April 12, 2018: NAWS 2018

  1. Summary
  2. From IRCA to NAWS
  3. Raw Ends, NAWS Continues
  4. US and California Data
    1. Demographics
    2. Employment and Earnings
    3. EDD and NASS
  5. NAWS Research
    1. Employment, Earnings and Migration
    2. Health
    3. Mexico
  6. Whither NAWS?
  7. Bibliography
  8. Agenda


NAWS Pic 1.jpgThe National Agricultural Worker Survey (NAWS) has become the most frequently consulted source of data on the characteristics, employment, and earnings of US crop workers. The purpose of the NAWS when launched in 1989 was to estimate the supply of labor to crop agriculture, but the NAWS has evolved into a survey that also collects detailed health and other data on non-H-2A workers employed on crop farms and their families.

NAWS data are used by many federal agencies, including those charged with predicting the effects of immigration reform and adjusting labor inputs for agricultural productivity measures. Researchers have used NAWS data to answer a wide variety of questions about farm workers.

After three decades, the NAWS is at a crossroads. The NAWS finds a largely Mexican-born, male, and unauthorized crop workforce that is aging and settling into families that often include unauthorized parents and US-born children. The NAWS cannot answer questions about several pressing farm labor issues, including the share of livestock workers who are unauthorized and the characteristics of H-2A guest workers.

The NAWS 2018 workshop reviewed the origins of the NAWS, including the decision to begin a new survey, current NAWS data, research using the NAWS, and the reactions of NAWS data users. Workshop participants hope that DOL will expand the NAWS sample and include livestock and H-2A workers as well as support research using NAWS data to better inform farm-labor policy making.


NAWS Pic 2.jpg The Immigration Reform and Control Act of 1986 included three major agricultural provisions:

  1. the Special Agricultural Worker legalization program for unauthorized farm workers who did at least 90 days of in Seasonal Agricultural Services (field work in perishable commodities) in 1985-86
  2. a revised H-2A guest worker program that allowed farmers whose need for labor was certified by the US Department of Labor to employ H-2A workers who are bound to them by contracts
  3. a 1989-93 free-agent Replenishment Agricultural Worker to replace exiting farm workers and prevent farm labor shortages

The SAW legalization program eventually legalized 1.1 million of the 1.3 million applicants,1 far more than qualified, the H-2A guest worker program shrank rather than expanded as unauthorized workers flooded into the US, and the RAW program was never implemented because there was no national farm labor shortage.

The NAWS grew out of the RAW program, which could have legalized unauthorized foreigners in the US or admitted new legal guest workers beginning October 1, 1989 if there were farm labor shortages.2 RAW workers would have been free agents allowed to live and work anywhere in the US and, if they did at least 90 days of SAS work a year, to become immigrants.

The formula to calculate the number of RAW workers was complex, taking up nine pages of IRCA, while the SAW legalization program took six pages. There was an absolute ceiling or maximum number of RAWS based on the number of SAWs, and a second shortage ceiling based on estimates of farm labor shortages in SAS in the next year, with the lesser of the two ceilings controlling the number of RAW visas issued.

For the absolute ceiling, the maximum number of RAWs in FY90 was  95 percent of the number of SAWs, minus the number of SAWs who did at least 15 days of work in SAS in FY89, and plus or minus the change in the number of H-2A workers in SAS crops in FY89 versus FY88.  For example, if the number of approved SAWs was 800,000, the ceiling on RAW admissions  in FY90 would be 760,000 minus, for example, 600,000 SAWs who did at least 15 days of SAS work in FY89, and minus say 10,000 additional H-2A workers if H-2A admissions in SAS crops increased from 20,000 in FY88 to 30,000 in FY89 (they did not). In this case, the absolute ceiling for FY90 RAW workers would have been 150,000.

Data for the absolute ceiling calculation were to come from the INS and farm employers, who were required to complete I-9 forms on all newly hired workers after October 1, 1989, note which newly hired workers had INS numbers in the A90 million series, and report the names, A-numbers, and days worked of such workers on an ESA-92 form to determine how many how many SAWs were staying in farm work and how many days they worked. The absolute ceiling calculation was made in terms of people, while the shortage ceiling was based on days worked, with the conversion from days to people made by dividing what was expected to be an average work year of 90 days. It is not clear what became of ESA-92 data.

