Department of Labor Logo United States Department of Labor
Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Article
May 2022

The challenges of seasonal adjustment for the Current Employment Statistics survey during the COVID-19 pandemic

For its seasonally adjusted data, the Bureau of Labor Statistics (BLS) seeks to remove all fluctuations that are caused by the yearly cycle of seasons. The coronavirus 2019 pandemic posed a challenge to this task, because its effect was strong, across the entire economy, and lasted for at least several months. To ensure that the effects of the pandemic were not being incorporated into the seasonal adjustment factors, the Current Employment Statistics program took additional actions such as splitting the seasonal adjustment into two runs (prepandemic and postpandemic) and incorporating additional types of outliers into its models.

Most series published by the Current Employment Statistics (CES) program have a regularly recurring seasonal movement that can be measured from past data. Seasonal adjustment eliminates the part of the change attributable to the normal seasonal variation to help provide a better interpretation of trends without the influence of regularly occurring seasonal patterns. Previous publications from the U.S. Bureau of Labor Statistics (BLS) have discussed the challenges of seasonal adjustment during the coronavirus disease 2019 (COVID-19) pandemic. Those publications focused on the approaches BLS has taken to mitigate the effects of the pandemic on the real-time seasonal factors BLS uses to produce the seasonally adjusted series.1 Those discussions address how we use outlier detection, our most commonly used tool, to intervene in realtime when we estimate seasonal factors each month.

COVID-19 had a dramatic effect on the labor market. Unlike a strike or weather event, the effect was nationwide and, for some industries, sustained over a long period. BLS needed to carefully construct its seasonal adjustment models to ensure that the effect of the pandemic did not get incorporated into the models as a normal seasonal fluctuation. This article addresses how the CES program seeks to mitigate the effects of the pandemic on the 5 years of historical seasonal factors that are revised annually for the CES program. To preserve the accuracy of the historical seasonally adjusted data in response to the effects of COVID-19, BLS combined a sequence of two runs (prepandemic and postpandemic) as an additional intervention, along with the normal outlier-detection treatment, to produce the revised series for the 2020 annual review. Since the time of the 2020 annual review, BLS time-series researchers have developed responses that provide a more standardized approach to mitigating the effects of the pandemic. This article also addresses the procedure used during the 2021 annual review, which includes the use of additional outlier types discussed in the responses developed by these researchers.

Background

The CES program is a monthly survey conducted by the Bureau of Labor Statistics. The program provides employment, hours, and earnings estimates based on payroll records of business establishments. Data produced from the CES survey include nonfarm employment series for all employees, production and nonsupervisory employees, and women employees, as well as average hourly earnings, average weekly hours, and average weekly overtime hours (in manufacturing industries) for both all employees and production and nonsupervisory employees.

Each year, BLS examines and revises the previous 5 years of seasonal factors and seasonally adjusted data that were originally calculated in realtime. The CES program uses 10 years of original data to produce seasonally adjusted data. Once a year, the CES program selects new model specifications and recalculates the previous 5 years of seasonally adjusted data using the new model specifications.2 The model specifications selected annually are not changed until the next annual review. After 5 years, the CES program no longer revises the seasonally adjusted series, unless there is a change to the industry that requires a reconstruction of the time series.3

How the seasonal adjustment process works

Prior to the normal treatment for seasonal adjustment, CES analysts remove quantifiable nonseasonal events, such as strikes, from the original data to ensure that those events are not included in the calculation of the seasonal factors. Additionally, X-13ARIMA-SEATS, the seasonal adjustment program BLS uses to produce seasonally adjusted data, detects outliers.4 The seasonal adjustment program removes the outliers detected and does not use them in calculating the seasonal factors. Removing the outliers detected prevents extreme values from distorting the seasonal factors. The seasonal adjustment program removes the outliers from the calculation of the seasonal factors but not from the seasonally adjusted series. After the seasonal adjustment program calculates the seasonal factors, it applies them to the original data to calculate the seasonally adjusted series. The seasonally adjusted series is the trend and the irregular components combined, which includes the outliers. CES analysts then uses the new model selections to revise the most recent 5 years of seasonally adjusted data. Exhibit 1 provides the historical time frame of the original series, used to produce the seasonally adjusted data, and the latest revision period that would be used in the normal treatment process.

Exhibit 1. Ten years of original data and the 5-year seasonally adjusted revision period for one seasonal adjustment run. The seasonal adjustment run is from 2011 to October 2020. This seasonal adjustment run is used to create the 5-year seasonally adjusted revised data, from 2016 to October 2020.

Additional intervention for the 2020 annual review

The COVID-19 pandemic is an event that has affected just about every part of the economy to such a degree that we now define many aspects of what we do into a prepandemic (prior to February 2020) and postpandemic (March 2020 forward) time frame. Given the pandemic, the CES program had to determine how best to maintain the integrity of the seasonally adjusted history prior to the pandemic. This task was particularly challenging in that the effects of the pandemic should not influence the seasonal adjustment model used to generate the seasonal factors that are prior to the pandemic. Ordinarily, outlier detection alone is used to prevent extreme values from affecting the calculation of the seasonal factors. However, additional intervention was necessary because of the extreme, pandemic-induced fluctuations in the estimates and the extended time frame over which the pandemic has occurred. The results provided show that, even with the outlier detection used as part of the normal treatment process, there were larger than normal revisions to the historical time series and unexpected changes to the historical normal seasonal movements.

