Treasury Board of Canada Secretariat
Symbol of the Government of Canada

ARCHIVED - Expenditure Review of Federal Public Sector - Volume One - The Analytical Report and Recommendation


Warning This page has been archived.

Archived Content

Information identified as archived on the Web is for reference, research or recordkeeping purposes. It has not been altered or updated after the date of archiving. Web pages that are archived on the Web are not subject to the Government of Canada Web Standards. As per the Communications Policy of the Government of Canada, you can request alternate formats on the "Contact Us" page.


6. Studies Comparing Federal Compensation to the Private and Broader Public Sectors

In the first part of this chapter, we look at three studies that provide overall comparisons between compensation in the federal public sector and the external economy. In the second part we review studies making such comparisons in relation to particular occupational groups or domains.

Comparing the federal public sector to the external labour market

We examine two types of studies, the first using data from the 2001 Census, and the second using Labour Force Survey data from the late 1990s and early 2000s. The latter study also includes the use of econometric modelling to generate comparisons while correcting for differences in workforce characteristics.

CFIB studies using Census data

The Canadian Federation of Independent Business (CFIB) published the first and most general study in our sample.[67] Based on the 2001 Census, it is the fourth in a series using Census data from 1986, 1991, 1996 and 2001. As reported by the CFIB, the main findings of the most recent edition as they relate to the federal public sector are:

Wage premium

Federal employees in public administration enjoy a 15.1% wage premium over their private sector counterparts–23.3% when non-wage benefits are included.

A significant overall increase in the wage premium favouring federal employees, over the 1995–2000 period is found ... Although federal public wage premiums fell to 8.9% by 1995, they dramatically increased to 15.1% in 2000.

By way of comparison, provincial employees in public administration enjoy a 9.1% wage premium over their private sector counterparts–14.8% when non-wage benefits are included ... Over the 1995–2000 period, provincial wage premiums decreased below those observed in 1990.

Regional variances

Federal public administration wage premiums substantially differ across the country ... from a high of greater than 25% in New Brunswick and Manitoba, to a low of 7.1% in Alberta, 11.5% in British Columbia, and 13.5% in Ontario.[68] At the level of metropolitan regions, the premium was reported as greatest in Winnipeg (25%), and smallest in Calgary (2.1%).

Population

In addition to the significant increases in federal public administration wage premiums since 1995, the federal public service has increased its workforce by over 20% from 1998 to the first quarter of 2003–reversing the cuts made in the mid 1990s.

The corresponding municipal premiums were 11.4% on wages, and 14.2% when non-wage benefits were included. The municipal wage premium was down from 11.8% in 1995.

Defining occupations

The CFIB methodology compares narrowly defined occupations that can be found readily in both the private and public sectors. Excluded are occupations that exhibit "excessive wage differentials"[69] between the sectors. Only full-time, full-year employees were included. At the national level, 257 occupations were included out of a total list of 514 occupational groups defined in the Standard Occupational Classification system. In deriving an aggregate estimate, occupational results were weighted by the relevant populations.

As a broad gauge of changes in relative compensation in particular, such studies as those published by the CFIB are interesting, particularly in highlighting trends that need to be understood. These studies provide a touch point, raising questions that deserve deeper examination. Moreover, they remind us that in comparing sectors, total compensation needs as much attention as salaries alone.

However, CFIB-type studies assume that the skill and experience requirements for the various occupations are the same in different worlds, which is unlikely to be true. Studies of this type have been published for many years. A 1984 internal Treasury Board Secretariat publication commented thus:

Frequently, comparisons of earnings, pay increases and compensation generally are made between the public service and other sectors of the Canadian economy ... These comparisons do not rely on matching job content ... They focus on a broad occupational grouping such as economists, engineers, or clerks ... The fundamental flaw in these broad or 'macro' comparisons is the failure to take into account the different occupational or job compositions in the constituencies being compared ... There is no valid reason why pay or compensation, or rates of increase for a broad occupational grouping such as engineer ... should be the same in different industries or organizations if the skill requirements within these occupations differ markedly among industries and organizations.[70]

As we will see later in this section, the more the specifics of particular jobs are controlled in comparing compensation, the more so-called "premiums" decrease. Accordingly, the macro nature of the CFIB studies suggests that we must interpret them with caution.

Furthermore, the private sector includes very diverse employers. It is reasonable to include employers of all sizes and types in macro assessments but when it comes to interpreting the results, we need to take into account that the federal government is a large, mainly unionized employer. It will likely always provide better salaries, benefits and job security than the majority of small and medium private businesses, as do larger, unionized private employers in the main.

Comparing median versus average income

A second critique relates to CFIB's selection of median income as the basis for comparison between sectors. The rationale is this:

Since the arithmetic average is subject to instability caused by extreme outliers, the median employment earnings of each occupation are used in calculating wage premiums.[71]

While there is merit to this viewpoint, it results in a larger wage premium than that derived from calculations using average salaries. Using a methodology similar to that employed by the CFIB study, the Treasury Board Secretariat derived a national federal public sector wage premium of 9.6% using arithmetic averages. This result, which is more than one third smaller than the premium reported in the CFIB study, arises because the distribution of salaries in the public sector is much more egalitarian than in the private sector. Since wide income dispersion is a basic characteristic of the private sector, use of the median tends to magnify differences in comparing income levels between the two sectors.

Figure 1031 below illustrates that the gap in wage premiums as measured by average versus median earnings is quite striking in some regions, especially Alberta, Ontario and the National Capital region. Apparently the difference between the average and the median is most pronounced in Alberta and Ontario, parts of the country where the private economy is most developed and entrepreneurial.