The shortage ceiling on RAW admissions was based on USDA projections of the demand for labor in SAS agriculture for the year ahead, DOL estimates of the supply of farm labor, and farm employer reports of how many SAW workers they employed and for how many days the previous year. For example, if USDA determined that there were 180 million man days worked in SAS in FY89 and expected no changes for FY90, and if DOL determined that 20 percent of the SAS man days were lost due to exiting workers and no new workers would be available, the RAW shortage ceiling would be 20 percent of 180 million or 36 million man days, which would have been divided by the expected average 90 days of farm work by SAW workers to generate a need for 400,000 RAW workers (36 million/90).

The lower of the two ceilings controlled, so only 150,000 RAWs could be admitted based on the absolute ceiling. All calculations were nationwide, so there could have been shortages and surpluses of farm labor in particular regions.

USDA used NASS’s Farm Labor survey, then known as the Quarterly Agricultural Labor Survey (QALS), to estimate SAS man days worked in FY89, how many additional man days would have been needed to prevent crop losses due to labor shortages in FY88, and how many more or fewer man days would be needed in FY90 because of planned changes in technology and personnel practices. These calculations generated about 152 million man-days worked in SAS in FY88, with no changes expected.

DOL faced the challenge of calculating the number of man days lost to SAS agriculture due to retirements and the exit of SAS workers and the additional man days that would be available to SAS agriculture if farm employers improved wages, working conditions, and recruitment efforts.  This required calculations for three groups of workers: (1) exiting workers who did SAS work in FY88 but not FY89, (2) entrants who did not do SAS work in FY88 but did SAS work in FY89, and (3) potential SAS workers who would do SAS work in FY90 if farm employers improved wages, working conditions, or recruitment methods.

USDA had been providing $750,000 a year to the Census to add a December supplement to the Current Population Survey in order to determine the number of hired farm workers and their characteristics. These Hired Farm Work Force surveys during the 1980s found a mostly young and white hired farm work force of 2.5 million in 1987 that included fewer than 10 percent migrants, defined as persons who stayed away from home overnight to do farm work for wages. Hired farm workers averaged 112 days of farm work a year (Olivera, 1989).

DOL considered the HFWF unable to reliably collect data on SAS workers, obtained the $750,000 a year from USDA for four years, and contracted with Aguirre (now JBS) to interview a sample of SAS workers.  NASS did not provide its list of 15,000 farm employers used for the Farm Labor survey, so Aguirre developed an employer sampling frame by ranking COA 1982 crop labor expenditures by county, grouping counties into crop reporting districts, stratifying the districts, and selecting 60 counties from high, medium, and low labor expenditure strata, including five in California: Imperial, Kern, Kings, Sonoma, and Yolo. Aguirre compiled a list of employers in each county and secured their cooperation to interview crop workers.3


  1. Unauthorized foreigners who did at least 90 days of qualifying farm work in the 12 months ending May 1, 1986 could apply for SAW status, and the estimate that there were 350,000 unauthorized farm workers was made the ceiling for Group I SAWs, who could become immigrants after December 1, 1990. The SAW application form provided space to list 10 farm employers; most SAW applicants listed only one. The median age of SAW applicants was 28, half were 20-29, and 85 percent were Mexican-born men. Surveys of applicants suggested that most earned $30 to $35 a day for 100 days of farm work a year. Half of SAW applications were filed in California.
  2. There were four priorities for issuing RAW visas: (1) first for unauthorized aliens who are the spouses or unmarried minor children of legalized workers (family unity criterion) and who did at least 20 days of US farm work in any of the previous five years (farm work criterion); (2) aliens who satisfy only the family unity criterion; (3) aliens who satisfy only the farm work criterion; and (4) all others who apply.
  3. Today the major source of employers used to sample workers is unemployment insurance data.

