Thought you would like to see this chart prepared by the Congressional Budget Office. It shows the accuracy of their previous budget deficit projections in the form of a confidence interval around the forecast. Note that they only show the confidence interval up to four years out.
In the document, they show that the confidence interval increases by about one percent of GDP ($120 billion) each time they add another year to their forecast horizon. In year 4, the confidence interval is about 5% of GDP ($600B) for that year’s projection alone. The detail shows that errors are also autocorrelated, a fancy term that means that initial errors tend to build up and get worse over time. Not good news for cumulative forecasts–exactly the ones CBO has to make for their Congressional masters to use on C-SPAN.
I have taken their numbers and estimated the error band around the 10 year projection. According to the CBO, the cumulative budget deficit over the next ten years is going to be about $2 trillion, plus or minus $13 trillion!
I’m not whining about the CBO; they certainly do the job better than I could. My point is that cumulative budget projection is a bad number. We should not use then to decide anything.