Agriculture, Forestry, and Fishing Research at NIOSH: Reviews of Research Programs of the National Institute for Occupational Safety and Health (2008)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

E Methods for Identifying the Agriculture, Forestry, and Fishing Workforce Population DISTINGUISHING BETWEEN EMPLOYMENT AND WORKFORCE POPULATION Efforts to measures the size of a population at occupational risk are often based on a determination of “employment” in the economic sector of interest. It is usu- ally assumed that that term refers to the number of people who are self-employed or are employed by others. It is implicit that those people do not change their employment status throughout all or most of the calendar year. Thus, measures of employment are sometimes thought of as more or less equivalent to determinations of the number of people actually working. However, the agriculture, forestry, and fishing (AFF) sector is unusual in three respects. First, many workers are employed for only a portion of the year. Second, there is a high rate of turnover of hired and contract workers; a great many workers are known either to enter or to leave the AFF sector workforce in the course of a year. Fully 16 percent of the nation’s hired crop-farm workers in 2001-2002 were determined to have been immigrant “newcomers”, that is, foreign-born persons who had been working in the United States for less than a year when interviewed by the U.S. Department of Labor (DOL). Presumably, the newcomers replaced people who had left crop-farm employment. Third, the AFF labor force includes  United States Department of Labor. Office of Assistant Secretary for Policy. Office of Program- matic Policy, Findings from the National Agricultural Workers Survey (NAWS) 2001-02. A Demographic and Employment Profile of United States Farm Workers, Research Report No. 9, March 2005, p. 8. 301 

302 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH children and the elderly, many of whom would classify themselves as retired but work actively on the farm during planting and harvesting. Many children and older adults do not receive money and do not appear on farm records. Confounding the difficulty is that the AFF labor force is known to be much larger than the corresponding level of employment; that is, at various times of the year, a great many in the AFF hired and contract labor force are unemployed and unable to find work. Finally, a large proportion of AFF workers are foreign-born, and a sizable “reserve labor pool” is in the countries of origin, including workers who may have temporarily returned to their homes. For those reasons, it is useful to distinguish measures of “population” from determinations of “employment”. Population refers to the number of people; em- ployment refers to their working status. Employment is often measured in terms of full-time equivalent (FTE) workers on the basis of temporal averages, usually derived from 12 monthly reports. Thus, two people who find half-time jobs in the AFF sector for a full year will be counted as a single FTE worker in measures of employment. Even for self-employed workers, the distinction is important. To illustrate, the Census of Agriculture asks farm operators to report the number of days on which they were employed off-farm. In the 2002 Census of Agriculture, more farm opera- tors reported having worked at least some days off-farm than reported no off-farm work. The majority of those who said that they worked off-farm at all said that they did so for 200 or more days per year; these farm operators might be counted twice in measures of employment, as would be the case for workers in any industry who “moonlight”, holding two jobs at the same time. HARNESSING THE QUARTERLY AGRICULTURAL LABOR SURVEY FOR AFF SURVEILLANCE The Quarterly Agricultural Labor Survey (QALS) is the only national survey of the agriculture workforce, conducted by the U.S. Department of Agriculture (USDA) and reported in the periodical Farm Labor. The QALS has recently been called the Farm Labor Survey (FLS). The survey is limited to farm employment. Initiated in 1910 and conducted with only a few interruptions nearly every year since then, the FLS is an employer survey that obtains reports of employment and other characteristics, such as wage rates paid and hours worked. A nationally representative sample of farm operators and agricultural service firms (mainly farm labor contractors) is contacted to determine the number of their hired farm  United States Department of Agriculture. National Agricultural Statistics Service, 2002 Census of Agriculture. Summary and State Data, Volume 1, Geographic Area Series, Part 51, June 2004. Table 55. Summary by Size of Farm: 2002, p. 69. 

