1 The Conditional Nature of Administrative Responsiveness to Public Opinion Julia Rabinovich Northwestern University EITM July 2007
2 Outline of the Paper Public opinion and administrative policies Gaps in the literature Theoretical model Congressional oversight as a link between public and the executive agencies Legislative responsiveness does not always induce administrative responsiveness Empirical model Future research challenges
3 Public Opinion and Policy Focus almost exclusively on legislative responsiveness Problem with the definition of policy Electoral connection as the main link between the public and policy Aggregate policy studies (e.g., Page and Shapiro 1983, 1992; Monroe 1979, 1998; Erikson, MacKuen and Stimson 2003) Individual voting behavior studies (e.g., Bartels 1991, Bailey 2001)
4 Administrative Policymaking Focus on the role of political elites Successful congressional oversight (e.g., McCubbins, Noll and Weingast 1987, Weingast and Moran 1983, McCubbins and Schwartz 1984) ex-ante and ex-post mechanisms Interest groups (e.g., Stigler 1971, Moe 1989, Carpenter 2002) Direct lobbying Indirect impact via legislative lobbying and campaign contributions
5 Administrative Responsiveness to Public Opinion A large share of policymaking is delegated Vague legislation Bureaucratic expertise There is no direct electoral connection between the public and administrative policymakers Congressional oversight is the main mechanism connecting the public and administrative policies Committee oversight Hearings Budgets
6 The Conditional Nature of Administrative Responsiveness (summary of theoretical results) Contrary to conventional wisdom, legislative responsiveness does not always translates into administrative policy responsiveness to public opinion In some cases, legislative responsiveness indeed induces administrative responsiveness however, under certain political circumstances, such as a divided government, greater legislative responsiveness to public opinion can result in less responsiveness from executive agencies.
7 A Model of Conditional Administrative Responsiveness to Public Opinion The model includes four actors: a legislature (L), a committee with jurisdiction over the agency’s policy realm (C), an agency (A) and the general public (P). The payoff functions: The decisive member of the agency: u A (x)= -(x – A) 2 The decisive member of the oversight committee: u c (x)= -(x – C) 2 The decisive member of the legislature: u L (x)= -α (x – P) 2 – (1- α)(x – L) 2
8 Sequence of Actions Stage 1: The agency makes a policy proposal (x). Stage 2: In response to the agency’s proposal, the committee either introduces a bill (b) or refrains from introducing a bill (gatekeeping). If the committee decides not to propose a bill, the game ends and the outcome is the originally proposed policy (x). Stage 3: If the committee decides to introduce a bill to the floor, the legislature can either accept, reject or amend the proposed legislation. If the legislature rejects the legislation, the outcome is the originally proposed policy (x). If the legislature accepts the legislation, the outcome is the committee’s proposal (c). If the legislature decides to amend the legislation, it will set it at its ideal policy (assuming open rule) and the final outcome would be an amended bill (b’).
9 Equilibrium Strategies Equilibrium strategies (SPE): A, if A (min (X (P, NP), C(X (P,NP)), max (X (P, NP), C(X (P,NP))) – regime 1 X*=max (X (P, NP), C(X (P,NP), if A max - regime 2 min (X (P, NP), C(X (P,NP), if A min- regime 3 where X (P, NP) = argmax xЄR U L (x)=αP + (1-α)NP; and C (X (P,NP)) = 2C – X(P,NP)
10 Two Views of Responsiveness : Responsiveness as a movement of public opinion and policy in the same direction To devise comparative static for this notion of responsiveness, I am solving for each one of the equilibrium regimes. The solutions for the aforementioned derivative can be summarized in the following 3 conditions: Condition 1: When X(P,NP) >A>C (X(P,NP)) or C(X(P,NP))>A>X(P,NP), the expectation is that there is no relationship between public opinion and policy. I call this condition “nonresponsiveness”. Conditions 2: When A>X(P,NP)>C(X(P,NP)) or C(X(P,NP))>X(P,NP)>A, we should expect public opinion and policy to move in the same direction. I call this conditions “responsiveness”. Condition 3: When A>C(X(P,NP) >X(P,NP) or X(P,NP)>C(X(P,NP)>A, we should expect to find negative relationship between public opinion and policy. Thus, I call this condition “adverse responsiveness”.
11 Two Views of Responsiveness : Responsiveness as a distance between policy and public’s preferences Under this notion of responsiveness, I am analyzing the impact of legislative responsiveness, which is represented in the legislature utility function as α, on the absolute distance between public’s preferences and policy, i.e. I analyze 3 conditions: 1. “Nonresponsiveness condition”: legislative responsiveness has no effect on administrative responsiveness to public opinion when either X(P,NP)<A<C(X(P,NP)), or C(X(P,NP)) <A<X(P,NP). 2. “Convergence condition”: legislative responsiveness increases administrative responsiveness when either A>X(P,NP)>C(X(P,NP), or C(X(P,NP))>X(P,NP)>A, or X(P,NP)<P<C(X(P,NP))<A or A<C(X(P,NP))<P<X(P,NP). 3. “Divergence condition”: legislative responsiveness decreases administrative responsiveness to public opinion when either P<X(P,NP)<C(X(P,NP)<A, or A<C(X(P,NP)<X(P,NP)<P, or X(P,NP)<C(X(P,NP))<P and C(X(P,NP)<A, or P<C(X(P,NP)<X(P,NP) and A<C(X(P,NP).
12 Comparative Statics Convergence Nonresponsiveness X* X(L,P) = argmax xεR U L (x)=αP + (1-α)L C(X(L,P))=2C-X(L,P) – the committee is indifferent between these point and the legislature's ideal policy point (X(L,P))
13 Equilibrium Strategies and Comparative Statics - Continued Convergence II X* Divergence X*
14 Statistical Model |Y t -Public Opinion| = β 0 + β 1 convergence*legislative responsiveness + β 2 divergence*legislative responsiveness + β 3 nonresponsiveness*legislative responsiveness + β i X i,t + ε t where Y t is a measure of administrative responsiveness to public opinion at time t. Convergence, divergence and nonresponsiveness are dummy variables set equal to one when the condition is present and zero otherwise. X i,t is a vector of control variables, including committee preferences, election year dummy and appropriation data and ε t is an error term.
15 To be continued…