What's the most "scientific" part of economics? i.e. has precise quantitative predictions that actually work out in the data? As far as I can tell it is lab behavioral economics. Does anything else come anywhere close?