Democracy and Militarization in Developing Countries: A Panel Vector Autoregressive Analysis

ABSTRACT

This study examines the dynamic relationship between democracy and the military in more than 40 developing countries from 1990 to 2017. We investigate the dynamic interaction between democracy and military institutions using a panel vector autoregressive model and impulse response functions as well as variance decomposition analyses. We show that democracy plays a significant role in the substitution of nonmilitary expenditures for defense expenditures. We also investigate the response of democracy to positive shocks in military and nonmilitary expenditures. We find that the responses of political systems and different indexes of democracy including electoral, liberal, participatory, deliberative, and egalitarian democracy to positive shocks in military expenditures are negative and significant, whereas their responses to the shocks in nonmilitary expenditures are not significant. This result suggests that the political behavior of governments in developing countries is influenced more heavily by their spending on the military sector than by their spending on the nonmilitary sector.

KEYWORDS:

• Democracy
• Military expenditures
• Panel VAR model
• Developing countries

JEL CLASSIFICATION:

• D74
• E02
• F51
• Q34

Acknowledgments

This study is part of the Economic Sanctions and Conflict Resolution research project, which received a research scholarship from the Gerda Henkel Foundation under the Security, Society and the State special program. Sajjad F. Dizaji greatly appreciates the financial support of the Gerda Henkel Foundation. We appreciate the helpful comments of Khusrav Gaibulloev (Editor) and two anonymous reviewers. We also thank Jhoana Ocampo for research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethical approval

This article does not contain any studies with human participants or animals performed by the authors.

Notes

1. The relationship between corruption and military spending in the members of the Gulf Cooperation Council and other countries in the Middle East and North Africa region is also investigated by Farzanegan (2018).

2. www.imf.org/external/pubs/ft/weo/2019/02/weodata/index.aspx.

3. Armed forces personnel (as a percentage of the total labor force) is stationary with respect to Im, Pesaran and Shin, ADF-Fisher, and PP-Fisher tests.

4. To test for the presence of cross-sectional dependence in our data, we implemented the simple test of Pesaran (2004) and computed the cross-section Dependence (CD) statistic. This test is based on the average of pair-wise correlation coefficients of the ordinary-least-squares (OLS) residuals obtained from standard augmented Dickey-Fuller (ADF) regressions for each individual. Its null hypothesis is cross-sectional independence and is asymptotically distributed as a two-tailed standard normal distribution. The results (available upon request) indicate that the null hypothesis of cross-sectional independence for our variables is rejected.

5. The exception is for militpolit, which our statistical software does not provide evidence regarding its cross-section independence and stationarity. However, we have estimated our PVAR model and related impulse response functions with and without this variable. The main results do not change.

6. In general, the response of variable j to a unit shock (forecast error) in variable k is depicted graphically by IRFs to provide a visual impression of the dynamic interrelationships within the system. As economic variables and political indexes have different scales, we have considered innovations of one standard deviation rather than unit shocks in this study (Lütkepohl 2005). In this case, IRFs choose the units at the vertical axes equal to the standard deviations of the residuals corresponding to the variables whose effects are considered. Such a rescaling offers a better picture of the dynamic relationships because the average size of the innovations occurring in a system depends on their standard deviation.

7. We have also estimated our PVAR model by applying two lags (according to HQ) and three lags (according to FPE and AIC). The overall results do not change. These results are available upon request.

8. These results are available upon request.

9. This is available upon request.

10. Instead of using the Polity2 index of democratization in our panel VAR model, we also apply varieties of democracy indexes (i.e. electoral, liberal, deliberative, participatory, egalitarian democracy indexes) in separate panel VAR models one by one. The overall results confirm our earlier estimations, indicating that increased military expenditures have a negative impact on components of democracy in developing countries.

11. We also examine the generalized impulse responses of the electoral, liberal, deliberative, egalitarian, and participatory indexes to positive shocks in per capita military expenditures. The results reveal negative and statistically significant responses of the democracy indexes to positive shocks in per capita military expenditures. These results are available upon request.

12. We also apply other components of democracy (i.e. liberal, deliberative, egalitarian, participatory indexes) in separate panel VAR models. The results confirm our previous findings, showing that non-military spending (% of GDP) has a negative and significant impact on military spending (% of GDP) and a positive and statistically significant impact on government revenues (% of GDP). However, the responses of the democracy indexes to the shock in non-military expenditures (% of GDP) are not significant. These results are available upon request.

13. The possibility of a U-shaped relationship between democracy and public spending has been noted in some studies (Plümper and Martin 2003; Hausken, Martin, and Plümper 2004; Aidt, Dutta, and Dauntan 2010).

Additional information

Funding

This work was supported by the Gerda Henkel Foundation [AZ 05/KF/18].