president.

While there is a rich scholarship on how Congress and other political elites can constrain or influence presidential decisions regarding overt military action and other national security issues such as alliance formation and defense expenditures, few studies have explored how these instruments affect the calculus related to covert action. There is good reason for this. It is reasonable to expect that some of these challenges Congress faces in influencing foreign policy are particularly acute in the realm of intelligence and covert action: Congress faces a substantial information asymmetry – even more than in other areas – legislators are unlikely to benefit electorally from pursuing oversight over intelligence, classification regulations and practices create barriers for information sharing, and interest groups and lobby activities are relatively scarce even at the height of scandals, leading experts to expect this to be an issue area where the Executive should have a particularly easy time retaining autonomy (Zegart, Reference Zegart2013; Colaresi, Reference Colaresi2014; Milner and Tingley, Reference Milner and Tingley2015).

Two recent studies have made important progress in excavating the effects of domestic politics on covert action and testing them empirically. Nandita Balakrishnan (Reference Balakrishnan2020) argues that institutional reforms in Congress, following a spate of scandals in the early 1970s, severely limited US ability to intervene abroad to support coups, causing a decline in military coup activity around the world. In another recent study, Gregory Smith (Reference Smith2019) argues that prior to these oversight reforms in the 1970s, patterns in US covert operations were best explained by partisan opposition in divided governments. Once oversight mechanisms are in place, however, he suggests that divided government plays a diminished role, or perhaps none at all. We argue that these studies are correct in highlighting the potential for domestic politics to affect the resort to covert action, but that neither tells a complete story and both fail to capture the central role that congressional attention, flawed and uneven though it may be, plays in reining in the president’s use of covert operations.

An account that focuses on partisan opposition Footnote ⁶ and divided government struggles to explain the exuberant use of covert FIRCs by the Eisenhower administration and the muted congressional response to those operations at a time when Democrats held both the House and Senate, as well as pitched battles over executive overreach in covert operations after reforms took place.Footnote ⁷ In the absence of sustained elite attention, early attempts at strengthening congressional oversight (pushed largely – but not solely – by liberal Democrats under Republican and Democratic presidencies) were met with intransigence and foot-dragging by the White House as well as by a broad bipartisan consensus that produced, by both omission and commission, what historiographies of US intelligence consider the golden age of legislative deference to the intelligence community and the Executive and what proponents of executive autonomy call the period of “benign neglect.”Footnote ⁸

Similarly, an emphasis on institutional reforms and formal oversight cannot explain the downward trend in covert FIRCs in the late 1960s to early 1970s, before new oversight mechanisms were instituted, nor can it explain the later increase in the 1980s, when many of the formal mechanisms were still in place. We do not argue that formal oversight or codified rules, such as the creation of the Senate and House Select Committees on Intelligence or the Foreign Intelligence Surveillance Act were unimportant. Whereas Zegart and Quinn (Reference Zegart and Quinn2010) have argued that congressional oversight is generally ineffective “by design,” we suggest that the effectiveness of these mechanisms is somewhat conditional on the level of congressional attention and that attention itself can have a powerful constraining effect even when formal oversight mechanisms are poorly institutionalized.Footnote ⁹

Covert and clandestine activities tend to have an affinity with executive autonomy, as they rely on remaining secret and/or deniable – this is, in fact, their defining trait. As such, widespread attention or publicity, whether supportive or critical, can constitute a major threat to covert operations, especially when coming from recognizable official sources.Footnote ¹⁰ A key implication is that covert activities are particularly vulnerable to nonlegislative actions by members of Congress such as floor speeches, press releases, leaks, and hearings. In other words, when it comes to covert action, broad congressional attention is not only an indicator of potential political costs or constraints but can also directly produce risks and costs for the Executive and its agents.

To explore congressional attention and its associated effects on the president’s authorization of FIRCs, we introduce, in the following sections, evidence from computer-assisted content analysis of Congressional Records, statistical analysis of covert operations, and causal-process observations from multi-archival primary sources. Following Quinn et al (Reference Quinn, Monroe, Colaresi, H Crespin and Radev2010), we operationalize congressional attention by examining how often a legislator talks about an issue on the Congress floor.Footnote ¹¹ Speeches, we argue, offer an imperfect but less biased measure of what issues members of Congress consider worthy of attention and, importantly, what they consider worthy of public consumption.Footnote ¹² Because floor speeches are public, taking something to the floor enters it in the official record and carries the likelihood that it will be picked up by other members of Congress, the media, or other interested parties. While committee hearings or bills, which other studies have used to measure the intensity of congressional interest in a topic, are potentially a more costly signal of priority and position, they can also be kept behind closed (sub)committee doors, redacted, and “sanitized” before being released to the public, as is the case with much of congressional activity on matters of intelligence and covert action. We show throughout this volume that administration officials and congressional supporters of executive autonomy have understood this of floor debates and have tried to prevent or quell them accordingly.

While we do not discount bills, resolutions, or hearings in our qualitative account of the fights over congressional oversight, there are also methodological reasons to favor floor speeches over these other instruments: they are more subject to systematic bias due to correlation with other variables of interest like partisanship and institutional capacity. Precisely because hearings and bills are time-consuming to organize and draft, they are much more reliant on the existence of committee staff, which means that we are far more likely to observe these after dedicated committees were established in the mid-1970s. If partisan opposition plays a role in attention and oversight, hearings and bills will be a very biased measure, less likely to be observed in times of unified government as majority parties are expected to block hearings and votes that they deem prejudicial to their president or preferred policies.

This is not to suggest that we expect partisanship or partisan opposition to be the main driver of legislative activity on intelligence and covert action – and as we show later it is not – only to guard against potential measurement bias as it relates to possible alternative explanations. As Zegart notes, “lackluster intelligence oversight from the 1940s to the 1970s remained remarkably resilient and consistent despite changes in party control of the House and Senate, despite fourteen years of unified government and sixteen years of divided government, and despite nearly an even split between Democratic and Republican presidents ”(Zegart, Reference Zegart2013, 22). Nor are we suggesting that low polarization or a supposed “bipartisan Cold War consensus” meant that politics stopped at the water’s edge.Footnote ¹³ Rather, we argue that politics was very much an important part of foreign policy-making during the Cold War, but that not all politics is partisan, and that Legislative attempts to reel in executive power transcended partisan divides even as polarization deepened. Our argument in this Element is in line with recent work that calls into question binary and static conceptions of partisanship and the tendency to read current partisan divides and levels of polarization back into the historical record, inadvertently exaggerating or misunderstanding its role (Friedrichs and Tama, Reference Friedrichs and Tama2022).