A pp e n d i x E 303 laborers during the week that includes the 12th day of the first month in each of the four calendar quarters (January, April, July, and October). Data on hired farm workers from other sources, such as the Census of Agriculture and the Census of the Population decennial population surveys, do not provide seasonal data on employed workers. The data provided by the FLS are not available from any other source. It provides the only national data on farm labor employment and wage rates and also provides regional and seasonal data. It is the analogue for agriculture of the monthly Current Employment Statistics (CES) payroll survey of the Bureau of L ­ abor ­Statistics (BLS), which is watched closely by policymakers and business econ- omists. Agriculture is deliberately excluded from the CES survey. Unlike the CES, the FLS is based on reports for only 4 months, one in each calendar quarter. The FLS is unique in that it seeks to survey a sample of all farm employers, no matter how small their payroll or number of employees. Other BLS data—such as the Census of Employment and Wages (CEW), which is based on quarterly unem- ployment insurance reports—are incomplete for agriculture because most small farms are excluded from the sample frame. (That is because many states specifically exclude small farm employers from unemployment-insurance [UI] coverage.) Nor is the FLS subject to the so-called small farm exemption that precludes the National Institute for Occupational Safety and Health from surveying farms that have fewer than 11 employees. All types of farms are included in the FLS, both crop and livestock producers. Data are reported for the United States as a whole and for each of the 18 USDA crop regions, including separate reports for each of California, Florida, and Hawaii, which constitute crop regions in their own right. The FLS relies on employer reports of their payroll, so all hired workers, irre- spective of age, are represented. Reports compiled by BLS usually report employ- ment only of persons who are at least 16 years old. The FLS sample frame comprises a comprehensive list of farms maintained by USDA’s National Agricultural Statistics Service (NASS) and an area frame (a random sample of parcels of farmland). The latter aspect is peculiar to the QALS: no other employment survey uses this method. It is important that the FLS survey includes all types of employers of hired farm workers, such as crew leaders and labor contractors, who have been deliberately ex- cluded from Economic Census coverage. The Census Bureau abandoned its Census of Agricultural Services, which had included farm labor contractors, after a failed effort in 1978. Agricultural service employers were first added to the FLS in 1987. For many years, the FLS included determinations of the numbers of self- employed farm workers and unpaid family workers. However, it ended that com- 

304 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH ponent of the survey in 2002 owing to budgetary constraints (Mark Aitken, USDA, private communication, February 19, 2007). The FLS asks employers to provide information about wage rates and the numbers of field workers and livestock workers. The FLS provides annual average wage data on directly hired farm workers in each of the 50 states. The data generated by the FLS are clearly limited to employment in each of 4 months of the year. No effort is made to determine how many people are employed year-round. However, respondents are asked to report the number of persons whom they expect to employ for 150 days or more and the number expected to be employed for less than 150 days. Because the excellent sampling frame used for the survey (list and area frame) was developed for use by other NASS surveys, the FLS does not incur special sam- pling costs. In fact, it is an efficient data-gathering tool. Information is collected by telephone and processed with other survey data gathered by NASS; thus, statisti- cians and other resources are already shared. The FLS provides excellent information on a regional and large-state basis that can be combined with data from other sources, such as the Agricultural Census, to estimate the numbers of farm workers in smaller geographic areas, such as counties. The FLS also can be combined with DOL’s National Agricultural Workers Survey (NAWS) to make fine-grained estimates of the demographics of farm workers by region and large state. Those estimates can even be made by season. The age, sex, place of origin, migration patterns, housing patterns, education levels, and use of social services can be estimated by combining the NAWS and the FLS. Furthermore, the FLS is used to support various other government programs. The NAWS itself uses FLS data to implement its sample and weight its results. The NAWS, FLS, and Agricultural Census are being used in various configura- tions, but always with a big role for the FLS, to allocate resources for the National Farmworker Jobs Program, the Migrant and Seasonal Head Start Program, the Legal Services Corporation migrant program, and the Migrant Health Program. Other agencies, such as the 1992 Commission on Agricultural Workers and the Congressional Budget Office, have used the FLS, usually in conjunction with other data sources, to describe the farmworker population. The H2A agricultural guest worker program uses the wage data from the FLS to set its adverse-effect wage rate for the visiting workers. Moreover, estimates from those data sources are used by farmworker programs and by state policymakers to design, implement, and obtain resources for their activities. Recently, the Congressional Research Service relied on the FLS to determine that there was, as of 2006, no national shortage of hired farm laborers. Citing the annual FLS reports for the period 1990-2006, the official report to Congress dem- onstrated that hired farmworker employment varied only slightly throughout the 

