Societal Impacts of Modern Conscription: Human Capital, Social Capital and Criminal Behaviour

(1)

Societal Impacts of Modern

Conscription

Human Capital, Social Capital and Criminal Behaviour

Daniel Almén

Societ

al Impacts of Modern Conscription

Department of Economics

ISBN 978-91-7911-176-2 ISSN 1404-3491

(2)

(3)

Human Capital, Social Capital and Criminal Behaviour

Daniel Almén

Academic dissertation for the Degree of Doctor of Philosophy in Economics at

Stockholm University to be publicly defended on Friday 12 June 2020 at 13.00 in sal G, Arrheniuslaboratorierna, Svante Arrhenius väg 20 C.

Abstract

Opportunity Costs and Conscription: An Unintended Progressive Tax?

Throughout history to present days, policymakers, social commentators and others have oftentimes viewed conscription as a natural extension of secondary education, and an important institution for vocational training. This paper uses Swedish administrative data and exploits a reform in 2004, implying a sudden downsizing of the military, to identify the causal effects of peacetime conscription on later labour market outcomes and education. I find that unemployment increased in the short run, and lasted up to four years after service. There are no significant overall effects on income or educational attainment. However, these average effects hide a large heterogeneity. High ability conscripts fall behind their counterparts who did not start military service, both in terms of income and employment. Furthermore, the results suggest that the effect is attributed to high ability conscripts assigned as privates. In contrast, no such evidence is found for conscripts assigned to officer training, despite the fact that all of them have a high ability, and a longer time in service. Plausibly, high ability conscripts have high opportunity costs of doing military service, and the civilian benefits from training as privates are too small to counteract these costs. The results highlight the importance of precise matching of aptitude to type of training or education, an insight that might be generalized to other contexts beyond conscription.

Citizenship, Social Capital and the Role of Conscription: Evidence from Sweden

Many scholars have argued that conscription has played an important role as a nation-builder throughout history. Today, advocates of conscription often put forward its potential to induce citizenship and civic engagement. This paper addresses this claim by studying the causal effects of military service on civic engagement by using Swedish administrative data on election participation, blood donation, and the payment of a mandatory, but highly evaded, fee to the public broadcasting service. I study two qualitatively very different conscription systems from two different eras in Sweden, yielding a high external validity. To study the effects of universal conscription (almost all healthy and fit men serve) during the early 1990s, I use an empirical strategy similar in spirit to work using randomly assigned judges as an instrument. To identify the effects of selective conscription (a small fraction of motivated and positively selected men serve), I exploit a reform in 2004, implying a sudden downsizing of the military. In contrast to the previous correlational literature, the results show small and insignificant point estimates for all outcomes in both populations studied. Hence, I find no evidence of any causal effects of military service on civic engagement in either a selective-, or in a universal conscription system

The Effect of Military Conscription on the Formation of Criminal Behaviour: Evidence from a Natural Experiment Conscription has been suggested to be a policy-tool to break young men's anti-social life-trajectories. This paper uses Swedish administrative data and exploits a reform in 2004, implying a sudden downsizing of the military, to identify the causal effects of peacetime conscription on contemporaneous, short- and medium-term crime. I find no evidence of any effects on criminal activity while in service. However, the post-service results show crime increasing effects of military service at the intensive margin (number of convictions), but not at the extensive margin (probability of conviction). The overall crime increasing effect seems to be primarily driven by thefts. This study finds no support for increased overall violent behaviour or that the military context per se induces anti-social behaviour. Rather, some suggestive evidence for worsened labour market opportunities for some groups is documented as a plausible mechanism behind the crime increasing results.

Keywords: Conscription, Military, Human Capital, Social Capital, Civic Engagement, Crime, Criminal Behaviour,

Labour Market, Election Participation, Nation-building, School for the Nation, Opportunity Cost, Welfare State.

Stockholm 2020 http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-181015 ISBN 978-91-7911-176-2 ISBN 978-91-7911-177-9 ISSN 1404-3491 Department of Economics

(4)

(5)

(6)

(7)

Conscription

Human Capital, Social Capital and Criminal Behaviour

(8)

ISBN PDF 978-91-7911-177-9 ISSN 1404-3491

Cover photo: Housing for conscripts at Ledningsregementet, 2019. Photo: Martin Billberg/Försvarsmakten.

(9)

(10)

(11)

I feel very privileged to have been given this great opportunity to pursue doctoral studies at the Department of Economics, Stockholm University, and to be a part of such a stimulating environment. I am grateful for everything that I have learned during the coursework and while completing this thesis. These years have sometimes been extremely challenging and tough. With a continual quest to escape a fog of ignorance one can, and does, get exhausted. However, an important insight from doctoral studies is that one can never escape it, but one can learn how to handle it, and make sure to enjoy when it once in a while clears up. This insight, experience, and the knowledge gained during my doctoral studies, have truly enriched my life in many ways! In contrast to finding out a consequence of a cause, it is generally an im-possible task to determine the causes of a consequence. These causes are some unknown weighted combination of everything that has happened in the past. Thus, writing acknowledgments reflecting all the great people that have contributed to, and facilitated, the completion of this thesis in one way or another, is by some means an impossible task. Hence, I will mention a few, far from all, that I know have meant a great deal.

Ph.D-studies are, more or less, just like any hike into unknown terrain. Before one takes off, it is necessary to know the essentials, or it will probably not end up well. Getting to know the tools and equipment, training to use them. The first year during the Ph.D-studies is much about that, and the year is equally valuable and frustrating. Doing this together with others makes it much easier and a much better experience. I’m sure it would not have been possible without the help and cooperation with many of my friends and colleagues in my cohort. We all worked together, but I would like to especially thank Mathias Iwanowsky and Richard Foltyn for being a great support in many ways throughout the years. Thanks!

Next, one actually has to start walking, deciding and knowing where to go. Easier said than done. This is a tiresome, overwhelming and very lonely process, and good guidance is crucial. One of the first persons to listen to my research idea, leading to this thesis, was Hel´ene Berg who later became my advisor. She helped me specify the econometrics and make it a realistic endeavour by struggling with me through applications for research funding and later data collection. Above all, she encouraged me to stick to my current path and assured me that it was a good one, although somewhat high-risk. Thank you for the great support and encouragement from the beginning to the end!

(12)

together. One has to navigate in a smart and precise way. M˚arten Palme is an excellent navigator and guide. With a map, compass, sextant, and night vision devices he stood along the track and later accompanied me all the way. As an advisor he has given me invaluable advice and insights about big and small things. Without M˚arten, the thesis would have been twice as long, but half as good. Jonas Vlachos also showed up along the way. He has acted as a third advisor, encouraged and helped me in many ways, as well as provided me with many very useful comments and ideas. I am ever grateful!

I have also had the possibility and privilege to make a visit to University of California, San Diego. I especially thank Eli Berman for hosting me, making me feel welcome, inviting me to conferences, and giving me a great deal of good advice both during my stay and afterwards. Visitors to UCSD share a somewhat dark and chilly (yes that’s right) office. By coincidence Matz Dahlberg also spent some time there, which enriched my stay in many ways. I would also like to show my gratitude to Judith and Donald Wesling who invited me and my family to their house in Pacific Beach many times. Your garden gave us awaited peace of mind, while Judith were picking flowers, and teaching our daughter the English colours of the nature. Wise, helpful and generous!

