Printer-friendly version (HTML)

Comparing A Firm’s Occupational Wage Patterns with National Wage Patterns
by John E. Buckley
Bureau of Labor Statistics

Originally Posted: September 29, 2006

This article explains how employers can use National Compensation Survey (NCS) data to compare the wage structure in their firms with occupational pay relationships at the national level.

The National Compensation Survey (NCS) publishes a wide variety of occupational wage data for selected metropolitan and nonmetropolitan areas, for broad geographic regions known as census divisions, and for the Nation.1 Occupational wage estimates also are presented by selected worker and establishment characteristics such as the following:

  • Full- or part-time status


  • Union or nonunion status


  • Paid on a time or incentive basis, and


  • Establishment employment size.

At the national level, NCS publishes estimates for more than 400 detailed white-collar, blue-collar, and service occupations. In 2006, NCS published average hourly earnings estimates for full-time workers in 401 detailed occupations in private industry.2 (See table 1.) Using NCS data, an employer can determine how the firm's wage structure compares with occupational pay relationships at the national level by completing a set of simple calculations. Smaller firms, which often lack the specialized expertise of wage and salary administrators, are more likely than larger firms to benefit from the method described here. While firms typically prefer to use local area or industry-specific levels for wage comparison purposes, using national data helps smooth out anomalies that may appear at the local level due to a smaller sample available for the local estimates.

An illustration

The illustration that follows shows how to calculate and compare occupational wages at the national level with those of a hypothetical firm. The first step is to select the occupations of interest. In this example, data entry keyers are used as the occupation with which the others are compared. As table 2 shows, messengers and janitors nationwide earned about 81 percent and 95 percent, respectively, of the earnings of data entry keyers, and payroll clerks and computer programmers earned 140 and 263 percent, respectively, of the average earnings of data entry keyers.

Table 2. Comparing national pay levels with levels in a hypothetical firm, selected occupations, June 2005
Occupation (1)
National average hourly pay rate (NCS data)
(2)
NCS pay comparisons (Data entry keyers = 100)
(3)
Hourly pay rate, firm X
(4)
Firm X's pay comparisons (Data entry keyers = 100)

Data entry keyers

$11.94 100 $17.25 100

Messengers

9.68 81 12.01 70

Payroll clerks

16.68 140 20.01 116

Janitors

11.32 95 12.93 75

Computer programmers

31.45 263 32.23 187

The NCS pay comparisons shown in column 2 were produced by dividing the national average hourly pay rates of each of the selected occupations by the hourly pay rate for data entry keyers, multiplying by 100, and rounding to nearest whole number.3 Computing pay relationships in the same way for a hypothetical firm ("firm X") produces a second set of comparative values. The hourly pay rate of data entry keyers within this hypothetical firm (column 3) is used as the basis of comparison for other occupational pay rates within the same firm (column 4). While earnings of computer programmers at the national level and at firm X are higher than workers in the other 4 occupations, the pay advantage of computer programmers over data entry keyers in firm X (87 percent) is less than the comparable advantage of computer programmers nationwide (163 percent).

The choice of which occupation to use as the "base occupation" is arbitrary: it might be selected because it's the largest occupation in the company, for example, or because it’s the first occupation on the company’s payroll; it might even be chosen at random. The choice merely provides a starting point for discussion. If the gap between the highest and lowest paid occupations is substantially different from what the NCS suggests is typical for those occupations (based on the pay relationships between similar occupations at the national level), the company might consider raising, lowering, or freezing the pay of some of its occupations or doing a combination of these things.

A note of caution is needed here. The method described is intended to be simple; consequently, it should be viewed as a rough tool rather than a precise mechanism for making decisions. For example, this method does not take into account all the factors that should determine a computer programmer’s hourly pay rate at firm X. The national estimate for this occupation includes entry level, mid-level, and senior programmers in small and large establishments in both high- and low-paying areas, while an individual firm might have a different mix of computer programmers.

These are some of the factors that users should keep in mind when making comparisons. The NCS publishes data for different levels of skill within each occupation. In the 2005 national bulletin,4 for example, six levels of computer programmers are presented, with average earnings in private industry ranging from $21.84 to $48.03 per hour. The bulletin includes information on how firms can determine the level of work of their own jobs, which may allow for more precise comparisons.

When comparing the earnings at an individual firm with those at the national level, users should consider such factors as employees’ length of service and special skills. For example, a particular firm might find it worthwhile to pay its data entry keyers more than its payroll clerks, although the latter earn more, on average, at the national level. In addition, as part of the decision-making process, users should consider the precision of a published estimate, as measured by its relative error.

Reliability of the data

Because the NCS is a sample survey, its estimates are subject to sampling errors. A measure of the variation among these differing estimates is called the "standard error" or "sampling error." The standard error indicates the precision with which an estimate from a particular sample approximates the average result of all possible samples. The relative standard error is the standard error divided by the estimate.

The standard error can be used to calculate a "confidence interval" around a sample estimate. For example, payroll clerks at the national level earned, on average, $16.68 per hour, with a relative standard error of 2.3 percent. Thus, at the 90-percent level, the confidence interval for this estimate is $16.05 to $17.31.5 If all possible samples were selected to estimate the population value, the interval from each sample would include the true population value approximately 90 percent of the time.

Table 1 includes the relative standard errors for all of the listed occupations. Smaller relative standard errors indicate that the true population value is likely to be found in a narrow range around the estimate. Because of sampling errors, small differences in reported averages should not be used to evaluate differentials. For example, the rate for an occupation averaging 97 percent of payroll clerks ($16.68 x .97 = $16.18) would fall within the confidence interval for payroll clerks ($16.05 to $17.31). That means that small differences in averages are not significantly different.6

 

John E. Buckley
Economist, Division of Compensation Data Analysis and Planning, Bureau of Labor Statistics.
Telephone: (202) 691-6299; E-mail: Buckley.John@bls.gov

 

Notes

1 The NCS sample consists of 152 metropolitan and nonmetropolitan areas representing the Nation's 326 metropolitan statistical areas (MSAs), as defined by the Office of Management and Budget in 1994, and the remaining portions of the 50 States. Data are published for about 90 of these areas each year.

2 Data were collected between December 2004 and January 2006. The average reference period was June 2005. For table source, visit the NCS website on the Internet at http://www.bls.gov/ncs/; supplementary table 2.2 presents data for full-time workers in private industry.

3 For example, to compare the pay of messengers and data entry keyers in column 1, divide $9.68 by $11.94, multiply by 100 and round. ($9.68 / 11.94 = .8107; .8107 x 100 = 81.07 = 81 rounded.) For ease of analysis, absolute earnings were converted to relative earnings.

4 National Compensation Survey: Occupational Wages in the United States, June 2005, Bulletin 2581 (Bureau of Labor Statistics, August 2006).

5 The confidence interval for payroll clerks is calculated as follows: $16.68 plus or minus 1.645 times 2.3 percent of the mean [that is, 1.645 x .023 x $16.68 = $0.63]; ($16.68 - $0.63 = $16.05; $16.68 + $0.63 = $17.31).

6 For more information on data reliability, see National Compensation Survey: Occupational Wages in the United States, June 2005.