There is something about big numbers that makes it difficult to keep a sense of proportion. Does it matter whether the tax giveaway in the Budget is pounds 1bn or pounds 4bn? Will government borrowing running a few billions ahead of where it ought to be make a difference?
To keep the billions in perspective, it is worth remembering that Britain's public finances are in good shape compared with most other industrial countries. The economic recovery has allowed the Government to make good progress in reducing the deficit from more than pounds 45bn two years ago. Keeping the public finances on this improving trend is more important than hitting a precise target.
Financial markets have a notoriously short collective memory, but the City well remembers that the reason Britain had a pounds 45bn public sector borrowing requirement in the first place was loose fiscal policy in the run-up to the 1992 general election.
We are now in the run-up to the 1997 (or earlier) general election. Markets are watching to see whether Kenneth Clarke is going to distinguish himself as the first post-war Tory chancellor who will not abandon fiscal prudence for electoral purposes.
This is the reason the precise size of the November giveaway is important. It will be a clue as to how far policy is being sacrificed to politics.
There are some hopeful signs that Mr Clarke will resist the traditional Tory temptation. The clearest came yesterday from the other half of the monetary policy double act, the Governor of the Bank of England. His speech displayed a carrot - no need for an interest rate rise now - and a stick, the possible need for one in the future.
The fact that under the present monetary arrangements the Governor acts as a kind of economic policy traffic warden - with the power to penalise but not to arrest an errant Chancellor - is the best safeguard we have against party political manipulation of taxes. That, and, give the man his due, Honest Ken Clarke's own desire to do right by the economy and safeguard the Government's inflation record.