From the TUC

Does the Treasury story on wages stack up?

02 Dec 2013, by Guest in Economics

Today sees the release (sort of!) of two important analysis of living standards.

The first comes from the ONS and shows that the median disposable income of a non-retired household has fell by 6.4% between 2007/8 and 2011/12.

The second comes from the Treasury and has been reported by the FT and Sky although the report is not yet available on the Treasury website. This second analysis looks fascinating but it hard for me to respond fully to as the Treasury press office referred me to the general enquires line where I was told it should be on the website in the next few days. (And I also asked the Tory Treasury Twitter account for a copy but have yet to recive a reply)

So – on the basis of a newspaper report and helpful blog from Sky’s Ed Conway – what are we to make of the Treasury analysis?

Chris Giles in the FT notes that:

The Treasury number crunching acknowledges that take home pay has slipped far behind GDP over the past decade but find this is not a result of exploitation of workers.

Companies have used money that previously would have gone into wages to fund pension deficits and increased employer national insurance bills, the calculations suggest. While take home pay fell behind productivity, employer costs of hiring workers did not.

Ed Conway gives further details (and a couple of graphs):

However, now the Treasury has provided a possible explanation. In research its economists have put together it suggests that the main reason households have seen their wages drop back in comparison with the country’s overall growth is not that they are missing out at the expense of their employers: it’s that they are being paid in other, rather less obvious forms. It’s right, they say, that wages have not risen as fast as overall productivity since 2000: but if you look at overall real compensation per employee, it has pretty much kept pace…

The difference between the two measures is that overall compensation includes not just wages but also the social contributions made by employers, including pension contributions and National Insurance Contributions. Technically-speaking, these are forms of payment, except that because they don’t go straight into your pocket they don’t feel particularly obvious.

So, the overall argument is that whilst real wages have fallen since the recession – and indeed began to stagnate pre-crash – this is mainly due to a rise in employer NICS and pension costs rather than any other factor. Households have not become disconnected from the proceeds of growth and indeed should benefit as growth returns. Or, in more partisan terms (as Chris Giles noted):

Using this analysis, the government will seek to pin much of the blame for stagnant living standards on Gordon Brown, who raised employer national insurance bills frequently as chancellor and increased the rate in 2003 to fund higher public spending.

But does this story actually hold up? To be honest it is very hard to say either way conclusively without access to the full document but on the basis of what is the public domain, I have my doubts.

This is best explained by looking at the two charts provided by Ed Conway. The first one tells the story as the Treasury would like us to understand it:


Whilst real wages (the purple line) has certainly lagged output per worker (the blackline), this is more than explained by the rising disconnect between real wages and real compensation per employee (the brown line).

Or at least this is what the chart suggest to a casual viewer.  I’d suggest things are not this simple.

First – and perhaps most importantly – I’d like to know why the chart is rebased so that 2000 is 100. What would it look like if it started in 1992? Or indeed 1980?

Basing can make quite a big bit of difference to this things and I’d actually be tempted to ignore everything pre-2000 when looking at this chart.

Given the that most claims about the breakdown in the relationship between productivity and earnings date this to the 1980s and the 1990s, then looking at only post 2000 might be considered a tad misleading.

Second – look closely at the chart from 2000 until 2008 (i.e. until the recession hit). Even on this basing the black line of output per worker had risen by more than the brown line of compensation per worker.  The fall in the black line post 2008 (i.e. the impact of the recession) is what pushes the brown line ahead.

Might not the current period of growth reassemble this? Even on the analysis as presented in this chart, it is perfectly possible that output per worker may well rise ahead of real compensation per worker. I.e. the link between growth and living standards may not be as strong as HMT are keen to claim.

Now take the second chart.


The first striking thing about this is that National Insurance contributions and government pension and other social contributions don’t show much of an increase. The big mover here is ‘funded pension schemes’.

The analysis then really comes down to ‘real wages have fallen because employers are putting more cash into funded schemes’.

To which I’d answer – ok then, but if you want to demonstrate this then you need to show two versions of the first chart – one for employers who offer a pension where you’d expect to see the disconnection between real compensation and one for those who don’t where you wouldn’t expect such a disconnect.

The question is – have the real wages of firms that don’t other a pension risen by less the productivity? If they have, then the Treasury analysis is at best incomplete.

