For what they’re worth and for anyone who doesn’t mind digging through the weeds, here are my comments on last week’s budget outlook from the Congressional Budget Office, which I previewed here.
In the baseline scenario that’s widely reported in the media (I’ll abbreviate it BS for this post), the CBO shows federal debt held by the public soaring from 76% of GDP at fiscal year-end in 2017 to 96% of GDP at the end of the ten-year forecast horizon in 2028.
The CBO also restored its alternative scenario (AS), which adjusts for certain constraints on what it’s legally allowed to include in the BS, making it more realistic than the BS. Unfortunately, the AS didn’t appear in annual reports between 2014 and 2017—in those years, the CBO highlighted areas where its BS projections were likely to be wrong but without producing an alternative. Even though it gets only a fraction of the attention given the BS, it’s good to see the AS back. It shows debt held by the public rising to 105% of GDP by the end of 2028, compared to the baseline’s 96% of GDP.
But the AS is also optimistic, partly because it uses the same rosy economic assumptions as the BS and partly because it includes less “emergency” spending than the BS.
As for the economic assumptions, the CBO’s unemployment rate projections are lower than they were last year in every projection year. They show an average unemployment rate for the next ten years of 4.4%, which is almost a third lower than the 6.2% historical average over the past forty years (see the chart below) and at least 0.4% lower than in any other ten-year projection the CBO has ever produced.
For perspective, consider that the Bureau of Labor Statistics shows only one historical ten-year period during which the average unemployment rate was as low as the CBO projects today. That one period was from 1948 to 1957 when the unemployment rate averaged 4.4% as in the CBO’s current projection, but it’s hard to imagine the average would have been that low had the Korean War not pushed unemployment firmly below 4% for 35 consecutive months.
For still more perspective, the CBO recently analyzed its forecasting record and discovered that its revenue projections for five years into the future averaged 1.7% of GDP more than actual outcomes after adjusting for tax law changes. That is, the CBO’s average revenue forecasting error was an overestimate of 1.7% of GDP. Why the huge errors? In the CBO’s words, they’re explained mostly by “the difficulty of predicting when economic downturns will occur.” Yet, the CBO followed that revelation by continuing to lower its unemployment rate projections—to all-time record lows in last week’s report—and by persisting with its practice of not including a genuine recession in its outlook. As I discussed last week, it includes only a tiny recession allowance, and I neglected to mention that the allowance doesn’t take effect until the forecast period’s back half.
As for emergency spending, I’ve suggested increasing the CBO’s spending projections by an average figure for bailouts, economic stimulus packages and supplemental appropriations, although adjusted for whatever spending in those three categories found its way into the particular year’s BS (by being appropriated in the short window between the start of the fiscal year and the compilation of the CBO’s annual outlook). I’ve suggested an “unexpected spending” allowance of about 0.5% of GDP, based on historical data beginning in 1980. But this year is different—with just over $100 billion in natural disaster funding having found its way into the BS, and $100 billion being about 0.5% of GDP, my adjustment isn’t necessary this year. In other words, the BS includes a typical amount of emergency spending, which in this instance happens to be skewed almost entirely to disaster relief. The problem, though, is in the AS…
For the AS, the CBO removed almost all of the current fiscal year’s $100 billion in emergency spending from future years, because it considers that spending atypical. That may seem reasonable on the surface, but the CBO defines emergency spending too narrowly. It doesn’t allow that the future will surely bring bailouts, economic stimulus packages and other types of unexpected appropriations that don’t meet its restrictive definition. As noted above, $100 billion for the 2018 fiscal year isn’t unusual—it’s about the average amount for unexpected appropriations in any given year, going all the way back to 1980.