An official website of the United States government
The U.S. Bureau of Labor Statistics (BLS) Employment Cost Index (ECI) is a measure of the change in compensation per hour worked for U.S. workers.1 It is published quarterly. Like many statistical programs, the ECI program publishes seasonally adjusted estimates along with its unadjusted estimates. The factors used in seasonal adjustment reflect how a series is expected to change given its historical annual pattern. Removing the series’ expected seasonal movement helps isolate short-term changes beyond those driven by seasonality.2
This article discusses the effect of seasonal adjustment on ECI quarterly estimates. Because the ECI program does not seasonally adjust all of its published estimates, the article begins by documenting the series for which the program publishes seasonally adjusted 3-month percent-change estimates in addition to unadjusted estimates. It then shows that seasonal adjustment in the ECI generally has only a small effect on the 3-month percent changes in compensation for the most aggregate series, which cover civilian and private industry workers. This small effect is demonstrated both by comparing the seasonally adjusted estimates with the unadjusted estimates and by comparing the seasonal adjustment in the ECI with the seasonal adjustment in the BLS Current Employment Statistics (CES) estimates for average hourly earnings. The small seasonal movement in the ECI is largely by design. The ECI measures the rate of change in compensation while holding the composition of workers in industry and occupational groups constant. The article ends by highlighting some industry groups that exhibit relatively large differences between seasonally adjusted and unadjusted estimates.
Each quarter, the ECI program publishes estimates of 3-month percent changes in compensation for 132 different sets of workers. These sets, referred to as domains, are defined by ownership, industry, occupation, and/or characteristic group. The sets of private manufacturing industry workers and private industry workers in professional and related occupations are two examples of domains for which the ECI program publishes estimates. The term “characteristic group” refers to census region or division, bargaining status, or basis of pay. The ECI program publishes 3-month percent changes in total compensation and in wages and salaries for all 132 domains. In addition, it publishes 3-month percent changes in total benefits for 14 of the domains. This publication scheme results in a total of 278 percent-change estimates published each quarter.3
Table 1 summarizes the seasonal-adjustment eligibility of the 278 estimates. The ECI program seasonally adjusts only about half of those estimates. The program does not publish seasonally adjusted estimates for any domains defined by one of the characteristic groups (census region or division, bargaining status, or basis of pay).4 This exclusion eliminates 78 estimates from consideration for seasonal adjustment. Of the remaining 200 estimates, 136 are eligible for seasonal adjustment. For these eligible estimates, table 1 summarizes their coverage by ownership group. Civilian workers include workers both in private industries and in state and local government.
Characteristic | Number of published estimates | Number of estimates eligible for seasonal adjustment | Coverage of seasonally adjusted estimates |
---|---|---|---|
Series defined by industry and/or occupational group | |||
Civilian | 51 | 25 | Broad industry groups and education and health services industries |
Private | 122 | 94 | Almost full coverage, except for series defined by combinations of occupational groups and broad industry groups |
State and local government | 27 | 17 | Education and health services industries, public administration |
Series defined by census region or division, bargaining status, or basis of pay | |||
Civilian | 2 | 0 | None |
Private | 76 | 0 | None |
Source: U.S. Bureau of Labor Statistics. |
For series defined by ownership and industry and/or occupational group, the seasonal-adjustment coverage is most extensive for private industry estimates. The only major exclusions from seasonal adjustment are domains defined by both occupational and broad industry group. For example, the ECI program publishes estimates for four occupational groups in manufacturing industries, and none of these estimates are eligible for seasonal adjustment. Also, the estimates for nursing care facilities, ECI’s newest series (introduced in December 2013), are not eligible for seasonal adjustment. All other private industry estimates are eligible.
For state and local government, seasonal adjustment focuses on education and health services industries, which have significant government employment. Fourteen published ECI estimates for government cover workers in those industries and are eligible for seasonal adjustment. The major exclusions from seasonal adjustment are 10 published ECI government estimates for occupational groups. None of these estimates are eligible for seasonal adjustment.
