Public Choice (2016) 168:55–74 DOI 10.1007/s11127-016-0349-0

Explaining the worldwide decline in the length of mandatory military service, 1970–2010 Danko Tarabar1 • Joshua C. Hall1

Received: 17 February 2016 / Accepted: 16 June 2016 / Published online: 25 June 2016 Ó Springer Science+Business Media New York 2016

Abstract Compulsory military service has declined considerably since 1970. This explained by changes on both the extensive and intensive margins by governments. While the decision to use conscription for military purposes has been studied extensively, in this paper we examine empirically the factors underlying the decline in the duration of military service obligations. Employing data from the Economic Freedom of the World index observed at 5-year intervals from 1970 to 2010, we find that the probability of a shorter military service time is positively associated with smaller country populations, smaller lagged army sizes, increases in primary schooling among young males, and having common law legal origins. Our empirical approach also highlights how the elderly population exhibits a nonlinear relationship with the length of conscriptees’ time in uniform. Keywords Conscription  Economic freedom  Ordered probit  All volunteer force JEL Classification D72  H56

1 Introduction For much of modern history, forced enlistment (conscription) has been used widely to meet national demands for military personnel. Conscription is an involuntary removal of individuals from civilian society by compelling young men and sometimes women to become a part of the active duty armed forces for a specified duration. Countries that do not utilize conscription for military purposes generally rely on an all-volunteer force (AVF).

& Danko Tarabar [email protected] Joshua C. Hall [email protected] 1

College of Business and Economics, West Virginia University, Morgantown, WV 26506-6025, USA

123

56

Public Choice (2016) 168:55–74

Over the past 45 years, a large number of countries have reduced the lengths of their compulsory military service obligations or moved entirely to an AVF. For example, the United States used conscription combined with voluntary enlistment until 1973, when then-President Richard Nixon ended the draft. The Economic Freedom of the World (EFW) annual report by Gwartney et al. (2015) provides a comprehensive measure of the extent of military conscription for a large number of countries going back to 1970.1 As an indicator of duration, the EFW index assigns each observed country a rating score that proxies for the length of compulsory military service. The EFW rating, however, is not expressed in terms of actual service obligations such as months or years. For example, a rating of 10 means that a country does not rely on conscription at all. A country in which compulsory service is in effect has a score less than 10, depending on the length of the term. The longer the duration of compulsory military service, the lower the score. Countries with terms of less than six months for draftees are assigned a rating of 5, while those with between six and 12 months are assigned a rating of 3. Military service obligations of between 12 and 18 months gets a rating of 1 and longer periods are given a rating of zero.2 The EFW index thus effectively captures the gradual relaxation of conscription terms for a broad cross-section of countries. Table 1 depicts the increase in the average EFW conscription rating from 3.0 in 1970 to 6.3 in 2010, indicating a significant reduction in the duration and use of conscription over this period. According to Mulligan and Shleifer (2005), virtually all countries today employ some professional military personnel. In countries that switched from conscriptmajority forces with lengthy service tours to AVFs, the change often was gradual, marked by sequential shortenings of military service obligations, armed forces’ downsizing, and greater reliance on volunteers. These downsizing and restructuring trends were particularly prominent among many of the former Eastern Bloc countries during their NATO integrations (Jehn and Selden 2002; Bove and Cavatorta 2012). For example, Poland (a NATO member since 1999) had a compulsory military service term of 2 years during the Cold War and still required service of 12 months as late as 2005. In that year, conscriptees’ service obligations were reduced to 9 months until the draft ultimately was phased-out in 2009. The decline in the reliance on conscription has been of great interest to economists given the role that they have played in the shift to an AVF in the United States (Henderson 2005). Henderson (2010) discusses the historical use of conscription in the United States, including the important influence of Milton Friedman. Beginning in the late 1960s, economists began to study more closely the economic costs of the draft and the benefits of an AVF (Oi 1967; Hansen and Weisbrod 1967; Friedman 1967; Fisher 1969; Borcherding 1971). This early literature generally concluded that an AVF has a lower social cost than a conscripted force (Lee and McKenzie 1992). More recent theoretical analyses, which take into account the deadweight cost of taxation necessary to fund a volunteer army, however, suggest that under certain circumstances social costs might be higher for an AVF (Warner and Asch 2001). Lau et al. (2004) show large dynamic costs to military conscription in

1

The EFW index has been used widely in the development and growth literature. See, for example, papers by Stroup (2007), Bergh and Nilsson (2010), and Gehring (2013).

2

While the conscription duration declines as the EFW score rises, there are two possible exceptions that are rarely encountered. First, if conscription is technically mandated but unenforced, or if the length of military service cannot be determined, the country receives a rating of 3. Second, if mandated military service of any length can be substituted for by compulsory civilian duty, the rating of 5 is given.

123

Public Choice (2016) 168:55–74 Table 1 The decline in conscription, 1970–2010

Source (Gwartney et al. 2015) and authors’ calculations. Number of countries for which data is available varies by year. Higher numbers correspond to reduced use (shorter periods or elimination) of conscription for military purposes

57

Year

Average rating

Median rating

1970

3.0

1

1975

5.3

3

1980

5.4

3

1985

4.7

3

1990

4.8

3

1995

5.2

3

2000

5.8

3

2005

5.8

5

2010

6.3

10

terms of human capital formation. For a general overview of the normative economic arguments related to the military draft, see Poutvaara and Wagener (2007a).3 In this paper we do not address the normative economic arguments regarding military conscription. Instead, our aim is to investigate empirically the demographic and macroeconomic forces underlying the decline in length of compulsory military service around the world. As early as 1970, Tollison (1970) pointed out that economists focused too much on the allocative and distributive effects of the military draft and ignored the political economy of conscription. While a number of papers in recent years have focused on the demographic, economic, and political factors underlying the use of conscription (Poutvaara and Wagener 2007b; Ng et al. 2008; Berck and Lipow 2011), to our knowledge only four extant empirical papers analyze the determinants of the reliance on the draft to fill military manpower requirements (Ross 1994; Mulligan and Shleifer 2005; Adam 2012; Asal et al. 2015). We build on this literature in two ways. First, the empirical literature has focused primarily on the extensive margin: the use of conscription.4 We focus our attention on the intensive margin: the duration of draftees’ military service obligations. Since the abolishment of conscription in many countries was not a one-time event but rather proceeded gradually, looking at the intensive margin gives greater insight into the factors underlying this change. Second, we improve on existing empirical studies by introducing new explanatory variables and developing new hypotheses about the decline in the duration of compulsory military service. Notably, we control for young males’ education levels and account for possible nonlinearities in the effects of the elderly population share. Our results reveal a richer dynamic of conscription than modeled in previous empirical literature. To preview one result, we find that the share of population 65 years of age and older exhibits an inverse-U relationship with the length of conscripted service. Similar to the previous literature, we find that larger shares of elderly in the overall population generally are associated with ‘‘more conscription’’ (that is, longer tours of duty for draftees in our case). However, for countries wherein the elderly comprise more significant fractions of the population, marginal increases in that population share are associated with shorter 3

Economists have also looked at the effect of conscription on voting (Cebula and Mixon Jr 2012), health and welfare (Angrist et al. 2011), attainment and wages (Hubers and Webbink 2015), higher education (Keller et al. 2010), and economic growth (Keller et al. 2009).

