Political Economy of Trade Policy Paola Conconi

Transcription

1 Political Economy of Trade Policy Paola Conconi Université Libre de Bruxelles (ECARES) and CEPR

2 No other area of economics displays such a gap between what policymakers practice and what economists preach as does international trade. The superiority of free trade is one of the profession s most cherished beliefs, yet international trade is rarely free. (Rodrik, 1995) The past decades have witnessed a remarkable decrease in the level of import tariffs. At the same time, there has been a surge in non-tariff barriers (NTBs), e.g. antidumping and countervailing duties, sanitary and phytosanitary (SPS) measures, technical barriers to trade (TBT).

3

4 Why do policymakers support protectionist policies? Economic explanations: welfare maximizing policymakers depart from free trade due to the existence of distortions (e.g. imperfect competition, dynamic economies of scale, revenue constraints). Political economy explanations: politically motivated policymakers respond to the interests of certain groups in society (e.g. industry lobbies, voters).

5 Plan of the lecture Review of political economy of trade literature, studying the role of Lobby groups Electoral incentives Institutional rules

6 Individual preferences (A) Interest groups (B) trade policy outcomes Policymakers preferences (C) Institutional structure of the government (D)

7 Standard trade model Small open economy All goods (but no factors) are tradable Consumer preferences are identical and homothetic Perfect competition in product and factor markets Alternative assumptions about the supply side: - Specific-factors (S-F) model - Hecksher-Ohlin (H-O) model

8 International trade generates winners and losers H-O model: an import tariff raises the real return to the economy s scarce factor and reduces the real return to the abundant factor (Stolper-Samuelson theorem) the distributional effects work along factor lines S-F model: an import tariff in sector i increases the return to the specific factor in that sector, lowering the returns to other specific factors through the induced increase in wages the distributional effects work along industry lines

9 Typology of political economy models (Rodrik, 1995) A) Median voter model (Mayer, 1984) B) Interest groups models: Tariff formation function (Findlay and Wellisz, 1982) Campaign contributions approach (Magee, Brock and Young, 1989) Political support function (Hillman, 1989) Political contributions approach (Grossman and Helpman, 1994, 1995)

10 Median voter model (Mayer, 1984) Tariffs determined by majority voting among the population Provided that preferences are single picked (i.e., each person has a unique maximum), the trade policy chosen maximizes the utility of the median voter. In a H-O model, if voters differ in their relative factor endowments and the median voter owns a lower capital-labor ratio than the one for the overall economy, we should observe Import tariffs, if imports are labor intensive Import subsidies, if imports are capital intensive

11 Lack of realism of the median voter model: - Trade policy not determined by majority voting - Import subsidies not observed - Tariffs observed even in small industries that do not have support from majority of voters

12 The role of lobby groups

13 Workhorse model of Grossman-Helpman (1994) Small open economy Specific factor model: numeraire good produced via CRS, using labor alone n goods produced via CRS, using labor and a specific factor N citizens with identical preferences: u = x 0 + n u i (x i ) i=1

14 In a subset L of sectors, specific-factor owners overcome freeriding problems (Olson, 1965) and form a lobby. Each lobby offers political contributions to the incumbent government to influence its trade policy choices. Trade policy consists of ad valorem tariffs/subsidies, driving a wedge between domestic and world prices t i = (p i p * * i ) / p i Gross-of-contributions aggregate utility of lobby i: W i (p) = l i + π i (p i )+α i N[r(p)+ s(p)]

15 Semi-benevolent politician sets trade policy to maximize a weighted sum of lobby contributions and aggregate welfare: G = C i (p)+ aw (p) i=l n W (p) = l + π i (p i )+ N[r(p)+ s(p)] i=1

16 The interaction between lobbies an policymakers takes the form of a common-agency game (Bernheim and Whinston, 1986): 1) Each lobby simultaneously and non-cooperatively presents the incumbent with a contribution function, giving a binding promise of payment conditional on the chosen policy. 2) Given lobbies contributions functions, the government sets trade policy and collects the corresponding contributions. In a political equilibrium, neither the government nor any lobby has an incentive to alter its behavior.

