Hardly gilt-edged

Central bank governors have to say such things, and a halfway fair interpretation of what Sir Mervyn really means is something like: "We're willing to let inflation go to hell in the short term if that brings even the slightest benefit to UK economic growth; especially as something good may come along in the medium term to curb inflation, though we haven't got a clue what that might be."

Certainly, London's financial markets are behaving as if 'flexible inflation targeting' is a euphemism for 'abandoning inflation targeting'. That will have three obvious effects: first, downward pressure on sterling; second, upward pressure on inflation; third, downward pressure on interest rates.

Of these, the third pressure is the only one that would sustain gilts' prices but its effect may be largely notional. After all, 'real' - ie, inflation-adjusted - interest rates are already negative out to 25 years on government bonds, so how much lower can they go? A bit lower, is one response, if gilts maintain their safe-haven status. But that status always excludes protection against inflation - that's how it is with all fixed-interest securities - so the more that inflation rises, the more that gilts' prices are threatened. Meanwhile, sterling's exchange rate and inflation have a well-known push-me-pull-you relationship that depresses gilts' values when inflation rises - why should overseas investors continue to favour gilts if they fear that their capital is tied up in a declining currency?

It is scenarios such as these that keep dragging me back to the thought that it would be great to be able to gain from falling gilts' prices (in the jargon, to be able to 'short sell' gilts). So thanks to those readers who pointed out that Deutsche Bank via its range of db x-trackers exchange-traded funds provides two funds that do this. There is its UK Gilts Short Daily ETF (code: XUGS) and its caffeine-powered sister, UK Gilts Double Short Daily ETF (code: XUSS), whose daily price changes are twice its sister's.

Let's focus on the basic short-gilts product, XUGS. This may be the nearest that private investors can get to short selling gilts. Even so, it does not replicate what would happen if I really sold short; that is, I found someone who would lend me some gilt-edged stock that I could sell on the understanding that I would make good the lender's foregone dividends and sell back the stock when I wanted, though at the prevailing market price. If I did that, I would both bet that gilts' prices will fall and would have borrowed funds cheaply to invest elsewhere. However, when I buy the XUGS ETF then I tie up capital rather than borrow cheaply. In addition, I take on risks that need explaining in detail, partly because they apply to many ETFs not just this one. So think of this as a generic exercise in the risks of 'daily' index-tracking ETFs as well as a critique of this specific fund.

First, rather than short selling a stock, or even a basket of stocks, with XUGS I am buying the inverse of the daily returns of an index, specifically the inverse of the Deutsche Bank Gilts Total Return Index. That means that if, say, the index fell 2 per cent in a day, the XUGS fund would gain 2 per cent and if the index rose 2 per cent, the fund would lose the same fraction. In other words, the fund's returns are calculated by putting a 'minus' sign in front of its daily returns. Apart from the fact that the logic behind this is flawed - it does not reflect what happens in real life - it creates other risks. In particular, I lay myself open to the risks of 'path dependency' - in which, over a short run, future returns will be influenced by recent returns - and the effects of compounded daily returns on an index.

If returns are compounded daily, then the result from day one is added to the fund's value as the starting point for day two and so on. Imagine that I bought the inverse of an index (much like the XUGS fund) at 100 and in the first day's trading the index fell 2 per cent (that's good for me) and rose 2 per cent in the second day (bad). Instinct might say that after two days trading I am quits - up 2 per cent, then down 2 per cent. But the effect of compounding is that the daily value of the fund moves from 100 to 102 and then 99.96. I have lost money.

Not just that, but if daily returns are too volatile, then I can lose money even though the index has moved in the direction I wanted. This is demonstrated in the table, which shows returns for a simplified and notional fund over 28 days trading under three scenarios. First, take the two scenarios, 'low volatility good' and 'high volatility bad'. They demonstrate pretty well what the labels imply. Under both of them, after 28 days the fund's value has fallen 8 per cent. That's what I wanted, so I should have made money. In the low-volatility scenario, I have done. The effect of daily compounding is that I am ahead by 7.3 per cent. But in the high volatility scenario I am down by 13 per cent. Sure, the fund's value has moved in the right direction, but the sum of the wild daily percentage movements is negative. Thanks to high volatility, I have lost money and the fund provider has made money.

And just to underline the idiosyncracies of betting on an index of daily returns, let's compare the effects of compounding highly volatile daily returns with what would happen if I was actually selling short and buying back after each day's trading. This is shown in the column 'In the real world'. After day one, the index falls 10 points and I gain 10 by selling at 100 and buying at 90. In day two it rises 13 points, so I lose that many and after two days trading my aggregate loss is three points or 3 per cent of the starting value. But in the model on which the XUGS fund is based, I am down by almost 6 per cent. After day one, I am up by 10 points - or 10 per cent - but in day two I lose 14.4 per cent of 110 because my loss is the inverse of the index's gain. And so it goes.

Granted, in the real world it is likely that daily price movements will be closer to my 'low volatility' data than the 'high volatility' numbers. So perhaps I would have a better chance of making money from the XUGS fund in a proper gilts bear market. Yet it does not have much of a track record to help make that assessment. It was only launched in May 2010 since when there has been a bull market all the way and the fund has lost 19 per cent of its value. Still, at least that's consistent with what a 'short' fund should do in a bull market.

And Deutsche Bank does not minimise the risks involved with the XUGS fund. True, it is eligible for both Sipps and Isa accounts, but the fund provider says it is only suitable for "financially sophisticated investors who wish to take a very short-term view on the underlying market", adding that it "is not intended to be a buy-and-hold investment".

From the table, it’s easy to see why, but the pity is, it's the long-term view on gilts that I want to pursue. Mean reversion alone indicates that we are close to an upward sweep in interest rates that may persist for many years. The question is not so much whether it will happen but when will it begin? So I could take the view that XUGS is the least bad option and probably the only realistic way to play the scenario of the secular bear market in government bonds, therefore I'll pop a holding into the Bearbull Global Fund.

Fine, except another question lingers: would I do this with real money? I'm not convinced. Using a daily index tracker is a high-risk way of playing a slow-moving long-term outcome that has the potential for all sort of perverse outcomes. So there is a case for concluding that, if I can't find the vehicle to fit the opportunity, then I should let it pass.

Short selling with exchange-traded funds