RAW Ends, NAWS Continues

NAWS Pic 3.jpg The SAW, H-2A, and RAW programs were relatively last-minute additions to IRCA. There had been extensive hearings and research on agriculture’s dependence on unauthorized workers, but little agreement on how the government should respond to what farmers asserted would be farm labor shortages with fewer unauthorized workers.

There were no farm labor shortages in the early 1990s for several reasons, including an upsurge in illegal migration, with workers using false documents to be hired.  Since employers did not have to determine the authenticity of documents presented by workers, they faced no penalties if the INS audited their I-9 firms and found unauthorized workers. Employers were obliged to fire any unauthorized workers.

Continuing unauthorized Mexico-US migration and false documents spread unauthorized Mexicans through out the US. Over time, many farm workers moved into nonfarm food processing jobs, and eventually into construction, manufacturing and services. Later arrivals moved directly to urban areas for nonfarm jobs.

The SAW and RAW programs began as ways to assure agriculture a legal farm workforce in ways that protected workers and avoided farm labor shortages. Newly legalized SAWs soon learned that they had more opportunities outside agriculture, and there were no farm labor shortages because illegal immigration accelerated.

US and California Data

The NAWS interviews 1,500 to 3,000 workers employed on US crop farms each year; it does not interview H-2A and livestock workers. About two-thirds of the employers approached by interviewers allow their workers to be interviewed, and workers are paid $20 to answer detailed questions about themselves and their families, their current and past jobs, and their health and access to insurance in interviews that typically last 45 minutes.

There is no direct federal mandate for the NAWS, but many agencies and others use NAWS data to understand who works on crop farms, their employment and earnings, and the characteristics of their families. The survey strategy is reviewed regularly, and the 14 JBS interviewers are trained to follow strict protocols when locating and interviewing workers.

The NAWS began interviewing crop workers in 1989. Many trends have a V-shape, beginning high, falling to a nadir around 2000 when unauthorized Mexico-US migration peaked, and rising since the 2008-09 recession with less unauthorized migration.  For example, the share of authorized crop workers was over 85 percent in 1990, fell to less than half in 2000, and has been rising to almost 55 percent today.


The share of US-born crop workers was 40 percent in the early 1990s, fell to less than 20 percent in 2000, and has most recently been 25 percent for the past decade.  The share of crop workers born in Mexico has been stable at two-thirds in recent years.  About three-fourths of crop workers are male.

California is different.  The share of US-born workers has been less than five percent for most of the past two decades. California’s share of unauthorized workers, which was similar to the 15 percent in the US in the early 1990s, has been 60 percent or more since the mid-1990s because the state has a higher share of foreign-born workers, most of whom are unauthorized.  About 85 percent of California farm workers were born in Mexico and five percent in Central America; three-fourths are male.

Most crop workers are not migrants.  There is no single federal definition of a migrant farm worker.  The NAWS, which considers a worker to be a migrant if he moved at least 70 miles for a farm job, finds a declining share of migrants, about 20 percent in both the US and California.  Over 70 percent of these migrants shuttle from a home base in the US or Mexico to their farm jobs; only 30 percent had at least two US farm jobs 75 or more miles apart. Domestic and international shuttle migrants typically have only one farm employer where they do farm work.

The crop workforce is aging.  In 1990 and 2000, over half of US crop workers were 20-34. The share of workers in 20-34 age group has dropped below 40 percent for the US, but remains over 50 percent in California.  Average years of schooling for US crop workers were eight in 1990, seven in 2000, and eight today; 60 percent of US crop workers have nine or fewer years of schooling. California crop workers are less educated, with an average seven years of schooling.

The share of crop workers who speak English well fell from a quarter in 1990 to less than 20 percent in 2000 and is now 30 percent.  In California, the share of workers speaking English well has been less than 15 percent over the past two decades.

Almost 60 percent of US and California crop workers are married parents, compared with 40 percent in 2000, when the crop workforce included many newcomers from Mexico. Median personal income is $15,000 to $20,000 for US and California crop workers; a seventh earned less than $10,000, and a seventh earned more than $30,000. Median family income was $20,000 to $25,000, and a third of crop worker families had incomes below the poverty line.

