Electoral incentives, public policy, and the New Deal realignment.

• Author: Fleck, Robert K.

1. Introduction

   This paper develops a model of the effects of electoral incentives on policy, then applies the model to the New Deal realignment. In the model, policy is the outcome of an agenda-setter game between the president and legislators. Specifically, the president sets policy subject to legislative approval. The president's ability to concentrate benefits in states with high electoral payoffs depends in part on his or her power to influence legislators' prospects for reelection. Regression analysis shows that New Deal spending and roll call voting patterns in the House of Representatives strongly support the model. Historical accounts of other aspects of New Deal policy, including positions on labor and civil rights issues, are also consistent with the model. Together, the theoretical results and the empirical evidence contribute to the understanding of several striking features of the policy and politics of the 1930s.

   Overview of the Theoretical Implications and Empirical Findings

   One of the model's key implications is that a reelection-seeking president has an incentive to seek policy more favorable to swing states than to states loyal to his or her party. Like the president in the model, once Roosevelt was in office, he, along with other Democrats who sought to build and maintain a broad national base of support for their party, had incentives to design policy that would win support in swing states. When applied to New Deal distributive policy, the model predicts that, to the extent that Roosevelt influenced the distribution of funds, spending data should show that more money went to swing states than to the traditionally Democratic South. This prediction fits Wright's (1974) findings and is confirmed by the empirical evidence in this paper.

   In the model, the president's influence over policy is constrained by the need to have policy approved by a majority in the legislature. The legislative constraint will be more relaxed when the president has greater ability to influence the reelection prospects of legislators. Consequently, when applied to the New Deal, the model predicts that spending data should show, first, an increase in the favoritism of swing states from the 1933-1934 period to the 1937-1938 period (when the legislative constraint was relaxed) and, second, a decrease in the favoritism of swing states from 1938 to 1939 (when the legislative constraint was tightened). The empirical evidence confirms both of these predicted changes in spending.

   The model also predicts the following: As a consequence of presidential incentives to favor swing states, the president is likely to face legislative opposition from his or her party's traditionally loyal states. The empirical evidence confirms this prediction. This explains why, with the rise of the conservative coalition in the late 1930s, the traditionally Democratic South provided the strongest Democratic congressional opposition to Roosevelt and the New Deal.

   Contribution to the Literature

   By developing a model of distributive politics and applying the model to the New Deal realignment, this paper contributes to several branches of research. One branch is distributive politics. With the assumption that the president determines policy subject to legislative approval, my model incorporates the basic principle of agenda-setter models (e.g., Romer and Rosenthal 1978; Rosenthal 1990). Modeling the president as the agenda setter adds to the insight provided by models of distributive politics within legislatures (e.g., Weingast, Shepsle, and Johnsen 1981) and models of distributive games between the president and Congress (Kiewiet and McCubbins 1985a, 1988). The findings also add to other work that has examined incentives to favor swing and loyal voters (e.g., Wright 1974; Kiewiet and McCubbins 1985b; Cox and McCubbins 1986).(1)

   Furthermore, my research contributes to the literature on critical elections and realignment (e.g., Key 1955; Burnham 1970; Ginsberg 1972, 1976; Sundquist 1973; Sinclair 1977, 1985; Clubb, Flanigan, and Zingale 1980; Brady and Stewart 1982; Brady 1988; Nardulli 1995). Compared to previous work, my approach provides a more detailed examination of the reelection motives of Democrats following their rise to power in the early 1930s. This approach helps to explain an essential part of the realignment because, while public disapproval of the Republican Party's handling of the Great Depression gave the Democrats a chance to govern, it did not guarantee that the Democrats would stay in power. Once in office, President Roosevelt and his fellow Democrats had to provide popular policy in order to be reelected.