4

This is primarily done through a simple indicator variable equaling 1 if the country has military conscription as law and zero otherwise, or the share of military personnel that are conscripts.

123

58

Public Choice (2016) 168:55–74

conscripted terms of service, on average. We also find that declines in the share of males 15–24 with primary school educations or less are associated with a shortening of draftees’ military service obligations. Our paper proceeds as follows. In the next section we discuss the related literature and our dataset. Section 3 outlines the empirical framework, while Sect. 4 reports the results. In Sect. 5 we offer concluding remarks.

2 Dataset and related literature The positive theoretical and empirical literature on the use of conscription for military purposes falls squarely in the public choice tradition. The use of conscription, as well as the length of draftees’ military service obligations, are in effect a government policy that acts like a tax on those individuals who are conscripted. As such, there is an underlying supply and demand curve for conscription based on the costs and benefits to different groups of individuals within a country. For example, the elderly usually are thought to be in favor of conscription because its economic costs fall largely on the young. Poutvaara and Wagener (2007b, 2011) provide a detailed overview of both the normative and political economy view of conscription. For any given military size, the choice is between a force largely comprised of conscripts paid below-market wages or volunteers paid market wages, both financed by general tax revenues.5 From a normative perspective, the choice between conscripts and a professional army can be thought as solving an optimal tax problem. An AVF introduces distortions because of the higher taxes necessary to fund market wages for volunteers. Conscription introduces deadweight losses because taking steps to avoid the draft results in inefficient occupational or schooling choices (Card and Lemieux 2001; Lokshin and Yemtsov 2008). From a normative perspective, the solution should be to minimize deadweight loss and it is generally thought that an AVF is superior to conscription in this regard. From a public choice perspective, democratic societies should overwhelmingly opt for conscription as the incidence of the ‘‘draft tax’’ disproportionately falls on young males while an AVF implies higher taxes on the entire population. As pointed out by Oi (1967), if voters were non-altruistic, under majority rule conscription should always dominate an AVF of a similar size. In practice, however, this is not the case, which has led to a large empirical literature trying to explain the determinants of conscription. Mulligan and Shleifer (2005) analyze conscription from the viewpoint of startup and enforcement costs. They argue that conscription involves a large fixed administrative cost. Holding the size of the armed forces constant the average fixed cost of conscription falls as population increases. More populous countries are therefore able to spread out the cost of recruitment offices, enforcers, and so on, over a larger tax-paying population. The cost of additional conscription regulation in their model is proxied by national legal origins. For instance, the United Kingdom’s common law system implies less government regulatory capacity for meeting social goals. A small government administrative apparatus thus has higher marginal costs of creating and enforcing conscription regulation. Following from their work, we enter as explanatory variables both population size in log form (LogPopulation) and legal origins (UKOrigins, SocialistOrigins).

5

We say ‘‘largely’’ comprised of conscripts, since even heavily conscripted armies need professional officers to train and lead.

123

Public Choice (2016) 168:55–74

59

Ross (1994) highlights the importance of controlling for the excess burden of taxation on the cost side of the political calculus. He did so by including a control for ‘‘central government share,’’ the ratio of total central government expenditures to GNP. The intuition is that more government spending implies larger marginal deadweight losses from taxation, holding incomes constant. Therefore countries with large government sectors may find that the deadweight cost of moving to an AVF is too high.6 In our empirical analysis we employ a similar measure of government spending as a share of GDP (GovConsumption). The World Bank defines government consumption expenditures as ‘‘outlays on goods and services, including employee compensation, and most expenditures on national defense and security.’’ Wealthier countries have more resources with which to finance an AVF. Ross (1994) points out, however, that higher incomes also suggest that the cost of paying market wages to an AVF would be higher as well. Thus, the predicted sign on a variable like GDP per capita is ambiguous theoretically and ultimately an empirical question.7 An increase in real income increases the cost of employing volunteers, other things being equal, creating an incentive on the margin for conscription or to lengthen draftees’ terms of service. For countries already employing conscription, longer military service obligations are a cost-saving measure, holding army size constant, as administrative and training costs are reduced.8 In democracies, public opinion carries significant weight in influencing policy. Poutvaara and Wagener (2011) point out that in a democracy citizens may prefer conscription for the purpose of improving social cohesion and equity (Poutvaara and Wagener 2011). The empirical literature, however, has found mixed results on the impact of democracy on conscription (Mulligan and Shleifer 2005; Adam 2012). For example, Adam (2012) finds democracy levels not to be associated with the likelihood of using conscription. He does find, however, that the number of years since a country has become democratic is negatively associated with the use of conscription. Mulligan and Shleifer (2005) do not find that democracy influences conscription at all. More recently, however, Asal et al. (2015) examine the long-term use of conscription in more than 100 countries over 200 years and find that democracies are less likely to rely on the military draft. To examine the effect of democracy, we use the well-known data set on political regime characteristics from the Polity IV Project (Marshall et al. 2013).9 The Polity IV variable ranges from -10 (strong autocracy) to ?10 (strong democracy). Although we employ the democracy variable in its levels, we also lend further context to both income and democracy it by considering their mutual interplay. Using interaction terms, we explore whether citizens exert more influence on conscription in democratic countries’ as income rises. In doing so, we are building on Ross (1994), who notes that if ‘‘freedom from compulsion’’ is a normal good, governments in higher-income countries would still experience political pressure to reduce draftees’ lengths of service or eliminate conscription altogether even in the face of the larger accounting costs of employing volunteers. Our expectation is that the interaction term will be 6

Bohanon et al. (2014) survey the literature and find that the excess burden of raising an additional dollar of revenue from a tax on labor is between $1.43 and $1.50.

7

See Ross (1994) for a longer discussion of the possible mechanisms through which GDP could affect the choice between conscription and an AVF.

8

We hold military size constant in our regressions.