To address these issues, CES analysts ran a sequence of two runs (prepandemic and postpandemic) for the purpose of mitigating the effect of the pandemic on the calculation of the historical seasonal factors. Informed by the combination of these two sequential runs, the CES program revised 5 years of seasonally adjusted data. Exhibit 2 provides the time frame of the original data BLS used to produce the seasonally adjusted data and the revision period of the latest annual review with the additional intervention. The revised seasonally adjusted data prior to March 2020, when the pandemic began, were produced from the prepandemic run. The revised seasonally adjusted data after the pandemic began, in March 2020, were produced from the postpandemic run. BLS used this combined approach to ensure that the effects from the pandemic were limited to the pandemic months and did not affect the seasonally adjusted history prior to the pandemic.

Exhibit 2. Ten years of original data and the 5-year seasonal adjustment revision period using the additional intervention. The prepandemic period is 2011 to February 2020, and it revises 2016 to February 2020 data. The postpandemic period is 2011 to October 2020, and it revises March 2020 to October 2020.

Normal seasonal movement

To better understand historical seasonal patterns, it is useful to view the normal seasonal movement in a series over time. The normal seasonal movement is calculated as the difference between the month-to-month change in the original, not seasonally adjusted series minus the month-to-month change in the seasonally adjusted series:

The normal seasonal movement is useful for analyzing historical seasonal patterns because it provides a rough estimate of the strength of the seasonal effect over time. The normal seasonal movements of the total employment series using the normal treatment are compared to the normal seasonal movements of the published total employment series using the additional intervention to show the effects of using the additional intervention as compared to the normal treatment.

Chart 1 highlights the normal seasonal movements of total nonfarm employment over time for March, one the months hit hardest, economically, by the pandemic. The rightmost bar in each year shows the normal seasonal movements as previously published before the annual review. In March of each year from 2011 to 2019, employment has increased by approximately 500,000–600,000 on account of the normal seasonal movement. In contrast, the previously published normal seasonal movement for March 2020 was 360,000, much lower than normal. This weakness comes from the effects of the pandemic without any additional intervention to mitigate them. The middle bar of the chart shows a similar pattern in the normal seasonal movements calculated with the normal treatment. As a result of the revision, the weakness from the effects of the pandemic is spread over the previous 5 years in addition to March 2020. The leftmost bar in each year of chart shows the normal seasonal movements revised 5 years using the additional intervention. The normal seasonal movements with the additional intervention are more stable and consistent with the history prior to the influence of the pandemic. Next, the results for the prepandemic period and the postpandemic period are discussed to explain the effects of the additional intervention compared with the normal treatment in more detail.

Prepandemic period

The normal seasonal movements in the prepandemic period, from January 2016 to February 2020, show that the additional intervention provides more consistency and smaller revisions compared with the results from the normal treatment. This is especially visible in the history of the series in the months of March and April. Without the additional intervention, the historical normal seasonal movements show a weaker seasonal increase than the previously published data. This weakening is due to the seasonal adjustment process factoring the effects of the pandemic into the history. The normal seasonal movements resulting from the additional intervention are more consistent with the previously published historical seasonal movements.

Postpandemic period

In the first month of the postpandemic period, March 2020, the expected seasonal increase made using the normal treatment appears to have been weakened by the effects of the pandemic. Using the additional intervention shows a larger revision more in line with the previously published data compared to the normal treatment, which appears to have shifted the weakness from the pandemic back into the historical normal seasonal movement. The estimates for the remainder of the postpandemic period from April 2020 forward are the same between the normal treatment and the additional intervention because all 10 years of data through October 2020 were used to calculate the seasonal factors for both approaches.

Additional intervention for the 2021 review

The two-step approach used during the 2020 annual review addressed the need to mitigate the effects of the pandemic in realtime with a limited amount of data. Now that we have more data available and additional guidance on using outlier intervention options within the X-13 ARIMA-SEATS software, CES has moved toward using a more standardized approach.

As the COVID-19 pandemic unfolded, BLS time-series researchers developed several responses to the challenges. They found that using additional outlier types could provide a more parsimonious outlier set for the pandemic period.5 Using additional outlier types provides users the simplest model with the least variables and greatest explanatory power. Additionally, there are a number of CES series with complex movements during the pandemic period that could not be adequately adjusted by using the additive outlier types only.

For 2021's annual review, two approaches were evaluated:

  1. Traditional outlier treatment–using additive outliers (AOs) only.
  2. Additional outlier treatment–using all outlier types: additive (AO), level shifts (LS), and temporary changes (TC)

Exhibit 3 provides details on the different outlier types:

Exhibit 3. Types of outliers. An additive outlier (AO) effects only a single observation. A temporary change (TC) has an effect that diminishes to zero over several periods. A level shift (LS) represents a permanent increase or decrease in the level of the series.

To compare model fits, the CES program uses a version of the Akaike information criterion (AIC). From Rebecca Bevans of Scribbr, "The Akaike information criterion is a mathematical method for evaluating how well a model fits the data it was generated from. In statistics, AIC is used to compare different possible models and determine which one is the best fit for the data. AIC is calculated from:

  • the number of independent variables used to build the model.
  • the maximum likelihood estimate of the model (how well the model reproduces the data).

The best-fit model according to AICc is the one that explains the greatest amount of variation using the fewest possible independent variables."6

The CES program uses the AIC corrected for small sample sizes, the AICc. As such, the AICc is used here to determine which outlier treatment provides the best model fit. Like the AIC value, a lower AICc value is an indicator of a better fit model. Table 1 provides a comparison of the number of series with a lower AICc from the model using the normal treatment with AOs only against those of the additional-treatment model using all outlier types.