Because I have been on leave for a couple of years, I have also had the opportunity to get to know Ph.D-students from other cohorts. Two of you are graduating about the same time as me. Thank you Louise Lorentzon and Niklas Blomqvist for exchanging ideas and teaming up during this last phase. My leave has also meant that I have switched offices a lot, and thus shared an office with quite a few. Dear fellow Ph.D-students, office-mates and friends, I cannot mention you all, you have all made my time in the basement and later on floor 4, a little easier and more enjoyable!

During my undergraduate studies at Lund University, I had Martin Nordin as an advisor. He encouraged me to apply to a Ph.D-programme and suppor-ted me very well in that process. Perhaps, this thesis would not have been written if it were not for his push. Matthew J. Lindquist did an excellent job as a discussant at my final seminar, and gave many valuable suggestions. Matthew’s research has turned out to be fruitful for my own research, I guess a good example of synergies in research! I have held a number of seminars at Stockholm University, where I have received a great deal of valuable feed-back from many committed members of the faculty, thank you all! A special

(13)

Finally, Anne Jensen has provided outstanding administrative support, as well as good work-out tips during these years!

Last but certainly not least, I would like to thank my family. My brothers Jonas and Marcus are always nice to hang out with. With you I always have funny and interesting conversations, our interests and knowledge are truly diverse, and that’s always fun! My uncle Benno has helped me with homework many times during high school, when struggling with math. Call Benno mom always said. He always answered with enthusiasm. Working behind the veil of the ordinary every day life, my grandmother Signild has been supportive and caring, with a great purpose and dedication beyond the limits of any successful Ted-Talker. A true hero!

I would not have had the opportunity, curiosity, nor the urge or capability to go on this hike if it wasn’t for my mom and dad. G¨oran has sown and spurred my societal interest, as well as inspired me to always go on and try out new things. Carina has been very caring and supported me many times, in many situations. Just like when I was about 10 years old, and wanted to learn all about scuba diving equipment. The store would only give out their thick fancy equipment catalogues to potential customers. Mom argued, I got the catalogue. Not yet bought my equipment!

Mom and dad have always encouraged and supported me, no matter what I have chosen to do. Sports, leisure activities, academics and career choi-ces. Both with the most important mental support, but also with the more practical, less exciting things that have to be dealt with in life, logistics and practicalities. Yet, perhaps the most important help is almost invisible, and easy to forget. Along the way on this hike (and many times before), I have always felt and known that if something goes wrong, support will be near. Indeed, with that safeguard, you can go faster, carry heavier, and take riskier paths than you would otherwise have done. No job is harder and requires such a determination as the job of being a good parent, I don’t know how to show you enough gratitude!

With me, since many years is my lovely wife Lisa. And, since not so many years, our beloved children Selma and Svante. Thank you Selma and Svante for conducting a forced shut down on my, sometimes overheated, brain every time I get home! Although you made calm moments a utopia, brought chaos and sleepless nights, without you, I’m sure I would have pushed myself too far down the never-ending road towards perfection. With children and

(14)

fa-read a single word. I guess that signifies how important support, good team-work, encouragement and love is for any achievement, big and small. Lisa has given me all this, despite having her own great ambitions and a much more important and difficult job than me. After all, it is you Selma would like to work with when she grows up, and she is truly convinced that I will become your assistant when I finally get my degree of doctor. Lisa, I’m so impressed by you and this thesis is equally yours!

The figures, tables, letter and words of this thesis are all put together by me. However, that part is just a drop in the ocean of the combined efforts and contributions done by many others throughout the years, of which many but not all, are mentioned above. After all, it is an impossible task to mention and acknowledge you all!

Daniel Alm´en Halmstad, Sweden April 2020

(15)

(16)

(17)

Introduction . . . 1 Chapter I:

Opportunity Costs and Conscription: An Unintended

Progressive Tax? . . . 9 Chapter II:

Citizenship, Social Capital and the Role of Conscription:

Evidence from Sweden . . . 115 Chapter III:

The Effect of Conscription on the Formation of Criminal

Behaviour: Evidence from a Natural Experiment . . . 213 Sammanfattning (Summary in Swedish) . . . 301

(18)

(19)

Introduction

The modern concept of nation-wide conscription was first introduced in France in the 18th century. Most European countries followed, and Swe-den introduced conscription in its modern form in 1901. In connection to its introduction, and throughout the 20th century, the belief in conscription fostering the ”good citizen” and being a ”school for the nation” in which citizens would learn portable skills has been prevalent in Europe, as well as in the U.S. It was seen as instrumental to nation-building, the overcom-ing of class and regional divisions within national borders, and a natural extension of secondary education.1 These ideas were also very prevalent in Sweden, and the virtues of a good citizen promoted by the military did, to a large extent, reflect the ideals promoted by the Swedish state.2 In fact, it has been suggested that Swedish conscription has been the largest fostering institution in Sweden’s history.3

As expressed by Forssberg & Fredholm (2014): ”the story of conscription is a story of Sweden”. Democratization, the creation of ”the people’s home” (the Swedish welfare state), the Cold War, and equality - all these social phenomena have in different ways interplayed with the military. Through-out most of the 20th century, almost all men in every cohort were drafted, and a remarkable consensus on conscription was well established (Dahlström & Söderberg, 2002a). Today, more than a million Swedish men (and some women) across geographic regions and of various backgrounds share the ex-perience of conscription (Forssberg & Kronberg, 2014). Reflecting that it was a well-integrated institution that most citizens in one way or another had to relate to, conscription has frequently been depicted, praised and criticized in books, movies and other media (Dahlström & Söderberg, 2002b).

By the end of the Cold War, the number of conscripts started to decline, and in 2009, the Swedish parliament decided to abolish (peacetime) conscription. Hence, within less than a generation, a major institution affecting almost half of every cohort was completely phased out. This is arguably one of the largest social policy changes to take place in modern Swedish history.

1

See e.g. Huntington (1957), Janowitz (1971, 1976), Weber (1976), Cohen (1985), Tilly (1990), Levi (1997), Frevert (2004), Krebs (2004), Hettne et al. (2006), Sturfelt (2014).

2_{Rudberg (2014), Sturfelt (2014) and Sundevall (2017). See Berggren & Tr¨}_ag˚_{ardh (2006)}

for an overview of how the Swedish history of ideas evolved at the time. From a historical point of view, Tr¨ag˚ardh et al. (2013) highlight the importance of compulsory schooling and conscription for the formation of social capital in Sweden.

3_{Klas Kronberg, editor of the book ”Lumpen - fr˚}_{an m¨}_{onstring till muck” (Forssberg}

& Kronberg, 2014) in an interview at https://www.forsvarsmakten.se/sv/information-och-fakta/var-historia/artiklar/100-ar-i-lumpen/.

(20)

In connection with the abolishment, and the following years up until the decision to reinstate conscription in 2017, conscription was a frequent topic in the public debate. Many of the arguments put forward echoed the wide-spread arguments of the early 20th century.