17 Characteristics of lobbies contributions Feasible, i.e. no greater than total income of lobby s members - Differentiable at least around the equilibrium - Truthful, i.e. the slope of the contribution function is equal to the marginal benefit of the policy for the lobby: C i 0 (p 0 )= W i 0 (p 0 ) The government chooses vector of tariffs/subsidies t that maximizes t W i (t)+ a t W (t) = 0 i L

18 Equilibrium tariff/subsidy in each industry i: t i 1+ t i = I i α L a +α L " $ # z i e i % ' & for i =1, 2,..., n I i : indicator variable equal to 1 if sector i is organized into a lobby α L 1: fraction of the population organized into lobbies z i = X i /M i : equilibrium ratio of domestic output to imports e i : elasticity of import demand (positive) or export supply (negative)

19 Testable predictions of Grossman-Helpman (1994) An organized import-competing industry can buy an import tariff The extent of protection should fall with import elasticity (e i ) import penetration (M i /X i ) in organized industries

20 Goldberg and Maggi (1999) Predictions of G-H (1994) in empirically testable form: t i 1+ t i = α L a +α L " $ # X i / M i e i % '+ & 1 a +α L " $ # I X i / M i e i % ' & i =1, 2,..., n Empirical model estimated by Goldberg and Maggi: t i e i! = γ # 1+ t i " X i M i $! &+δ I X i # % " M i $ & +ε i i =1, 2,..., n (17) % Predictions: γ <0, δ >0, γ + δ >0

21 Main variables Protection measure (t i ): US NTB coverage ratio (1983) Import demand elasticities (e i ): estimates by Shiells et al. (1986) Political organization dummy (I i ): various thresholds of corporate contributions to determine whether and industry is organized. Import penetration ratio (X i /M i )

22 Variable Estimation results X i /M i ** I i X i /M i 0.01 ** N 107 Coefficients γ and δ have predicted signs and are statistically significant Estimates of γ and δ used to compute the implied structural parameters - government s weight on social welfare (β) as opposed to contributions (1-β) - proportion of the population organized in a lobby (α L )

23 Limitations of Goldberg and Maggi (1999) NTB coverage ratio inaccurate measure of the level of protection (e.g. non-binding quotas) Overall corporate contributions inaccurate measure of contributions (e.g. not trade related) exclusion of important sources of contributions (e.g. labor unions) Government s weight on welfare (β =0.98): do lobbies really matter? The G-H (1995) model might be the appropriate model to test since the US is not a small country and tariffs are often determined cooperatively (but need for data on different sets of countries).

24 Strengths of Grossman-Helpman (1994): Strong micro-foundations for government-lobby interactions Applicability to other economic problems (e.g. environmental taxation) Directly testable (e.g. Goldberg and Maggi, 1999) Limits of Grossman-Helpman (1994): No role for international negotiations (Grossman and Helpman, 1995) No analysis of lobby formation (Mitra, 1999; Bombardini, 2008) Other roles of lobbies: transmission of issue-specific information, access to politicians (Blanes-i-Vidal et al, 2012; Bertrand et al, 2014) No role for voters and electoral incentives (Conconi et al, 2014)

25 Trade Wars and Trade Talks (Grossman-Helpman, 1995) Two large open economies, home (no *) and foreign (*) Production and demand side of each economy as in G-H (1994) π i : international price p i = π i t i : domestic price Product markets clear when M i (π i, t i ) + M i * (π i, t i* ) = 0, i =1, 2,, n. (7) M i (p i ) = D i (p i ) - X i (p i ): net domestic imports

26 Two forms of interaction Non-cooperative (trade wars) Cooperative (trade talks)

27 Trade wars Stage 1: in each country, lobbies simultaneously and noncooperatively offer contribution functions relating their binding promise of political support to the selected policies Stage 2: given lobbies contribution functions and the other country s policy, each government sets its trade policy and collects the corresponding contributions

28 Equilibrium condition: tw i= L i * * ( t, t ) + a W( t, t ) = Equilibrium trade policies: t 0 (8) t I α X i 1 + for i 1,2,..., n * π M e il L i 1 = = * a + α L i i i (9) The first component of (9) represents the political support motives for trade intervention (as in G-H, 1994). The second component captures the terms-of-trade motives for trade intervention and represents the optimal tariff (or export tax) adopted by a large country (Johnson, 1954).

29 In the case of an import competing lobby, terms-of-trade considerations reinforce the industry s lobbying efforts An organized import-competing industry emerges from a trade war with an higher tariff than in the Johnson equilibrium

30 Trade talks Stage 1: in each country, lobbies present policymakers with contributions schedules tying their promised political support to the policies that emerge from international talks. Stage 2: given lobbies contribution schedules, governments set cooperative trade taxes and collect corresponding contributions.

