Taxpayers Alliance fact check
Let's play 'fact check the Taxpayers' Alliance'.
They have shocking new "research" out which claims that 'the borrowing plans announced in yesterday's Pre-Budget Report will run up double the debt that Britain incurred winning the First World War.'
Happily, if unwisely, they included their sources and methodology for this. They read one book and looked at one website, which is actually far more extensive research than some of their past efforts. Unfortunately, they either didn't understand the website, or deliberately and dishonestly misrepresented the information it provided.
Their methodology was to adjust for inflation the amount that the government borrowed during the First World War, and compare it to the amount that the government will be borrowing over the next few years. They came up with the figure of £255bn for borrowing during world war one, compared to £512bn for the current crisis.
But the problem is that the website that they used to calculate this figure specifically explains that they've used the wrong comparison. The Measuring Worth website lets you enter an amount from any given year, and then calculate different measures of how much that amount would be worth now. The results vary considerably depending on different measures, particularly over 90 year periods, so it is absolutely crucial to use the correct measure for what you are trying to find out. Happily, they have a guide to explain which measure to use:
"If the amount you are asking about is the construction of a church, the cost of a war, or a new highway, again the context is important. If the question is how much it cost compared to the present cost of materials or labor, you would use the GDP deflator and/or the wage or earning index. However, you may be more interested in how important this project was to the community or the country. In the past there were less amounts of materials and labor available for all projects. So to measure the importance of this project (compares to other projects) use the share of GDP indicator."
To get the figure of £255bn, the Taxpayers' Alliance used the retail price index measure. As an economically valid comparison, this is about as useful as if they'd used a Random Number Generator or just made the number up themselves. It's clear from the explanation above that the most useful comparison is about 'how important this project was to the community or the country', and hence the indicator to use is the share of GDP indicator.
And if you use the measure of GDP indicator, you find that the relative cost of the First World War, in today's money, is £1,939bn, or nearly four times the cost of borrowing planned by the government now. (For what it's worth, if you instead use the GDP deflator or wage or earning index, you get numbers ranging from £311bn to £1,480bn).
It tells you everything that you need to know about the Taxpayers' Alliance that given the choice between cherry picking data and making a dishonest comparison which suited their argument or making an honest comparison based on the facts, they chose the former. Alternatively, if it were a genuine error, it shows that they don't understand the difference between different measures of inflation and other basic economic concepts. Thank goodness for the other Taxpayers' Alliance.
posted by donpaskini @ 6:10 pm
9 Comments
9 Comments:
We were talking about how much it cost; comparing the cost of the war to the cost of Darling's fiscal stimulus. We weren't comparing how important the two debts were to the economy in their respective eras.
That's the thing about this kind of adjustment for the falling value of money. There isn't a right answer. They just tell you different things. As the website says, it's about the "question". We set out, from the beginning, to study cost.
You can throw insults at our work all you want but its more credible if you find a genuine error first rather than thinking that a simplistic reading of a website's instructions can substitute for expertise.
Do you really think the best way to compare the cost of war ninety years ago to the cost of the fiscal stimulus now is using an indicator which is derived from a bundle of goods and services purchased by a "representative" household, and which excludes business and government expenditure? Why?
And it's funny that the disclaimer about how "there isn't a right answer" and the statement about why you chose one indicator over another didn't appear in the "full methodology".
If it is genuinely the case that no one at the Taxpayers' Alliance is aware of how to do proper research, then surely your rich backers could have a whip round and sort out some training.
Does it really matter if it works then we will repay the trillion or what ever it is, if it fails we might not need to repay anything since we might be at war again, I wonder who will invade who this time.
donpaskini,
Because that gives a pretty reasonable idea of the value of money at any one time. The basket of goods and services it can be exchanged for. It is an obvious place to start for such calculations. That might be why its series on Measuring Worth goes back further (another possibility is that historians can construct it more reliably).
Using another index such as the GDP deflators might yield a different answer but not by nearly enough to invalidate the basic answer that Darling is setting out plans to borrow significantly more than Britain did in WW1. Its not cherry picking, I just started with an obvious - and well understood - measure of inflation.