We also draw on scholars such as Gries (Reference Gries2020) and Jeong and Quirk (Reference Jeong and Quirk2019) who suggest that ideology, aside and apart from – and maybe more so than – partisanship, plays a critical role in explaining congressional preferences over foreign policy. Our qualitative and quantitative analysis both speak to the importance of disentangling ideology and partisanship in this way to understand the fault lines over covert action, intelligence, and executive authority during the Cold War – and perhaps beyond.Footnote ¹⁴ We also find, consistent with other recent scholarship, that presidents often worry more about co-partisan than about opposition critics.Footnote ¹⁵ A recent study focusing specifically on congressional oversight of intelligence notes that “for Congress to impose its will over the executive’s preferences [ … ], members of Congress from the president’s party must also agree that the executive’s relationship to the CIA needs restructuring” (Haas, Reference Haas2023, 8). Tama refers to this as anti-presidential bipartisanship – the cooperation between the two major parties to constrain or oppose presidential policies – and finds that even in today’s extremely polarized environment, bipartisanship is more common than traditionally recognized on matters of foreign policy (Tama, Reference Tama2024).

To be sure, some studies have found that the partisan composition of Congress can affect US foreign and security policy and that opposition politics can be a constraint on the president. For example, Kriner finds that when a president is faced with less partisan support, they are less likely to use force abroad (Kriner, Reference Kriner2010), and Howell and Pevehouse (Reference Howell and Pevehouse2007) demonstrate that when leaders face strong majority opposition, they issue fewer executive orders. These effects, even on overt action, however, are not universal but contingent. Howell and Pevehouse (Reference Howell and Pevehouse2007) further find that the composition of Congress matters only for major uses of force and not for minor deployments.Footnote ¹⁶ This is consistent with our expectation and findings that partisanship plays a smaller role when considering covert operations that are usually far cheaper and have a much smaller physical footprint than even the most limited overt military operations. We also find evidence that the extent to which intelligence and covert action matters become polarized across party lines is itself partially a product of how much attention these issues get, both within Congress and among the broader public. This means that politicians have a degree of choice in how much they want fights over covert action to become partisan issues and that they often choose the mode and venue of the fight accordingly, to either minimize or maximize polarization.

The resulting picture is one in which congressional attention affects executive decisions regarding covert action independently from formally codified oversight, and, second, that congressional attention is not a manifestation of partisan disputes or simply a reflection of opposition politics, or a purely reactive or opportunistic response to policy failures or scandals. We propose that congressional attention creates particular problems for covert operations, namely making it difficult to keep them covert, raising the expected political costs and risks associated with these operations.

2.1 A Conspiracy of Silence

Between 1947 and the mid-1960s, congressional attention to and scrutiny of CIA activities was extremely limited. The structure of the Agency (Zegart, Reference Zegart2000), informational asymmetries between Congress and the Executive, and lack of congressional incentives (Johnson, Reference Johnson2005) made inquiries challenging and rare (Lester, Reference Lester2015). CIA operations were rarely discussed outside of the Armed Services and Appropriations subcommittees – the two congressional groups that exclusively oversaw intelligence activities. Even within those small secretive groups, debates were often nonexistent, as congressional watchdogs abdicated their responsibility, gave the Agency a wide berth, and repeatedly blocked attempts by the few legislators who deigned to call on the Agency’s representatives to account for their finances or activities (Durbin, Reference Durbin2017). Sustained campaigns to increase intelligence oversight were few and far between. Most notable among them were Democratic Senator Michael Mansfield’s attempts to establish a Joint Intelligence Committee, modeled after the Joint Committee on Atomic Energy, between 1953 and 1955 (Barrett, Reference Barrett2005, esp. 171–176). Mansfield’s resolutions lacked broad support among his colleagues, with the last failing in a vote of twenty-seven to fifty-nine despite a narrow Democratic majority in the Senate. The Eisenhower administration’s discussion of the dangers of such a committee, however, give critical insight into the political calculations regarding covert operations and congressional oversight. A 1956 memorandum from DCI Allen Dulles, and the Office of the President to the National Security Council regarding Mansfield’s proposals for an Oversight Committee on Intelligence is worth quoting at length,

Dulles went on to describe the risk of increased tensions between the Executive and Congress were such a committee to be established:

Mansfield would relentlessly continue to push the idea of a joint committee through the 1960s, encountering strong bipartisan opposition each time and rarely being able to bring bills or resolutions to a vote. It would be more than fifteen years before open discussions of more extensive oversight gained traction in Congress, and these calculations continued to color executive decision-making regarding covert operations.

Despite the widespread use of covert action in the early years of the Cold War and the proliferation of spy thrillers in novels, movies, and TV, the CIA’s activities only became a substantial subject of debate within elite policy circles in the mid-to-late 1960s. We can look to the Council on Foreign Relations’ Foreign Affairs magazine as a proxy for broader elite attention. Foreign Affairs includes articles authored by influential policymakers and academics and was consumed by the same audiences.Footnote ²³ In the period from 1945 to 1950, there was only one article published that mentioned the CIA; a 1946 article titled “Our Armed Forces: Merger or Coördination?” which focused on post-WWII adaptation and the need for a central intelligence agency. From 1951 to 1960 one article was published that mentioned the Agency, but only in passing. From 1961 to 1965, two articles mentioned the CIA. The first “Toward a New Diplomacy” (1962), mentions it in the context of foreign policy integration, among a range of other agencies and offices. The second “Slow Down at the Pentagon” (1965), mentions the CIA briefly in the broad context of expanding government bureaucracy. From 1966 to 1970 we see a spike in the number of articles and the extent to which they discuss the agency substantively. This trend would then hit a peak in the 1976 to 1980 period, with a quarter of the articles published in 1978 and 1979 mentioning the Agency, after which these discussions became relatively less frequent but never quite disappeared. Figure 3 plots the number of articles mentioning the CIA each year between 1946 and 1990.Footnote ²⁴

Figure 3 Foreign Affairs articles mentioning the CIA.