Over half of US and California crop worker families received some type of needs-based public assistance during the previous two years, a sharp jump from less than a quarter in 1990 and 2000. In many cases, this needs-based assistance was for US-born children in farm worker families, including Medicaid, SNAP (Food Stamps), and WIC for mothers and infants. Only 10 percent of crop workers reported receiving unemployment insurance during the previous 24 months.

These demographic data portray an aging male and generally married crop workforce that is settling with families near their jobs.  The slowdown in unauthorized entries of new young workers means fewer migrants and more crop worker families receiving public assistance benefits, often for US-born children.

Employment and Earnings

Type of employer data also follows a V-shaped trajectory, starting high, dipping in 2000 and rebounding since.  For example, about 85 percent of US crop workers were hired directly in 1990, 73 percent in 2000, and 80 percent today; the California direct-hire shares were 73, 55, and 66 percent, that is, the California direct-hire share has not yet returned to 1990 levels.  

Average years of farm work experience fell from 10 in 1990 to eight in 2000 and is now 14 across the US. In California, average farm work experience fell from 11 years in 1990 to nine in 2000 and is now 16 years.  US and California workers were employed an average seven years for their current farm employer.

Average worker-reported earnings were $5.25 an hour in the early 1990s for US crop workers, $6.50 in 2000, and $10.20 today, compared to $5.55, $6.55, and $10.10 for California.  The California wage premium of earlier years has turned into a wage deficit.

US crop workers averaged over 190 days in 35 weeks of farm work recently, suggesting more than five days of work a week. California crop workers had even more days of farm work, an average 205 days in 36 weeks in recent years.  The share of US crop workers with at least one nonfarm job was over 30 percent in 1990, 15 percent in 2000, and 25 percent today.  The California shares are 16, six, and 17 percent, that is, California crop workers are less likely to have nonfarm jobs.

Most NAWS crop workers are employed in fruits and vegetables, about 60 percent of US crop workers and almost 90 percent of California crop workers.  In 1990, 35 percent of US crop workers were in vegetables and 28 percent were in fruits and nuts.  By 2000, it was 25 percent vegetables and 37 percent fruits and nuts, and today 35 percent of workers are in vegetables and 30 percent are in fruits and nuts.  In California, the 1990 vegetable share was 27 percent and the fruit and nut share 55 percent; these shares were 19 and 70 in 2000, and 26 and 63 percent today, that is, almost 90 percent of California workers are in fruits and vegetables.

The share of US and California crop workers in harvesting has been falling.  For the US, harvesting was the primary task when interviewed of 40 percent of US workers in 1990, 30 percent in 2000, and less than 20 percent today.  For California, the harvesting share fell from almost half to 30 percent to 25 percent today.  The most common current job today is semi-skilled, such as equipment operator: a third of US workers, and 37 percent of California workers, had such jobs when interviewed.

NAWS workers averaged 196 days of farm work during the previous 12 months, with unauthorized workers averaging the most days, 215, and migrant workers the fewest, 144.

Three fourths of workers plan to continue to do farm work for at least five more years.  In 1990, two-thirds of US workers said they would continue to do farm work as long as they could, in 2000 this dipped to 55 percent, and today over 75 percent of workers plan to continue to do farm work indefinitely; the California shares are 75 percent, 65 percent and 80 percent.  A declining share, about a third of US workers and a quarter of California workers, say they could find a nonfarm job within a month.

The NAWS data do not conform to stereotypes. For example, fruits and nuts are worth twice as much as vegetables, and hire large numbers of piece rate workers.  The NAWS interviews more non-harvest than harvest workers, and more vegetable than fruit workers, which may help to explain why almost 90 percent of workers reported being paid an hourly wage (piece rate workers are guaranteed a minimum hourly wage, but their hourly earnings are typically higher to provide them with an incentive to work).

The relatively few harvest workers paid piece rates may also help to explain why the hours worked averaged 46 for men and 40 for women, more than many piece rate workers are employed.  The relatively few piece rate workers in the NAWS averaged only two percent more than hourly paid workers.

