Abroad and At Home: Foreign and Domestic Sources of U.S. Migration Policy -- Marc Rosenblum
Foreign and Domestic Sources of U.S. Migration Policy
University of California, San Diego
PRELIMINARY DRAFT: Please contact author before citing; comments very welcome.
Abstract: Although migration is a defining characteristic of the twentieth century, and migration policy remains highly controversial, there is little consensus on the actual determinants of migration policy. Current explanations of migration policy-making focus either on Congress or, in a few cases, on international regimes; but neither of these mono-causal arguments accounts for the diversity of outcomes over time, or for the evidence cited by the other side in the debate. I argue that migration policy is not set unilaterally by Congress, but that it is the product of strategic interactions between Congress, the president, and migrant-sending states. I expect the relative ability of each to bring outcomes in line with preferences to vary in predictable ways as a function of their preferences relative to the status quo, the salience of migration to each actor, the actions available to each, and the existence of cooperative institutions. I conduct a case study of U.S.-Mexican migration policy-making since 1942 which provides empirical support for the model.
As a "nation of immigrants," the United States is also a nation of migration policy. The high stakes of migration policytens of millions of people migrate each year, and millions are on waiting lists to enter the United Stateshave assured that it has historically been highly divisive; and never more so than in the 1990s. Despite its importance and the often vicious rhetoric surrounding the immigration debate, there is little consensus on the forces that influence migration policy-making.
The issue deserves attention not only because migration is a defining feature of the twentieth century, but also because migration policy displays some puzzling characteristics. First, there is no obvious relationship between the state of the economy and U.S. migration policy: migration policy became more restrictive during the high-growth 1920s, 1950s, and 1990s, but also during the slower 1980s; and while efforts to restrict migration during the economic crisis of the 1970s failed, migration quotas expanded during the recession of 1990. A second puzzle concerns the relationship between migration and foreign policy: border control is central to U.S. sovereignty, but the effects of migration policy are largely felt in migrants' countries of origin. As a result, restrictive policies often conflict with foreign policy interests, including in the post-war and post-cold war transitional periods of the 1950s and 1990s, creating problems for U.S. foreign policy-makers. Finally, migration policy is often internally inconsistent: 1996, for example, saw both the passage of one of the most restrictive migration bills in history, and the initiation of significant new U.S.-Mexican cooperation on the subject.
All three of these empirical questions relate to a theoretical question: who are the actors that influence U.S. migration policy? While migration is fundamentally an international issue, the overwhelming majority of the literature focuses on domestic determinants of migration policies, typically treating it as a product of congressional interest group politics. It is difficult, however, to reconcile the Congress-dominated argument with a growing body of theory which explains migration policy changes as a function of globalization and international regimes. Moreover, there is considerable evidence that the president's preferences differ from Congress's, and that he is independently able to influence outcomes. In short, migration policy raises a theoretical question about the ability of Congress to act aloneas is commonly assumedversus the ability of the president and international actors to influence outcomes.
I argue that migration policy is the product of a strategic interaction between Congress, the president, and migrant-sending states. All three of these actors have strong, distinct preferences; and all have the ability to influence outcomes. I draw on bargaining theories to propose hypotheses about how differences among actors are resolved. By expanding the analysis beyond the existing boundaries of interest groups or international regimes, my approach accounts for evidence cited by both sides in the prevailing debate, and offers insights into the changes in migration policy-making over time.
I begin with a review of the literature on migration policy-making. Section three presents my argument in two parts: I show that migration policy-making is a three-player game, and then derive hypotheses about the ability of each actor to influence outcomes. Section four describes U.S.-Mexican migration policy since 1942, and analyzes the case in light of my model. Section five concludes.
Theories of migration policy-making may usefully be broken down into two broad camps: migration policy as a purely domestic process, and international/global explanations for migration policy. Domestic explanations have thoroughly dominated the debate.
2.1 Domestic Explanations for Migration Policy
Domestic theories of migration policy typically begin with economic interest groups. In general, this literature holds that business lobbies for loose migration policies, while unions are restrictionist (Fix 1991, Fuchs 1990, P. Martin 1994b). Many interest group arguments emphasize that the policy-making process has been captured by pro-migration forces which guarantee permissive policiesor large loopholes in restrictive policiesexplaining the failure of states to control their borders (Cornelius et al 1994, Briggs 1994). Freeman (1995) explains this tendency by arguing that migration has concentrated benefits and diffuse costs, so that pro-migration forces are better able to overcome collective action problems.
Two variations on the interest group approach are worth mentioning. First, Jeannette Money (1998) notes the well-documented geographic concentration of migrants. Money shows that policy changes result when migrant-receiving localities are politically important at the national level. While Money looks at electoral institutions, Hollifield (1992) and Joppke (1998) emphasize the role of the judiciary, showing that liberal Western institutions give migrants disproportionate political power.
Three points should be emphasized with respect to these interest group arguments. First, as Joppke (1998) and others have argued, states do have the capacity to control migration, though never perfectly. Programs like "Operation Wetback" in 1954 and Operations "Hold the Line" and "Gatekeeper" today affect the expected costs of migrating illegally (also see Hanson and Spilimbergo 1998). Effective policies are possible, in part, because the model is not as simple as Freeman and others hold: while beneficiaries of migrant labor are often diffuseconsumers who like low pricesunions lobby for restrictionism (Mink 1986; Haus 1995). Recent demographic changes also mean that migrants are increasingly dispersed and likely to work for small, unorganized firms (Cornelius 1998). As a result, it is notoriously difficult to develop a concise interest group model which actually explains behavior (but see Wong 1996).
A third problem with interest group approaches is highlighted by analysts who focus on the executive branch. Congress does not make migration policy alone: "Between 1882 and 1952 ten immigration laws were vetoed by six presidents with widely differing styles, perceptions of executive-legislative authority, and attitudes about the use of veto power" (Morris 1984, 34). Manuel García y Griego (1992, 75-6) argues that "the chief executive and a small group of advisers and public officials" make "policy choices relatively free of societal pressure." Kitty Calavita (1992) emphasizes that bureaucrats, especially within the I.N.S., initiate policy on their own, at times in conflict with Congress. The critical point here is that there is more to migration policy-making than intra-Congress haggling, though it is often unclear when the executive branch attempts and/or is able to defy Congress.
