My post on the national income identity has raised lots of interesting discussion, particularly focused on the notion of the ‘Laffer’ curve [scare quotes reflect the fact that neither the curve nor the underlying idea was original with Laffer]. The basic theory underlying the curve, that both zero and 100 per cent tax rates will raise no revenue, is close to self-evident and certainly accepted by all mainstream economists*

What might be called the Laffer proposition is that the US is, or was, near the point at which taxes can be cut without reducing revenue. This is empirically false, as was demonstrated under Reagan.

Standard economic estimates suggest that the loss of output associated with each additional dollar of tax revenue is somewhere between 0 and 50 cents, with the best estimates (IMO) around 20 cents. This isn’t all a loss of measured output. Some of it is due to time spent in compliance and some reflects losses in consumer welfare because of misallocation of resources. So, I’d guess that a $1 reduction in tax revenue might produce, in the long term, an increase in output of 10 cents.
[This would be offset if the money was taken from high-return investments like education, or if the deficit was allowed to grow, but let’s suppose the cuts are applied to panem et circenses – in Australia, fireworks are the obvious example].

On average, because of things like progressive income taxes and social insurance, about half of any change in output arising from greater capacity utilisation flows through to government budgets. So, of the 10 cents worth of extra output, governments would get about 5 cents. But if you used a high-end estimate of welfare loss, and assumed that most of it was reflected in measured output, you might get a clawback of 15 cents. So Laffer is between 5 per cent and 15 per cent right.

All of this is relevant because, as a result of consistent Republican pressure, the Congressional Budget Office is likely to implement what has been called ‘dynamic scoring’, incorporating assumed supply responses into estimates of the budget impact of fiscal policies such as tax cuts. You can read a critical assessment by Stan Collender here. While we have yet to see how, and if, this will work in practice, the numbers that are being bandied about justify Collender’s “fear for the future of the republic”. Here for example, is Bruce Bartlett of the National Review

At present, there is a systematic bias in the revenue-scoring process that encourages tax increases and discourages tax cuts. Look at it from the point of view of members of Congress. In the former case, they are told that a 10% increase in tax rates will raise 10% more revenue, when in fact it will raise perhaps 7%. In the latter case, they are told that a 10% tax-rate reduction will lose 10% revenue, when it probably will only lose 7%.

That is, dynamic scoring will reduce the measured cost by 30 per cent. As I’ve argued this is twice as much as a reasonable upper bound estimate and five or six times as much as is likely. This is for the favorable case when the tax cut is financed by reducing current government consumption. If it is financed by cutting investment, printing money, or issuing debt, the overestimate is even worse. In these cases, the final cost of the tax cut will typically be equal to or greater than the first-round effect.

Finally, I should say that I’m talking about the long run (the average over a full business cycle) here. In a recession, tax cuts may provide a Keynesian stimulus and produce a short run increase in output. Conversely in a boom, there are no unemployed resources to bring into action, so there’s no possibly dynamic payoff.

*It’s obviously true at zero. With 100 per cent taxation, a few idealists might keep working for nothing in the hope of providing enough revenue to keep themselves and everyone else from starving, but economists don’t like to count on this kind of thing.