When Presidents Limit Bureaucratic Power: Evidence from Abortion Bans in Foreign Aid

Abstract

Presidents possess vast authority to change policy unilaterally. But their power depends on the compliance of unelected officials, who typically have more information. We examine the conditions under which presidents invest in costly oversight of the bureaucracy. We identify an underlying political process—the polarization of an agent’s potential principals—and argue that because this increases the potential for future policy losses for presidents, they create new means of overseeing agency behavior. To test this argument, we examine abortion restrictions in foreign aid, leveraging archival records and interviews with former officials. Most importantly, this allows us to study prohibitions that were proposed but not adopted. We find that as abortion became an essential part of party competition for the presidency, Republican presidents increased their investment in oversight, which eroded bureaucrats ability to moderate presidential power over abortion abroad.

Presidents act unilaterally with effects that transcend the borders of the United States. They impact migration flows, global supply chains, and the security conditions thousands of miles away—all without prior approval of Congress (e.g., Moe and Howell 1999; Howell 2003; Mayer 2001). But for presidential directives to be impactful, they must induce compliance from bureaucratic agents (Lowande 2018; Acs 2020; Turner 2020; Benn 2022).

Scholars have long acknowledged this is no easy task. In standard models of delegation, principals defer to agents with different views because those agents are better informed. Were they equally informed, principals could give their agents the amount of discretion that would make the principal best off, and nothing more (or less). But this information asymmetry is endogenous, and has implications for the future. As Gailmard and Patty (2013) show, it depends on the agent’s willingness to acquire expertise, and their ability to use that expertise over time. In this study, we investigate the other side of this asymmetry—the principal’s willingness to make costly investments in oversight, which change the nature of future interactions with bureaucrats.

To investigate the conditions that lead presidents to invest in oversight, we study abortion restrictions to family-planning foreign aid from 1965 to 2021. Since the latter half of the twentieth century, U.S. presidents have influenced the global availability of abortion—and family planning services, generally—affecting the health of millions of women in the process. Through their control over government grant-making and contracting (Gitterman 2017), presidents have set the conditions for the kinds of family planning and health services the federal government may sponsor. Though scholars frequently mention these restrictions as exemplary cases of presidential unilateral power, no study has investigated how those restrictions came to be, or whether they reflect the president’s true preferences.

We analyze thousands of documents from six presidential libraries and over ten hours of interviews with former USAID officials who served under both Republican and Democratic administrations. This approach offers two key sources of inferential leverage. By examining the implementation and reversal of similar restrictions over time, we hold constant the policy space in question while varying the political circumstances of interest. Most importantly, our qualitative approach reveals counterfactual policies that go unobserved when one looks solely at the (public) policies signed by presidents. It allows us to pinpoint where the president and USAID wanted, but failed, to move policy. In short, if one looked only at the published record, it would appear presidents got what they wanted. As it turns out, this is only partly true.

We argue that the polarization of political issues provides an explanation for why the current president would invest in oversight, even when that oversight might be used by a future president. For presidents facing a distal bureaucratic agency, polarization increases policy losses they expect in the future. They hedge against those losses by investing in new means of oversight, which most importantly, are likely to reduce the information gap between the agencies and all its principals, not just future presidents.

Our evidence is consistent with this argument. Early in our period of study, unelected officials can and did limit the power of presidents over abortion. Presidents’ proposals were either negotiated away, compromised, or moderated. As abortion became a more polarized issue, these policy loses significantly increased. Thus, while bureaucrats’ preferences were stable (and left-leaning), Democrats and Republicans polarized, which generated incentives for Republican presidents to invest in oversight. This resulted in a variety of presidential efforts to increase the accountability and monitoring of officials in the Office of Population within the U.S. Agency for International Development (USAID). After decades of such investments, presidents of both parties now impose, and bureaucrats comply with, abortion rules in foreign aid.

Our research has important implications for the study of presidential power, bureaucratic behavior, and American political development, more broadly. Though much has been written about presidential unilateral power (see Lowande and Rogowski 2021), observing the failed proposals and negotiations within the executive branch is rare (but see Rudalevige 2021). The Mexico City policy illustrates how the apparent success of presidents to propose new, unilateral policies is sometimes moderated by bureaucrats—often in ways unseen by initial observers (Lowande 2024). Second, and relatedly, our theory and evidence illustrate a developmental process that other policy areas might follow. Our framework suggests that emerging areas of political conflict initially result in presidents experiencing policy losses, but over the longer term, motivate investments in oversight that significantly change future presidents’ interactions with bureaucrats.

Finally, our study highlights the importance of religion and gender in the study of American political development (e.g., Skocpol and Ritter 1991; Canaday 2009; McDonagh and Nackenoff 2015; Teele 2018). We demonstrate how unelected officials were able to advance a family planning agenda and moderate limitations on reproductive rights funding, which altered the strategies of family planning organizations and the experiences of women worldwide. Political conflicts over abortion restrictions also significantly shaped the development of USAID, along with both secular and religious-affiliated non-profit organizations. Thus, our study illustrates how the study of religion and reproductive rights improves scholars’ understanding of American political development.

Theory

Interactions with bureaucrats present a classic challenge for presidents. Unelected officials can help the president implement desired policies without prior approval from Congress or Courts—but bureaucrats also have agency, and can drift. This is a well worn problem with standard answers about how presidents, in particular, resolve it.

The most obvious tool might be to change the unelected officials doing the implementing. That could be done by appointing loyalists to the agency in question, or “politicizing” it. Lewis’ (2008) definitive work on this strategy shows that presidents balance the trade-off between shared policy goals and competence. Another might be to let officials within the president’s immediate orbit make the decisions—or to “centralize” policymaking—which Rudalevige (2002) argues is limited mostly by presidents’ need for relevant information. A related answer particular to executive action is to mimic Congress: attempt to write a complete contract by including provisions within presidential directives that attempt to limit the agent’s discretion (Lowande 2018; Acs 2020; Turner 2020). Recent empirical work by Benn (2024), for example, measures the degree of discretion granted in recent executive actions, which suggests that (at a minimum) this behavior is quite common.