The first article to substantively discuss the CIA,“Intelligence and Foreign Policy: Dilemmas of a Democracy,” was published in January of 1969 and is worth quoting at some length. In the article, historian William BrandsFootnote ²⁵ identified the moment when elites first started to take notice of CIA activities as the downing of Francis Gary Powers’ U-2 spy plane in 1960 and the spectacular failure at the Bay of Pigs in 1961. Brands explains, “From then on there were periodic revelations of past U.S. intelligence operations, and after each disclosure there was a new outcry for more control over CIA and less reliance on it.” Brands criticized certain alleged CIA activities from 1965 to 1967 and acknowledged that intelligence and covert operations can sometimes conflict with democratic ideals and principles, but rejected the notion that such covert activity should therefore be rejected out of hand. More importantly, he questioned whether more congressional oversight was likely to improve policy or resolve the inherent dilemma. Rather, he cautioned, that “(a) formal Joint Committee would also create additional pressure on (the) CIA to adopt a cautious and bureaucratic approach in a field where imagination and flexibility are important qualities” and would “reduce the willingness of foreign intelligence services to pass intelligence to, and cooperate with, CIA, because of their fear that such relationships would become known and create domestic political problems” (Brands, Reference Brands1969, 294–295).

Brands’ article is notable because it reflects the old bipartisan elite consensus and marks the point at which it started to visibly erode, nearly a decade after the U-2 incident and Bay of Pigs. The increased attention after each episode was short-lived and failed to generate enthusiasm for oversight. The U-2 hearings in the Foreign Affairs committee were designed less to uncover the causes of the incident or question the CIA’s or the administration’s management of intelligence-gathering operations and more to discuss potential implications of the event for the Paris Summit and US-USSR relations going forward. The proceedings were kept as distant from the public eye as possible – given the publicity the incident had already received by the USSR and the media – and deliberately avoided discussions about congressional oversight and partisan blame games. In 1960, at a May 26th White House meeting between Eisenhower and congressional leaders from both parties to discuss the incident and the upcoming congressional inquiry, Senator Mansfield asked the president if he would support the creation of a joint committee to oversee the CIA. Eisenhower rejected the proposal emphatically and said that CIA operations were “so delicate and so secret” that he would support “a bipartisan group going down occasionally and receiving reports from the CIA on their activities, but that he would hate to see it formalized.” Perhaps more importantly, the president’s position was echoed by every other congressman in the room, including the powerful members of Mansfield’s own party.Footnote ²⁶ Senator Fulbright, who would be heading the hearings in the Senate Foreign Relations Committee, emphasized that he would like to keep the inquiry “on track and not let it stray” and avoid turning the matter into a political one. The only criticism aimed at Eisenhower, by Fulbright himself, was the president’s decision to take full responsibility and not disavow the operation.Footnote ²⁷ The hearings, initiated the next day, were held in Executive Session (behind closed doors and limiting non-committee members to observer status). As Fulbright promised, the Committee report was only published after White House staff had thoroughly “sanitized” its contents and excluded the entirety of the testimonies by CIA personnel.Footnote ²⁸ Hearings following the Bay of Pigs and the Cuban Missile Crisis followed similar dynamics.

It was only the Johnson administration’s fateful overt military intervention in the Dominican Republic in 1965 that finally prompted Fulbright to question the wisdom of the intelligence and analysis provided by the CIA and to seriously consider the need to establish closer oversight to avoid future unnecessary, costly – the United States lost more than 40 Marines during the botched operation – and damaging interventions. The Dominican fiasco helped push Fulbright, McCarthy, and other Senators in the Foreign Relations committee – mostly Democrats who were already beginning to question American covert and overt involvement in Vietnam – to revisit Mansfield’s proposals for a joint committee or perhaps create a new intelligence subcommittee under Fulbright’s Foreign Relations Committee. The move, orchestrated largely behind closed doors in meetings of congressional leadership in both parties, provoked fierce opposition not only from the CIA and the White House, but also from senators in the Armed Services and Appropriations subcommittees – especially conservative Democrats and Republicans like Richard Russell and Everett Dirksen, who believed that existing instruments were more than sufficient. Mike Mansfield, who had for years pushed tirelessly for a joint committee on intelligence found himself torn between his personal convictions and his new responsibilities as Senate majority leader. He tried to strike a compromise, advocating first the expansion of the existing CIA subcommittees to include members of the Foreign Relations Committee and later the creation of a two-man Foreign Relations subcommittee consisting of the chair and ranking member. Mansfield’s main goal in brokering the compromise was to avoid a fight on the floor of Congress. He warned Johnson that a floor debate would hurt the CIA, the Senate, and the White House:

Johnson agreed that a floor fight was undesirable, but felt that he could not trust Fulbright, his erstwhile close ally, and other members of his committee not to leak information furnished by the CIA. The president rejected Mansfield’s compromise and chose instead to work with Russell and Dirksen to stall and kill the proposals in committee.Footnote ³⁰

2.2 Attention Starts to Mount

Two events in 1967 heralded the shifting grounds. The first was the “Ramparts affair,” which CIA director Richard Helms would later identify as the “first sort of unzipping of covert operations that the Agency was involved in.”Footnote ³¹ On February 1967, Ramparts, a wildly popular left-leaning magazine, published an article documenting CIA involvement with the National Students Association, in supposed violation of the Agency’s charter (Hathaway and Smith, Reference Hathaway and Jack Smith1993; Jeffreys-Jones, Reference Jeffreys-Jones2003; Prados, Reference Prados2006). The story was picked up by other major outlets and led to a cascade of revelations of clandestine involvement with various groups and organizations operating abroad and at home. Criticism came primarily and most vocally from Democrats, with Republicans, including minority leaders, vehemently defending the importance of supporting the cause of freedom by any means necessary. Having learned about the story ahead of time, the Johnson administration was able to anticipate it and enlist support from legislators of both parties, ultimately overwhelming the (primarily co-partisan Democratic) critics at home and counterbalancing the mounting public outcry in foreign capitals. While vigorously defending its actions, the administration also rushed to revise its policy on clandestine funding of voluntary organizations to avoid further embarrassment, but the genie would not go back in the bottle. As Helms later put it, things “settled down again but never to be precisely the same.”Footnote ³²

The second event in 1967 that proved significant was the creation of the task force that would produce the “Pentagon Papers.” Set up inside the Department of Defense to review US military and political involvement in Vietnam from 1945 to 1967,Footnote ³³ the study would reveal the extent of CIA activities in Indochina. In 1971, MIT researcher and former RAND analyst Daniel Ellsberg leaked the Pentagon Papers to the New York Times (Moynihan, Reference Moynihan1998; Moran, Reference Moran2015). The leaked documents, and the extensive coverage they received, opened a veritable floodgate of further leaks, many of which were received and publicized by journalist Jack Anderson, who already figured among Nixon’s worst enemies. In the following years, Anderson published secret memos revealing, among other things, US covert support for autocratic Pakistan against democratic India during Bangladesh’s War of Independence in 1971Footnote ³⁴ and collusion between the CIA and I.T.T. Inc., a Connecticut-based multinational corporation that had allegedly been involved in attempts at subverting democratic elections in Chile against Allende in 1970, and in the United States in 1972.Footnote ³⁵ In May 1973, these revelations sparked an investigation by a subcommittee under the Senate Foreign Relations Committee into I.T.T.’s and the Nixon administration’s activities in Chile. This “Subcommittee on Multinational Corporations” was chaired by Democratic Idaho Senator, Frank Church.

