Skip to navigation

Skip to main content

 

Changes in the California Farm Labor Market in the 1990s -- Ed Taylor and Philip Martin

Changes in the California Farm Labor Market in the 1990s

Ed Taylor and Philip Martin, UCD

October 3, 2000


Abstract. 1

Introduction................. 1

NAWS US Data, FY97-98.................... 2

Worker Characteristics... 3

Migration................ 3

Employment, Earnings, and Assets.... 4

Employers................... 5

NAWS Regional Data, FY98 6

Farm Labor in California 7

Conclusions................... 10

Bibliography................ 10

Abstract
The National Agricultural Workers' Survey (NAWS) began as an effort to help estimate changes in the supply of farm workers to determine if additional foreign worekrs were needed; the NAWS was continued after 1993 as a mechanism to chart changes in the farm work force and to examine specific issues, including the health status of farm workers.

The NAWS highlights the importance of foreign-born and Hispanic workers among hired workers in US crop agriculture. The three major changes in the California farm work force in the 1990s illustrated by the NAWS are:

· A rising percentage of foreign-born and unauthorized workers in the 1990s, from about 10 to over 50 percent.

· The growing gap between nonfarm wages and benefits and farm wages and benefits, as illustrated by (1) the declining weeks of US farm work, (2) a shrinking gap between the average hourly farm wage and the minimum wage, and (3) fewer workers reporting access to fringe benefits. This growing gap may be speeding up exits from the farm work force, reducing the average tenure in crop work from 10 to 15 years to 5 to 10 years.

· Virtually all new entrants to the farm work force are immigrants, which means that the future farm work force is growing up outside the US, and a major policy issue will be the terms on which US farm employers gain access to foreign workers.

Introduction
The National Agricultural Workers' Survey (NAWS), which interviews workers employed on crop farms during three cycles of about 10 weeks each, has become the most cited source of data on farm worker characteristics. The purpose of this paper is to compare NAWS data on farm workers with other farm and farm labor data to understand what happened to the California farm labor market and farm workers in the 1990s.

NAWS interviews workers in 85 counties across the US with the permission of their employers, collecting data on the demographic characteristics of farm workers, their current and previous farm and nonfarm jobs, and on their income and assets. The NAWS does not generate the number and distribution of farm workers, but the profile compiled is believed to describe about 1.8 million or 72 percent of US hired farm workers. Comparisons of NAWS data across regions can highlight e.g. the high percentage of unauthorized workers in the southeastern states.[1]

The purpose of the NAWS between 1989 and 1993 was to estimate entries into and exits from the work force in Seasonal Agricultural Services, defined to include most crop farms--NAWS data provided one element for a complex formula used to determine whether Replenishment Agricultural Workers (RAWs) were needed (none were during the four-year RAW program). The NAWS was continued after 1993 to monitor farm worker characteristics and to obtain health and other information on farm workers and their families.

The NAWS interviewed about 20,000 workers between 1989 and 1998, and continues to interview about 2,000 workers a year. During this period, most of the workers interviewed have been young male immigrants from Mexico employed in US fruit and nut, vegetable and melon, and horticultural specialty (FVH) commodities. During the nine years that workers were interviewed, they earned an average of $0.50 to $1 an hour over the federal minimum wage for 25 to 26 weeks or 1000 hours of farm work a year.[2] The major change in the NAWS in the 1990s was the declining share of workers legalized in 1987-88, Special Agricultural Workers, and their replacement with the newly arrived unauthorized workers.

NAWS US Data, FY97-98
The most recent US Department of Labor report based on the National Agricultural Workers' Survey (NAWS) included 4,199 workers who were interviewed in FY97 and FY98, i.e., between October 1, 1996 and September 30, 1998. The six-chapter DOL report, the eighth, follows the outline of previous NAWS reports: sections one through three cover farm worker characteristics; sections four and five cover earnings and jobs; and section six focuses on worker and family participation in US welfare programs.