Figure 1031
Average versus median wage premiums by region based on data from the
2001 Census

Display full size graphic

Average versus median wage premiums by region based on data from the

The 2000 pay-equity data anomaly

A third critical observation relates to an important accident of timing that appears to have artificially increased federal public service incomes reported in the 2001 Census. The year 2000, for which income was reported in the 2001 Census, was an unusual year for federal public sector earnings. As we note in Volume Two, about $1.3 billion was paid during 2000 in retroactive and interest payments to employees affected by the 1999 settlement of the Public Service Alliance of Canada pay equity complaint, as well as other wage settlements. Statistics Canada has confirmed to us that wages and salaries reported by federal government clerks in the annual Survey of Labour and Income Dynamics (SLID) show a definite anomaly in 2000 compared with the years immediately before and after.[72] Table 1032 provides the detailed evidence.

Table 1032

Average wages and salaries for clerks and other federal government employees, Survey of Labour and Income Dynamics, 1998 to 2001

 

Clerks

Non-clerks

All

Men

Women

All

Men

Women

1998

$31,201

$38,064

$28,302

$38,791

$42,861

$32,791

1999

$36,124

$40,020

$35,145

$42,567

$46,679

$35,599

2000

$45,570

$46,090

$45,418

$46,841

$48,852

$43,695

2001

$36,428

$40,218

$35,281

$47,522

$51,609

$40,792

Female and male clerks constitute more than 20% of the population of the combined core public service and separate employer domains. To remove the effect of the one-time pay equity payments in 2000, the annual wages and salaries for clerks in the federal government must be scaled down, using a ratio of 35/45 for women and 40/46 for men. With this correction, there is no longer an increase in the federal public service wage differential versus the private sector between 1995 and 2000. The correction makes the data for clerks equivalent to that of other public servants, for whom no transient salary increase is evident in SLID data.

Pursuing the analysis further, Statistics Canada calculated that if we do not correct for the receipt of one-time employment equity retroactive payments, the federal-private sector differential for clerks rises from 12% in 1990 to 14% in 2000. However, if we correct the data using the scaling factors noted above, the differential for this group falls from 12% in 1990 to 6% in 2000. The conclusion of this analysis is that there is little support for the notion that the federal-private sector wage differential has increased for workers in the group comprising secretaries and administrative officers, clerks and professionals. Finally, we note that if we apply the CFIB analysis by gender, the wage difference between the federal public sector and the private sector was much larger for female (18.1%) than for male employees (3.2%). In effect, then, the overall federal government "wage premium" mainly reflects relatively higher female wages, which in turn reflects in part various federal social policies. These include pay equity, as well as family-friendly policies such as income supplementation of Employment Insurance to 93% of annual income for up to a year for combined maternity and parental leave.

This insight raises an important issue in how we should interpret private sector compensation levels and practices as a standard for comparability. The private sector no doubt reflects market realities in determining salary levels. However, the market itself may reflect some discriminatory attitudes and practices that disadvantage women. To the extent this may be true, any equitable policy on comparability needs to take account of this phenomenon.

CPRN and Gunderson studies using Labour Force Survey data

A second type of general study uses Labour Force Survey data and econometric modelling to generate compensation comparisons that correct for differences in the workforce characteristics of the sectors being compared. The Canadian Policy Research Networks published such a study in 2000.[73] The Treasury Board Secretariat commissioned a follow-up study from Morley Gunderson in 2003.[74] Our principal focus is on the more recent study.

Among the most pertinent findings of these studies are the following points:

Higher earnings

After controlling for other wage-determining variables such as education, age, length of tenure, gender, coverage by collective agreement, and size of establishment, public sector employees "invariably earn more than do employees in the private sector who have the same endowment of other wage-determining characteristics."

In all three levels of government, the public sector premium increased over the 1997 to 2003 period. The increase was most pronounced in the federal government, going from 6.9% in 1997 (where it was in the middle of the public sector premiums) to 16.2% in 2003 (at the high end of the premiums).

"The federal government premium in 2003 ranges from 12.4% when controlling for collective bargaining coverage and 47 occupations, to 23% when not controlling for collective agreement coverage or any occupation groups."

Higher premiums for some groups

"The federal premium, as is the case for most elements of the public sector, is generally larger for females than for males, and for low-wage service jobs. It is lower (often negative) for trades, labour and primary occupations, and about average for managerial/professional occupations and for clerical/secretarial/administrative occupations."

Other public sector premiums

In descending order, the "pure public sector premiums were: utilities 17%, provincial government 16.7%, federal government 16.2%, local government 13.5%, other public 9.0%, health 7.4%, and education 6.7%."

In the CPRN 2000 study, Gunderson, Hyatt and Riddell discuss the relative merits of the Census and the Labour Force Survey (LFS) as data sources for comparing public and private sector earnings. Both are tried-and-true surveys with long histories and well-established reputations both within Statistics Canada and elsewhere. Both have a wide range of control variables, with the Census including indicators of visible minority or immigrant status, for example, that are not available in other surveys. However, for the purpose of this analysis the "class of worker" variable included in the LFS but not the Census is more pertinent.

