Start with some background. In boom times, federal spending on discretionary programs often climbs -- the government's flush, people are doing well, and they don't see the need for Washington to skimp on improving schools, health programs, infrastructure and the like. During lean times, the government often bows to political pressure and institutes austerity measures, reducing the growth in spending on these programs, freezing it, or even cutting them in real terms. That's why it's called discretionary spending: it's up to Congress to decide every year how much of it to do, and what to prioritize.
There's thus no clear way to project what future Congresses will choose to do, and therefore how much the government is likely to spend on this part of the budget for the next 10 years. Some analyses peg expected growth of discretionary spending to GDP growth, but that's a pretty high target in times like these, when, for instance, President Obama had already vowed to freeze discretionary spending for the next few years.
CBO took into account our austere circumstances and concluded a more realistic growth measure over the next decade is inflation. Relative to that projected growth, the debt deal, which locks in discretionary spending cuts, and then institutes near-term caps, saves $2.1 trillion. About $1 trillion of that comes from expected discretionary savings, and the rest will come from a Congressional deficit committee tasked with making up the difference, largely on the tax and entitlement side.
Enter S&P, which took CBO's conclusion -- $2.1 trillion -- and, according to Treasury, applied it, uncorrected, to the completely different, aforementioned baseline, assuming spending typically grows with GDP. Of course, if in absence of the debt limit deal, Congress would have increased discretionary spending at such a high rate, then the deal itself would actually cut vastly more than $2.1 trillion.
S&P didn't initially take that into account, didn't recalculate the expected savings, lopped $2.1 trillion off the wrong projection, and thus overestimated growth in future deficits by about $2 trillion, and debt as a percentage of GDP by about 8 percent.
Treasury turned S&P's original and corrected analysis into a graph, to illustrate the magnitude of the error.
S&P officials acknowledged the error, but rejected Treasury's assessment of its import, and revised their rationale for the downgrade to emphasize Congressional dysfunction, the growth of entitlement costs and GOP reluctance to raise taxes. David Beers, the head of the Standard & Poor's government debt rating unit, told ABC he has no regrets about the downgrade. But there's little difference between $2 trillion three weeks ago and $2 trillion now, and S&P hasn't yet explained why that arbitrary figure was so important in the run up to the deal, but relatively unimportant now.