Each of these strategies hinges on, and is limited by, the information asymmetry between presidents and bureaucrats. Lewis, for example, argues that presidents balance the need for effective government and political control, which is driven by the fact that career bureaucrats typically have more information. But this asymmetry is also endogenous—it is a function of choices by both presidents and bureaucrats. Gailmard and Patty (2013) show that the (costly) decision to acquire expertise by bureaucrats is, in part, determined by the policymaking latitude and autonomy they enjoy. What interests us is the flip-side of this informational gap: When do presidents invest in the capacity to conduct oversight?

Presidents and their administrations have a finite amount of effort to spend while in office. Oversight is any effort by a principal spent reducing the informational asymmetry between themselves and an agent. An “oversight technology” then, is any costly structural action that makes that effort more efficient and effective. For presidents, the canonical example is executive branch budgeting. Prior to the centralization of budgeting in the Bureau of the Budget and later the Office of Management and Budget, there was nothing in principle that limited presidents from expending effort to observe the budgeting process of a line agency. What changed was that the president now had a new, structural means of learning about the agency’s proposed budget figures. In this instance, the oversight technology also came with a shift in the locus of decision-making about the budget. In other words, oversight technologies often complement the centralization of policymaking.

Politicization, meaning the appointment of loyalists in agencies, can also enhance future oversight. The appointees themselves take actions that make their own efforts to get information more effective. They might change reporting procedures, shift responsibilities to adjacent bureaus, set-up routine audits of programs, or consolidate duplicative functions to make bureaus less independent. These efforts are all costly in the short run, especially those that involve reorganization. But when appointees and the presidents they serve want to conduct oversight in the future, less effort will go farther toward reducing the information gap between bureaucrats and their principals. ¹

Finally, so-called “fire-alarm” oversight presumes an underlying structure that qualifies as an oversight technology. Interest groups are supposed to make oversight cheap and easy, by alerting principals when bureaucrats step out of line (McCubbins and Schwartz 1984). In other words, they minimize the effort needed to monitor agencies for misbehavior by identifying that misbehavior on the principal’s behalf. But that assumes an interest group environment where the informational gap between outside groups and government bureaucrats is sufficiently small. Otherwise, the president will not be able to free-ride on their effort. “Fire-alarms” are useless if there is no one to there to pull them. Typically, examinations of interest-groups and the bureaucracy start by presuming a particular interest-group environment (e.g., a “dominant” or contested one, à la Moe [1989]). Here, we argue it is possible for presidents to encourage the development of an interest group environment that reduces the effort needed for future oversight.

Specifically, presidents attempts to control policy can have second-order effects that help interest groups meet their needs as organizations. For example, if the president successfully influences the grant-making and contracting behavior of agencies, those resources help favored interest groups survive and grow. Even when the president does not successfully influence policy, they can grant interests access, which helps leaders demonstrate their worth to donors and followers. A noteworthy example beyond our study is the George W. Bush administration’s faith-based initiatives, which sought to channel federal spending to religious-affiliated organizations that wanted to run social welfare programs. The fact that these organizations received these resources meant they would exist into the future. The fact that the president signed executive orders and met publicly with them helped them to show their supporters that they were successful. And, as a result, the interest group environment in the future is changed. So, too, is the president’s capacity for oversight, because the groups capable of pulling the fire-alarm exist. Put differently, this oversight technology is a specific example of how policy makes politics.

This all points to another complication. Investing in the capacity to oversee an agency is not just costly up front. The payout is sometimes difficult to keep exclusive to the president. When presidents set up structures that help them monitor the bureaucracy, they give other principals an opportunity to leech. Moreover, presidents must consider the fact that they will eventually not be the president. In other words, akin to Gailmard (2009), we argue that oversight technologies present collective benefits with localized (in this case, upfront) costs to the president.

This dynamic is obvious in each of the examples we have mentioned. The routine preparation of agency budgeting and rulemaking dockets ² makes it easier for all external actors to learn about agency activities. Suppose, for example, a political appointee consolidates the program evaluation roles played by multiple bureaus into a single office. This makes it easier for the appointee to conduct oversight inside their agency, but it also makes it easier for congressional investigators, who now also have a single bureau to query. They create a consolidated paper trail that can be accessed by outside parties. Likewise, when the president’s actions foster “fire-alarm” pulling interest groups, these groups communicate publicly about the activities of agencies. Even when they do not, the interest group landscape in the United States is sufficiently politically balanced that complaints come from all sides.

Under what conditions would presidents pay these internalized costs, for a benefit that is non-exclusive and delayed? We argue for two necessary conditions. The first condition is distal preferences. The president and the bureaucrat need to have known disagreements. If the president and the bureaucrat had the same vision about policy, the information asymmetry would not matter to the president—the bureaucrat would use their expertise to act as a faithful agent. The president can hand down a directive, provide the agent with wide latitude, and which would make the president better off. For investing in oversight technologies to be worth the cost, at a minimum, the president must expect some significant agency loss by relying on the bureaucrat.

But we argue that this condition is typically not sufficient. A loss on any one policy will not be worth the administrative headache needed to build up the capacity to monitor an agency. What pushes the president over the line, in our view, is the prospect of future losses—and in particular, the losses the president expects will become more significant. In the contemporary era of American politics, we argue there is one secular trend that drives this changing calculus: polarization.

More specifically, we argue there is a second necessary condition, the polarization of competing principals on the issue in question. To be clear, this differs from the classic observation that partisan voting in the U.S. Congress has monotonically increased since the mid-1970s (e.g., McCarty 2019). We define the polarization of potential principals from the perspective of the bureaucrat to mean that the two major parties (i.e., the potential principals) hold more zero-sum, more internally disciplined views on the issue area implicated by the choices of the bureaucrat. ³ Either party may control the presidency or Congress in the future. That means that either party, in turn, may hold the presidency and give orders to the bureaucrat. Likewise, either party may be on the outs, and have to attempt to influence the agency from Congress, or more informally.

When polarization is relatively low, and the president and bureaucrat disagree, the current president does not stand to gain anything additional from investing in oversight. The fact that the presidency can direct the agency, and retains a privileged informational position relative to other actors, matters less because the next president—even when they are from another party—will not be giving radically different orders. Moreover, the current president can expect that the future, opposite-party president will incur a similar degree of agency loss.