31 Objective functions of the two governments: G = i L C i (t, t * ) + a [W (t, t * )] (10) G * = i L* C i* (t *, t) + a * [W (t *, t)] (11) Efficient cooperative policies must be such that G cannot be raised without lowering G *. The objective of a mediator or supra-national government is a * G + a G * (12)

32 Equilibrium conditions (13) 0 )], ( ) ( [ ), ( ) ( * * * * * = t t t,t t t t,t * * * * t t t t W W a a W a W a L i i L i i (14) 0 )], ( ) ( [ ), ( ) ( * * * * * * * * * = t t t,t t t t,t * t * t * t * t W W a a W a W a L i i L i i

33 From (12) and (13), we can determine the equilibrium trade policy ratios: % t i t * i = I α il L ' & a + α L X i π i M i ' ( % * I * * α il L ' ) a * * & + α L X i * π i M i *' ( * ) for i =1,2,...,n (15) Foreign elasticities do not enter into (15) because there are no terms-of trade motives for trade intervention. Protection rates reflect domestic and foreign lobbying.

34 Higher domestic tariffs if home lobby is stronger than the foreign The domestic lobby has more political power if foreign industry not organized (I il =1, I il* =0) greater stake in the negotiations (X i >X i* ) smaller government s weight on welfare (a i < a i* ) smaller proportion of population organized (α L < α L* ) If lobbies are equally strong, their political influences cancel out and international prices are the same as in free trade.

35 The role of electoral incentives

36 In last presidential campaign, Obama stated NAFTA has been devastating on the community ( ) I don t think NAFTA has been good for America, I never have. He later admitted that his rhetoric had been overheated and amplified ( ) Politicians are always guilty of that, and I don t exempt myself. On September 17, 2012, less than two months before facing re-election, and the same he was campaigning in the crucial swing state of Ohio, President Obama lodged a complaint against China at the World Trade Organization, alleging that it unfairly subsidizes car-part exports. There was nothing subtle about it but then subtlety does not win many elections (The Economist, Chasing the anti-china vote: a suspiciously timed dispute, September 22, 2012).

37 Policymakers Horizon and Trade Reforms: the Protectionist Effect of Elections Paola Conconi Université Libre de Bruxelles (ECARES) and CEPR Giovanni Facchini Nottingham University, Università di Milano, LdA, CEPR and CES-Ifo Maurizio Zanardi Lancaster University Management School

38 We study the determinants of U.S. congressmen s votes on all major trade liberalization reforms since early 1970s and provide systematic evidence that electoral incentives lead politicians to take a protectionist stance. In particular, we show that the political horizon of U.S. congressmen their term length and how close they are to facing elections crucially affects their support for trade liberalization reforms. Focus on the United States: Data availability Institutional features of the U.S. Congress: 6-year terms for Senate, 2-year terms for House Staggered structure of Senate

39 Main results House members are less likely to support trade reforms than Senate members. This is not true for senators in the last two years of their mandate, suggesting that inter-cameral differences are driven by differences in term length. The last generation of senators is more protectionist, both when comparing different senators on same bill and individual senators on different bills. This effect is pervasive: it applies to members of both parties and representatives of import-competing and export constituencies. No cycling behavior for senators with safe seats or retiring, suggesting that the protectionist effect of election proximity is driven by re-election incentives.

40 Outline of the presentation Related literature Dataset and variables Empirical methodology and results House versus Senate Different generations of Senate members Discussion Conclusions

41 Related literature Influence of domestic political factors on trade policy outcomes (e.g., Grossman and Helpman, 1994; Maggi and Rodriguez-Clare 1998). Political viability of economic reforms (e.g., Fernandez and Rodrik, 1991; Alesina and Drazen, 1991; Dewatripont and Roland, 1995; Drazen, 2000). Political business cycles (e.g., Rogoff and Siebert, 1988; Rogoff, 1990). Election proximity and legislative behavior (e.g., Amacher and Boyes, 1978; Thomas, 1985; Bernhard and Sala, 2006).

42 Related literature (cont.) Term length and legislative behavior (e.g., Titiunik, 2008; Dal Bo and Rossi, 2011). Analysis of U.S. congressmen s voting behavior on trade policy (e.g., Bloningen and Figlio, 1998; Baldwin and Magee, 2000; Karol, 2007). other economic policies (e.g., Peltzman, 1985; Mian, Sufi and Trebbi, 2010).

