Wednesday, 13 April 2011

Efficient taxes hypothesis

This article on pensions from the Economist made me wonder about something.

It used to be that people in the UK were compelled to buy an annuity on retirement with their pension savings. Well, almost compelled - if they took it out as cash, they'd have to pay tax on the lump sum.

So if I saved for ten years at 22% tax relief, and I then withdrew the sum, I'd have to pay 22% tax on the withdrawal.

Fair enough - the idea of pension tax relief is to encourage people to build up an income which would stop them having to draw on state benefits in retirement. If I can withdraw the cash, as if it were a standard savings plan, then the tax treatment should be the same as if I'd saved the money outside of a pension.

But could this be used to game the system?

What if today - with top tax rates at 50% - I think that the income tax rate is higher than its long-term average. Then I might save in a pension now, intending to withdraw later when tax is cut back to 40% or lower - so that I avoid paying today's 50% rate and pay the 40% rate at the time of withdrawal instead.

And conversely - if rates are cut to 33% but I don't believe it will last, I might decide not to save now (or even withdraw my pension savings early) and then wait for tax to go back up before saving again.

(I know this is a simplified picture of the complexities of pension tax treatment, but imagine it's this straightforward for now).

Now this requires predicting whether tax rates are high or low compared to their long-term average. If I were trying to do the same with oil prices, or the stockmarket, or Treasury bonds, or gold, or the Japanese yen, we'd say it was a mug's game. The efficient market hypothesis (which I broadly believe in, despite behavioural reservations) says that the market's best forecast of future prices is already built into today's price, and unless I have inside information or a really good reason to think the market is wrong, I shouldn't make investments based on my predictions of future prices.

So is it equally foolish to make predictions of future tax rates? Or, because there is no market in tax rates (except a bit of cross-border competition and arbitrage) is it possible to actually make accurate forecasts of future taxes?

Is there an efficient taxes hypothesis?

1 comment:

Kayla said...

In the States you can have a retirement account where you don't pay tax now, but pay tax later, or you can pay tax now but pay no tax later. I know many in high income brackets prefer the 'don't pay now' option, as they are in a higher tax bracket but expect their income will drop and pay much less tax later. Nothing similar in the UK?