But when polarization is high, the potential loss due to bureaucratic agency bites now, and in the future. This motivates presidents who disagree with those bureaucrats today to ensure that future oversight will be more efficient and effective—for all political actors. Presidents who disagree with the implementing bureaucrat must always confront the possibility that the bureaucrat will have a new chief executive in the future. That future president may not share their views precisely, and may be from the other side of the political spectrum. When the ideological gap between the current president and that future president grows, the risk of allowing that future president to command the bureaucracy grows. But the current president can do something about that risk. They can set up oversight mechanisms that effectively diminish the bureaucrat’s informational advantage over all actors. They ensure that even if they are on the outs, their allies will have the capacity to monitor and perhaps influence the decision-making of the bureaucrat in the future.

For example, in the case we will describe, Republican presidents beginning with Reagan faced a distal agency, whereas Democrats were ideologically proximate to that agency. Why would these Republican presidents invest in oversight which they know that a Democratic president could use in the future? According to this scheme, there are two answers. The first is that the Democratic presidents did not need oversight to get what they want. When Democrats and bureaucrats largely agree on the proper policy, Republican investments do not make a future Democratic president better off. The second reason is that some of the benefits of new oversight technologies are non-excludable. In this case, this helps Republican presidents ensure that even if they are out of power, their allies will have better access to information.

In this way, both preference misalignment and polarization of principals are necessary. Presidents who agree with the bureaucrat would prefer that the bureaucrat retain an informational advantage over their opponent and all outside actors. Conversely, presidents expecting consistent agency loss have the incentive to diminish the bureaucrat’s informational advantage for all principals, now and in the future.

Research Design and Case Selection

Typically, observing a developmental process like the one we describe is difficult, which makes a longitudinal study of abortion restrictions on foreign aid ideal. The Mexico City Policy, known by opponents as the “Global Gag Rule,” prohibits foreign organizations from receiving federal funds if they promote abortion as a method of family planning. As table 1 shows, since implemented by President Ronald Reagan in 1984, it has been rescinded by every Democratic president and re-established by every Republican president in the first days of their term.

Table 1 – Observed Abortion Prohibitions in Foreign Aid

This gives us the variation needed to assess our theory. First, it gives us variation on the preference alignment of the bureaucrat and president. Using our data, we can approximate the preferred policies of the president and the bureaucracy. As the former changes over time, so does the ideological proximity of the president and the agent. Second, it gives us variation on political polarization, as abortion politics has polarized in the United States from 1960 to the present. Thus, by focusing on one policy over time, we are best able to hold constant all policy-invariant factors that might influence these relationships, and instead focus on how the change in our explanatory variables influenced the oversight investments exercised by presidents over bureaucrats.

The Mexico City policy is sometimes thought of as a textbook-case of unilateral power (e.g., Cooper 2002; Dodds 2013; Lowande 2014). However, scholars have focused on the policy proposed, not on policy development or implementation. The policy affects reproductive rights globally, and thus, has been studied in the field of public health. These studies document the change in U.S. policy (Fox 1986), and focus on the effect of the policy on births, maternal death, abortion, and contraception use—mostly in Sub-Saharan Africa (e.g., Jones 2015; Brooks, Bendavid, and Miller 2019). Studies argue that abortion and birth rates rose, while contraceptive use declined, as a result of the policy. Thus, previous research has focused more on outcomes, rather than on policy development or implementation. By including important counterfactual information on the policies that could have been enacted, and highlighting the procedure by which it is implemented, our study fills significant gaps left by earlier work.

Evidence

We present evidence from presidential archives and elite interviews. Each source has its own limitations. The Presidential Records Act effectively embargoes all relevant records for twelve years after U.S. presidents leave office. This means that archival coverage only runs through the second Bush administration. For each open archive, we contacted archivists and searched all indexed materials for records relating to foreign aid and abortion. Naturally, the vast majority of records are irrelevant for our analysis or duplicate what is available outside of archives. To render this approach more transparent, we make all documents used in the following descriptions available in a separate data repository. The advantage of these records is that most documents contain private commentary and recommendations of presidential subordinates, in a historical record unadjusted by memory or retrospective biases.

Alongside these materials, we conducted elite interviews of former USAID officials who had a role in implementing the policies. We contacted every living, former USAID administrator for an interview. We then proceeded with “snowball” sampling, and obtained more interviews by referral. Ultimately, we interviewed eleven former officials for a total of over ten hours. Lower-level officials served in both Republican and Democratic administrations, whereas we were able to speak with one USAID head appointed by a Democratic president—and two appointed by Republicans.

We were careful to avoid common pitfalls in elite interviews. Most relevant in our case are present-day claims by former officials that may make their actions look better in the “eyes of history.” In particular, it may be that these officials have an incentive to present their efforts to moderate the moves of conservative presidents as more successful than they actually were. There are two ways to guard against this particular bias. First, by speaking to both conservative and liberal officials, we can validate accounts when they are told similarly by both sides. Second, we check them against archival materials and public records. Finally, if we are unable to do either check, we can exclude the material as hearsay or note it in our summary. These are the rules we follow in the following analytic narrative.

Background: Family Planning in Foreign Aid, 1964–1983

Prior to the first presidential directive in 1984, the issue of family planning in foreign aid was shaped by large informational gaps between presidents and unelected officials. The result was largely unchecked bureaucratic discretion. In the 1960s, demographers argued that population growth would hinder economic development and lead to significant food shortages and unrest worldwide. These concerns reached the White House, but it responded with caution, leaving agencies alone to implement their preferred policies.

In 1965, for example, Horace Busby, a close advisor to President Lyndon Johnson, wrote to the president with concerns about the number of “women in the child bearing” years, arguing that the administration ought to discuss the consequences of unchecked population growth. ⁴ However, political aides advised President Johnson to remain strategically silent on whether population growth should be addressed by contraception and family planning services. ⁵ Johnson took this advice, making general public statements about the “population problem” without policy prescriptions, behavior that he maintained until the end of his presidency. ⁶

Even as the international community in the West came to a consensus that global population was a problem, and key advocates successfully introduced the issue to officials at the White House and Department of State, the president and his advisers believed a public stance might reduce their policy options. One memo to the president is particularly revealing:

The whole area of family planning, birth control, and population growth is marked with potential political booby traps. … . It has not been necessary to join “ideological” battle. Getting the President involved in an ideological fight might arrest progress. (emphasis original) ⁷

Johnson established a Presidential Commission on Population, but only as he was leaving office and had decided not to run for re-election. The commission, which mostly focused on family planning domestically, turned in its report less than two months before the president left office. Given that many of its participants were outsiders, it is unlikely the report reflected the president’s preference. ⁸ It could have developed into a presidentially-controlled unit tasked with overseeing family planning policies, but it was not organized again. In line with our theory, LBJ did not build the capacity for oversight.