Table 1. Comparison of the number of series with lower AICc for models with all outlier types and AO-only outlier types
CES seriesAICc lower (better model fit)
All outlier typesAO onlyTotal number of series
Number of series with better fitPercentage of series with better fitNumber of series with better fitPercent of series with better fit

Total

2,363482,555524,918

All employees

5716826332834

Average weekly hours for all employees

2333740463637

Average hourly earnings for all employees

2153442266637

Average weekly overtime hour for all employees

43357965122

Production employees

4356820232637

Average weekly hours for production employees

2183441966637

Average hourly earnings for production employees

2013243668637

Average weekly overtime hours for production employees

40338267122

Women employees

4076224838655

Note: AO = Additive outliers. AICc = Akaike’s information criterion (corrected for small sample size). CES = Current Employment Statistics.

Source: U.S. Bureau of Labor Statistics.

We found that, overall, the number of series with a better model fit when we used all outlier types (48 percent of series) is approximately the same as those with a better model fit when we used the AO-only outlier types (52 percent). However, there are substantial differences in these numbers by data series type. For the employment series, the model using all outlier types provided a better fit for most series: all employees (68 percent of series), production employees (68 percent), and women employees (62 percent). The pandemic most strongly complicated the behavior of the employment series, which required additional outlier types to fully adjust. For the hours and earnings series, the model using AO-only outliers provided a better fit for most series. The impact from the pandemic on the hours and earnings series were less substantial in most cases and do not require the additional outlier types to provide a better model fit. The model with the lower AICc was used to produce the seasonally adjusted data for each series. Additionally, CES analysts performed a review of the series on a case-by-case basis to validate the seasonally adjusted series as part of the usual annual review procedure.

Prepandemic period

The normal seasonal movements in the prepandemic period, from January 2017 to February 2020, show that including the additional outlier types provides more consistency and smaller revisions compared with the normal treatment that uses the AOs only. The smaller revisions are broadly visible throughout the history of the series. Without the use of the additional outlier types, the historical normal seasonal movements show a weaker seasonal increase than those previously published. This weakening is due to the seasonal adjustment process incorrectly factoring the effects of the pandemic into the history as normal seasonal movement. The normal seasonal movements calculated with the additional outlier types are more consistent with the previously published historical seasonal movements.

Postpandemic period

The normal seasonal movements in the postpandemic period, from March 2020 to October 2021, show that including the additional outlier types provides more consistency and smaller revisions compared with the normal treatment that uses the AOs only. In the first month of the postpandemic period, March 2020, and in the following March 2021, the expected seasonal increase calculated with the AOs only appears to have been weakened by the effects of the pandemic. Using the additional outlier types shows a normal seasonal movement more in line with the previously published history compared with the results from the AOs only model, which appears to have shifted the weakness from the pandemic back into the historical normal seasonal movement. The revisions to March 2020 through the remainder of the postpandemic period are consistently smaller when including the additional outlier types.

Conclusion

The COVID-19 pandemic is unusual in its severity and duration when compared with other shocks to the economy. As a result, additional intervention was required to preserve the accuracy of the historical seasonally adjusted series. The results show that the effects of the pandemic were not completely removed from the seasonal factors as part of the normal treatment in which we used additive outlier detection alone. As part of the 2020 annual review, additional intervention and a sequence of two runs (prepandemic and postpandemic) was used to mitigate the impact of the pandemic on the historical seasonal factors. The additional intervention is an extension of the normal outlier-detection treatment and is an additional step towards effectively isolating the prepandemic period. The point of the additional intervention is to ensure that the postpandemic period does not influence the seasonal model prior to the pandemic—so that we preserve the historical seasonal patterns. The results show that the historical normal seasonal movements calculated with the the additional intervention are more in line with the expected normal seasonal movement—especially in the months in which the pandemic had its largest economic effect (March and April 2020). This approach provided the most accurate revisions to the seasonally adjusted data in realtime with the limited amount of postpandemic-period data available during the 2020 annual review just after the start of the pandemic. As the pandemic unfolded, BLS time-series researchers were able to develop a more standardized approach to mitigating the effects of the pandemic. In lieu of using the two-step approach used during the 2020 review, CES incorporated the use of additional outlier types available in the X-13 ARIMA-SEATS software as part of the most recent annual review to mitigate the effects of the pandemic. The models available by the use of the additional outlier types provide a better fit and allow better adjustments for the complex movements in the CES series that are due to the pandemic. These new models result in the calculation of more stable seasonal factors used to produce the seasonally adjusted series.

Appendix

2020 annual review

The appendix provides the components used to calculate the normal seasonal movement for total nonfarm employment for the following:

  • Table A-1: Not seasonally adjusted series;
  • Table A-2: Revised seasonally adjusted series using the normal treatment;
  • Table A-3: Revised seasonally adjusted series using the additional intervention;
  • Table A-4: Seasonally adjusted series as previously published prior to revision.

Tables A-5 through A-7 provide the normal seasonal movements calculated using the components from Tables A-1 through A-4. The difference in the normal seasonal movement between the normal treatment and the additional intervention is provided in table A8. The difference between the normal seasonal movement as previously published and the revised normal seasonal movement calculated by means of the normal treatment and the additional intervention is provided in tables A-9 and A-10, respectively. Some months are affected by the variable survey week adjustment, sometimes referred to as the 4- versus 5-week effect.7

Table A-1. Month-to-month change, not seasonally adjusted, in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011   

  -2,863       814       906      1,217[1]   684       489[1]   -1,274       275       746[1]     923       335      -175[1]

 2012   

  -2,606       954[1]   904       883       814       337[1]   -1,209       379[1]     628       847       389[1]    -75   

 2013   

  -2,882     1,038[1]   799      1,005       893[1]   409     -1,174       437[1]     608       922       525[1]   -251   

 2014   

  -2,804[1]   740       958      1,138       912[1]   587     -1,079       380[1]     686      1,054[1]   465        -4   

