Master’s Thesis Geography Human Geography THE ROLE OF UNEMPLOYMENT IN THE ARAB SPRING UPRISINGS Kari Paasonen 2016 Supervisors: Markku Löytönen University of Helsinki Halvard Buhaug Peace Research Institute Oslo UNIVERSITY OF HELSINKI DEPARTMENT OF GEOSCIENCES AND GEOGRAPHY GEOGRAPHY P.O BOX 64 (Gustaf Hällströmin katu 2) FI-00014 University of Helsinki HELSINGIN YLIOPISTO – HELSINGFORS UNIVERSITET – UNIVERSITY OF HELSINKI Tiedekunta/Osasto – Fakultet/Sektion – Faculty/Section Laitos –Institution –Department Faculty of Science Department of Geosciences and Geography Tekijä –Författare – Author KariPaasonen Työn nimi –Arbetets titel – Title The role of unemployment in the Arab Spring uprisings Oppiaine –Läroämne – Subject Human Geography Työn laji–Arbetets art – Level Aika–Datum –Month and year Sivumäärä – Sidoantal – Number of pages Master’s Thesis December 2016 180pp. + appendixes Tiivistelmä – Referat – Abstract It is often assumed that unemployment increases the probability of armed conflicts, riots and political instability in general. However, there is a considerable shortage of empirical research on this topic and generally the evidence available does not confirm the positive linkage between unemployment and political instability. Many have considered unemployment to be an important factor also behind the so-called Arab Spring. This thesis aims to analyse more closely what role unemployment played in the spark of these uprisings. From a theoretical viewpoint, relative deprivation theory and opportunity-cost approach can link unemployment to political instability. On the other hand several factors, stigmatisation and lack of collective resources, for example, are theorised to hinder the mobilisation of the unemployed. The approach of this thesis is quantitative and its principal method is regression analysis. Most of the analyses are conducted using existent and extensive survey data sets, most notably the third wave of the Arab Barometer released in late 2014. In addition, the thesis explores descriptively data about the trends and patterns of the unemployment rates, protests, and organised violence. The study area of the thesis covers those Arab countries, where this data is available. Tunisia, Egypt, Libya, and Yemen are studied more closely. The key finding of the thesis is that an unemployed respondent has not taken part in the protests of the Arab Spring more likely than an employed respondent. Several alternative operationalisations confirm this finding. Survey data also suggest that unemployed people are less satisfied with their lives and politically less active compared to the employed. Further, satisfaction with life does not affect an individual's likelihood of protest attendance, but a higher interest in politics is a strong predictor of protest participation. As it seems that dissatisfaction does not turn to protesting, this might explain the low participation of the unemployed. This thesis finds scantly support that a higher unemployment rate would lead to a higher number of protests or a higher level of organised violence in the North African Arab countries. These results do not inevitably signify that unemployment was an irrelevant factor in the eruption of the Arab Spring uprisings. However, results suggest that unemployment was not a major driving force behind the Arab Spring. More generally, the outcome of this thesis together with earlier empirical research around the world suggests, that unemployment does not increase the likelihood of political instability as forcefully as it is often proposed. Avainsanat – Nyckelord – Keywords Arab Spring, unemployment, political instability, armed conflicts, political geography, Arab countries Säilytyspaikka – Förvaringställe –Where deposited Helsinki University Library, Kumpula Campus Library Muita tietoja–Övriga uppgifter–Additional information HELSINGIN YLIOPISTO – HELSINGFORS UNIVERSITET – UNIVERSITY OF HELSINKI Tiedekunta/Osasto – Fakultet/Sektion – Faculty/Section Laitos –Institution –Department Matemaattis-luonnontieteellinentiedekunta Geotieteidenjamaantieteenlaitos Tekijä –Författare – Author KariPaasonen Työn nimi –Arbetets titel – Title Työttömyyden rooli arabikevään kansannousuissa Oppiaine –Läroämne – Subject Kulttuurimaantiede Työn laji–Arbetets art – Level Aika–Datum –Month and year Sivumäärä – Sidoantal – Number of pages Progradu-tutkielma joulukuu 2016 180s. + liitteet Tiivistelmä – Referat – Abstract Usein oletetaan, että työttömyys kasvattaa aseellisten konfliktien, mellakoiden ja myös muunlaisen poliittisen epävakaisuuden todennäköisyyttä. Asiaa on kuitenkin tutkittu empiirisesti huomattavan vähän ja yleisesti ottaen olemassa oleva näyttö ei osoita positiivista yhteyttä työttömyyden ja poliittisen epävakaisuuden välillä. Monet ovat pitäneet työttömyyttä myös merkittävänä tekijänä niin kutsutun arabikevään taustalla. Tämän tutkielman tavoitteena on tarkastella lähemmin, minkälainen rooli työttömyydellä oli näiden kansannousujen käynnistymisessä. Teoreettisesti työttömyys voidaan yhdistää poliittiseen epävakaisuuteen suhteellisen deprivaation tai vaihtoehtoiskustannusten näkökulmasta. Toisaalta useiden tekijöiden, esimerkiksi resurssien puutteen ja stigmatisaation, on teoreettisessa keskustelussa esitetty ehkäisevän työttömien aktivoitumista. Tutkielma on luonteeltaan kvantitatiivinen ja sen ensisijaisena tutkimusmenetelmänä on regerssioanalyysi. Tarkastelu perustuu pääosin laajoihin valmiina oleviin kyselyaineistoihin, joista keskeisin on vuoden 2014 lopulla julkaistu Arabibarometrin kolmas kierros. Lisäksi tutkielmassa tarkastellaan kuvailevasti dataa protestien, järjestäytyneen väkivallan ja työttömyysasteen vaihtelusta ja jakautumisesta. Tutkimusalue kattaa ne arabimaat, joista aineistoja on saatavilla. Tunisiaa, Egyptiä, Libyaa ja Jemeniä tarkastellaan lähemmin. Tutkielman