43 Data: votes on trade liberalization bills ( )

44 Data: control variables Legislators characteristics Senate Senate1, Senate2, Senate3 Party affiliation Gender and age Safe seat Retiring Lobby contributions from labor and corporate groups Constituencies characteristics Export ratio Concentration of exports and import industries Percentage of high skilled Population

45 Trade policy interest of constituency j captured by its export ratio: X t j Y t j Two difficulties to construct this measure for House members: District-specific information must be obtained by aggregating countylevel data; counties may be split in different districts; Redistricting: every ten years 435 congressional districts assigned across the United States depending on population.

46 Santa Clara County, California

47 Santa Clara County, California

48 House vs Senate To verify the role of term length, we study the run the following probit model: Pr(Vote t j =1) = φ(β 0 + β 1 Senate j + β 2 Χ t j + β 3 Z)

49

50 House versus different generations of senators To verify whether inter-cameral differences are driven by differences in term length, we compare House members with different generations of senators: Pr(Vote t j =1) = φ(γ 0 + γ 1 Senate1 t j + γ 2 Senate2 t j + γ 3 Senate3 t j + γ 4 Χ t j + γ 5 Z)

51

52 Different generations of senators To study the role of election proximity, we focus on Senate votes and estimate Pr(Vote t j =1) = φ(δ 0 +δ 1 Senate2 t j +δ 2 Senate3 t j +δ 3 Χ t j +δ 4 Z) Pr(Vote t j =1) = φ(λ 0 + λ 1 Senate2 t j + λ 2 Senate3 t j + λ 3 Χ t j + λ 4 Z + λ j )

53

54

55

56 The pervasiveness of senators cycling behavior Inter-generational differences are observed among senators from both parties and representing both import-competing and export constituencies.

57

58 The role of re-election incentives To verify if re-election motives are behind the protectionist effect of election proximity, we focus on the voting behavior of senators who hold very safe seats have announced their retirement

59

60 Additional robustness checks Additional political controls for legislators (incumbency, years in Congress, membership in Appropriations and Finance committees). Additional economic controls for legislators constituencies (real GDP per capita, unemployment rate, percentage of population above 65). Differences in the margin of passage of votes (close versus lopsided votes). Different subsets of trade votes (e.g., only votes on the ratification of the most important trade agreements negotiated by the US). Alternative measures of trade exposure for preferential trade agreements. Alternative measures of ideology (ACU indexes, DW-Nominate scores).

61

62 Main results Existing models in the political economy of trade do not consider the role of term length and election proximity and thus cannot explain our findings: House Representatives are generally more protectionist than senators, but inter-cameral differences disappear for the last generation of senators, facing re-election at the same time. Election proximity reduces support for trade liberalization, a result that holds when comparing different senators voting on the same bill and individual senators over time. Inter-generational differences in senators voting behavior are pervasive, i.e., apply even to representatives of export constituencies, in which most voters should benefit from trade liberalization. Election proximity has no effect on the voting behavior of senators who are retiring or hold very safe seats.

63 Discussion (cont.) Our findings suggest that office-motivated politicians pander toward the interests of protectionist voters when they approach re-election. This explanation requires the existence of two biases: Protectionist bias: when deciding whether to support trade liberalization, politicians are more responsive to the interests of protectionist voters. Recency bias: when deciding whether to re-elect incumbent politicians, voters are more responsive to recent policy choices.

64 Protectionist bias Deviations from free trade almost invariably aim at constraining imports rather than subsidizing them (Rodrik, 1995). Different mechanisms can give rise to a protectionist bias: Loss aversion in voters preferences (Freund and Ozden 2008; Tovar 2009) Information bias: voters are more informed about trade barriers that help them as producers than those that hurt them as consumers (Ponzetto 2011) Heterogeneity in voters preferences: trade policy is only salient to individuals who strongly oppose liberalization (e.g., Guisinger 2009).

65 Recency bias Behavioral economics: disproportionate salience of recent stimuli or observations is one of the cognitive biases affecting belief formation, decision making, and human behavior in general (e.g., Lee 1971). Political economy: Voters follow the what have you done for me lately? principle (e.g., Fiorina 1981; Weingast, Shepsle, and Johnsen 1981; Ferejohn 1986; Lewis-Beck and Stegmaier 2000; Eisenberg and Ketcham 2004). Imperfect information about politicians ability and the policy environment can lead voters to optimally attach more weight to recent policy choices (e.g., Rogoff and Sibert 1988).