Officials at USAID then had wide latitude to implement policy, and began aggressively pursuing family planning services as a means of limiting population growth. This led to the establishment of the Office of Population within USAID in 1969, which was headed by Reimert T. Ravenholt—an epidemiologist described by many as a family planning “zealot.” ⁹ It focused on contraception to the exclusion of alternatives. Former USAID officials still refer to this as the “Ravenholt approach.” ¹⁰ Moreover, as a later head of the Office of Population put it:

[Ravenholt] had enormous power, unlike that of any other office director of a central program in USAID, because he had the power to hire and fire field officers working in the program. It was the only technical program in USAID that was not managed, geographically by the regional bureaus. ¹¹

The Office of Population began to fund the development of contraceptives, advocate for sterilization experiments, ¹² and develop new methods of abortion. ¹³ More specifically, menstrual regulation kits (MRKs)—a handheld, disposable means of abortion were developed under contract of the Office of Population and were widely distributed. ¹⁴

These policies were eventually curbed by conservatives in Congress, not the White House. Outraged by the funding of MRKs, Senator Jesse Helms (R-NC) introduced an amendment to the Foreign Assistance Act in 1973, which would prohibit the direct funding of abortion. Months after the Supreme Court’s decision in Roe v. Wade (410 U.S. 113, 1973), the Helms amendment prohibited the use of foreign assistance to pay for abortion as a method of family planning, or to coerce any person to practice abortion. Every USAID official interviewed stated that they complied with the Helms amendment, even if they were opposed to it. However, most agreed there was some time before they settled on what an abortion service was, and that even then, there was “grey area” in the prohibition. In this initial period, USAID retained the ability to implement policies it desired because officials like Ravenholt used information gaps to their advantage, with little interference or oversight from presidents.

The polarization of Republican and Democratic elites over the issue of abortion, in general, was relatively low. As O’Brian (2019) writes, before the 1980 election, “abortion was a Catholic concern, and Catholics were Democrats”(1043). LBJ, Nixon, Ford, and Carter all took moderate positions, resisted lobbying from the extremes, or did their best to avoid the issue entirely. This was particularly true of abortion policy relating to federal funding, which is what the Mexico City policy would later implicate. Adams (1997), for example, documents a gradually widening partisan gap in abortion votes in Congress since the 1930s. But he excluded “gag rule” related votes like the Helms amendment because “even some adamant pro-life legislators viewed the issue in terms of free speech and the sanctity of doctor-patient relations”(722). In short, even though competing principals held the presidency over this period, there was no partisan imperative to radically change family planning policy in foreign aid. So presidents neither imposed explicit policy directives, nor did they invest in oversight.

How Bureaucrats Moderated the Mexico City Policy

As abortion policy mapped on to party competition over the presidency, the behavior of USAID became more politically costly to social conservatives. As O’Brian (2019) shows, popular polarization over abortion existed prior to the Reagan administration, but it was this period—and in particular, the election of 1980—which saw elite officials leverage that popular polarization into an electoral strategy. More specifically, the Christian Right began to mobilize on the abortion issue, and the Reagan campaign saw conservative Catholics who traditionally voted for Democrats as a part of the coalition that put him in the White House. Officials at the Office of Population had been supporting family planning services while treading as close as possible to activities prohibited by the Helms Amendment. In the Carter administration, they permitted organizations that performed abortions to continue receiving funds, as long as the funds themselves did not go toward abortion services.

In response, the Reagan White House, attempting to incorporate conservative Catholics and Evangelical voters in their voting coalition, took measures to limit and supervise how foreign aid could be used in family planning. At this stage, however, bureaucrats still held a large informational advantage that forced the White House to stray from its initial policy preferences. Following the development of the initial policy, the administration invested in novel forms of oversight which reduced the informational advantage of USAID in the future.

Policy development. Reagan wished to expand the prohibitions of the Helms amendment. The 1984 World Population Conference in Mexico City gave him the opportunity. John A. Svahn, Assistant to the President for Policy Development, chose Carl A. Anderson to oversee writing the Mexico City statement (Fox 1986). Anderson was a former congressional staffer for Jesse Helms, and at the time, taught part-time at the Pontifical Lateran University—known colloquially as the “Pope’s University.” ¹⁵ The White House then chose James L. Buckley, former Senator from New York, to head the conference delegation. ¹⁶ Buckley led the 1973 attempt to pass a constitutional amendment that would have applied the Fourteenth amendment to human embryos (H. J. Res. 769, 93rd Congress). According to the Administrator of USAID at the time, “there was lots of pressure on the White House on this, and Buckley was a very trusted person.” ¹⁷

But the leadership of USAID was chosen for reasons that had nothing to do with abortion. M. Peter McPherson, a lawyer for the Reagan transition team, was chosen as USAID administrator because of his experience in the Peace Corps. At the time, USAID was the major funder of family planning programs internationally. ¹⁸ Republicans believed they were interpreting the Helms amendment in ways that allowed the United States to indirectly fund abortions. In the words of one former legal counsel under Reagan,

They came to us and said, words to the effect of “We can’t do abortions. We know we can’t do abortions. We want this program here to train doctors on how to do abortions. Now, that’s not doing abortions, you understand. We’re just giving them medical training.” There’s a red line that Congress had drawn. And, the professionals wanted to get as—they wanted the lawyers to tell them, “How close can we get to that red line without crossing over it?” ¹⁹

The Department of State and the National Security Council (NSC), while not involved in the abortion debate, considered population to be a national security issue. The NSC received memos from the CIA recommending long-term strategies to control global population growth. ²⁰ It was their concern that growing populations in developing countries would destabilize governments, leading them to pursue authoritarian forms of government and partnerships with the Soviet Union.

