Friday, 5 June 2009

Mythbusters 101: Wages in the Irish public sector

Sli Eile: Are incomes, on average, higher in the public sector than in the private sector? Note the reference to ‘incomes’ and not ‘wages’ in this question.

The context is claims made by some commentators that wages (note wages) are about 20% higher in the public sector than in the private (and the strong implication that wages in the public service need to be cut by this amount for reasons of restoring public finance as well as encouraging cost competitiveness in the traded sectors). Allied to this is a broader agenda about reducing wages in general. It is no accident that media and political commentators focus in not only on the supposed wage difference but also the claimed benefits of security, pension entitlement and outmoded and inflexible work practices.

By focussing, narrowly, on wage costs as distinct from incomes (much harder to measure with up to date statistics) the terrain for making comparisons is restricted. Thus, income of the self-employed, dividends, rental income, profits etc are excluded. But, lets stay with that narrow terrain of wages for the moment.
An important reference was the paper entitled Public-Private Wage Differentials in Ireland, 1994-2001 by Gerry Boyle, Rory McElligott and Jim O’Leary published in the ESRI Quarterly Economic Commentary in 2004. The data are dated but it remains the single most used reference point for comparing wages. Recourse to published CSO data on average hourly earnings for specific sectors is possible to compare trends since 2001, the last year covered in the Boyle et al comparison (even though the latter uses total monthly earnings).

In comparing the public and private sectors (sticking with a broad definition of public sector to include education, health, semi-state companies etc) account needs to be taken of differences in age-profile, educational attainment, unit size and occupational structure. Boyle et al. factored this in by using a multi-variate approach based on through-time data from the European Community Household Panel Survey (the comparator nowadays would be the EU Survey on Income and Living Conditions (EUSILC). As in any comparison of income care is needed with regard to quality of reported earnings – perhaps more so at the top and lower end of the income scale where some survey under-reporting is possible. By focussing, only, on employees there is relatively less scope for under-reporting. The other difficulty arises from comparing various ‘benefits-in-kind’ not all of which may appear in the reported data.

Data issues aside, the Boyle et al. paper concludes, clearly and convincingly, that on average wages are higher than in the private sector for employees. They conclude following extensive regression analysis on earnings data as follows:
We estimate that in 2001, the latest year for which we have completed our analysis, the premium enjoyed by public servants was about 13 per cent, on the basis of gross monthly earnings.

More recent work by Eilish Kelly, Séamus McGuinness and Philip O’Connell (‘Benchmarking, Social Partnership and Higher Remuneration: Wage Settling Institutions and the Public-Private Sector Wage Gap in Ireland’ ESRI Working Paper no 270, December 2008) also shows a significant premium for public sector workers compared to private – statistically allowing for various factors that differentiate these sectors (educational level, occupation, age etc). Using more recent data from a different data source to that used by Boyle et al.: the National Employment Survey, Kelly et al. report an increase in the estimated public sector pay premium (after adjustment for other factors) from 7.7 to 23.5% between 2003 and 2006. Furthermore they report, as did Boyle et al., higher premia still for lower paid workers in the public sector (32% in 2006 compared to 24% in 2003 in Kelly et al). It should be noted that Kelly et al restricted their sample to full-time, permanent employees who are aged between 25 and 59, and exclude semi-state body employees
(All of these studies use standard econometric regressions with employee wages regressed on a range of explanatory variables including age/experience, education, gender, size of enterprise etc. However, there are some differences in the list of explanatory variables included. For example Kelly et al include a measure of ‘professional body membership’ to control for the effect of union membership on wages. This variable, alone, accounts for a premium of 14% in 2006 in favour of public sector workers.).
Why such a premium exists and what its implications are in the current economic climate could be explored further. Historically, it is likely that the existence of such a premium is associated with:
- Strong trade union membership across the whole public service (picked up anyway in the Kelly et al paper);
- Allied to this the role of centralised pay agreements in maintaining a relatively stable pattern (between public and private) over time; and
- Benchmarking which had the effect of raising senior and top-manager pay in the public service relative to lower-paid public and private sector workers.
Boyle et al. showed that the difference between high and low earners is smaller in the public sector (in 2001). Furthermore they showed that the ‘premium’ to public service workers was higher at the lower end of the wage distribution which illustrates that the meagrely-paid among particular groups of public service workers like some industrial workers, clerical staff at point of entry etc on close to something around €25,000 a year are relatively well paid compared to the working poor or in the private sector on rock-bottom wages close to the national minimum wage (and sometimes less in the ‘informal’ economy). They estimate a premium of about 17% for the bottom 10% of earners (the lowest 10% of earners in the public service earned 17% more than the lowest 10% of earners in the private sector – after adjustment for factors such as age, education, occupation etc).