keskeinen tulos on, että työtön kyselyn vastaaja ei ole osallistunut arabikevään protesteihin työssäkäyvää vastaajaa todennäköisemmin. Useat vaihtoehtoiset mittaustavat vahvistavat tämän tuloksen. Kyselyaineistojen perusteella näyttää myös siltä, että työttömät ovat tyytymättömämpiä elämäänsä ja poliittisesti passiivisempia työssäkäyviin verrattuina. Tyytyväisyys elämään ei kuitenkaan vaikuta siihen, kuinka todennäköisesti yksilö osallistuu protesteihin, poliittinen aktiivisuus puolestaan on vahva osallistumisen todennäköisyyttä lisäävä tekijä. Koska tyytymättömyys ei näytä johtavan protestointiin, tämä saattaa selittää työttömien vähäistä osallistumista. Tulokset tarjoavat niukasti tukea sille, että korkeampi työttömyysaste johtaisi korkeampaan protestien määrään tai korkeampaan järjestäytyneen väkivallan tasoon Pohjois-Afrikan arabimaissa. Tulokset eivät välttämättä tarkoita, että työttömyyden rooli arabikevään kansannousujen puhkeamisessa olisi ollut merkityksetön. Tulosten perusteella näyttää kuitenkin siltä, että työttömyys ei ollut keskeinen tekijä arabikevään alkamisen taustalla. Yleisemmällä tasolla tutkielman lopputulokset yhdessä aiempien eri puolilta maailmaa saatujen empiiristen tutkimustulosten kanssa viittaavat siihen, että työttömyys ei kasvata poliittisen epävakaisuuden todennäköisyyttä niin merkittävästi, kuin usein on esitetty. Avainsanat – Nyckelord – Keywords arabikevät,työttömyys, poliittinen epävakaisuus, aseelliset konfliktit, poliittinen maantiede, arabimaat Säilytyspaikka – Förvaringställe –Where deposited Helsinginyliopistonkirjasto,Kumpulankampuskirjasto Muita tietoja–Övriga uppgifter–Additional information TABLE OF CONTENTS List of tables, figures, and maps vii List of frequently used abbreviations ix 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2. Unemployment and the Arab Spring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.1. Arab world . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.2. Arab Spring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.3. Labour markets and unemployment in the Arab world . . . . . . . . . . . . . . . . . . . . 10 2.4. Unemployment as a driver behind the Arab Spring in the literature . . . . . . . . . . 13 3. Unemployment and political instability: theoretical background and earlier empirical research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.1. Theoretical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.1.1. Theoretical approaches supporting positive linkage of unemployment and political instability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.1.2. Theoretical approaches challenging positive linkage of unemployment and political instability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2. Empirical evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.2.1. Empirical research finding positive linkage between unemployment and protesting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.2.2. Empirical research finding negative or no linkage between unemployment and protesting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.2.3. Empirical research finding positive linkage between unemployment and organised violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.2.4. Empirical research finding negative or no linkage between unemployment and organised violence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.2.5. Summarising empirical evidence available. . . . . . . . . . . . . . . . . . . . . . . . . . 34 4. Research questions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.1. Research questions and hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.2. What new information could this thesis give us compared to the existent research?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 5. Research design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.1. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 iv 5.2. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 5.2.1. Data in research question I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 5.2.2. Data in research questions II, III, and IV . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 5.3. Study area. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 5.3.1. Study area in research question I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 5.3.2. Study area in research questions II, III, and IV. . . . . . . . . . . . . . . . . . . . . . . 53 5.4. Variables in research question I and their validity. . . . . . . . . . . . . . . . . . . . . . . . 54 5.4.1. Protesting and organised violence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.4.1.1. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.4.1.2. Validity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 5.4.1.3. Excluded events in hypotheses 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . . 57 5.4.2. Unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 5.4.3. Population density. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5.5. Variables in research questions II, III, and IV and their validity . . . . . . . . . . . . . 64 5.5.1. Protesting, main explained variable. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 5.5.2. Unemployment, main explanatory variable . . . . . . . . . . . . . . . . . . . . . . . . . 66 5.5.3. Being satisfied with life and politically active, variables serving both as explanatory and explained variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 5.5.4. Control variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5.5.5. Validity.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 5.6. Data reliability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 5.6.1. Data reliability in research question I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 5.6.2. Data reliability in research questions II, III, and IV . . . . . . . . . . . . . . . . . . . 