Disagreements between the White House and bureaucrats led to several weeks of draft negotiations. The State Department produced the original draft policy. The White House edited the original, concerned its emphasis on the urgency of the population problem would lead to increased spending on family planning. In a memo to Svahn, he wrote

The arrangement of certain paragraphs has been changed and some words deleted in order to … reduce the sense of crisis that pervades sections of the statement. Also gone are terms such as “essential priority element” and “urgent” which seem to establish a basis for budget increases. ²¹

The next day, the draft was returned to the State Department, now including an explicit abortion prohibition:

The United States … does not consider abortion an acceptable element of family planning programs and will not contribute to those of which it is a part. Nor will it any longer contribute directly or indirectly to family planning programs funded by governments or private organizations that advocate abortion as an instrument of population control. ²²

USAID, State, and some at the NSC did not want any kind of abortion restriction. It was absent from their early drafts. ²³ The policy proposed in the White House draft would have prohibited family planning funds for any organization that provided abortions. This meant international NGOs, foreign governments, and domestic NGOs. Officials at the NSC, State Department, and USAID objected strenuously on the grounds that such a provision would infringe on the sovereignty of other nations. ²⁴

This was the first aspect of the White House proposal the others sought to remove. According to McPherson, “the White House had a draft that said we couldn’t … give money for family planning generally to countries whose official policy was to support abortion. And I did go to the White House and get them to exclude the countries [governments].” ²⁵ In a hand-written note attached to his own draft, Richard Levine, Director for Policy Development at NSC, wrote, “I understand from [McPherson] that the original language on no US aid for countries or groups that practice abortion has been removed from the Svahn draft.” ²⁶ Buckley continued to insist that the prohibition be included in the statement, and threatened to exit the delegation several times. ²⁷ It was not included, but he headed the delegation anyway. The preferences of USAID, NSC, and the State in this instance, overrode those of the White House.

The officials working outside the White House were also able to change the White House’s proposed position on population growth. The White House statement was dismissive of the dangers of population growth and the utility of family planning programs. The language was eliminated, with less negative and more concise wording replacing it. ²⁸

Greater restrictions were, however, imposed on other family planning funding. Abortion opponents considered the United Nations Population Fund (UNFPA) a backdoor means of funding abortions. There were reports of forced abortion and sterilization as a result of China’s one-child policy. Some thought that UNFPA’s programs in China supported the policy. In the first drafts, UNFPA was left out entirely. In the final draft, UNFPA would be required to show it was not engaged in abortion or coercive family planning programs, or it would lose its largest single funder. ²⁹

Policy implementation. Policy moderation did not stop at the writing stage. As written, the statement applied to “non-governmental organizations” that conducted family planning programs abroad. This left open whether NGOs headquartered in the United States would be subject to the policy. Shortly after the conference, officials at USAID solicited legal advice from the Department of Justice on potential litigation from U.S. NGOs. The strongly worded advice indicated that the government would likely lose in court on first amendment grounds, and advised USAID to exclude U.S. NGOs from the policy.

We should also advise [US]AID that the degree of risk of an adverse court decision depends on whether the regulations reach the speech and conduct of U.S. and U.S.-based NGOS or only foreign NGO’s. The risk would be highest if, for example, [US]AID wrote regulations which would disqualify U.S. NGO’s who use private funds to speak in favor of abortion in foreign countries as a method of family planning. ³⁰

USAID followed this recommendation, and the advice proved correct, as the final policy survived court challenges. Domestic NGOs were not the only ones spared, as USAID continued funding of UNFPA for one more year—to the surprise of the White House. UNFPA refused to end its support of China’s population program, a pre-condition for continued funding in the Mexico City policy. McPherson ordered a comprehensive internal review of UNFPA activities, and withheld just $10 million dollars from the earmarked $46 million destined for UNFPA. ³¹ His public position was that since UNFPA itself was not engaging in involuntary family planning, removing the whole $46 million would not stand up in court. Only subsequent amendments to the Foreign Affairs Act in 1985 (Nowels 2000) resulted in the defunding of UNFPA. ³²

Ultimately, just two organizations with existing USAID contracts—International Planned Parenthood Fund (IPPF) and the Pathfinder Fund—lost funding. Even in the case of IPPF, its affiliates in Latin America (where abortion was mostly illegal) continued to receive and accept USAID funding under Mexico City.

Oversight investment. Most importantly, the Reagan administration identified shortfalls in oversight capacity, and took steps so that both it, and future presidents, would have more information in the future. Republican appointees in the agency began requesting more information from organizations that publicly acknowledged funding abortions with their own funds. The agency’s contract with Pathfinder Fund (now Pathfinder International), for example, came up for renewal. Pathfinder claimed 3% of their funds were spent on abortion services, and that they had set up a separate accounting process for those funds to comply with the Helms amendment. Derham suspected this figure had been significantly deflated by creative accounting, and requested a more detailed breakdown of expenses so to verify they’d been following Generally Accepted Accounting Principles. ³³

They also participated in an external review by the Department of Justice, a departure from past administrations. It seemed only to reveal how little direct control the Secretary’s office had over the agency’s worldwide programs. More specifically, it was unclear how abortion prohibitions like those generated by the Mexico City conference would be carried out, which grantees and contractors they would apply to, or when they would apply. Legal counsel at the Department of Justice cleanly identified the main issue:

Although the books of the 40 prime grantees and contractors and 1,800 sub-grantees and sub-contractors must be made available to [US]AID auditors, [US]AID apparently has only about three auditors to conduct such audits. While in theory audits may also be conducted by [US]AID’s Inspector General, who has regional offices, the IG’s chief mission is to uncover internal waste, fraud, and abuse. Similarly, the audit rights of the General Accounting Office apparently are not regularly exercised. Therefore, [US]AID must rely on good-faith compliance by the primes and subs. ³⁴

Compliance with the final policy would require USAID career officials who opposed the policy to implement it faithfully.

In a move that helped address this gap, the Reagan administration supported the development of a broader non-profit community that could render “fire-alarm” oversight more reliable. At this time, the international NGO community had not polarized over abortion, as even segments of IPPF (mostly in Latin America) still sought and accepted USAID funding under the Mexico City policy. But the Reagan administration channeled foreign aid dollars to Catholic non-profit organizations. The best example related to family planning is the administration’s insistence that the Office of Population fund providers of “natural” family planning, who were often opposed to any kind of contraception. By the end of the Reagan administration, the Secretary of State’s Office and the White House was fielding regular complaints from religious charities that USAID dollars were being used in violation of the Mexico City policy. These were then passed on to McPherson.