66 Combining a protectionist bias in trade policy with a recency bias in voting provides a simple explanation for our findings: when deciding whether to support trade reforms, politicians pander toward the interests of protectionist voters; this bias is more pronounced at the end of their mandate, when their policy choices are more likely to affect voters decisions. This mechanism can also explain why the protectionist effect of election proximity is so pervasive: representatives of export constituencies pander toward the interests of a vocal minority that opposes liberalization.

67 Cycles in campaign contributions from lobbies could in principle explain electoral cycles in trade policy votes. Lobbying models do not examine calendar effects; even if there are cycles in campaign contributions, it is unclear how they could explain our findings, unless they systematically differ between pro-trade and anti-trade lobbies. Our empirical analysis shows that Contributions from labor (corporate) groups are associated with votes against (in favor of) freer trade. The protectionist effect of election proximity is not driven by cycles in campaign contributions: it is robust to controlling labor and corporate contributions received by senators throughout their mandates.

68 Conclusions We have examined the role of term length and election proximity on policymakers willingness to support trade reforms. Our results show that electoral incentives decrease legislators support for trade liberalization. Support for trade liberalization could be enhanced by Lengthening terms of elected politicians Delegating trade policy to non-elected officials

69 Ongoing research on the role of electoral incentives New theoretical models to shed light on the mechanisms through which electoral incentives affect policymakers choices. Empirical analysis of the role of electoral incentives on legislators votes on other secondary policy issues: - Gun control (Bouton, Conconi, Pino and Zanardi, 2014) - Environmental policy (Conconi, Pino and Zanardi, 2016)

70 Avenues of future research in political economy of trade Emergence of global value chains Does it fundamentally change the nature of political economy?

71 Fast Track Authority and International Trade Negotiations Paola Conconi Université Libre de Bruxelles (ECARES) and CEPR Giovanni Facchini Nottingham University, Università di Milano, LdA, CEPR and CES-Ifo Maurizio Zanardi Lancaster University Management School

72 Introduction The conclusion of international agreements between the U.S. and partner countries involves both Congress and the President. In the case of trade agreements, all deals negotiated by the President must be approved by Congress, which can amend their content. Art. I: Congress has the power to regulate commerce with foreign nations... and to...lay and collect taxes, duties, imposts, and excises.

73 Introduction Since 1974 U.S. legislators can decide to give up their power to amend trade agreements, by granting fast track authority (FTA) to the executive. Main features of FTA: Congress can only accept or reject deals negotiated by the President Deals must be ratified by Congress within 90 legislative days We examine the determinants of congressmen s FTA voting decisions.

74 Motivation In the absence of fast track, the inability to submit pacts to Congress without having them picked apart line-by-line diminishes trading partners enthusiasm for negotiations. Financial Times, September 25, 2007 In the absence of fast track, the prospects for completion of the Doha Round of global trade talks, as well as several proposed bilateral U.S. trade deals, remain bleak. Wall Street Journal, June 29, 2007 Without it (fast track authority), lawmakers can wreck carefully negotiated deals with toxic amendments. The Economist, February 22, 2014

75 Conferrals of FTA

76 Outline of the presentation Related literature Theoretical model A simple model of trade negotiations FTA and trade negotiations Empirical analysis Empirical methodology Data Results Conclusions

77 Related literature Interaction between domestic politics and trade negotiations Mayer (1981) Grossman and Helpman (1995) Maggi and Rodríguez-Clare (2007) Evolution of U.S. trade policy Lohmann and O Halloran (1994) Hiscox (1999) Determinants of congressional trade policy decisions Kahane (1996) Box-Steffenmeyer et al. (1997) Baldwin and Magee (2000)

78 Related literature (cont.) Strategic delegation in bargaining: principals may gain by delegating decision making power to status quo-biased agents, to increase their bargaining power in negotiations with third parties: The power of a negotiator often rests on a manifest inability to make concessions and to meet demands. (Shelling, 1956)

79 A simple model of trade negotiations Specific-factor trade model between Home and Foreign Three goods, i = 0, 1, 2, three factors of production. Good 0 is the numeraire, produced using only labor and freely traded Goods 1, 2 are produced using labor and a sector-specific factor Individuals share quasi-linear, separable preferences Home is abundant in the specific factor used to produce good 2 Foreign is abundant in the specific factor used to produce good 1