Worker Characteristics
The 1996-98 NAWS found that 81 percent of US crop workers were born abroad, including 77 percent born in Mexico. About half of the US-born crop workers were Hispanics and half were non-Hispanic whites. This implies that, of the estimated 1.8 million persons who worked on US crop farms, about 1.4 million were born in Mexico and 340,000 were born in the US.

Workers were asked a series of questions about their place of birth and legal status— authorization to work in the US was inferred from their answers. About 48 percent (864,000) of the crop workers interviewed in 1997-98 were authorized to work in the US; 52 percent (936,000) were not. The legal workers were divided almost equally between US citizens (22 percent) and legal immigrants (24 percent).

In 1997-98, about 1.1 million unauthorized foreigners were legalized under the Special Agricultural Worker (SAW) program; they were supposed to have done at least 90 days of farm work in 1985-86. Studies based on a variety of farm labor data concluded that at least half of the foreigners who received SAW status did not do this requisite farm work, a conclusion also suggested by the NAWS. In 1989-90, the NAWS found that about 31 percent of crop workers were SAWs, meaning that about 558,000 SAWs were employed in agriculture if the NAWS covered 1.8 million crop workers. By 1997-98, the number of SAWs was down to 16 percent, or 285,000. The movement of SAWs out of agriculture almost exactly matches the increase in unauthorized farm workers.

Farm workers are unlike other US workers. In 1998, about 54 percent of US workers were male, and 39 percent were under 35 years of age. Some 80 percent of crop workers were men, and 67 percent were under 35. Despite their youth, about 52 percent of crop workers were married, and half of them had children. Almost half of the married farm workers left their families in Mexico while doing US farm work; other farm workers left their families in South Texas, for example, while they migrated north to Michigan to do farm work. About 38 percent of farm worker parents, who did not have their children with them while they did farm work, had three or more children.

All US Workers and Farm Workers (%)


US Workers
Farm Workers

Male
54
80

Under 35
39
67


About 84 percent of farm workers speak Spanish and 12 percent speak English; 85 percent of farm workers, compared to 11 percent of all US workers, have not completed high school. The median years of schooling for farm workers was six and most farm workers completed their education in Mexico. Many farm workers are eligible to participate in US adult education courses, and 22 percent did, usually attending GED or English classes, but less than five percent of foreign-born farm workers reported that they could read or speak English well.

Migration
The 4,199 workers interviewed between 1996 and 1998 held a total 7,697 farm jobs. About 60 percent had only one job, and only seven percent had four or more jobs, suggesting that relatively few workers fit the stereotype of a migrant who follows the ripening crops from south to north, working on many farms.

In the nonfarm labor market, 143.2 million workers had at least one job in 1997, and 15.6 million experienced unemployment— annual average employment was 129.6 million, and annual average unemployment was 6.7 million. An average eight million workers held two or more jobs simultaneously in 1998—about 4.5 million had one full-time and one part-time job. For more information: http://www.bls.gov/news.release/work.nws.htm

NAWS defined a migrant as a worker who moved 75 miles or more from his usual residence to find a US farm job; an overnight stay away from home was not required. The largest group of crop workers interviewed, 44 percent (792,000 if the NAWS applies to 1.8 million crop workers) were not migrants. Another 39 percent (702,000) were shuttle migrants— their usual homes were in generally in Mexico, and they traveled more than 75 miles from their homes to their US farm jobs.[3] Only 17 percent (306,000) of the crop workers interviewed were stereotypical follow-the-crop migrants who have one US farm job and then travel at least 75 miles for another US farm job. These migration data help to explain why programs established to assist follow-the-crop farm workers and their families often deal with workers who, for example, arrive in a migrant camp from Mexico and remain for the entire six months that the camp is open.

Employment, Earnings, and Assets
Workers averaged $5.93 an hour for 38 hours a week in FY97 and FY98, for weekly earnings of $225— average weekly earnings for all private sector workers were $442 in 1998 (the federal minimum wage rose from $4.25 to $4.75 on October 1, 1996 and to $5.15 on September 1, 1997). The quarterly USDA publication, Farm Labor, reports higher average hourly earnings and hours worked— an average $6.98 an hour for field and livestock workers in 1998, and 40 hours a week. About 77 percent of the 7,697 farm jobs paid hourly wages; 20 percent paid piece rate wages. Piece rates are most common in harvesting jobs.