The Mexico City policy also had a structural impact on non-profits who had provided abortions. At the time Pathfinder Fund came up for renewal, it relied on USAID for the vast majority of its revenue. McPherson strongly suggested it remove the most vocal proponents of abortion from its board, and it complied (Fox 1986, 623). It also cut ties with abortion providers in countries where the practice was legal.

In summary, with a new Republican administration opposed to abortion, the administration took steps to restrict funding. Yet given large information asymmetries, the final policy and outcomes differed from the preferences revealed by the original White House drafts in several ways, even as it shifted the status quo to the right. The Republican appointees in USAID could see that the sprawling agency would only present problems for enforcement, under the backdrop of an abortion issue sorting along party lines at the highest levels. Oversight on the part of the presidency and other actors was needed to limit future policy loss, so the administration invested in structural changes that would eventually make it easier to oversee the agency in the future.

Clinton and the Cairo Shift

George H.W. Bush took over the presidency four and a half years after the conference, and stood firm in his support for the Mexico City Policy and the accompanying Kemp-Kasten amendment. United Nations officials, family planning organizations, and some members of Congress pushed for an alternative formula to save the UNFPA from being completely defunded. But the president promised to veto any law that would weaken the policies, ³⁵ a promise that he kept when he vetoed the FY 1990 Foreign Operations Appropriations bill. ³⁶ The oversight infrastructure that Reagan had developed was enough to address any problems within USAID that he could foresee. Instead, the White House focused its attention on keeping Congress in check and established a precedent by taking a hardline stance in support of the policy. ³⁷ Thus, the Mexico City Policy had been in place for almost ten years when Clinton won the presidency.

On January 22, 1994, Clinton rescinded the policy via presidential memorandum. The next eight years solidified abortion’s polarization across party lines. A key example of this polarization came with changes to the discourse around population issues. It would no longer center on the economic threat of a population explosion, but instead, on a women’s right to bodily autonomy. The Clinton administration and USAID officials tended to agree on abortion issues, but notably, the agency’s remaining autonomy still led to some minor policy loss. As expected, however, the administration did not invest in new oversight technologies—in fact, they did their best to ensure the agency would have more information by providing them more unrestricted funds for population programs. This came in response to challenges from conservatives in Congress.

During this period family planning itself was reframed as a womens’ rights issue, which further mapped it onto existing partisan cleavages in the United States. The International Conference on Population in Cairo in 1994 presented Clinton with the same policy opportunity the Reagan administration had. The United States’ message on population no longer focused on the population growth “problem,” but on women’s control over their bodies. This Cairo shift was mainly brought by transnational women’s activists who had minimally organized around international family planning in the 1960s and 1970s, but changed direction, in part, due to the Mexico City Policy (Higer 1999). It reflected an international shift over the discourse over abortion and family planning policy as “the voluntary nature of family planning, for one thing and the right of women to have a full menu of reproductive rights…became very, very important.” ³⁸ This linked family planning not only with abortion, but the broader women’s rights movement. Notably, there was a 9-point gap in feelings towards the womens’ liberation movement across Republicans and Democrats in the 1980 ANES. But by 1992, it had widened to 15 points (ANES 1980, 1992).

But the final policy statement differed from the Clinton administration’s objectives in a notable way: it limited the extent the administration publicly supported abortion. The president took strong positions supporting the provision of abortions at military hospitals. ³⁹ Timothy E. Wirth, then Under Secretary of State for Global Affairs at the State Department, “was very insistent on pushing the U.S. Government to say that abortion was a method of family planning.” ⁴⁰ As J. Brian Atwood, then-Administrator of USAID later told us, he “saw that as a real threat to our family planning program, and so … strongly opposed him.” Here, USAID officials prevented policy movement in a more liberal direction.

By 1996, Republicans had control of both chambers in Congress. International abortion policy had become domestically salient and sorted along party lines. Republicans imposed greater restrictions on USAID’s family planning program. While USAID officials were able to initially negotiate the terms and creatively find solutions to the funding limitations, they were significantly affected by budget cuts and new limitations imposed by Congress.

House Republicans tried to reinstate the Mexico City Policy in the FY1996 Foreign Operations Appropriations, but there were not sufficient Republicans in the Senate to move the policy forward in anticipation of a presidential veto. ⁴¹ For FY1997, Congress tried again, and for the third year in a row, the conflict over abortion bans on international family planning funds delayed the foreign operations spending bill until the end of session. ⁴² They succeeded in limiting USAID’s family planning program. Originally, the bill provided $356 million for family planning activities (a 35% cut), “but none of the money could be spent until July 1, 1997, unless both chambers of Congress voted to release the aid by Feb. 28.” ⁴³

Atwood drafted and Clinton signed a determination concluding that the delay in funding would have “significant negative impacts on the proper functioning of the U.S.-supported international population program.” ⁴⁴ Congress agreed to early release of the money, and increased the budget to a total of $385 million, but it implemented a system in which “funds were to be disbursed at a maximum rate of 8 percent a month.” ⁴⁵ This was known to USAID officials as “metering,” a practice that continued through FY1998.

On its face, this practice generated significant obstacles for the contracting and grant operations of the Office of Population. It is typical for foreign aid programs to provide substantial financial commitments, up front, to enable contractors and grantees to pay for start-up costs. Some up-front costs are unavoidable, and cannot be defrayed on a month-to-month basis, which meant that many who would normally receive funding could not operate. There is no public accounting reasoning for this outlay structure. In the words of Scott Radloff, then Deputy Director of the Office of Population,

It’s all for pain—to inflict pain on the program. [They] couldn’t impose Mexico City. … we thought they were going to cut our funding. That’s what you usually do. … Well, they did that, then they added this metering. … I don’t know any other program that’s had to work with those conditions. ⁴⁶

Despite this, officials at the time adapted to the practice to prevent it from impacting their programs. This required the help of the USAID director and very careful accounting. Atwood authorized the use of roughly $70 million in discretionary funds controlled by the administrator, which allowed them to fund programs most in need of up-front commitments, as well as partly compensate for the budget cut. ⁴⁷ Margaret Neuse, Deputy Director of the Office of Population and Reproductive Health at the time, described how this was done, along with its effects: “[Y]ou would have to figure out which programs, which countries, which projects are going to need what money when so that they can keep going … they thought this was going to be a really crippling thing for the program.” ⁴⁸