Against the backdrop of this rapid policy change and the lively public de-bate, Swedish conscription is surprisingly understudied by quantitatively oriented researchers. In contrast, the more qualitatively oriented literature on Swedish conscription is somewhat larger.4 This research has documented many individual testimonies of former conscripts who are convinced that their time in the military (paraphrasing): ”gave them invaluable knowledge for life”; ”made them grow”; ”strengthened their self-confidence”; ”taught them how to co-operate and the value of fellowship”; and ”changed their values and beliefs”. Few claim that conscription had an adverse impact on them. Even fewer believe that they were unaffected (see Forssberg & Kronberg (2014), especially chapter 5 (Wollin Elhouar, 2014)).

This thesis seeks to add to the scarce quantitative literature on the societal effects of conscription, and is the first comprehensive quantitative work on the topic to date.5 It consists of three self-contained articles presented in three chapters. All of them make use of rich administrative data and study Swedish peacetime conscription during the post Cold War era. Each article concerns the effects on distinct aspects of society: i) The labour market and education; ii) civic engagement and social capital (pro-social behaviour); and iii) criminal (anti-social) behaviour. A first look at the data reveals that former conscripts, on average, perform better on the labour market, obtain higher levels of education, are much more prone to civic engagement and less prone to criminal behaviour. On the other hand, the data also shows that individuals selected into military service have better pre service health, score better on the enlistment-tests and have parents with higher levels of education and income.6 That said, the aim and challenge of the articles

4_{See e.g. Rehn (1999), Ericsson (1999), Malmer (2013), Forssberg & Kronberg (2014),}

Sturfelt (2014), Sundevall (2016) and Sundevall (2017).

5_{Internationally, some work is being done on the causal effects of conscription. Most}

studies investigate the impacts on labour market outcomes and crime. See later chap-ters in the present thesis for further references and a more thorough description of the literature.

6

In 1990, Swedish conscripts represented the overall male population fairly well. However, as the number of conscripts declined during the 1990s and early 2000s, the positive selection increased substantially and hence the military gap grew wider. A civil-military gap has been problematised by, for example, Samuel P. Huntington and Morris Janowitz (Huntington, 1957, Janowitz, 1960). See Section A in the Appendix of the first chapter for an overview of how the civil-military gap has evolved in Sweden.

(21)

are to take the process of selection into account and hence, to isolate and estimate the causal effects of conscription. A causal effect of a treatment is estimated if the observed differences in outcomes can be attributed to whether an individual has been treated or not, and nothing else. To achieve this, the articles make use of natural experiments generating an as good as random allocation of individuals into treatment (i.e. military service). The first chapter, ”Opportunity Costs and Conscription: An Unintended Progressive Tax?”, studies the causal effect of (post 2000) conscription on labour market outcomes and educational attainment. The main results sug-gest some secondary costs of conscription. I find that conscription increased post service unemployment in the short run. No significant overall effects on income or educational attainment are found. However, these average effects hide a large heterogeneity. High ability conscripts fall behind their counterparts who did not start military service, both in terms of income and employment. To that end, Swedish modern conscription has operated as an implicit progressive tax. Furthermore, the results suggest that the effect is attributed to high ability conscripts assigned as privates. In contrast, no such evidence is found for conscripts assigned to officer training, despite the fact that all of them have a high ability, and longer time in service.7 Plau-sibly, high ability conscripts have high opportunity costs of doing military service, and the civilian benefits from training as privates are too small to counteract these costs. Hence, the secondary costs of conscription plausibly arise from a discrepancy between the ability of the conscript (his oppor-tunity cost) and the complexity of the tasks performed during service and training.

The second chapter, ”Citizenship, Social Capital and the Role of Conscrip-tion: Evidence from Sweden”, studies the causal effects of military service on civic engagement and social capital. This is done by using administra-tive data on election participation, blood donation, and the payment of a mandatory (but highly evaded) fee to the public broadcasting service. The article studies two qualitatively very different conscription systems from two different eras, post 2000 and the early 1990s. Hence, both selective- and uni-versal conscription are studied. The study finds no evidence of any causal effects of military service on civic engagement in either a selective-, or in a universal conscription setting.

The third chapter, ”The Effect of Military Conscription on the Formation of Criminal Behaviour: Evidence from a Natural Experiment”, studies the

7

Hence, in line with previous research, qualified military training such as officer training might entail some benefits on the civilian labour market (Gr¨onqvist & Lindqvist, 2016).

(22)

causal effect of (post 2000) conscription on criminal behaviour. The study finds overall crime increasing effects, primarily driven by increased theft. The study finds no support for increased overall violent behaviour or that the military context per se induces anti-social behaviour. Rather, some sug-gestive evidence for worsened labour market opportunities for some groups is documented as a plausible mechanism behind the crime increasing re-sults.

In addition to adding knowledge to the literature on the effects of conscrip-tion, the three chapters also give valuable insights into each literature fo-cusing on the different aspects of the society studied. The first article con-tributes to the fields of labour economics and economics of education. The secondary costs of conscription plausibly arise from a mismatch. This insight might be generalized to other contexts such as labour market programmes and education. Analogous to military training, a better matching of, for example, job training to aptitude does plausibly have a high return.

The second article speaks to the broad literature in the social sciences on civic engagement and social capital. The levels of social capital differ substan-tially across countries, and Sweden together with the other Nordic countries exhibit exceptionally high levels (see e.g. Algan & Cahuc (2014), Tr¨ag˚ardh (2007) and Tr¨ag˚ardh et al. (2013)).8 However, how social capital is gener-ated does yet remain to be clarified. As pointed out by Rothstein & Stolle (2008), the fact that citizens in some countries and regions are able to trust each other and thereby overcome their collective action problems while oth-ers are not, is one of the most interesting puzzles in the social sciences. On both historical and theoretical grounds, conscription is an appealing candi-date and this has indeed been put forward in the public debate. However, the results of the second article show no support for the long-standing view that modern conscription provides a setting in which civic engagement or social capital is largely enhanced. Finally, the third article gives insights into the formation of criminal behaviour and contributes to the literature on education-based policies and crime.

To sum up, the general finding of this thesis is that modern Swedish con-scription has had a modest impact on most conscripts, and thus arguably also on society at large. This is certainly a surprising result, given the long standing ideas of conscription as an important fostering institution, teaching valuable portable skills, and the many documented testimonies of a perceived life-transforming event. However, a quantitative approach like that in the

(23)

present thesis does neither confirm nor falsify individual testimonies of good or bad experiences, or the perceived benefits or disadvantages of being con-scripted. Rather, it measures the average causal effect of conscription on conscripts, and in some cases relevant sub-groups. Yet, these measures are valuable and relevant measures for public policy. Specifically, the analysis of the post 2000 conscription is relevant for Sweden today. The conscription system implemented in 2018 resembles the system in place during the early 2000s in many ways.

As is often the case with research, new findings give rise to new questions. The first imminent question that arises from this thesis is what explains the discrepancies between the documented results and conclusions of previous qualitative work, and the conclusions of the present study. With this thesis, we know a little more, yet we still know very little of the societal impacts of conscription.

References

Algan, Y. & Cahuc, P. (2014). Trust, growth and well-being: New evidence and policy implications. In P. Aghion & S. Durlauf (Eds.), Handbook of Economic Growth chapter 2. Oxford: Elsevier.

Berggren, H. & Träg˚ardh, L. (2006). Är svensken människa? Stockholm: Nordstedts.

Cohen, E. A. (1985). Citizens and Soldiers: The Dilemmas of Military Service. Ithaca: Cornell University Press.