9

Cited thousands of times, the data has been used in a number of papers in comparative political economy such as Coyne (2008), Leeson et al. (2012), Dutta et al. (2013), and Coyne and Williamson (2015). Democracy, in the Polity IV reports, has three elements: institutions through which citizens can express preferences, institutionalized constraints on the executive, and the guarantee of civil liberties for all citizens in both politics and their daily lives.

123

60

Public Choice (2016) 168:55–74

positive, suggesting a mutually reinforcing effect: higher democracy levels work to reduce the use of conscription as incomes rise, and that higher incomes promote shorter compulsory terms of service where democracy is more advanced. In the empirical literature on the determinants of conscription demographics play a significant role. In both Mulligan and Shleifer (2005) and Adam (2012), the share of the population over the age of 65 is shown to be positively associated with the likelihood of using conscription. Adam (2012) notes that the elderly are doubly burdened when conscription is abolished because they had to serve in the military previously and now pay higher taxes to finance an AVF. As a result, the elderly are expected to exert political pressure to maintain the status quo with respect to military conscription. Unlike the earlier literature, however, we see this relationship as being less straightforward. Holding population size constant, a larger over-65 population share (Age65) in all likelihood means a shrinking of the labor force. Similarly, longer compulsory terms of service reduce and distort the labor supply of younger workers who help finance health care and pension benefits for a growing elderly population (Poutvaara and Wagener 2011). The rational elderly voter might then find it in his or her own interest to support the relaxation of conscription requirements. This observation suggests that the share of the population over 65 is related nonlinearly to draftees’ service obligations. In order to account for these nonlinearities we also enter Age652 as an explanatory variable. To the elderly variable, we add a new demographic variable, the percentage of lesser educated young males (UnskilledYouth). Calculated from Barro and Lee (2013), the variable is the percentage of males 15–24 who have at most primary schooling as their highest educational attainment. Our prior is that UnskilledYouth will be associated with longer mandatory service for two reasons.10 First, the social costs associated with the misallocation of labor into the military under conscription are less when this group is large (Mulligan and Shleifer 2005). Second, the employment prospects of this group in the formal sector are poor and thus conscription might be a way mitigate against social unrest caused by widespread unemployment or underemployment.11 Lastly, we control for armed conflict (War) and size of the armed forces (ArmySize) as is standard practice in the empirical literature. Both variables are associated with the demand for conscripts associated with the likelihood of armed conflict across countries. Army size is calculated as the ratio of armed forces to the total male population aged 15–24, while armed conflict is captured by major episodes of political violence from Marshall (2015). Major episodes of political violence are coded on a scale 0–10 capturing both the existence and intensity of inter- and intra-state conflicts resulting in a minimum of 500 deaths in a given year. We calculate five-year moving averages of political violence to allow for both the past and anticipated incidences of armed conflict to affect the length of conscripted service.12 Since conscription policies may feed back into decisions on military size, we lag ArmySize.13 10 Keller et al. (2010) find that conscription discourages enrollment in tertiary education in a sample of OECD countries. This might lead to concern that conscription is influencing UnskilledYouth, not the opposite. We believe that reverse causality is not an issue here since this variable is individuals who have only the primary education or less. It is highly unlikely that military conscription is influencing the decision to complete primary school. 11

On the relationship between unemployment and civil unrest see Sayre (2009) and Oyefusi (2010).

12

This variable takes into account the major episodes of political violence directly present within a country that resulted in a minimum of 500 deaths. Seven categories of conflict are recognized: international violence, international war, international independence war, civil violence, civil war, ethnic violence, and ethnic war. 13 While draftees are inducted into different branches of service in different countries around the world, the army is the most likely branch to rely on conscription to fill its ranks.

123

Public Choice (2016) 168:55–74

61

Table 2 reports summary statistics and Table 7 in the appendix gives the sources for the variables included in our analysis. Appendix Table 8 lists the countries observed in our regressions along with the years in which they are observed.

3 Empirical methodology Our measure of conscription comes from the Economic Freedom of the World (EFW) annual report by Gwartney et al. (2015). Conscription is one of the 42 distinct variables that comprise the EFW index. The overall EFW index broadly measures countries’ commitment to ‘‘personal choice, voluntary exchange, freedom to compete, security of privately owned property, and the freedom to enter and compete in markets’’ (Gwartney et al. 2015).14 The draft, in effect, represents forced labor through involuntary removal from the civilian labor force and as such represents a violation of free choice and economic freedom. As noted earlier, the EFW data set does not report the actual length of service obligation in months or years. Rather, the EFW’s authors cluster countries similar lengths of draftees’ service obligations and assign them interval rating scores based on a specific formula using information from The Military Balance by International Institute for Strategic Studies (2013) and World Survey of Conscription and Conscientious Objection to Military Service by War Resisters International (2005). Conscription ratings with corresponding intervals of compulsory military service length are summarized in Table 3. The EFW report, while having annual data for more than 150 countries since 2000, only has data observed at fiveyear intervals for a smaller number of countries from 1970 to 2010. Given that conscription policy and demographics change slowly, we feel that looking at five-year intervals from 1970 to 2010 is most appropriate. We have a maximum of 801 total observations in up to 127 countries.15 Table 2 Summary statistics, 1970–2010

Variable Conscription

SD

Min

Max 10

5.3

4.4

0

GovConsumption (%)

16.0

6.2

3.7

64.4

LogPopulation

16.3

1.5

12.8

21.0

LagArmySize (%)

7.4

8.0

.1

65.1

LogRGDPPC

8.1

1.6

4.3

11.3

PolityIV

3.2

7.0

-10

10

SocialistOrigins

.12

.33

0

1

UKOrigins

.31

.46

0

1

UnskilledYouth See Table 6 for data sources. N = 801, from Table 4 specification (1)

Mean

War Age65 (%)

41.7 .71 7.3

26.7

.16

95.8

1.6

0

11.8

4.9

.7

22.9

14 The EFW has been used as an explanatory variable in hundreds of scholarly studies in economics, finance, political science, and many other fields (Hall and Lawson 2014). 15 These countries comprise most of the world’s population. A full list of countries included in our analysis can be found in Table A2. The EFW data is only observed every five years from 1970 to 2000.

