Is Policy Exchange right to claim that the IFS is wrong on public sector pay?
Today’s Telegraph led with the claim that ‘workers are 40% better off in public sector’. The claim that public sector wages are ‘out of control’ is based on this research from Policy Exchange. But in February the IFS concluded (in research which Policy Exchange have referenced, and therefore presumably read) that the gap was 6%, that there were still methodological reasons which limited the validity of this estimate (such as the impossibility of controlling for gender discrimination in the private sector and the difficulty of taking into account actual ability rather than proxy measures for productivity) and that:
Before the financial crisis, public sector employees were, on average, paid at levels roughly in line with their private sector counterparts once observed differences in skill composition were taken into account.
So, have Policy Exchange managed to improve on the IFS’s methodology?
A brief review of their report suggests that they have not. The Telegraph’s statistic is derived from a straight analysis of hourly pay in the public sector compared to the private. But as all academics who work in this area acknowledge, this is a meaningless comparison. It doesn’t control for the different types of jobs in each sector, for the age of workers, for their gender, for their qualifications, their length of service and other variables which make a difference to pay levels. In particular, it doesn’t control for the nationalisation of several UK banks which are currently public sector employers. In their research, the IFS state that:
Drawing conclusions on the public sector pay premium from such raw comparisons is not appropriate. The raw differential does not take into account the fact that the skill compositions of the two sectors are markedly different: it is like using the average pay of neurosurgeons and the average pay of bartenders to conclude that neurosurgeons are overpaid!
Interestingly, when commenting on their basic comparisons, the authors of the Policy Exchange report do note that ‘the above figures can only be indicative since they do not account for compositional differences in the workforce between the private and public sector’ and go on to assess that once age, qualifications, gender, hours and region are controlled for the premium falls to 8.8%. It is impossible to assess the validity of this figure as no details are provided of the methodology used – all we know is that it remains higher than the statistic produced by the IFS, and that the former (and completely inappropriate) Policy Exchange figure is that one that has been flagged to journalists.
It is also unclear how much this overall hourly pay premium figure is affected by better pay in the lowest paid occupations. Policy Exchange’s own report acknowledges that ‘public sector pay premiums are highest for those with relatively low earnings’, a conclusion that the TUC’s own analysis has also identified. Without seeing the same analysis of pay (with age, qualifications etc controlled for) broken down by decile it’s impossible to say, but it seems likely that if there is a pay premium it’s likely to be one that benefits staff who get slightly above NMW in the public sector but get nothing above the legal minimum in private sector jobs.
Policy Exchange are clearly entitled to argue for low-paid public sector staff to get paid even less, but if this is their policy ambition it would seem to make more sense for them to state it more clearly in future, rather than making sweeping statements about pay comparisons in the sector as a whole.
There are a number of other questions about the report’s methodology which I don’t have time to cover in detail. The report claims that once pensions are taken into account the public sector premium doubles to 43%. No information is provided on how the pension premium has been calculated (or whether the analysis used controls for any of the variables outlined above), but as it is significantly higher than the difference cited by IFS (14%) it may be questionable.
The report also compares different categories of worker, but again it’s unclear how meaningful the comparison is. Policy Exchange tell us that public sector ‘primary and nursery education teaching professionals’ earn £33,140 a year (median) while in the private sector their annual pay is only £21,159. But surely such a large differential simply suggests that the former category contains different job roles to the latter – given primary school teachers are far more likely to be employed in the public sector, and childcare providers far more likely to have a private sector employer, this seems likely to be responsible for the difference. Are Policy Exchange really suggesting that private school primary teachers are paid over £10,000 a year less than their colleagues in the state sector?
Policy Exchange end their report by claiming ‘public sector pay premiums are large and have grown in recent years’. This is simply wrong. At best the evidence shows that some workers in the public sector have a small wage premium, which is a result of the lowest paid staff in public sector employers being paid slightly more in the public than the private sector.