2.2 International Explanations for Migration Policy
A growing number of analysts have begun looking to international explanations for migration policy decisions. First, interdependence may force policy-makers to consider the migration consequences of foreign policy. Efforts to consider the migration effects of foreign economic policy include the Caribbean Basin Initiative (Diaz-Briquets 1995) and NAFTA (Bach 1990), both designed in part to reduce regional "push" factors. Policy-makers may be unable to ignore linkages: "some governments . . . claim that the developed world's right to export to their countries is contingent upon the third world's right to export their human problems to the developed world." "Similarly, a U.S. decision to establish a guestworker program or to provide special refugee assistance . . . affects that country's economy; but it also constitutes de facto support of that country's government"(McCarthy and Ronfeld 1983, 388; 394-5).
If migration and foreign policy are linked, is migration policy also used as a foreign policy tool? There is a consensus in the literature that "refugee policy . . . is an area where the administration is relatively free to insert its foreign policy concerns into migration policy" (Schoultz 1992, 197; also Kraly 1990, Loescher and Scanlan 1986, and Zolberg 1990). On the other hand, "there [has been] constant friction [between Congress and the president] over the use of parole power to admit refugees in excess of" refugee quotas and on a "political," rather than humanitarian basis (Keely 1983, 340; also Anker and Posner 1981, Loescher and Scanlan 1986).
Finally, a group including Jacobson (1996) and Sassen (1996) argue that globalization has created conditions under which international regimes weaken the ability of the state to regulate migration. Jacobson argues that commitment to international human rights and refugee regimes forces liberal states to grant larger-than-desired concessions to unwanted migrants.
2.3 The Relationship between Domestic and International Sources of Migration Policy
The literature on migration and international relations is a good starting point, but it raises as many questions as it answers. Empirically, migration has more often been a source of conflict between the United States and its neighbors than an opportunity for cooperation; but important exceptions exist (Tucker 1990, Bach 1990, Glazer 1990, Pastor 1984). When is migration a foreign policy tool, and when do domestic interests dominate? The answer implicit in much of the literature, and stated explicitly by Rico (1992), is that Congress dominates whenever it wants to; and foreign policy only "matters" when Congress is, for whatever reason, "uninterested" in migration.
An alternative approach observes that different migration issue areas seem to display different explanatory logics. Keith Fitzgerald (1996) argues that different issue areas are dominated by different branches as a result of path-dependent inter-branch struggles, which reflect the institutional balance of power at the moment new policies are demanded. For example, poor institutionalization of policy networks in the early twentieth century meant that front-gate policy in response to the surge in European migration was captured by ethnic interest groups lobbying Congress. By the mid-twentieth century, however, when World War Two produced a surge in refugees and demand for refugee policy, a well-developed national bureaucracy was in place, and the executive has dominated that policy area.
Neither of these efforts to unify domestic and international approaches explanations for migration policy are satisfying, however. The president often appears to make migration policy against Congress's preferences; and it is not only refugee policy that demonstrates this tendency (Mitchell 1992a, Dominguez 1990, Stepick 1990). In short, the existing literature fails to systematically address the variety of ways in which migration and foreign policy interact, focusing predominantly on refugee policy; and few theories account for the balance between foreign and domestic sources of U.S. migration policy.
I argue that migration policy-making is a strategic interaction between Congress, the president, and migrant-sending states. Each of the theories discussed above is partially right, because each captures part of this complex process. My argument consists of two parts. First, I show that migration policy cannot be explained simply by looking at Congress. Rather, the president and migrant-sending states are also key players. If all three actors influence migration outcomes, it raises the question of when each has more or less influence. The second part of my argument draws on bargaining literature and "two-level games" to propose hypotheses about the relative influence of each player. Outcomes are a function of the preferences, the salience of migration to each player, the actions they may take, and the institutions which govern the bargaining game. I begin this section by outlining the assumption I make to simplify analysis.
3.1 Modeling Assumptions
One Dimensional Policy Space
For ease of exposition, I conceptualize migration policy as consisting of a one-dimensional policy space which ranges from a perfectly open migration policy, under which there is no border enforcement and migrants enjoy complete protection; to a perfectly closed policy, under which the border is effectively sealed and migrants are denied all citizenship rights. I make this assumption because it allows me to draw on a wide literature which uses spatial models to derive hypotheses about the relationships between preferences, institutions, and outcomes.
In making the assumption of unidimensionality, I collapse two distinct issues: control policies, which regulate who enters the country; and integration policies, which determine the extent to which migrants are granted full citizenship rights. Despite the fact that these issues are distinct and may vary independently of each other, they may be treated as a single issue for the purposes of analysis because bi- (or multi-) lateral bargaining over multiple issues has been shown, theoretically and empirically, to collapse under most circumstances to a single bargaining dimension (Tsebelis and Money 1997). Some of the specific factors which locate migration outcomes in one-dimensional space are summarized in table one.
The Equilibrium Concept
Each player attempts to maximize his own utility. When all three players' actions maximize their own utility given the actions of the other players, the resulting outcome may be described as a Nash equilibrium, and should be stable until the underlying conditions which created it change. Conversely, whenever the policy outcome is such that one or more players is able to act to bring outcomes more in line with preferences than the status quo (i.e., the situation is not a Nash equilibrium), I expect the player who is able to improve their payoffs to do so.
In applying the Nash concept to empirical questions, multiple equilibria often exist. Moreover, policy-making actually occurs under conditions of uncertainty and policy-makers often collapse multiple dimensions into a single piece of legislation. As a result, even given a single set of actions which produce a particular equilibrium, there is often a range within which policies may fall. For both of these reasons, making predictions about bargaining outcomes and the relative ability of each player to influence outcomes requires us to look to a class of models Young (1975) refers to as "economic bargaining models," in which two players bargain within a known contract zone over the division of a single good.
3.2 Congress, the President, Migrant-sending States, and Migration Policy-making
Congress, the president, and migrant-sending states all care about U.S. migration policy; and each has the potential to influence outcomes. It follows that all three "matter" in migration policy-making.