Farm workers interviewed in the NAWS averaged 24.4 weeks of farm work, for farm earnings of $5,500 in 1996-98. They also averaged 4.6 weeks of nonfarm work, for nonfarm earnings of about $1,000. Farm workers averaged 10 weeks unemployed in the US and 12 weeks abroad. Weeks of farm and nonfarm work in the US declined in the 1990s, while weeks abroad increased, perhaps reflecting the rising share of unauthorized farm workers who are sojourners.

The NAWS recorded the employment status of workers on a month-by-month basis. The percentage of workers doing farm work peaked in the summer months at 55 to 60 percent; the unemployment rate for workers available for work was at least 15 percent. During the winter months, the percentage of workers doing farm work was between 35 and 40 percent, and the percentage of workers unemployed was 20 to 23 percent. About one-third of farm workers reported that they were outside the US in the winter months.

Farm work is often described as a job, not a career. The crop workers interviewed between 1996 and 1998 had an average eight years of US farm work experience, an average that may be misleading, since the half of the workers who were authorized had an average of 13 years of US farm work experience, while the half who were unauthorized had an average four years of US farm work experience. That is, workers were concentrated at the two extremes of the experience spectrum.

About half of the crop workers interviewed between 1996 and 1998 said that they intended to remain farm workers as long as possible; the other half intended to exit the farm work force within five years, which would imply that half of the crop workers will have to be replaced by 2003. About 60 percent of farm workers said they had relatives or friends with nonfarm US jobs, and 35 percent thought they could get a nonfarm US job within one month.

Experienced farm workers may get out of agriculture as they learn about nonfarm alternatives. They may also find farm work less attractive— the average weeks of farm work for US-born farm workers fell from 24 in 1990-92 to 22 in 1996-98, while the average weeks of unemployment rose from 15 to 19. Foreign-born farm workers saw their weeks of US farm work fall as well, from 28 to 25, but they responded by spending more time outside the US— average weeks outside the US rose from 10 to 15.

NAWS obtains income data by range. Half of farm workers had 1997 incomes of less than $7,500, and half of farm worker families had incomes of less than $10,000. Most individuals and families were below the poverty line— $8,350 for an individual in 1997, and $12,800 for a family of three. About 20 percent of farm workers said that they or someone in their family received UI benefits within the past two years. Since 50 percent of farm workers are unauthorized, and 14 percent work year-round, this means that over half of the 36 percent who are eligible for UI benefits received them.

About 17 percent of farm workers received benefits through means-tested programs. Unauthorized workers are generally not eligible for such programs, although their US-born and US-citizen children may be. The NAWS reports that about one-third of the legally authorized farm workers received means-tested benefits-- the three most common benefits were Medicaid (Medi-Cal), Food Stamps, and the Women, Infants and Children (WIC) program. Far fewer farm workers received cash payments under AFDC/TANF or public housing assistance.

The increasing presence of unauthorized workers also shows up in home buying and assets. The percentage of farm workers who owned or were buying a US home dropped— in 1994-95, about one-third owned or were buying, compared to 14 percent in 1997-98. The percentage who owned a vehicle in the US dropped from 49 to 44 percent.

Employers
The NAWS selects workers to be interviewed on the basis of a special list of farm employers developed by Aguirre. About 80 percent of the workers interviewed were employed directly by the growers where they were working. About 33 percent of workers were employed in fruits and nuts, 28 percent in vegetables, and 14 percent in horticultural specialties— these FVH crops employed 75 percent of the crop workers.

Most farm employers must provide some benefits to workers— Social Security, unemployment insurance and worker's compensation— and some provide additional benefits, including pensions, health insurance and vacation pay. For example, about 60 percent of the workers interviewed said that they could not get UI benefits, which may be explained by the fact that, in many states, only workers employed on the largest farms must be covered by UI, and unauthorized workers are not eligible for benefits, even if their employers pay UI taxes on their wages. Workers compensation pays for medical costs associated with work-place injuries and provides payments to workers who cannot work as a result of work-place injuries. About half of the states do not require farmers to provide workers compensation coverage for farm workers, and two-thirds of farm workers interviewed in the NAWS said that they were not covered.

