Why the CBO is Frequently Wrong

 During the hectic, unreal days of Watergate, Congress decided it needed a trusty sidekick to wrangle the federal budget.  Thus, the Congressional Budget Office (CBO) was born on July 12, via the Congressional Budget and Impoundment Control Act.  Why? Well, President Nixon was playing fast and loose with funds, "impounding" money Congress had earmarked, causing a Capitol Hill kerfuffle…Congress wanted its wallet back!

The “how” was simple: lawmakers crafted a nonpartisan agency to give them the straight dope on budgets and economic forecasts, that was to be free from executive spin.  Modeled after California’s budget wizards, the CBO became Congress’s go-to for number-crunching. 

The CBO’s mission?  To arm Congress with clear, unbiased data for smarter spending decisions, without whispering policy advice. It’s like a financial crystal ball, helping lawmakers navigate the fiscal fog with confidence.  And so, the CBO was supposed to be a nonpartisan Capitol Hill hero, quietly crunching numbers to keep the budget battles light and bright while giving the voters the truth about pending legislation—without political spin.

Unfortunately, it hasn’t always worked out that way.

The CBO tries very hard to be nonpartisan, and there is no better proof that it generally is than the fact that each political party in turn has complained long and loudly about bias whenever its ox is gored but staunchly defends the office when the CBO’s projections ding the opposing party.  Even if we assume the CBO operates without appreciable bias—as I believe it does —that still doesn’t mean that its estimates are to be believed.

The problem is this:  from its inception, the CBO has created financial models based on static forecasting (sometimes called, “static scoring”).  So, what is “static forecasting”?  Let me give you a real-world example:

Back in 1990, Congress passed a luxury tax on high-dollar items that typically were only bought by the wealthy:  large yachts costing over $100,000, private jets, and very expensive cars.  Congress wanted to raise tax revenue without increasing taxes on either the middle class or the poor.  The bill, the Excise Tax on Luxury Goods was part of the Omnibus Revenue Reconciliation Act of 1990.  The CBO estimated that the tax on yachts alone would bring additional tax revenue of over $120 million.

At the time, American shipyards sold thousands of yachts annually, so the CBO just estimated that 10% of the existing revenue would be paid to the government annually.  Of course, nothing like that happened.

The global yacht business was competitive and when American-made yachts suddenly went up in price, customers immediately changed their buying behavior.  Since they were just rich and not stupid, consumers bought used yachts (which were not subject to the new tax) or they went overseas—particularly to Europe—to purchase their luxury boats.   New yacht sales in the United States dropped by more than 70%, causing many small and medium boat builders—especially in Florida, Maine, and the Pacific Northwest—to close or drastically downsize.  An estimated 7,600 to 9,000 jobs were lost in the yacht-building industry alone, with ancillary industries (marinas, suppliers, etc.) losing many thousands more.

Well, at least the government brought in more revenue, right?

Nope!  The total amount collected on the sale of yachts in the first year was $12.7 million—a figure less than the government spent enforcing and collecting the new tax.  The Government Accounting Office (GAO) said this was primarily because the IRS had to track complex sales of high-end goods and pursue evasions, such as shell companies buying boats offshore.

It won’t surprise you that these dismal results were duplicated on all the luxury goods that were mentioned in the new law.  Whether it was jewelry, expensive furs, private jets, or the latest Ferrari, the rich altered their purchase patterns in ways not predicted by static modeling.  A Congressman summed up the failure, “We intended to tax the millionaires, but we actually ended up punishing the middle-class boat builders.” 

Congress eventually repealed the law, but not before the disruption had long-lasting effects.  Some economists argue that the yacht industry still has not completely recovered, thirty-five years later!

Since static forecasting does not take into account the behavioral response of the taxpayers to new taxation, what is really needed is dynamic forecasting.

Static models are simpler: raise taxes, get more money.  But dynamic models say: “Wait—people might work less, invest differently, or shift money offshore!”

Dynamic modeling tries to account for how people, businesses, and markets adapt—sometimes in weird, unexpected ways.  It’s like trying to predict not just how the pool table balls move after the break, but also how the table might warp mid-game, or your pet cat might jump up on the table and attack the ball.

Why doesn’t the CBO just always use dynamic models?  Because they’re wildly complex.  To do it right, you need to make hundreds of assumptions—about interest rates, about labor markets, about consumer psychology, about unicorn migration patterns (okay, maybe not about that last one, but you get the picture).  And if Congress doesn’t like the answer?  Suddenly your assumptions are under fire and your budget is cut. 

So, the CBO sticks mostly to static scoring—not because they think it’s perfect, but because it’s safer, more transparent, and less politically explosive.  After all, nobody in the CBO wants to explain to a Senate committee on live television that the entire economic forecast turned into crap because millions of Americans suddenly went out and bought pet rocks.

We could—but we won’t—change the way the CBO makes economic forecasts.

Imagine the Congressional Budget Office (CBO) as a weather forecaster for the federal budget.  If Congress passes a tax cut, for example, the CBO could use dynamic scoring to say, “Here’s what might happen—if everything goes great, if things go as usual, or if the economy trips on a rake.”

Instead of a static forecast of the tax cut, they give three dynamic forecast scenarios:

·      High projection: Everyone starts businesses, hires like crazy, and the GDP soars.  Tax revenues fall at first, but the booming economy makes up for it.  Unicorns dance, it rains in Death Valley, and the Seattle Mariners win the World Series.  It’s the “rosy scenario.”

·      Medium projection: The tax cut boosts spending and work a bit, but not dramatically.  The economy grows, but mostly as expected.  This is the Goldilocks forecast—not too hot, not too cold.  Both proponents and opponents of the tax cut claim to be correct.

·      Low projection: People pocket the extra cash, but don’t do much with it.  The economy doesn’t pick up, and tax revenues drop more than hoped.  Cue the budget shortfall and awkward committee hearings while the talking heads with perfect hair speak confidently on television about the failure of “trickle-down economics.”

Dynamic scoring would let the CBO show how real people might react to policy changes—rather than just assuming that everyone keeps behaving the same as always.  Unfortunately, predicting human behavior is a tricky business, which is one reason the CBO avoids dynamic scoring.  The other reason—and the main reason—is that Congress doesn’t want to be in the position of explaining economics to voters who want simpler answers.  Lawmakers prefer a clean, single “score”—even if it’s less nuanced.

Or to put that more succinctly, Congress wants answers that can fit on a bumper sticker.  Even if they are wrong.