As far as not explaining this in the methodology, I thought it would be obvious when we gave a cost in pounds that we weren't expressing something in % of GDP. I did specify that we were using the RPI. After that, people can make up their own minds about whether or not the value of the two borrowing sprees is interesting, or if they don't care about absolute costs and just want to think in percentage points of GDP. I don't need to start every note from Stats 101.
1. The basket of goods and services is not 'the obvious place to start' for comparisons over time about levels of government debt. It is a measure which specifically excludes government spending, for a start.
2. You claim that 'using another index such as the GDP deflators might yield a different answer but not by nearly enough to invalidate the basic answer that Darling is setting out plans to borrow significantly more than Britain did in WW1.' But three of the five indicators which were available to you show that Darling is borrowing less than was borrowed during World War One, so it does matter which indicator is used. And the indicators which don't support your argument include the indicators which look at earnings, which are absolute measures as opposed to percentage of GDP, and which are directly relevant when it comes to pronouncing about the 'burden on families'.
3. I can't see anywhere in the original article which mentions that you were using the RPI - could you refer to the relevant bit?
1. It excludes some things but gives a decent account and is reliable (bear in mind national accounts don't exist until around WW2 so all these figures are historians' reconstructions). It certainly isn't like using a Random Number Generator. It's a legitimate measure of long term inflation.
Given that the GDP deflator doesn't change the conclusion that the borrowing Darling announced is significantly more than the borrowing used to finance WW1 my estimate is clearly robust enough that the particular measure of inflation used is not critical to the findings.
2. Comparing to average earnings would, like percentage of GDP, be a measure of affordability, not cost.
You might have double my income but we would generally both still measure cost in pounds. The important thing here is to produce an account that won't mislead the reader. Average earnings would mislead more than a straight inflation measure, I believe, as people expect, for example, that their incomes will grow in "real terms". At the aggregate, incomes don't grow in real terms if you're measuring inflation by average earnings.
3. Sorry, that isn't in there. This was just a short release and clearly it didn't take you long to work out we were using the RPI. I assumed you'd taken it from our release. What I was getting at is that we made it clear we were using an inflation index and not something like percent of GDP or average earnings. Again, given that our basic finding isn't particularly altered by the particular index used that doesn't seem critical.
In the end, all I'm doing in this note is comparing two figures and making a pretty basic adjustment for inflation. Hence, there wasn't a huge amount of detail in the methodology. In other reports, where the method is more controversial, we've gone into exhaustive detail. E.g:
http://tpa.typepad.com/home/files/the_burden_of_green_taxes.pdf
Or:
http://tpa.typepad.com/home/files/wasting_lives.pdf
Now, you can make a reasonable case for using the GDP deflator or, I believe, the RPI. Neither is really right or wrong. We could have a reasonable debate over which is more appropriate. To pretend that using the RPI leads to our results being some kind of "Random Number Generator" is just absurd, though, and prevents us having a reasonable discussion. It is no accident that my result is in the same region as the one obtained with the GDP deflator. They're measuring a similar thing.
Don
If you're feeling geeky, have a look at the TPA reports on pensions. They also include some very basic errors.
Just one particular howler - their most recent report claimed that the normal retirement age for all public sector schemes is still 60, when in fact it is 65 for new entrants, and has been 65 for local government for years. this stuff can be fact checked in seconds.
The worst thing about this gutrot is that journalists love these headline figures and often just quote them uncritically.
The TPA's backers know what they're paying for, don't they?
The other day, I was busy doing something with the radio playing in the background. This particular factoid (bailout = more than WWI) managed to get through and lodged in my head the way that George Michael singing 'Careless Whisper' does.
As long as journalists are prepared to use stuff published by pressure groups, Matthew will be happy to put up with the odd bit of flak in a comment-thread.
///Matthew Sinclair said...
...The important thing here is to produce an account that won't mislead the reader...///
That had me laughing out loud.
Surely Matthew what you meant to say was, "The important thing here is to produce an account that supports our headline-grabbing agenda"
Post a Comment
Subscribe to Post Comments [Atom]
<< Home