Effects of Hurricanes Harvey, Irma, and Maria on BLS Data Collection and Reporting

Several U.S. states, Puerto Rico, and the U.S. Virgin Islands suffered heavy damage from three major hurricanes in late August and September 2017. Hurricanes Harvey, Irma, and Maria have affected data collection in many BLS surveys. Here is a summary of those effects.

Hurricane Harvey first made landfall on Friday, August 25, in Texas and caused catastrophic damage there. After returning to the Gulf of Mexico, Harvey was downgraded to a tropical storm and then made landfall a second time on Wednesday, August 30, in Louisiana.

Before Hurricane Irma hit the lower Florida Keys on Sunday, September 10, Irma already had caused severe damage in St. Thomas and St. John in the U.S. Virgin Islands. Puerto Rico also suffered damage from Irma. After moving north from the Keys, Irma made landfall again later on September 10 on Florida’s southern coast. Irma was a Category 5 hurricane when near the U.S. Virgin Islands and Puerto Rico. Irma dropped to a Category 4 when passing through the Florida Keys, then a Category 2 hurricane when hitting mainland Florida.

Hurricane Maria made landfall in St. Croix in the U.S. Virgin Islands and in Puerto Rico on Wednesday, September 20, causing catastrophic damage. These areas already had suffered damage from Hurricane Irma earlier in the month.

Current Employment Statistics (Establishment Survey)

The reference period of the establishment survey is the pay period that includes the 12^th of the month. People are not counted as employed if they are not paid for the entire pay period that includes the 12^th of the month. Hurricane Harvey made landfall before the survey’s September reference period. Hurricane Irma made landfall during the September reference period. As a result, Hurricanes Harvey and Irma affected September payroll estimates for the nation, Texas, Florida, and other areas.

National data

We have more information about how Hurricanes Harvey and Irma affected September 2017 national nonfarm payroll data, including survey response rates. Data for September were released Friday, October 6, 2017.

National nonfarm employment estimates do not include Puerto Rico or the U.S. Virgin Islands. For more information about employment estimates for Puerto Rico and the Virgin Islands, see the State and Metro Area information below.

State and Metro Area

The estimates of employment, hours, and earnings for Puerto Rico and the U.S Virgin Islands are published by BLS in the State Employment and Unemployment news release. Puerto Rico also produces geographic area estimates with guidance from BLS. These estimates are published in the Metropolitan Area Employment and Unemployment news release.

Because of the devastation caused by Hurricanes Irma and Maria, Puerto Rico and the U.S. Virgin Islands could not conduct normal data collection for their establishment surveys. Enough data have been collected for Puerto Rico to produce preliminary estimates for December 2017. There were no changes to the normal estimation method for December. Response rates for December were slightly lower than average in Puerto Rico. See Hurricanes Irma and Maria and Puerto Rico and the U.S. Virgin Islands payroll data for more information.

The U.S Virgin Islands was able to collect adequate sample data to produce final estimates for September, October, and November 2017. Along with preliminary December 2017 estimates, BLS released these data on January 23, 2018. There were no changes to the normal estimation method. Response rates for December 2017 were lower than average in the U.S. Virgin Islands.

Employment, hours, and earnings estimates for Texas, Florida and other areas affected by the hurricanes were released in the State Employment and Unemployment news release.

No changes were made to the normal estimation methods for the employment, hours, and earnings estimates for Texas and Florida for the September 2017 final through December 2017 preliminary estimates. With the release of January preliminary estimates on March 12, 2018, September 2017 estimates for both states have been replaced with universe data from the Quarterly Census of Employment and Wage (QCEW), as part of the 2017 annual benchmark process. For more information on the CES annual benchmark see www.bls.gov/web/laus/benchmark.pdf. Also, based on information from the QCEW data, the methodology for the birth/death component of the October and November 2017 re-estimates was changed for Florida.

Current Population Survey (Household Survey)

Hurricane Harvey made landfall well before the September reference period for the household survey. Hurricane Irma struck Florida on September 10, at the start of the September reference period (Sunday, September 10, through Saturday, September 16, 2017). The U.S. Census Bureau began collecting data for the household survey on Sunday, September 17. Data collection started somewhat later in areas severely affected by Hurricane Irma. Throughout the collection period, the Census Bureau closely monitored response rates in these areas and in areas severely affected by Hurricane Harvey.