Given "new institutionalist" assumptions, preferences are a function of players' institutionally-determined constituencies. For each of the players discussed here, migration policy is likely to be important; but players' distinct bases mean that preferences should vary according to different causal variables. My assumptions about preferences are summarized in tables two through four.
I assume that congress members have an over-arching desire to be re-elected (Mayhew 1974). As such, I assume that they seek to supply a combination of material and ideological goods which appeal to winning coalitions in their own districts (Arnold 1979). At the district level, Congress members face three types of pressures when it comes to migration policy. First, migration often has short-term, local economic costs as an influx of labor causes a bump in unemployment at the district level, and a short-term drop in wages. Moreover, while the net fiscal impact of migration (including undocumented migration) is positive, costs are disproportionately borne by states and localities, while fiscal benefits flow to the federal government (Chang 1998, Fix and Passel 1994).
In addition to these financial calculations, Congress members respond to economic interest group pressures, as described in section 2.1, often for permissive policies. Whether immigrants are mainly perceived in Congress as a threat to organized labor or an asset to businesses depends on the changing role of immigrant labor in the economy, and the relative strength of competing groups in Congress.
A final factor is the ideological cost of migration. Jeannette Money (1997) argues that migration surges are likely to produce a backlash from natives who see their culture under attack. The faster the inflow of new immigrants, and the less diverse the receiving community, the higher the perceived ideological costs of migration. Immigrants eventually become assimilated, so that in the long run, and at the national level, diversity is valued. While Money focuses on Europe and Australia, the history of backlashes against each major new wave of migrants to the United States supports her argument.
Thus, congressional preferences are a function of three factors. First, opposition to migration should be inversely proportional to the state of the economy, especially in migrant-receiving areas. I assume that economic pressures operate slowly, so that it often takes years before changes are reflected in congressional preferences. Second, preferences are a function of the interest group pressures on members. Third, opposition to migration should be directly related to the level of recent migration and rate of increase in migration, and inversely related to the overall level of migrants from the current ethnic group already assimilated into the country or particular area.
The president's preferences differ from Congress's in three fundamental ways. First, presidents must build a "heterogeneous national constituency [which] leads them to think in grander terms about social problems and the national interest" (Moe and Wilson 1994, 11; also Lohmann and O'Halloran 1994, Milner and Rosendorff 1997). Second, in contrast to Congress members, the president is not primarily concerned with reelection. Second-term presidents (except for Franklin Roosevelt) have been formally or informally barred from running for reelection. More generally, presidents are held accountable by the public and by history for their overall handling of the "ship of state" (Skowronek 1993), and must therefore consider the long-term consequences of policy decisions. Finally, the president differs from Congress in that the former is more concerned about foreign policy issues.
Relative to Congress, then, the president tends to be relatively open to migration for both domestic and international reasons. On the domestic level, I expect the president to prefer a flow of migrants that meets current demand for labor and supports economic growth. Presidents also tend to favor higher levels of migration than Congress because of the relationship between migration and foreign policy. As noted above, executive branch actors may also use migration as a foreign policy tool, and migration relations with particular countries have frequently been the subject of bilateral negotiations. The significance of foreign policy depends upon the strategic importance to the president of the particular migrant-sending state, and the importance to that state of emigration.
Presidential enthusiasm for open migration policies is checked when there is widespread popular demand for migration controls. The president must be sensitive to demand for migration controls since it is the executive branch which is held responsible for enforcing migration policy. Thus, in general, I expect the president's preference for migration flows to be a function of three factors: the overall state of the economy, the president's foreign policy agenda, and the national demand for migration controls.
On a fundamental level, "other countries . . . respond to the fact that what we do on immigration affects them" (Glazer 1990, 15). First, U.S. migration policy affects sending state economies: migration represents a "safety valve" which may absorb unemployed workers; and migrant remittances are a major source of foreign exchange. Second, if sending states care about their citizens, this impacts migration preferences in two ways. On the one hand, if would-be migrants prefer to migrate, sending states should favor open policies. At the same time, however, migrantsespecially undocumented onessuffer human rights abuses, and sending states have an obligation to protect citizens from such abuses, at home or abroad.
Migration may have ideological costs to sending states. Emigration suggests a failure by the state to create sufficient employment. In the case of Mexico, it is also necessary to consider the specific relationship with the United States. At least since the Mexican Revolution, state leaders have relied on anti-Americanism as a unifying national theme. Thus, when Mexico fails to deliver migration goods, stakes are raised by the fact that the conflict is with the United States. Good migration relations are essential to Mexican legitimacy: the ability to migrate North is considered a birthright, the economy suffers when migration flows are challenged, and the state is expected to protect flows and migrants' rights; but too much migration cooperation remains problematic.
Thus, migrant-sending states tend to prefer open U.S. policies, and that the rights of migrants abroad be protected. The intensity of that preference varies inversely with domestic economic conditions, which determine the emigration demand. Mexican preferences and preference intensity also vary with regime legitimacy and as a function of the overall bilateral relationship with the United States.
Congress, the president, and migrant-sending states each may influence migration outcomes. Here, I outline the actions available to each of these players, individually and in conjunction with others.
Congress's greatest asset is its ability to pass legislation. For the most part, this consists of Congress's agenda-setting power in bilateral policy-making with the president. While the relationship between legislation and policy outcomes is imperfect, I assume that there is a direct correlation between legislative changes and changes in outcomes. In addition to acting with the president to enact legislation, Congress may cooperate with the president and international actors to pass and ratify treaties or enact legislation in support of other international agreements.
Congress may also take unilateral action. With two-thirds support, Congress may override a presidential veto. Second, Congress can indirectly influence outcomes, including by holding hearings and using congressional research services. These unilateral actions are important as a tool for monitoring executive branch enforcement agents to ensure that executive shirking is kept to a minimum. Congress may also appeal directly to voters, but Congress's ability to "go public" in this way is limited (Kernell 1997).
The president may also act multilaterally or alone. First, the president's veto power makes her an important legislative actor, even given Congress's potential to override (Eskridge and Ferejohn 1992, Cameron 1996). Likewise, the president plays the critical role in cooperating directly with international actors through treaties, and executive agreements. In contrast to the legislature, presidential efforts at bilateralism often benefit from dense networks of institutions connecting executives between countries.