The Watergate Scandal in 1972–1974 would bring new levels of congressional attention to the president’s use and abuse of the CIA and other agencies tasked with foreign and domestic intelligence activities, including surveillance, intimidation, and assassination attempts against political opponents in the United States.Footnote ³⁶ Vice President Nelson Rockefeller, Henry Kissinger, James Schlesinger, Phillip Buchen, and James Lynn, in a 1975 memo to President Ford, noted that “[o]ne of the most serious consequences of Watergate was that the intelligence community became a topic for congressional investigation, as well as public and press debate.”Footnote ³⁷

In 1972 and 1973, Congress also held various hearings sparked by revelations of secret operations in North Vietnam, as well as Laos and Cambodia during the Vietnam War. Faced with electoral defeat and massive public pressure, Republicans voted in large numbers for the Case-Church Amendment, which officially ended the flow of funds for the war. That same year, James Schlesinger, the new director of the CIA, directed Agency employees to report all activities since the Agency’s founding that may be considered outside the scope established by law, especially those in the United States. Schlesinger was, above all, determined to get a sense of what other skeletons might be waiting to jump out from the closet and sink the Agency and the administration. The resulting collection of reports came to be known as the “Family Jewels.”Footnote ³⁸ William Colby, a career intelligence officer and Schlesinger’s successor, was intent on keeping the Family Jewels from being publicly released and not having his confirmation hearings become a spectacle. He did not have to try hard; the Democratic leadership in the Armed Services Committees in the SenateFootnote ³⁹ and the HouseFootnote ⁴⁰ were happy to oblige and keep potentially damning CIA activities out of the public eye (Prados, Reference Prados2003, 262–265). Though Colby’s honeymoon with Congress would not last long.

It was under this radical new level of scrutiny that the Nixon administration had to operate when deciding whether, how, and much to support, coordinate, or conspire with Chilean opposition and the military in the Chilean congressional elections in March 1973 and in the lead-up to the September 1973 coup that led to Allende’s ouster. As Peter Kornbluh notes, State Department and “senior CIA officials […] feared the consequences of precipitous military action and believed in the prudence of caution given the ongoing congressional committee investigation into ITT.”Footnote ⁴¹ In reviewing continued support for the opposition and the signals that could send of US support for a coup, Acting Assistant Secretary of State for Inter-American Affairs, John H. Crimmins, noted in May 1973 that “[a]nother important factor was the increased sensitivities in the US and in Chile to covert activities of this kind. This sensitivity could well raise the risk level of the enterprise.”Footnote ⁴² In a later meeting on June 11, Harry Shlaudeman, the incoming Deputy Assistant Secretary of State for Inter-American Affairs, fresh from his four-year stint as Deputy Chief of Mission in Santiago, noted that “the risk of our assistance becoming known centered here rather than in Chile. [ … ] exposure here would receive a far bigger play and far more credence in Chile than anything that could be said there.” The Assistant Secretary of State for Inter-American Affairs, John Kubisch, agreed with his outgoing Deputy, Robert Hurwitch, that “new sensitivities in the US and in Chile to US covert activities should not necessarily drive us to abandon all these, but that they would make it necessary to analyze in a much more critical way than hitherto the importance of the objectives that we were trying to achieve through them,” and noted that “in this particular instance he did not believe that the benefits outweighed the potential cost.”Footnote ⁴³

A month later, after a failed military coup seemed to imperil the survival of the opposition, Kubisch would become more supportive of continued financial support for political parties and private actors opposing Allende, but would again note that “[r]ecent disclosures and allegations about U.S. activities in Chile in 1964 and 1970, together with current developments and attitudes in the U.S. towards covert government activities, make the potential damage to the USG from exposure of this program far greater than in the past.”Footnote ⁴⁴ In these debates, as in the cases we will discuss at greater length in the next sections, congressional attention, even absent institutional reform or legislation, directly influenced the perception of political and operational risk and shaped decisions regarding not only whether to intervene but how and how much. A month before the coup that brought Pinochet to power, the 40 Committee would approve the continuation of financial support but refrained from expanding the support to new partners in Chile or making more explicit commitments to coup plotters to preserve deniability. In the days after the coup, Kubisch, Shlaudeman, and others would be called upon to testify in front of the House and Senate committees on foreign affairs.

In 1974, Congress passed the Hughes-Ryan Amendment to the Foreign Assistance Act, establishing firmer oversight over covert operations. The vote was close but, importantly, did not fall along partisan lines, in either the House or the Senate.Footnote ⁴⁵ The most important provision in the amendment, which would remain a key focus of contention for decades to come, was the requirement that covert operations were to be reported, “In a timely fashion,” not only to the intelligence subcommittees in the Senate and the House Armed Services and Appropriations committees, but to six full committees, before appropriated funds could be spent on these operations.Footnote ⁴⁶ With “timely fashion” left undefined, the main implication of the amendment was to vastly increase the number of legislators who would have access to information related to covert operations – and therefore were able to leak them to the press or disclose them in floor speeches – and could be expected to demand clarification on a regular basis (Durbin, Reference Durbin2017).

As information from the “Family Jewels” was revealed to the public by the New York Times’ Seymour Hersh, and as revelations about the CIA’s involvement in Chile mounted, congressional attention swelled. The straw that broke the camel’s back, however, was Hersh’s reporting, in December 1974, on the CIA’s domestic activities, such as its infiltration of antiwar groups and widespread surveillance of US citizens, including members of Congress. Less than a month after Hersh’s first article came out, in January 1975, the Senate impaneled the Select Committee to Study Governmental Operations with Respect to Intelligence Activities, also known as the Church Committee, by a bipartisan vote of 88 to 6.Footnote ⁴⁷ The House created its own Select Committee, led by Otis Pike (D-NY), operating in parallel. The first House Select Committee, chaired by Lucien Nedzi (D-MI), was abolished after the NYT revealed that the congressman had been briefed by Colby on the Family Jewels in 1973, prompting members of his own party to question his ability to investigate the Agency. The ensuing battle saw Nedzi reaching across the aisle to sabotage his own committee (Haines, Reference Haines and Chauhan2004). The Democratic “rebellion” against Nedzi received ample coverage, further fueling public attention.Footnote ⁴⁸ The proceedings and findings from both committees would go on to shine a light on decades of abuses by the CIA, NSA, and FBI and would be highly publicized, making Church and Pike household names and prompting 1975 to be dubbed the “Year of Intelligence.”

Throughout, the White House jealously guarded its prerogatives and repeatedly tried to block Congress’ access to key information and witnesses on grounds of protecting national security. Gerald Ford came to office in 1974, at the zenith of congressional attempts at clawing back authority over foreign policy. After more than a year of institutional jockeying, in February 1976 Ford issued Executive Order 11905 – reorganizing the intelligence community and issuing a prohibition against assassinations – to ward off further legislative action (Poznansky et al., Reference Poznansky, Downes and O’Rourke2017). As Durbin argues, “While it served to define the organizational and operational boundaries of the various components within the intelligence community, the order was most notable for several provisions that responded directly to the criticisms leveled against the CIA by Congress” (Durbin, Reference Durbin2017, 148). Later that year, the Senate created a Permanent Select Committee on Intelligence, granting it subpoena and investigatory powers to examine intelligence activities, including “covert or clandestine activities affecting the relations of the United States with any foreign government, political group, party, military force, movement or other association.”Footnote ⁴⁹ The House followed suit in 1977, creating its own Permanent Select Committee on Intelligence. The votes establishing both committees were bipartisan. In a memo to Brent Scowcroft that spring, NSC staffer John Matheny wrote of the new attention to Agency activity: “Covert action ills are used to justify increased accountability up and down the chain of command – to the point of risking management paralysis by virtue of the vast increase in administrative workload on high-level decision-makers, not to mention increased risks of public disclosure.”Footnote ⁵⁰ That summer, DCI George H. W. Bush described his sense of the new oversight initiatives in a letter to Ford,

As we show later, congressional attention in the 1970s tempered covert FIRCs substantially, though not absolutely or permanently. Covert activities were reignited by shifts in both domestic and international politics in 1979–1980, namely the Iranian Revolution, the Soviet invasion of Afghanistan, and the election of Ronald Reagan. These events put a nail in the coffin of an already-strained détente and reignited demand for US covert operations while also temporarily mitigating institutional battles over foreign policy. Under these geopolitical circumstances and given new authority and responsibility, congressional watchdogs grew weary of overly tying the president’s hands. Attention substantially decreased for a spell, and legislators reverted to a more cooperative or delegative stance vis-à-vis intelligence and covert action while still attempting to set some limits. For example, the Intelligence Oversight Act of 1980, which effectively repealed Hughes-Ryan, reduced from eight to two – the Senate and House intelligence committees – the number of committees that needed to be briefed on covert operations.Footnote ⁵² Additionally, it gave the president the option of restricting notification in particularly sensitive cases – as determined by the White House – to only a select few individuals: the chairman and ranking members of each Intelligence Committee and the House and Senate majority and minority leadership, also known as the “Gang of Eight.”

The most emblematic case in this period of executive-legislative cooperation over intelligence and covert action was, of course, the “secret” war in Afghanistan, where Congress played a supporting, perhaps even instigating, role. However, as we explain in later sections, throughout the intervention in Afghanistan and the broader “Second Cold War” (Halliday, Reference Halliday1983), policymakers in the Executive continued to worry that increased attention to covert action, even in support of the administration’s policies, would limit their options and imperil operations. And while Afghanistan did bridge institutional divides, this truce did not automatically translate to executive control over policy in other parts of the world. Unhappy with limitations they still saw as excessive, and weary of reporting requirements, the Reagan White House proceeded to authorize new operations with only minimal deference for Congress’ formal oversight role while working hard to further “unleash” the CIA (Jeffreys-Jones, Reference Jeffreys-Jones2003, 227). When the White House pushed for authorization to intervene in Central America, Congress pushed back. Reagan moved ahead anyway. In 1982, as the civil wars in Nicaragua, El Salvador, Honduras, and Guatemala raged on, the CIA pursued several operations ultimately curtailed by leaks associated with the Iran–Contra Scandal, where the agency’s activities were not only in direct violation of the rules Congress had established but also struck a powerful chord with popular opinion.Footnote ⁵³ Congressional attention intensified again, with investigations and hearings dominating the airwaves and the Congress floor during the summer of 1987. The fallout from the investigations dragged on for years, including a spate of indictments of top White House officials well into George H. W. Bush’s presidency. The scandal also sparked negotiations between Congress and the White House that ultimately produced the Intelligence Authorization Act of 1991, which introduced a much stricter and clearer set of oversight rules, and which Bush signed despite serious misgivings. We briefly return to how these dynamics have played out in the post–Cold War era in the last section.

3.1 Estimating Attention and Its Correlates

We used computer-assisted content analysis of the congressional Record to estimate the amount of attention paid to issues of intelligence and covert activities throughout the Cold War. To start, we built a corpus of approximately 1.8 million speeches made on the floor of Congress between 1945 and 1991 (sessions 79 through 101).Footnote ⁵⁴ The approach we follow is similar to Quinn et al. (Reference Quinn, Monroe, Colaresi, H Crespin and Radev2010), but differs in important regards, most notably the fact that we are dealing with a much longer and earlier time period, as well as analyzing both House and Senate speeches. To be sure, using floor speeches to measure attention is not without its complications, especially over such a long time period. For example, it is possible that legislators spent less time discussing intelligence and covert action on the floor of Congress after the creation of additional dedicated committees and subcommittees. If that is the case, our measure might underestimate congressional attention after 1975, leading us to similarly underestimate the correlation between congressional attention and covert FIRCs. It is also possible, however, that the existence of specialized committees spurs floor debates on those topics by increasing legislative activity, creating professional and electoral incentives for members to position publicly, and increasing the number of members and congressional staff with experience, expertise, and personal investment in the topic. In fact, Quinn et al. (Reference Quinn, Monroe, Colaresi, H Crespin and Radev2010) found that, among Senators, floor speeches are positively correlated with their committee membership. Additionally, floor speeches serve a different, more public function than the work and discussions that happen largely behind closed doors in committees, especially those dealing with national security.