Dahlström, J. & Söderberg, U. (2002a). Inledning. In J. Dahlström & U. Söderberg (Eds.), Plikt, politik, och praktik. Värnplikts försvaret under 100 ˚ar chapter 1. Stockholm: Krigsarkivet.

Dahlström, J. & Söderberg, U. (2002b). Plikt, politik, och praktik. Värnplikts försvaret under 100 ˚ar. Stockholm: Krigsarkivet.

Ericsson, L. (1999). Medborgare i vapen – v¨arnplikten i Sverige under tv˚a sekel. Lund: Historisk media.

Forssberg, A.-M. & Fredholm, A. (2014). Inledning. In A.-M. Forssberg & K. Kronberg (Eds.), Lumpen: fr˚an m¨onstring till muck, Arm´emuseums ˚arsbok. Stockholm: Atlantis.

Forssberg, A.-M. & Kronberg, K. (2014). Lumpen: fr˚an m¨onstring till muck. Arm´emuseums ˚arsbok. Stockholm: Atlantis.

(24)

Frevert, U. (2004). A Nation in Barracks: Modern Germany, Military Con-scription and Civil Society. Oxford: Berg.

Gr¨onqvist, E. & Lindqvist, E. (2016). The making of a manager: Evidence from military officer training. Journal of Labor Economics, 34(4), 869–898. Hettne, B., S¨orlin, S., & Østerg˚ard, U. (2006). Den globala nationalismen.

Stockholm: SNS.

Huntington, S. P. (1957). The Soldier and the State: The Theory and Politics of Civil-Military Relations. Cambridge: Harvard University Press. Janowitz, M. (1960). The Professional Soldier, a Social and Political

Por-trait. Glencoe: Free Press.

Janowitz, M. (1971). Basic education and youth socialization in the armed forces. In R. W. Little (Ed.), Handbook of Military Institutions chapter 5. Beverly Hills: Sage Publications.

Janowitz, M. (1976). Military institutions and citizenship in western soci-eties. Armed Forces & Society, 2(2), 185–204.

Krebs, R. R. (2004). A school for the nation? how military service does not build nations, and how it might. International Security, 28(4), 85–124. Levi, M. (1997). Consent, Dissent, and Patriotism. Political Economy of

Institutions and Decisions. Cambridge: Cambridge University Press. Malmer, E. (2013). Hemmet vid nationens skola V¨ackelsekristendom, v¨

arn-plikt och soldatmission, ca 1900-1920. PhD thesis, Stockholm University, Stockholm.

Rehn, S. (1999). Unintentional learning - the sheet-anchor of the Swedish armed forces? An examination of the learning of conscripts during their training period and of the function of the Swedish national service system in society. PhD thesis, Lule˚a University of Technology, Lule˚a.

Rothstein, B. & Stolle, D. (2008). The state and social capital: An institu-tional theory of generalized trust. Comparative Politics, 40(4), 441–459. Rudberg, P. (2014). Arm´en m˚aste blifva en skola f¨or hela folket:

Krigsmak-tens folkfostrande ambitioner och praktiker 1901-1950. In A.-M. Forssberg & K. Kronberg (Eds.), Lumpen : fr˚an m¨onstring till muck, Arm´emuseums ˚arsbok chapter 3. Stockholm: Atlantis.

(25)

narrative of conscription during the early cold war. Milit¨arhistorisk Tid-skrift, (1), 23–58.

Sundevall, F. (2016). Lgbt in the military: Policy development in sweden 1944–2014. Sexuality Research and Social Policy, 13(2), 119–129.

Sundevall, F. (2017). Military education for non-military purposes: Economic and social governing projects targeting conscripts in early twentieth-century sweden. History of Education Review, 46(1), 58–71. Tilly, C. (1990). Coercion, Capital and European States, AD 990–1990.

Oxford: Basil Blackwell.

Tr¨ag˚ardh, L., Ed. (2007). State and Civil Society in Northern Europe - The Swedish Model Reconsidered. New York: Berghahn Books.

Träg˚ardh, L., Wallman Lund˚asen, S., Wollebæk, D., & Svedberg, L. (2013). Den svala svenska tilliten - Förutsättningar och utmaningar. Stockholm: SNS Förlag.

Weber, E. (1976). Peasants into Frenchmen: The Modernization of Rural France 1870-1914. Stanford: Stanford University Press.

Wollin Elhouar, E. (2014). Det bästa som hänt mig men aldrig mer -lumpen som ambivalent upplevelse. In A.-M. Forssberg & K. Kronberg (Eds.), Lumpen: fr˚an mönstring till muck, Armémuseums ˚arsbok chap-ter 5. Stockholm: Atlantis.

(26)

(27)

Opportunity Costs and Conscription: An Unintended

Progressive Tax?∗

∗

First and foremost, I would like to thank Hel´ene Berg, M˚arten Palme and Jonas Vlachos for invaluable support in writing this article. I am also very grateful for many helpful comments by Matthew J. Lindquist and Eli Berman. This article has also benefited a lot from helpful comments by seminar participants at: Stockholm University; Univer-sity of California, San Diego; the ENSA-session at the 94th WEAI Annual Conference, San Francisco; and the 3rd Stockholm–Uppsala Education Economics Workshop. I’m also grateful for comments received when presenting at the Swedish National Audit Office. Several at the Swedish Defence Recruitment Agency have been very generous in helping me with data and other general questions. Financial support from the Jan Wallander and Tom Hedelius’ Foundation and the Lars Hierta Memorial Foundation is gratefully acknowledged.

(28)

”The real cost of conscripting a soldier who would not voluntarily serve on present terms is not his pay and the cost of his keep. It is the amount for which he would be willing to serve. He is paying the difference. This is the extra cost to him and must be added to the cost borne by the rest of us.”

— Milton Friedman, 1966b

1. Introduction

During the 20th century, many viewed conscription as a natural extension of secondary education and an important institution for vocational train-ing. Sweden was not an exception.1 _{However, in the U.S. in the 1960s,}

the draft received attention from economists and the analysis of the costs and benefits of conscription became predominantly influenced by economics which restricted the analysis to military efficiency, although not without crit-icism (Cohen, 1985, Cowen, 2006).2 Milton Friedman and Walter Oi, among others, argued that the costs of the draft were underestimated due to the implicit tax (i.e. opportunity cost) levied on the conscripts (Friedman, 1967, Oi, 1967).3 Following the recommendations of the Gates Commission’s final report, in which these arguments were pivotal, the draft was abolished in

a

See Cohen (1985, p. 129-130).

b

Presentation at a conference on the draft. Later published as Friedman (1967, p. 204).

1_{See e.g. Janowitz (1971) and Cohen (1985) for the U.S. President Roosevelt, President}

Truman and General Eisenhower supported this view (Cohen, 1985). Furthermore, a policy in spirit of this view was ”Project 100,000”, initiated in 1966. The aim was to bring 100,000 “underprivileged” youths into the U.S. armed forces and teach these youths skills, discipline, reduce unemployment and make them eligible for veterans’ welfare programmes (Segal, 1989). For Sweden, see e.g. Sundevall (2017), Sturfelt (2014) and Forssberg & Kronberg (2014).

2_{Charles Moskos and Morris Janowitz were two distinguished sociologists criticizing the}

economic approach (see e.g. Janowitz & Moskos (1979), Moskos (1981), Moskos & Faris (1982), Moskos (1988) and Moskos & Wood (1988)).