 2015   

  -2,815[1]   831       757     1,183[1]   942       475      -952[1]     192       558      1,148[1]   421        -2   

 2016   

  -2,974[1]   831       895     1,094[1]   649       664[1]    -973       253       666[1]     882       432      -212[1]

 2017   

  -2,876     1,029[1]   656     1,024       839       643[1]   -1,103       317       388[1]    1,013       574[1]   -247   

 2018   

  -3,098     1,237[1]   703       977       935       666[1]   -1,164       470[1]     312      1,003       481[1]   -200   

 2019   

  -2,953       805[1]   675      1,062       673[1]     622     -1,058       436[1]     417       989       595[1]   -249   

 2020   

  -2,791[1]   913     -1,016   -19,701[1]  3,168      5,082       606[1]     1,621      1,218      1,622[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A-2. Revised month-to-month change, seasonally adjusted (normal treatment), in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011   

     19       212       235       314[1]    101       236[1]      60       126       233[1]    204       132       202[1]   

 2012   

    354       262[1]    240        82       100        73[1]      152       172[1]     187       159       156[1]    239    

 2013   

    191       278[1]    139       191       222[1]     181       112       242[1]     187       225       264[1]     69    

 2014   

    175[1]    166       254       325       218[1]     326       232       188[1]     309       252[1]   291       268    

 2015   

    191[1]    271        71       284[1]    331       174       302[1]   125       155       306[1]   237       273    

 2016   

     67[1]     192       292       207[1]     51       255[1]     383       151       287[1]     92       115       190[1]

 2017   

    212       200[1]    132       236       161       212[1]     242       205        27[1]     210       169[1]    152    

 2018   

     85       389[1]    220       183       288       224[1]     166       226[1]      84       176        52[1]     203    

 2019   

    229       -41[1]    233       263        88[1]      193       205       188[1]     184       147       182[1]    119    

 2020   

    260[1]    244     -1,487   -20,679[1]   2,833      4,846     1,726[1]  1,583       716       680[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

Table A-3. Revised month-to-month change, seasonally adjusted (additional intervention), in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011    

    19       212       235       314[1]    101       236[1]      60       126       233[1]    204       132       202[1]

 2012    

   354       262[1]     240        82       100        73[1]      152       172[1]     187       159       156[1]    239    

 2013    

   191       278[1]     139       191       222[1]    181       112       242[1]     187       225       264[1]     69    

 2014    

   175[1]    166       254       325       218[1]    326       232       188[1]     309       252[1]   291       268    

 2015    

   191[1]    271        71       284[1]    331       174      302[1]    125       155       306[1]   237       273    

 2016    

   108[1]    212       237       197[1]     41       258[1]    371       143       289[1]    118       130       214[1]

 2017    

   197       183[1]     139       220       141       211[1]    228       190        42[1]     249       196[1]    179    

 2018    

    81       378[1]     195       153       270       214[1]    149       229[1]     105       212        92[1]     240    

 2019    

   237       -50[1]     168       219        63[1]      175       193       195[1]     221       195       234[1]    161    

 2020    

   315[1]    289     -1,683    -20,679[1]  2,833      4,846    1,726[1]   1,583       716       680[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

Table A-4. Month-to-month change, seasonally adjusted (as previously published), in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011    

    19       212       235       314[1]   101       236[1]    60       126       233[1]    204       132      202[1]

 2012    

   354       262[1]    240        82       100        73[1]      152       172[1]    187       159       156[1]    239   

 2013    

   191       278[1]    139       191       222[1]    181       112       242[1]    187       225       264[1]     69   

 2014    

   175[1]    166       254       325       218[1]     326       232       188[1]    309       252[1]    291       268   

 2015    

   191[1]    271        71       284[1]   331       174      302[1]    125       155       306[1]    237       273   

 2016    

    73[1]     263       229       187[1]    42       267[1]    354       135       269[1]    145       151      230[1]

 2017    

   185       188[1]    129       197       155       216[1]    215       184        18[1]     267       225[1]    130   

 2018    

   121       406[1]    176       137       278       219[1]    136       244[1]     80       201       134[1]    182   

 2019    

   269         1[1]     147       210        85[1]      182       194       207[1]    208       185       261[1]    184   

 2020    

   214[1]    251     -1,373   -20,787[1]  2,725      4,781    1,761[1]   1,493       711       654[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A-5. Revised normal seasonal movement (normal treatment) in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011   

[2]   602       671       903[1]    583       253[1]  -1,334       149       513[1]    719       203     -377[1]

 2012   

  -2,960       692[1]    664       801       714       264[1]  -1,361       207[1]    441       688       233[1]   -314   

 2013   

  -3,073       760[1]    660       814       671[1]    228     -1,286       195[1]    421       697       261[1]   -320   

 2014   

  -2,979[1]   574       704       813       694[1]    261     -1,311       192[1]    377      802[1]    174      -272   

 2015   

  -3,006[1]   560       686       899[1]    611       301    -1,254[1]     67       403      842[1]    184      -275   

 2016   

  -3,041[1]   639       603       887[1]    598       409[1]  -1,356       102       379[1]     790       317     -402[1]

 2017   

  -3,088       829[1]    524       788       678       431[1]  -1,345       112       361[1]     803       405[1]   -399   

 2018   

  -3,183       848[1]    483       794       647       442[1]  -1,330       244[1]    228       827       429[1]   -403   

 2019   

  -3,182       846[1]    442       799       585[1]    429     -1,263       248[1]    233       842       413[1]   -368   

 2020   

  -3,051[1]   669       471       978[1]    335       236    -1,120[1]     38       502      942[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A6. Revised normal seasonal movement (additional intervention) in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011   