Is the Treasury trying to suggest that a cut in employer NICS contributions is the best way to boost wages? I’d happily listen to that argument but I’d also be concerned that many employers would simply pocket the extra cash or use it to rebuild margins rather than increase pay.

Of course over all of this there is the big question about sort of wages we are measuring – mean or median?

What we know from ONS analysis is that median and mean incomes present a different pattern – any serious analysis of living standards needs to focus far more on the median and look less at the mean. Perhaps (and I’m being generous here) this has not been done because of data limitations, but that still makes the analysis less striking than it appears.

Overall I remain unconvinced by this analysis – or at least the parts of it I’ve seen. I’d really like to see that first chart rebased to 1992 and 1980 and see if the picture holds, I’d like to see an analysis of real compensation versus output per worker for firms that pay and do not pay pensions and I’d like to see an analysis that looks specifically at median incomes.

But in the absence of all of this, this looks more like the Treasury trying to score a political point (the living standards squeeze is all Labour’s fault) rather than a serious attempt to explain what is happening to pay.


Of all the potential problems identified above the single biggest one I suspect is the first chart and the decision to base it to the year 2000.

To give an example of the kind of imapct this could have, I have included two charts below showing the median income of a non-retired household from wages, benefits and investment & pension income from 1977 to 2011. The data here actually doesn’t matter – I just happened to have this ONS data to hand this afternoon.

The first chart bases this data to 1977 and the second to 2000. The charts look rather different despite being the exact same information.

This is why I suggest caution when looking at that first Treasury chart. Basing in 2000 rather than, say, 1992 or 1980 might seriously impact what it appears to show.

incomes basing 1

Incomes basing 2

5 Responses to Does the Treasury story on wages stack up?

  1. Postkey
    Dec 3rd 2013, 9:11 am

    A similar argument is made by the Heritage foundation?

    “This picture is misleading because comparing wage and productivity growth is like comparing apples to oranges. The two are not directly comparable for two reasons.
    First, the government measures productivity and wage growth differently. The Bureau of Labor Statistics uses a different method to adjust productivity for inflation than it uses to adjust wages.[4] A simple comparison of these two series reveals the differences between these two measures of inflation, not the actual difference between wage and productivity growth.
    Second, wages are only part of what workers earn. Benefits, such as health coverage, 401(k) plans, and paid sick leave are an increasingly large part of workers’ earnings. Economic theory says that companies will raise workers’ earnings when their productivity rises, but it does not say that those increased earnings will take the form of cash wages. The correct comparison is between productivity growth and workers’ total compensation, including benefits, not just the cash wages portion of that compensation.
    To make an apples to apples comparison of productivity and workers’ pay, one needs to look at total compensation, and use the same measure of inflation to adjust both series. Chart 2 shows such a comparison. Over the past forty years compensation per hour and output per hour-that is, productivity-have moved almost in unison. Productivity rose 110 percent since 1968, and total compensation rose 103 percent.”

  2. Postkey
    Dec 3rd 2013, 9:21 am

    ” . . . to fund higher public spending.”

    Generally, it is the lower income groups that benefit from increased health and education spending? It increases their ‘social wage’?–historical-tables.html
    Table 14.

  3. Wages, Compensation & Growth | ToUChstone blog: A public policy blog from the TUC
    Dec 4th 2013, 9:26 am

    […] I’ve already blogged on the Treasury analysis of pay, productivity and total employee cost. As I said earlier this week, I don’t find it especially convincing but the debate is rumbling on and is a very important one. […]

  4. Andrew Sherwood
    Dec 4th 2013, 10:09 am

    Thanks Duncan, great analysis. I would have been interesting to see the top 1% of earners wages and total compensation received on these graphs.

    PostKey, It seems a bit of a leap to claim that spending on public education and health is part of people’s wages.
    The reason the state spends money on these is to benefit the people, basic human rights (which have been denied to people until recently in the US), and the economy, to provide a educated and healthy workforce.

  5. Postkey
    Dec 4th 2013, 11:30 am

    Social wage

    “The term used to denote social benefits available to all individuals, funded wholly or partly by the state through taxation and received free or at subsidised cost. The social wage includes free education, National Health Service treatment, etc. Trade unions have generally been concerned to support activities that increase the social wage. During the Social Contract agreement between the Labour government and the unions in the mid-1970s, agreement to wage restriction was sought by the government in exchange for measures to increase the social wage through increased government expenditure on health, education, etc.”