A major part of the problem is that the thing that macro-economists study, the economy of the world, changes its behavior in response to the reported observations of the economists.
It's worse than that, at least according to Austrian economics. For them, the theoretical basis of macro-economics, or at least of almost all of them, namely, the assumption what money (as a measurement unit) is what matters, is simply false. So, all the models starting from it will suffer no matter what.
The main criticism Austrian economists throw at macro-economics, econometrics etc. is that money isn't an objective unit, it's just a representation of subjective preferences that very from person to person, and within a single person from instant to instant, only resembling something minimally solid, and thus as a fake unit, as a kind of surface effect. Thus, the same way that a sociology divorced of any concept of human psychology is bogus, so is any macro-economics that doesn't base itself entirely on that small subset of human psychology that is micro-economics. True economics is at best a sub-field of psychology, and macro economics is at most a sub-field of sociology.
Austrian economists only do math after they manages to understood reasonably well the psychological mechanisms behind a set of atomic exchanges (not necessarily involving money), provided it shows itself as something that can have calculations done, which most often than not isn't the case.
Mainstream economics doesn't like this, at all. Keynesians, marxists, econometrists etc. all believe they have a unit of measurement, and that they can turn this unit and its measurement into a hard science. They cannot. It's wishful thinking, if not outright bullshit. But it's a bullshit so full of technobabble, so enchanting in its seeming seriousness, that the self-deception simply proceeds, unchecked.
Now, that doesn't means Austrian economics isn't full of bullshit too. It's own model of human psychology they call praxeology is extremely flawed, since it's based on philosophical assumptions more than on actual psychological research. Much of it is quite useful, or at least inspiring, but non-scientific anyway. But what they do very well, and the reason I keep reading them, is their debunking of mainstream economics pseudoscientific assumptions. They are at the top of their game when they take a macro-economics equation, break it down in its main components, and proceed to show analytically how what it describes is pure, glorified nonsense.
So, here's what must be done to really turn economics into the mostly scientific discipline it can ever be: take Austrian criticism of macro-economics, add the state of the art in cognitive sciences, develop an actually valid psychology-based micro-economics from both (it won't be Austrian's praxeology), and then, by way of reductionist thinking, build from then, step by step and floor-by-floor, a bottom-up macro-economics that's based on something actually relevant and universally valid (which "money" most definitely isn't).
Then, and only then, at least a semblance of actually working models.