123

62

Public Choice (2016) 168:55–74

Table 3 EFW conscription rating description Conscription duration

Rating

No conscription

10

\6 months

5

6–12 months

3

12–18 months

1

[18 months

0

Source Gwartney et al. (2015). If conscription is mandated but not enforced, countries receive a 3. If compulsory civilian duty can be substituted, a rating of 5 is given

Since the EFW conscription index assigns ranked categorical values 10, 5, 3, 1, 0 to different levels of military conscription this calls for the use of an ordered probabilistic estimator in econometric modeling.16 We hence estimate the following baseline panel random effects ordered probabilistic model and add more controls in additional specifications17: Conscriptionit ¼ b1 GovConsumption þ b2 LogPopulation þ b3 LagArmySize þ b4 LogRGDPPC þ b5 UnskilledYouth þ b6 UKOrigins þ b7 SocialistOrigins þ b8 PolityIV þ b9 War þ b10 Age65 þ mi þ uit ; where Conscriptionit is the latent unobservable variable indicating the country i’s ‘‘true’’ underlying propensity to use conscription of a certain length at time t, mi is the random effect, and uit is the error term. The latent dependent variable takes an observable value only when it crosses a certain threshold T estimated along with other parameter coefficients. In general, the dependent variable takes the value Conscriptionit ¼ j if T1 \Conscriptionit  T2 , and the probability of this is given by: PðConscriptionit ¼ jÞ ¼ PðT1 \Conscriptionit  T2 Þ 0

0

0

0

¼ PðT1  X b\u  T2  X bÞ ¼ FðT2  X bÞ  FðT1  X bÞ; where F is the cdf of a logistic distribution FðzÞ ¼ ez =ð1 þ ez Þ if the error term u is distributed logistically, and the cdf of a standard normal distribution U if the error is normally distributed. We use both multinomial ordered logit and probit estimators in our regression analysis as a robustness check. The coefficients b can be immediately interpreted only for the direction and significance of the impact on the latent variable, but not for magnitudes. Since the observed dependent variable takes j different alternative values, there will be j different sets of marginal effects. The marginal effect of the kth regressor on the probability that the dependent variable takes the value Conscription ¼ 5 is therefore given by:

16

Becker and Kennedy (1992) provide a survey of the ordered probit regression.

17

We cannot employ country fixed effects as the country fixed effects estimator in non-linear probabilistic models is not consistent.

123

Age652

Age65

War

UnskilledYouth

UKOrigins

SocialistOrigins

PolityIV  LogRGDPPC

PolityIV

LogRGDPPC

LagArmySize

LogPopulation

GovConsumption

Dep. Var.: Conscriptionit (4)

(.007)

(.180)

(.007)

(.076)

-.344*

(.083)

-.143*

(.005)

-.012**

(.756)

1.612**

(.590)

.023***

(.179)

(.077)

.317

(.087)

-.099

(.005)

-.011*

(.903)

1.843**

(.530)

-.266

.019***

-.167

(.088)

.373***

(.091)

(.005)

-.166*

-.130

(.006)

(.727)

-.016***

(.815)

-.015**

3.697***

(.757)

(.761)

3.995***

-1.994***

-3.000***

(.016)

(.122)

-.179

(.256)

-.117

(.021)

-.047**

(.185)

.093

(.015) -1.416***

(.020)

.010

(.272)

-.285

(.023)

-.054**

(.197)

-.074

-.040 (.027)

.027*

(.118)

-.041 (.028)

.024

(.020)

(.276)

-.157

.012

(.287)

(.022)

-.399

-.691**

(.024)

(.195)

-.045**

(.206)

-.050**

-.348*

-.531***

-.044

(.028)

-.044**

(.028)

-.090*

(.140)

.721***

(.167)

-.370**

(.010)

-.026**

(1.623)

7.713***

(1.514)

-5.787***

(.034)

.036

(.538)

-1.335**

(.041)

-.094**

(.373)

-1.069***

(.054)

(.013)

.033**

(.326)

-.230

(.166)

-.425**

(.010)

-.027***

(1.359)

7.100***

(1.417)

-3.967***

(.027)

.037

(.203)

-.226

(.501)

-.925*

(.037)

-.088**

(.335)

-.692**

(.055)

-.088

(6)

(5)

(3)

(1)

(2)

Random Effects Ordered Logit

Random Effects Ordered Probit

Table 4 Determinants of conscription term lengths

(.139)

.624***

(.167)

-.301*

(.010)

-.018*

(1.770)

3.663**

(.970)

-2.770***

(.035)

.029

(.493)

-.834*

(.041)

-.101**

(.366)

-.166

(.055)

-.081

(7)

(.013)

.042***

(.329)

-.548*

(.166)

-.367***

(.009)

-.019**

(1.486)

3.235**

(1.058)

-.652

(.028)

.045

(.213)

-.287

(.473)

-.302

(.035)

-.091***

(.339)

.181

(.055)

-.078

(8)

Public Choice (2016) 168:55–74 63

123

123 (4)

127

Full

-670.10

63.67***

512.10***

Sample

Log Pseudolikelihood

Wald v2

LR test

497.80***

140.82***

-661.98

Full

127

801

263.87***

40.99***

-619.12

Consc.

86

559

243.90***

56.27***

-606.07

Consc.

86

559

801

524.51***

63.95***

-654.17

Full

127

801

516.31***

44.55***

-604.95

Full

127

282.45***

65.83***

-687.31

Consc.

86

559

(7)

268.32***

54.97***

-592.50

Consc.

86

559

(8)

* p \ 0:10; ** p \ 0:05; *** p \ 0:01

Cluster-robust standard errors are in parentheses. With the exception of UnskilledYouth, the significance of estimated coefficients are not sensitive to the introduction of period dummies. A significant LR (likelihood-ratio) test statistic suggests random effects model appropriateness over standard pooled ordered model alternatives. Specifications (1)– (4) estimated using ordered probit, and (5)–(8) using order logit random effects estimators. Full sample and countries that used conscription at least once

801

Countries

(6)

(5)

(3)

(1)

(2)

Random Effects Ordered Logit

Random Effects Ordered Probit

No. of observations

Dep. Var.: Conscriptionit

Table 4 continued

64 Public Choice (2016) 168:55–74

Public Choice (2016) 168:55–74

65

i oP h 0 0 0 0 ¼ F ðTConscription¼3  X bÞ  F ðTConscription¼5  X bÞ bk : oxk In this particular case, the marginal effect bk may be interpreted as the change in probability of a country having compulsory service of less than six months for a unit increase in the independent variable xk . The ordered conscription variable offers several advantages over the standard dichotomous approach. By discriminating between different lengths of draftees’ terms of service, we do not simply estimate the likelihood of a country having conscription. Instead we are able to account for the relaxation of conscription regulations by reducing the lengths of draftees’ service obligations for a broad cross-section of countries. Different lengths of compulsory military service allow us to gauge the determinants of the choices of different quantities of conscription in response to changes in demographics, income, and external threats.