80 6. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 6.1. Research question I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 6.1.1. Hypothesis 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 6.1.2. Hypothesis 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 6.2. Research question II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 6.2.1. Hypothesis 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 6.2.2. Hypothesis 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 6.2.3. Hypothesis 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 6.2.4. Results from other data sets and additional operationalisations . . . . . . . . . . 117 6.3. Research question III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 6.3.1. Hypothesis 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 6.3.2. Hypothesis 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 6.3.3. Results from other data sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 6.4. Research question IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 v 6.4.1. Hypothesis 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 6.4.2. Hypothesis 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 6.4.3. Results from other data sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 6.5. Control variables in research questions II, III, and IV . . . . . . . . . . . . . . . . . . . . . 130 7. Interpretation of the results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 7.1. Central results and possible explanations for them . . . . . . . . . . . . . . . . . . . . . . . 133 7.1.1. Research question I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 7.1.2. Research question II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 7.1.3. Research question III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 7.1.4. Research question IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 7.1.5. Explaining results of all research questions together . . . . . . . . . . . . . . . . . . 141 7.2. Reliability and validity of the results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 7.2.1. Research question I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 7.2.2. Research questions II, III, and IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 7.2.2.1. Time lag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 7.2.2.2. Validity of unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 7.2.3. Use of linear regression in research question III. . . . . . . . . . . . . . . . . . . . . . 151 7.3. Comparisons with earlier research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 8.1. What and how well did the thesis answer the question about the role of unemployment in the Arab Spring? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 8.2. What new knowledge has this thesis brought?. . . . . . . . . . . . . . . . . . . . . . . . . . . 159 8.3. Directions for future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 Appendixes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 vi LIST OF TABLES, FIGURES, AND MAPS Tables 1 Study area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 2 Share of unemployed respondents in ABIII by country and actual unemployment rates in the same countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 3 Correlation matrix for governorate level variables of unemployment, protests, fatalities of organised violence, and population, Pearson's r. . . . . . . . . . . . . . . . . . . 100 4 Binary logistic regression, protest participation of the unemployed (measured with the question of employment status) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 5 Binary logistic regression, protest participation of the unemployed (measured with the question of absolute income) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 6 Binary logistic regression, protest participation of the unemployed (measured with the question of relative income) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 7 Binary logistic regression, protest participation of the unemployed among the youth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 8 Binary logistic regression, protest participation of the unemployed among the educated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 16 9 Linear regression, satisfaction with life of the unemployed. . . . . . . . . . . . . . . . . . . 121 10 Linear regression, the influence of unemployment on being politically active (measured by interest in politics) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 11 Binary logistic regression, the influence of unemployment on being politically active (measured by voting) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 12 Binary logistic regression, protest participation of the dissatisfied . . . . . . . . . . . . . 126 13 Binary logistic regression, protest participation of those politically active (measured by interest in politics) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 14 Binary logistic regression, protest participation of those politically active (measured by voting). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Figures 1 Unemployment rates in the Arab world, the European Union, and globally 1991–2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2 Unemployment rate in 2010 in Arab League member states . . . . . . . . . . . . . . . . . . 12 3 Unemployment rates for Mauritania and Morocco 1991–2014 according to the ILO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 4 ABIII respondents by age and gender. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5 ABIII respondents’ answers to the question concerning satisfaction in government . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 4 6 Unemployment rate, protests, and fatalities caused by organised violence in Algeria 1991–2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 6 7 Unemployment rate, protests, and fatalities caused by organised violence in Egypt 1991–2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 vii 8 Unemployment rate, protests, and fatalities caused by organised violence in Morocco 1991–2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 9 Unemployment rate and protests in Tunisia 1991–2014. . . . . . . . . . . . . . . . . . . . . . 89 10 Unemployment rate and fatalities caused by organised violence in Yemen 1991–2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 11 Protest participation and employment (measured with the question of employment status) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 12 Protest participation and employment (measured with the question of absolute income) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 13 Protest participation and employment (measured with the question of relative income). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 14 Satisfaction with life among the unemployed and the employed . . . . . . . . . . . . . . . 120 15 Interest in politics among the unemployed and the employed . . . . . . . . . . . . . . . . . 122 16 Protest participation and satisfaction with life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 17 Protest participation and interest in politics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 18 Protest participation and country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 19 Protest participation and age. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 20 Protest participation and education level. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 21 Findings of research questions II, III, and IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Maps 1 Demands and fatalities of the Arab Spring uprisings in 2011. . . . . . . . . . . . . . . . . . 6 2 Study area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 3 Unemployment by governorate in Morocco in 1999 and 2004. . . . . . . . . . . . . . . . . 61 4 Population density in North African Arab countries in 2010 . . . . . . . . . . . . . . . . . . 94 5 Unemployment at governorate level and urban and rural unemployment rates in North African Arab countries before the Arab Spring uprisings . . . . . . . . . . . . . 95 6 Protests in North African Arab countries by governorate in 2006–2011 . . . . . . . . . 96 7 Fatalities caused by organised violence in North African Arab countries by governorate in 2006–2011. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 viii LIST OF FREQUENTLY USED ABBREVIATIONS AB Arab Barometer survey ABII Second wave of the Arab Barometer survey ABIII Third wave of the Arab Barometer survey ACLED Armed Conflict Location & Event Data ILO International Labour Organization UCDP GED Georeferenced Event Dataset of Uppsala Conflict Data Program SCAD Social Conflict in Africa Database WDI World Development Indicators data set WVS World Values Survey WVS6 Sixth wave of the World Values Survey ix 1. INTRODUCTION Unemployment is often seen as a factor considerably increasing the likelihood for various forms of political instability from rioting to terrorism and civil wars. This assumption is made both by analysts and politicians, both in relation to the developing and developed world (United Nations 2004: 54–70; World Bank 2011a: 72–96; International Labour Organization 2013a). For example, the two previous interior ministers of Finland have identified unemployment as a potential threat to security. According to them unemployment can or is likely to lead rise in violent extremism, criminality and terrorism (Eurooppatiedotus 2013; Orpo 2015). This thesis will focus on the specific case of the Arab Spring uprisings in 2011, but unemployment has been blamed for riots in the Arab world already before that (Sadiki 2000; Middle East Online 2008; Ottaway & Hamzawy 2011: 4–6). As it is assumed that unemployment causes political instability, this is then followed by an assumption that employment creation is a way to foster peace in fragile states and in conflict-affected areas. Indeed, according to Berman et al. the vast majority of aid money spent by governments to restrain political violence is aimed at hindering the recruitment of insurgents and fighters. The underlying logic is that those gainfully employed participate less probably in political violence and on the other hand, employment lessens grievances and this dampens the potential for political instability (Berman et al. 2011: 496–497). However, it has been pointed out that there is scant empirical evidence to confirm either this link between unemployment and political instability or the link between job creation and increased stability. In 2010 Cramer argued that “There is no remotely convincing evidence at the cross-country, large-N level, at the quantitative case study level, or at the ethnographic, ‘qualitative’ level, for any bold claims that unemployment is a mechanistic causal factor in violent conflicts in developing countries. The evidence on youth unemployment is even weaker” (Cramer 2010: 24). It seems that after 2010 not much has changed; the substantiation is still missing. As Holmes et al. gathered literature for their review on the impacts of employment creation on stability and poverty reduction in fragile states, they found close to 10 000 articles to assess. As they restricted their inclusion to studies providing primary quantitative or qualitative evidence about these issues, only seven studies out of these 10 000 met the criteria. Of these seven studies one examined the impacts of job creation on stability, others studied impacts on poverty. Consequently it is possibly not surprising that the authors write that the policy of employment creation as a mean to stabilise fragile states is primarily based on 1