Overall, Gunderson et al. prefer the LFS. Among their principal reasons are three points:

  • The LFS measures hourly wages and permits control for differences in hours worked, whereas the Census measures annual earnings.
  • The LFS is based on telephone interviews. With experienced interviewers, errors are less likely than with respondents filling in a survey form and returning it as with the Census.
  • Although the Census has many more observations than the LFS, the LFS provides current information monthly, whereas the Census data is not released for public use until three or more years after the period on which it reports.[75]

These two studies provide an intriguing empirical approach to mining complex data for comparative purposes. They involve developing econometric equations that allow the researcher to use the relevant Statistics Canada database to estimate the impact on earnings of selected variables such as age, length of tenure, education and unionization, among others. This method can be used to compare the government and private sectors or compare between industries. In this way, the researcher attempts to strip away factors that can legitimately explain differences in earnings between sectors. After controlling for these variables, if the wage difference favours the public sector this may be interpreted as a wage premium. In effect, the Gunderson studies assert that the government wage premium is an economic rent that is extracted by the public sector over and above the wages that are determined in the overall Canadian labour market.

Wage premiums in the private sector

As noted by Gunderson et al., general studies are not without important limitations. Most significant is the assumption that the econometric model constructed by the researchers includes all the key variables that determine market wages and can explain differences between the public and private sectors. As a way to test whether this is the case, Treasury Board Secretariat analysts replicated the Gunderson model and disaggregated the private sector. The purpose was to determine whether there are significant wage discrepancies between different parts of the private sector that are unexplained by the model's other variables.

Assuming that private sector wages are driven by an overall labour market, and that the model includes all the pertinent explanatory variables, wages differences within the private sector should be relatively small. Conversely, large and unexplained wage premiums between different parts of the private sector may indicate that key determinants of market wages are missing in the equations. Thus, the remaining difference between the public and the private sectors would not necessarily reflect a wage premium or that public sector wages are out of line with the market.

In fact, the Treasury Board Secretariat estimates showed that many of the wage differences between parts of the private sector were larger than the difference between the private sector as a whole and the federal public sector. Table 1033 presents the results of this analysis. We note that the professional/finance occupations in the private sector differ from the mining and oil sector and the retail sector, for example, by a greater amount than the federal public sector differs from the private sector as a whole. These results suggest that key factors other than those included in the Gunderson equations could explain the variations in the private sector, and might also contribute to an explanation of the gap between the public and private sectors.

Table 1033

Wage differences between areas in the private sector based on Labour Force Survey data*

Private sector area

% Wage difference
vs. professional/finance
using 10 occupational groups

% wage difference
vs. professional/finance
using 47 occupational groups

Professional/finance (ref. group)

Agriculture

-11.2

-4.8

Mining & oil

19.8

17.7

Construction

12.0

9.0

Manufacturing

0.4

2.0

Wholesale

3.5

-2.9

Retail

-14.5

-10.1

Storage

-0.6

-2.2

Miscellaneous

-14.6

-10.9

Federal Government

13.9

12.2

* Source: TBS Compensation Policy Division.

Closely matching occupations in the private and public sectors

Gunderson focuses on regression results using only 10 broadly defined occupational groups, although he calculates results using 47 occupational groups. Using the latter approach, the estimated federal public sector premium declines by almost 4 percentage points to 12.4%.

Gunderson et al. suggested that using 47 occupations could result in underestimating wage gaps because there may be a tendency in government to overclassify positions as a way of paying higher wages. This could be pertinent if vertical stratification were an important factor in the data, for instance, if data relative to junior and senior positions for the same types of jobs were reported as separate occupations at this level of disaggregation. However, a cursory examination of the data indicates that this is not the case. Moreover, almost 97% of federal government positions in the sample are concentrated in 20 of the 47 occupations. The Treasury Board Secretariat analysis found that using the 47 occupations reduces wage differences across private sector areas as well. This suggests that the closer you match occupations, the less there is a difference. These points support the view that using 47 occupations may better use all the data available in assessing the presence and size of any wage premium.

Measuring regional and seasonal variations in wage differences

Government wages are generally the same for the same work regardless of location. The Labour Force Survey (LFS) database, in contrast, disproportionately includes areas where private sector wages are relatively low compared to major metropolitan areas. This is because the LFS oversamples smaller jurisdictions[76] to report on them. Then, to allow the aggregation of observations coming from regions that are not sampled proportionately, the survey design assigns smaller sampling weights to oversampled smaller jurisdictions. The regression analysis in the Gunderson study did not take account of the sampling weights embedded in the LFS micro-data.

To assess the impact of ignoring sampling weights, Treasury Board Secretariat first repeated the unweighted regressions performed by Gunderson et al. and successfully replicated their results.[77] Then TBS ran regressions that incorporated the LFS micro-data sampling weights. The analysis yielded a narrowing of the wage premium:

  • Using April 2003 data, the wage premium for the federal government when controlling for 10 occupations was estimated at 14.5%, down 1.7 percentage points from the 16.2% estimate achieved using unweighted data. When controlling for 47 occupations, the wage premium narrowed to 10.9%, compared to 12.4% with the unweighted approach.
  • Similarly, when April 1997 data was used, the wage premiums associated with the models using 10 and 47 occupations dropped by 2.0 and 1.5 percentage points respectively.

We concluded that using the LFS sampling weights to make the data more representative of the regional distribution of the Canadian population reduced the reported wage premium significantly. The difference ranged from 1.6 percentage points when controlling for 47 occupational groups with April 1997 data, to 3.2 percentage points when controlling for 47 occupational groups with April 2004 data.

We also observed that Gunderson et al. only used LFS data for April in their analysis. To test how sensitive the analysis is to seasonal variation, we looked at October 2002, October 2003 and April 2004 data, in comparison with Gunderson's April 2003 data. Because only one sixth of the LFS sample is renewed each month, using sample periods six months apart provides samples with completely different observations. Table 1034 summarizes the results of the test of seasonal sensitivity. In summary, we found the following:

  • When controlling for 47 occupational groups, the estimated federal government wage premium using unweighted data declines from 12.4% using the April 2003 data as in the report by Gunderson et al., to 11.2% and 8.8% using the data for October 2002 and October 2003 respectively.
  • If we incorporate the sampling weights as well, the wage premium drops further, from 10.9% using the April 2003 data, to 8.7% and 6.6% with the October 2002 and October 2003 data respectively.