We first identified all speeches that contain words that reference the CIA or intelligence activities.Footnote ⁵⁵ We coded each speech as “1” if it mentions one of the intelligence-related words and “0” otherwise. Using speaker-related information from Gentzkow et al. (Reference Gentzkow, Kelly and Taddy2019) as well as additional information on party composition of different branches, we briefly explore the correlates of congressional attention to intelligence as a proportion of all speeches on the floor of Congress. We report some of the results of that analysis here and provide additional historical context to help validate the measure. The historical overview earlier in Section 2 and the qualitative evidence in Sections 4 and 5 provide further evidence that variation in attention cannot be explained by previous covert FIRCs, partisanship, or partisan opposition, or even a single set of laws or institutional reforms, alleviating concerns with endogeneity, multicollinearity, and (post-)measurement bias in the next section.

Figure 4 plots the results from a semi-parametric model that estimates the probability that any given speech mentions intelligence accounting for the speaker characteristics and allowing for nonlinear interactions between partisanship and time. The two main reasons one could expect temporal variation in the effect of partisanship are, on the one hand, the partisan sorting and realignment that take place during this period, and, on the other, opposition politics, as Democrats might be more likely to call attention to intelligence operations under Republican presidents and vice versa.Footnote ⁵⁶ We find evidence for the former and little evidence for the latter. We report the full models and discuss the findings further in Appendix A. While our measure of attention is, in theory, neutral with regard to sentiment or position, the correlations and patterns sketched out here all lend support to our contention, drawn from existing scholarship and confirmed by the actors themselves, that when it comes to intelligence and covert operations there is no such thing as good publicity.

Figure 4 Estimated partisan differences in attention to intelligence over time.

Figure 4, as predicted, shows low attention in the first three decades, with a first small bump in attention surrounding the creation and strengthening of the CIA with the 1947 National Security Act and the 1949 Central Intelligence Agency Act. Notably, both acts passed with almost universal bipartisan support. The 1947 act creating the CIA with virtually no congressional oversight authority passed by a voice vote under a Republican-dominated Congress, which Truman nicknamed the “do-nothing Congress” in his re-election campaign. The 1949 CIA Act, which revamped the Agency and gave it access to secret funds for operations, was voted into law by a Democratic-majority Congress, with full Republican support. The only “no” votes in the House were Democrats – and one American Labor Party congressman. The hearings in the Committee on Armed Services illustrate the extent of bipartisan support for the CIA and for keeping its activities shielded from public and broader congressional attention. Chairman Carl Vinson (D-GA) and ranking member Dewey Short (R-MO) agreed that, in Vinson’s words, “We cannot have a Central Intelligence Agency if you are going to advertise it and all of its operations from the [ … ] Empire State Building. So the Congress just has to go along or else not have any confidence in us.”Footnote ⁵⁷ The subsequent bumps in attention coincide with Mansfield’s failed pushes for congressional oversight and the Bay of Pigs fiasco, respectively. These are followed by a surge of attention in the late 1960s peaking in 1975, appropriately dubbed the “Year of Intelligence,” before receding and peaking again in 1987 during the Iran–Contra Scandal. We also find that while congressional attention is on average higher after the reforms of the mid-1970s, there is substantial variation in the periods before and after Hughes-Ryan. We explore the post-1974 variation in greater detail in later sections.

The next step we took to create a measure of congressional attention that we can use to test for constraining effects on presidential behavior was to take all intelligence-related speeches and filter out those speeches that are likely irrelevant to congressional attention to the CIA itself, such as those in which members are merely incidentally referencing intelligence activities in the context of other issues in foreign or domestic affairs. To do so, we extracted these speeches and subjected them to unsupervised topic modeling. We converted the subset of the corpus into a Document-Feature Matrix, from which we removed standard English stopwords, symbols, numbers, and punctuation, as well as words that frequently occur in congressional speeches, but have low information value.Footnote ⁵⁸ Next, we used the R package stm, developed by Roberts et al. (Reference Roberts, Stewart and Tingley2019), to estimate a series of topic models on just this subset of speeches.Footnote ⁵⁹

A key challenge in unsupervised topic modeling is selecting the appropriate number of topics, or k. Existing scholarship offers some heuristics and useful diagnostic tools, but ultimately suggests that selecting the “right” k is more art than science. After trying models with between five and thirty topics, we settled on a fairly simple ten-topic model, as this produced topics that are easy to interpret, are readily discernible from each other, score highly on exclusivity and semantic coherence, and produce relatively small residuals.Footnote ⁶⁰ Perhaps more importantly, these various models all identify a single topic that captures speeches relating to congressional oversight of intelligence activities. Figure 5 plots the ten topics, in descending order of prevalence, and the three most distinctive words for each. A cursory glance reveals topics related to energy (10), Soviet nuclear capabilities and strategy (6), international communist subversion (8), and specific interventions like Nicaragua and Vietnam (5). Figure 6 is a word cloud of the 100 most distinctive words to load in Topic 1, the most prevalent topic in the corpus, which we are labeling the “oversight” topic.Footnote ⁶¹

Figure 5 Topic proportions, 10-topic model

Figure 6 Word cloud, oversight topic

3.2 Does Congressional Attention Restrain the Executive?

Having produced a measure that approximates the level of attention paid by legislators to issues relating to the oversight of intelligence and covert activities, we can now test our theory that higher levels of congressional attention can constrain the president’s use of covert FIRCs. In the remainder of this section, we offer a set of quantitative tests of our main hypothesis that:

• H₁: Increased congressional attention results in fewer covert FIRC operations.

This stands in contrast to other existing domestic-politics explanations of temporal variation in US covert FIRCs. As we suggested in previous sections, these alternative explanations can be summarized as:

• H_a1: When the government is divided (different parties control the Executive and Congress), presidents authorize fewer covert FIRCs;

• H_a2: Oversight reforms in 1974–75 drastically reduce the incidence of covert interventions henceforth;

  as well as a null hypothesis which holds that:

• H₀: Congressional attention and oversight have no meaningful impact on covert interventions.