The most common voluntary benefit provided to farm workers was housing-- about 20 percent of crop workers received free housing from their employers. About five percent of crop workers received health insurance for off-the-job injuries to themselves or their families and 10 percent received vacation pay.

NAWS Regional Data, FY98
There are many data elements collected from workers, but some of the most cited involve hourly wages and weeks worked in the worker’s current job. In FY98, the 2099 workers earned an average $6.26 an hour in their current job (unweighted), and hourly wages ranged from a low of $5.51 in the east to $6.88 in the northwest. The standard deviation around average wages was quite high--$1.63 an hour for the US, and $2.27 an hour in the northwest.

Some of the average data in the NAWS may be influenced by extreme observations. For example, workers are asked how long they were employed in their current job, and their responses range from just started to having been employed in the same job for 20 years. The average length of employment in the current job was 11 weeks, with a standard deviation of 30 weeks. To determine the impacts of extreme observations, we removed the 29 employees who were in their current job for more than 53 weeks--the average weeks in the current job dropped to 9 weeks, and the standard deviation to 11 weeks.

Average Hourly Wages and Weeks Worked in Current Job: FY98


Region

Wages
Standard Dev
Weeks
Stan Dev

1
East
5.51
1.09
7.9
10.4

2
Southeast
6.12
1.63
12.5
32.6

3
Midwest
6.69
1.76
12.0
18.0

4
Southwest
6.03
1.65
12.3
13.8

5
Northwest
6.88
2.27
10.8
67.8

6
California
6.43
1.29
9.8
9.4


US
6.26
1.63
10.7
29.6

Source: NAWS, 2099 Interviews in FY98


Average Hourly Wages and Weeks Worked in Current Job: FY98

Difference--2099 vs 2070 workers

Region

Wages
Standard Dev
Weeks
Stan Dev
Weeks
Stan Dev

1
East
5.51
1.10
6.7
6.7
15%
36%

2
Southeast
6.11
1.62
8.2
15.1
34%
54%

3
Midwest
6.69
1.77
11.0
14.4
8%
20%

4
Southwest
6.04
1.66
11.7
12.5
6%
10%

5
Northwest
6.89
2.28
6.6
7.6
39%
89%

6
California
6.43
1.29
9.8
9.4
0%
0%


US
6.26
1.63
9.0
11.4
16%
61%

Source: NAWS, 2070 Interviews in FY98

Excludes workers employed more than 53 weeks in current job

Farm Labor in California
California agriculture is different from the rest of U.S. agriculture in many ways, but three stand out: size, labor intensity, and seasonality:

· Size. California has been the number one farm state since 1950, as measured by farm sales, largely because of the production of so-called fruit, vegetable, and horticultural specialty commodities. In US agriculture as a whole and in most states, the $200 billion in annual farm sales is about 50 percent livestock and 50 percent crops, and the major crops grown are low value per acre and mechanized field crops such as wheat and corn. In CA, by contrast, crop sales were three-fourths of the $27 billion annual farm sales in 1997, and two-thirds of the crop sales were high value fruit and nut, vegetable and melon, and horticultural specialty (FVH) commodities, such as flowers and mushrooms.

· The second California uniqueness is labor intensity. Commodities that are high in value also tend to be labor intensive. Labor intensive is not an adjective that is usually defined, but labor costs are typically 20 to 40 percent of the cost of producing lettuce or peaches, versus 10 percent in manufacturing and 70 to 80 percent in services. Labor is thus the largest single “controllable” expense for many FVH growers, in the sense that farm employers can negotiate whether they will pay $0.20 or $0.21 cents to have a 25-pound tray of raisin grapes harvested, but they cannot usually negotiate the price of the fertilizer used on the grapes.

