## MPs’ Pay and the Law of Misleading Averages

By now you’ll surely have seen the that IPSA, the Independent Parliamentary Standards Authority, has recommended that MPs should receive a 9% increase in their pay in 2015/16 which, on top of the 1% increases they’re due to receive in the next two year, will raise the basis gross salary for a backbencher to £74,000 a year from its current figure of £66,366 a year.

IPSA have also proposed that from 2015 onwards MP’s pay should be indexed to changes in *average* earnings and, as justification for this increase, has put forward the argument that since 1980, the amount we pay our backbench MPS has fallen relative to *average* earnings from 3.16 times the *average* in 1980 to just 2.7 times the average today.

Now when people start taking about averages, particularly bureaucrats, the very first question that needs to be asked is ‘which average are you talking about?’.

If you drag your memory back to the days when you did maths at school then you should recall that there are three different type of average one can calculate, the mean, mode and median, each of which will give us a different average based on the distribution of incomes, in this case, across the whole population.

For those who switched off in maths class, let me explain that a little further – imagine we choose 100 people at random and ask them to tell us their annual gross salary (i.e. before tax) and then put those figures in a simple table so we can calculate their average salary.

As already mentioned there are three different ways in which we can calculate the average.

We can take the *mode*, which is simple the most common or frequent salary figure that appears in our table and take that to be our average salary – in practice we’d group salary’s together into £1,00o or £5,000 bands to avoid running into problems if everyone in sample has a slightly different annual salary but you get the general principle here, if more people in our sample earn between £15,000 and £20,000 a year than in any other group then that’s our average figure.

The problem with taking the mode as our average is that salaries are not evenly distributed across society.

We live in a country where a very large number of people work in jobs that pay only very modest salaries while a relatively small number of people command very much larger salaries so the overall earning distribution is skewed heavily toward the bottom ends of the distribution, which mean that if we take the mode as out average, that will give us an unreasonably low figure for our average salary.

Our second option is to use the *mean*, which we calculate simply by adding up all the individual salaries in our table and dividing that figure by the number of people in the group (100).

That’s the type of average that people will be most familiar with and will probably consider to be the most reasonable one to use but there is a problem when we come to use the mean here because one of people in our (hypothetical) sample works as a commodities trader in the City of London and, when you factor in their annual bonus as part of the salary, it turns out that they earn more in a single year than all of the other 99 people in our sample put together – in statistics we call this an ‘outlier’, a figure that is wildly different from the rest of the data in a particular dataset. In this example, our commodities trader has the effect of pushing up the mean salary figure to the extent that 75 out of the 100 people in sample earn less than the average salary for the whole group – giving us another heavily skewed distribution – when most people would assume that if they earn the average salary they would fall about halfway between the highest and lowest earners.

That leave us with just the *median* to consider, and the median figure is calculated by sorting the salaries in our table into order from the lowest to the highest (or vice versa – it doesn’t matter) and then taking the salary that falls right in the middle (50th place) as our average.

That is the median, and across large swathes of economic and demographic data in which we know that whatever we’re trying to measure is unevenly distributed across the population it is the media that is the preferred method of calculating the average because, of the three, it is the one that it least susceptible to the influence of skewed or variable distributions or outliers.

So, bearing all that in mind, when IPSA is talking about linking MPs pay to *average* earnings, which *average* are they referring to?

Okay, based on data from the Office for National Statistics, the mean figure for gross weekly pay (excluding overtime) in the United Kingdom in 2012 was £476.40, which scales up to an annual salary of £24,772.80 – and if we multiply that figure by three, which appears to be the figure that IPSA and other have in mind for the ‘correct’ relationship between MPS pay and average pay, then we end up with a annual figure of £74, 318.40, which is as near as damn it that figure that IPSA have put forward.

So I think we can safely conclude that IPSA’s average is the *mean*.

But what about the median?

Well, the ONS figure for median gross weekly pay (excluding overtime) is just £391 and we scale that up we get an annual salary of £20,322 which, when we apply the three times multiplier for MP’s give us a figure of just £60,996.

All of which means that, just on their current salaries, ** backbench MPs are paid 3.27 times the average (median) salary for the UK and that will rise to 3.57 time the average (median) salary** if IPSA’s proposed pay increase for MPs goes ahead, assuming that average (median) earnings rise by just 1% per annum between now and 2015.

As the great John Lydon once asked, back in the days when he went by the name Johnny Rotten:

‘Ever feel like you’ve been cheated?’

What this also proves is people like Mark Ferguson, the editor of LabourList, need to pay rather more attention to the details, and especially the figures, before speaking out in favour of index-linking MPs pay to average earnings:

There is a simple solution though – and one that be used to restore, rather than erode, the esteem in which MPs are held by the public.

Why not set MPs’ salaries at a level that is a multiple of the average UK salary? (For example 3x). That would mean that being an MP would continue to be a lucrative profession, but their salaries would only rise at a rate at which they were delivering prosperity for ordinary working people. It would also make future pay rises completely defensible, and might also encourage some MPs to take a greater level of interest in the real average national salary – and how to increase it – than is currently the case.

Mark’s argument that linking MPs pay to average earnings would provide MPs with an incentive to deliver prosperity for ordinary working people might have some merit if the link was based on median earnings as this would at least force MPs to work to improve pay for people right in the middle of the income distribution but its one that falls apart completely if the link is made to changes in mean earnings as this figure can be inflated simply by increasing income inequality and creating greater prosperity just for those at the top of the income distribution while allowing the wages paid to ordinary working people to remain flat, if not decline.

If the last few years should have taught us anything it’s that just about the last thing we need to doing in providing backbench MPs with a financial incentive to work towards increasing the bonus payments paid to bankers and city traders in order to up their own salary.

Linking MPs pay to changes in median earnings would be a distinct improvement on any link to mean earnings, particularly if the link operated in such a way as to introduce a performance-related element by reducing MP’s salaries in line with any falls in median earning but for many people that still doesn’t quite far enough because it fails to create any kind of incentive for MPs to deliver prosperity for those at the very bottom of the income distribution, people who rely on benefits for either the totality of their income or to top-up the income they receive from taking on low-paid and often part-time work.

If we want to create an incentive for MPs to do more for people at the very bottom of the income distribution then our preferred link in terms of future changes to MPs pay should be one any pay rise for MPs is determined either by changes in median earning or the annual uprating in welfare benefits, whichever is the lower figure – and to avoid MPs sneaking in a pay rise on the back of sucking up to pensioners, I’d suggest that the link to benefits should be made to any changes in the single person over 25 rate of Jobseeker’s Allowance.

In fact, and while we’re kind of on the subject of welfare benefits, we also have the issue of MPs moonlighting and taking on second jobs to take into consideration as well, so why no take the link to welfare benefits a stage further by dispensing with salary payments to MPs altogether and, instead, pay them a mean-tested allowance with the same earning taper as Universal Credit (65%, I think). That way we create a positive incentive for MPs to ensure that when they make changes to the benefits system on the pretext that they will ‘make work pay’ those changes will actually make work pay.

That’s not an idea that I expect would prove popular with the vast majority of MPs but it does strike me as one that might be worth pursuing with a petition in the hope gaining enough signatures to force MPs to debate such a proposal, if only to see exactly how they react to it.

With IPSA’s proposals due to go out to public consultation, at the very least the point needs to be made, and made forcefully, that any links to average pay or earnings need to be based on median pay/earnings and not mean pay/earnings if the system is to operate fairly and create an incentive for MPs to increase prosperity for the many and not just few, including themselves of course.

Pingback: How IPSA is misleading us all in trying to raise MPs wages | Liberal Conspiracy()