Once the Republicans took the majority in Congress, they instructed the budget scorekeepers to do “dynamic scoring:” building macroeconomic feedback effects into their revenue estimates from proposed changes to tax policy. Because economists broadly assume that taxes distort decisions about activities with growth impacts, like labor supply or capital investment, cutting taxes often generates more growth in macro models. Since cutting taxes also loses revenue, a key motivation behind dynamic scoring is to make tax cuts look cheaper than they are under static scoring, which omits such feedback effects.
If this all makes you nervous, it should. There is a potential fistful of thumbs on that scale, from reckless tax cutters to questionable modeling assumptions to a level of uncertainty that should, IMHO, be left out of the official scores. There’s nothing wrong with asking the scorekeepers to give you a range of dynamic estimates, but given how unsure we are about the magnitude of such estimates, there’s something quite wrong with building them in to the scores thus and the future budget baseline.
Well, thanks to the Joint Committee on Taxation, we now have one of these dynamic scores firmly in hand. In this case, the JCT scored the cost of the “tax extenders,” that big bunch of preferential tax rules for businesses and individuals, like credits for research, energy, and investments that must be renewed every few years to stay alive (this proposal is for a two year extension).
Because they’re not assumed to continue, the cost of the extenders—the revenue lost to the Treasury—is not included in the budget baseline. And because there’s very little in the way of revenue offsets in the proposed extension, it adds $97 billion to the 10-year budget deficit.
At least, that’s the static (non-dynamic) estimate. The “addition effects resulting from macroeconomic analysis” amount to about $10 billion, offsetting the 10-year deficit number by 11%.
That’s not a huge effect and it certainly gets nowhere close to paying for the cost of the bill. If you were unhappy about unpaid-for tax cuts adding $97 billion to the 10-year deficit, I’m not sure you’re much assuaged by adding $87 billion instead.
But neither is it trivial, especially in a climate where legislators are so boxed in by competing priorities—no new taxes (tax cuts are, of course, encouraged), protect the defense budget, sequester caps, balance the budget—that every dollar of offset is worth its weight in gold.
But is it “correct?” That’s admittedly a high bar. Given the plethora of economic and political moving parts over a decade, no one knows whether the static score is right either. But dynamic scoring adds a whole other layer of uncertainty on top of that.
First, let’s be clear: no one’s accusing the able analysts at the JCT of cooking the books. They’re applying a model that they believe most accurately meets the demands of the Congress.
But there is something very important missing from the analysis: a range of estimates, based on plausible assumptions that differ from the ones they chose to highlight. For example, JCT decided to use the “high elasticity substitution” parameter for labor supply. That translates into more people working and thus more growth and revenue relative to the other, lower choice.
These future growth projections depend a lot on what the Federal Reserve is up to. JCT chose to simulate a “neutral Fed,” meaning the Fed doesn’t raise rates to offset the alleged stimulative fiscal boost from all these tax cuts over the next couple of years. That may be reasonable, especially given that most people expect the extenders to live up to their name (i.e., to be extended), so the Fed may well build such expectations into their forward thinking. On the other hand, the Yellen Fed’s been explicit about beginning to raise rates starting any month now, so this too is an important reason the dynamic estimate could be too high.
The most positive feedback effects here come from the increased near-term investment in capital stock engendered by certain tax cuts, like expensing and bonus depreciation. But while this is a standard assumption, it too is open to question in today’s economy. The cost of capital has been really low in historical terms, and yet firms’ recent investment record has been quite weak. Again, that doesn’t mean JCT is wrong. It means that our models are necessarily limited and while we should employ them to get a broad sense of how a tax change might play out, their predictive power re nuanced feedback effects is simply not up to the task of prudential budget planning.
Then there’s the long term. JCT notes that “in the second and third decade after enactment,” the bill raises the debt level. In these models, that leads to higher interest rates which would reduce the capital stock and thus lead to slower growth. JCT expects that the negative impact would be minimal relative to the size of the economy, but it’s a very large economy, and very bad stuff can still be small relative to the overall.
On the other hand, who knows what would happen that far down the road? But that’s kind of the point.
I recently wrote about the fact that if you do dynamic scoring the way I think you should, accounting for and reporting a range of results based on different assumptions about these macro feedback effects, you’d end up with a wide range:
“For example, a dynamic score of Rep. Dave Camp’s tax reform plan estimated that it would raise the level of GDP over ten years by between nothing (0.1%) and a lot (1.6%). Guess which end of that spectrum fans of his plan would choose to highlight?”
As per political outcomes, we’re stuck with dynamic scoring. Even with a Congressional majority that I trusted on fiscal issues, I’d be against the idea based on what I know of the models’ limitations. With the current group, I worry that they’ll try to make tax cuts look a lot cheaper than they really are. For that reason, I urge JCT and CBO to at least provide us with a range of estimates to supplement the single official dynamic estimate that Congress now requires. Let’s be as transparent as we can be about the uncertainties inherent in this exercise.