Also surprising is that a higher share of workers employed in vegetables, 60 percent, were hired by FLCs than those employed in fruits, 35 percent, even though FLCs are often associated with fruits.


The NAWS interviews workers, while EDD and NASS collect data from employers. EDD collects 12 monthly snapshots of employment during the payroll period that includes the 12th of the month, and finds that average employment has been rising in California agriculture, from 396,000 in 2012 to 424,000 in 2017, with 85 percent of this increase in support services for agriculture, mostly farm labor contractors. This increased employment was concentrated in the San Joaquin Valley and the Central Coast, the regions that have 50 and 16 percent of the state’s average farm employment, respectively.

The San Joaquin Valley in 2017 had 35 percent of the state’s average direct-hire crop employment, 72 percent of average livestock employment, and 59 percent of statewide support employment.

Employer reports generate average employment; the total number of workers employed sometime during the year is larger.  Analysis of all SSNs reported by agricultural employers suggests two unique workers per average job, almost 850,000, including 705,000 or 83 percent “primary farm workers” whose had their maximum earnings in California agriculture. Almost 60 percent of primary farm workers had their maximum earnings with a crop support service firm such as a FLC, followed by 22 percent with a fruit and nut employer, and seven percent with a vegetable and melon employer.

About 60 percent of all farm workers had one farm job, 25 percent had two jobs, and 18 percent had three or more jobs.

EDD also conducts a Current Employment Statistics survey of 3,0000 farmers each month that generates employment for six regions of California. In contrast to the NASS survey, the CES includes nonfarm service firms who bring workers to farms and reports the data for detailed NAICS codes. The CES is generating timely employment estimates, but not earnings data. EDD is working with DOL to obtain hours worked data from employers paying UI taxes.

The NASS surveys 14,000 US farm employers, including 1,000 in California, in April and October for its Farm Labor survey; employers are asked about employment, earnings, and hours worked for the week containing the 12th of the month. Earnings are divided by hours worked to generate hourly earnings estimates for all hired workers, field workers, and field and livestock workers.

Average hourly earnings of field workers in California have generally been higher than those in the rest of the US, reflecting the state’s higher minimum wage and other factors. However, between 2009 and 2015, the California premium disappeared, only to re-emerge in 2017, when California hourly earnings of $13 for field workers were four percent higher than the US $12.50. In California, livestock hourly earnings are consistently higher than field worker hourly earnings.

The average number of hired workers on US and California farms has been trending downward, from 900,000 in the US  and 275,000 in California in 2000 to 725,000 in the US and 175,000 in California in recent years.  In California, workers hired directly by growers and employed on their farms more than 150 days outnumber those employed 149 days or less by almost four to one, down from two to one in 2000. California hired workers are employed more hours a week than US hired farm workers, an average 43 versus 40 hours.

The 2017 Census of Agriculture asks two questions about farm labor. Section 29 asks how many workers were hired, and distinguishes those employed more or less than 150 days on the responding farm and asks for the number of migrants (away from permanent home overnight to do farm work). Section 30 asks for the total cost of hired farm labor, including employer-paid benefits and payroll taxes, and the cost of contractor labor.

Neither the Farm Labor survey nor the Census of Agriculture collect data on the employment and earnings of workers brought to farms by nonfarm firms such as labor contractors. In California, more workers have been brought to farms by crop support service firms than were hired directly since 2009.

NAWS Research

Employment, Earnings and Migration

Perloff has done the most research using NAWS data. He emphasizes that the NAWS worker sampling strategy is based on farm employment rather than housing, as with the CPS. Some of Perloff’s most important findings include:

Other research using NAWS found that the supply of unauthorized workers in the US is inelastic, that is, unauthorized workers offer to work the same number at high and low wages, likely because they are in the US to work.


The NAWS regularly asks farm workers about their health and often includes supplemental questions to address Congressional mandates and federal agency priorities  The NAWS has been used for occupational agricultural injury surveillance and to gather information on occupational mental health, farm worker hygiene and clothes laundering practices, as well as farm worker parents’ awareness of and access to child care services, such as those provided by the Migrant and Seasonal Head Start program.  New questions introduced in October 2017 ask about access to preventive health care, participation in education and training, and access to and use of digital information sources.