· The third uniqueness is seasonality. Peak employment on most farms growing FVH commodities is 5 to 10 times larger than trough employment. Matching seasonal workers with seasonal jobs in a manner that minimizes uncertainty for farm employers and unemployment for workers has always been difficult, and raises the question of who should pay for the cost of seasonality. In practice, workers generally bear most of the costs of seasonality—they wait for work to begin, and then return to their country of origin, collect UI benefits if they are eligible (farmers and other employers pay for UI benefits), find nonfarm jobs or nonfarm sources of income in the US, or drop out of the labor force when there is no seasonal work available.

One measure of seasonality is the so called peak-trough ratio, employment in the peak month of the year divided by employment in the trough month. California’s Employment Development Department stopped publishing monthly estimates of the employment of seasonal and regular workers in 1994; the 1994 report showed that the employment of hired farm workers on the state’s farms ranged from a low of 160,000 to a high of 349,000, a peak-trough ratio of 2.2.

A different EDD data series shows more farm employment, but the same seasonality. If all workers reported to state unemployment insurance authorities by farm employers are considered “farm workers,” (they include the managers of corporate farms as well as mechanics and clerks and farm workers), then farm worker employment trended upward in the 1990s, with considerable seasonality. Farm worker employment as measured by the UI data ranged from a low of about 300,000 in January-February to a high of about 475,000 in August-September, a peak-trough ratio of about 1.6; that is 475,000 divided by 300,000. Peak-trough ratios are larger for particular counties and commodities.

The smallest peak-trough ratios, indicating the least seasonality, are found in the most urban counties--San Francisco, San Mateo, San Diego, Los Angeles counties, which have peak-trough ratios just a little bit over 1, meaning that the nurseries and other agricultural employers in those counties tend to hire about the same number of workers each month of the year.

Seasonality is closely linked with migration. Given a peak-trough ratio of 2 to 1, there are 200 workers employed in the peak month for every 100 employed in the trough month. These extra employees in the past were often people who are not usually in the labor force, such as housewives and students, or workers who move from one area of the US to another, so-called follow the crop migrants. Today, many seasonal workers move into the US to handle peak labor needs, so called shuttle migrants. Follow-the-crop migration has decreased for three major reasons:

· the harvesting of some crops has been stretched out for marketing and processing reasons,

· temporary housing for migrants is scarce and

· the availability of unemployment insurance and other farm worker services programs makes migration less necessary.

There are many ways to think about a labor market in which 800,000 to 900,000 individuals find jobs with 35,000 farm employers in the course of the year, as occurs in California agriculture. One way is to focus on 3 words that begin with C: concentration, contractors, and conflict:

· Concentration refers to the fact that most farm workers are hired by a relatively small number of large farms.

· Contractors refers to the presence in many farm labor markets of “intermediaries” who match seasonal workers with jobs.

· Conflict refers to the fact that immigrant workers dis-satisfied with farm wages and working conditions tend to exit agriculture for the nonfarm labor market, often within 10 years of arriving in the US.

The Census of Agriculture reported that 36,450 California farms hired 549,265 workers in 1997—a worker would be counted twice if he was hired by two farmers. However, workers brought to farms by FLCs and other agricultural service firms are not recorded in the COA, which records only the amount of money paid to labor contractors and agricultural service firms. The 9,500 farm employers who hired 10 or more workers accounted for 85 percent of the workers hired in California, i.e., 26 percent of farm employers accounted for 85 percent of workers hired, and more detailed data would likely show that the largest 1,000 farm employers accounted for over half of farm worker employment.

Contractors or intermediaries are the second C of the California farm labor market; they evolved from one of the crew to independent businesses that seek to profit from the difference between what a farmer pays to have a job done and what the worker gets. In 2000, there are about 1200 contractors registered with the state of California, but they employed about one-third of the farm workers reported to Unemployment Insurance authorities. Most farm worker advocates want to eliminate “merchants of labor,” or at least make the operators of farms to which FLCs take workers jointly liable for labor and other violations. In addition to FLCs, many other agricultural service firms bring workers to farms to provide crop and harvesting services.

There is an inevitable Conflict between employers and workers over the “fair” wage and the level of effort a worker must expend to keep his job. The classic responses to work place conflict are worker exits, government regulation, and collective bargaining. The most common responses in agriculture have been for dissatisfied workers to exit for the farm labor market, and for government regulation of the farm labor market to increase.