The president also has a set of unilateral actions. Most importantly, divided powers means that Congress must delegate to the president the enforcement of legislation. While Congress tries through a variety of means to check agency slack (McNollgast 1987, McNollgast 1988, McCubbins and Schwartz 1984 ), the nature of the delegation problem insures that the president is able to "shirk" to a significant degree (McCubbins and Page 1987, Moe and Wilson 1994). That is, the president always has some ability to enforce a policy other than that passed into law. Finally, the president has extensive ability to "go public" to change public opinion and thereby influence Congressional preferences.
Depending on the issue, international actors have an almost unlimited number of actions available which impact U.S. policy; migration theorists have given relatively little thought to this subject. At the most extreme level, emigration has been a tool of conquest, including in 1975 by Morocco into the Spanish Sahara, Israeli policy in the West Bank, and Texas' colonization of Mexico (Teitelbaum 1984). More common has been the use of migrant expulsions as a foreign policy tool (by Vietnam, East Germany, Cuba, and others). Cuba, in particular, has significantly influenced U.S. policy by regulating exit flows (Dominguez 1992).
In addition to these unilateral policies, international actors interact with U.S. actors in at least three general ways. First, to the extent that preferences overlap, direct cooperation through treaty or executive agreement is possible, subject to qualifications found in the literature (Oye 1986; Krasner 1991). Second, even when preferences about an issue conflict, international actors may "link" cooperation between two separate issue areas (Keohane and Nye 1977; Sebenius 1983; Lohmann 1997). Finally, international actors can influence U.S. politics by directly or indirectly lobbying U.S. policy-makers (Evans, et. al. 1993; Milner 1997; Pahre and Papayoanou 1997; Schoppa 1993; Sebenius 1992; etc.). Knopf (1993) distinguishes between "transgovernmental" interactions, when state actors support like-minded factions in another; "transnational" interactions, when non-state actors strengthen non-state actors abroad; and "cross-level" interactions, which occur between state and non-state actors in different states.
While Congress's influence over migration policy consists mainly of passing legislation with the cooperation of the president, the latter may act alone, or with migrant-sending states, to enforce policies other than those agreed to with Congress. I expect all three actors to anticipate moves by the others, so that the policy-making process may usefully be conceptualized as a strategic game of the following form: Congress passes a new migration policy or maintains the status quo; if Congress passes a policy, the president may sign or veto it, after which Congress sustains or overrides the veto. These three moves are the "legislative round" of the game. Next, in the "enforcement round," the president either enforces the exact policy passed by Congress, shirks to enforce a slightly different policy, or cooperates with migrant-sending states to change outcomes. At the same time, migrant-sending states may also act unilaterally or cooperate with the president. Finally, Congress may respond in the following legislative round as the game repeats. The game is depicted graphically in figure one.
3.3 Migration Policy as Strategic Bargaining Game: Independent Variables and Hypotheses
If Congress, the president, and migrant-sending states all "matter," it raises the question of their relative influence over outcomes. That is, when players' preferences conflict, what outcomes result? I consider the influence of four types of variables on bargaining outcomes: preferences; issue salience, or preference intensity; specific actions available to players; and institutions.
Preferences and Bargaining Outcomes
Players' preferences may be thought of as defining the universe of possible outcomes: no equilibrium policy will ever allow less migration than is desired by the most restrictionist actor, or more than the level desired by the least restrictionist. In a basic way, preferences determine the actions players may feasibly pursue. When two players prefer policies on opposite sides of the status quo (i.e., one player prefers a more restrictionist policy than the status quo and another prefers a more open policy), the players cannot possibly agree on a new policy.
Preferences are an important independent variable because changes in players' preferences always threaten to disturb an existing equilibrium since the same set of possible actions may now offer a more-preferred outcome. Given that players take note of others' expected actions, it is possible that a change in any player's preferences will produce new actions, possibly by a different player. Note, however, that not all preference changes will disturb an existing equilibrium, so this variable does not imply a "causal" hypothesis in the sense of creating necessary and sufficient conditions for changes in outcomes. Given the discussion of the overall policy-making game in section 3.2, this discussion of preferences produces two hypotheses which are the essential argument of this paper, and which contrast with the null hypothesis (H0) that currently dominates the literature:
H0: U.S. migration policy is (only) a direct function of congressional preferences
H1A: U.S. migration policy is (also) a direct function of presidential preferences
H1B: U.S. migration policy is (also) a direct function of Mexican preferences
Finally, as Schelling (1960) and numerous other in his wake (e.g., Putnam 1988, Evans et. al. 1992, Milner 1997) have observed, when the president's preferences lie between those of Congress and international actors, the former may be able to credibly claim that his "hand are tied" by a provincial Congress, and thereby extract concessions from the international actors. More generally:
H2: The player whose preferences are "in between" those of the other two players has the most influence over migration policy
Salience and Bargaining Outcomes
The basic Rubinstein (1982) bargaining model predicts how a unit is divided by two players with independent "discount factors" making alternating offers. A number of subsequent authors have given more thought to how and why players discount the future, emphasizing two factors: the salience of the issue to each player, discussed in this section, and the actions available to each, discussed in the following.
McMillan (1992) notes that given uncertainty about whether an agreement will be reached at all, the key factor determining outcomes is each side's "reversion point," that is, the cost to each side of failing to reach an agreement. Sebenius agrees: "many ad hoc discussions using the term 'power' can be reduced to precise and rigorous statements about the parties' no agreement alternatives" (Sebenius 1992, 334). For Tsebelis and Money (1997), this issue is described as the "impatience" displayed by each side. Given my assumption that politicians are accountable to their constituencies, I assume that "no-agreement costs" are a function of the salience of migration to politician's constituencies. That is,
H3A: As the salience of migration increases (decreases) for a given player, his bargaining strength decreases (increases), and outcomes should be less (more) favorable to him.
Given the duality of the president's preference function, policy outcomes should also vary depending on how the president weights foreign versus domestic sources of his preferences. In particular, the more weight the president attaches to foreign policy in generalor U.S.-Mexican politics in particularthe more she should work to bring outcome in line with Mexican preferences, above and beyond the president's own preferred policy.