Table 1034

Wage differences between the federal government and the private sector, based on Labour Force Survey data for October and April in selected years using weighted and unweighted regressions

 

% Wage difference

 

Unweighted regressions

Weighted regressions

 

10 occupational groups

47 occupational groups

10 occupational groups

47 occupational groups

October 2002

14.3

11.2

11.6

8.7

April 2003

16.2*

12.4*

14.5

10.9

October 2003

12.4

8.8

10.2

6.6

April 2004

15.3

12.3

12.3

9.1

* Results reported in Gunderson's report. 

Wage premiums appear to be quite sensitive to seasonal factors. A likely explanation is that April falls in the peak tax return season, when the Canada Customs and Revenue Agency uses thousands of seasonal workers in clerical positions for which there is a wage premium that is relatively high compared to the private sector.

The studies by Gunderson et al., like the CPRN study based on the Censuses, raise important points deserving further investigation. Notwithstanding the methodological issues that they raise, the upward trend in any federal public sector wage premium identified in both the CFIB and the Gunderson studies must be taken seriously.

Institut de la statistique du Québec Study using job matching

L'Institut de la statistique du Québec (ISQ) [Quebec Statistical Institute] is mandated to compare the compensation of unionized employees in the provincial administration québécoise [Quebec public sector], which includes the public service, school boards and CEGEPS, hospitals and other health and social service providers, with that of other unionized and non-unionized employers in Quebec. The Institute's annual comparative study looks at average salaries adjusted for hours worked, as well as total compensation. The analysis is based on matching specific jobs through an annual survey of establishments with more than 200 employees. Separate comparisons with private sector and federal government employees are also presented in the report.

The Institute's 2003[78] report offers the following points relevant to the federal public sector:

  • Employees of l'administration québécoise were paid an estimated 9.1% less in salaries compared with l'administration fédérale, [federal government] and 7.5% less on total compensation.[79]
  • The relative change compared with the 2002 report saw the gap between l'administration québécoise and l'administration fédérale widen by 1.1% in salaries, and by 0.9% on total compensation.
  • L'administration québécoise trailed the Quebec private sector by 10.6% on salaries, and 3.5% on total compensation. For the unionized private sector, the gaps were wider at 13.9% on salaries and 10.8% on total compensation. For the non-unionized private sector in Quebec, the differences were much smaller, at 4.4% and parity respectively.

In its methodological appendix to the 2003 report, the ISQ provided an extensive description of how it compares jobs and compensation.[80] The comparisons are based on 60 benchmark jobs, representing 20% of the administration québécoise, in such areas as financial administration professional, engineer, laboratory technician, administrative support staff, motor vehicle mechanic and cook. For most jobs, two or three levels of complexity and responsibility were analyzed.

The forms of remuneration included in the total compensation assessment included salaries, pension contributions, insurance of all kinds and statutory programs. Bonuses, overtime and job security are not part of the analysis. Comparisons are made on the basis of employer expenditures for the various components of total compensation. The normal hours of work are taken into account in adjusting the comparisons. The federal public sector had the longest working hours among the groups examined by the ISQ.

Comparing federal wages to the private sector

The detailed data released in the ISQ study enable a comparison of wages for federal government employees and those of Quebec private sector employees. The overall comparison with this provincial private sector can be extended to the whole Canadian private sector by taking into account the salary differentials between the private sectors in Quebec and the whole country. Based on June 2003 Labour Force Survey data, Quebec private sector wages for firms with more than 100 employees was 5.5% below the national private sector average for the same type of employees.[81]

Table 1035 summarizes a Treasury Board Secretariat comparison of federal government wages and Quebec private sector wages using the ISQ data. The total wage differential and those for occupational categories are weighted averages based on the federal government's populations[82] for the jobs included in the survey. These data can differ substantially from those reported by the ISQ because they are based on federal government rather than provincial government populations. However, both the compensation data for the positions used to calculate these aggregates, and the job matching are from the ISQ's work.

Table 1035

Wage differences between federal government employees and comparable private sector employees in Quebec, 2003

 

Federal
gov.
emplt.

Federal
gov.
wage

Private
sector  wage

Wage difference

Federal
gov. tot. comp.
hourly

Private
 sector tot. comp.
hourly

Total
comp.  difference

Professionals

9,484

$62,052

$62,672

-1.0%

$45.79

$43.78

4.6%

Technicians

9,299

$52,044

$49,262

5.6%

$39.20

$34.84

12.5%

Office personnel

22,175

$36,987

$36,319

1.8%

$28.23

$26.45

6.7%

Service

962

$32,626

$29,258

11.5%

$22.51

$17.13

31.4%

Trades

1,638

$41,295

$54,568

-24.3%

$29.82

$37.93

-21.4%

Total

43,558

$45,725

$45,350

0.8%

$34.33

$32.24

6.5%

Source: Private sector wages, job matching and total compensation data from the Institut de la Statistique du Québec

Using this approach, federal government employees in 2003 earned on average 0.8% more than comparable Quebec private sector employees. The wage differential ranged from –24.3% for trades to a premium of 11.5% for service employees, most of whom are cooks and cafeteria workers. When taking account of the difference between provincial and national private sector wage averages, we can reasonably interpret these findings as indicating that federal government employees were on average paid about 5% below private sector employees in Canadian firms with 200 employees or more. 