The NAWS provides important insights into how the farm labor market deals with three essential tasks: recruitment, remuneration or motivation, and retention. Recruitment in the farm labor market is typically handled informally by bilingual intermediaries—either independent FLCs or foremen/supervisors employed by farm operators—who recruit seasonal workers by word of mouth, usually by asking current workers to recruit friends and relatives. In the nonfarm labor market, by contrast, most workers are recruited formally, with employer ads, written applications, interviews and tests, and employers checking out references.

Remuneration or motivation on the job is encouraged in several ways. For tasks in which (1) quality is less important than speed, and (2) there is no easy way to monitor the pace of individual workers, such as picking lemons or peaches in trees so that a supervisor cannot easily see all workers at once, piece rate wages are common, e.g. $10 to pick a bin of peaches. Piece rate workers must work fast enough to earn at least the minimum wage, $5.75 per hour in CA in 1999, meaning they must fill the bin with peaches in less than two hours, to keep their jobs. Piece rate wages keep grower costs constant even if the workers have differing productivity and minimize the need for close supervision of the speed of work, but require monitoring of the quality of work done.

Hourly wages, on the other hand, can encourage higher quality but slower work. Crews of workers paid hourly wages often include a foreman or pusher who keeps e.g. all workers hoeing weeds to at the same pace. In some cases, employers control the speed of work by having workers follow a conveyor belt harvesting machine—hourly wages can be paid to melon or broccoli or lettuce crews in such situations because the employer controls the speed of the machine. Some employers mix both wage systems, guaranteeing an hourly wage, and offering a piece rate as an incentive, as in harvesting strawberries and table grapes.

Retention/promotion is not usually an issue for most farm employers and farm workers. The job ladder in CA ag is very short, resembling a pyramid with a wide base and short sides, so there are relatively few promotion opportunities. This means that most seasonal farm workers do not expect to get promoted on the farms where they work, since there are relatively few slots for year-round equipment operators or irrigators, and few farmers have personnel systems that identify the best field workers to fill such jobs. Instead, both the seasonal and the better jobs tend to be filled in a word-of-mouth fashion. Most farmers are more interested in ensuring that enough seasonal workers will be available for all of agriculture than in ensuring that a particular worker returns to the farm next year, i.e., there is a tendency to think in terms of a collective response to assure an adequate supply of labor for all of agriculture then to think of developing a seasonal work force for a particular farm.

Conclusions
The NAWS began as an effort to help estimate changes in the supply of farm workers; it has continued as a mechanism to chart changes in the farm work force and to examine specific issues, including the health status of farm workers. The three major changes in the crop labor force in the 1990s, as determined by the NAWS, include:

· A rising percentage of foreign-born and unauthorized workers in the 1990s

· Declining weeks of US farm work and a shrinking premium over the minimum wage

· Relatively few follow-the-crop migrants who travel significant distances within the US to hold two or more farm jobs.

The NAWS highlights the importance of foreign-born and Hispanic workers among hired workers in US crop agriculture. The interviews seem to be strongly oriented toward FVH agriculture, and highlight the speed with which foreign-born workers spread in this sector.

Bibliography
US Department of Labor. National Agricultural Workers Survey.

http://www.dol.gov/dol/asp/public/programs/agworker/naws.htm

Appendix
When using NAWS data for descriptive or statistical analysis, it is important to be aware of the limitations of this data set. While the NAWS data are useful for many purposes, they may not be very well suited for others. There is a danger of obtaining unreliable estimates if the limitations of NAWS—like any data set—are not taken into account. The main questions we have about the quality of the NAWS data have to do with sampling. There also is suppression of information that the USDL deemed might permit farm worker identification (e.g., county); however, in light of the sampling concerns, it is not clear how useful the omitted information would be.