[2]   602       671       903[1]     583       253[1]  -1,334       149       513[1]    719       203      -377[1]

 2012   

  -2,960       692[1]    664       801       714       264[1]  -1,361       207[1]    441       688       233[1]  -314    

 2013   

  -3,073       760[1]    660       814       671[1]   228     -1,286       195[1]    421       697       261[1]  -320    

 2014   

  -2,979[1]   574       704       813       694[1]   261     -1,311       192[1]    377       802[1]    174      -272    

 2015   

  -3,006[1]   560       686       899[1]     611       301    -1,254[1]     67       403       842[1]    184      -275    

 2016   

  -3,082[1]   619       658       897[1]     608       406[1]  -1,344       110       377[1]    764       302      -426[1]

 2017   

  -3,073       846[1]    517       804       698       432[1]  -1,331       127       346[1]    764       378[1]  -426    

 2018   

  -3,179       859[1]    508       824       665       452[1]  -1,313       241[1]    207       791       389[1]  -440    

 2019   

  -3,190       855[1]    507       843       610[1]   447     -1,251       241[1]    196       794       361[1]  -410    

 2020   

  -3,106[1]   624       667       978[1]     335       236    -1,120[1]     38       502       942[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A-7. Normal seasonal movement (as previously published) in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011   

  -2,882       602       671       903[1]    583       253[1]  -1,334       149       513[1]     719       203      -377[1]

 2012   

  -2,960       692[1]    664       801       714       264[1]  -1,361       207[1]     441       688       233[1]   -314    

 2013   

  -3,073       760[1]    660       814       671[1]    228     -1,286       195[1]     421       697       261[1]   -320    

 2014   

  -2,979[1]   574       704       813       694[1]    261     -1,311       192[1]     377       802[1]   174      -272    

 2015   

  -3,006[1]   560       686       899[1]    611       301     -1,254[1]    67       403       842[1]   184      -275    

 2016   

  -3,047[1]   568       666       907[1]    607       397[1]  -1,327       118       397[1]     737       281      -442[1]

 2017   

  -3,061       841[1]    527       827       684       427[1]  -1,318       133       370[1]     746       349[1]   -377    

 2018   

  -3,219       831[1]    527       840       657       447[1]  -1,300       226[1]     232       802       347[1]   -382    

 2019   

  -3,222       804[1]    528       852       588[1]    440     -1,252       229[1]    209       804       334[1]   -433    

 2020   

  -3,005[1]   662       357      1,086[1]   443       301     -1,155[1]   128       507       968[1][2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A-8. Difference in normal seasonal movement (revised with normal treatment versus revised with additional intervention) in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011    

     0         0         0         0[1]        0         0[1]      0         0         0[1]        0         0         0[1]  

 2012    

     0         0[1]       0         0         0         0[1]      0         0[1]        0         0         0[1]       0    

 2013    

     0         0[1]       0         0         0[1]        0         0         0[1]        0         0         0[1]       0    

 2014    

     0[1]        0         0         0         0[1]        0         0         0[1]        0         0[1]       0         0    

 2015    

     0[1]        0         0         0[1]        0         0         0[1]      0         0         0[1]       0         0    

 2016    

   -41[1]     -20        55        10[1]      10        -3[1]     12         8        -2[1]      -26       -15       -24[1] 

 2017    

    15        17[1]     -7        16        20         1[1]     14        15       -15[1]     -39       -27[1]    -27    

 2018    

     4        11[1]     25        30        18        10[1]    17        -3[1]     -21       -36       -40[1]    -37    

 2019    

    -8         9[1]     65        44        25[1]      18        12        -7[1]     -37       -48       -52[1]    -42    

 2020    

   -55[1]     -45       196         0[1]        0         0         0[1]      0         0         0[1]  [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A-9. Difference in normal seasonal movement (revised with normal treatment versus as previously published) in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011    

     0         0         0         0[1]        0         0[1]      0         0         0[1]        0         0         0[1]  

 2012    

     0         0[1]        0         0         0         0[1]      0         0[1]        0         0         0[1]        0    

 2013    

     0         0[1]        0         0         0[1]       0         0         0[1]        0         0         0[1]        0    

 2014    

     0[1]       0         0         0         0[1]       0         0         0[1]        0         0[1]       0         0    

 2015    

     0[1]       0         0         0[1]        0         0         0[1]      0         0         0[1]       0         0    

 2016    

    -6[1]     -71        63        20[1]       9       -12[1]    29        16        18[1]      -53       -36       -40[1]

 2017    

    27        12[1]       3        39         6        -4[1]     27        21         9[1]      -57       -56[1]      22    

 2018    

   -36       -17[1]      44        46        10         5[1]     30       -18[1]       4       -25       -82[1]      21    

 2019    

   -40       -42[1]      86        53         3[1]      11        11       -19[1]     -24       -38       -79[1]     -65    

 2020    

    46[1]     -7      -114       108[1]    108        65       -35[1]     90         5        26[1] [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table A-10. Difference in normal seasonal movement (revised with additional intervention versus as previously published) in total nonfarm employment (in thousands)
Year    Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2011    

     0         0         0         0[1]      0         0[1]      0         0         0[1]        0         0         0[1]  

 2012    

     0         0[1]       0         0         0         0[1]      0         0[1]        0         0         0[1]        0    

 2013    

     0         0[1]       0         0         0[1]        0         0         0[1]        0         0         0[1]        0    

 2014    

     0[1]       0         0         0         0[1]        0         0         0[1]        0         0[1]       0         0    

 2015    

     0[1]       0         0         0[1]        0         0         0[1]       0         0         0[1]       0         0    

 2016    

    35[1]    -51         8        10[1]      -1        -9[1]     17         8        20[1]      -27       -21       -16[1]