The seasonal-adjustment coverage of the series for civilian workers largely mirrors that for state and local government workers. Several of the civilian series eligible for seasonal adjustment refer to education and health services industries, which employ significant numbers of both government and private industry workers. The other civilian series eligible for seasonal adjustment cover broad industry groups: manufacturing, goods-producing industries, service-providing industries, and public administration.
In summary, the ECI program seasonally adjusts only about half of its published estimates. However, the seasonal-adjustment coverage is extensive for industry and occupational groups in private industry. The coverage is less extensive for government and civilian estimates, although it does focus on education and health services industries, which have a high concentration of state and local government employment.
It should also be noted that seasonally adjusted ECI estimates are not published for all series eligible for seasonal adjustment. For these series to be published, they must also show stable and identifiable seasonality. Series are evaluated annually for conformity with this condition. As of the most recent (March 2023) revision to the ECI seasonal factors, 2 of the 136 series eligible for seasonal adjustment failed to meet the criteria for stable and identifiable seasonality; consequently, seasonally adjusted estimates for these series are not currently published.5
Chart 1 shows 3-month percent changes in the ECI for total compensation for civilian workers, presenting both seasonally adjusted and unadjusted estimates from March 2014 to December 2023. (For greater precision, the percent changes in this article are calculated by using unrounded index values.) As seen in the chart, seasonal adjustment smooths the movement of the series over time. However, the difference between the two series is relatively small, generally in the range of one-tenth to two-tenths of a percentage point each quarter.
Table 2 more fully demonstrates the small effect of seasonal adjustment on the estimates for civilian workers. The table shows the average difference between seasonally adjusted and unadjusted rates of change across all quarters from March 2014 to December 2023. A negative value in the table indicates that the seasonally adjusted rate of change is lower than the unadjusted rate; in such cases, the recurring seasonal trend leads to a larger quarterly change, which seasonal adjustment offsets. For example, seasonal adjustment reduced the December-to-March rate of change in the ECI by an average of 0.15 percentage point, offsetting the typically higher rate from December to March. For both wages and salaries and total benefits, seasonal adjustment reduced the December-to-March rate of change and increased the September-to-December rate of change. The effects of seasonal adjustment were typically smaller for the March-to-June and June-to-September rates of change.
Reference period | Total compensation | Wages and salaries | Total benefits |
---|---|---|---|
December–March | -0.15 | -0.11 | -0.24 |
March–June | 0.00 | -0.01 | 0.04 |
June–September | -0.06 | -0.08 | 0.00 |
September–December | 0.20 | 0.20 | 0.21 |
Source: U.S. Bureau of Labor Statistics. |
To provide further perspective on the size of the ECI seasonal adjustment, charts 2a through 2d compare the seasonal adjustment in the ECI for wages and salaries with the seasonal adjustment in another BLS wage series, the CES series for average hourly earnings. Also, to create a better match between the workers covered by the two programs, the charts use data for private industry workers. For this comparison, which is broken down into four separate charts to isolate the seasonal adjustments for March, June, September, and December, the size of the seasonal adjustment for the ECI is measured as the percent difference between seasonally adjusted and unadjusted index values, while the size of the seasonal adjustment for the CES is measured as the percent difference between seasonally adjusted and unadjusted average hourly earnings.6 For example, if the seasonally adjusted index value for the ECI equals 119.4 and the unadjusted index value equals 119.3, the size of the adjustment for the ECI is 100 × (119.4/119.3 – 1) = 0.08 percent. Similarly, if the seasonally adjusted estimate for CES average hourly earnings equals $24.32 and the unadjusted estimate equals $24.49, the size of the adjustment for the CES is 100 × (24.32/24.49 – 1) = −0.69 percent.
