FROM CONSCRIPTION TO VOLUNTEERS: BUDGET SHARES IN NATO DEFENCE SPENDING

Abstract

A growing number of NATO countries suspended compulsory military service during the past decade or are now phasing it out, moving to an All Volunteer Force (AVF). An AVF can free resources available for investment in up‐to‐date equipments, thus improving operational capabilities. Our paper investigates shifts in NATO military expenditure shares on personnel, equipment, infrastructure and other costs over the period 1970–2008 and explores the impact of the transition to AVFs on these shares of the defence budget. Results suggest that while the end of conscription did not reduce the share of spending on personnel, NATO forces are increasingly less reliant on soldiers and more on capital.

Keywords:

• Volunteer forces
• Defence budget
• Military personnel
• NATO

JEL Codes:

• C33
• H56
• J20

ACKNOWLEDGMENTS

We thank Ron Smith for his help and expert guidance throughout this research. We are grateful to two anonymous referees for their comments. We would like to thank Yunus Aksoy and the seminar participants at Birkbeck College and the 14th Economics and Security Conference, Izmir. The responsibility for any remaining errors or omissions is our own.

Notes

¹In the US, money for post‐service education (e.g., college, technical schools) helped to attract people with higher‐than‐average skills and cognitive aptitudes, who are likely to be more successful in the military (Williams, 2005).

²The Romanian Parliament, for example, voted to abolish conscription in October 2005, with the vote formalising one of many military modernisations that Romania agreed to when it joined NATO.

³Simon and Warner (2007), using US data over the 1988–2000 (a period of decreasing unemployment rate), find no disproportionate recruiting from minority groups.

⁴Nelson (1986) suggests that this increase was due to a rise in the number of first‐term enlistees who left the service before the end of their contract.

⁵One could use a logit transformation of the shares to ensure that predicted shares lay in the zero‐one interval, but then adding up would be lost.

⁶The dataset is part of the Defence Expenditure Studies – Data on NATO Defence Expenditures and Economic Development 1949–2001 by the Department of Security Policy and Strategy – Division of Defence Analysis, Defence Ministry of Sweden.

⁷SIPRI provides estimates of real defence expenditures in million USD in constant 2005 prices and constant exchange rate. These are made compatible with the earlier series expressed in constant 1995 prices. Using a constant exchange rate assumes that PPP holds. To transform the series in constant prices, SIPRI deflates the military expenditure expressed in local currency by a local price index. Local price indexes might be poor approximations of the evolution of military prices. However, there is no available index for military prices.

⁸The 2‐way FE model is superior to the Pooled OLS model and the 1‐way FE model. The 2‐way FE allows for more flexible estimates since it allows the constants to vary. However, it imposes the equality of slopes and variances. Heterogeneity in slopes might be an issue. If slopes are heterogeneous, the fixed effect model produces unbiased estimates of the average effects, provided that the differences in coefficients are independent of the regressors.

⁹According to the BIC, estimating two separate sub‐periods leads to the selection of the 2‐way FE model. AIC selects the 2‐way FE model for the share on infrastructure and other cost and the 1‐way FE model for the share on personnel and equipment. Since the differences are very small, we select a 2‐way FE specification and we estimate the relationship accounting for the break in 1989. The estimates are shown in Table VII.