So I've been thinking a lot about how the Joint Committee on Taxation projects its excise tax revenue, and the more I think about it, the more I realize just how much of the revenue they're projecting comes from a pretty elaborate dynamic effect on the labor market.

I got some clarity from sources in the know about how the CBO calculated that somewhat cryptic "Other Effects on Tax Revenues and Outlays" that generates $83 billion to defray the costs of the program over ten years.  Basically, the CBO's assumption is that the attractiveness of the subsidies will encourage several million people to drop their employer insurance and buy subsidized insurance on the exchange.  Their employers will pay them more money instead, and this money will be taxed.

Similarly, the Joint Committee on Taxation, which calculated the revenue from the excise tax, assumed that almost three quarters of the benefit would come from employers driving down their insurance costs, passing on the savings to their workers as compensation, and that compensation getting taxed.  And because the pressure to cut costs is expected to gather steam over time, the effect is larger in the out years--by 2019, the JCT predicts $46.3 billlion worth of revenue, but only $7.9 billion of it comes from people actually paying the tax.  The rest is dynamic effects.

The CBO projects $180 billion in "gross cost" of the coverage expansion--i.e., what the government will spend on Medicaid and SCHIP expansions, operating the exchanges, and subsidizing the purchase of standardized policies therein.  It turns out that about $58 billion, or just about one third of that cost, is being paid for by expecting employers to substantially lower their costs and pass through all the savings to their employees, who then pay taxes on them.

As revenue-raising mechanisms go, this is pretty indirect.  And with so many links in the chain, you can see lots of places where this could go wrong.  What if employers just cut their costs and don't raise their employees wages, because they're in a dying unionized industry?  What if they shift workers to other forms of tax-deferred compensation, like 401(k) matching or HSAs?  What if smart lawyers figure out a way to structure health plans to avoid the tax?  What if all the people who leave employer insurance for the exchange, or have their benefits cut, are low wage workers with low marginal tax rates?

Moreover, these effects are modeled against a hypothetical world.  What if insurance costs would have leveled off in high cost states without the excise tax?  We'll never know.  It's going to be hard, ultimately, to know whether this thing was revenue neutral or not.

There's a reason that the CBO has historically been wary of dynamic effects--they're a lot easier to model in a classroom than in real life.  That's presumably why they haven't modeled the dynamic effect of the huge increase in marginal tax rates on low wage workers (more about this later).  I can see the logic here, and it's very compelling as economic theory.  But it's not exactly revenue I want to count on.