 2017    

    12        -5[1]      10        23       -14        -5[1]     13         6        24[1]      -18       -29[1]      49    

 2018    

   -40       -28[1]     19        16        -8        -5[1]     13       -15[1]      25        11       -42[1]      58    

 2019    

   -32       -51[1]     21         9       -22[1]      -7        -1       -12[1]      13        10       -27[1]     -23    

 2020    

   101[1]    38      -310       108[1]    108        65       -35[1]     90         5        26[1] [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

2021 annual review

The appendix provides the components used to calculate the normal seasonal movement for total nonfarm employment for the following:

  • Table-B1: Not seasonally adjusted series;
  • Table B-2: Revised seasonally adjusted series using additive outliers only;
  • Table B-3: Revised seasonally adjusted series using all outlier types;
  • Table B-4: Seasonally adjusted series as previously published prior to revision.

Tables B-5 through B-7 provide the normal seasonal movements calculated using the components from tables B-1 through B-4. The difference in the normal seasonal movement between the additive outliers only and all outlier types is provided in table B-8. The difference between the normal seasonal movement as previously published and the revised normal seasonal movement calculated with the additive outliers only and all outlier types is provided in tables B-9 and B-10, respectively. Some months are affected by the variable survey week adjustment, sometimes referred to as the 4- versus 5-week effect.8

Table B-1. Month-to-month change, not seasonally adjusted, in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012   

  -2,606       954[1]     904       883       814       337[1]   -1,209       379[1]     628       847       389[1]    -75   

 2013   

  -2,882     1,038[1]    799      1,005       893[1]     409     -1,174       437[1]     608       922       525[1]   -251   

 2014   

 -2,804[1]    740       958      1,138       912[1]     587     -1,079       380[1]     686     1,054[1]    465        -4   

 2015   

 -2,815[1]    831       757      1183[1]    942       475      -952[1]     192       558     1,148[1]    421        -2   

 2016   

 -2,974[1]    831       895      1,094[1]   649       664[1]    -973       253       666[1]     882       432      -212[1]

 2017   

  -2,876     1,029[1]    656      1,024       839       643[1]   -1,103       317       388[1]    1,013       574[1]   -247   

 2018   

  -3,098     1,237[1]    703       977       935       666[1]   -1,164       470[1]     312      1,003       481[1]   -200   

 2019   

  -2,953       805[1]     675      1,062       673[1]     622     -1,058       436[1]     417       989       595[1]   -249   

 2020   

 -2,791[1]    913     -1,016   -19,699[1]  3,169      5,085       598[1]     1,622      1,231     1,607[1]    550      -510   

 2021   

 -2,631[1]   1,155      1,179      1,050[1]   946      1,189       -41[1]      495       704[1]    1,659   [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

Table B-2. Revised month-to-month change, seasonally adjusted (additive outlier only), in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012    

   354       262[1]    240        82       100        73[1]      152       172[1]     187       159       156[1]     239   

 2013    

   191       278[1]    139       191       222[1]     181       112       242[1]     187       225       264[1]      69   

 2014    

   175[1]   166       254       325       218[1]     326       232       188[1]     309       252[1]    291       268   

 2015    

   191[1]   271        71       284[1]    331       174       302[1]   125       155       306[1]    237       273   

 2016    

   108[1]   212       237       197[1]     41       258[1]     371       143       289[1]    118       130       214[1]

 2017    

    76       186[1]    150       320       474       216[1]     139       104        30[1]       3        91[1]       63   

 2018    

    84       431[1]    315       408       694       232[1]     -21       107[1]    -117       -31       -79[1]      115   

 2019    

   213        82[1]    377       639       283[1]     378        50        14[1]     -103      -163        -3[1]       89   

 2020    

   154[1]   336     -1,119    -19,090[1]  2,322      4,500     1,369[1]  1,436       564       301[1]     77      -177   

 2021    

   343[1]   643      1,059      1,719[1]   219       577       633[1]   256       109[1]    290    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

Table B-3. Revised month-to-month change, seasonally adjusted (all outlier types), in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012   

    354       262[1]    240        82       100        73[1]      152       172[1]     187       159       156[1]     239   

 2013   

    191       278[1]    139       191       222[1]     181       112       242[1]     187       225       264[1]      69   

 2014   

    175[1]   166       254       325       218[1]     326       232       188[1]     309       252[1]   291       268   

 2015   

    191[1]   271        71       284[1]   331       174       302[1]   125       155       306[1]   237       273   

 2016   

    108[1]   212       237       197[1]    41       258[1]     371       143       289[1]    118       130       214[1]

 2017   

    213       190[1]    142       205       223       197[1]     183       145        99[1]     141       200[1]     176   

 2018   

    133       402[1]    225       179       333       183[1]      66       219[1]      57       145       102[1]     248   

 2019   

    279        24[1]    224       288        77[1]      130        78       160[1]     163        93       252[1]     200   

 2020   

    339[1]   376     -1,498   -20,493[1]  2,642      4,505     1,388[1]  1,665       919       647[1]   333      -115   

 2021   

    520[1]   710       704       263[1]   447       557       689[1]   517       424[1]    677    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

Table B-4. Month-to-month change, seasonally adjusted (as previously published), in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012   

    354       262[1]    240        82       100        73[1]      152       172[1]     187       159       156[1]      239   

 2013   

    191       278[1]    139       191       222[1]     181       112       242[1]     187       225       264[1]       69   

 2014   

   175[1]    166       254       325       218[1]     326       232       188[1]     309       252[1]   291       268   

 2015   

   191[1]    271        71       284[1]    331       174       302[1]   125       155       306[1]   237       273   