A pp e n d i x E 305 TABLE E-1  Directly Hired Farm Workers and Agricultural Service Workers in the United States, 2006 Workers, 150 Workers, Less Agricultural Total, All Hired Week Days or More Than 150 Days Service Workers Farm Workers Jan. 8-14, 2006 512,000 102,000 180,000 794,000 Apr. 9-15, 2006 581,000 139,000 241,000 961,000 Jul. 9-15, 2006 630,000 246,000 320,000 1,196,000 Oct. 8-14, 2006 592,000 205,000 280,000 1,077,000 SOURCE: USDA, Farm Labor. See http://usda.mannlib.cornell.edu/MannUsda/viewDocumentInfo.do?documentID=1063. 17-year period (Levine, 2007). Table E-1 presents the latest findings of the FLS, covering all 4 sampling weeks of 2006. It is possible that some individual workers may be double-counted because they performed jobs for two or more employers during a sampling week, but effect is probably small, especially because a majority of the reported employment is accounted for by persons working directly for farm operators for 150 days or more (sometimes described as regular or permanent employees). NATIONAL CENSUS OF POPULATION AND HOUSING The most comprehensive effort to enumerate the U.S. population and its characteristics is the decennial Census of Population and Housing (commonly re- ferred to as the Census). The Census is important in its own right but also because additional surveys—such as the Current Population Survey (CPS), the American Community Survey (ACS), and the Hired Farm Work Force Report—have relied on its sample frame. The Census 2000 Special Equal Employment Opportunity (EEO) Tabulation also relied entirely on an analysis of Census data and provided detailed occupation and education data. As further described below, the Census data and surveys based on its sample frame are probably accurate with respect to self-employed workers in the AFF sec- tor and for most regular or year-round workers in this sector. However, it is well established that the Census does not accurately enumerate a great many hired and contract farm laborers, as was officially delineated by its senior administrator in 1994. Scholarship has also demonstrated that deficiency (Gabbard et al., 1993).  Letter from Everett M. Ehrlich, Administrator, Economics and Statistics Administration, U.S. Department of Commerce, October 24, 1994. 

306 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH The Census relies primarily on a mail-return short form requested of all Ameri- can households and a long form from a random sample of about one-sixth of them. For literate English-speakers who reside in a dwelling with a physical address, the response rate is quite high. In contrast, for non-literate, low-income, non-English speaking immigrants, who may be undocumented or reside in informal dwellings that lack physical addresses, the response rate is low. The Census undercount is a continuing problem to which considerable effort and resources have been allocated in an effort to account for those missed by the mail-return forms. As a result of those considerations, government programs that serve hired farm laborers no longer rely on Census figures to estimate the size of this component of the AFF workforce. As a consequence, the use of Census data to measure the number of American hired farm laborers by government surveys, such as the CPS and the ACS, that rely on the Census sample frame have also become suspect with respect to their ability to enumerate hired farm laborers. In the decennial Census, employment status information is collected for the workweek before the reference date of April 1. The information is reported on the “long form”, gathered from a roughly one-sixth sample of all households, or about 18 million of the 105 million households. The specific question determining employment status in the 2000 Census read as follows: LAST WEEK, did this person do ANY work for either pay or profit? Mark the “Yes” box even if the person worked only 1 hour, or helped without pay in a family business or farm for 15 hours or more, or was on active duty in the Armed Forces. Note the careful wording regarding possible unpaid work on a family farm. Inclu- sion of that phrase makes it possible for the Census to seek to enumerate unpaid family workers in the AFF sector. However, the inquiry regarding industry of employment refers to either the current job or, if there is no current employment, the most recent job held, even if it was as far back as 1996. The question regarding industry status read as follows: Describe clearly this person’s chief job activity or business last week. If this person had more than one job, describe the one at which this person had the most hours. If this person had no job or business last week, give the information for his/her last job or business since 1995. Similarly, the inquiry regarding occupation status refers to the respondent’s industry status, clearly, albeit implicitly, referring to the most recent job since 1995.  United States Department of Commerce. Bureau of the Census, United States Census 2000, Form D-2, Question 21, p. 6.  Ibid. Question 27, p. 7. 

