(Original post by rochey)
I think the question you have to ask yourself is really, what is your particular point of view on economic methodology? If you're curious about critiques of neoclassical economics and want an all-round development economics training, then I'd go to SOAS (that's why I'm going!). On the other hand, if you very much consider yourself a mainstream economist who wants to apply standard mathematical anaytical tools to a focus on developing economies, and be as quantitative as possible, then Nottingham might be better.
Honestly, I do not like these "Mainstream vs. the bravehearted other" discussions. I want to do science! I spoke to a few development and trade economists who are in a high research position and nobody acutally believed in neoclassical models. The criticism to neoclassical world is obvious, but the crucial point is how we adress it!
We could say: "Ok, look, neoclassical models assume blablabla, so this assumption never holds, so we do not need to take care about neoclassical models". Or we could say: "Ok, let us investigate what happens if we relax this and that assumption". If you read some papers, for instance from the "post-autistic economists", the choose the first way of adressing neoclassical models. This is no science.
I am used to the other way. At my university, we look at different models of behaviour, for instance expected utility (which could be named as "classical but rejected by evidence" aproach), prospect theory, rank-dependent models, PT of the third generation and so on. We model altruistic preferences in order to investigate what happens in a neoclassical model if preferences are not only self-regarding. We compute and simulate highly complex agent-based models in different markets in order to investigate what happens, if we assume heterogeniety among agents. We also investigate what happens in standard trade models if firms are different. All this we do within mathematical models.
So, please tell me, what is mainstream economics? Do you mean the standard undergraduate textbook examples by Mankiw? Or do you mean the modeling of systematic deviation from rational behaviour as is done frequently in recent research papers?
I fear that, at SOAS, I would get "forget about neoclassical models". But I would like to get in touch with formal models that approach reality and I believe that it is worth to formulate models mathematically. That is why I decided to go elsewhere.
Additionally, there are two points where I would like to disagree.
Regarding applications to other universities for PhD programmes, it is of utmost importance who writes your LORs, at least if you try to reach the good departments.
Regarding the quantitative amount of work at SOAS, they state that they do not require formal proofs or that like. However, my research experience is that is favorable to know the exact mathematical formulation of some statistical tools in order to evaluate critically what has been done in some paper. I think knowing issues about weak instruments, Granger causality tests (you should look for a paper named "Chickens, Eggs and causality, or which came first by Thurman and Fisher, its funny to read) or Bayesian statistical methods (they are far better in small samples than the normal ML estimators, which is indeed a problem in development economics) is also very important, especially if your policy recommendations (that might be derived from empirical evidence) affect the life of thousands!