8. Conclusions
8.1. What and how well did the thesis answer the question about the role of
Unemployment is often considered to be a forceful driver behind various forms of political instability, ranging from protests and riots to terrorism and armed struggles. Also in the case of the Arab Spring, unemployment is one of the most often mentioned factors which are seen to have stoked the uprisings. Still, the empirical research studying the relationship of
unemployment and political instability is scattered and its results provide scantl support for the assumed linkage.
It is important to study the significance of unemployment to political instability. Firstly, this is important per se to get better understanding of mechanisms behind rebellions.
Secondly, this is important to provide guidance for policies which aim at preventing political instability or fostering peace.
My thesis has focused on studying the role of unemployment in the Arab Spring
uprisings. Even though unemployment is repeatedly labelled as an important factor behind the Arab Spring, there are—as far as I know—no prior studies which would have this clearly focused on the significance of unemployment in the context of the Arab Spring.
If unemployment would be a forceful causal factor behind political instability, it could be assumed that a higher unemployment rate would lead to higher numbers of protests and a higher level of organised violence. More precisely this could signify that in areas where the unemployment rate is higher there is more political instability, that the Arab Spring uprisings were preceded by a rise in the unemployment rate, or that there are more protests and
organised violence in times of high unemployment rates and less political instability in times of lower unemployment rates. I was looking for these and some other possible indications in several North African Arab countries. In general the evidence to support the assumed
relationship between the unemployment rate and political instability remained weak. There are some concerns about data reliability, whose impacts on results are hard to estimate.
However, if the relationship between unemployment and political instability would be robust, I would assume that it would have appeared more clearly also in these analyses. It is possible that political instability is mainly determined by other factors and unemployment only has a minor influence. If this is the case, it is obvious why the effect of unemployment has not appeared in my analyses; my hypotheses were based on the assumption that unemployment
has a strong impact on political instability.
After comparing the unemployment rates and level of political instability I used survey data to test if the unemployed of the Arab countries are more active protesters than the
employed. I am much more confident with this survey data than with the data which I used to study the relationships between the unemployment rates and the appearance of protests and organised violence. The survey data is more extensive and its reliability and validity are superior. From now on I used regression models which included a set of control variables, for education, for gender, and for some other factors which previous studies have shown to have an impact. Thus I was here able to detect also if being unemployed had a lesser influence in protesting.
In short, I found very little support that unemployed people were more active protesters compared to the employed in relation to the Arab Spring protests or in relation to protesting in general. Generally, in those occasional models, where the results indicated significant
differences in protest attendance between the unemployed and the employed, the unemployed turned to be less likely to protest than the employed. This was the case especially in Egypt, Libya, Tunisia, and Yemen, countries hard hit by the Arab Spring, which we studied more closely. The finding that the unemployed are not more active protesters compared to the employed seems robust and can be considered as the key result of the thesis. Different operationalisations, different model specifications, and three different data sets suggest similar results. The results showed clearly and consistently, that on the other hand, being younger and being more educated are factors which throughout the models increase the likelihood of an individual to protest. Further, it appears, that the same socio-demographic groups have been active in the Arab Spring protests and in demonstrations more generally.
The results also hinted that there are differences between the Arab countries in how actively the unemployed have mobilised.
If only the examination of unemployment rates and political instability would have indicated a weak or no relationship, issues with data could have been blamed. But as the survey data indicate alike, that at the individual level the unemployed are not more active protesters, this cannot be explained with data quality. The results of survey data support the results from earlier analyses of the thesis.
Further analyses which were based on the same survey data sets revealed that the unemployed are significantly less satisfied with their lives and also somewhat less active in politics. It also appeared that being dissatisfied in life does not increase the likelihood of an individual to take part in protests, but being politically more active is clearly linked to higher
protest attendance at the individual level. Finding that it is not more likely for the less satisfied to protest contradicts explicitly with relative deprivation theory. This, however, provides one possible explanation for why the unemployed are not more active protesters in the first place and why the unemployment rate did not generally seem to be connected to political instability. The results suggest, that the unemployed are less satisfied, but this will not erupt as protesting due to their low interest in politics. This is apparently still a partial picture. Theoretical considerations provide also several other mechanisms to explain relationships between unemployment and political instability which are not tested here.
Possibly the unemployed do not mobilise because they do not have social networks, they do not want to identify themselves as unemployed, they are marginalised, they lack political skills and other resources, and they believe that protesting is unlikely to bring them jobs.
Taken together, although my results do not provide a complete picture, all the results of my thesis form a logical whole, where one part supports another.
The coefficients of determination in the regression models were overall low, which was not surprising taken into account how complex phenomena protest participation, satisfaction in life or political activity are to model. The explanatory power of the regression analyses does not lie in any single model or its coefficient of determination but instead in the abundance of survey data, in the number of the models and the hugely similar results they indicate.
Unemployment might have some role in the spark of the Arab Spring and in the protests and insurgencies of the Arab world more largely, but based on these results the role does not appear to be substantial. The question then emerges whether some other operationalisations or some other approaches would have led to different outcome. Maybe the results would reveal that the unemployed are still eager to take to the streets, if we would have observed how recently employment status of an individual has changed, if an individual has many
acquaintances who are unemployed, or if we would have studied the unemployment rate at the city or at the neighbourhood level. Perhaps the unemployed were protesting first and big masses followed? Or possibly we would have found a correlation with political instability by using the disaggregated unemployment rates for different ethnic groups. This is naturally possible, but also fully speculative.
Notably, the control variables all operated as we expected throughout the analyses.
Education and age were clear drivers of protest no matter if this was observed with regression analyses, chi-square tests or by charts. Even if an additional analysis would find that with some specific operationalisation unemployment predicts more protests or more political
instability, I would still think that its impact remains minor compared to age and education for example. In addition, results refuted the hypothesis of the dissatisfied being more eager to protest. Although this was not a premise for all theoretical approaches suggesting higher protest participation among the unemployed, this still partly shakes the basis to even expect a relationship between unemployment and protesting.
What can we finally say based on these results about the role of unemployment in the Arab Spring uprisings? We have studied whether the unemployed are active protesters and whether the levels of organised violence and protesting correlate with the unemployment rate spatially and temporally. But these analyses do not explicitly tell, whether unemployment was an influential factor. As discussed earlier, it might be that people are frustrated even if things have become better, and it is also possible that people who are not unemployed protest because of unemployment. Still, the results presented above are noteworthy. I guess many would agree that if unemployment did play a major role, it should have been better observable in these analyses. Thus, based on my thesis it seems that unemployment was not a major driving force behind the Arab Spring.