However, when all other compensation benefits are included such as pension plan contributions by employers, and wages are adjusted to reflect the number of hours worked, the difference in compensation turns into a premium of 6.5% for federal government employees relative to Quebec private sector employees. The total compensation differential ranged from –21.4% for trades to 31.4% for service workers.

The main caveat with this comparison exercise is that the jobs used for comparison were limited by the ISQ data. The benchmark positions surveyed by the ISQ were selected to reflect the wide range of positions in the administration québécoise. These positions do not correspond to the job structure of the federal public service. Therefore, this wage comparison between the federal government and the private sector should only be seen as indicative of the actual wage differential between federal government employees and private sector employees

Overall comments on general studies

In this section we have reviewed three distinct approaches to comparing salaries and total compensation across sectors of the economy.

First, the most general comparison, a CFIB study derived from 2000 Census data, found a relatively large wage premium (15.1% on wages and 23.3% on total compensation) between the federal public sector and the Canadian private sector.

Second, Gunderson et al. reported a somewhat larger federal salary premium (16.2% on salaries), albeit for 2003,[83] based on Labour Force Survey data. However, this gap was derived when focusing on 10 broad occupational groupings. When a set of 47 ccupations was used, the wage difference between the federal public sector and the overall private sector declined to 12.4% for 2003.

Third, with the 2003 ISQ report, we have an assessment based on direct comparison of specific jobs. At this level of detailed comparison, we find a negligible difference between federal public sector wages and those in the Quebec private sector, and by extension, the Canadian labour market as a whole. Overall, these results suggest a general principle that the more rigorously we define the characteristics of particular occupations and jobs in comparing compensation, the less we find a general salary premium for the federal public sector.

Earlier in this chapter we concluded that there was no evidence of a general lag in federal public service salaries compared with the Canadian private sector. We came to this view based on two points. First, the 1992 Treasury Board Secretariat study calculation based on work by the Pay Research Bureau during its last years, that "on a national basis, average salaries in the federal public service were 8.3% behind those in the private sector." We view this as a worst case, with a strong likelihood that compared to the overall private sector–not just large unionized private sector employers–any lag would likely have been smaller than 8.3%. Second, we noted that federal public sector average salaries had increased more rapidly in the late 1990s and early 2000s than broad indicators of private sector salary increase. Putting these points together suggests that a small premium in favour of the federal public sector had emerged by the early 2000s.

While both the CFIB and the CPRN/Gunderson studies report a substantial federal public service wage premium in comparison with the Canadian private sector, we have seen that both studies may overstate the extent of any average salary gap. In the CFIB study, for example, using average instead of median salaries, and discounting for the one-time pay equity payments in 2000, accounts for most of the reported salary difference. For the CPRN/Gunderson study, we note that controlling for 47 versus 10 occupational categories reduces the wage premium by about one quarter. The ISQ, meanwhile, found very little federal public sector lead versus the private sector, based on analysis of specific job matches for 60 benchmark jobs. Taking all the evidence into account, we conclude that as of 2003 the overall federal public service wage premium was likely well under 10%.[84]

Importantly, however, all the studies report a relative increase over recent years in federal public sector salaries in comparison with the various comparator groups. For the CFIB study, the increase was from an 8.9% federal wage premium in 1995 to the reported 15.1% in 2000. For Gunderson et al. the growth was from 6.9% in 1997 to 16.2% in 2003. For the ISQ, the gap on salaries between the federal and Quebec administrations grew by 1.1% between 2002 and 2003 alone.

Whatever the actual absolute difference in salary levels between the federal public sector and other sectors, our own analysis in this report confirms a relatively rapid and unprecedented rate of increase in real average salaries in the federal public sector. If federal public sector average salaries continue to grow faster than salaries in the private sector or elsewhere in the economy, a substantial wage premium will certainly open up in favour of the federal public sector.

Macro studies are not definitive tools for assessing by how much one broad employment sector lags or leads other sectors. But they are clearly valuable as bellwethers of directional change. They can also serve to highlight issues requiring more in-depth analysis. In the end, however, we need to complement macro-analysis with careful studies of particular occupational groups. We now turn our attention to describing and assessing a sample of such studies completed in the past few years.

Studies comparing compensation for particular occupational groups

In this section we look at four types of compensation studies focusing on particular occupational groups:

The first is a joint employer-union study relating to the Operational Services (SV) bargaining unit.

The second is a study related to the Foreign Service (FS) bargaining unit carried out jointly on behalf of the union, the main employing departments, and the Treasury Board Secretariat.

The third is a study initiated by employing department Transport Canada in regard to the Aircraft Operations (AO) bargaining unit.

The fourth type involves two studies sponsored by the Treasury Board Secretariat as employer. The affected groups are the Computer Systems (CS), and the Executive (EX) groups.

The fifth relates to studies respecting RCMP compensation.

In concluding this section, we describe briefly the pilot compensation studies being undertaken on certain occupations by Statistics Canada on behalf of the National Joint Council's Joint Compensation Advisory Committee.

Union-employer joint study (Operational Services group)

The best example of a recent joint union-employer study of compensation for a particular bargaining group is the Morneau-Sobeco national compensation survey relating to the Operational Services (SV) group. This group includes a wide variety of trades-oriented occupations such as firefighters, ships' crews, hospital service workers, labourers, electricians, plumbers, carpenters, cleaner/janitors, mechanics and others. The study covered wages and benefits for a group of 31 job titles representing over half of the approximately 12,000 employees in the SV bargaining group.[85] The survey instrument was sent to 700 private and public employers across the country. A total of 172 organizations responded, about two thirds in the private sector, employing a total of 38,000 workers.