 2016   

   108[1]    212       237       197[1]     41       258[1]     371       143       289[1]    118       130       214[1]

 2017   

    197       183[1]    139       220       141       211[1]     228       190        42[1]     249       196[1]      179   

 2018   

     81       378[1]    195       153       270       214[1]     149       229[1]     105       212        92[1]      240   

 2019   

    237       -50[1]    168       219        63[1]      175       193       195[1]     221       195       234[1]      161   

 2020   

   315[1]    289     -1,683   -20,679[1]   2,833      4,846     1,726[1]  1,583       716       680[1]   264      -306   

 2021   

   233[1]    536       785       269[1]    614       962     1,091[1]   483       379[1]    648    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table B-5. Revised normal seasonal movement (additive outlier only) in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012   

[2]   692[1]    664       801       714       264[1]  -1,361       207[1]    441       688       233[1]     -314  

 2013   

  -3,073       760[1]    660       814       671[1]    228     -1,286       195[1]    421       697       261[1]     -320  

 2014   

  -2,979[1]   574       704       813       694[1]    261     -1,311       192[1]    377       802[1]    174      -272  

 2015   

  -3,006[1]   560       686       899[1]   611       301     -1,254[1]    67       403       842[1]    184      -275  

 2016   

  -3,082[1]   619       658       897[1]   608       406[1]  -1,344       110       377[1]    764       302     -426[1]

 2017   

  -2,952       843[1]    506       704       365       427[1]  -1,242       213       358[1]   1,010       483[1]     -310  

 2018   

  -3,182       806[1]    388       569       241       434[1]  -1,143       363[1]    429      1,034       560[1]     -315  

 2019   

  -3,166       723[1]    298       423       390[1]    244     -1,108       422[1]    520      1,152       598[1]     -338  

 2020   

  -2,945[1]   577       103      -609[1]   847       585      -771[1]    186       667      1,306[1]   473      -333  

 2021   

  -2,974[1]   512       120      -669[1]   727       612      -674[1]     239       595[1]   1,369    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table B-6. Revised normal seasonal movement (all outlier types) in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012  

[2]   692[1]     664       801       714       264[1]  -1,361       207[1]    441       688       233[1]    -314   

 2013  

   -3,073       760[1]     660       814       671[1]    228     -1,286       195[1]    421       697       261[1]    -320   

 2014  

   -2,979[1]   574       704       813       694[1]    261     -1,311       192[1]    377       802[1]   174      -272   

 2015  

   -3,006[1]   560       686       899[1]   611       301     -1,254[1]    67       403       842[1]   184      -275   

 2016  

   -3,082[1]   619       658       897[1]   608       406[1]  -1,344       110       377[1]    764       302      -426[1]

 2017  

   -3,089       839[1]     514       819       616       446[1]  -1,286       172       289[1]    872       374[1]    -423   

 2018  

   -3,231       835[1]     478       798       602       483[1]  -1,230       251[1]    255       858       379[1]    -448   

 2019  

   -3,232       781[1]     451       774       596[1]    492     -1,136       276[1]    254       896       343[1]   -449   

 2020  

   -3,130[1]   537       482       794[1]   527       580      -790[1]     -43       312       960[1]   217      -395   

 2021  

   -3,151[1]   445       475       787[1]   499       632      -730[1]     -22       280[1]    982    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table B-7. Normal seasonal movement (as previously published) in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012   

  -2,960       692[1]    664       801       714       264[1]  -1,361       207[1]    441       688       233[1]     -314   

 2013   

  -3,073       760[1]    660       814       671[1]   228     -1,286       195[1]    421       697       261[1]     -320   

 2014   

 -2,979[1]    574       704       813       694[1]   261     -1,311       192[1]    377       802[1]     174      -272   

 2015   

 -3,006[1]    560       686       899[1]    611       301     -1,254[1]    67       403       842[1]     184      -275   

 2016   

 -3,082[1]    619       658       897[1]    608       406[1]  -1,344       110       377[1]    764       302     -426[1]

 2017   

  -3,073       846[1]    517       804       698       432[1]  -1,331       127       346[1]    764       378[1]     -426   

 2018   

  -3,179       859[1]    508       824       665       452[1]  -1,313       241[1]    207       791       389[1]     -440   

 2019   

  -3,190       855[1]    507       843       610[1]   447     -1,251       241[1]    196       794       361[1]     -410   

 2020   

 -3,106[1]    624       667       980[1]    336       239     -1,128[1]    39       515       927[1]     286      -204   

 2021   

 -2,864[1]    619       394       781[1]    332       227     -1,132[1]    12       325[1]   1,011    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table B-8. Difference in normal seasonal movement (revised with additive outliers only versus revised with all outlier types) in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012    

     0         0[1]       0         0         0         0[1]       0         0[1]        0         0         0[1]        0   

 2013    

     0         0[1]       0         0         0[1]        0         0         0[1]        0         0         0[1]        0   

 2014    

     0[1]       0         0         0         0[1]        0         0         0[1]        0         0[1]       0         0   

 2015    

     0[1]       0         0         0[1]       0         0         0[1]       0         0         0[1]       0         0   

 2016    

     0[1]       0         0         0[1]       0         0[1]      0         0         0[1]       0         0         0[1] 

 2017    

  -137        -4[1]       8       115       251        19[1]    -44       -41       -69[1]   -138      -109[1]     -113   

 2018    

   -49        29[1]     90       229       361        49[1]    -87      -112[1]   -174      -176      -181[1]     -133   

 2019    

   -66        58[1]    153       351       206[1]    248       -28      -146[1]   -266      -256      -255[1]     -111   

