Using dummy variable as my dependent variable I'm planning on fitting a model which finds the relation between conflicts that happen in a country (such as the Arab Spring) - my dependent variable - and other socioeconomic indicators such as income inequality, poverty, GDP growth ...
Is it possible to construct my dependent variable's data as 1 if a conflict occurred in a particular year, and 0 if not.
Will my model be meaningful if I use such data for my dependent variable?
 A: You can have 0/1 encoded dependent variable. That's exactly what is modeled in mortgage defaults, for instance. Each observation is the delinquency state of the loan. It can be 0 for current loans, and 1 those in default. 
Compare the likelihood function of logit to that of an ordinary linear regression. You don't need to worry about this too much since the stat package will deal with it, but it's good to know.
A: In my opinion, you can construct your response variable in that way, but I'm not sure about how do you could model it. If you could consider each year as an independent Bernoulli trial, your response variable would follow a binomial distribution, and then you could use a logistic regression to model the probability of the conflict to happen. Nevertheless, I’m not sure that you could consider each year as an independent Bernoulli trial. If a conflict occurred in a particular year, is it independent of having occurred the previous year, or the next year? I think it is not. Maybe you could use some kind of autorregressive model?
A: I have been thinking about this for a while, and I believe that it depends on a few factors which I will talk about.
Your study sounds very interesting! And I understand that you are trying to predict a whether conflict will occur or not, but what is your Outcome Window, or rather your specific objective? And how far are you looking back? For instance: 


*

*You could be trying to predict whether a conflict in Iran in the next 2 years will happen based on data you have of the past 5 years... You could have socio-economic predictors to number of conflicts in neighbouring countries in the past 1, 2, 3, or 5 years etc. In this case you can go ahead and use a logistic regression with ease since you will have dynamic variables in your model that can be updated. So yes, you can model your dependent variable as 1 or 0.


Questions with regards to outcome window, and observation period (it would be 5 years if you are looking at data from the last 5 years), etc, will help you specify your objective and in turn lead you to choose the right modelling technique.
Hope this helped!
