I don't have the time atm to get into much detail (but will come back in 8ish hours after work), but very quickly:
- This is just a simplified description of probably the simplest macro-economic growth model you'll learn if you study economics. It's basically the model you learn in Macro 101. I didn't really want to write a super-long post going through all the extensions of the base model (not to mention the more complicated micro-economically founded concepts, models and theories you'd learn if you study developmental economics) as one of the predictions of the simple model that I thought relevant, that economies experience slower rates of growth as they reach capital saturation, is empirically supported.
- It's not intended to accurately reflect all aspects of a actual economy. No model is. If a model did this, it wouldn't be a model. We would probably just call it 'the economy'. It's just a simplified model from which a number of more complex extensions have been made, but I didn't feel were worth going into.
- It doesn't purport to make realistic assumptions, and no-one with any sense would think that to be the case. In addition to what you've pointed out, it also assumes the economy only produces a single good, has no government, no international trade, and constant returns to scale. All of these are almost certainly incorrect to varying degrees. Despite these simplifying assumptions, that does not mean the model is useless, or that it does not make any testable predictions. No-one in their right mind would think these assumptions are correct.
I'm honestly not sure what level of realism you expect an economic model to have. Just because a model is strictly wrong (as all models are; hence the name: model), does not mean it isn't useful. You mention empirical results from behavioural economics, which strongly suggest the neoclassical assumption of perfect actor rationality is incorrect. Again, this is just a simplifying assumption under which models can still yield accurate predictions in the large. Is Newton's law of gravitation completely correct? No, and it's been superseded by general relativity. However, is it still useful? Yes.
I can't really think of any economists (and this is certainly true for those I work with), let alone any people in general, who would believe any of these assumptions accurately reflect reality. There are a whole bunch of others too, like the implicit assumption that the labour-capital split of income remains constant. This is also almost certainly not correct as well.
It, and its extensions which account for 'human capital' (which has a multiplicative effect when combined with physical capital) which the simplified model just rolls into the solow residual (i.e. total factor productivity), do make a number of testable predictions, some of which have actually been tested in cross-country econometric studies. For example, here's a paper that examines a number of extended models, and among other things, examines the evidence for absolute vs conditional convergence (http://www2.stat.unibo.it/brasili/file/2009-2010/COSDI/chap1..., see pages 48 & 49).
You can literally just look at the model equation, do some simple algebra to turn one of the independent variables into the sole dependent variable and, presto, you have a prediction. Here are a few:
- An economy will converge to a balanced-growth equilibrium, regardless of its starting point. At this point, the growth of output per worker is determined solely by the rate of technological progress.
- At equilibrium, the capital/output ratio depends only on savings, growth, and depreciation rates.
- Countries with higher savings rates will accumulate capital at a faster rate (if all other relevant features of these economies are the same, which they aren't).
- If productivity is the same across countries (which it isn't), then countries with less capital per worker will have a higher marginal productivity of capital and provide a higher return on capital investment. Consequently, in a world of open market economies and global financial capital (which does not exist, and probably can't), investment will flow from rich countries to poor countries, until capital/worker and income/worker equalise across countries.
All that said, if you are able to produce a model that makes no incorrect assumptions, accurately accounts for bounded rationality, and all its predictions in every real world scenario are accurate, you'll win a nobel prize (and will also become filthy rich by investing according to this super-model's predictions).
Given us economist have simply just been 'doing it wrong' this whole time, because we foolishly didn't realise SF computer programmers held a much deeper understanding of macroeconomics, I'm really excited to being a first-hand witness to, what will be, a quantum leap in our understanding of macroeconomic dynamics and development economics.
I'll try to make sure to list every relevant caveat (the above is only a partial list, at best, for this particular model btw), before daring to discuss an unscientific subject like economics (which I apparently know nothing about).
EDIT: Oh fuck it. I quit.