For each of the 31 job titles, Morneau and Sobeco provide mean hourly rates and the 10th, 25th, 50th (median), 75th, and 90th percentile rates, along with the number of incumbents included in the sample for that job title. The report documents in a similar fashion such matters as hours of work, vacation, and access to employer-supported insurance, pensions and other benefits.

The report does not compare its findings to the existing wage rates for the federal government's Operational Services employees. According to the Public Service Alliance of Canada:[86]

The average hourly wage for the 31 jobs surveyed by Morneau-Sobeco is $21.41 (when weighted in accordance with the population for these jobs in the federal government). The current average wage paid for these jobs at Table 2 [the union name for the SV bargaining group] is $17.78 an hour, which is $3.63 an hour behind the external market. In percentage terms, the external average is 20 per cent ahead of the Table 2 average."[87]

In an internal assessment, the Treasury Board Secretariat observed:[88]

Overall, the findings of the joint study suggest that for many positions, the Federal Public Service rates are comparable and even lead external markets (8 out of 31 positions).... For most positions, however, the Federal Public Service rates are below those of the external markets (23 out of 31 positions). It should be noted, however, that more than half of the job comparisons suggesting that the Federal Public Service is above market are to be interpreted with caution due to significant regional distribution discrepancies between the SV population and the population in the joint study,[89] which may create a bias.[90]

This internal assessment compared the Morneau-Sobeco findings with data from other sources such as the 2001 Census and the Fair Wages Survey regularly completed by the Labour Branch of Human Resources and Social Development Canada (HRSDC) relating to construction occupations. For 16 of the 31 jobs, the Treasury Board assessment found lower wages in one of the other sources, sometimes substantially lower. For example, for labourers the Morneau-Sobeco study found a 26.3% gap versus the external market, whereas the HRSDC survey reported only a 4.7% gap. For cleaner/janitors, interestingly, the joint study reported a 29.4% gap, but the 2001 Census wage information suggests federal public service and external market rates are comparable.

Another source of caution highlighted by the Treasury Board Secretariat is the reliance on information from organizations with fewer than 10 employees for 15 out of 31 jobs. Such unrepresentative sources introduce a significant risk of error.

Most fundamentally, however, the Morneau-Sobeco survey report does not explicitly come to terms with the reality that wages for trades-oriented jobs vary widely on a regional basis in Canada. The main report presents hourly rates for various percentile levels as already noted but there is no attempt in the study to weight wages regionally to align with the locations where the federal SV employees work. In fact, for 22 of 25 jobs paid according to pay zones, the survey sample population distribution varied by at least 10% in at least one zone from the distribution of the SV population.[91] For cooks, for example, only 23% of surveyed workers were in zone 2, while 70% of federal public service cooks worked there.

The Public Service Alliance of Canada (PSAC) has long been forthright in seeking to eliminate regional wage differences and they have achieved considerable success through collective bargaining over the years. However, there is no homogeneous Canada-wide external labour market for most trades. So logically, trades should either be paid regional rates coinciding to the local markets, or at least according to national or zonal rates reflecting the actual geographic distribution of federal public servants in the SV group.

Our review of the Morneau-Sobeco survey on the Operational Services group is instructive. This is clearly a serious piece of work. It certainly supports well the general notion that trades-oriented federal employees lag the external wage market.[92] However, determining with precision what that gap is can be debated at length. There is no doubt that the SV study contributes usefully to the debate but it does not resolve definitively the issue of comparability as it relates to the trades-related occupations in the public service.

Studies co-sponsored by the union, the employer, and the main affected departments

Here we review briefly two studies that were carried out in the past few years through cooperation among the relevant union, the main employing departments, and the Treasury Board Secretariat, the formal employer. The first study relates to the Foreign Service (FS) bargaining unit, and the second to the Aircraft Operations (AO) group.

Foreign Service study

Pricewaterhouse Coopers (PwC) conducted the Foreign Service study on behalf of the Department of Foreign Affairs and International Trade, Citizenship and Immigration Canada, the Treasury Board Secretariat and the Professional Association of Foreign Service Officers.[93] This was a wide-ranging study covering issues well beyond compensation per se, including the impact of rotational assignments abroad on Foreign Service officers' spouses and elderly parents, and recruitment and career progression.

On the compensation side, the report described how the Foreign Service (FS) pay ranges juxtapose with those of other federal public service classification groups such as Economics, Sociology and Statistics (ES), Commerce (CO), Program Administration PM), Law (LA), Administrative Services (AS), Career Assignment (CA) and Management Trainee (MM). PWC also reported the findings of a survey of compensation and other employment conditions for other foreign services, international organizations and private sector companies.

On the internal comparability side, the comparator groups identified had classification levels that reached pay maximums considerably higher than the FS group. Among countries that shared compensation data relating to their foreign service, five were reported as paying higher salaries than Canada, including two with salaries more than 60% higher; six had fairly similar salary levels; and three had somewhat lower salaries.

What is most striking about the FS study is its unclear focus. In effect, a wide range of information that might or might not be pertinent was assembled, but without any expressed theory of what comparators are most appropriate and why. For example, the study was unclear in choosing between internal equity and external competitiveness as the main driver of wage setting. In the case of Foreign Service officers, their work is much more like that of other public servants than any Canadian external groups. Still, it is not obvious to which public service analogue group we should compare Foreign Service salaries. Both the FS population and their actual work blend aspects of the work of economists, lawyers, commerce officers, program administrators and administrative service officers.

