Studying time series of a single variable can be a hazardous activity - as indeed can any forecasting exercise. But a reasonable forecast of the 2008 recession could have been made using a neural network analysis of the index of industrial production. (I tried with 24 lags of the dependent variable, 2 neurodes in the hidden layer, and a hyperbolic tangent squasher, using monthly data that go back to 1968 - just in case you wanted to know.) This same model predicted the recovery quite well too. The same model shows the recovery in industrial output stalling somewhat from now onwards, staying at just above zero growth over the next 15 months. But then the forecast indicates a nasty dive in the last quarter of 2011.

One should not place too much confidence in long range forecasts that come out of such a simple model. But we know enough about the general macroeconomic environment to know that we are far from being out of the woods just yet. The growth rate of the Eurozone economies is likely to be severely dampened by the need to address fiscal deficits of the southern states - and the UK trades a lot with (and so its fortunes depend a lot on) the Eurozone. At the same time, the deficit reduction policies being implemented at home will themselves have a deleterious impact on growth. On balance, I would still expect the next few years to be characterised by a long, slow recovery. But a double dip cannot be ruled out, and is, if anything, getting to look more rather than less likely.