H3B: As the salience of foreign/bilateral relations to the president increases (decreases), outcomes should be more (less) favorable to migrant-sending states.
Possible Actions and Bargaining Outcomes
The analysis here is greatly complicated by two facts which separate the empirical case of interest from the formal bargaining literature: the sets of available actions vary by player, and available actions may vary over time as players have the ability, at least potentially, to change the rules of the game themselves. The ability to change the game inevitably lowers the cost of no agreement; "alternative opportunities . . . are a source of bargaining power" (McMillan 1992, 48), In other words:
H4: The more (fewer) actions a player may potentially take, the more (less) favorable outcomes should be to him.
Institutions and Bargaining Outcomes
A final factor which influences the ability of actors to credibly commit to agreements is the existence ofor the ability to createmultilateral institutions. Even given the dense institutional structure within the United States, Congress must frequently create additional oversight institutions to insure that outcomes reflect policies, and the problem is more severe in the poorly institutionalized realm of international relations. As a result, international actors create institutions to promote transparency, bind players to agreements, and thereby raise the expected utility of cooperation (Powell 1998, Keohane 1984, Axelrod 1984). While I assume that institutions do not change outcomes in that they reflect the preferences and power relationships of players at the time that are created, institutions are important for their ability to stabilize an equilibrium outcome:
H5A: When equilibrium agreements are accompanied by formal institutions, they should be more resistant to changes in underlying conditions.
Institutions themselves are more resistant to change when they embody the preferences of a larger number of veto players (Tsebelis 1995):
H5B: Agreements accompanied by multilateral institutions should be more resistant to change than those accompanied by bilateral institutions.
The literature on migration policy-making has typically startedand stoppedwith Congress because "it has been universally understood that Congressand not the presidentpossessed the constitutional authority to set conditions for entry and fix quota numbers" (Loescher and Scanlan 1986, 56). But at the same time, "[p]robably no other major area of public policy has been the cause of as much conflict between Congress and the executive branch as immigration" (Morris 1985, 33-4); and it is a policy area whose effects are felt at least as strongly in migrant-sending states as at home. I argue that migration policy is not, in fact, set unilaterally by Congress, but that it is the product of strategic interactions between Congress, the president, and migrant-sending states. I expect the relative ability of each to bring outcomes in line with preferences to vary in predictable ways as a function of their preferences relative to the status quo, the salience of migration to each actor, the actions available to each, and the existence of cooperative institutions. These hypotheses are summarized in table five.
In this section, I trace the history of U.S. migration policy-making with respect to Mexico, broken down by decade since 1942. For each decade, I discuss the independent variables identified above, summarize the actual actions which players undertook, and then analyze migration outcomes. The argument is summarized in table six.
4.1 The 1940s
In 1942, the United States and Mexico initiated the "Bracero" guestworker program which dominated bilateral migration relations for 25 years. The early years of the program were characterized by intensive interaction between Mexico and the U.S. executive, with the former initially dominating the terms of the agreement. The close of World War Two marked significant changes in the underlying structure of migration issues, the preferences of all three players, and, eventually, the nature of migration relations.
Congress was not very concerned about migration policy, especially with respect to Mexico, until after 1946. The economy was strong after 1941 (especially in migrant-receiving states), and there was little migration during the war. With the end of the war, a return to isolationism, the United States' rise to superpower status, and the influx of over a half million displaced persons from Europe, created a strong feeling that "Americaness" was something to protect by excluding "others" (McCarthy and Ronfeld 1983, 384). Racism and ethnic nationalism were rampant, especially in the Senate (Loescher and Scanlan 1986), but migration policy was not highly salient, especially with regard to Mexico (See Simon 1984).
Presidential preferences, in contrast, were strong and unambiguous during the War years: cooperating with Mexico on a guestworker program served the dual purposes of satisfying domestic demand for steady access to migrant farm labor, and solidifying foreign policy ties with Mexico, which remained neutral in early 1942. With the close of the War, foreign policy concerns shifted primarily to rebuilding in Europe, and Mexico received less special consideration; but economic growth and demand for migrant labor continued.
In the 1940s, Mexico's economy was recovering from the conflict surrounding the appropriation of American and European oil interests by Lázaro Cárdenas in 1938; and debt restructuring was a high priority (Garcia y Griego 1988, 105). The government's anti-imperialist legitimacy remained high in the wake of the Cárdenas sexenio, and successful cooperation during the early years of the Bracero program kept U.S. relations uncontroversial. Although Mexico was initially ambivalent about the Bracero program, within a few years leaders came to see it as a critical source of revenue and control over migration flows.
The Executive agreement of July 23, 1942 which initiated the Bracero program was accomplished completely "in the absence of Congressional legislation or public debate" (Calavita 1992, 2). The program was formalized in an exchange of diplomatic notes August 4, and up and running for almost nine full months before Congress authorized the program by the Act of April 29, 1943. Congress's only effort to shape the program was to pass separate legislation authorizing unilateral recruitment of workers migrating to the state of Texas, which had been excluded from the bilateral agreement. The executive blocked the action, however, under pressure from Mexico, by instructing agents at consulates and the border to block Mexicans who lacked authorization from the Mexican government.
Congress became more involved in bilateral migration relations with the end of the war, passing Public Law 40 in April 1947 mandating an end of government-to-government Bracero contracting. Following a flood of grower requests, however, a new bilateral agreement was negotiated between the two executives in February, 1948. The agreement again bypassed Congress, establishing a new program which allowed direct contracting between workers and employers, supervised by state actors. Though formally bilateral, recruitment was highly decentralized, mainly occurring within the United States where precedence was given to "drying out" undocumented immigrants.
The 1948 agreement was short-lived, however, thanks to unilateral action taken by low-level I.N.S. officials. In the so-called El Paso Incident of October, 1948, border personnel responded to a massing of would-be braceros in Juárez by opening the border and paroling thousands of Mexicans into the United States as legal workers. In response, Mexico abrogated the agreement. Although the two sides re-opened talks immediately, it was not until August, 1949, after Sen. Clinton Anderson (D-NM) introduced a unilateral guestworker recruitment bill, that a new agreement was signed.