Comparisons to other countries' foreign services or to international organizations are, in principle, logical in regard to work content. However, there is little indication that care was taken to ensure accurate job matches; overly generic benchmark jobs cover a wide spectrum of potential job value and invite interpretation difficulties. In any case, each country has its own living standard and approach to remunerating public servants. The fact that United Nations employees or diplomats representing the United States or Switzerland[94] earn substantially more than Canadian diplomats has a certain propaganda value. However, it has little evident relevance in determining Canadian salaries. Moreover, using simple currency conversion to compare these or other salaries is overly simplistic. All countries base their compensation on labour market influences in the home country, which can vary dramatically from one country to another. At a minimum, it would be necessary to compare on the basis of purchasing power parity.

Aircraft Operations study

During the late 1990s, some specialized occupations became hot in the external labour market to the point where employing departments initiated compensation reviews to address problems in recruiting and retaining qualified staff. This was largely a result of the position of the Treasury Board Secretariat that it would consider temporary recruitment and retention allowances for particular groups, where a department could make a compelling case that it was experiencing problems to ensure an adequate supply of skilled employees.

One such example is the Pricewaterhouse Coopers study regarding the civil aviation inspector community commissioned in 1998 by Transport Canada.[95] The study was undertaken in the context of the expanding aircraft and airline industries of the late 1990s, which gave rise to general optimism that growth would take off. Within this set of expectations, PwC surveyed 25 air operators using three benchmark positions. The 14 companies responding were grouped into four tiers: national flag carriers, national charter carriers, regional carriers and commuter operators. Compared with the full set of respondents, the PwC survey found Transport Canada (Aircraft Operations group) salaries between the 21st and the 30th percentiles.

The Treasury Board Secretariat considered the PwC survey market data to be invalid and unreliable for several reasons. First, the job matching on the three benchmark positions, was considered weak. Second, TBS analysts felt the salary comparisons did not consider issues relating to extra duty pay, the existing terminable allowances, differences in benefits, and the less stressful working conditions of a government regulatory operation versus those of operational pilots employed by airlines. In summary, TBS argued that the survey did not meet a simple face validity test because the disparity between the AO positions and the broader market were purportedly enormous. If true, this should have resulted in an exodus from Transport Canada; however, substantial departures did not materialize.

As a result of basic disagreements on the PwC study's validity, negotiations with the AO bargaining unit were protracted, spanning several years. This made the vulnerability of such studies evident: the transience of hot labour markets. In the aftermath of September 11, 2001, and the subsequent restructuring of the international and Canadian airline industries, all of the forecasts for pilot shortages and rapidly increasing salaries were reversed as layoffs replaced recruitment and salaries were rolled back.

Studies conducted for the core public service

Under this rubric we look at two quite different studies commissioned by the Treasury Board Secretariat as the employer for the core public service. The first relates to the Computer Systems (CS) bargaining unit, and the second to the Executive (EX) classification group for whom the Treasury Board sets the salary ranges based on the recommendations of an external advisory group.

Watson Wyatt study for the Computer Systems group

In preparation for the 2002–2005 bargaining round for the CS bargaining group, the Treasury Board Secretariat asked Watson Wyatt to review published surveys on salaries in the informatics area.[96] Despite their inherent limitations, such studies can assist in positioning public service compensation in the context of the wider labour market. Based on job capsules describing succinctly the duties of 15 jobs widely used in the industry, Watson Wyatt gathered published survey data on compensation for comparable external jobs, with data from various sources adjusted to a common September 2002 date. The surveys included were those conducted by Watson Wyatt itself, Mercer, Morneau-Sobeco, the Toronto Board of Trade, and Towers Perrin.

The study looked at base salary and total cash compensation, which included the federal government's recruitment and retention allowances. On base salary, the federal public service was found to pay about the same as the external market for jobs at the first three classification levels of the Computer Systems group (i.e. CS 1 to CS 3), but less at the CS 4 and CS 5 levels. Looking at total compensation, Watson Wyatt found a federal public service premium at the lower three CS levels, and an increasing lag at the CS 4 and CS 5 levels.

It is important to emphasize that studies of this type essentially compare salary structures rather than actual salaries. For greater precision, Watson Wyatt provided information on salary ranges at the 25th, 50th and 75th percentiles, as well as the arithmetic average. Overall, Watson Wyatt concluded:

We are able to say that the [federal] public sector is not underpaying the CS1 to CS3 levels. However, the public sector is underpaying the CS4 and CS5 levels. This conclusion is based both on the base salary comparison and on the total cash compensation comparison.

Interestingly, this study had no discernible impact on the ensuing substantive negotiations with the Professional Institute of the Public Service of Canada (PIPSC), representing the CS bargaining group. The eventual settlement made no distinction in salary adjustments between CS1 to 3 and CS4 and 5. On the other hand, the existence of this study may have affected the process of negotiation, for example, by making PIPSC reluctant to proceed to a conciliation board, where they would have needed to counter the survey report's findings.[97]

Hay Associates study for the EX group

Finally, we report on a comparison of Executive (EX) compensation with that of comparable executives in the private and broader public sectors, completed by Hay Associates in 2001.[98] The study was commissioned by the Treasury Board Secretariat in support of the work of the Advisory Committee on Senior Level Retention and Compensation. The comparator market comprised two sub-markets:

  • The first is a group of 75 organizations in the broader public sector including municipalities, provinces, universities, school boards, colleges, health agencies, Crown corporations, government agencies, and not-for-profit organizations.
  • The second comprises 283 private companies in the industrial-and-financial category included in Hay Group's compensation data base.