4 Results Table 4 reports the empirical results from random effects ordered probit and logit regressions between the latent EFW measure of conscription duration and a set of hypothesized determinants in a panel of up to 127 countries observed at five-year intervals between 1970 and 2010. Standard errors are clustered at the country level in all specifications. In the interest of parsimony, we do not report the threshold parameter estimates. The results from the likelihood-ratio (LR) test favor the use of the random effects model as opposed to the standard pooled ordered models. We look at two samples in Table 4: the full sample for which we have data [specifications (1)–(2) and (5)–(6)], and the sample in which we include only countries that used conscription in at least one of the observed years [specifications (3)–(4) and (7)–(8)].18 The results from Table 4 are generally in line with our expectations. When analyzing the results it is important to note that the dependent variable is ordered so that higher values represent shorter compulsory military service obligations. A positive sign on an independent variable thus indicates reduced reliance on conscription (either shorter terms of service or none at al, i.e., an AVF). For example, we find a negative and statistically significant coefficient on LagArmySize across all specifications. This means that as the number of individuals in the military relative to the 15–24 old male population increases, a country is less likely to shorten draftees’ terms of service—in the limit to zero with an AVF. Similarly, we find negative and statistically significant results for LogPopulation in the specifications that have all countries, but not when our analysis is limited to countries that had conscription at any point from 1970 onward. Democracy, as measured by PolityIV, has no consistent relationship with the duration of military conscription and the interaction between democracy and GDP per capita is only significant in one specification. Countries with UK legal origins are more likely to have reduced draftees’ military obligations over this period across all samples, while those with socialist origins were less likely to reduce service time, other things being equal. Finally, in all specifications we find reductions in the proportion of 15–24 year-olds with less than a primary education (UnskilledYouth) to be positively 18 We also re-estimated these regressions using a dummy variable for NATO membership and period effects and obtained similar results.

123

66

Public Choice (2016) 168:55–74

100

Probability (percents)

Fig. 1 Probability density of different conscription terms given shares of Age65

< 6 months 6-12 months 12-18 months > 18 months

80 60 40 20 0

0

2

4

6

8 10 12 14 16 18 20

Share of 65+ (percents)

related to declines in compulsory military service. As this is a new addition to the empirical literature our results add insight beyond our focus on draftees’ mandatory times in uniform. Our empirical results also suggest nonlinearities with respect to the elderly population. In specifications that include only Age65 we find a positive and often statistically significant relationship with the duration of conscripted military service. Countries are more likely to reduce the lengths of compulsory military obligations as the population percentage of citizens over age 65 increases. While this finding is largely at odds with the literature, such as Adam (2012), our positive results are robust across specifications.19 In unreported regressions, we use related proxies for age structure, such as the dependency ratio and population median age, and find similar results.20 Allowing for nonlinearities, however, we observe that Age65 has the expected sign based on the previous literature, while the squared term Age652 is positive and consistently statistically significant. This suggests a parabolic relationship such that when the elderly population share is small, marginal increases in that share lead to longer conscription terms, but that relationship reverses when the elderly population share is larger.21 These findings help explain why developed countries with rising life expectancies and large elderly population shares generally have reduced conscription lengths or eliminated them entirely. Fig. 1 displays the probability density of having different conscription lengths given different shares of population over 65, while holding all other variables constant at their respective means.22 Although the probability of having the longest conscription length ([18 months) unambiguously falls with elderly population share, this is not the case for the other three conscription lengths. Higher GDP per capita is associated with longer compulsory military service requirements for the average observed country in the full sample. This finding is in line with the view that accounting costs are more relevant than total economic costs in the choice between conscription and an AVF. To explore the linkage between income and conscription lengths further, we present results in Table 5 from two regression subsamples 19

It should be noted, however, that neither Adam (2012) nor Mulligan and Shleifer (2005) obtain consistent signs for this variable’s coefficient across different specifications.

20

Unreported regressions available upon request.

21

Looking at specification 4 from Table 4, the elderly population share is associated with longer compulsory military obligations until it reaches 7.5 %, at which point a larger elderly population share reduces conscription length.

22

Plot constructed from Table 4 specification (3), using a sample of 87 countries that used conscription at least once in the observed period.

123

(4)

559

94

Countries

94

559 33

33

242

(.013)

(.365)

-.812**

(.270)

-.033

(.006)

-.008

(1.042)

4.372***

(.878)

1.687*

(.068)

.061

(.632)

1.127*

(.070)

-.195***

(.279)

.085

(.008) 242

(.215)

.496**

(.342)

.153

(.007)

-.014**

(1.894)

4.860***

(.964)

.138

(.073)

.013

(.655)

.026

(.086)

-.179**

(.348)

.148

(.032)

.002

.042***

(.193)

(.096)

-.000 (.047)

.020**

-.091

.305

(.088)

(.087)

(.007)

-.133

-.112

(.008)

(.826)

-.020***

-.015*

(.863)

(.904)

2.975***

3.135***

(1.007)

(.018)

-3.419***

-3.493***

(.018)

(.326)

.016

.015

(.358)

(.022)

-.545*

-.562

(.022)

(.213)

-.017

(.220)

-.022

-.428**

-.546**

-.046*

(.028)

-.045

(.028)

94

559

(.168)

.597***

(.163)

-.316*

(.015)

-.026*

(1.603)

6.005***

(1.881)

-6.708***

(.030)

.034

(.672)

-1.129*

(.040)

-.042

(.397)

-1.079***

(.051)

-.092*

94

559

(.014)

.033**

(.325)

-.080

(.164)

-.355**

(.014)

-.033**

(1.512)

5.701***

(1.664)

-6.487***

(.030)

.033

(.605)

-1.093*

(.038)

-.034

(.374)

-.837**

(.051)

-.092*

(6)

(5)

(3)

(1)

(2)

Random effects ordered logit

Random effects ordered probit

No. of observations

Age652

Age65

War

UnskilledYouth

UKOrigins

SocialistOrigins

PolityIV

LogRGDPPC

LagArmySize

LogPopulation

GovConsumption

Dep. Var.: Conscriptionit

Table 5 Determinants of conscription term lengths

33

242

(.443)

.798*

(.757)

.098

(.012)

-.020

(3.282)

8.482***

(1.579)

.550

(.157)

.063

(1.143)

.221

(.202)

-.348*

(.564)

.282

(.086)

-.027

(7)

33

242

(.027)

.076***

(.699)

-1.500**

(.560)

-.236

(.011)

-.012

(1.764)

8.019***

(1.642)

3.201*

(.139)

.147

(1.164)

2.087*

(.139)

-.376***

(.480)

.188

(.052)

-.009

(8)

Public Choice (2016) 168:55–74 67

123

123 (4)