3

Similar inefficiency arguments were pointed out already by Adam Smith and John May-nard Keynes (Poutvaara & Wagener, 2009, Johnson, 1960).

(31)

1973.4 Although still very widespread, conscription has been on the decline since the Cold War.5 Nevertheless, it has very recently been reintroduced in some countries (e.g. United Arab Emirates, Lithuania, and Sweden) while others are contemplating this (e.g. France). Today, scholars and policy-makers debate about the costs and benefits of conscription, some on pure military grounds, while others stress its wider potential beneficial impact on society.6

Apart from the education system, no other institution has engaged as many individuals for such a long time as conscription. The military provides the largest government training programme with a fair share of every cohort participating.7 Moreover, military service coincides with a critical time in life when the individual is young, more adaptive and about to make important choices in life. The service is long, typically about a year and it is intense. Conscripts are trained for more than regular full-time work hours and live together at base. Despite the potentially large effects on later outcomes of those assigned to military training through conscription, there are very few studies empirically examining these potential effects.

This paper uses a natural experiment provided by a government reform in

4

The Gates Commission (The President’s Commission on an All-Volunteer Armed Force (U.S., 1970:2)) was initiated by President Nixon in 1969 and both Milton Friedman and Walter Oi became members of the commission. See e.g. Cohen (1985), Warner & Hogan (2016) and Hendersson (2005) for a discussion on the influence and importance of economic arguments in the Gates Commission and the abolition of the draft.

5

As of today, eight EU-member-, five NATO-member- and more than 60 countries have conscription. See the CIA’s World Factbook https://www.cia.gov/library/publications/the-world-factbook/fields/333.html and http://chartsbin.com/view/1887.

6

In the U.S., recent advocates of conscription include retired Army Maj. Gen. Dennis Laich, retired Army Gen. Stanley McChrystal and former Chairman of the Joint Chiefs of Staff Adm. Mike Mullen (Laich, 2013, Foreign Policy , 2012, Military Times, 2017). A type of non-military conscription was also promoted (through the campaign ”A New Call to Service”) by the 2020 democratic presidential candidate Pete Buttigieg. Scholars such as Charles Moskos have also spoken in favour of conscription (e.g. Moskos (2001, 2005a,b)). In Europe, prominent voices promoting conscription are Prince Harry and President Macron (The Telegraph, 2015, Le Monde, 2017). Positive effects on the workforce constituted a driving factor behind the recent introduction of conscription in the United Arab Emirates (CSIS, 2017).

7

Labour market programmes described by Heckman et al. (1999) as ”work experience-” and ”on-the-job training”-programmes are in many aspects similar to military training. The former are designed to ”introduce young people to the world of work” (Heckman et al., 1999). In the case of Sweden, the number of individuals participating in such programmes is far below the number of conscripts during the conscription era (AF, 2018).

(32)

2004, implying a sudden downsizing of the armed forces, to isolate the causal effects of conscription on short- and medium-term labour market outcomes and education. The background to this policy change was that, following the Cold War, Sweden downsized its armed forces. The 2004 reform was particularly large, immediately implemented and what units to be closed was not foreseen. On the day when the reform was announced, many in-dividuals had already been tested and assigned (but had not yet started service) to units subsequently closed down by the reform. These individu-als had a significantly lower probability of starting service in comparison to those assigned to other units, and thus serve as control group. The outcomes studied are unemployment, university enrolment, years of schooling, income and earnings up until age 29.

The results suggest some secondary costs of conscription. I find a signif-icant positive effect on unemployment that amounts to about 35 days of additional unemployment, or a 0.25 additional year when the individual has been registered as unemployed between the ages 22 and 29. A dynamic ap-proach, where the effects are estimated by year after service-entry, shows a significant positive effect on unemployment for up to four years after entry. Hence, military service interrupts and delays the entry on the labour market and the effect seems larger than the mechanical offsetting effect of entering the labour market later. I find no statistically significant effect neither on income nor on earnings although the point estimates show an initial drop and a subsequent catch up. Moreover, I find no evidence of overall effects on university enrolment or educational attainment.

These overall effects hide a large heterogeneity between different groups. The results show very small, or insignificant, effects for low ability conscripts. However, after service, high ability conscripts fall behind their counterparts who did not start military service, both in terms of income and unemploy-ment. Taking the heterogeneity analysis further, I find that it is the high ability privates that are driving the results. In contrast, no such evidence is found for conscripts assigned to officer training, despite the fact that all of them have high ability. Plausibly, for high ability conscripts, the opportu-nity costs for doing military service are high and the civilian benefits from training as privates are too small to counteract these costs. As for educa-tional attainment, I find no clear evidence of an effect on years of school-ing. However, for high ability conscripts, and especially those assigned as platoon-level officers, I document a postponed university enrolment.

A well-known strand of the literature on conscription studies the effects of conscription during the WWII or Vietnam-era (e.g. Angrist (1990), Angrist

(33)

& Krueger (1994), Angrist & Chen (2011), Angrist et al. (2011), Lemieux & Card (2001)). However, wartime service is conflated with combat experi-ence, or the risk of being exposed to it, thus limiting the external validity. In order to address the question of the effect of peacetime conscription, which is the most frequent and arguably the most policy relevant setting, stud-ies on peace- and wartime conscription have to be held apart. Turning to the scant peacetime conscription literature, two studies are closely related to this study. First, Bingley et al. (2020) study Danish conscripts and find negative effects on earnings in the short run, driven by high ability con-scripts. Second, Hjalmarsson & Lindquist (2019) study Swedish conscripts and find positive effects on income post age 30. Further analysis suggests that individuals from disadvantaged backgrounds lose while individuals from advantaged backgrounds benefit. None of the studies find any effects on over-all educational attainment although Hjalmarsson & Lindquist (2019) find a negative effect for individuals with a criminal history prior to service, and Bingley et al. (2020) find a delay for high ability men. In sum, previous literature shows mixed results and the context seems to matter.

This paper makes two main contributions to the previous literature. While many scholars stress the inefficiency of conscription due to the opportunity costs of the conscripts, others stress the potential beneficial impact military training might have on later outcomes. The first and main contribution of this paper is to acknowledge, and empirically document, that both factors are at play and counter each other. The reform together with detailed data allow me to study heterogeneous effects with respect to individual type (ability and SES-background), treatment (type of military training), their interaction, and thus separate the effects of the two.

As theoretically pointed out by Warner & Asch (2001), the inefficiency (i.e. the opportunity costs of conscripts) of conscription depends on who is drafted and thus, the selection method. In line with this insight, my analysis shows that the effects of conscription indeed depend on the opportunity cost (prox-ied by ability) of the conscript. However, the quality or civilian benefit from the particular type of military training also matters. Hence, the opportunity costs can potentially be counteracted if the conscription system matches in-dividuals to positions and training with well-balanced complexity, demands and requirements in relation to the aptitudes of the conscripted individu-als.

Second, as opposed to previous studies, I study positively selected individuals that are conscripted into a more modern and high-tech armed force that earlier studies. Today, many armed forces conscript a much smaller and more

(34)

positively selected fraction of each cohort than during the 20th century’s labour intense armed forces. This makes the results of this paper policy relevant for today’s public policy debate on conscription.

