The Canadian Federation of Independent Business (CFIB) has received a lot of publicity for their report claiming that wages of public sector workers far exceed pay levels in the private sector for all levels of government and in all regions of Canada. Columnists and media outlets in Canada have frequently reported these claims as fact, with little or no critical analysis.
These claims include:
- “Government and public sector employees are typically paid between 8% and 17% more than similarly-employed individuals in the private sector”i.
- “When benefits are included, government employees earn 25% to 40% more than their private-sector counterpartsii.
- City workers enjoy generous wages, “lavish benefits” and are a “pampered bunch”iii.
- “Public sector employers have a combined wage and benefits bill that is $19 billion higher than if they had kept costs to private sector norms.iv”
However, the CFIB’s Wage Watch study—the basis for these claims—is flawed and distorts the truth in a number of different ways.
The study manipulates the data by selectively excluding more than half of the public sector workforce and using methods to exaggerate wage differences for the remainder. However, since the CFIB doesn’t make any of the underlying data or wage comparisons at the occupational level available, it is impossible to determine how much this has skewed the results without completely replicating their study.
In addition, the study makes no adjustment for other major factors that influence wage levels: education, skills and experience, unionization, the size of the employer, and for the averages, the size of the city where the workers live. Each of these factors on their own could explain the difference in wages, but they were not taken into consideration.
Any economist or social scientist knows that if you ignore key factors such as these in the analysis of aggregate data, you get invalid results.
- The analysis excludes many occupations and workers, covering only a third of the total public sector workers. It includes more lower-level occupations where the public sector generally provides higher wages, while excluding many higher paid management and executive occupations where pay is considerably higher in the private sector.
- The CFIB study further distorts the results by using median earnings, instead of average earning for the occupationsv. A federal Treasury Board study that replicated a previous version of the CFIB study for federal government workers found that the use of medians instead of averages magnified the “wage premium” by an average of over 50%
- No adjustment is made for education, skills or experience within occupation groups. Public sector workers often have higher levels of education and experience than workers in similar occupations in the private sector, but this wasn’t accounted forvi. The federal government’s analysis of a previous Wage Watch study stated: “The fundamental flaw in these broad or ‘macro’ comparisons is the failure to take into account the different occupational or job compositions.vii”
- A large part of what the study really reflects is not a public sector wage premium, but a union wage premium. Wages for unionized workers are an average of 10% to 50% higher than wages for non-unionized workers, with greater benefits for women and relatively lower-waged workers (who appear to be over-represented in this study). Because labour unions represent 74% of public sector workers and only 18% of private sector workers, these greater benefits have a big impact on public and private sector averages.
- No adjustment is made for the size of the employer. It is well known that wages, salaries and benefits for larger employers tend to be higher than those provided by small employers. Weekly earnings are an average of 18% higher for those employed in workplaces of 500 or more compared with those in workplaces of less than 500. About 70% of public sector workers are employed in workplaces of 500 or more, compared with about 37% of private sector employeesviii.
- The comparisons between public and private sector wages are skewed again by the fact that public sector employment is more concentrated in larger cities where the cost of living and wages are generally higher. For instance, over 55% of the federal public sector workers included in the CFIB study lived in five of the largest cities in Canada, while less than 25% of the private sector workers they were compared to lived in these cities. Statistics Canada calculates the basic cost of living for large cities in Canada as 20% higher than the cost of living in medium-sized cities, 33% higher than the cost of living in smaller urban areas, and 53% higher than the cost of living in rural areasix.
- The calculation of the additional cost of paid and unpaid benefits—which the CFIB says adds another 7% to 24% to the public sector wage premium—is very roughly calculated and based on aggregate nation-wide estimates. It has little or no relevance to any of the specific examples and should not be applied to them. Most of the additional benefits are in the form of pensions. The financial crisis has clearly demonstrated the dangers of depending on private investments for retirement security: even conservative experts now admit we need to strengthen, and not weaken, public pensions.
Instead of using averages, lumping diverse jobs together, and accounting for these other factors (or manipulating the data and ignoring these other factors as the CFIB study does), a much more accurate way is to make direct comparisons for very specific jobs.
The Institute of Statistics of Quebec does this analysis on an annual basis, making detailed data publicly availablex. Its annual surveys have found that provincial government salaries have on average trailed other salaries, including those in the private sector, for directly comparable jobs. The latest survey found that provincial government salaries were similar to comparable non-unionized jobs, but that their total compensation, including benefits was about 11% higher. In comparison with other unionized jobs, Quebec government salaries trailed both in terms of salaries and total compensation.