Everything about the years 1942-1946 worked in Mexico's favor, and, not surprisingly, migration outcomes closely reflected Mexican preferences. The salience in Mexico of a guestworker agreement was low, while the president's demand was high for both domestic reasons (concern about labor shortages) and as a foreign policy goal (fear that Mexico would fail to join the war effort). Congress was not pleased by the terms of the agreement, but was shut out of the process: presidential action in Mexico's favor, negating Congress's effort at unilateral recruitment in 1943, contrasts with the events of 1954 (see below).
The end of the war changed the president's preferences and strength relative to Congress, and the result was a fundamental shift in the bargaining game. At the international level, the president no longer considered Mexico's support a national security issue; and at the domestic level the president no longer faced a labor shortage. Not only was the president's demand for a migration agreement lower for these reasons, but Congress was also less willing to defer to the president. Following the "El Paso Incident," negotiations between the U.S. and Mexican executive branches wore on for eight intense months, with neither side willing to make concessions. As Garcia y Griego explains, "what finally broke the logjam was a threat of unilateral action by the United States [Congress]" in the form of the Anderson bill, after which Mexico conceded all of its previously non-negotiable points (Garcia y Griego 1988, 185-6). It was only after the passage of the Anderson billdrawing attention to the importance of players' ability to change the gamethat the president was strengthened by his relative position as predicted by hypothesis two.
4.2 The 1950s
The 1950s saw the Bracero program evolve from a series of executive agreements to a genuine interaction between all three actors. Major policy discontinuities were triggered by the start and finish of the Korean War, with the latter producing an equilibrium which continued unchanged until 1964.
As the scope of the Bracero program expanded throughout the 1950s, growers demanded continued access to migrant labor, increasingly lobbying the Agricultural Committees of the House and Senate. Congressional preferences were strong because (as Freeman would predict) bracero employers were concentrated beneficiaries. Bracero workers were also still a relatively small population, and mainly "circular" (i.e., returning to Mexico every year), meaning they represented little cultural threat to natives.
The president's attitude toward guestworkers went through three phases in the 1950s. Frustration over the effects of undocumented migration on U.S. workers made the president disenchanted with the program (President's Commission on Migratory Labor 1951). With the onset of the Korean War, however, the Labor Department forecast record shortages, and presidential preferences resembled 1942 again. After the war, concern over the "wetback problem" resurfaced: undocumented migration was considered harmful to workers and a security threat since migrants might include "subversives."
Mexico's demand for the migration "safety valve" increased over the course of the decade as "stabilizing development" failed to generate sufficient employment, and because the program developed a bureaucratic constituency. However, weak enforcement after the 1940s created poor working conditions for Braceros, raising concerns about human rights. Mexico was also concerned about undocumented migration. Finally, especially to the extent that early disputes were resolved in favor of the United States, the bilateral relationship threatened domestic legitimacy.
Mexico's twin frustrations about human rights and undocumented migration prompted negotiators to offer an ultimatum in January, 1951: the agreement was to be institutionalized through government-to-government, congressionally sanctioned contracting, or Mexico would end emigration to the United States. The Chairmen of the House and Senate Agricultural Committees, at these negotiations, started hearings on the subject in March; PL78 was passed in June; and a bilateral Migrant Labor Agreement (MLA) was signed immediately thereafter.
While Mexico achieved its main goal of government-to-government contracting, fundamental disputes remained over how contracts were to be administered and Mexico's desire to blacklist known discriminators. At the same time, "a sense of crisis regarding unsanctioned migration was beginning to impact upon U.S. and Mexican public awareness in 1950 and 1951" (Garcia y Griego 1988, 209). In response, the two executives cooperated in a series of bilateral border patrol efforts and "deep repatriation" programs which deported undocumented migrants by plane and rail to the Mexican interior. Joint enforcement climaxed in "Operation Wetback," a sweep of immigrant-dependent agricultural areas which expelled hundreds of thousands of (mostly) undocumented immigrants in 1954.
With the end of the Korean War in June, 1953, U.S. negotiators sought to limit Mexico's aggressive contract dispute tactics when the program came up for renewal in December. With no new agreement by January, the United States initiated an "interim policy" of unilateral recruitment. In response, Mexico suspended the treaty and placed troops on the border to prevent unauthorized emigration. At the same time, the I.N.S. facilitated entries, allowing undocumented Mexicans to literally step across the border and re-enter as "unilateral bracero recruits." As Garcia y Griego (1988, 672-3) notes: "It was a bizarre way to express the idea that the U.S. really desired to contain the 'wetback invasion;' it was a stranger way for Mexico to communicate that it really desired to keep the door open for further emigration." Within a week, Mexico proved unable to control flows; and by March of 1954 a new agreement was signed on American terms. Fortunately for the Eisenhower administration, Mexico gave in a week before the Comptroller General ruled that unilateral recruitment was explicitly illegal under the terms of PL78, prompting a week of contentious hearings in the House Agriculture Committee before the law was amended to maintain the U.S. bargaining position.
Two changes occurred at the beginning of the decade, dramatically altering the bargaining game: by around 1950 Congress and the president had traded places in terms of their migration preferences; then America's entry into the Korean War made the president once again dependent on the program. Mexican negotiators correctly perceived that they could benefit from the former change if Congress was included in negotiations. Consistent with hypotheses 3A and 3B, Mexico could take advantage of migration's salience to the president to bring Congress into the game.
The terms of PL78 and the 1951 MLA contained two important victories for Mexico: the program returned to a system of government-to-government contracting, and recruitment was to take place primarily within Mexico. Congress benefited from officially legalizing the import of contract workers, eliminating the Attorney General's discretionary role in the program. Congress also rejected several presidential recommendations, including the establishment of employer sanctions and guidelines for determining "prevailing wages" and demand for labor. Finally, the president obtained two concessions: interior recruitment was supplemented by border recruitment centers; and Mexico could no longer blacklist areas.
The 1951 agreement is interesting on an additional level besides these compromises over terms: By bringing Congress into the bargain, Mexico was able to institutionalize the relationship as it had failed to do in the 1940s. Mexico was willing to make concessions in pursuit of this institutionalization, especially since it could regain some ground by pressing individual contract disputes. The president was willing to accept the outcome despite Mexico's post-agreement haggling because labor availability remained a concern, and because Mexico cooperated with the executive on border enforcement.
The end of the war destroyed this equilibrium. With no more threat of a labor shortage (i.e., a decline in salience), the president was willing to risk the agreement itself in pursuit of less Mexican intervention (supporting hypothesis 3A). By initiating unilateral recruitment, the executive demonstrated that Mexico had no effective veto over bracero recruitment (contrary to all three players' prior belief). In effect, then, the president narrowed Mexico's set of available actions, and in so doing weakened Mexico's bargaining position, as predicted (Hypothesis 4). This change in available actions also caused preferences to return to their previous alignment: Mexico and the U.S. executive still desired some agreement, especially if it meant more effectiveness against undocumented migration; but Congress was only concerned about the availability of bracero labor, which no longer seemed to require Mexican cooperation.
The equilibrium reached in the wake of the 1954 crisis was highly stable. Once all three players knew that Mexico lacked veto power over the agreement, and given the low salience of U.S.-Mexican relations, Congress and the president designed an ideal program from a purely domestic standpoint; but Mexico still benefited from bilateral recruitment. Finally, Mexico and the president benefited from cooperative border enforcement: only as a bilateral program could Operation Wetback legalize large numbers of migrants. Once this program became institutionalized in U.S. law and international agreement, it would take a substantial shift in both congressional and presidential preferences to bring it down.
4.3 The 1960s
The 1960s saw the termination of the Bracero program, and the beginning of 30 years of unilateral migration policy-making by the two sides.
Although access to migrant labor remained a priority, by the 1960s the terms of the debate had shifted. The impact of large-scale migration throughout the 1950s on wages and "culture" filtered up to Congress, which became convinced that braceros depressed wages and took jobs from natives. With over 3,000,000 Bracero contracts issued during the 1950s, and migrants' spread beyond the agricultural southwest, the argument was persuasive to "Great Society" Democrats.
The president's preferences also turned against bilateral migration policy in the 1960s. On the one hand, relations with Mexico took downturn as the latter was a primary opponent of U.S. Cuba policy, and was generally confrontational throughout the decade. Domestic pressures on the executive were basically unchanged from the 1950s: while labor demand continued, costs of legal and undocumented migration made guestworkers a mixed blessing. As with Congress, the executive's shift to the left in the 1960s tended to place workers ahead of business.
Bilateral relations were difficult from the Mexican perspective as well, beginning with the controversy over Cuba. While benefits of bilateral recruitment and reliance on remittances meant Mexico opposed the Bracero program's termination, the success of the "Mexican Miracle" economy and weak border enforcement made its loss less damaging than had been feared.
Labor activists and liberals in Congress finally overcame agricultural interest groups in 1963, though intensive lobbying of the president by Mexico, pushed the decision to terminate the program back a year. The termination fell within a general context of rationalizing U.S. migration policy through the 1965 amendments to the INA, replacing variable by-country (Eastern Hemisphere) quotas in favor of a flat 20,000 migrants-per-country system. Congress also capped overall Western Hemispheric migration, despite executive branch objections. Steady growth, and a series of more pressing foreign and domestic issues kept migration off the American agenda for the next ten years.
Mexico's reaction to congressional moves against the Bracero program was to press for its continuation by lobbying Kennedy, and then propose new programs in subsequent years. At the same time, Mexico took unilateral steps. Anticipating a glut, Mexico created a duty free zone along its northern border to absorb returning braceros ("the Maquiladora program"), and anticipating a rise in abuses of now-undocumented migrants in the U.S., Mexico almost doubled its U.S. consulate presence.
The termination of bilateral policy-making in 1964 is easy to explain: congressional and executive preferences had both been moving against the program for years, and Mexico's foreign policy influence was low. Even so, the low domestic salience of migration at the time (Simon 1984) allowed Díaz Ordaz to successfully extend the program an extra year.
Unilateralism stabilized by the late 1960s as Mexico, Congress, and the president all considered it optimal. Employers' labor needs were met, while the official policy of migration control satisfied constituencies in both branches. In Mexico, low demand for emigration and high demand for nationalism also supported unilateralism. The low salience of migration in both countries throughout the decade insured that neither government faced pressure to act. Thus, even though policy was not "cooperative" at this time, mutual indifference and peaceful coexistence were probably a stabilizing force in the bilateral relationship.
4.4 The 1970s
The 1970s brought the United States major economic crisis and a surge in migration. These issues failed to resonate with Congress and Mexico, however, and presidential initiatives on migration went unanswered. By the late 1970s, however, economic shocks brought Mexico back to the bargaining table; but tentative efforts by both executives failed to bilateralize policy.
A minority in Congress expressed concern about undocumented migration in the early 1970s, but the problem was small in scope and not a high priority. It was only in the late 1970s, after several years of stagflation and high levels of migration, that salience increased significantly.
As early as 1973, the coincidence of economic crisis and increases in legal and illegal migration made cooperation with Mexico attractive to the president. The dramatic demonstration of U.S. vulnerability to international shocks made Mexico an attractive ally as a medium-sized regional power with newly-discovered oil reserves. Increasing political activism by Mexican-Americans also raised Mexico's stock in the United States.
Developments in the early-1970s made Mexico uninterested in bilateral migration policies, though its economic crisis of 1976 prompted a temporary reversal in this attitude. Continued strong performance by the Mexican economy, and the discovery of additional oil reserves, appeared to place the country in a position of "managing prosperity." At the same time, Mexico realized that the end of the Bracero program had not substantially affected migrants' ability to work in the United States. Nonetheless, migrants' rights in the United States remained an important issue, partly driven by the nationalism characterizing Mexican politics at the time. While the crisis of 1976 caused a softening of anti-American rhetoric, the well-publicized (in Mexico) murder of three Mexicans in New Mexico in 1977 kept rights on Mexico's agenda.
