Monday, September 1, 2014

From the U.K. GDP page at MeasuringWorth

Maybe you saw this already, at MeasuringWorth. I just found it now:

The data published here for the years after 1948 are published by the UK Office of National Statistics. Each year this office publishes the latest update of these data in the United Kingdom National Accounts, The Blue Book. It is surprising that each year they have a new value for years past and this is not just because they are using a different “base year.” They have continuously revised the nominal observations as well. For a complete discussion of this, see How the Post War United Kingdom's Annual Estimates of GDP Have Changed.

The revisions in the series published in 2012 were very large and not explained well and so we decided to not use them. In 2013 they published a new version of real and nominal GDP (only for) 1997 to 2012 and again there are changes. (For example the value for nominal GDP for the year 2008 is now 28,000 million pounds larger than it was two years ago.) ... Until this is resolved, we will use the same data for 1948 to the present that was published in 2011 and extrapolate ...

I have used data from MeasuringWorth for a long time, and I have come to trust them. If they choose not to use the revised data, I have to at least be concerned that they may be right.

It makes me wonder also about the mid-2013 comprehensive revision of U.S. data.

A brief excerpt from How the Post War United Kingdom's Annual Estimates of GDP Have Changed:

For example, the nominal value for GDP in 1958 is different in every report from 1959 to 1974, starting at £22,493 million and ending at £22,843 million. After that it “only changes” ever two three years. For the 2006 to 2012 reports is stayed at £23,050 million. To repeat, without seeing the tables use for construction in the earlier years, it is not possible to know why there were so many changes.

It is also unlikely that there have been 18 different sets of new data to use in the last 50 years that would explain how the growth rate between 1958 and 1959 changed so often.

The opening paragraph:

How fast did the United Kingdom real GDP grow between 1958 and 1959? By consulting the official numbers of the 51 UK annual government reports since those two years, you can choose 18 different values in a range between 2.66% to 4.68%. Which is correct? Or perhaps the question is which of these 18 choices is “relevant?” If you are interested in how people were reacting to the reports at the time, then the relevant number is 2.66%, the growth computed for the 1960 report. If you are interested in the most consistent time series, then perhaps it is the 4.68% from the 2012 report. But in 2011 it was 4.28%, which does not give one much confidence that the 2013 report will not be different again.

And an observation from the second paragraph, which just happens to say exactly what I was thinking:

In general, we might expect estimates of nominal GDP to change for a few years after the initial estimate, as new data become available, and to then stay fixed.

You know there are many people who think inflation is much higher than the official U.S. numbers indicate. Now there may be reason to doubt not just inflation, but also GDP. And the unemployment numbers also, because they changed the way they figure it some years back. Each data series that comes to be doubted strengthens the doubt elsewhere.

Maybe they are not doing anything wrong. But it looks wrong. It looks like they are changing the numbers to fake a better economy.

I guess this is just one more reason we really ought to fix the economy: so that the state stats don't have to be tweaked any more to make things look less bad. That would be nice, huh?

// "Citation"

Lawrence H. Officer and Samuel H. Williamson, "What Was the U.K. GDP Then?" MeasuringWorth, 2014

No comments: