It is not always comfortable being grim in grim times – I’d rather be jolly. But the job of brickonomics is as much to challenge as it is to – hopefully – help inform views.

So it is some comfort when challenges made stand the test of time.

Here I have provided three example which I hope support the case for being critical of the numbers. While I may be smug about them in my weaker moments, I am realistic enough to know that I could so easily have ended up as a carping misguided buffoon had reality chosen a different path.

But the point I am trying to make here (other than the shabby and shameful attempt to buff my image as a soothsayer) is a serious one.

Don’t accept data, the output of analysis or forecasts unchallenged.

At the heart of most forecasts there is too much randomness, too many variables and too many unknowables to be able to accurately predict to a decimal point the detail of ups, downs and sideways movements.

This does not make forecasts pointless, on the contrary, they are extremely valuable because they explore what ifs and likelihoods and they provide insight and judgment which help readers make up their minds.

So here are the three revisions I have chosen to focus on.

Inflation

When it comes to understanding the meaning of forecasted annual rates of change, particularly when there are base effects influencing the figures, it’s well worth doing some simple sums from the base data and maybe a quick graph to check the real implication of forecasts.

Here’s one example of the rather sad record of the UK’s leading economists to forecast CPI inflation.

In November 2009 we were being asked to accept the consensus view among economic experts that CPI inflation in 2010 would end up at below 2%.

The collation put together by Treasury of forecasts from City and non-City experts put the median forecast for CPI at the end of 2010 at 1.8%. The highest forecast was 3.7%, the lowest 1.1%. And that 3.7% forecast from Citigroup was a very lone outlier being the only one above 3%.

Indeed that November the vast majority of the forecasters couldn’t seem to pick accurately the final quarter figure for CPI inflation in 2009, even though simple sums would have suggested a figure of 2.0% or above was very likely.

More worrying was the November Inflation Report from the Bank of England with its central projection for CPI inflation peaking at 3% before falling away. Mind you, this was an improvement on the August guesstimate made by the Bank, which suggested inflation would reach a mini-peak in 2010 of 2.5%.

At the time these forecast seemed curious to me and others and it prompted the question in brickonomics: “Are expectations of inflation too low?”

The answer we now know is yes. Inflation has not fallen below 3% all year.

As you can see from the blog, with some simple sums and a few fairly rational assumptions, my graph ended up looking a bit like what actually happened. What puzzles me is what were the experts doing coming up with outcomes that proved so wrong?

Anyway it was a relief to see revisions to thinking gain some momentum in the summer of 2010 . And as we see in the latest forecasts published by the Treasury that the consensus is 3.1% with the Office of Budget Responsibility at 3.0%.

Sadly this follows months when leading economists have been scratching their heads as to what might be driving inflation higher than their models might suggest while the Monetary Policy Committee of the Bank of England – the nation’s lever puller on inflation – has lost much needed credibility.

Stamp Duty

Here’s an example of why it’s worth checking individual lines in forecast data and asking yourself: “Has anyone really looked at this in detail?”

The folk putting together broad economic forecasts can only know so much in detail and can quite easily overlook something.

Here is an example of one revision, not totally unexpected, from the Office of Budget Responsibility, which got a bit of stick for the forecast it produced following the June Budget.

One line of numbers that caught my eye at the time was the forecast for Stamp Duty Land Tax. This suggested to me and to the more econometrically literate among my friends and contacts I spoke to that there was a bit of unchecked “reverting to mean” going on. That is to say the model assumes that things will broadly bounce back to normal, whatever that normal comes out at with the data series used. Often this is a reasonable assumption, but not always.

Now, I am only speculating on this as a possible reason why OBR came out with what appeared to be an exceptionally unlikely and optimistic projection for how much money the Treasury would take from Stamp Duty on property sales. But there is a whiff of it.

Anyway the details of my concerns were expressed in a blog, which pointed out that the OBR was wildly optimistic.

And so it was pleasing to see the OBR’s November forecast showing a  significant downward revisions of £2 billion in the annual take from SDLT in the later years of its forecast.

I suspect I was far from the only person a bit concerned about the original numbers.

Output

Sometimes data just looks and feels wrong and doesn’t fit with experience. In that case it is worth treating it with great caution rather than accepting it wholesale.

On this note, it will not have escaped readers of brickonomics that I’ve taken a fairly robust approach in my criticism of the figures for construction output.

Indeed this rather echoed the considerable concern within the industry at the suggestion that construction could have grown in the second quarter of 2010 by almost 10%. The growth rate over the second and third quarters dominated the official figure for growth across the UK economy as a whole.

Well, criticisms were voiced and it was naturally a huge relief to see revisions made in November. Yes growth now appears to have been strong, but not outlandishly so.

I’d like to think I played my part. I put together at great expense to my timetable a list of 10 possible reasons why the output figures might be wrong.

And you’ll be pleased to hear that in the end it was none of these, I am reliably informed by those who put the numbers together.

So what do I know?