-450.45

41.02***

359.78***

Wald v2

LR test

334.76***

60.89***

-447.84

Non-OECD

OECD

27.12***

92.91***

-180.69

OECD

30.69***

89.32***

-172.17

Non-OECD

370.36***

42.72***

-441.21

Non-OECD

361.07***

61.13***

-438.65 31.54***

89.44***

-180.00

OECD

(7)

34.42***

78.81***

-171.01

OECD

(8)

* p \ 0:10; ** p \ 0:05; *** p \ 0:01

Notes: cluster-robust standard errors are in parentheses. With the exception of UnskilledYouth, the significance of estimated coefficients are not sensitive to the introduction of period dummies. A significant LR (likelihood-ratio) test statistic suggests random effects model appropriateness over standard pooled ordered model alternatives. Specifications (1)–(4) estimated using ordered probit, and (5)-(8) using order logit random effects estimators. OECD and Non-OECD countries

Non-OECD

Log Pseudolikelihood

(6)

(5)

(3)

(1)

(2)

Random effects ordered logit

Random effects ordered probit

Sample

Dep. Var.: Conscriptionit

Table 5 continued

68 Public Choice (2016) 168:55–74

Public Choice (2016) 168:55–74

69

Table 6 Marginal effects/changes in probabilities (in percentages) for specific conscription outcomes Category:

GovConsumption LogPopulation LagArmySize LogRGDPPC PolityIV SocialistOrigins UKOrigins UnskilledYouth War Age65

[18 months dy/dx

12–18 months dy/dx

6–12 months dy/dx

\6 months dy/dx

No conscription dy/dx

.49

.11

.04

-.02

-.62

(.31)

(.08)

(.15)

(.02)

(.43)

5.85***

1.36**

.53

-.31

-7.43**

(2.08)

(.55)

(1.77)

(.24)

(3.15)

.55*

.13

.05

-.03

-.70*

(0.28)

(.08)

(.16)

(.03)

(.36)

7.61***

1.77*

.70

-.41

-9.67**

(2.74)

(.95)

(2.37)

(.35)

(4.83)

-.14

-.03

-.01

.00

.17

(.22)

(.05)

(.04)

(.01)

(.28) -41.95***

31.03***

7.67**

3.04

-1.79

(5.94)

(3.40)

(10.01)

(1.49)

(13.80)

-43.97***

-10.22***

-4.04

2.39

55.85***

(13.78)

(3.22)

(12.31)

(2.49)

(5.56)

.17**

.04*

.01

-.00

-.22**

(.07)

(.02)

(.05)

(.00)

(.10)

1.44

.33

.13

-.07

-1.82

(.94)

(.25)

(.46)

(.07)

(1.37)

-4.11***

-.95***

-.37

.22

5.22***

(.97)

(.32)

(1.20)

(.20)

(1.17)

Delta-method standard errors are in parentheses. Marginal effects are multiplied by 100 to obtain percentage changes. Coefficients can be interpreted as the incremental change in probability that a particular outcome is observed, given a unit change in regressors. Values rounded to nearest one-hundredth * p\0:10; ** p\0:05; *** p\0:01

consisting of 33 OECD (usually considered as ‘‘rich’’), and 94 non-OECD countries.23 Notably, we find that the sign of the marginal effect of real GDP per capita differs across the samples. Among OECD countries the sign is positive, indicating that higher levels of real GDP per capita are associated with shorter conscripted military service terms. Increases in real GDP per capita among the non-OECD countries are, on average, associated with longer conscription terms. We take these findings as indicative of the view that richer countries place more weight on the economic costs of conscription and poorer countries on accounting costs in considering conscription lengths. Table 6 displays marginal effects of choosing each conscription outcome for a unit change in regressors. Here we report only the results from Table 4 specification (1) for brevity. For those unfamiliar with calculating marginal effects in multinomial probit models it is important to note three things. First, by construction the marginal effects of each variable sum to zero. Second, the sign of a marginal effect does not necessarily correspond in signs on the overall coefficients in Table 4. Third, the interpretation of the 23

We thank a referee for this suggestion.

123

70

Public Choice (2016) 168:55–74

marginal effect coefficients is that a one unit increase in the independent variable raises or lowers the probability of a country being in that category. For example, a one unit increase in the percentage of 15–24 year old males without primary education increases the probability of conscription duration by 0.17 %.

5 Conclusion The world has changed considerably since 1970. It has generally become more open, more democratic, and more economically free along many dimensions (Gwartney 2009; Sheehan and Young 2015). One of those dimensions is the use of conscription for military purposes. Unlike the previous literature on the determinants of conscription, we examine the factors that influence the intensive margin of conscription: the duration of compulsory military service. This is in contrast to the empirical literature that has focused almost exclusively focused on the likelihood of using conscription or the share of conscripts in the military. Using multinomial ordered probit and logit methods in a panel setting, our results generally are consistent with the existing literature on the determinants of the use of conscription. While our general empirical results largely are consistent with the literature, we extend this literature by providing more insight into how changes in the demographic structure and education levels of young males affect the lengths of conscripted miliary service across countries. Most notably, we find that the effect of the share of the population over the age of 65 is nonlinear. Increases in the elderly population share increase the probability of longer compulsory military service, lengths but at some point this effect turns negative and a larger elderly population share reduces draftees’ time in uniform. We also introduce a new variable to the literature that captures the percentage of 15–24 old males without a primary school education. We find that reductions in this variable have contributed to the decline in compulsory military service obligations. An interesting area of future research would be to investigate the effect of tax capacity and evasion on the decline in conscription, given the importance of being able to pay the higher wages necessary to fill an all-volunteer force. To provide but one example, while of relatively high income, Greece is well known for widespread tax evasion Skouras and Christodoulakis 2014 and still requires military conscripts to serve terms of nine months. In addition, our results could be used to inform future case studies of particular countries’ experiences with conscription in the spirit of Henderson (2005) and Perri (2013). Acknowledgments The authors would like to thank Adam Nowak for helpful comments and suggestions as well as attendees at the 2014 APEE, Public Choice, and WVU CFE Brown Bag. Part of this research was conducted while Hall was a Big XII Fellow with the Free Market Institute at Texas Tech University. Hall and Tarabar would like to acknowledge the Center for Free Enterprise at West Virginia University for general research support.

Appendix See Tables 7 and 8.