The household survey has two measures that show some of the impact of bad weather on employment and hours at work. The survey collects data on the number of people who had a job but were not at work for the entire reference week due to bad weather. These people are counted as employed regardless of whether they were paid for the time off. The household survey also provides a measure of the number of people who usually work full time but worked part time (1 to 34 hours) during the survey reference week due to bad weather.

Current and historical data on people who were not at work for all or part of the reference week due to weather are available on the household survey's most requested statistics page. September data were released Friday, October 6, 2017.

The Current Population Survey is conducted only in the 50 states and the District of Columbia. National household estimates do not include Puerto Rico or the U.S. Virgin Islands. For more information about labor force estimates for Puerto Rico and the U.S. Virgin Islands, see Local Area Unemployment Statistics below.

Local Area Unemployment Statistics

Puerto Rico conducts its own monthly household survey, patterned after the Current Population Survey, to produce labor force information. The estimates are published in the State Employment and Unemployment news release. Puerto Rico also produces labor force estimates for geographic areas, with guidance from BLS. These estimates are published in the Metropolitan Area Employment and Unemployment news release.

Because of the devastation caused by Hurricanes Irma and Maria, Puerto Rico could not conduct its household survey for September 2017. Data collection resumed for October 2017, but the response rate was somewhat lower than average, partly because of difficulties reaching some remote areas with significant hurricane damage. BLS interpolated the September 2017 estimates for Puerto Rico and published them on February 27, 2018. We combined these interpolated estimates with the survey-based estimates for the other months to calculate official annual averages for 2017. We also used the interpolated estimates to fill in the missing monthly observation for the seasonally adjusted series. These estimates will serve as controls for local area estimates in September 2017, which BLS will release on April 20, 2018.

The U.S. Virgin Islands does not conduct a household survey and, therefore, does not participate in the Local Area Unemployment Statistics program.

Puerto Rico and the U.S. Virgin Islands statistics at a glance are available at:

www.bls.gov/regions/northeast/puerto_rico.htm#eag

www.bls.gov/regions/northeast/virgin_islands.htm#eag

Job Openings and Labor Turnover Survey

The reference periods of the Job Openings and Labor Turnover Survey (JOLTS) are the pay period that includes the 12^th of the month for employment, the last business day of the month for job openings, and the entire calendar month for hires and separations.

Hurricane Harvey made landfall in Texas at the end of August, before the August reference period for job openings and during the August reference period for hires and separations. Hurricane-related closures occurred late in the month, leaving little time for additional hires and separations to occur. Our research found the following:

• Survey nonresponse for August was not notably higher than in prior months in Texas.
• The August job openings, hires, and separations were not notably different from prior months in the South region. (Texas is in the South region.)

Hurricane Irma made landfall in Florida and Georgia in September during JOLTS data collection for August. The storm struck after the reference periods for August, but it was difficult to reach some establishments in September to collect August data. Our research found the following:

• Florida establishments had a higher incidence of nonresponse for August. Georgia did not show a notable increase in nonresponse. Although the nonresponse had minimal impact on survey estimates, BLS continues to try collecting more August data in the affected areas.
• The August job openings, hires, and separations were not notably different from prior months in the South region. (Florida and Georgia are both in the South region.)

The September reference month for JOLTS will be most affected by Hurricane Irma. The September data will be published in November.

We have more information about how Hurricanes Harvey and Irma affected August and September 2017 JOLTS data, including survey response rates. Data for August were released on Wednesday, October 11, 2017. Data for September were released on November 7, 2017.

JOLTS estimates do not include Puerto Rico or the U.S. Virgin Islands.

Quarterly Census of Employment and Wages

The Quarterly Census of Employment and Wages is not a survey but a complete count of all 9.9 million worksites covered by the unemployment insurance system. More than 90 percent of employment and wages typically are reported each quarter.

Hurricanes Irma and Maria reduced the response rates for Puerto Rico and the U.S. Virgin Islands. No changes to estimating techniques were needed. The Virgin Islands response rate was affected starting with the second quarter of 2017.