Other direct comparisons of specific jobs by private human resource firms—such as Morneau Sobeco, Watson Wyatt and the Hay Group—for the federal government have had generally similar findings: no evidence of consistently higher public sector wages, with some studies showing public sector wages lagging for most jobsxi.
One common finding is there is much greater equality of wages in the public sector than in the private sector: with relatively fewer poorly paid workers and fewer very highly paid executives and professionals. Interestingly, the only area where the CFIB study says public sector wages are too low is for senior public servantsxii.
The CFIB study is based on the assumption that public sector wages and salaries should all follow “private sector norms.” These private sector norms are a world where top executives in Canada take home on average 259 times as much as the average Canadian is paid, 596 times the minimum wage and where women are paid, on average, only 84% of what men are paidxiii.
Given that the public has strongly supported increasing minimum wages and is often appalled by high levels of executive pay, the more equitable range of pay provided in the public sector more likely represents our society’s social norms than the large and growing inequalities of the private sector: a world that the CFIB wants the rest of us to slavishly imitate.
[i] CFIB Wage Watch, December 2008; repeated in various news stories. www.cfib.ca/research/reports/rr3077.pdf
[ii] Margaret Wente columns in Globe and Mail, June 9 and July 16, both on page A17 and CFIB Wage Watch.
[iii] Catherine Swift, quoted in the National Post, Frustration grows over public vs. private pay; Reverse Robin Hood p. A1, July 2, 2009; also quoted in CFIB open letter, July 21, 2009. www.cfib.ca/en/media/releases09/on072109.asp
[iv] CFIB Wage Watch December 2008, P. 1. www.cfib.ca/research/reports/rr3077.pdf/>
[v] The median is the middle observation of a series of variables. The “mean” or arithmetic average is the more common way of calculating averages: the sum divided by the number of observations, or in the case of this study it would be total wages and salaries divided by the number of employees. For example, consider a comparison of two series with three observations: A with (5, 6, 8) and B with (4, 5, 12). For Series A the median is 6 and the mean average is 6⅓ (= (5+6+8)/3). For Series B, the median is 5 while the mean average is 7 (= (4+5+12)/3). In this example, the median of A is higher than the median of B. At the same time the opposite is true for the mean average: B has a higher mean average than A. Sectors with more equitable wage distributions almost always have higher median wages than sectors with less equitable wage distributions but similar average wages.
[vi] The study includes some adjustment for differences in average wages between the public and private sectors, but this is not the same as accounting for differences in education or skill.
[vii] See chapter 6 in Treasury Board of Canada Secretariat Expenditure Review of Federal Public Sector Compensation Policy and Comparability www.tbs-sct.gc.ca/report/orp/2007/er-ed/er-ed-eng.asp
[viii] Calculations made using 2008 data from Tables 9 and 10 in Statistics Canada’s Employment, Earnings and Hours. www.statcan.gc.ca/pub/72-002-x/2009004/tablesectlist-listetableauxsect-eng.htm
[ix] The calculation of the basic cost of living comes from the estimates of the low-income cut-offs, which measures the after-tax income at below which most Canadians spend at least 20 percentage points more than the average on food, shelter and clothing. Large cities are defined as urban areas with populations of 500,000 and higher, medium-sized cities are 30,000 to 100,000 and 100,000 to 500,000, and smaller urban areas are those with less than 30,000. www.statcan.gc.ca/pub/75-202-x/2007000/t098-eng.htm
[x] The 2008 report of Rémunération des salariés. État et évolution comparés is available at: www.gouv.qc.ca/portail/quebec/pgs/commun/actualites/actualite/actualites_081128_wages/?lang=en and comparative salary data by occupation are available to view at: www.stat.gouv.qc.ca/donstat/societe/march_travl_remnr/remnr_condt_travl/emploi_repere/index_an.htm
[xi] For a review of some of these studies, see Treasury Board of Canada Secretariat Expenditure Review of Federal Public Sector Compensation Policy and Comparability www.tbs-sct.gc.ca/report/orp/2007/er-ed/er-ed-eng.asp
[xii] Wage Watch, December 2008, page 5. />www.cfib.ca/research/reports/rr3077.pdf
[xiii] Hugh Mackenzie, Banner Year for Canada’s CEOs, Canadian Centre for Policy Alternatives, December, 2008. www.policyalternatives.ca/reports/2009/01/reportsstudies2057/2008 Labour Force Survey for average hourly earnings by gender.