Winning Proposal to the Pierce Scholar's Fund: 2011 Competition

Project Title: Party Nationalization, Partisan Pork and Public Goods Provision in Thailand

Faculty Sponsor and PI: Brian Min

Ph.D. Student: James Atkinson

Project Description

What accounts for variation in public goods provision within a given country? Some of the redistributive literature claims that parties may reallocate public funds by focusing on districts with certain types of voters. The loyalty voter theory argues that politicians will redistribute according to decreasing levels of risk, with the least risky group, supporters, receiving the most funds and the most risky group, opposition, receiving the least (Cox & McCubbins 1986). Ansolabehere & Snyder (2006) provide evidence from U.S. county transfers that supports this loyalty voter theory. Lindbeck & Weibull (1987) present a different model, which focuses on swing voters. Using a formal model, they find that, in equilibrium, the parties propose the same redistributive policy focused on groups with weak party preferences. Furthermore, groups are more desirable for redistribution if the group is relatively numerous at the cutpoint between the parties and the group is poor and more responsive to economic benefits (Dixit & Londregan 1998). This theory finds empirical support in research on intergovernmental transfers in Sweden (Dahlburg 2004) and in India (Dasgupta 2008).

Alternatively, several scholars have focused on how institutional variation affects the distribution of public goods in cross-country studies. One explanation focuses on the effect of party nationalization on the distribution of public goods (Hicken, Kollman & Simmons 2010). The authors argue that nationalized parties have a broader national constituency that expects a focus on more national-level policies. As the party system overall becomes more nationalized (less fragmented), parties should provide higher levels of public services at a nationwide level. The authors present cross-country evidence suggesting that higher levels of party fragmentation result in a decreased provision of several health indicators.

This project considers how institutional variation in party nationalization over time and within a given country affects the distribution of public goods. Since 1992, Thailand has had seven different governments, ranging from an unsteady coalition of several parties to a single-party majority. A key institutional change in the party system occurred in 1997 with the formation of a new constitution. Some of the constitutional reforms focusing on creating stronger, more nationally-oriented parties seemed successful in that they reduced the effective number of parties at the national level from 7.2 during the period of 1986-1992 to 2.6 in 2005 (Hicken 2006). Thus, these constitutional reforms transformed the Thai party system from a fragmented party system to a more nationalized party system in a short time period.

This project seeks to adjudicate between the partisan voter theories (swing and core) and the party nationalization theory of redistributive politics using electoral data and nighttime lights satellite imagery across 400 electoral districts in Thailand from 1992-2009. The key dependent variable will be the change in electrification level from the 1992 base line in a given district. This variable comes from the Nighttime Lights Time Series data set of the Department of Meteorological Satellite Program’s Operational Linescan System (DMSP-OLS). Visible and infrared imagery from DMSP-OLS instruments are used to monitor the global distribution of clouds and cloud top temperatures twice each day. The nighttime lights imagery, however, also records the aurora, city lights, manmade and natural fires, and natural gas flaring (NGDC 2010). The images can be examined at a pixel level that covers approximately 1 square kilometer with each pixel encoded on a scale of 0 to 63 that designates the relative measure of brightness (Min 2010). The key independent variables for this project will include measures for party system nationalization, vote share of the national party (coalition) in power, and the competitive level of a given district election.

If either of the partisan voter theories were true, we would expect either the national party in power’s vote share or the level of competitiveness in a given district (but not both) to have a positive effect on the change in brightness level in the district. We, however, expect that neither of these partisan theories provide an adequate explanation. Instead, we expect that higher levels of party system nationalization will result in increases in electrification levels in a given district. Thus, the overall level of public goods should increase in each district when the party system is less fragmented. Furthermore, we suspect that parties may choose partisan redistributive strategies conditioned on a given institutional environment. For example, at high levels of party nationalization, voters may view a party that provides more electricity to areas of strong core support as failing to meet their expectations of a focus on broad national-level policies. In the same institutional environment, however, voters might view a party that provides more public goods in swing districts more favorably. Thus, we will test whether the effect of vote share or competitiveness on the change in brightness levels in a given district depends on party nationalization.

In summer 2011, we propose to turn this idea into a conference paper to be presented at MPSA 2012 and to subsequently submit this paper for publication consideration in a peer-reviewed journal. The scope of the project in summer 2011 will consist of three main phases. In the first phase, Atkinson and Min will further develop the theoretical arguments for this paper. During the second phase, Atkinson will construct a data set consisting of electoral, economic and satellite imagery data for 400 districts in Thailand from 1992-2009. Throughout this phase, Min will provide methodological support as Atkinson continues to learn the ArcGIS software. The final phase will include data analysis and the drafting of a conference paper by both Atkinson and Min, which they will submit for presentation at MPSA in 2012.

This experience will provide several benefits for Atkinson. He will learn from Min’s expertise in ArcGIS and with the Nighttime Lights Dataset and will develop better spatial analytics skills. Furthermore, Atkinson will benefit from the preparation of a paper that he can present at MPSA the following year. Finally, this project allows Atkinson to incorporate his study of the Thai language and of Southeast Asian politics into a project with the potential to develop into a broader dissertation topic for Atkinson.