A fifth of farm workers reported poor or fair health, almost all showered or bathed after every work day, and 10 to 20 percent of workers re-wore work clothing from the day before.

Hamilton found that unauthorized farm workers report fewer doctor-diagnosed illnesses and less back pain than US citizens and legal immigrants, even after controlling for age, farm work experience, and other factors. There are several hypotheses about why the unauthorized report better health, including less time to acculturate to US dietary habits and the unauthorized being less willing to report aches and pains.


Both Mexican and US agriculture rely on Mexican-born workers. Surveys conducted in rural Mexico find that the number of workers available to work in agriculture in Mexico or the US is declining by 150,000 a year.

The number of Mexicans employed in Mexican agriculture is about 6.5 million, including up to three million workers employed for wages on farms sometime during the year (some may be small farmers as well).  There are many reasons why fewer rural Mexicans want to work in agriculture, from the fact that there are fewer rural Mexicans due to declining birth rates to rising education levels and more nonfarm job opportunities in Mexico.  

The number of Mexicans willing to cross the border illegally for $5,000 or more and work in US agriculture is shrinking, but there appear to be many Mexicans willing to work legally in the US as H-2A workers for the California AEWR of $13.18 an hour in 2018.

Whither NAWS?

Industry and government agency staff see the NAWS as the most reliable source of data on crop workers, and find the data ever more useful as the number of newcomer unauthorized workers shrinks. Users want more timely release of the data, expanded coverage of livestock and H-2A workers, more employer data, and larger annual sample sizes so that characteristics of workers can be examined by commodity and size of farm rather than just geography.  If the supply of farm labor continues to shrink, the NAWS could help to identify workforce skills development gaps amidst more labor-saving technologies.

There was also discussion of how to capture all of the people who work on farms, especially as firms that may be considered wholesale traders create crews to harvest and pack produce in fields, or as trucking and logistics firms take on farm labor tasks, or as farms create “captive” FLCs to provide workers. Sorting out the complexity of who is a farm employer and who is a farm worker is a major challenge for agriculture that an expanded NAWS can help to answer.


Olivera, Vic. 1989. Trends in the Hired Farm WorkForce, 1945-87. USDA. ERS. AIB 561.


Andrews Conference Room, 2203 SSH

The NAWS was begun in 1989 to help DOL determine the supply of labor available to US crop agriculture so that, in conjunction with USDA estimates of the demand for crop workers, the proper number of free-agent RAW workers could have been legalized or admitted to prevent farm labor shortages. There were no labor shortages, and USDA stopped estimating the demand for crop workers, but DOL continued to collect data on the characteristics and earnings of crop workers, which find a mostly unauthorized farm workers that is aging and settling in one place with families that include US-born children.

This workshop will review almost 30 years of NAWS data. Session 1 will explain why DOL chose to begin a new survey rather than continue USDA's supplement to the December CPS, the current state of NAWS, and the role of the evolving NAWS in the farm labor data landscape. Session 2 will compare US and CA trends in worker characteristics, employment and earnings, and other variables. Session 3 highlights analyses using NAWS data, and Session 4 provides perspectives on the survey from organizations that represent agricultural employers and conduct farm labor research.

This workshop, and a farm labor conference April 13, 2018, is sponsored by the UCD Gifford Center and the Giannini Foundation.

8:00 AM: Breakfast available

8:30 AM: Welcome and overview, Philip Martin, Daniel Carroll, DOL, and Tim Beatty, UC-Davis

8:45 AM: The Long View: 30 years of NAWS Data (15 minutes)

10:15 AM: Break

10:30 AM: NAWS Trends: California and US (20 minutes and 15 minutes)

12:00 PM: Lunch

1:00 PM: Analyses using NAWS Data (20 minute presentations)

2:30 PM: Break

2:45 PM: Perspectives on NAWS (10-15 minute presentations)

4:30 PM: Adjourn