Farmworkers are notoriously difficult to survey. Most are seasonal, many are migrants and some leave the US during the off season. A challenge in the NAWS has been to survey a random sample of farm workers without having a universe or total population from which to reliably draw a random sample. The goal of the NAWS sampling process was to select a random sample of farmworkers that would be nationally representative. The NAWS data are almost certainly representative of some farm worker population—the question in, which one? We have looked closely at the NAWS data for California, one of the 6 NAWS regions (aggregated from the original 12 NAWS regions) and the one we know best. There, the sample seems to over-represent some groups (e.g., workers in fruit and nut crops) while under-representing others (e.g., supervisors). This raises questions about when the use of NAWS data is appropriate for descriptive or statistical analysis and when it is not.

Given the inherent difficulty of sampling farm workers, the NAWS makes a serious effort to design its sample scientifically. The key sample design question addressed in the NAWS materials is how to select employers to target in the survey and how many workers to survey in each NAWS region and crop cycle (winter, summer, or fall), in order to obtain a representative sample of the farm worker universe. Some government data sources were available to assist in this. USDA was able to provide quarterly estimates of hired and contract employment by quarter for each of the 12 NAWS regions, drawing from the Quarterly Agricultural Labor Survey (QALS). Those data make it possible to allocate NAWS interviews across regions and crop cycles, i.e., in proportion to the region/cycle shares of total US annual farm employment.

Once NAWS determined how many workers to interview in each region and crop season, the next step is to decide where in each region to survey workers. Within the 12 NAWS regions, there are 47 crop-reporting districts (CRDs)—that is, an average of about 12 CRDs per region. The CRDs are aggregates of counties that USDA determined have similar agricultural characteristics and are roughly similar in the size of their farm labor expense. For the CRDs and the counties within them, the Census of Agriculture (COA) provides estimates of total seasonal farm labor expenses. Within the 12 NAWS regions, CRDs were selected using probabilities proportional to the size of their total seasonal farm labor payrolls. The COA labor expenses were adjusted seasonally using estimates, which identify the percentage of labor expenses that fall into each of the NAWS cycles, fall, winter/spring, and summer. Interviews were allocated to the selected CRDs proportionally to the amount of seasonal farm expenses within the CRD.

Having determined how many interviews to conduct within each CRD, the NAWS then had to decide which county or counties within CRDs to target with their interviews. Within each CRD, counties were selected randomly without replacement using probabilities proportional to the county’s farm labor expenses. In almost all cases, interview quotas were filled within the first county selected in this way. This means that NAWS data come mostly from the single county having the largest farm labor expenses within each CRD. For example, in California, there was a very high a-priori likelihood that Fresno County, the state’s largest agricultural county, would be selected for the NAWS survey.

Finally, the NAWS had to determine how many and which agricultural employers to target within each county. This requires employer lists for the NAWS counties. The NAWS appears to have invested considerable effort into developing lists of farm labor employers, using information from the Bureau of Labor Statistics (BLS), Agricultural Soil and Conservation Service, and other administrators’ data sources. The NAWS documentation notes that “these lists were reviewed, updated and supplemented using information from county extension agents, local employment agencies, grower organizations, and farmworker service providers.”

Employers were selected from these lists with probabilities proportional to the size of their seasonal farm workforce. This means that, within the county with the highest farm labor expenses in each CRD (e.g., Fresno County in California), the sampling procedure favors the largest farm employers. In Fresno County, those would disproportionately be table grape producers.

The final sampling decision, which is not discussed in the NAWS documentation, involves which workers in the Fresno grape vineyard to interview. Picturing ourselves in a vineyard with 300 workers cutting grapes and laying them out on paper trays to dry in the sun, it is not obvious how to draw a random sample. It would be useful to have additional information about how this was done. What are the odds of the busiest workers being interviewed? Of crew leaders and supervisors being chosen?

Some basic characteristics of the California NAWS sample raise some concerns about which population of farm workers NAWS represents. The total California sample size is 3,768 for all 6 years of NAWS and 592 for 1988. Tables 1a-b and 2a-b summarize the distribution of the sample across crops and tasks. In 1998 (Table 1a), 4 out of 5 California farm workers in the NAWS were in fruits/nuts. Only 14.5 percent were in vegetables, even though in terms of wages paid vegetables would rank higher than fruits/nuts in California.

