Monday, 24 January 2011

How well has the Bank of England forecasted the inflation rate?

How have the Bank of England's recent inflation forecasts compared with actual out-turns?

Before that question gets an answer, we need to acknowledge that economic forecasting is extremely difficult. So we shouldn't be too hard on the Bank simply because it can not precisely forecast the inflation rate 24 months ahead.

The Bank of England present their forecasts with a considerable degree of modesty. In addition to their central forecast, they also publish upper and lower bounds that indicate the degree of uncertainty that the Bank attaches to its forecast.

Therefore, one shouldn't place too much emphasis on whether the forecast is right or not. Instead, it is much better to think of the forecast in terms of what it reveals about what Bank of England is thinking and how it views recent macroeconomic developments.

So what do recent forecasts reveal? The chart above illustrates the Bank's forecast between May 2008 and February 2010, along with the actual out-turn for inflation. Ironically, the most accurate long-term Bank forecast is the oldest. The May 2008 inflation forecast was rather pessimistic. The Bank expected inflation to remain above the 2 percent forecast for several years ahead. Although this forecast missed the slowdowin of inflation in 2009, it wasn't too far off for 2010.

From November 2008 onwards, the Bank sharply revised their forecasts. They began to expect a sharp deceleration in inflation. They got this right; inflation did come down sharply. In September 2009, it was just 1.1 percent.

Then things went off track for the Bank's forecasters. Inflation picked up sharply in the closing months of 2009. It is given a further boost with the resumption of the higher VAT rate in January 2009. Curiously, the VAT hike was pre-announced, so it was surprising that the Bank didn't capture the uptick in their forecast during the early part of last year.

However, the most revealing forecast is the one from February 2010. By then, the Bank had caught up with the VAT shock, and managed to get the q1 forecast more or less spot on. Thereafter, things start to go wrong. The Bank expected inflation to fall, very much as it had in the winter of 2008. Instead, the rate went the other way.

In forecasting terms, this was a near term miss. In other words, the recent surge of inflation caught the Bank of England completely by surprise. The normal response to such situations is to completely rethink one's assumptions. The next inflation rate will make particularly interesting reading as the Bank struggles to explain this divergence between their recent forecasts and the inflation out-turn.

So what does this tell us about UK monetary policy? Two things; a further round of quantitative easing is now extremely unlikely and an interest rate hike is coming sooner than previously expected.


DavidB said...

If they could genuinely get inflation down to that common forecast of about 1% then interest rates for savers as low as they are might be at least tollerable. The devaluation of my savings irritates me no end.

Nick Drew said...

Economic forecasting is bunk - astrology with numbers. Show me the econometric forecaster who dares to publish his track record. There are none.

The most honest that I know of is the EIA (US Energy Information Administration) which keeps all its historical forecasts on its site.

Turns out they are good at forecasting supply and demand, but their long-run performance on oil-price forecasting is dire: their predictions have been approximately 50% lower than out-turn prices over a sustained period. (which two observations, taken together, pretty much prove that 'fundamentals' are a poor guide to future prices)

Needless to say, their forecasts were, pretty much in the range of 'consensus' at the time of publication - so their failure is a proxy for everyone else's

As to the BoE's modesty, if I understand correctly, the 'fan' of uncertainty superimposed on some of their forecasts is based on back-calculated (implicit) variability as "assessed" by the market, based on the prices of relevant options and an option-pricing model