We draw on two recent studies of US covert regime-change interventions discussed in Section 1 (Downes and O’Rourke, Reference Downes and O’Rourke2016; Smith, Reference Smith2019). Since we are primarily interested in the effects of domestic political factors on the propensity to engage in these covert interventions abroad, for our main analysis we use the panel data from Downes and O’Rourke, which includes all known instances in which the United States initiated covert FIRC operations from 1947 through 1989.

The variables of interest for this portion of the study are as follows:

• DV: Onset is coded 1 if the US initiates a covert FIRC operation against that country and 0 otherwise.Footnote ⁶²

• IV: Congressional Attention. We used the estimates from the topic model to produce an indicator of congressional attention to intelligence oversight in a year, calculating the percentage of intelligence-related speeches that relate to oversight. This variable then, is the estimated percentage of speeches related to intelligence oversight out of all congressional floor speeches in a given year, producing a number between 0 and 1.

• Domestic controls: Divided Government is coded as 1 if the congressional majority is from a different party than the president’s and 0 otherwise. Election Year is coded as 1 if it’s a presidential election year. Q1 Approval is the president’s mean approval for the first quarter of a given year.Footnote ⁶³ We account for president-fixed effects by including dummies for presidents from Truman to Carter, with Reagan-Bush as the excluded category.Footnote ⁶⁴

• Dyadic and Target Controls: Relative capabilities are the target’s capabilities (measured by the Correlates of War Composite Index of National Capabilities) as a share of US capabilities. Democracy is coded as 1 if the country is a democracy and 0 otherwise. Active Soviet Operation is coded as 1 if the country is the target of an ongoing Soviet covert operation according to Berger et al. (Reference Berger, Easterly, Nunn and Satyanath2013).

• Time controls: Peace years is the number of years since the last onset of a US-supported covert FIRC operation in the country, and its squared and cubic terms (Carter and Signorino, Reference Carter and Signorino2010).

Figure 7 plots the results from several models.Footnote ⁶⁵ Models 1 through 3 are logistic regressions with robust standard errors. The data for these models are structured as a panel, with the units of analysis being all possible target countries over time. Model 1 includes only Congressional Attention and time controls. Model 2 adds the additional domestic control variables described earlier. In all models, Congressional Attention is negatively and significantly correlated with the likelihood of covert FIRC onset. The fact that we find consistently negative correlations between congressional attention and covert action onset as our theory predicts also relieves concerns about potential reverse causation. After all, if attention itself were primarily or largely a product of past or current covert FIRC operations, we would see a positive relationship in the data. As we suggest in previous sections and reiterate in the cases in the next section, Congress can have their attention drawn to intelligence oversight for various reasons, including particularly spectacular operations, unrelated scandals involving intelligence agencies, and broader battles over foreign policy authority.Footnote ⁶⁶

Figure 7 Models of covert FIRC onset, coefficients with 95% CIs

Models 1 through 3 do not find statistically significant results for Divided Government or other domestic-level controls. This is consistent with Zegart’s conclusion that “Divided government does not appear to dramatically increase Congress’ incentives or ability to oversee intelligence agencies.” (Zegart, Reference Zegart2013,119)Footnote ⁶⁷ Model 3 adds dyad- and target-level control variables. Among these, only Soviet Operations is statistically significant, with the likelihood of US-backed covert FIRCs increasing substantially in countries experiencing Soviet operations. In some models, we find limited evidence for the effect of democracy in the target country and relative capabilities, but those findings are not robust.Footnote ⁶⁸

Next, in Model 4, we change the estimation technique to deal with a key potential issue with this analysis. Specifically, using panel data with observations at the level of the target country when the variable of interest is measured at the level of the initiator, the United States, might unduly inflate the number of observations without sufficient theoretical justification. In other words, for models without target-level covariates, it makes no difference if the covert operation targets Chile or France. And even the inclusion of those covariates changes little, as the characteristics of any individual country are reasonably exogenous to the changes in US politics.Footnote ⁶⁹ We suggest that a more appropriate analysis may be a time-series analysis, with observations only at the US level and the DV being a count of covert operations initiated in a given year, normalized as a proportion of the number of states in the international system that year. We present these findings in Model 4 in Figure 7. Since this new DV is a proportion, bound between 0 and 1, we can also model it with a Generalized Linear Model in the binomial family and a logit link 7.

We again find that even with a much smaller sample size the coefficient for Congressional Attention is still negative and statistically significant $(p<0.05)$, while other domestic-level variables fail to reach significance.

Figure 8 explores the substantive effect of Congressional Attention by displaying the predicted probability of onset of a covert FIRC operation (y-axis) conditional on the level of attention (x-axis) from Model 3, with other covariates held at their mean/median values. The effect is not only statistically significant; it is also fairly large. The estimated probability of onset when Congressional Attention is at its 90th percentile is less than one-third of the probability when Congressional Attention is at its 10th percentile.Footnote ⁷⁰

Figure 8 Effect of congressional attention on likelihood of covert FIRC onset, predictive margins with 95% CIs

In the next two sections, we offer causal process observations from declassified documents pertaining to the attempts to covertly intervene in the civil war in Angola starting in the mid-1970s, as well as additional evidence from later cases. We show direct evidence that heightened congressional attention constrained the president’s use of covert FIRCs during the Cold War, affecting where, when, how, and how much covert intervention took place.

4.1 Attention Fluctuates, Oversight Recedes

These dynamics persisted throughout the Reagan years, in a period of extensive reassertion of executive autonomy over foreign policy and through repeated attempts at circumventing or weakening Congress’ foreign policy role not only in Nicaragua but elsewhere around the world. The new administration had decided as early as 1981 that Qaddafi had to be overthrown – it was not a question of whether but how. Reagan’s more hawkish advisors, including William Casey and John Pointdexter, would find that statutes and regulations, especially the Ford and Carter prohibitions on assassinations, were practical impediments to some of their plans, but not necessarily impossible to get around. To do so, however, would require significant effort. Concerns with congressional opposition were partly responsible for repeated decisions not to escalate CIA support of dissidents in Libya, and to equivocate in aiding the abortive coup against Qaddafi in 1984. The Reagan administration also seriously considered a Bay-of-Pigs-style invasion by CIA-supported forces in Chad and Egypt – going as far as writing up the speech to be delivered by the president – but ultimately had to pull the plug on the operation after details of the operation leaked to the press.Footnote ⁹⁷ Some of the internal opposition to these plans came not from those who rejected opposing Qaddafi – the covert intervention to support the anti-Qaddafi rebels in Chad elicited little opposition within the administration or in Congress – but a concern that these attempts at unseating him would result in his death. Reagan administration officials dealt with these concerns not only by insisting that Qaddafi’s death was likely unavoidable but not intentional policy but also by limiting the direct lethal support for dissidents. During the overt intervention in 1986, when the United States bombed various government buildings, including Qaddafi’s primary residence and headquarters, they did so while insisting that the goal was not to kill the Libyan dictator. While it might appear to be an absurd denial of the obvious and there is plenty of evidence to believe that many in the government hoped that Qaddafi would perish in the attack, there is also evidence to suggest that steps that could have been taken covertly to increase the chance of that outcome were not taken due to a fear that details of the attempt would later come to light.Footnote ⁹⁸ Expected congressional opposition also contributed to the reluctance to intervene in Suriname between 1980 and 1987.Footnote ⁹⁹ Reagan considered lending support to a Dutch overt intervention in December 1986, but the operation was called off before he could make a decision.Footnote ¹⁰⁰