123

Public Choice (2016) 168:55–74

71

Table 7 Data description and sources Variable

Description

Source

Conscription

Ordered discrete variable taking values from most to least restrictive conscription regulation: 0, 1, 3, 5, 10

International institute for Strategic Studies & World Survey of Conscription and Conscientious objection to Military service

GovConsumption

The percentage share of government final consumption real expenditures in real GDP. Derived using constant 2005 USD

UN National Accounts Database

LogPopulation

The natural log of total population size

UN Demographic Statistics

LagArmySize

The ratio of total army size to total number of males 15–24 years of age

Barro-Lee dataset (2014) World Develop. Indicators

LogRGDPPC

The natural log of real GDP per capita. Derived using constant 2005 USD

UN National Accounts Database

PolityIV

Level of democracy ranging from -10 (complete autocracy) to ?10 (consolidated democracy)

Marshall et al. (2013) Dataset version p4v2013d

UKOrigins

Dummy taking the value 1 for countries with British legal tradition, 0 otherwise

LaPorta et al. (1999)

SocialistOrigins

Dummy taking the value 1 for countries with socialist legal tradition, 0 otherwise

LaPorta et al. (1999)

UnskilledYouth

Share of male population 15–24 with no or only primary education as the highest educational attainment

Barro-Lee dataset (2014) Authors’ calculations

War

Five-year moving average of inter- and intra-state conflict intensity

Marshall (2015)

Age65

Percentage share of total population aged 65 years of age and older

World Develop. Indicators

Table 8 List of countries with observed periods, full sample (N ¼ 127) Africa (N ¼ 33)

Asia & Oceania (N ¼ 34)

Europe (N ¼ 34)

Americas (N ¼ 26)

Algeria (1970–2010)

Armenia (2005–2010)

Albania (1970–2010)

Argentina (1970–2010)

Benin (1985–2010)

Australia (1970–2010)

Austria (1970–2010)

Barbados (1970–2010)

Botswana (1975–2010)

Azerbaijan (2005–2010)

Belgium (1970–2010)

Belize (1980–2010)

Burundi (1975–2010

Bahrain (1975–2010)

Croatia (1995–2010)

Bolivia (1970–2010)

Cameroon (1975–2010)

Bangladesh (1975–2010)

Cyprus (1980–2010)

Brazil (1970–2010)

Centr. Afr. Rep. (1980–2010)

Cambodia (2010)

Czech Rep. (1995–2010)

Canada (1970–2010)

Chad (1975–2010)

China (1970–2010)

Denmark (1970–2010)

Chile (1970–2010)

123

72

Public Choice (2016) 168:55–74

Table 8 continued Africa (N ¼ 33)

Asia & Oceania (N ¼ 34)

Europe (N ¼ 34)

Americas (N ¼ 26)

Congo, Dem. Rep. (1970–2010)

Fiji (1975–2010)

Estonia (1990–2010)

Colombia (1970–2010)

Congo, Rep. of (1975–2010)

India (1970–2010)

Finland (1970–2010)

Costa Rica (1975–2010)

Cote d’Ivoire (1970–2010)

Indonesia (1970–2010)

France (1970–2010)

Dom. Rep. (1970–2010)

Egypt (1970–2010)

Iran (1970–2010)

Germany (1970–2010)

Ecuador (1970–2010)

Gabon (1975–2010)

Israel (1970–2010)

Greece (1970–2010)

El Salvador (1970–2010)

Gambia (2010)

Japan (1970–2010)

Hungary (1970–2010)

Guatemala (1970–2010)

Ghana (1970–2010)

Jordan (1970–2010)

Ireland (1975–2010)

Guyana (1990–2010)

Kenya (1970–2010)

Kazakhstan (2005–2010)

Italy (1970–2010)

Haiti (1970–2010)

Lesotho (2005–2010)

S. Korea (1970–2010)

Latvia (1990–2010)

Honduras (1970–2010)

Malawi (1975–2010)

Kyrgyzstan (2005–2010)

Lithuania (1990–2010)

Jamaica (1975–2010)

Mali (1975–2010)

Malaysia (1970–2010)

Luxembourg (1970–2010)

Mexico (1970–2010)

Mauritania (2005–2010)

Mongolia (2005–2010)

Macedonia (2005–2010)

Nicaragua (1970–2010)

Mauritius (1975–2010)

Myanmar (1970–2010)

Moldova (2005–2010)

Panama (1975–2010)

Morocco (1970–2010)

Nepal (1975–2010)

Netherlands (1970–2010)

Paraguay (1970–2010)

Mozambique (2005–2010)

New Zealand (1970–2010)

Norway (1970–2010)

Peru (1970–2010)

Namibia (1990–2010)

Pakistan (1970–2010)

Poland (1970–2010)

Tr. & Tob. (1975–2010)

Niger (1970–2010)

P. N. Guinea (1980–2010)

Portugal (1970–2010)

USA (1970–2010)

Rwanda (1975–2010)

Philippines (1970–2010)

Romania (1970–2010)

Uruguay (1970–2010)

Senegal (1970–2010)

Qatar (2010)

Russia (1990–2010)

Venezuela (1970–2010)

Sierra Leone (1975–2010)

S. Arabia (2010)

Serbia (2005–2010)

S. Africa (1970–2010)

Sri Lanka (1975–2010)

Slovakia (1995–2010)

Tanzania (1970–2010)

Syria (1970–2010)

Slovenia (1995–2010)

Togo (1985–2010)

Tajikistan (2010)

Spain (1970–2010)

Tunisia (1970–2010)

Thailand (1970–2010)

Sweden (1970–2010)

Uganda (1970–2010)

Turkey (1970–2010)

Switzerland (1970–2010)

UAE (1980–2010)

Ukraine (1995–2010)

Vietnam (2005–2010)

UK (1970–2010)

Zambia (1970–2010)

