Dynamic Scoring of Taxes: Simon Johnson's Simply Wrong on This
Simon Johnson (ex-IMF wallah) has an extremely strange piece at the New York Times arguing against the dynamic scoring of tax changes. The argument, in so far as there is one at all, seems to be that because it's difficult we shouldn't do it.
That's not really the correct response, the one we'd be looking for, from someone who used to be the chief economist at the IMF. Mr. Johnson was in fact hired by all of us to work out the tough things for us.
The argument in favour of dynamic scoring is obvious and in fact irrefutable.
Ah, sorry, let me explain what the argument actually is all about. When we make plans about what taxes might be in the future we need to decide whether we're just going to do a straight line analysis, a static one, or a dynamic one.
Do we just say, OK, income tax here is 30%, if we raise it to 40%, we'll get one third more money (for 10% is one third of 30%)? Similarly, if we cut rates from 30% to 20% do we think we'll get one third less money?
Or do we think that, well, if we change tax rates then people will change their behaviour? If we raise them, perhaps some people will think that those last few hours of work aren't worth it, maybe better to sit in hte garden with a nice book? If we lower them, maybe some people will do a few hours more work?
This argument also works the other way as well: perhaps people have some target income they desire to get. So if we raise taxes on them, they'll work even more so as to get their target income? This would mean that we wouldn't get one third more in taxes, we'd get more than that from a tax rise. Or, if we lower taxes, people would work less instead of more.
There are more formal names for these concepts (income and substitution effects) but don't worry about that. Our question is, do we want straight scoring, no one changes their behaviour as a result of the new tax rates, or dynamic, where we do the best we can to work out what will really happen?
Consider three cases.
Revenue losses to NYC and NYS from untaxed smoking are substantial. Losses from packspurchased without a stamp or with a counterfeit stamp are estimated to be between $241million and $272 million per year, or about half of the total revenue collected by the stateand city from cigarette taxation in NYC.
When we discuss a rise (or even a lowering) of NYC cigarette taxes, do we take account of how much people dodge the taxes or not? Do we consider how much more people might dodge them if they are raised or not? Given that from the moment anyone even discusses a change there are people planning on how to spend the money, yes, I think we probably should consider the actual amount we're going to get, not just the amount we'll get if we assume no one will try to dodge the tax.
The Commission paper estimates the tax itself would raise €16bn to €43bn, but the figure is very dependent upon the degree of dislocation, and previous reports suggest the Commission's original estimate was €10bn. Revenues would be shared between Member States and the EU (partly reducing national contributions).The Commission does not however estimate the reductions in receipts from other taxes. Stamp duty revenues are currently quite significant – £4bn in the UK alone. Add to this reduced corporation and personal tax receipts – the Commission's impact assessment anticipates a reduction in economic output of almost 1.8% – and it seems likely the revenue effect of the FTT will be negative.
So how do we count the revenues from this tax? The €16 billion we might get from the actual tax? Or the €16 billion minus the £4 billion lost in tax in the UK? Or the €16 billion minus all of the revenue lost from having an economy 1.8% smaller? It's worth noting that the range, the €16 to €43 billion is all about dynamic scoring. The question being, how many financial trades will there be after financial trades are taxed?
It would be very strange indeed to try and do straight or static scoring on this tax for part of the aim of the tax is actually to reduce financial activity. So we have to use dynamic scoring.