Distributive politics of public goods networks through multistage voting

• Published date: 01 February 2020
• Author: Kengo Kurosaka
• Date: 01 February 2020
• DOI: http://doi.org/10.1111/jpet.12395

J Public Econ Theory. 2020;22:146–169.wileyonlinelibrary.com/journal/jpet146

|

© 2019 Wiley Periodicals, Inc.

Received: 28 January 2019

|

Revised: 20 July 2019

|

Accepted: 2 August 2019

DOI: 10.1111/jpet.12395

ORIGINAL ARTICLE

Distributive politics of public goods networks

through multistage voting

Kengo Kurosaka

Department of Economics, Hokkaido

Musashi Women’s Junior College,

Sapporo, Hokkaido, Japan

Correspondence

Hokkaido Musashi Women’s Junior

College, Kita22 Nishi13 Kita‐ku, Sapporo,

Hokkaido 001‐0022, Japan.

Email: kurosaka@hmjc.ac.jp

Funding information

Japan Society for the Promotion of

Science, Grant/Award Number: 23‐8274

Abstract

Public goods networks create conflict between districts

needing the network and districts that pay the costs.

Thus, it is difficult for pork barrel politicians to achieve

logrolling with other districts. I show that logrolling is

possible for such politicians if they set their agenda

properly. To show this, I model the centralized decision‐

making process as a multistage vote among representa-

tives from different districts. I assume the districts are

better off if and only if the public good “connects”them

to the network. Public goods can be provided universally

by setting the agenda properly, even when costs exceed

the benefits.

1

|

INTRODUCTION

Public goods, such as irrigation systems and transportation infrastructures, are often regarded

as a target of pork barrel politicians logrolling with other districts and bringing projects to their

own. However, it can be difficult for pork barrel politicians to cooperate with other districts, as

public goods networks trigger conflict between those who benefit from the relevant project and

those who merely pay its costs. For instance, the construction of the Erie Canal in the 19th

century was opposed by farmers located alongside the Hudson River, as they already had a

connection to New York City. In the case of the high‐speed railroad project in Kyushu, Japan,

the cities already connected to Tokyo were reluctant to approve the extension. How, then, can

pork barrel politicians cooperate with geographically distant colleagues and bring local public

goods to their own districts?

This study examines the power of agenda‐setting and determines to what extent it affects the

shape of public goods networks. In the case of the Erie Canal, the project was divided into three

districts, from the nearest to New York City: (a) east section, (b) middle section, and (c) west

section. The construction began from the middle section, then moved on to the west section,

before finally ending with the east section. The high‐speed railroad in Japan was also divided

into three districts: (a) Tokyo to Hakata, (b) Hakata to Shin‐Yatsushiro, and (c) Shin‐Yatsushiro

to Kagoshima‐Chuo. Its construction began from the Tokyo to Hakata section, then moved on

to the Shin‐Yatsushiro to Kagoshima‐Chuo section, and finally ended with the Hakata to Shin‐

Yatsushiro section. In both cases, the decision‐making process was divided into small steps, and

some steps were executed in the reverse order from what was initially planned. Many historical

records suggest that these orders were deliberately chosen by authorities so that they could

design the networks, which suggests the power of agenda‐setting.

1

To examine the power of agenda‐setting explicitly, I use a model in which the

representatives from each district sequentially vote on the public goods network. The key

assumption of my model is that the benefits from public goods are strongly dependent on

other districts. More precisely, voters are better off if and only if the public goods connect

voters to the origin or to those who already have a connection—the former can be a water

sourceforanirrigationnetworkorthecitycenterforacommuterrail.Anotherkey

assumption of my model is sophisticated voting, which refers to the backward‐inductive

actions taken by voters, first advocated by Farquharson (1969). I find that a public goods

network can be universally provided if and only if the voting on the median interval, which

is roughly the middle of the network, occurs at the end of the multistage voting. Likewise, I

find that almost any type of public goods network can be approved by multistage voting as

long as the network is connected and longer than half of the full range. I also find that a

public goods network can be universally provided even where the costs of construction

exceed the benefits from connection.

To the best of my knowledge, the problem of providing a publicly funded network through

collective decision‐making was first discussed by Tullock (1959). He demonstrates the

possibility of logrolling by describing the case of farmers who wish to be connected to a

major road through publicly funded local roads; however, Tullock (1959) explicitly deals with

neither the collective decision‐making process among farmers, nor the property of network

externality.

2

Rational choice models later developed various collective decision‐making models

to explain pork barrel politics in legislature. These models include the agenda‐setting model of

Ferejohn, Fiorina, and McKelvey (1987), and the bargaining model of Baron and Ferejohn

(1989).

3

The topics of distributive politics and network externality meet in the field of public

economics courtesy the seminal work of Oates (1972), which addressed the question of who

should finance local public goods with interdistrict externalities. Lockwood (2002) focuses on

the decision‐making process in the central government, applying the framework of Ferejohn

et al. (1987), and investigates under which condition the decentralized provision of local public

goods can be supported. Xie and Levinson (2009) detail the preference of residents on the

network depending on the distance to the city center; however, they do not consider the

structure of public goods networks.

The contribution of this paper is twofold. First, this is the first work to describe the impact of

agenda‐setting explicitly in the context of distributive politics. Ferejohn et al. (1987) model

the conflict among districts in the provision of local public goods as multistage voting by the

representatives. However, the difficulty of multistage voting is that the outcomes depend on

the procedure—that is, the agenda‐setting—especially when voters are backward inductive in

the sense of Farquharson (1969). They find the conditions under which local public goods are

1

See Appendix H for the episodes of Erie Canal and of the high‐speed railroad in Kyushu, Japan.

2

See Mueller (2003) for the development of the literature of logrolling.

3

There are many other types of collective decision‐making models. See Collie (1988) for a survey of the early studies on this topic.

KUROSAKA

|

147