In providing[1]useful guidance on how to think about the various classes of macroeconomic tools, Olivier Blanchard, Professor Emeritus of MIT's economics department, makes a very important point: different tasks require different tools.
On forecasting tools, he notes that their straightforward purpose makes evaluating their quality obvious: we ask how well they perform in practice, out of sample. This type of evaluation falls to standard statistical criteria like the bias-variance trade-off or forecast errors based on backtesting, and it is a fundamental part of EA's approach. In enhancing our models last year based on the results of our backtesting, we have continued to hone our forecasting tools to their purpose.