Since it is political “silly season”, and every politician under the sun is broadcasting plans for programs designed to catch the voter’s eye, we thought it was worth looking at how accurate cost forecasts of these programs are. Large projects such as TARP or the Affordable Care Act represent significant outlays by taxpayers, affecting government budgets for a decade or more. We wanted to know whether what we hear about their cost ahead of time is right, or not.
The majority of the work for this project was performed by our excellent intern, Jacob Godshalk, this last summer. The task we set to Jacob was to research the accuracy of the government’s own forecasting arm, the Congressional Budget Office. Jacob found that the CBO is adequately accurate when the forecast time frame was short. For instance, for revenue projections for the next one to two years, the CBO tended to forecast about 1% to 2% too high versus reality. So tax revenues are somewhat chronically falling short of forecasts, but not by much. Of course, over time, even 1% compounds to real money.
Over more intermediate time frames – five or six years – accuracy decayed. The CBO was generally five or six percent too high on revenue projections, though it managed to precisely forecast the cost of the ACA by 2015 in 2010.
Still, we were most curious about the huge numbers we sometimes hear out of the CBO for larger programs with muti-decade impacts. For instance, for the ACA, the longer term forecast was abysmal. The ACA cost over $30 billion MORE than the CBO had estimated, by the time 2016 rolled around, a forecasting error of 23%.
In the 1990’s, many Savings & Loan companies received a bailout. The CBO forecast that the bailouts would cost $120 billion, give or take a few billion. Instead, the total cost was $480 billion. More recently, TARP was expected to cost about $430 billion. Instead, it cost the government nothing; in fact, a net profit of over $15 billion (after program costs) flowed to Treasury.
Furthermore, the CBO is no better than anyone else at predicting a recession. Its forecast for government revenue in 2009, a recession year, was a hefty overestimation of 25%!
So next time you read those gasping articles containing CBO cost projections, remember that their numbers are not facts. Instead, they are guesses, perhaps worse guesses than you yourself might make, and most often, guesses that put the taxpayer at a disadvantage. Chronically overestimating revenue costs us all in the long run, and underestimating program costs is even more deleterious.