Two takeaways from charts 2a through 2d are apparent. First, the seasonal adjustments for the ECI for wages and salaries are much smaller than those for CES average hourly earnings. Across all 4 months presented in the charts, the seasonal adjustments for the ECI always remain below two-tenths of a percentage point. Second, upward adjustments to the ECI for wages and salaries are most consistently made for December. As a result, the ECI 3-month percent changes are seasonally adjusted upward for the September-to-December period and then seasonally adjusted downward for the December-to-March period. In contrast, upward adjustments to CES average hourly earnings are most consistently made for June. For the ECI, the estimates for June are adjusted slightly downward, with the adjustment becoming larger for more recent years, to offset stronger June growth in ECI wages and salaries.
The design of the ECI explains the difference between the seasonal patterns of the two series and why the ECI shows much smaller seasonal variation. The ECI is calculated with a Laspeyres index formula, which holds employment for industry and occupational groups constant. Thus, the ECI is a measure of the change in employer labor costs that is independent of the influence of employment shifts among occupational and industry categories. The measure of CES average hourly earnings equals the total employee payroll divided by total employee hours across all workers.7 Therefore, CES average hourly earnings are affected by month-to-month employment shifts among industries, which leads to a seasonal pattern that is considerably different from that for the ECI.
Although seasonal adjustment has a relatively small effect on the ECI 3-month percent changes in compensation for the most aggregate civilian and private industry series, it has a larger effect on the ECI published estimates for workers in some industry groups. The tables presented in this section highlight domains for which the ECI program consistently adjusts the 3-month percent change for one of the reference periods by half a percentage point or more. As in table 2, the size of the seasonal adjustment is measured as the average difference, by reference period, between the seasonally adjusted and unadjusted 3-month percent changes from March 2014 to December 2023. Negative values in the tables indicate typically larger changes in compensation that are adjusted downward.
Table 3 shows the average difference, by reference period, between seasonally adjusted and unadjusted ECI estimates for private industry workers in two industry groups in the finance and insurance sector: credit intermediation and related activities and insurance carriers and related activities. For both industry groups, the seasonal adjustments to the 3-month percent changes are downward for the December-to-March and March-to-June periods, offsetting recurring seasonal increases in compensation for these periods. The seasonal adjustments are then upward for the June-to-September and September-to-December periods, offsetting the smaller seasonal increases in compensation for these periods.
Reference period | Credit intermediation and related activities | Insurance carriers and related activities | ||
---|---|---|---|---|
Total compensation | Wages and salaries | Total compensation | Wages and salaries | |
December–March | -0.12 | -0.07 | -0.38 | -0.35 |
March–June | -0.94 | -1.06 | -0.54 | -0.50 |
June–September | 0.20 | 0.21 | 0.42 | 0.40 |
September–December | 0.88 | 0.95 | 0.51 | 0.47 |
Source: U.S. Bureau of Labor Statistics. |
Table 4 shows average differences between seasonally adjusted and unadjusted ECI estimates for civilian workers in the educational services industry group. As in the previous example, the seasonal adjustments to the 3-month percent changes for this group are relatively large. The largest adjustment is for the June-to-September period. The changes for the other periods are adjusted upward to offset the large downward adjustment for September. A similar pattern of seasonal adjustment exists for the detailed ECI series under educational services, such as the series for elementary and secondary schools in state and local government.
Reference period | Total compensation | Wages and salaries |
---|---|---|
December–March | 0.26 | 0.31 |
March–June | 0.32 | 0.25 |
June–September | -0.72 | -0.76 |
September–December | 0.15 | 0.19 |
Source: U.S. Bureau of Labor Statistics. |
As a final example, table 5 shows average differences between seasonally adjusted and unadjusted ECI estimates for private industry workers in utilities. As in the previous two examples, the seasonal adjustments for this industry group are relatively large. Because the largest seasonal increases in compensation for workers in utilities typically occur in March, the December-to-March percent-change estimates are downward seasonally adjusted by a little over half of a percentage point, on average. The estimates for June-to-September and September-to-December periods are both upward seasonally adjusted, to offset the typically smaller changes in compensation for these periods.