The remainder of this paper is organized as follows. Section 2 gives an overview of previous literature. Section 3 provides a background to con-scription in Sweden and The Swedish Armed Forces (SAF) together with some key details of the Swedish enlistment procedure. Section 4 presents the empirical strategy followed by a section describing the data. An em-pirical analysis of the validity and relevance of the reform is provided in Section 6. The results are presented in Section 7. Finally, Section 8 dis-cusses the results vis-`a-vis the existing research, and Section 9 concludes the paper.

(35)

2. Previous Literature

The most influential papers study U.S. war veterans who served in either World War II or Vietnam (Angrist & Krueger, 1994, Angrist, 1990).8 The general finding is that conscription has a negative effect on future earnings or income, and lost civilian job experience is the proposed main mechanism for the Vietnam veterans. However, while white Vietnam veterans suffered substantial earnings losses, no evidence of this is found for black veterans. Overall, the negative effects are more pronounced for Vietnam veterans than for WWII-veterans, although the negative effect seems to vanish in the long run (Angrist et al., 2011). A positive effect of conscription on schooling is also documented, either as a consequence of draft avoidance as suggested by Lemieux & Card (2001) or by a veteran benefit programme, the GI-Bill (An-grist & Chen, 2011). Draft avoidance and the presence of the GI-bill make this a quite specific institutional setting. Moreover, wartime conscription studies are necessarily conflated with combat exposure, or the risk of being exposed together with other wartime contextual particularities. Therefore, when assessing the effects of conscription, wartime conscription and peace-time conscription ought to be held apart.

A related but small literature concerns the effects of voluntary military ser-vice on labour market outcomes. Although voluntary military serser-vice is dif-ferent from conscription, in practice, some conscription systems have larger or smaller parts of voluntarism.9 Angrist (1998) studies the effects of vol-untary military service in the U.S. on labour market outcomes and finds that soldiers who served in the early 1980s experienced a positive effect on employment during and after service.10 However, the effects on earnings after service were positive but small for nonwhites while actually negative for white veterans.11

8_{Instrumental variables estimation strategies are used to exploit the fact that quarter}

of birth was correlated with the probability of being drafted (WWII) and that some cohorts were drafted through a draft lottery based on date of birth (Vietnam War).

9

In particular, this applies to conscription systems similar to that studied in the present paper, where conscription does not coexist with other recruitment systems. Thus, volunteering is not an option.

10

The early 1980s was a period with only minor U.S. military operations. Two different empirical strategies are used. The first strategy compares applicants who did not enlist to applicants who enlisted using matching methods while the second uses instrumental variables that were generated by an error in the scoring of the exams that screen military applicants.

11_{In line with the result of increased employment, Figinski (2019) conducts a resume}

study and finds that there is a civilian labour market benefit, in terms of callbacks and requests for interviews, from military experience in the Reserve Component of the U.S.

(36)

There are few studies outside the U.S. studying the effects of conscription on labour market outcomes and educational attainment. Regarding edu-cational attainment, the draft avoidance behaviour is of major concern in many countries. Studies on the effects on educational attainment where draft avoidance is not a possible mechanism are Hubers & Webbink (2015) (The Netherlands), Hjalmarsson & Lindquist (2019) (Sweden) and Bingley et al. (2020) (Denmark). The former finds an overall negative effect while the latter two find no effects although Bingley et al. (2020) identify a de-lay.12 Turning to labour market outcomes, Grenet et al. (2011) and Bauer et al. (2012) use a regression discontinuity design on cohorts in the 1950s and find no effect on long-run earnings in the U.K. or Germany. Exploit-ing the same reform in the Netherlands, Imbens & van der Klaauw (1995) and Hubers & Webbink (2015) both find negative effects in the medium and long run, respectively.13 In comparison to the present study, these studies on labour market effects study much older cohorts and only identify mean effects.14

However, some studies find positive effects for selected subgroups, e.g. low educated Portuguese men (Card & Cardoso, 2012) and Arab Druze men in Israel (Asali, 2017). Few studies study heterogenous effects in terms of type of treatment. Exceptions are Hanes et al. (2010) and Goldberg & Warner (1987). The former study conscripts in Sweden during the early 1990s and find a positive effect on earnings at age 30 for privates. Goldberg & Warner (1987) study veterans of the Vietnam era in the U.S. and find that the returns to military experience vary by military occupations. However, they both have to rely on the selection on observables in a cross section analysis.

Two studies on peacetime conscription, briefly mentioned in the introduc-tion, study heterogeneous effects extensively. Using a draft lottery, Bingley et al. (2020) study Danish conscripts serving during the late 1990s and early 2000s and find a negative effect on earnings already in the short run.

Al-Armed Forces.

12_{Di Pietro (2013) finds no effect on university enrolment of the abolition of compulsory}

military service in Italy. Maurin & Xenogiani (2007) find that the educational attain-ment decreased after the abolition of conscription in France. Similar effects are found by Bauer et al. (2014) in Germany. In all three studies draft avoidance behaviour is at play and in the latter two papers, the results seem to be driven by it.

13_{The entire cohort born in 1959 was exempt from military service.} 14

Siminski (2013) uses the Australian draft-lottery during the Vietnam war to study the long-run effects of service. He finds large negative effects on employment measured at age 55-60. However, the results are attributed to the Australian veteran programme providing disincentives to work and are thus very context specific.

(37)

lowing the effect to vary across the cognitive ability distribution, they find that the negative effect is driven by high ability conscripts.15 Part of the effect is due to postponed educational attainment for this group. Hjalmars-son & Lindquist (2019) study Swedish conscripts serving during the early 1990s, and use the exogenous variation in the chance of serving that arises from random assignment to enlistment officers who are more or less likely to enlist individuals. In contrast to Bingley et al. (2020), they present re-sults suggesting positive overall effects on income, post age 30.16 However, further analysis suggests that individuals from disadvantaged backgrounds are worse off in terms of income and unemployment while individuals from advantaged backgrounds are better off. In general, the diverging income effects are found post age 30 while the unemployment effects appear earlier. When studying heterogeneous effects on educational attainment, Hjalmars-son & Lindquist (2019) find a negative effect on educational attainment for individuals with a criminal history prior to service, and Bingley et al. (2020) find that the delay in overall university enrolment is driven by high ability individuals.

The two latter studies are distinct from the present study in three main aspects. First, the draft lottery used by Bingley et al. (2020) only applies to those who are not volunteers, arguably implying that the motivation among conscripts is lower. Second, there is less heterogeneity of treatment in terms of duration and service categories in the study by Bingley et al. (2020). Third, Hjalmarsson & Lindquist (2019) study conscripts during a period of time when almost all men fit for service were drafted and thus, the conscripts were recruited from the whole ability distribution and were much less positively selected in all dimensions.17 These differences are further discussed in Section 8.

15

Similar effects are found using parental background. That is, individuals from higher SES-backgrounds are hurt the most.

16_{The focus of their study is criminal behaviour, however.} _{They do, in general, find}

crime increasing effects driven by individuals from disadvantaged backgrounds and labour market outcomes are studied when exploring potential mechanisms. Income is measured as the average income between ages 30-34.

17_{Section 3 and Section A in the Appendix give an overview of the structural change of}

the Swedish Armed Forces and its implications for the selection and composition of enlisted conscripts in terms of SES-background and ability.