We find similar dynamics at play in Reagan’s and Bush’s initially timid approach to Panama’s Manuel Noriega. Relations with Noriega, a long-term collaborator in CIA anti-communist activities in Central America, deteriorated as revelations mounted about his various extracurricular activities. By the mid-1980s it was clear that he had been feeding false information to the CIA, playing all sides of the conflict, receiving money and weapons from the United States and from Cuba while in bed with drug cartels in Colombia and Central America (Robinson, Reference Robinson1988; Dinges, Reference Dinges1990). While both presidents counted on congressional support in their determination to see Noriega ousted – in the form of condemnations, indictments, and sanctions – both also struggled with congressional opposition to more direct and forceful covert efforts to overthrow him. While getting castigated in the public eye for not being forceful enough, senior White House officials repeatedly traded accusations with leaders of the Senate Select Committee on Intelligence, David L. Boren (D-OK) and William S. Cohen (R-ME). Brent Scowcroft, the president’s National Security Advisor, and Dick Cheney, his Secretary of Defense, complained that congressional “micromanagement” of covert operations, restrictive interpretations of rules against assassinations and kidnappings, and demands for real-time updates constrained the Executive’s options and ultimately prevented successive covert plans to remove Noriega (Engelberg, Reference Engelberg1989; Hoffman and Goshko, Reference Hoffman and Goshko1989).

The institutional impasse was only overcome when, in December 1989, Panama declared a state of war with the United States and Panama Defense Force troops attacked and killed a US soldier stationed in Panama City, providing the Bush administration the excuse it was awaiting to execute an overt invasion and extradition. Though the operation was planned in secret without consulting Congress, it involved the extraordinary rendition of Noriega and drew condemnations from the Organization of American States and the UN General Assembly for violating International Law. Nonetheless, congressional opposition was muted and public support was plentiful (Cramer, Reference Cramer2006, 194–196) This case is another useful illustration of how the White House can sometimes tame Congress and wrest control of foreign policy tools if it is free to use the bully pulpit to defend and frame its preferred policies, something it can easily do for overt, but not covert interventions.Footnote ¹⁰¹

The evidence from these cases also shows that partisan divides regarding covert interventions – as well as overt – grew wider in the 1980s, tracking the growing polarization and ideological sorting in American politics. Even then, though, this continued to be largely secondary to ideological and institutional disputes. The increased partisan polarization of covert action also owed, in part, to the greater attention to the topic itself, which in turn was often driven by leaks and disclosures from members of Congress. Senator David Durenberger, the Republican chairman of the Senate Intelligence committee in 1985, went so far as to say that oversight became a partisan issue when “covert becomes overt.”Footnote ¹⁰²

Even in this increasingly polarized environment, bipartisanship and ideological divides continued to trump pure partisanship. While key legislation on Nicaragua was mostly introduced by Democrats such as Tom Harkin and Edward Boland, the first Boland Amendment, amending the FY1983 House Defense Appropriations bill to prohibit aid to rebels with the goal of promoting regime change – and thus precipitating the Iran-Contra scheme – was characteristically opposed by conservative Democrats and supported by then nearly extinct liberal Republicans. The appropriations bill itself was adopted 411–0. A 1983 bill introduced by a Republican congressman to repeal the Boland Amendment failed to even come to a vote, as did Democratic-sponsored bills that would have closed the loopholes in the Boland Amendment and cut off all aid to the Contras. In 1984, after Reagan authorized the mining of Nicaraguan deep-water ports without notifying the Senate Select Committee, it was its Republican chair in the new Republican-majority Senate, Barry Goldwater, who became the public face of congressional condemnation and lent weight to the passage of the second Boland Amendment prohibiting support for all paramilitary operations in Nicaragua, regardless of their stated objective (Mayhew, Reference Mayhew2000, 95). Between 1981 and 1983, the Reagan administration dealt with these restrictions in their messaging, both publicly and privately with members of Congress, by denying that the goals of the campaign in Central America included overthrowing the Sandinistas. Rather, they insisted that the goal was primarily to interdict weapons going to El Salvador and Honduras and prevent the spread of violence in the region. Congress’ response was to allow aid to be directed openly toward interdiction and diplomacy, curbing the administration’s ability to use those funds for lethal assistance to the Contras. This policy of denial became harder over time as revelations of the scope of American covert involvement in the region mounted. By 1985, as we noted in the previous section, the “secret” wars in Nicaragua, Afghanistan, and Angola had lost all but the pretense of deniability, and the administration and their supporters in Congress mounted a full-fledged campaign for the authorization of lethal aid to the Contras. In February 1986, Reagan asked Congress for 100 million dollars to be funneled from the Pentagon for overt military assistance, including training and equipment. The House approved the request in March and the Senate in August, clearing the way for a major escalation of direct involvement in the conflict. The vote was split and narrow, but not along clean partisan lines.Footnote ¹⁰³

While the case of Angola illustrates how an attentive Congress can, through formal and informal means, limit the scope and goals of covert operations, other cases of intervention and nonintervention since 1980 further illustrate the continued tensions over executive autonomy and legislative oversight, the substantial variation in attention over time, and the importance placed on avoiding attention to preserve secrecy and deniability. As we will see in the next section, even in the unusual case of Afghanistan, where Congress found itself openly pushing for, not against, intervention, there is clear evidence that the White House, the CIA, and Congress often disagreed bitterly on the appropriate level of congressional involvement, and the risks and benefits of publicity stemming from the attention of Congress.