 2020    

  -185[1]   -40       379      1,403[1] -320        -5       -19[1]   -229      -355      -346[1]  -256       -62   

 2021    

  -177[1]   -67       355      1,456[1] -228        20       -56[1]   -261      -315[1]  -387     [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table B-9. Difference in normal seasonal movement (revised with additive outliers only versus as previously published) in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012    

     0         0[1]       0         0         0         0[1]       0         0[1]        0         0         0[1]        0    

 2013    

     0         0[1]       0         0         0[1]       0         0         0[1]        0         0         0[1]        0    

 2014    

     0[1]       0         0         0         0[1]       0         0         0[1]        0         0[1]       0         0    

 2015    

     0[1]       0         0         0[1]        0         0         0[1]       0         0         0[1]       0         0    

 2016    

     0[1]       0         0         0[1]        0         0[1]       0         0         0[1]        0         0         0[1]  

 2017    

  -121         3[1]      11       100       333         5[1]     -89       -86       -12[1]   -246      -105[1]    -116    

 2018    

     3        53[1]    120       255       424        18[1]    -170      -122[1]   -222      -243      -171[1]    -125    

 2019    

   -24       132[1]    209       420       220[1]    203      -143      -181[1]   -324      -358      -237[1]     -72    

 2020    

  -161[1]    47       564      1,589[1]  -511      -346      -357[1]  -147      -152      -379[1]  -187       129    

 2021    

   110[1]    107       274      1,450[1]  -395      -385      -458[1]  -227      -270[1]  -358    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 Table B10. Difference in normal seasonal movement (revised with all outlier types versus as previously published) in total nonfarm employment (in thousands)
 Year       Jan       Feb       Mar       Apr       May       Jun       Jul       Aug       Sep       Oct       Nov       Dec   

 2012   

      0         0[1]       0         0         0         0[1]       0         0[1]       0         0         0[1]        0    

 2013   

      0         0[1]       0         0         0[1]        0         0         0[1]       0         0         0[1]        0    

 2014   

      0[1]      0         0         0         0[1]        0         0         0[1]       0         0[1]       0         0    

 2015   

      0[1]      0         0         0[1]        0         0         0[1]       0         0         0[1]       0         0    

 2016   

      0[1]      0         0         0[1]        0         0[1]      0         0         0[1]       0         0         0[1]  

 2017   

     16         7[1]       3       -15        82       -14[1]    -45       -45        57[1]   -108         4[1]       -3    

 2018   

     52        24[1]     30        26        63       -31[1]    -83       -10[1]    -48       -67        10[1]        8    

 2019   

     42        74[1]     56        69        14[1]     -45      -115       -35[1]    -58      -102        18[1]       39    

 2020   

     24[1]    87       185       186[1]   -191      -341     -338[1]     82       203       -33[1]     69       191    

 2021   

    287[1]   174       -81        -6[1]     -167      -405     -402[1]     34        45[1]     29    [2][2]

Note:

[1] These values come from a 5-week survey interval.

[2] Not applicable at time of annual review.

Source: U.S. Bureau of Labor Statistics.

 


Suggested citation:

Nicole Hudson, Jeannine Mercurio, and Jurgen Kropf, "The challenges of seasonal adjustment for the Current Employment Statistics survey during the COVID-19 pandemic," Monthly Labor Review, U.S. Bureau of Labor Statistics, May 2022, https://doi.org/10.21916/mlr.2022.14

Notes


1 For an example of one such publication, see “The challenges of seasonal adjustment during the COVID-19 pandemic” Commissioner’s Corner (U.S. Bureau of Labor Statistics, December 21, 2020), https://blogs.bls.gov/blog/2020/12/21/the-challenges-of-seasonal-adjustment-during-the-covid-19-pandemic/.

2 For more information on seasonal adjustment, see “Technical notes for the Current Employment Statistics survey: seasonal adjustment" Current Employment Statistics—CES (national) (U.S. Bureau of Labor Statistics, February 4, 2022), https://www.bls.gov/web/empsit/cestn.htm#section6e.

3 All CES series are evaluated annually for sample size, coverage, and response rates. Reconstruction of time series may result from a reevaluation of the sample and universe coverage for CES industries. In this case, seasonal adjustment is rerun for the entire history of the reconstructed time series.

4 X-13 ARIMA-SEATS is publicly available from the U.S. Census Bureau at https://www.census.gov/data/software/x13as.html.

5 Brian C. Monsell, "Time series responses to the COVID-19 pandemic at BLS for monthly and weekly series," Statistical Survey Paper (U.S. Bureau of Labor Statistics, December 2021), https://www.bls.gov/osmr/research-papers/2021/pdf/st210040.pdf.

6 Rebecca Bevans, “An introduction to the Akaike information criterion,” Scribbr, March 26, 2020, https://www.scribbr.com/statistics/akaike-information-criterion/.

7 See “Technical notes for the Current Employment Statistics survey: seasonal adjustment" Current Employment Statistics—CES (national) (U.S. Bureau of Labor Statistics, February 4, 2022), https://www.bls.gov/web/empsit/cestn.htm#section6e.

8 See “Technical notes for the Current Employment Statistics survey: seasonal adjustment" Current Employment Statistics—CES (national) (U.S. Bureau of Labor Statistics, February 4, 2022), https://www.bls.gov/web/empsit/cestn.htm#section6e.

 

 

 

 

article image
About the Author

Nicole Hudson
hudson.nicole@bls.gov

Nicole Hudson is an economist in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

Jeannine Mercurio
mercurio.jeannine@bls.gov

Jeannine Mercurio is an economist in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

Jurgen Kropf
kropf.jurgen@bls.gov

Jurgen Kropf is a former Division Chief in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

close or Esc Key