MPC's new dove ruffles a few feathers over sterling
A forecast based on the random walk model would be pretty good until it turned out to be catastrophically wrong
Monday 20 September 1999
His was an argument of two halves. The first was an attack on the conventional assumption used in preparing the Bank's inflation forecast, which has sterling declining gently during the next two years. The second was his preferred exchange-rate forecast.
The Bank's current convention uses the catchily titled concept of "uncovered interest parity" (UIP), which is based on the instinct that if there were a consistent opportunity for profit, traders would arbitrage it away. Suppose you can earn 6 per cent putting your money in the UK for a year and 4 per cent on a similar investment in Germany. If the exchange rate is expected to remain unchanged, the choice is a no-brainer. Investors would borrow in euros to deposit in sterling. Enough of them would do it, in theory, to send UK market interest rates lower and euro rates higher until they were equal.
So an observed 2 per cent interest differential ought to imply an expected 2 per cent depreciation in sterling over the year. And this is how the Bank calculates the exchange rate forecast it uses in making its predictions for inflation. In August's Inflation Report, prepared when the sterling index stood at 103.1, this method implied a gentle decline in the index to 96.6 by mid-2001.
As the chart suggests, the exchange rate actually rarely behaves in this way. Not surprisingly, detailed econometric research confirms the eyeball impression. The UIP theory provides only a poor forecast of actual exchange rate movements. Dr Wadhwani went on to show that a simple alternative - the random walk assumption that the exchange rate today is equal to the exchange rate yesterday plus a random shock - performs better in terms of forecasting actual moves in the pound.
Again, simply looking at the chart suggests intuitively that this should quite often be true. For the level of sterling does move in fairly stable if narrow ranges for long periods, and not only during the ERM episode. It has done precisely this since early 1997, bobbing around between 100 and 110 on its index.
Does that mean it would be right to agree with the MPC's newest and biggest dove that the Bank's inflation forecast should be based instead on the conventional assumption that the exchange rate will stay constant? For if so, that has a dramatic implication for interest rates. It would bring the inflation forecast down by 0.4 percentage points, which would allow a far lower level of rates.
Well, not necessarily. For a forecast based on the random walk model would be pretty good until it turned out to be catastrophically wrong. A session on economic forecasting at last week's conference of the British Association for the Advancement of Science (yes, economics is a science and has been officially since 1897) shed some light on predicting exchange rates.
David Hendry of Nuffield College, Oxford, noted that simple forecasting rules can be very successful but nevertheless not very interesting. Take two friends who play a game of predicting every 30 seconds whether somebody standing at a bus stop will stay or not. One decides to predict this person will stay, until he leaves, and then she will switch her forecast to predicting that he will not be at the bus stop. She will be right every time except the single one when she has predicted he will stay but he goes. It will look like a pretty good forecasting record.
The other friend figures people wait at bus stops for the bus, so she predicts the person will stay until the bus comes and then leave on the bus. But the bus arrives, the person stays and a minute later is picked up by a car. This forecast was right for a bit and then went badly ashtray. Like the UIP, not such a good record.
However, it would not do for the Bank of England to be right about forecasting the exchange rate in the simple way, and predicting a constant exchange rate is exactly this kind of forecast. It would be right until the pound fell, and a different constant forecast might well be right after the pound fell. But all of the significance for the economy would be packed into the bit in between, the single episode of incorrect forecasting.
As Ray Barrell of the National Institute for Economic and Social Research pointed out in another paper at the conference, the behaviour of the pound is characterised by sudden, sharp shifts from one band to another. Sterling will stay strong, and a constant forecast will be perfectly reasonable, until it falls sharply. Interest rate decisions somehow need to take account of the likelihood of a sharp depreciation as these are exactly the episodes that have in the past lit the blue touch paper on inflation.
Dr Wadhwani's paper goes on in its second half to set out a theory that explains why the pound has been strong, and could therefore be used to predict when and why it might fall. He argues that the fundamental level for sterling depends on trade deficits, unemployment rates and equity returns; and empirically the current account turns out not to matter much. The rise in sterling can be explained by the increase in German unemployment relative to UK unemployment. The most likely event to send the pound falling sharply would be a drop in US and therefore UK share prices.
A good forecast for the exchange rate now would put perhaps a 60 per cent chance on its staying around its current level and a 40 per cent chance it will fall 20 per cent. Quibble about the probabilities, but the two most likely outcomes are far apart from each other. The Bank's middling UIP forecast, in between the two, is the least likely thing to happen. It might be unsatisfactory, but without a consensus on which of the alternatives to plump for, it must still be the least controversial.
As Lord Burns, the former permanent secretary to the Treasury, remarked at the British Association: "The exchange rate has been one of the biggest complications for policy makers in the last 25 years." That certainly hasn't changed. And ironically, the currency markets sent sterling higher in reaction to Dr Wadhwani's speech even though, if he were to win the argument on the MPC, interest rates would be lower. After all, if one of the Bank's eggheads is saying the pound will stay strong, it must be true - mustn't it?
As Voltaire once said, “Ice cream is exquisite. What a pity it isn’t illegal”
- 1 Mother fed her daughter tapeworms to make her skinny for beauty pageant
- 2 Crystal Palace next manager latest: Palace consider Ally McCoist - EXCLUSIVE
- 3 Richard Dawkins on babies with Down Syndrome: 'Abort it and try again – it would be immoral to bring it into the world'
- 4 ALS ice bucket challenge co-founder Corey Griffin drowns, aged 27
- 5 Tim Sherwood column: 'It started as a three-horse race but turned into the Grand National'
'Alien thigh bone' on Mars: Excitement from alien hunters at 'evidence' of extraterrestrial life
West poised to join forces with Assad in face of Islamic State
Pamela Anderson rejects ice bucket challenge because of ALS experiments on animals: 'Mice had holes drilled into their skulls'
James Foley 'beheaded': Isis video shows militant with British accent 'execute US journalist' – as hunt begins for killer
ALS ice bucket challenge co-founder Corey Griffin drowns, aged 27
Richard Dawkins on babies with Down Syndrome: 'Abort it and try again – it would be immoral to bring it into the world'
Scottish independence: English people overwhelmingly want Scotland to stay in the UK
Isis threat: Cameron wants an alliance with Iran
Michael Brown shooting: Chaos erupts on the streets of Ferguson after autopsy shows teenager was shot six times – twice in the head
Disgusting, frustrating, but intriguing: how the country really feels about its politicians
Bin bag full of cats' heads discovered near Manchester's Curry Mile
iJobs Money & Business
£50000 - £60000 per annum: Harrington Starr: Network Engineer - CCNP, Hedge Fu...
£60000 - £70000 per annum: Harrington Starr: Senior Network Engineer-CCIE, Mul...
£30000 - £50000 per annum: Harrington Starr: Network Infrastructure Engineer (...
£35000 - £45000 per annum: Harrington Starr: Network Engineer (CCNP, BGP, Mult...