The Hay system for evaluating the scope and challenge of jobs was used to ensure the comparison of jobs at the same level of difficulty. The compensation elements included in the analysis were:

  • salary,
  • bonus,
  • the actuarial value of employer paid benefits,
  • perquisites and paid time-off, and
  • the actuarial present value of any long-term incentive.[99]

The findings used median salaries for comparison.

  • Federal public service Executive salaries were within +/- 5% of those in the Canadian broader public service, except at the Deputy Minister (DM) 2 level, where there was a 13% lag.[100]
  • Compared with the Canadian private sector, there was a 10% public sector lag for comparable positions equivalent to the EX 1 level, with the gap increasing from level to level, reaching 42% at EX 5 and 84% at DM 2.[101]
  • The target percentages for performance bonuses in the federal public service were generally found to be half the level in the Canadian private sector, and equal to or below those in the Canadian broader public sector.[102]
  • Long-term incentives such as stock options were not available in the federal public service and were relatively rare in the broader public sector. They were, however, prevalent in the private sector, especially at higher levels of executive management. The median present value at the equivalent of the DM 2 level was 35% of salary.[103]
  • On non-cash benefits as a proportion of base pay, federal public service executives enjoyed a lead on the broader public sector, especially at the DM-2-equivalent level, and lagged the private sector somewhat.[104]
  • On total remuneration, the Hay Group found a lead ranging from 2% at the EX 3 level to 8% at the EX 1 level for the federal public service versus the broader Canadian public sector. Compared with the Canadian private sector, however, the federal public service lagged by progressively larger amounts, starting at 17% at the EX 1 level and rising to 147% at the DM 2 level.[105]

This is a credible study based on a proven and widely used methodology. There are some subjective elements that could be the subject of critique:

  • First, the Hay system for evaluating jobs in itself can be applied differently in different circumstances. For example, the Hay Group discounts federal public service point ratings by dividing by a factor of 1.113 in standardizing them for comparison with the ratings for jobs in their Canada-wide database.
  • Second, the methods used to evaluate non-cash benefits follow Hay standard methods, which could certainly be debated. Hay does not, for example, assign a value for retiree health and dental coverage, since it is "uncommon in the general market."

Nevertheless, the annual updates of this total compensation comparison for executives provides a consistent basis for the Advisory Committee on Senior Level Retention and Compensation to formulate recommendations on adjustments to federal executive compensation.

Mercer study for the RCMP

Starting in 1993 and annually since 1998, William M. Mercer has produced an annual comparison of total compensation among major Canadian police forces, for the RCMP Pay Council. As described in Chapter 11 of Volume Two, this survey covers six large municipal police forces across the country, the two provincial police forces, and the RCMP itself. The reports provide details on cash compensation, hours of work, leave, pension and savings plans, group benefits of all kinds, and other compensation, allowances and perquisites. This information is used to determine what salary and benefit adjustments should be recommended for approval to the Treasury Board.

This survey is strongly factual in approach. A small number of appropriate comparators are fully documented. There can be some question about whether the work of all these police forces is strictly the same. The RCMP, for example combines what could be called "classic" provincial and municipal policing, with a national role that goes beyond what other forces undertake, particularly in such areas as national security. Nevertheless, similarities are likely as great as can be established in other comparative studies.

General comments on studies of particular groups

In contrast with the general studies described in the previous section of this chapter, studies of particular groups can be very useful in principle for comparing apples and apples in regard to compensation. However, as we have seen in reviewing the very different surveys and studies analyzed in this section, there are no established approaches or standards that are applied consistently in undertaking such studies. As could be expected, the studies tend to be commissioned in the lead up to collective bargaining, or some other compensation determination process, rather than regularly scheduled. Most studies are ad hoc, undertaken once and not repeated, which reduces greatly their utility.

The need to recreate a standing capacity to conduct group-specific compensation studies relating to the federal public service, to assist in building a common information base for collective bargaining, has been recognized for some time. The Public Service Modernization Act, passed by Parliament in November 2003, provides for the establishment of a pay research unit in the new Public Service Labour Relations Board.

In recent years, the National Joint Council has sponsored a union-management Joint Compensation Advisory Committee (JCAC) to undertake pioneering work relating to systematic compensation surveying. Since 2001, in particular, the JCAC has contracted with Statistics Canada to prepare a pilot survey on compensation in private establishments with at least 50 employees, the federal public sector, and the broader public sector in Ontario.[106] The pilot expected to survey a sample of about 2,900 organizations, with 1,650 in the private sector and 800 in the non-federal public sector. The collection method involved personal interviews using a questionnaire for nearly 90% of the respondents, and the remaining, more distant, organizations were surveyed using paper-assisted telephone interviews.

It is too early to confirm how well the survey worked. However, this experience is encouraging in several ways. The approach was developed by Statistics Canada in active consultation with both Treasury Board and bargaining agent representatives. Anecdotally, it appears that those involved participated on a very professional basis. The cut-off of 50 employees was reasonable. Both unionized and non-unionized employers were included and the sampling method should result in representative results. Interviews in person or by telephone should result in reliable and consistent reporting.

On the other hand, the process has been excruciatingly long. From the start of joint work on the pilot until the first results are received will be at least three or four years. It is sensible and necessary to take the time to invest in sound methodology and confidence building in such an area. However, to be useful in the long run, it will be necessary for faster turnaround. It may also be feared that either the management or the union side may reject the process if the initial results are not to their liking.

With these remarks, we conclude our review of general and group-specific compensation-comparability studies, especially in relation to matters of salary and wages.