A marked feature of public sector pay has been the very large growth in senior and top management salaries– some of this linked directly to ‘benchmarking’ awards in two phases over recent years. The theory was that the public service needed to attract or retain the best talent by aligning salaries with comparable roles in the private sector. This ‘theory’ is very questionnable because:

- The public service can never be simply run like a business centered on the ethos of profit maximisation
- The inequality between the lowest and highest paid is a justifiable source of resentment among many {the ratio of salaries for Clerical officer staff (civil service, health and local authority) to CEO/Secretary-General has increased from a factor of 3.5 to 5.6 (Source: IMPACT/CPSU – Table 10) between 1987 and 2006 for examples}.
- Public service labour markets remain rigid and internal so that inflows from the private sector at the very top level is the exception more than the rule.
All of this needs to be borne in mind since the ‘premium’ to public sector workers compared to private sector workers is one dimension. The other ‘premium’ is the ratio of high to low earners in each of the public and private sectors has gone up appreciably over time. I don’t see too many people in the economic fraternity complaining about that. (We have seen obscene growing differences in regard to top remuneration in Corporate USA and mimicked in Ireland and elsewhere from the era of ‘greed is good’ onwards.)

What sorts of factors has driven salaries in recent decades and what implications can be drawn for incomes policy? I will look at these issues as well as some evidence on non-wage incomes, in a subsequent 'Mythbuster'.


Jane Gray said...

Very interesting. I hope you will also look at the distribution of gender-typed occupations in the public and private sectors. In addition to the fact that you are more likely to find unionized manual workers in the public sector (as Michael Taft pointed out in an earlier comment), I suspect that the public service accounts for a high proportion of professional 'women's jobs' - that is female-typed occupations where there is a relatively high return to education (nursing, teaching). And of course we know that labour markets are highly segregated by gender. This is why I suspect that the kinds of models used by the authors you cite are not sufficiently complex to capture wage differential between the public and private sectors.

Slí Eile said...

Thanks for these obs. The model used in both papers appears to be standard - OLS regression with a wide range of explanatory variables including gender, occupation, size of enterprise and, in the case of the Kelly et al. paper, membership of a professional association. Hence, the relevant factors you mention - high education and predominantly more women are factored into this analysis. It seems that the net remaining point is the impact of public sector union membership and its role in negotiating - effectively - stable and increasing wages in recent years. Where some private sector workers has lost out is the way that low wages have been traded for tax cuts over the last two decades leaving many such workers in a relatively low paid position compared to public sector workers (and private sector workers abroad - a point that will be looked at in future blogs.

Jane Gray said...

My point was, though, that OLS models that treat gender as a variable cannot explain wage variations that occur through processes of labour market segmentation. Put simply, I suspect that most female employees working in sex-typed occupations in the public sector are different from most female employees working in sex-typed occupations in the private sector on a whole range of dimensions - professional status, work history and so on. So to figure out whether or not there is a public sector pay premium one would have to figure out how working in the public/private sector interacted with working in female/male labour market segments (amongst other interactions).
I suspect that a significant part of the public sector wage 'premium' may be accounted for by the fact that most good 'women's jobs' are found there. That's still a premium, of sorts, but it does put rather a different complexion on things.

Slí Eile said...

Certainly public and private sectors vary in regard to the proportion of women employed, their education level as well as concentration in particular grades where, as you correctly point out, segmentation may occur. Obvious examples include Clerical Grade staff in central and local government as well as nurses. The point of multi-variate regression analysis such as that used in the papers cited is that it controls for differences in characteristics such as gender, education etc. While there are proportionately more women employed in the public sector (e.g. primary teachers, nurses, Clerical officers) their average educational attainment level is considerably higher. However, when comparing third level graduates women (e.g) in public and private sectors there is a clear public sector advantage in pay. On average women third level grads earn about one third more than men without third level in public and private sectors each. The premium to women when all other factors are controlled for is still around 24% as it is 23% for men.
None of this is to suggest that public sector workers are over-paid or that their wages should be cut. Starting from the facts and understanding how public and private sector workers differ is important.

Jane Gray said...

There may be a limit to the fruitfulness of our debate, but I continue to disagree with the assertion that multivariate regression can, by itself, account for wage differentials in segmented labour markets, where occupations are sex-typed and there is relatively little mobility between segments. My hypothesis is that, compared to in the public sector, being a woman may be more disadvantageous in the private sector, where proportionally more women may be confined to 'secondary' occupations. If this is true, treating gender as one variable amongst others is not going to capture the processes giving rise to wage differentials; you would need some kind of hierarchical modelling that took account of the fact that the processes in question were occurring both at the level of individuals and at the level of labour market segments.
Not everyone agrees with the theory of labour market segmentation, of course, but it is intriguing to me that the argument about the public/private wage gap is fundamentally an argument about segmentation. The idea is that public sector workers are able to use methods of 'closure' - union representation, social partnership, internal promotion rules, unspecified 'favouritism' on the part of government - to garner higher wages than in the private sector. But of course this fails to take account of the fact that labour markets are multiply segmented (across both public and private sectors) - and the statistical model used to supposedly test the claim about a public sector wage premium assumes that there is a single market for labour!

Slí Eile said...

Thanks for your comments. This merits more research and discussion. Certaintly some degree of market segmentation is present with very internal labour markets at work. As you point out some segments are predominantly male/female. What is not clear is how this explains a general pattern of wage differential when comparing similar occupations, levels of education and gender between public and private sector. Perhaps different modelling as you say is required.