The Shifting Determinants of Defense Spending Preferences Between 1980 and 2008

Abstract

This paper analyzes defense spending preferences using ordered logit regression analysis of American National Election Survey data from 1980 through 2008. Our results indicate that as opposed to having the ideology of isolationism, political party identification towards the Republican Party or having economic stakes in defense spending always play a significant role in increased preference towards defense spending. Demographic groups such as Native Americans, Hispanics, and retired women, a demographic subgroup, display generally positive preferences towards defense spending. Somewhat surprisingly, another demographic subgroup, ‘security moms,’ do not show a preference. Our analysis also displays lower (higher) preference in the early 1990s (2000s) for defense spending compared to the year 2008.

Keywords:

• Defense spending
• Public good

• JEL Codes
• H53
• H41

ACKNOWLEDGMENTS

The paper has benefited from the comments and suggestions of two anonymous referees on earlier versions. The usual disclaimer applies. Sencer Ecer and Nicholas J. Veasey thank former colleagues at Georgetown Public Policy Institute. Sencer Ecer’s research was supported by a Marie Curie International Reintegration Grant within the 7th European Community Framework Programme.

Notes

¹ Bartels, 1994; Higgs and Kilduff, 1993; Hartley and Russett 1992.

² It is important to note that we are analyzing defense spending preferences, which is only a factor that effects demand. An analysis of defense spending, on the other hand, would require analysis supply and demand factors. See Ali (2012).

³ Bartels, 485.

⁴ Mintz, 1992.

⁵ Russett , Hartley, and Murray. 1994.

⁶ Ladd, 2007.

⁷ Bartels, 1994.

⁸ ‘The “notworking” dummy variable includes homemakers and students, while the unemployed variable covers those who have been laid off that are seeking work as well as those who are not seeking work.’

⁹ The base race is white and OtherRace is all other than White and those that are explicitly specified in the model.

¹⁰ USA Today Editorial, ‘Candidates ignore ‘security moms,’ at their peril,’ 20 July 2004, ‘According to GOP pollster David Winston, ‘security moms’ now make up between 11 and 14% of the electorate. The trend has manifested itself in increased concealed-weapons-permit applications among women; the rise of national-security-focused Web logs published by hard-hitting female ‘war bloggers’; and an upsurge in political activism by women on core homeland-defense issues, such as immigration enforcement.’ http://usatoday30.usatoday.com/news/opinion/editorials/2004-07-20-malkin_x.htm See Also http://www.huffingtonpost.com/news/security-moms Huffington Post, accessed 13 June 2013

¹¹ p-values of statistical significance were generally much lower than 10%.

¹² Isolationist variable is not preset for the year 1982, so the regression analysis excludes data from that year.

¹³ Wong, 2007.

¹⁴ Klein, 2003.

¹⁵ Elder, 2007.