A pp e n d i x E 307 TABLE E-2  Classes of Workers: Civilian Population 16 Years Old and Older Employed in Agriculture, Forestry, Fishing, and Hunting (AFFH) Class of Worker, AFFH Sector Total Workers, Male and Female Self-employed in own business 836,417 Employees of private for-profit business 938,663 Unpaid family workers 75,938 Government workers 63,848 Private not-for-profit wage and salary workers 14,817 Total, all types of workers (above) 1,929,683 SOURCE: Census 2000, SF 3, Table P51. Thus, the data on industry status and occupation status refer to the full popula- tion whereas the data on employment status refer only to persons employed during the 1-week period prior to April 1. The data on the nation’s employed workforce are summarized in Tables P49, P50, and P51 of Summary File 3 (SF 3), in which cross-tabs by sex are reported for all employed persons at least 16 years old by industry, occupation, and class of worker, respectively. A brief overview of the data on the AFF workforce derived from Table P51 is reported in Table E-2 above. In addition, the Census reports include summary data on the size of the rural population and the size of the farm resident population. Finally, the Census long form has an intriguing inquiry regarding the sale of agricultural products. Specifi- cally, for residents of a single-family dwelling or mobile home, respondents were asked to report “the actual sales of all agricultural products from this property” in 1999. OCCUPATIONAL CENSUS REPORT The Census Bureau also prepares a Special Equal Employment Opportunity (EEO) Tabulation on behalf of the federal EEO Commission, colloquially known as the “Occupational Census” report. It is subject to the same limitations as the Census in seeking to enumerate AFF workers, but with respect to English-speaking persons who are long-term residents of a dwelling that has a physical address the findings are likely to be reasonably accurate. However, because of the Census un- dercount, it is likely that the Occupational Census does not accurately reflect the full population of hired farm laborers.  Ibid. Question 44(c), p. 9. 

308 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH Table E-3 presents the EEO Occupational 1990 and 2000 tabulations for AFF occupational categories. It is important to note that the findings regarding hired farm laborers likely reflect a substantial undercount. Moreover, self-employed fam- ily workers are not represented in the table, because they do not correspond to an officially recognized occupation. CURRENT POPULATION SURVEY The CPS is a monthly survey of about 60,000 randomly selected occupied American households in 754 primary sampling units (out of the nation’s roughly 3,000 counties, or groups of contiguous counties). Its main purpose is to deter- mine changes in population, demographic characteristics, and economic status, especially employment and unemployment. In keeping with standard practices of the BLS, the reference week is the one that includes the 12th day of the month. The BLS adopted the North American Industry Classification System (NAICS) in 2002, bringing its industry and occupation categories into conformity with those of all other federal agencies, including NIOSH (Bowler et al., 2003). One important change resulted from adoption of the NAICS: forestry, fishing, and hunting had been classified in the much broader “services” sector, separately from agriculture, but are now included with agriculture to form the new AFF sector. TABLE E-3  Census Special Equal Employment Opportunity (EEO) Tabulations, United States, 1990 and 2000 Occupation 1990 2000 Farmers and ranchers 829,919 587,015 Farm, ranch, and other agricultural managers 257,446 201,980 First-line supervisors of farm, fishing, and forestry workers 55,503 57,440 Hired farm workers 797,405 806,075 Fishing and hunting workers 58,493 51,100 Forestry and conservation workers 20,431 18,980 Logging workers 115,524 105,675 Total, all types of occupations (above) 2,134,721 1,828,265 NOTE: Redefinition of some occupational categories in 2000 required combining some 1990 figures to correspond to the new definitions. Moreover, “Hired farm workers” were termed “Farm workers” in 1990 and “Miscellaneous agricultural workers, including animal breeders” in 2000. SOURCE: http://www.census.gov/hhes/www/eeoindex/eeoindex.html. 

