Reevaluating Spending in Gubernatorial Races: Job Approval as a Baseline for Spending Effects

• Published date: 01 March 2005
• Author: Kedron Bardwell
• DOI: 10.1177/106591290505800109
• Date: 01 March 2005
• Subject Matter: Articles

In both state and federal campaigns, incumbents have

high success rates when they run for reelection. Mem-

bers of the media and many scholars attribute this fact to

the “perks of the office” and other inherent advantages of

incumbency, especially incumbents’ ability to raise and

spend more money than challengers. But is an advantage in

campaign spending the main source of incumbents’ near

invulnerability? Undoubtedly, there is a large disparity

between incumbent and challenger fundraising in most state

and federal elections. From 1980 to 2000, on average,

gubernatorial challengers raised and spent only 61 percent as

much money as their incumbent opponents did. In this same

period of time, U.S. Senate challengers faced comparable

fundraising disparities in their bids to unseat incumbents.

And gubernatorial and senatorial challengers both fared

better than challengers in U.S. House races (Herrnson 2000).

Even so, there are conflicting research findings on the

real effects of campaign spending in these races. Scholars

agree that challenger spending reduces incumbents’ share of

the vote, but findings regarding the effects of incumbent

spending have varied based on how models of the vote are

specified. Research using ordinary least-squares (OLS)

regression tends to show that spending by incumbents has

little independent effect on the vote (Abramowitz 1988,

1991). But because candidates’ and contributors’ expecta-

tions about the closeness of a race influence fundraising and

spending levels, most scholars use two-stage least-squares

(TSLS) regression to correct for simultaneity bias. Some

TSLS analyses confirm OLS findings (Jacobson 1978, 1980,

1985), but others indicate that the dollar-for-dollar impact

of incumbent spending is just as large as that of challenger

spending (Green and Krasno 1988, 1990; Gerber 1998).

This article uses new data and tests a theoretical per-

spective that aims to shed light on this debate. It builds on

previous research in two ways. First, it extends the scope of

spending effects research to gubernatorial elections. Second,

it uses newly available data on gubernatorial job approval

ratings to control for incumbent vulnerability in the year

before the election. This measure sets a baseline for the

expected vote and keeps variables that intervene in a cam-

paign from taking credit for pre-campaign levels of incum-

bent popularity. After a literature review that highlights

methodological issues in research on spending effects,

instruments for incumbent and challenger spending are

constructed and used in a TSLS regression analysis of races

from 1980 to 2000. By controlling for the governor’s vul-

nerability, the analysis shows that incumbent spending does

not have an independent effect on the vote. This implies

that governors’ spending buys little new support in the

course of general election campaigns.

RESEARCH ON SPENDING EFFECTS

The effects of campaign spending on the vote are hard to

measure. This is due in part to the fact that spending is

strategic: incumbents who are vulnerable spend more than

incumbents whose seats are safe (Green and Krasno 1990).

This relationship between spending and the expected vote

causes basic OLS models to underestimate the impact of

incumbent spending and overestimate the influence of chal-

lenger spending (Erikson and Palfrey 2000; Gerber 1998).

As such, recent research has focused on how to correct for

the simultaneity bias between incumbent spending and elite

expectations of the vote (Abramowitz 1988, 1991; Erikson

and Palfrey 1998, 2000; Gerber 1998; Green and Krasno

1988, 1990; Jacobson 1978, 1980, 1985, 1990). Still,

scholars have struggled to find instruments for spending

that predict well and are derived from factors exogenous to

the vote (Jacobson 1990: 342).

Jacobson (1978, 1980, 1985, 1990) authored the semi-

nal research on spending effects in congressional elections.

After analyzing House and Senate races from the 1970s and

1980s, his conclusion is that while spending by challengers

has a large impact on incumbents’ share of the vote, incum-

bent spending has a much smaller effect. Jacobson’s

97

Reevaluating Spending in Gubernatorial Races:

Job Approval as a Baseline for Spending Effects

KEDRON BARDWELL, GRAND VALLEY STATE UNIVERSITY

Research on campaign spending has tried to resolve the issue of differential effects for incumbent and chal-

lenger spending. This analysis offers two new perspectives to the spending effects literature: (1) It extends the

scope of this research to include gubernatorial elections, and (2) it uses job approval ratings to control for gov-

ernors’ popularity. This approval measure sets a pre-campaign baseline for the expected vote that keeps cam-

paign variables from being credited with pre-existing levels of incumbent popularity. A two-stage least squares

(TSLS) analysis of two decades of gubernatorial races shows that incumbent spending (unlike challenger

spending) does not have a significant effect on the vote. If incumbent spending wins little new support in the

course of campaigns, generous public funding for statewide candidates will boost spending by challengers

without reducing challenger competitiveness via higher incumbent spending.

Political Research Quarterly,Vol. 58, No. 1 (March 2005): pp. 97-105

PRQ_March05_III 3/24/05 9:19 AM Page 97