Discretionary Military Action in Political Time

Abstract

This study investigates whether a president’s place in Stephen Skowronek’s theory of political time affects the likelihood that the president will use his power as commander-in-chief to take discretionary military actions abroad. The study presents a new approach to an age-old question: what makes the president more or less likely to decide to use military force abroad? The study’s theory section combines Skowronek’s framework and the larger political time the literature with insights from literature on diversionary theory, foreign policy decision-making, and executive-legislative relations. The political time literature argues that the president’s relationship with the dominant party regime of their day, and the strength of that regime, creates systematic incentives for presidents who serve during similar places in political time. The theoretical framework presented in this study argues that presidents who are unaffiliated with the dominant party regime of their day will be more likely than others to engage American armed forces in minor military conflicts. The study uses two datasets to test this argument and finds evidence that presidents who serve during the politics of preemption and the politics of reconstruction are more likely than other presidents to order the military into relatively small-scale, minor international conflicts. This finding builds on Skowronek’s argument that presidents who serve during the politics of articulation are the most likely to engage in major wars.

Notes

1 Nichols and Myers (2010) argue that McKinley and Teddy Roosevelt should be considered reconstructive presidents because, taken together, they institutionalized the “System of 1896” as the dominant partisan regime of their day. For the purposes of this article, we follow Skowronek’s classification scheme.

2 These sources include Blechman and Kaplan’s 1978 study Force Without War, Zelikow’s (1984) follow up study, studies conducted by the Defense Department, and data collected and aggregated by the Federation of American Scientists (see Blechman and Kaplan 1978; Meernik 2004; Zelikow 1984).

3 A president’s term was cut short during nine years in the dataset—1841 (Harrison), 1850 (Taylor), 1865 (Lincoln), 1881 (Garfield), 1901 (McKinley), 1923 (Harding), 1945 (Roosevelt), 1963 (Kennedy), and 1974 (Nixon). Fortunately, for seven of the nine years, the president’s successor fell into the same political structure. Only in 1865 and 1945 did this pattern not hold. Only one event occurred in 1865 during Johnson’s term. There were zero events during 1945.

4 The Incidents Per Year variable is a continuous count variable while the US Use of Force variable is a dichotomous variable. Accordingly, we use Poisson regression models to analyze the MID dataset and logit regression models to analyze the Meernik dataset.

5 Preemptive President equals 1 if a president serves during the politics of preemption and 0 otherwise. Reconstructive President equals 1 if a president serves during the politics of reconstruction and 0 otherwise. We use the values assigned by Skowronek to determine the political structure in which a president operated. We also analyzed models that used Unaffiliated President as an independent variable. Unaffiliated President is a dichotomous variable with a 1 indicating that a president is unaffiliated with the dominant partisan regime of the day (i.e. a preemptive or reconstructive president) and a 0 indicating that the president is affiliated with the dominant regime (i.e. a disjunctive or articulator president). These results are substantively similar. Appendix A presents these results. We present the results that use Reconstructive President and Preemptive President as independent variables in the main body of the paper because there are substantive and statistical differences between how reconstructive presidents and preemptive presidents decide to use military force abroad.

6 Disjunctive President equals 1 if a president serves during the politics of disjunctive and 0 otherwise.

7 Time equals 1 in 1818 and counts upward with each passing year. This variable controls for the fact that the US military has engaged in more international actions over time.

8 Lagged Incidents measures the number of conflicts in the previous year. We also analyzed this data using Poisson models with random effects and robust standard errors. The results are nearly identical so we decided to present the simpler model in our paper.

9 Post World War II President, that equals 1 if the president served after 1945 and 0 otherwise.

10 The p-value for Preemptive President is 1.68 so this coefficient is statistically significant at the p < 0.1 level in a one-tailed test.

11 At first glance, Figure 1 makes it appear that preemptive presidents are the least likely to take discretionary military action. The differences between preemptive presidents, disjunctive presidents, and articulators are not statistically significant.