After the imposition of significant administrative obstacles, USAID used creative accounting to limit the effects of political intrusion. Yet Congressional interference did not stop there. In 1999, Congress passed what came to be known as “Mexico City Lite.” The provision, which expired at the end of fiscal year 2000, “barred any of the $385 million in U.S. international family planning assistance for organizations that perform abortions—except in cases of rape, incest or where the life of the woman is in danger—or lobby to change abortion laws or government policies in other countries.” ⁴⁹ The president had vetoed a bill that contained this language the previous year. ⁵⁰ It was “lite” because it still allowed organizations that talked about or gave information about abortions to receive funding, provided the funding did not exceed a cap and the organization agreed to not actively promote abortion. ⁵¹ Officials at USAID worked to ensure these new restrictions were less onerous on those that typically received funding. ⁵²

Thus, the Clinton administration and USAID officials reversed the direction of Reagan-era restrictions on abortion in foreign aid. They met resistance from conservatives in Congress. Because the administration and agency agreed, however, Clinton did not invest in new means of oversight. Instead, they released unrestricted discretionary funds to compensate for losses imposed by Congress, which the agency used without external audit. At the same time, family planning programs in foreign aid were being mapped onto domestic polarization over the abortion issue. The new, Republican-controlled Congresses were elected by a political base largely opposed to abortion. Meanwhile, the Cairo meeting on population explicitly re-branded family planning assistance as a reproductive rights issue for women, which helped to further polarize the elites who could serve as principals for USAID.

George W. Bush and Investments in Oversight

George W. Bush took office on January 20, 2001, reinstating the Mexico City Policy two days later. ⁵³ Once an obscure program during Johnson’s term, family planning in foreign aid had been linked to one of the most salient and polarized issues in American politics. The Republican party was now far to the right of USAID, and the new president invested in oversight to further close any information gaps.

Policy development and implementation. The second Bush administration did not decrease USAID’s funding or expand the Mexico City Policy. However, most of the USAID officials we spoke with described a “chilling” effect. Some grantees believed that if they gave medical attention to women who were hemorrhaging after a poorly performed abortion, they would be in violation of the policy. Officials at USAID organized a study to demonstrate the chilling effect on post-abortion care, and notified the White House. ⁵⁴

Thus, unlike previous administrations, president George W. Bush detailed implementation guidelines for the policy in a second presidential memorandum issued March 28, 2001. It included what USAID officials asked for: “[E]xcluded from this definition [of abortion services] is the treatment of injuries or illnesses caused by legal or illegal abortions, for example, post-abortion care.” ⁵⁵ Officials were quick to notify partners of the exemption and emphasize that the president saw family planning programs as the best way to “prevent” abortions abroad. ⁵⁶ They went on to form a post-abortion care (PAC) working group that advised and evaluated new post-abortion care programs. ⁵⁷ Thus, after signing an initial directive that would have reinstated the Reagan policy, George W. Bush signed another directive that carved out a new exemption and opened a door to a new initiative, all at the request of the agency. We interpret this as a minor, but important example of bureaucrats moderating the president’s initial preferences. It is also worth noting that it occurred under an acting official, before G.W. Bush had nominated a USAID Administrator.

Oversight investment. After conceding on post-abortion care, the G.W. Bush administration took steps to improve presidential oversight. This began with the appointment of Andrew Natsios as administrator of USAID. Neither McPherson nor Atwood were known to have strong opinions about family planning prior to their nominations. ⁵⁸ However, by 2001, USAID candidates were being selected on the basis of their abortion positions. Natsios had been a Massachusetts state legislator and had a public anti-abortion stance. Moreover, as the former head of World Vision, he also had deep connections to the faith-based NGO community, the kind that had been promoted by foreign aid policies that began in the Reagan administration. According to him, he later learned “there [were] a group of Conservative Catholics who were supporting [his] nomination.” ⁵⁹

Moreover, by G.W. Bush’s term, the Mexico City Policy itself had helped change the landscape of family planning providers abroad. During the Reagan administration, USAID officials described the difficulty of finding care providers who would have both the expertise and mission-orientation desired by the White House. ⁶⁰ By the early 2000s, there were no shortage of providers willing to accept family planning funds when the Mexico City Policy was reinstated. There were more than forty Christian NGOs working on the issue. In addition, by 2004, USAID finalized new rules to implement G.W. Bush’s “faith-based initiatives,” which were designed to ensure that religious organizations could more easily receive federal funding. ⁶¹

As administrator, Natsios took steps to ensure that auditing programs would be easier. Historically, the Office of Population had a great deal of autonomy from the rest of the USAID programs, dating back to the leadership of Ravenholt. Originally located in Rossyln, VA, it was even physically separate from other agency bureaus. It moved prior to Natsios’ tenure, but his reforms ensured that the office lost more of its original independence. At the beginning of 2001, Natsios moved management of the budget out of the Bureau of Management and into the Bureau of Policy and Program Coordination (PPC). ⁶² This concentrated both budget and planning functions in single assistant administrator who reported to Natsios. He also created the Bureau of Global Health in 2002, and placed the Office of Population inside it. Part of the purpose, according to him, was to “take all the technical people in the regional bureaus and move them all into the central bureaus.” ⁶³ The central bureaus had “developed substantial technical expertise” over time, but there remained technical experts in regional bureaus. Both moves helped centralize decision-making, while also providing more reliable avenues for obtaining information about how USAID funds were being spent.

Furthermore, the heightened party polarization in this period made ceding on abortion policy more costly not only to presidents, but to proponents of abortion care services. These organizations changed their tactics to mitigate the effects of the policy, while remaining uncompromising. For example, IPPF prohibited its Latin American affiliates from accepting USAID funds. Instead, it took a hardline stance and used the G.W. Bush reinstatement to fundraise, since other countries where in a position to provide foreign aid to international family planning organizations.

Ultimately, G.W. Bush compromised less than Reagan had, in part because USAID programs could be more effectively monitored—both by Republican appointees and the broader international aid community. We attribute this to structural investments in oversight, which began during the Reagan administration, but were ratcheted up under the second Bush. These factors restructured the interactions of the Office of Population with the White House, and ultimately, led to fewer opportunities for unelected officials to moderate policy proposals of the president.