123

Public Choice (2016) 168:55–74

73

References Adam, A. (2012). Military conscription as a means of stabilizing democratic regimes. Public Choice, 150(3–4), 715–730. Angrist, J. D., Chen, S. H., & Song, J. (2011). Long-term consequences of Vietnam-era conscription: New estimates using Social Security data. American Economic Review, 101(3), 334–338. Asal, V., Conrad, J., and Toronto, N. (2015). I want you! The determinants of military conscription. Journal of Conflict Resolution. http://jcr.sagepub.com/content/early/2015/09/23/0022002715606217.full. Barro, R. J., & Lee, J. W. (2013). A new data set of educational attainment in the world, 1950–2010. Journal of Development Economics, 104, 184–198. Becker, W. E., & Kennedy, P. E. (1992). A graphical exposition of the ordered probit. Econometric Theory, 8(1), 127–131. Berck, P., & Lipow, J. (2011). Military conscription and the (socially) optimal number of boots on the ground. Southern Economic Journal, 78(1), 95–106. Bergh, A., & Nilsson, T. (2010). Good for living? On the relationship between globalization and life expectancy. World Development, 38(9), 1191–1203. Bohanon, C. E., Horowitz, J. B., & McClure, J. E. (2014). Saying too little, too late: Public finance textbooks and the excess burdens of taxation. Econ Journal Watch, 11(3), 277–296. Borcherding, T. E. (1971). A neglected social cost of a voluntary military. American Economic Review, 61(1), 195–196. Bove, V., & Cavatorta, E. (2012). From conscription to volunteers: Budget shares in NATO defence spending. Defence and Peace Economics, 23(3), 273–288. Card, D., & Lemieux, T. (2001). Going to college to avoid the draft: The unintended legacy of the Vietnam War. American Economic Review, 91(2), 97–102. Cebula, R. J., & Mixon, F. G, Jr. (2012). Dodging the vote? Empirical Economics, 42(1), 325–343. Coyne, C. J. (2008). After war: The political economy of exporting democracy. Palo Alto, CA: Stanford University Press. Coyne, C. J., & Williamson, C. R. (2015). Foreign aid and the culture of contracting. Eastern Economic Journal, 41(1), 102–125. Dutta, N., Leeson, P. T., & Williamson, C. R. (2013). The amplification effect: Foreign aid’s impact on political institutions. Kyklos, 66(2), 208–228. Fisher, A. C. (1969). The cost of the draft and the cost of ending the draft. American Economic Review, 59(3), 239–254. Friedman, M. (1967). Why not a volunteer army? New Individualist Review, 4(4), 3–6. Gehring, K. (2013). Who benefits from economic freedom? Unraveling the effect of economic freedom on subjective well-being. World Development, 50, 74–90. Gwartney, J. (2009). Institutions, economic freedom, and cross-country differences in performance. Southern Economic Journal, 75(4), 937–956. Gwartney, J., Lawson, R., & Hall, J. (2015). Economic freedom of the world: 2015 annual report. vancouver: Fraser Institute. Hall, J. C., & Lawson, R. A. (2014). Economic freedom of the world: An accounting of the literature. Contemporary Economic Policy, 32(1), 1–19. Hansen, W. L., & Weisbrod, B. A. (1967). Economics of the military draft. Quarterly Journal of Economics, 81(3), 395–421. Henderson, D. R. (2005). The role of economists in ending the draft. Econ Journal Watch, 2(2), 362–376. Henderson, D. R. (2010). From ‘‘porous’’ to ‘‘ruthless’’ conscription, 1776–1917. Independent Review, 14(4), 587–598. Hubers, F., & Webbink, D. (2015). The long-term effects of military conscription on educational attainment and wages. IZA Journal of Labor Economics, 4(1), 1–16. International Institute for Strategic Studies. (2013). The military balance 2013. London: International Institute for Strategic Studies. Jehn, C., & Selden, Z. (2002). The end of conscription in Europe? Contemporary Economic Policy, 20(2), 93–100. Keller, K., Poutvaara, P., & Wagener, A. (2009). Military draft and economic growth in oecd countries. Defence and Peace Economics, 20(5), 373–393. Keller, K., Poutvaara, P., & Wagener, A. (2010). Does a military draft discourage enrollment in higher education? FinanzArchiv, 2, 97–120. LaPorta, R., de Silanes, F. L., Shleifer, A., & Vishny, R. (1999). The quality of government. Journal of Law, Economics and Organization, 15, 222–279.

123

74

Public Choice (2016) 168:55–74

Lau, M. I., Poutvaara, P., & Wagener, A. (2004). Dynamic costs of the draft. German Economic Review, 5(4), 381–406. Lee, D. R., & McKenzie, R. B. (1992). Reexamination of the relative efficiency of the draft and the allvolunteer army. Southern Economic Journal, 58(3), 644–654. Leeson, P. T., Sobel, R. S., & Dean, A. M. (2012). Comparing the spread of capitalism and democracy. Economics Letters, 114(1), 139–141. Lokshin, M., & Yemtsov, R. (2008). Who bears the cost of russia’s military draft? Economics of Transition, 16(3), 359–387. Marshall, M. G. (2015). Major episodes of political violence (mepv) and conflict regions, 1946–2014. Vienna, VA: Center for Systemic Peace. Marshall, M., Gurr, T., & Jaggers, K. (2013). Polity IV Project: Political regime characteristics and transitions, 1800–2012. Vienna, VA: Center for Systematic Peace. Mulligan, C. B., & Shleifer, A. (2005). Conscription as regulation. American Law and Economics Review, 7(1), 85–111. Ng, Y.-K., et al. (2008). Why is the military draft common? Conscription and increasing returns. Annals of Economics and Finance, 9(2), 373–384. Oi, W. Y. (1967). The economic cost of the draft. American Economic Review, 57(2), 39–62. Oyefusi, A. (2010). Oil, youths, and civil unrest in Nigeria’s Delta: The role of schooling, educational attainments, earnings, and unemployment. Conflict Management and Peace Science, 27(4), 326–346. Perri, T. J. (2013). The evolution of military conscription in the United States. Independent Review, 17(3), 429–439. Poutvaara, P., & Wagener, A. (2007a). Conscription: Economic costs and political allure. Economics of Peace and Security Journal, 2(1), 6–15. Poutvaara, P., & Wagener, A. (2007b). To draft or not to draft? Inefficiency, generational incidence, and political economy of military conscription. European Journal of Political Economy, 23(4), 975–987. Poutvaara, P., & Wagener, A. (2011). The political economy of conscription. In C. Coyne & R. Mathers (Eds.), The handbook on the political economy of war (pp. 154–174). Cheltenham: Edward Elgar. Ross, T. W. (1994). Raising an army: A positive theory of military recruitment. Journal of Law and Economics, 37(1), 109–132. Sayre, E. A. (2009). Labor market conditions, political events, and Palestinian suicide bombings. Peace Economics, Peace Science and Public Policy, 15(1), 1. Sheehan, K. M., & Young, A. T. (2015). It’s a small world after all: Internet access and institutional quality. Contemporary Economic Policy, 33(4), 649–667. Skouras, S., & Christodoulakis, N. (2014). Electoral misgovernance cycles: Evidence from wildfires and tax evasion in greece. Public Choice, 159(3–4), 533–559. Stroup, M. D. (2007). Economic freedom, democracy, and the quality of life. World Development, 35(1), 52–66. Tollison, R. D. (1970). Political economy of the military draft. Public Choice, 9(1), 67–78. War Resisters International. (2005). World survey of conscription and conscientious objection to military service. Cape Town: War Resisters International. Warner, J. T., & Asch, B. J. (2001). The record and prospects of the all-volunteer military in the United States. Journal of Economic Perspectives, 15(2), 169–192.

123