Table 1a. Distribution of California NAWS Sample, by Crop, 1998


Crop
Number of Cases
Percentage of All California Cases

Field
11
1.86%

Fruits/Nuts
472
79.73%

Hort
8
1.35%

Vegs
86
14.53%

Misc/Mult
15
2.53%

Grand Total
592
100.00%

Aggregating across years, the crop distribution of California farmworkers in NAWS looks a little bit more like what we would expect, but it is still skewed very heavily towards fruits/nuts:

Table 1b. Distribution of California NAWS Sample, by Crop, All Years

Crop
Number of Cases
Percentage of All California Cases

Field
167
4.43%

Fruits/Nuts
2559
67.91%

Hort
101
2.68%

Vegs
842
22.35%

Misc/Mult
94
2.49%

Grand Total
3768
100.00%


For some categories of workers, sample sizes are very small. The 1998 NAWS California sample includes only one supervisor. Given the job-skill pyramid in California agriculture, it is not surprising that there would not be many supervisors in a truly random sample. However, this worker group comprises only 0.17 percent of the NAWS California sample, which is too low. Clearly, the NAWS data are inappropriate to examine such issues as the mobility of workers into supervisory jobs. It is not clear why so few supervisors were not included in the California NAWS sample—or why a stratified approach was not used to target under-represented groups of workers.

Table 2a. Distribution of California NAWS Sample, by Task, 1998


Task
Number of Cases
Percentage of All California Cases

Pre-Harvest
120
20.27%

Harvest
133
22.47%

Post-Harvest
50
8.45%

Semi-Skilled
278
46.96%

Supervisor
1
0.17%

Other
10
1.69%

Grand Total
592
100.00%

The number of supervisors is higher than one per year in the full, 6-year sample, but it is still very low—an average of between 4 and 5 individuals per year.

Table 2b. Distribution of California NAWS Sample, by Task, All Years


Task
Number of Cases
Percentage of All California Cases

Pre-Harvest
562
14.92%


Harvest
1235
32.78%


Post-Harvest
320
8.49%


Semi-Skilled
1508
40.02%


Supervisor
27
0.72%


Other
112
2.97%


Grand Total
3768
100.00%


These sampling issues need further investigation. How might the skew towards fruit crops and non-supervisory workers influence average farmworker characteristics in the NAWS sample? How can one generalize from the NAWS sample to the universe of US farmworkers, in light of these concerns? Most importantly for the future, how can one improve the NAWS sample design to include a more representative mixture of crops, tasks, and perhaps other characteristics? Answering these questions is likely to be difficult because of the lack of a well-defined universe from which to randomly draw farmworker samples and calculate weights that can be used for stratified sampling to target sub-populations of farm workers that may have been under-represented in the NAWS.


--------------------------------------------------------------------------------

[1] The NAWS share of the hired farm work force is derived as follows. During the 1970s and 1980s, USDA had questions about hired farm workers attached to the December Current Population Survey, so that migrant workers could be interviewed in their usual homes, and the CPS data were used to estimate that there were 2.5 million hired workers employed on US farms during a typical year for wages. In 1997, the Census of Agriculture reported the labor expenditures of crop and livestock farmers; these labor expenditures were divided by the average hourly earnings of field and livestock workers reported in the NASS Farm Labor survey for 1997 to estimate that 72 percent of the hours worked adjusted by hourly earnings were worked on crop farms, and 72 percent of 2.5 million is 1.8 million.

[2] For example, in 1989-90, when the federal minimum wage went from $3.35 in 1989 to $3.80 in 1990, NAWS workers averaged $4.85 an hour in their current job. In 1997-98, when the federal minimum wage was $5.15 after September 1, 1997, workers averaged $5.93 an hour.

[3] NAWS defined shuttle migrants as persons who spent at least 28 days a year outside the US, so that a worker who was interviewed soon after arrival in the US could be considered a shuttle migrant even if the move to the US was permanent. The home base for 88 percent of shuttle migrants and 43 percent of the follow-the-crop migrants is Mexico, usually rural areas.