A pp e n d i x E 309 As previously indicated, the Census sample frame, the Master Address File (MAF), updated with additional physical addresses, is used for the CPS. However, unlike the decennial Census, the CPS has no procedure for updating the MAF re- garding informal dwelling locations that lack physical addresses. A cross-sectional statewide survey of hired farm laborers in California found that a substantial portion reside in informal dwellings that lack physical addresses (such as shacks, garages, and illegal trailers) and would probably be missed by the CPS (Villarejo and McCurdy, 2008). About 70 percent of the CPS is conducted by telephone, using local residents of the areas to be covered. But community-based survey research indicates that one-fifth of dwellings occupied by hired farm laborers in California lacked land- line telephone service (Villarejo et al., 2001); this may lead to bias in findings with respect to hired farm laborers in a telephone survey. Another bias in telephone surveys of hired farm laborers is the reluctance of some workers to be interviewed by strangers asking personal questions. A recent analysis compared findings in California obtained from a statewide, population- based telephone survey of all adults with findings of a statewide cross-sectional household survey of hired farm laborers in which biliterate, bicultural staff con- ducted in-person interviews (Mines, 2005); the former survey appeared to be successful in reaching English-speaking, home-owning farm workers but failed to adequately include non-English speaking farm laborers who were renters. The CPS employment-status findings are limited to people at least 16 years old (BLS, 2007b). Farm employment includes some workers under 16 years old and is legally permissible for persons as young as 12 years old and, in exceptional circumstances, even younger (DOL, 2004). Hence, the AFF workforce will be un- dercounted by an unknown amount in the CPS. Employment and unemployment findings from the CPS are closely monitored by government officials and economists and are even cited by Wall Street analysts who regard them as indicators of the health of the American economy. It is gener- ally agreed that the CPS findings are very accurate with respect to people who speak English and do not migrate to find work. But scholarship has demonstrated that the CPS does not accurately represent foreign-born, non-English speaking farm laborers (Mines, 1998; Larson et al., 2002). In that regard, the CPS suffers from the same deficiencies as the Census. The annual March supplement to the CPS seeks to determine detailed demo- graphic and other characteristics of the American population. Farm labor scholars have pointed out that March is not the best month in which to survey hired farm laborers working in the United States, and that the March supplement is likely to yield findings on ethnicity, race, and foreign-born status of hired farm laborers that differ substantially from those of the full population. However, it is likely that 

310 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH the CPS accurately reports the employment status of other segments of the AFF workforce, especially self-employed workers and unpaid family workers. Thus, with the caveat that directly hired and contract workers are not properly enumerated in the CPS, the other segments of the AFF workforce are likely to be accurately determined. In comparing the findings of the CPS with those of the FLS, it has been demon- strated that the FLS systematically reported hired farmworker employment about 30 percent higher than the CPS during the period 1984-1997 (Mines, 1998). The employer reports in the FLS are probably more accurate with respect to this seg- ment of the AFF workforce than the partial sample obtained by the CPS. A recent report by the Congressional Research Service provides updated infor- mation for the period after 1997 and confirms the earlier findings (Levine, 2007). As indicated in Table E-4, the FLS reports from employers regarding their hired farm workers are 29-55 percent larger than the findings from the CPS, as analyzed by the USDA Economic Research Service (ERS). Note that the ERS analysis includes hired farm workers at least 15 years old whereas findings published by the CPS are for workers at least 16 years old. The committee has identified other survey efforts that could be considered for specific, limited, single-purpose use when conducting surveillance of AFF sectors, including the Hired Farm Work Force report, the Current Employment Statistics, and the Census of Employment and Wages. The 5-year Census of Agriculture is considered separately below. TABLE E-4  Hired Farmworker Employment, 1998-2006, Annual Average Economic Research Service (ERS) Analysis of Current Population Survey (CPS) (at least 15 years old) versus Farm Labor Survey (all ages) Year Hired Farm Workers, ERS (CPS) Hired Farm Workers, QALS 1998 875 1,126 1999 840 1,162 2000 878 1,133 2001 745 1,125 2002 793 1,111 2003 777 1,072 2004 712 1,102 2005 730 1,062 2006 748 1,007 SOURCE: Levine, 2007, Table 1, p. CRS-10. 