80 A simple model of trade negotiations Home imports good 1, exports good 2 (vice-versa for Foreign) Trade policies consist of ad valorem import tariffs/subsidies, τ and τ* Home and Foreign aggregate welfare: ), ( ) ( ) ( ) ( 1 ), ( * * 2 1 * τ τ τ τ τ τ τ Ω = T R R W ), ( ) ( ) ( ) ( 1 ), ( * * * 2 1 * * τ τ τ τ τ τ τ Ω = T R R W

81 Totally differentiating W, we obtain the slope of Home s indifference curves: Substitution yields where It follows that * * 2 1 * τ τ τ τ τ τ τ Ω + Ω + + = R T R d d * 2 2 * 1 1 * ) (1 τ π τε τ π τ τ d d M d d M d d = * 1 * 1 * 1 * 1 * M p dp dm = ε * * 1 )0 ( ε τ τ τ < d d

82 Home s indifference map τ * W 1 W 2 τ W 3

83 Negotiations between Home and Foreign executives W * Z τ * Z W Z 0 τ

84 Assumptions on international agreements Assumption 1: Executives can only agree to trade deals that make each of them at least as well off as they are in the status quo. Assumption 2: Executives can only agree to tariff combinations such that no additional welfare gains can be achieved by one of them without the other one losing. Assumptions 1 and 2 imply that the executives agree to combinations of import tariffs (subsidies) on the AB arc of the contract curve CC.

85 Negotiations between Home and Foreign executives τ * W Z * C Z W Z A 0 τ B C

86 Bargaining between the two executives To derive the equilibrium outcome of the trade negotiations, we employ the generalized Nash bargaining solution. Domestic and foreign tariffs are chosen as the solution to the following maximization problem: max( W W Z ) γ W * * W Z W,W * ( ) 1 γ where γ captures Home s relative bargaining strength.

87 Bargaining between the two executives max( W W Z ) γ W * * W Z W,W * ( ) 1 γ W * A 0 E Z B W

88 Congressional preferences D electoral districts characterized by identical population shares but different production shares of the two goods. Congressmen represent the interests of their constituencies, which differ due to the uneven geographical distribution of industries: Import districts (M): a fraction β M of the districts is relatively specialized in the production of the import-competing good; Export districts (X): a fraction β X of the districts is relatively specialized in the production of the export good; Non-specialized districts (N): the remaining fraction β N of districts has equal stakes in the production of the two goods.

89 Congressional preferences W M (τ,τ * ) = h + α M 1 R 1 (τ) + α M 2 R 2 (τ * ) + h( T(τ) + Ω(τ,τ * )) with α 1 M > h > α 2 M W X (τ,τ * ) = h + α X 1 R 1 (τ) + α X 2 R 2 (τ * ) + h( T(τ) + Ω(τ,τ * )) with α 2 X > h > α 1 X W N (τ,τ * ) = h + hr 1 (τ) + hr 2 (τ * ) + h( T(τ) + Ω(τ,τ * ))

90 Preferences of Home congressmen τ * W Z M W Z X Z W Z N τ

91 The model: stages of the game FTA voting International negotiations Congressional approval t = 1 t = 2 t = 3

92 Game Tree

93 FTA and trade negotiations Each legislator in Home votes for or against FTA so as to maximize his expected utility, anticipating the impact that FTA (or lack thereof) will have on the outcome of the negotiations with Foreign. Absent FTA, the game s outcome is the same as if the foreign President negotiated directly with Home s Congress majority. FTA voting decisions thus imply a choice between two country representatives: the President and Congress majority. Assumption 3: In the absence of FTA, the foreign executive and a majority of Home legislators agree to tariff combinations such that no further welfare gains can be achieved by one without the other one losing. Assumption 4 (ratification constraint): trade agreements must make a majority of Home legislators at least as well off as in the status quo.

94 No majority When β i < 1/2 for all i {M, X, N}, trade deals can only be amended in the third stage of the game by a coalition of representatives. The only possible amending coalition is between N and M (under mild condition of Lemma 1): W M,N = β M W M + β N W N An increase in β N makes the indifference curves of the coalition flatter, making N representatives less likely to vote in favor FTA.

