The Sunday Business Post ran four stories last weekend- including a front-page banner headline - attacking not only public sector workers’ living standards, but workers in public enterprise as well.
Semi-States Enjoy Pay Increases During the Recession
Sitting Pretty in the Semi-States
Public Sector: the Insider Story
The Special Protections of the Semi-States
The Sunday Business Post is determined to outdo the Sunday Independent in public sector worker bashing.
And the most interesting thing about these articles is that they are based on a survey and a reading of wage numbers that are not only completely wrong – but make the most basic statistical mistakes. This is poor analysis, masquerading as informed commentary. Let’s look at some of the claims and see where they went off the rails (unfortunately the SBP is behind a paywall).
The SBP Survey on Public Enterprise Wages
The SBP did a survey. It purported to show the average wage in a number of public enterprises for 2009 and 2013. From this they deduced whether the average wage rose or fell. Here’s what their survey found.
To get the average wage they took the total wage bill and divided it by the number of employees (at least in a number of cases). The problem is that they forgot the compositional effect – that is, how using group averages can paint a misleading picture about how individuals in that group are doing.
Take Bord na Mona, for instance: in the SBP survey, they show employees (workers and management) getting a ‘pay rise’ of 2.5 percent. The problem is that they didn’t. Workers have had their pay frozen since 2009 (Unite represents many workers here). So how did the SBP survey get it wrong?
That ol’ composition effect. Here’s a simple example of how it works. Take a group of 10 people who individually earn a wage of increments of €10,000: €10,000, €20,000, €30,000 all the way up to €90,000 and €100,000. Their average wage would be €55,000.
Now take that same group of people earning the same amount – but this time remove the person earning €10,000 (through downsizing). The average of the remaining nine rises to €60,000 – or nine percent. BUT: no one received a pay increase. The average rose because of the change in the composition of the group, not the individual wage. (You can do this in reverse – remove the top earning person: the average wage drops by nine percent but no one had their pay cut.)
Changes in the composition of the group can change the average but doesn’t reflect individuals’ outcomes: low-income workers leave, some retire and are replaced by workers at a lower scale, etc.
This is a similar case with the Dublin Airport Authority – it shows a marginal increase but actually workers there have had their pay frozen. I can’t speak to the other public enterprise companies but given the fact that the SBP survey didn’t take account of the compositional effect in these two companies – or actually go to each company or relevant trade union to find out what the real situation is – this taints the whole exercise.
Just one more point before we leave this survey: they show the ‘average industrial wage’ falling from €34,349 to €31,716 (or 8 percent). I have no idea where they got these numbers (and they don’t say). Fortunately, we can go to the CSO for assistance.
The SBP surveyed the payroll of public enterprises between 2009 and 2013 but they only present data on the average industrial wage between 2010 and 2013 – they don’t say why they omit 2009. Even so:
- They underestimate the average industrial wage by nearly 25 percent
- They show the average industrial wage falling by 7.7 percent between 2010 and 2013 – the CSO shows it rising by 0.9 percent and 1.9 percent since 2009 (for 2013 I have used the average of quarterly figures). But don’t forget the compositional effect – this measures the average of the total wages. Some workers will have had pay cut or frozen during this period, some new start-ups may pay higher wages, some low-paid firms may have closed, etc.
The SBP survey cannot be treated seriously.
Public – Private Sector Pay
David McWilliams also uses wage statistics carelessly, comparing the weekly income of public and private sector workers as produced by the CSO.
‘Public servants are paid an average of €919 per week compared to €622 in the private sector. Just take that in for a second. What justifies this significant disparity?’
Wow – public sector workers are paid nearly 48 percent more than private sector workers. The problem is that there is no significant disparity, and may be none at all. Indeed, many public sector workers may be paid less. Just take that in for a second. Now, let’s explain.
The CSO numbers that McWilliams quotes are correct. However, this compares workers in sectors in which there is no counterpart in the public sector (e.g. hotel, restaurant, retail workers) and where there is no counterpart in the private sector (e.g. army, Gardai). This is a headline number that tells us little about the story.
To clarify the real story, CSO researchers produced a like-for-like comparison. They took into account: occupation, full-time / part-time, gender, professional / trade union membership, age, permanent / fixed-term, length of service with current employer, educational achievement, over-time, shift work, nationality, size of enterprise and more.
When they do this exhaustive work the results tell a far different story. The CSO researchers found the following.
Yes, there is a public sector premium but there are some noteworthy caveats.
- For all workers, the public sector premium is 10 percent. However, most of this is accounted for by women being paid better (that is, less gender pay inequality) which should surely be a good thing.
- When the CSO analysis focuses on permanent full-time workers, the premium falls to 7 percent – with public sector men being paid only marginally more than their private sector counter-parts.
However, none of the above includes the pension levy introduced in 2009. The pension levy was a pay cut by another name. The average pension levy is approximately 7 percent. If you were to subtract that from the above figures (admittedly a bit crude), we’d find:
- For all workers, the premium is only marginal and pay is the same for men.
- For permanent, full-time workers there is no premium and public sector males actually earn less than their private sector male counterparts
This is not the type of analysis many commentators want to hear, never mind present. It runs counter to the narrative. But while we keep a hold of the details, we must look at the larger picture. There is an attempt to hide the fact that private sector wages are low by EU-15 standards – and setting up a false debate helps hide this. It is a classic case of misdirection.
According to National Accounts,(see national accounts breakdown by sector under Economy and Finance) Irish Public Administration employee compensation - which include all civil servants and many local authority personnel – are over 9 percent below the EU-15 average. But industrial workers are over 22 percent below the average.
The issue is not the public sector premium (which is minimal and in some calculations non-existent). The issue is that both public and industrial employees are paid well below the EU-15 average – especially industrial workers. But don’t mention that. Better to contrive a false divide and try to set one group of workers against another.
Much of this material has been canvassed by others. In particular, Impact trade union has produced a helpful paper - Making sense of Ireland’s public-private pay gap - while Impact’s Bernard Harbor has a provocative piece in The Journal. It’s not that more robust and relevant analysis is available – it’s just that some commentators decide that evidence and facts are not essential to their unsubstantiated assertions.
Of course, we never get retractions. We shouldn’t expect the Sunday Business Journal to run a headline on the front page: ‘Apologies – We Mislead You’, or ‘Sorry – We Got That Wrong’. In the economic debate – especially when it comes to wages and labour costs – that game plan is to make claims, move on, and make more claims.
Regardless of wrong these claims may be.