(38)

3. Conscription in Sweden

3.1. The Perception of Military Training as Schooling

Sweden introduced large-scale conscription in 1901 and relied on this to supply its armed forces with soldiers until it was abolished in 2010 (Ericsson, 1999, Kronberg, 2014). In connection to its introduction, it was argued to be ”a school for the nation”, to improve hygiene, to be a tool to control alcohol consumption and instil physical discipline (Rudberg, 2014). In the following years during the first half of the 20th century, policymakers and social commentators on education recurrently advocated military service as a tool to deal with various economic and social problems. Examples were mass unemployment, uneducated voters as well as a shortage of skilled personnel in major work sectors (Sundevall, 2017). In the 1930s, as the Social Democrats developed the vision of Sweden as a Folkhem (people’s home) emphasising social rights, equality and social mobility, conscription was promoted as an important institution in the nascent welfare state (Leander, 2005, Sturfelt, 2014).18

Over the years, there has been a remarkable consensus on the existence of conscription in Sweden, and the question of why was seldom articulated (Dahlstr¨om & S¨oderberg, 2002, Sturfelt, 2014). However, Rehn (1999) inter-prets the strong support for conscription in view of the concept ”the hidden curriculum”.19 With support from both policymakers and conscripts, SAF introduced a new evaluation system during the late 1990s. The purpose was to stress the civil merits of service, and facilitate the assessment of the conscripts’ qualities by future civilian employers. In addition to evaluating military skills and knowledge, the conscripts’ ability to take initiatives, co-operation skills and independence etc., were evaluated and explicitly graded

18

Another important institution in the creation of the welfare state, also with an emphasis on equality and social mobility, was the compulsory schooling reform implemented in the 1950s and 1960s. The reform is shown to have increased earnings and decreased crime for men from lower SES-backgrounds (Meghir & Palme, 2005, Meghir et al., 2012). Hence, to the extent that policy makers actually managed to design the con-scription system according to similar ideas, one would expect similar effects.

19_{”The hidden curriculum” refers to what is learned in an educational context besides}

the learning objectives of the formal curriculum. One of the first studies explicitly discussing this phenomenon was Jacksson (1968) and the literature is vast. However, already Immanuel Kant mentioned the essence of the phenomenon: ”Children, for in-stance, are first sent to school, not so much with the object of their learning something, but rather that they may become used to sitting still and doing exactly as they are told” (Kant, 1803).

(39)

although not being in any formal military curriculum. Rehn (1999) argues that this can be understood as a public belief that these skills and abilities were learned during service.

Despite the declining number of individuals conscripted during the 1990s and 2000s, the idea of conscription was never challenged up until the late 2000s (Leander, 2005). In 2009, the parliament voted (with three votes in favour) to abandon conscription. Nevertheless, conscription was reintroduced in Sweden in 2018 and although policymakers’ justification was ”worsened in-ternational security”, many nonmilitary arguments described above were put forward by conscription advocates in the public debate.20

3.2. The Military and Conscription

From a strict military point of view, the purpose of military service was to train military units ready to be deployed in case of war. During peacetime, professional officers functioned as instructors of conscripts while functioning as commanders in the wartime organization. By the end of their service, conscripts were evaluated by their officers who graded them and decided whether the conscript fulfilled the requirements for being assigned a position in the wartime organization. If so, the conscript was placed in a specific position, in a specific unit in the wartime organization. Hence, in contrast to many other conscription systems, the active service consists of training making the conscript ready to occupy a position from which he is formally on leave after active service. In that way, a small country like Sweden could, during the Cold War, have a large wartime organization of about the same size as the military forces of Britain, France or West Germany (Agrell, 2010, SOU, 1984:71, ˚Aselius, 2005).

During the Cold War, almost all Swedish physically and mentally fit men were conscripted. The principle of conscripting and training the entire male cohort every year was initiated during WWII, and was strong throughout the Cold War. However, by the end of the Cold War, SAF began a struc-tural change and moved towards a much smaller but more technologically

20_{See SOU (2016:63). The report gives a background to the need for more military}

per-sonnel, proposes reintroduced conscription and provides an analysis of various conse-quences. Although the purpose of the report was not to investigate possible effects on nonmilitary outcomes and society at large, it does, however, comment on consequences for civilian employers: ”Civilian employers will most likely not be affected[...] The com-mittee’s assessment is however that civilian employers will benefit from the training and the personal development acquired during service”, (SOU, 2016:63, pp. 191-192).

(40)

advanced organization. Along with this structural change, in every accepted defence proposition during the 1990s and early 2000s, the parliament decided on major cut-backs. Consequently, the need for conscripts to fill positions in the wartime organization was reduced. Hence, the needs of SAF and not the size of each cohort would determine the number of conscripts.21 This together with the downsizing drastically reduced the number of conscripted during the 1990s and 2000s (Agrell, 2010). This is shown in Figure 1.

Figure 1: Trends of Individuals Tested, Conscripted, and Started Service A: No. tested, conscripted, and

started service 0 20000 40000 60000 1990 1995 2000 2005 Test Year

Started Service Tested Conscripted

B: Share (of tested) conscripted, and started service 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1990 1995 2000 2005 Test Year

Started Service Conscripted

Notes: A: Number of tested, conscripted (assigned a service category), started service by test year. B: Share conscripted and started of those who were tested. All physically and mentally fit men were required to do the enlistment tests. During the 1980s and early 1990s, 95% of each male cohort were tested (Hjalmarsson & Lindquist, 2019). During the period 2001-2004, between 85% and 90% were tested (TPVFS, 2004:1).

3.3. The Enlistment Procedure

Every year, The Swedish Tax Agency was obliged to send information to The National Service Administration (NSA) on which Swedish young men that are turning 17 years old. These individuals had to go through the en-listment process. NSA had the responsibility to test and enlist young men

21

A new law (SFS, 1994:1809), stated that the number of individuals conscripted each year should not exceed what is necessary for the wartime organization of SAF. A new service category, ”utbildningsreserven” (”training reserve”) was also introduced. Individuals fit for service but not needed in the wartime organization were placed in this category. Those individuals were never intended to start service except after a government decision.

(41)

to SAF in a service category (e.g. reconnaissance, rifleman, mechanic, radio-operator etc.) that best matched the individuals’ test results and preferences (if possible) with the needs of SAF (TPVFS, 2002:1). Each test-office was responsible for calling and testing individuals within their designated geo-graphic region and for conscripting individuals to units within this region. Thus, most conscripts were conscripted to units located in the same geo-graphical region as their domicile, but some service categories and units had conscripts from all over the country (HKV, 2001). The enlistment process consisted of several steps: First, the young man received a letter calling him to one of the five test-offices to which he geographically belonged. In general, NSA tried to assign the test-day in connection to the testees’ 18th birthday (Carlsson et al., 2015, TPVFS, 2004:1).22 Figure 21 (A) in the Appendix (Section D) shows the age distribution of testees in this study

Second, he had to fill out a health declaration together with other general questions on leisure activities etc., and bring this to the test-office on the test-day(s).23 NSA could decide not to call an individual if the individual had a medical certificate of a health status not suitable for service, for exam-ple. In such a case, the enlistment process ended.24 During the test day(s), the testee went through a number of rigorous tests, ranging from medical examination by a doctor, physical tests under supervision of a nurse, com-puterized cognitive tests and a 30-minute interview with a psychologist who assessed the testees ability to fulfill the psychological requirements of mil-itary service. Finally, the testee met an enlistment officer who, based on the test-results, decided to enlist the testee or not, and if so in what ser-vice category and unit. Enlistment officers used computers to help make the first match between a testee’s scores and a set of available service categories (TPVFS, 2004:1, SOU, 2004:5). The enlistment officer also noted the testee’s motivation towards service which could be used in the selection.