   By explaining the links between reelection incentives and New Deal policy, my work complements the previous literature's emphasis on critical elections. The critical election of 1932 gave Democrats a chance to govern and, therefore, played a fundamental role in realignment. But to understand the massive political changes of the 1930s, it is essential to consider two phenomena, both of which can be viewed as realignment. The first is a change in party dominance: The Democrats replaced the Republicans as the dominant party. The second is a shift in the alignments within the Democratic Party: With the rise of the conservative coalition in the late 1930s, many southern Democrats in Congress became frequently disloyal to their own party.

   Understanding the link between these two realignment phenomena is facilitated by an understanding of the reelection motives analyzed in this paper. While policy favoring swing states was valuable for Roosevelt's reelection and for maintaining the Democratic Party's dominance at the national level, it caused legislators from traditionally loyal Democratic states to oppose Roosevelt and the New Deal. Thus, policy based on the incentives of Democrats to maintain the realignment with respect to party dominance led to the realignment within the Democratic Party. These findings complement not only the works cited above but the vast historical literature on the political economy of the New Deal (e.g., Freidel 1965, Patterson 1967, 1969; Sitkoff 1978; Schulman 1991; Brinkley 1995).

   The findings in this paper also add to the empirical literature on the political economy of New Deal spending. This paper builds most directly on the work of Wright (1974), who constructs measures of political productivity and argues that spending was higher where it produced greater expected electoral benefits for the administration. He shows that per capita spending tended to be low in states that had been traditionally Democratic. Spending was greater in states where elections were more likely to be influenced by the level of benefits and, as a consequence, the funds would be more productive in maintaining the Democratic Party's new base of support.(2)

   Structure of the Paper

   Section 2 presents the model and derives the testable hypotheses that form the basis of the empirical analysis. Section 3 explains how the model applies to the New Deal; it also describes the spending, constituency, and roll call data used to test the hypotheses derived from the model. Section 4 presents the empirical tests. Section 5 discusses the rise of the conservative coalition, explaining the Democratic Party's division over civil rights issues, as well as economic policy, in the framework of the model. Section 6 concludes the paper.

2. Reelection Motives and Distributive Policy: A Model

   This section presents the model's assumptions, along with a discussion of the model's solution and implications. For a mathematical presentation of the model, see Appendix A.

   Objectives and Assumptions of the Model

   The model's assumptions are based on the following objectives. First, the model is intended to focus on the effects of electoral incentives. Second, the model must allow for executive and legislative influence on policy. Specifically, it should capture the effect of interaction between the president and legislators when the president has influence as an agenda setter. Third, the model should consider the value of policy to voters. It should incorporate the fact that the policy that wins the most votes in one state may be unpopular in another state, and it must have a useful spatial interpretation (i.e., indicate whether some policy position is close to what voters want). Fourth, the model should capture two intuitive points about reelection of the president: (i) By implementing policy that pleases voters in a state, the president will obtain a larger vote share in that state, and (ii) the president's vote share is not known with certainty when policy decisions are made. Fifth, the model should reflect two factors influencing the reelection of legislators: (i) A legislator is likely to receive more votes if, ceteris paribus, the legislator votes for the outcome with the higher value to his or her state, and (ii) a legislator is likely to receive more votes if a popular president assists (or refrains from opposing) the legislator's campaign.

   Assumptions

   (A.1) Electoral Incentives. The president and legislators are concerned exclusively with the next election. The president seeks to maximize the expected number of states in which he or she receives a majority vote. Each legislator seeks to maximize his or her own expected vote share.

   (A.2) Executive and Legislative Influence on Policy. Policy is the outcome of a one-round game between the president and legislators. First, the president selects a proposed policy. Second, legislators vote under majority rule to approve or reject the president's proposal. Rejection causes policy to revert to a known outcome (the status quo) equally distant to all states' ideal points.

   (A.3) The Value of Policy to Voters. To capture conflicting interests between states, the number of policy dimensions is assumed to equal the number of states. On each dimension of policy, one state has a high marginal net value curve...