Reference period | Total compensation | Wages and salaries |
---|---|---|
December–March | -0.61 | -0.63 |
March–June | -0.06 | -0.02 |
June–September | 0.29 | 0.26 |
September–December | 0.36 | 0.38 |
Source: U.S. Bureau of Labor Statistics. |
Other domains for which the ECI program publishes seasonally adjusted estimates show smaller differences between seasonally adjusted and unadjusted estimates. Typically, these differences are similar to those for the aggregate civilian series.
The ECI program publishes seasonally adjusted 3-month percent changes in compensation for only about half of its published estimates. However, seasonally adjusted estimates are available for nearly all industry and occupational groups for which the program publishes a series for private industry workers. Seasonally adjusted estimates are also available for civilian and government workers in education and health services industries, which have a high concentration of state and local government employment. Seasonal adjustment generally has only a small effect on the 3-month percent changes in compensation for the most aggregate ECI series, although some published ECI series, such as those for workers in financial activities and educational services, show more substantial seasonal adjustment.
Michael K. Lettau, "Seasonality in the Employment Cost Index," Monthly Labor Review, U.S. Bureau of Labor Statistics, February 2024, https://doi.org/10.21916/mlr.2024.4
1 See John W. Ruser, “The Employment Cost Index: what is it?,” Monthly Labor Review, September 2001, https://www.bls.gov/opub/mlr/2001/09/art1full.pdf.
2 See E. Raphael Branch and Lowell Mason, “Seasonal adjustment in the ECI and the conversion to NAICS and SOC,” Monthly Labor Review, April 2006, https://www.bls.gov/opub/mlr/2006/04/art3full.pdf.
3 The Employment Cost Index (ECI) program additionally publishes estimates for 12-month percent changes in total compensation and in wages and salaries for 15 metropolitan areas. The program also publishes 12-month percent changes in the cost of health insurance for private industry workers. Because only 12-month percent changes are published for these series, they are not candidates for seasonal adjustment.
4 The ECI computation procedures for indexes defined by census region or division, bargaining status, or basis of pay use a modified ECI formula and differ from those defined by ownership, industry, or occupational group. See “National compensation measures: calculation,” Handbook of Methods (U.S. Bureau of Labor Statistics, December 15, 2017), https://www.bls.gov/opub/hom/ncs/calculation.htm. There is no theoretical reason preventing the seasonal adjustment of 3-month rates of change based on the modified ECI formula. However, in practice, the ECI program has not made these rates eligible for seasonal adjustment.
5 The two series that failed the criteria for identifiable and stable seasonality (as of the most recent revision of the ECI seasonal factors) are those for (1) private industry wages and salaries for workers in sales and related occupations and (2) total benefit costs in private aircraft manufacturing. For more information on the ECI seasonal-revision process and the procedures for determining seasonality, see E. Raphael Branch, “Changes in the publication of seasonally adjusted Employment Cost Index series,” Monthly Labor Review, March 2013, https://www.bls.gov/opub/mlr/2013/03/art5full.pdf.
6 The Current Employment Statistics (CES) program uses concurrent seasonal adjustment in which data up to and including the current month are used to develop the seasonal-adjustment factors. In contrast, the ECI seasonal adjustment forecasts seasonal factors four quarters into the future. However, this difference is unlikely to be a large factor affecting the difference between the two programs’ seasonal factors. See, for example, Jurgen Kropf, Christopher Manning, Kirk Mueller, and Stuart Scott, “Concurrent seasonal adjustment for industry employment statistics” (U.S. Bureau of Labor Statistics, 2002), https://www.bls.gov/osmr/research-papers/2002/pdf/st020110.pdf.
7 For the formula used to calculate the measure of CES average hourly earnings, see “Technical notes for the Current Employment Statistics survey” (U.S. Bureau of Labor Statistics), https://www.bls.gov/web/empsit/cestn.htm#section6.