The Wall Street Journal just had a great article on the pathetic job the Congressional Budget Office (CBO) has done with predicting costs and coverage from the Affordable Care Act. It’s about time someone points out the idiocy of these projections from Washington (not just the CBO). If you’ve had any statistics classes, you know the standard error for a prediction gets bigger the further away it is from your data. For a time-series model, the error as little as three periods into the future is so big that any predictions are practically meaningless.
I’ve never understood why we always get 10-year forecasts from every agency in Washington. I suspect the reason can be summed up by National Security Advisor Jeffrey Pelt in The Hunt for Red October: “I’m a politician, which means I’m a cheat and a liar, and when I’m not kissing babies I’m stealing their lollipops.” This raises the question: do lollipop makers take election years into account when forecasting future sales?
Anytime you see “10-year forecast,” you can replace that with “Once upon a time,” and you’ll be better prepared for what comes next.