Off the Mark. How can CBO miss its projections so quickly?

NRO ^ | December 12, 2003, 9:01 a.m. | By Rep. Mark Steven Kirk (R., Ill.)

[.cnI redruM]

More than a century ago, British Prime Minister Benjamin Disraeli observed, "What we anticipate seldom occurs; what we least expected generally happens." Disraeli's maxim has become a mantra for current federal spending. Every year, Congress depends on the Congressional Budget Office (CBO) to provide forecasts that will help the House and Senate set federal budget priorities for the current fiscal year and beyond. Six months later, when Congress is working on legislation that reflects these new priorities, CBO tells Congress that its forecasts were wrong. This is not acceptable.

Forecasters justifiably point out that their numbers cannot be expected to reflect policy changes enacted later in the year. However, even after adjusting for legislative changes, such as the Bush tax cuts, CBO's one-year estimates have been seriously off the mark. Sometimes legislative cash flow has actually improved CBO's accuracy. The catch-all classification known as "economic and technical variables" cause surplus and deficit projections to miss their marks.

How can CBO miss its projections so quickly? From the perspective of overall economic performance, variables used to predict the performance of the economy can be too high or too low. Since 1999, CBO has released an annual study that compares its economic assumptions with those published by other prominent forecasters. Despite this work, its projections have not improved. The answer goes beyond an examination of forecasting variables.

From the perspective of government revenues and expenditures, analysts can misjudge the impact of tax-policy changes or the cost of new subsidies. A change in tax policy influences the decisions that individuals and businesses make — thereby affecting federal revenues. Similarly, budget projections that fail to take into account what will happen when Congress subsidizes something will also be off, sometimes badly. Medicare can provide many examples; the costs of a renal-dialysis benefit is one case. Congress added this benefit to Medicare in 1972, depending on forecasts that predicted program enrollment would level out at about 90,000 patients by 1995. By 2005, Medicare actuaries expect enrollment to exceed 400,000.

Whether we are talking about correctly measuring taxpayer responses to tax-rate reductions, or increases or beneficiary responses to new subsidies, Congress needs real-world estimates that account for the interaction between federal taxes, federal programs, and individual behavior. This year, in a first step toward incorporating real-world feedback into budget projections, CBO included alternative analysis that predicted a range of economic responses to government actions. More work needs to be done.

Ultimately, accurate forecasts will come from stricter accountability for past mistakes. Every year, CBO should complete a "look back" review of the previous fiscal year, analyzing the variance between its forecasts and actual budget results, with the goal of isolating which forecasting variables cause the greatest error. It must also identify the technical factors that consistently contribute to forecasting inaccuracies. Most importantly, this review must include recommendations on how CBO plans to improve forecasting accuracy in the future.

A push for better budget forecasts is more than a "good-government" initiative. The credibility of the projections Congress uses to do its business is on the line. According to the New York Times's September 14 article, "Dizzying Dive to Red Ink Poses Stark Choices for Washington," from the late 1990s and through 2001, CBO's long-term surplus estimates were the result of ebullient "forecasting error, parallel to the ‘irrational exuberance' the Federal Reserve chairman Alan Greenspan saw among investors during the 1990's" that Washington used to justify "a major change in tax policy." The experts find many reasons why past budget projections have been wrong. Who can say that the current ones are any better?

Certainly, CBO's long-term forecasting track record needs improvement. Examination of cumulative five-year forecasts completed under the oversight of four different directors shows a pattern to its long-term error. When the economy is headed into a cyclical upswing, CBO's five-year projections overstate the total multi-year surpluses and deficits. On the other hand, when the economy is poised toward recession, accumulated deficits will exceed forecasts. CBO itself pointed out its failure to miss economic turning points in "A Supplement to The Budget and Economic Outlook: Fiscal Years 2003-2012": "CBO's forecasts of real growth over the long run have alternated between periods of optimism and pessimism. The five-year forecasts produced during the late 1970s turned out to be too optimistic, averaging about 2 percentage points a year too high. Forecasts from the early 1980s, in contrast, were too pessimistic by a little less than half a percentage point. Forecasts from the late 1980s were overly optimistic again, but the projections made between 1992 and 1996 have been too pessimistic by more than a percentage point on average."

In July, the National Bureau of Economic Research announced that the U.S. economy was on a recovery path out of the recession that had begun in March 2001. Economic growth jumped to over 3 percent during the first half of 2003, and soared to above 8 percent in the third quarter — thanks to the economic stimulus provided by the Bush tax cuts. If CBO has continued its 20-year practice of missing turning points, it is overstating the cumulative deficits we should expect during the next five years.

This is good news indeed, but no sign that Congress should abandon its commitment to control discretionary spending. Growth-centered policies have shortened the severity of our last recession. However, the urgent obligations our country faces in its war against global terrorism gives Congress a renewed mandate to spend taxpayer dollars wisely. We need accurate budget projections to do this job.

[.cnI redruM]

Interesting from the standpoint of mathematical modeling.

1) When forced to extrapolate values beyond the known range of the function in question, you are going to stink. There are more eloquent mathematical formulations thereof, but that gets the point across.

>>>>>>>>>The five-year forecasts produced during the late 1970s turned out to be too optimistic, averaging about 2 percentage points a year too high. Forecasts from the early 1980s, in contrast, were too pessimistic by a little less than half a percentage point. Forecasts from the late 1980s were overly optimistic again, but the projections made between 1992 and 1996 have been too pessimistic by more than a percentage point on average."

2) The CBO has clearly failed to develop adequate proxies for estimating future variables. In other words, we've had pretty much the same set of monetary variables, within almost the same set of preconditions for going on 65 years now. We still have no clue how to model these variables.

At least the IPCC has a reputation for lying. We know that when the UN tells us that they have proof that the the Earth is warming 1.4-5 degrees Centigrade within a range of plus or minus 10 degrees, we can turn around and go back to watching the boob tube. Someone needs to either straighten the CBO out, or convince our economic policy makers to take them about as seriously as they would take The Delphic Oracle.

[farmfriend]

bttt