During the period studied in this paper, there were approximately 600 ser-vice categories. Serser-vice categories were also classified by the serser-vice level which was related to rank which, in turn, was also related to length of service. The levels were private, squad-level officer, platoon-level-officer and company-level-officer.25 Each service category had a well-defined description

22_{Those born in the last quarter were sometimes tested at the beginning of the year}

in which they turned 19. And, in cases when the future testee was expected to finish high-school later than what is usually the case, the test day might have been postponed (TPVFS, 2004:1).

23_{The test procedure lasted for 1-2 days.} 24

During the period 2001-2004, between 85% and 90% were tested.

(42)

together with specific test-result requirements. For example, reflecting the fact that conscripts assigned to higher levels had more responsibility and/or more complex tasks, the cognitive test was explicitly used when assigning a testee to different service levels. To be eligible for officer training, the testee had to score above a certain threshold on the cognitive test. The overall aim was that no conscript was to be underqualified or overqualified and although motivation was taken into account, it was only considered when two candi-dates were equally qualified in every aspect. At the end of the test-day(s), the testee had a preliminary decision on what service category and level he was to serve in, what unit he was to serve in, a preliminary start date and for how many days he would be in service.26 The enlistment procedure had now been completed.

In order to make it possible for NSA to assign the correct number of indi-viduals to relevant service categories and units, SAF were required to place a detailed order to NSA every year regarding the upcoming three years, typ-ically in December. Thus, SAF placed an order year t for the number of conscripts needed at each unit and service category year t + 1, t + 2 and t + 3, and the order was updated with minor changes every year (TPVFS, 2004:1, SOU, 2004:5). After receiving these orders, NSA planned its ac-tivities and started to conscript individuals for future service according to placed orders.27 As shown in Figure 21 (B) (Section D in the Appendix), most individuals were planned to serve within two years after the test year and a majority when they were 19 years old, i.e. in connection to their high school graduation. The enlistment procedure together with the hasty gov-ernment decision to downsize SAF gives rise to the empirical strategy of this paper.

in Section B in the Appendix.

26

The correct terminology is regiments, flotillas, schools or centres, but to simplify I will refer to these as units.

27

NSA has stressed the importance of getting accurate information on the needs of SAF well in advance to be able to properly plan its activities (see NSA (2004)).

(43)

4. Empirical Strategy

The government reform in 2004 was neither foreseen by the large group of individuals affected, nor by the NSA. A government bill (Prop., 2004/05:5) was handed over to the parliament on September 23, 2004. It concerned the downsizing of SAF by about one third and explicitly stated which units that were to be closed, or moved, together with a discussion of why. On December 16, 2004, the government bill was accepted by the parliament and soon after that, the government issued an appropriation letter concerning the directions and guidelines for the activities of SAF in the upcoming year (Regleringsbrev, 2005). It listed a large number of units, across branches and geographically spread across the country, that were to be closed, or moved, already on December 31, 2004.28 _{The closures were to be completed no later}

than June 30, 2006.

Most of these units trained conscripts every year and due to the closures no, or very few, entries could take place. Because of this hasty government decision and how the enlistment procedure was carried out, many individuals had already been tested and assigned to units that were to be closed. In order to minimize the damage, NSA decided to stop assigning individuals to units proposed to be closed on the day when the government bill was handed over (September 23, 2004). Nevertheless, despite this, NSA estimated that 4,500 (out of 16,000) individuals were planned to enter one of the units and immediately started the process of cancelling service calls in January 2005.29

The population is restricted to those tested and conscripted with the planned entry year 2003, 2004, 2005 and 2006.30 I also restrict the sample to those

28

A map with test-offices, their administrative regions, units pre-reform and units to be closed or moved is shown in Figure 20 in Section D in the Appendix.

29

NSA stated that the aim was to notify all individuals affected by the downsizing be-fore the end of March, well bebe-fore the application deadline to universities and other institutes of higher education (April 15). NSA reported that this was accomplished (NSA, 2004). Evidence that the closures were hasty and unforeseen by SAF (and thus by NSA as well as the conscripted individuals) can be seen in the documents from SAF Headquarters in Section E in the Appendix. If one compares the plan for the number of conscripts for the year of 2005 in the years 2002 (Figure 28), 2003 (Figure 29) and 2004 (Figure 30), a dramatic downward adjustment in the 2004 plan for the concerned units can be seen (written in December in connection with the acceptance of the government bill, Prop. (2004/05:5)).

30_{I do not include earlier cohorts due to the risk of including individuals affected by the}

downsizing followed by the government reform and the defence decision in the year of 2000.

(44)

conscripted before the government bill was handed over to the parliament (September 23, 2004).31 This is necessary since those conscripted after that date faced a different competition and were never at risk of being assigned to a to-be closed base.

To identify the individuals affected by the unit closures of the reform, I define a binary variable, Cancelled Position=1 if an individual was assigned to a (to-be) closed unit with a planned entry date in the year of 2005 or 2006, i.e.,32

1(Cancelled Pos.) = 1(Entry year ≥ 2005)·1(Asgd. to a closed unit). (1) Figure 2 shows the number and the shares of individuals affected by the reform per test year and test month. The majority of the affected individuals were tested during the year of 2004, but some individuals tested in 2003 were also affected. As seen in the figure, the probability of being affected by the reform also varies within years.33

Because the units were not closed by policymakers at random, it is rather likely that the reform is correlated with unit-specific characteristics. How-ever, the military context provides unusually good opportunities to control for these characteristics. Individuals were assigned to units solely based on information available in the data.34 _{Moreover, the downsize hit all}

branches, types of units and geographic regions. Nevertheless, to make the case stronger and reduce the risk of having an instrument correlated with un-observable characteristics, I identify unit fixed effects using several previous cohorts, in a difference-in-difference approach.35

Thus, the key identifying assumption is that the type of training (treat-ment) received, the characteristics of conscripts assigned to each unit, and

31

The final and formal decision of whom to conscript, in what category and where was taken about four times a year. The enlistment officers ranked and compared all testees from the previous quarter and assigned them to a service category, a unit and a start date (TPVFS, 2002:1, SOU, 2004:5). Most individuals are thus formally conscripted at a later date than the test-day, but for most conscripts the final decision confirmed the preliminary decision and the information the conscript received by the end of the test-day(s).

32

All units are listed in the government appropriation letter (Regleringsbrev, 2005). De-tails are given in Section F in the Appendix.

33

The same information but by test-office (geographic region) is shown in Figure 22 and Figure 23 in Section D in the Appendix. The figures show that all offices (i.e. geo-graphic regions) were affected by the reform.

34_{Except for any preferences, other than motivation towards service (which is available in}

the data), revealed during the interview with the enlistment officer.

Societal Impacts of Modern Conscription: Human Capital, Social Capital and Criminal Behaviour