95 Trade negotiations between foreign executive and coalition of M and N legislators

96 Main results Proposition 1: When β i < 1/2 for all i {M, N, X}, representatives of M districts will never vote in favor of FTA; representatives of X districts will vote in favor of FTA whenever N representatives do.!

97 Preferences of X and N legislators under no majority

98 Main results Proposition 1: When β i < 1/2 for all i {M, N, X}, representatives of import districts will never vote in favor of FTA; representatives of export districts will vote in favor of FTA whenever non-specialized district representatives do. Proposition 2: When β i < 1/2 for all i {M, N, X}, N representatives will vote against (in favor of ) FTA if their share of seats in Congress is above (below) a threshold. N β^

99 Empirical predictions When no district type has a majority, the likelihood that a U.S. congressman votes in favor of FTA increases with the degree to which his own constituency is relatively export-oriented compared to the U.S. as a whole. When no district type has a majority, the likelihood that representatives from non-specialized constituencies vote in favor of FTA decreases with their relative share in Congress.

100 Empirical methodology Probit model to explain FTA voting decisions by congressmen: i i P( vote t ) = Φ( α + β Xt + β2 1 Z X ti : district-specific and time variant variables Z: other control variables ) Main explanatory variable of interest: ratio of employees in export industries to import-competing industries, relative to the U.S: L i US x, t x, t i x US x i λ t =, λ i t =, Λ US t = Lm, t Lm, t m m L λ λ i t US t

101 Data 13 votes over the period granting or extending FTA. Bill Description Vote in House Vote in Senate H.R Trade Act of 1974 First approval of FTA Other provisions: escape clause, antidumping, countervailing duties, trade adjustment assistance, GSP Dec. 11, 1973 ( ) Dec. 20, 1974 (72-4) H.R Trade Agreements Act of 1979 Extension of FTA Other provisions: implementation of Tokyo Round July 11, 1979 (395-7) July 23, 1979 (90-4) H.R Omnibus trade and Competitiveness Act Approval of FTA Other provisions: strengthening of unilateral trade retaliation instruments, authority of USTR July 13, 1988 (376-45) Aug. 3, 1988 (85-11) H.Res. 101 Disapproval of extension of FTA May 23, 1991 ( ) S.Res. 78 Disapproval of extension of FTA May 24, 1991 (36-59) H.R Extension of FTA June 22, 1993 ( ) June 30, 1993 (76-16) H.R Extension of FTA Sept. 25, 1998 ( H.R Trade Act of 2002 Approval of FTA Other provisions: Andean Trade Preference Act, trade adjustment assistance, GSP July 27, 2002 ( ) Aug. 1, 2002 (64-34)

102 Data From roll call voting records, we collect votes in favor or against FTA, identity of congressmen, Congress branch, party affiliation. Two difficulties to construct Λ t i for House members: 1) District-specific information must be obtained by aggregating county-level data; counties may be split in different districts;

103 Santa Clara County, California

104 Santa Clara County, California

105 Data From roll call voting records, we collect votes in favor or against FTA, identity of congressmen, Congress branch, party affiliation. Two difficulties to construct Λ t i for House members: 1) District-specific information must be obtained by aggregating county-level data; counties may be split in different districts; 2) Redistricting : every ten years 435 congressional districts assigned across the United States depending on population.

106 Congress composition by district type

107 Empirical results trade exposure

108 Empirical results trade exposure

109 Empirical results trade exposure

110 Empirical results N constituencies

111 Robustness checks Alternative definitions of legislators identity (i.e., g values) Falsification exercise on non-fta votes Including Agriculture; services; other economic controls Within-county heterogeneity Clustering of standard errors Regressions by decades

112 Conclusions The current U.S. institutional setting involves the recurrent choice between partial delegation (FTA) and no delegation (no FTA). Our model shows that no delegation tends to skew trade agreements in favor of the U.S., and can thus explain why other countries are reluctant to negotiate with the U.S. in the absence of FTA. Our results are in line with the literature on strategic delegation, which shows how principals may gain by delegating to status-quo biased agents to increase their bargaining power in negotiations with other parties (e.g. Schelling, 1956).

113 Conclusions The foreign country might be tempted to adopt similar institutional procedures, leaving trade negotiations in the hands of protectionist legislators, to try to skew trade agreements in its favor. In this case, both countries would end up being worse off than if they could commit to delegate trade negotiating authority to their executives (e.g. Jones, 1986). In a multi-country setting, lack of FTA would be more costly for the U.S., leading its trading partners to negotiate with other countries.