Perils of Predicting Future Budget Surpluses
A commenter asks how seriously we should take 10-year budget forecasts. I think the answer is that (1) the forecasts are generally honest though (2) they aren't very accurate (can easily be off by trillions); but (3) conclusions about where the surplus went are still fairly solid and (4) the forecasts are a good way to compare different policy proposals.
On honesty: I'm inclined to believe that the CBO (Congressional) and OMB (White House) forecasters are professionals who are fairly insulated from politics. Fiscal expert Alan Auerbach
[102k pdf] has found that both CBO and OMB forecasts, though often wrong, are unbiased, or at least enough so that any systematic biases can't be distinguished from mistakes.
Auerbach writes that errors can be divided into three categories: policy (e.g. unforeseen spending and tax changes), whether the economy booms or busts, and "technical" errors.
Technical revisions [i.e. errors] are residual, containing changes that the agency attributes neither to policy nor to macroeconomic changes. For example, a technical revision in revenue would result from a change in the rate of tax evasion, a shift in the composition of capital income from dividends to capital gains, or a change in the distribution of income. In general, 10-year budget forecasts depend on 3 things: the economy, policy, and tax revenue given the economy and policy.
Forecasting the economy, I think, is the smallest problem. It's hard to know exactly when a recession will hit, but over 10 years, errors tend to cancel out, and it's fairly easy to forecast average growth. According to the study I discussed yesterday, $700 billion of the $5 trillion surplus disappeared because of the bad economy. This is a fairly small number compared to "technical" errors and tax cuts, despite the fact that the economy has been shockingly slow to recover from the recession.
"Technical" errors seem to be a pretty big problem. Over half of the surplus proved to be illusory. Perhaps tax evasion increased, perhaps the falling stock market reduced capital gains, perhaps income of the very rich fell. The CBO's explanation
for the error speculates about these things, but ultimately concludes that they're not sure what happened.
Policy changes make a big difference, but as long as the analyst is upfront about what assumptions are made, these aren't really errors. The budget numbers I discussed yesterday originally came from the CBO. The CBO is legally constrained to make the silly assumption that current law will continue into the future, even if it is very likely that new laws will be passed. For example, no one really believes that the Bush tax cuts will expire in the bizarre way that's currently scheduled.
The discussion of the disappearing surplus
from yesterday was put out by the Center on Budget and Policy Priorities, the Concord Coalition, and the Committee for Economic Development. These groups are, respectively, center-left, centrist, and center-right proponents of a balanced budget.
That study started with CBO numbers and made different policy assumptions. For example, the baseline CBO projections assume that the tax cuts expire at various points in the future. The study assumed that they all become permanent (as Bush has proposed) and that "alternative minimum tax relief" passes. The CBO projections follow Bush's budget in projecting little increase in defense spending. The study says that Bush's defense plan (put out by the Dept. of Defense) calls for substantial increase in military spending, and assumes that this is what will happen. Whether the tax cuts become permanent and military spending soars is obviously unpredictable -- it depends on who is elected this year and in 2008 too. But this isn't really an error: the study is warning us what will happen if Bush gets his way, it's not really predicting that Bush is going to get his way.
Similarly, if we started with the CBO numbers and conducted a similar analysis of the Kerry plan, we'd have a valid comparison. Suppose we found that Bush's plan will lead to a $4.4 trillion deficit (as the CBPP/CC/CED found) but Kerry's will lead to a $2 trillion deficit (pulling a figure out of a hat for the sake of concreteness). It might turn out that the CBO made, say, $1 trillion in forecast errors. But the error would be the same in both the Bush and Kerry forecast, so it wouldn't affect the conclusion that Bush's deficit is $2.4 trillion bigger than Kerry's.