Most of the actions in the decade were initiated by the president, and many attempted to adopt a bilateral tone. In a reversal from the 1960s, the Ford administration sent feelers in 1974 regarding Mexico's interest in a new guestworker program in exchange for U.S. access to Mexican oil reserves. Bilateralism looked promising with the start of the Carter and López Portillo administrations. During a 1977 summit, the executives established "binational consultative mechanisms" to promote high-level dialogue; and Carter honored a Mexican request by sending a letter asking each U.S. governor "'to deal fairly and humanely with any persons accused of being undocumented workers'" (Pastor 1984, 48).
But efforts at bilateralism failed to produce results, and both executives continued to act unilaterally as well. From the U.S. side, Ford initiated a deep repatriation program in 1976, deporting undocumented Mexicans in a well-publicized airlift directly to Mexico City. Carter proposed major reforms in 1977, focusing on employer sanctions (anticipating the 1986 IRCA); but Congress rejected the program. Carter also entertained the idea of erecting a barbed wire fence along the border in 1978, but rejected the idea as likely to harm bilateral relations.
Mexico rejected Ford's oil-for-guestworkers offer in 1974, but was both more receptive to bilateralism and concerned about access after 1976. Mexico's crisis thus prompted a brief surge in outreach by that country, including an unsuccessful effort to persuade Congress to pass special quotas for "contiguous neighbors" when by-country quotas were extended to the Western Hemisphere in 1976. By late 1977 when Mexico's economy entered another boom cycle, Mexico again prioritized unilateral efforts to protect migrants, more than doubling its budget for consul advocacy and legal training for Mexican experts on U.S. migration policy.
The priority Carter attached to Mexico relations was reflected in his creation of the Office of the U.S. Coordinator for Mexican Affairs, responsible for coordinating all interactions with Mexico; and in the drafting of Presidential Review Memorandum 41 (PRM-41), which sought to identify linkages between migration and other areas of mutual concern. The central question for the 1970s is why were the executives unable to bilateralize migration policy despite their preference to do so?
The answer to this question is twofold. First, the president was the only actor consistently concerned about migration policy, and he was in an exceptionally weak bargaining position throughout the decade. The 1970s "was a time in which U.S. global hegemony was visibly in decline and there was a marked sense of national vulnerability to external political and economic pressure. Rising inflation added to domestic discontent. The Carter administration was by mid-1979 under siege at home and abroad" (Thorup 1989, 140). In the wake of Vietnam and Watergate, Congress increasingly considered its role to include checking executive initiatives, including on foreign policy (Lindsay 1994). In short, the president's inability to take advantage of his position "in between" the other players (contra H2) is explained in part by looking to H3 and H4: the overall crisis increased salience to the executive and limited his available actions, while Congress expanded its actions through increased oversight.
A second obstacle to bilateral policy-making was that even though both sides desired cooperation per se, it was still difficult to find common ground. On the U.S. side, Carter's attention to the issue, including PRM-41 failed to produce concrete proposals: "Agreeing on immigration policy was hard enough. Working out compromises between it and other issue areas became almost impossible" (Rico 1992, 279). Mexican politics did not make Carter's task any easier. The crisis of 1976 caused an softening in an otherwise extremely nationalistic period; this was not an easy time for Mexico to embrace bilateralism even if attractive offers had been made.
In part, the failure of the 1970s followed from the prior decade of strict unilateralism and the consequent lack of institutionalized structures for discussing bilateral migration issues (H5). Negotiators faced a problem of talking past each other, with Mexico seeking to protect migrants rights, and the United States emphasizing control of undocumented migration. Thus, all the two sides accomplished was to address the prior problem by founding the binational consultative mechanism, which would eventually evolve into the Binational Commission that continues to hold inter-agency cabinet-level meetings today.
4.5 The 1980s
Mexico again embraced bilateralism after the 1982 economic crisis, though its bargaining power was weak. In the United States, the combination of Mexico's crisis and a surge in migration flows in the early 1980s brought migration policy to the forefront of the national agenda at roughly the same moment that both branches of U.S. government turned against Mexico.
By the 1980s, congressional preferences were solidly restrictionist, finally reflecting the pressures which had built over a decade of economic crisis and high migration. Not only had migration increased, but it increasingly consisted of permanent settlers, representing an ideological threat to a growing number of local communities. The 1970 and early-1980s also saw a massive influx of other "third world" migrants, especially from Cuba and Vietnam.
While Reagan initially shared Carter's preference for bilateralism, a number of factors after 1982 turned him against Mexico. First, the crisis of 1982 (more so than 1976) was highly costly to U.S. businesses, and it called into question whether Mexico could be of "use" to the United States. Continued high migration rates kept migration in the public eye, especially after the 1980 Mariel boatlift, politicizing the issue for the president. And, for the first time since the 1960s, the United States had a major foreign policy beef with Mexico over the latter's refusal to support Reagan's Central America policy.
On the Mexican side, the crisis of 1982 initiated a transition from statist nationalism in the 1970s to free market liberalism in the 1990s. The crisis created a rift in the PRI between economic liberals and more traditional corporatist "dinosaurs." The election of Miguel De la Madrid in 1982, and his reversal of long-standing policy by bringing Mexico into the GATT in 1985, marked a victory in this dispute for economic liberals, and Mexico increasingly prioritized bilateralism thereafter.
Starting in 1982, the U.S. executive acted in ways which could only be described as hostile to Mexico. As late as 1981, Reagan's contribution to the IRCA debate, the proposed Omnibus Immigration Control Act of that year, included a doubling of Mexican and Canadian migration quotas, "diplomatic efforts to encourage Mexico and other U.S. neighbors to help curb illegal immigration," and a two-year experimental guestworker program for. But as the IRCA debate wore on, the administration made little effort to bring Mexican issues to the table in Congressional testimony, and instead adopted a confrontational position by repeatedly raising issues which had previously been "untouchable" in the bilateral relationship, such as democratization and foreign policy (Chabat 1989).
While a series of bills introduced in the 1970s foreshadowed IRCA, the issue only started to gain momentum in the 1980s, especially once passage of the Refugee Act of 1980 cleared other migration issues from the agenda. After coming close in 1982 and 1984, Congress finally passed the IRCA in 1986. It is significant that the 1986 act was one of the only iterations of the bill which did not include any concessions to Mexico; as late as 1984, the House had passed a version of IRCA which included an amendment to