A pp e n d i x E 311 THE HIRED FARM WORK FORCE The USDA ERS summary, The Hired Farm Work Force, sought to provide pe- riodic reports on the demographic and economic status of the nation’s hired farm laborers, but was discontinued after 1987 and replaced by the NAWS. The 2000 ERS report, Profile of Hired Farmworkers, 1998 Annual Averages, has indicated that those analyses have important limitations, associated mainly with the difficulty of enumerating Hispanic farm laborers (Runyan, 2000). The data for these reports were derived from the CPS and regarded by scholars with the same degree of skep- ticism as the CPS (Larson, 2002). AMERICAN COMMUNITY SURVEY The ACS is the most recent initiative of the Census Bureau. It is a monthly survey of 100,000 randomly selected households designed to replace the “long form” of the Census. A major portion of the sample is rotated each month, and this enhances the statistical power of the findings by accumulating data for a sequence of several months. ACS uses the Census sample frame. Like those of the CPS, its sampling methods and survey methods lead to underreporting of hired farm laborers. However, the cumulative sample over many months yields findings that are more statistically stable than some findings of the CPS. CURRENT EMPLOYMENT STATISTICS The CES is a monthly survey of 160,000 businesses and government agencies, representing 400,000 worksites and seeks payroll and employment information for the week that includes the 12th day of the month. It is sometimes referred to as the payroll report of employment. All agricultural workers and self-employed workers are excluded. Therefore, the CES is not helpful for the purposes of the NIOSH AFF program. CENSUS OF EMPLOYMENT AND WAGES The BLS also compiles the CEW, which is based on the quarterly reports of monthly employment and quarterly total wages required of employers subject to unemployment insurance requirements. As in the case of other BLS employment data, monthly employment refers only to persons on the payroll during the pay period that includes the 12th day of the month. 

312 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH The main problem in relying on CEW data for the AFF sector is that many farm employers are exempt from UI requirements. That follows from the fact that each state sets its own criteria for UI coverage. Many states, such as California and New York, have “universal” coverage, typically requiring every private-sector employer that pays at least $100 in wages or salaries in a calendar quarter to pay UI taxes. But many states specifically exempt farm employers that have quarterly payrolls below a specified threshold, typically $50,000. Thus, in effect, an unknown portion of the AFF hired farm labor workforce is not reported, because some employers are exempt from payment of UI taxes. Those workers are also ineligible for qualifica- tion to receive UI benefits. CENSUS OF AGRICULTURE The committee has specifically reviewed the use of the Census of Agriculture (CoA) as a sector-surveillance tool. It is part of the 5-year Economic Census and is based on responses to a mail-return census form that NASS sends to its master mail list of farm operators. The findings are regarded as the most comprehensive body of systematic data on farmers and the farm sector. Most important, nearly all findings are reported at the county, state, and national levels. Farm operators report such factors as the number of days worked off farm, whether they consider their principal occupation to be “farmer”, and whether they reside on farm. Age, sex, and other demographic features of farm operators are also reported. Data are reported on hired-labor and contract-labor expenses, the number of directly hired workers, the number of persons employed for 150 days or more, and those employed for less than 150 days, at each geographic level. Cross-tabs of these data items are also available for each NAICS category of farms and for other standard measures of farm operators. A given worker may find employment with two or more farm operators in the course of a single year, so the CoA’s “number of workers” is more properly regarded as the number of farm jobs, not workers. Clearly, farm employment and the hired worker population cannot be determined from these data items (Gabbard et al., 1993). NATIONAL AGRICULTURAL WORKERS SURVEY The NAWS is a national cross-sectional survey of hired crop-farm workers; livestock workers are excluded. The survey is employer-based, and farm employ- ers, whether farm operators or labor contractors, of all payroll sizes are included in the sample frame. In-person interviews, normally conducted away from the 