Quarterly Census of Employment and Wages reporting history for Puerto Rico and the U.S. Virgin Islands (employment for third month in quarter)

Year/Quarter	Reported	Imputed	Missing	Prorated
Puerto Rico
2018/2	91%	3%	0%	6%
2018/1	91%	4%	0%	6%
2017/4	92%	3%	0%	5%
2017/3	79%	15%	0%	6%
2017/2	89%	6%	0%	5%
2017/1	90%	4%	0%	6%
2016/4	93%	3%	0%	4%
2016/3	93%	3%	0%	4%
U.S. Virgin Islands
2018/2	91%	9%	0%	0%
2018/1	90%	10%	0%	0%
2017/4	77%	23%	0%	0%
2017/3	50%	49%	0%	1%
2017/2	54%	26%	0%	20%
2017/1	95%	5%	0%	0%
2016/4	94%	6%	0%	0%
2016/3	95%	5%	0%	0%
Definitions Reported: Worksite data collected directly from employer(s). Imputed: Worksite data not reported, replaced by estimated values from reporting establishments. Prorated: Individual worksite data not reported, replaced by proration of data to worksite based on data reported for a master record. Proration based on previously reported distribution to individual worksites. Master Record: Usually a statewide report by an employer who has multiple worksites in a given state.

Producer Price Indexes

Hurricanes Harvey and Irma did not affect the overall September 2017 response rate for the Producer Price Index (PPI). However, some respondents from the affected areas may have been unable to report their data. As a result, BLS may receive their information late, or it may need revision. BLS imputes missing prices with the price changes for similar products for which prices were reported. All PPI data are subject to revision 4 months after first publication to allow for late reports and corrections. The recalculated September 2017 indexes would reflect any late pricing information when they are published with the release of January 2018 PPI data.

Consumer Price Index

The Consumer Price Index (CPI) program collects data throughout the month. BLS collects prices for commodities and services during three pricing periods, each for about a third of the month. BLS collects prices for shelter during the entire month.

Hurricane Harvey had a small impact on survey response rates in August. Price collection late in the month was disrupted in 2 of the 87 collection areas, both in Texas (Houston and Beaumont). The impact on collection in September was minimal.

Hurricane Irma made landfall during the second pricing period in September. Data collection was affected in the following areas in Florida that were on-cycle for collection in September: Tampa, Fort Myers, Melbourne, and Ocala. The Miami area was off-cycle for data collection in September.

Hurricane Maria did not affect CPI data collection because the CPI does not collect price data in Puerto Rico or the U.S. Virgin Islands.

A list of collection areas and their pricing cycles is available in Appendix 4 in the BLS Handbook of Methods.

BLS released September CPI data on October 13. The data are available from the news release, interactive charts, and supplemental files.

Import and Export Price Indexes

Hurricanes Harvey and Irma had a small impact on data collection for import and export price indexes for September 2017. Some respondents from the affected areas could not report their data. Hurricane Maria did not affect data collection in September 2017 because the reference period is the first week of the month.

Hurricane Maria did affect data collection for October 2017 from Puerto Rico because most respondents could not report their data. Trade between Puerto Rico and other countries counts as imports into and exports from the United States.

BLS may receive information on import and export prices late, which could lead to revisions in the indexes. BLS first imputes missing prices using the price changes for similar items for which prices were reported. All data are subject to revision in each of the 3 months after first publication to allow for late reports and corrections. Recalculated September 2017 indexes would reflect any late pricing information when BLS publishes the indexes for October, November, and December 2017. Recalculated indexes for October would reflect any late pricing information when BLS publishes the indexes for November and December 2017 and January 2018.

Employment Cost Index

Hurricanes Harvey and Irma had minimal impact on overall September 2017 Employment Cost Index response rates. In the affected areas, however, there was an increase in the number of establishments that temporarily closed and instances where BLS could not contact respondents. BLS imputes missing data using the methods described in the National Compensation Measures Handbook of Methods and Accounting for missing data in the Employment Cost Index.

To collect the most data possible, BLS changed its procedures in the hurricane-affected areas to allow establishments to report employer costs for wages and salaries and benefits outside the standard reference week (the week including September 12). This change was only used when the respondent could not provide data for the standard reference week and the collected costs were more recent than those collected for the June 2017 Employment Cost Index.

Hurricane Maria did not affect data collection because the Employment Cost Index does not include Puerto Rico and the U.S. Virgin Islands.

The National Compensation Survey, which produces the Employment Cost Index, uses a national sample of private industry and state and local government establishments. Private industry and state and local government establishments together are called civilian. Excluded from the survey are workers employed in federal and quasi-federal agencies, the military, agricultural workers, volunteers, unpaid workers, people receiving long-term disability pay, and those working overseas. In addition, private industry excludes workers in private households, the self-employed, workers who set their own pay (for example, proprietors, owners, major stockholders, and partners in unincorporated firms), and family members paid token wages.

Last updated: January 30, 2019