Recent Presidents and Minimal Policy Loss

There is no evidence after the George W. Bush administration that USAID officials were able to moderate abortion policies demanded by the White House. The Obama administration rescinded the Bush policy in its first days in office, as well as the restrictions to UNFPA. ⁶⁴ Officials at USAID favored the removal of restrictions, and even set up a “fast-track” system for re-funding organizations previously denied under the previous administration. ⁶⁵ Thus, consistent with Clinton, we find no evidence that the Obama administration invested in oversight, because it did not need to. Allowing USAID officials to find creative means of refunding population programs starved during the previous eight years served their political interests.

The opposite was true for the presidency that followed. The Trump administration reinstated and expanded the Mexico City Policy—this time, applying it to all global health assistance. ⁶⁶ Specifically, the directive ordered the Secretary of State to develop and implement a plan “to extend the requirements of the reinstated Memorandum to global health assistance furnished by all departments or agencies.” Then-Secretary of State, Rex Tillerson, subsequently convened senior officials later that year to implement the expansion. ⁶⁷ A GAO report confirms that the policy had been expanded to include over 1,300 global health awards, both those that had been “grandfathered” in under previous Republican administrations, as well as new awards. It also estimated that under the new policy, about $261.6 million in global health funding had been declined by aid organizations unwilling to accept the Trump administration’s terms. This represented a significant expansion of the application of the policy. ⁶⁸

Furthermore, the Trump administration attempted to merge USAID with the Department of State. Officials quickly claimed that this extreme attempt at oversight would destroy decades of expertise and infrastructure. One of them was former administrator Natsios. He argued that “the misguided absorption of USAID into State is a much graver danger to the effectiveness of U.S. government aid program [sic] … A merger, however, could permanently diminish our ability to help and save lives around the world” (Natsios 2017). Tillerson later backed off a plan for a merger, but it is also notable for how it would have changed the task of monitoring USAID programs. The few details of the initial plans that emerged suggested it would involve more consolidation of budgeting, planning, and auditing of foreign aid programs. Thus, during this period, investments in oversight continued. Furthermore, in the Trump presidency, the expansion of the Mexico City Policy was successful. There is no indication that officials at USAID or elsewhere were able to secure concessions or to moderate the president’s preferences.

One potential concern, however, is that our coverage of the earlier periods relied heavily on first-hand accounts and archival material. Both are scarce during the Obama and Trump years. Practitioners working in the Obama and Trump administrations were less likely to respond to requests for interviews—or, in some cases, still worked for USAID, so were not permitted to be interviewed. Under the Presidential Records Act, records for presidents Obama and Trump will not be available until 2029 and 2033, respectively. This raises the possibility that there were moderating influences on the Obama and Trump policy changes; we simply have not observed them, because we only have public records and news reports.

We do not think this is the case, for several reasons. First, in the Reagan, Clinton, and G.W. Bush presidencies, the moderated policy was readily observable. What could not be determined, however, was that those moderating effects were attributable to officials outside the White House. In the cases of both Obama and Trump, the publicly announced policy did not differ from the one USAID implemented. Second, during the earlier periods, leaks to journalists often confirmed the political conflict going on behind the scenes. In the Trump administration, there are plenty of anonymous sources claiming to be from within USAID and State, who voiced discontent over the expansion of the Mexico City policy. But none of that discontent appears to have resulted in moderation. As the GAO report suggests, a significant conservative expansion of the policy was imposed despite the preferences of the agency.

Discussion

Presidential power is often only as real as the compliance of unelected officials. With superior information, bureaucrats may drift from the president’s preferences in the development and implementation of policy. By observing both the outcome of the policy and counterfactuals that would have happened if not for bureaucratic interference, we highlight the conditions that lead presidents to attempt to solve this problem by investing in the capacity for oversight.

As family planning programs were mapped onto party polarization over abortion, Republican presidents stood to lose more under the existing autonomy of USAID and its Office of Population. In these circumstances, these presidents sought not just to direct the activities of the agency, but to invest in new and innovative forms of oversight. They appointed like-minded officials, and those officials set up routine audits. They fostered like-minded non-profit groups around the world, who would eventually perform “fire-alarm” oversight on the cheap. They attempted and sometimes succeeded at restructuring USAID so that population programs came under the watchful eye of political appointees. At the start, the Reagan administration’s preferred policies were significantly moderated by USAID officials. This is a key finding of this study, as from the published record, there is nothing that would lead one to that conclusion. By the Trump administration, however, presidents were able to secure the policies they desired. We attribute this turnaround to decades of investment in oversight, driven largely by a key secular change over time—the polarization of competing principals.

Beyond this application, our more general argument is subject to several scope conditions. First, our argument is a developmental one: durable shifts in the political environment and governing structures make reversal highly unlikely. For a variety of reasons, investments in oversight tend to persist over time. The next administration, of course, will inherit the standard operating procedures of the previous administration, as well as any structural reorganization. Moreover, as with other cases of policy feedback, the interest groups empowered by previous presidents will be more difficult to disregard in the future.

Second, our theory should apply best to policy areas where the president has similar discretion. These are policy areas where the United States government funds a service or program, and the president can attach strings under their authority as chief executive of the federal government. This is quite common. As Gitterman (2017) shows, every modern president has used their authority as chief executive officer to achieve policy wins, including on healthcare, discrimination on the basis of race, class, and gender, as well as on the minimum wage.

These findings also imply some important considerations for future work. This longitudinal case study of abortion bans demonstrates that even in a case widely considered an example of unilateral presidential power, bureaucrats were able to negotiate with the president and to significantly shape the policy in its early years, in a way that held decades after. It was only after costly investments in oversight that the policy moved closer to presidential expectations. Furthermore, as we demonstrate, family planning is complex and has centered debates about both population control and women’s rights. As reproductive health policy is becoming the purview of chief executives at the federal and state level, we encourage political scientists to use institutional frameworks to study its development and effects taking into consideration all affected actors.

Our work also builds a framework through which to study the impacts of presidential interventions in emerging policy areas, such as pandemic response and electoral administration. Pandemic response is an area that has received a large influx of federal funds, with NIH receiving $4.9 billion to fund COVID-19 research on diagnostic tests, vaccines, and treatments. Another area is election administration, with policy becoming more polarized across party elites with time. Ultimately, our argument may be applied to explain changes in presidents’ efforts in these areas to address agency problems in executive action.