A pp e n d i x E 313 worksite, with workers who agree to cooperate, are administered by professional staff members (bilingual, biliterate, and mostly bicultural). Three seasonal cycles of interviews are conducted in each federal fiscal year (FY) to provide persons who are employed in only part of each calendar year an opportunity to participate. The most recently published report of the NAWS is based on findings from 6,472 interviews conducted during FY 2001 and FY 2002 (DOL, 2005). The NAWS was not intended or designed to enumerate workers or to provide quantitative reports of total hired crop-worker employment. Begun in 1988 by DOL, it was a response to the congressional mandate of the Immigration Reform and Control Act of 1986 (IRCA). The specific purpose of the NAWS under the IRCA mandate was to determine whether persons who were newly authorized for employment in the United States under the seasonal agricultural worker (SAW) visa program continued in farm work after their immigration status had been adjusted. If it were determined that persons holding SAW visas were leaving seasonal crop work and being replaced by unauthorized workers before federal FY 1992, a short- age of legally eligible workers would be officially declared, and a new replenishment agricultural worker (RAW) visa program would automatically be triggered. During the first 3 years of NAWS surveys, it was determined that the “exit rate” of SAW-visa holders from crop agriculture was negative, that is, more SAWs entered than left agricultural crop work in the United States each year. As a consequence, the RAW visa program was allowed to “sunset” because it was not needed. The NAWS is an unusual survey in that it seeks to obtain detailed work and family histories; information on workplace compliance with labor regulations; current income, workplace and job conditions, and immigration status; and other hard-to-obtain information. Such information has been regularly gathered from NAWS participants in every federal fiscal year since the October 1988 start date. Owing to budgetary and policy considerations, the number of interviews con- ducted each year has varied considerably. When combined with data from other sources, such as the Census of Agricul- ture or the FLS, the NAWS has been effectively used to provide otherwise hard- to-estimate numbers, such as estimates of the number of persons who qualify for participation in federal programs intended to serve migrant farm laborers. In addition, the NAWS conforms its sampling procedure to the same 18 USDA crop regions as form the basis of the FLS (and in fact uses FLS data to assist in determin- ing the proportion of interviews required in each crop region). As a result, data are available for a few individual states, such as California, that are themselves distinct crop regions. The NIOSH AFF Program entered into an interagency agreement with DOL to add an occupational safety and health supplement to the NAWS for federal FY 1999 (three NAWS interview cycles, starting in October 1998 and concluding in 

314 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH September 1999) supported by NIOSH funding. The interviews included inquiries specifically designed to probe the occupational safety and health status of hired farm laborers, and some of the queries were permanently added to the NAWS after the end of NIOSH supplementary support. A NIOSH report on the findings of the occupational health supplement, by Andrea Steege and Sherry Baron, has been completed and has undergone extensive outside review. At this writing, the report is moving through the NIOSH internal approval process. EMPLOYMENT DATA ON INDIVIDUAL STATES A number of important farm states, such as California and Washington, have universal UI requirements for virtually all private-sector workers. Thus, compre- hensive employment data are regularly published or otherwise made available on workers in the major industry sectors, often at both the state and regional levels and sometimes also at the county level. Employment data are compiled for the pay period that includes the 12th day of each month of each calendar quarter with corresponding payroll summaries. For example, California publishes a monthly Agricultural Bulletin and an annual supplement providing annual average employ- ment data for all the major NAICS categories related to farm employment (CA EDD, 2007). The findings are based on a random cross-sectional sample of firms that file quarterly reports to meet their UI obligations. Findings for the state as a whole and each of the state’s six crop regions are reported. The California findings can be compared with the findings for that state as determined by federal surveys, such as the FLS. Such a comparison can yield useful information on the effectiveness of coverage of federal and state surveys. Table E-5 shows one such comparison; it compares a special compilation of all individual quarterly reports submitted by farm employers (universal reporting), Agricultural Bulletin survey data, and FLS survey data on California employment by farm labor contractors. Because quarterly tax reports of employment and payroll by employers determines their UI tax obligations, whereas the FLS and the Agricultural Bulletin are based on self-reports of total employment, both surveys are likely to understate a substantial portion of farm-labor contractor employment in California. WORKERS’ COMPENSATION INSURANCE REPORTS More than a dozen states require workers’ compensation insurance coverage for nearly all private-sector employees. Some states, such as Washington, require the  Sherry Baron, private communication, March 6, 2007. 

