Econometrics should include error bars in their statistics.

In the IMF Financial Programming and Policies MOOCs (Parts One and Two), we were taught to make econometric predictions of GDP and other measures, using their relationship to GDP to predict future trends. Thus GDP's error bars are multiplied as GDP projections are used in projecting other variables (effective tax, national debt, interest rates, unemployment, etc.). To be intellectually honest, the IMF should report confidence intervals on its projections; but they confidently report one figure, as if it represents the 100% correct prediction.

Economic models ignore the perverse motivations of individual actors in markets, serving to make pricing arbitrary and chaotic, prejudicial and labile, rather than efficient and as close to fundamental value as you can get.