A pp e n d i x E 315 TABLE E-5  Hired Farmworker Employment, Farm Labor Contractors, California, 2000: Comparison of Farm Labor Survey, Agricultural Bulletin Surveys, and ES202 Reports Agricultural Universal Reports from Month and Week Farm Labor (FLS) Bulletin (CA-EDD) All Employers January 9-15 75,000 69,700 96,017 April 9-15 85,000 86,600 134,475 July 9-15 99,000 150,300 177,409 October 8-14 86,000 109,100 135,949 coverage through the public sector, in this case the Department of Labor and Indus- tries. Other states, such as California, permit coverage through private insurance providers or through a public agency, the State Compensation Insurance Fund. Actuarial reports are annual reports that summarize claim frequencies by occupational risk category. There are 14 nationally recognized risk categories related to agriculture, and claim records of exposure of all workers in each risk category (based on payroll totals) and numerous other details are regularly pro- duced to enable insurance providers to base premium rates on claim frequency and experience. Paid workers’ compensation claims have been reviewed and analyzed to obtain reasonably accurate estimates of employment and estimates of the cumulative prevalence of injury and illness in hired farm laborers in California and other states (Villarejo, 1997). The actuarial agency responsible for rate-setting in California has used its own analysis of the surveillance data available from the records of paid claims to determine which factors are important in reducing workplace injuries and illnesses (WCIRB, 2002). To illustrate the potential usefulness of those data for surveillance purposes, Table E-6 presents the 10-year summary of hired farm-laborer paid claims under workers’ compensation insurance in California. It is not known whether all eli- gible persons filed such claims, but it is very likely that the most serious incidents, fatalities and permanent disabilities, resulted in payment under workers’ compen- sation. The data in Table E-6 refer to all paid claims in the 14 classification codes that refer to on-farm work, both crop and livestock. The term “Claim Frequency Report (Level 5)” refers to summary information on paid claims through the fifth year following the policy year. It is necessary to carefully consider the number and actual costs associated with paid claims through a long period after the year in which the incident took place because some claims are initially challenged by the 

316 A g r i c u lt u r e , F o r e s t r y , and Fishing Research at NIOSH TABLE E-6  Paid Claims Under Workers’ Compensation Insurance, Hired Farm Workers, California, 1990-1999, Claim Frequency Report (Level 5) Type of Claim Number of Claims Fatalities 455 Major permanent disability 12,932 Minor permanent disability 31,958 Temporary disability 68,357 Medical only/no indemnity payment to claimant 185,029 Total 298,731 SOURCE: Classification Experience Reports, 1990-1999 (Level 5), Workers’ Compensation Insurance Rating Bureau of California, San Francisco, California. employer or the insurer (“open claims”) whereas others may involve lengthy, multi­ year rehabilitation or medical treatment. Under California law, an occupational injury or illness that requires hospitalization or leads to loss of more than 3 days of work results in an indemnity payment in lieu of lost wages; if the number of lost workdays is lower, no indemnity is paid. Table D-6 does not show the actu- arial analysis of the most serious incidents (those amounting to a loss of $5,000 as measured by combined medical and indemnity costs), which includes the nature of an injury, the nature of an accident, the body part injured, and other variables. The latter data could inform occupational safety research. Surveillance of AFF workers would be best accomplished by thinking first in terms of all workers in each AFF sector. From there, it makes sense to consider surveillance of all five categories of workers—self-employed workers, unpaid family workers, directly hired laborers, contract laborers, other employees of large-scale firms—in each of the three AFF subsectors. Datasets on workers’ compensation are reasonably reliable for surveillance of fatalities and serious injuries or illnesses among hired workers but less reliable for minor cases.