Since such disorders are similar to those caused by intestinal parasites that workers could bring from Mexico or that could result from poor sanitation in a worker’s living environment, we used statistical techniques to isolate the effects of poor sanitation in the work environment. Even if poor sanitation leads to physical discomfort, the health problems may not have a a significant impact on an individual’s ability to work productively. If these health problems are debilitating, individuals suffering from them should be more likely to be on welfare or unemployment compensation or to have lower earnings. This hypothesis is tested in a model where the probability of being in a welfare program and earnings are a function of personal characteristics and poor health. The next section discusses the survey and the data set utilized in this study. The following section, describes the estimation techniques used. Next, three probit equations for gastrointestional disorders, repiratory problems, and muscular problems conditional on measures of demographic characteristics, living environment, and work environment are presented. Conditional on these health measures, the probability of receiving welfare or unemployment compensation is calculated. Next, the effect of these health measures on earnings is examined. The paper concludes with a discussion of the policy implications of these findings.Our data come from Mines and Kearny’s 1981 survey, “The Health of Tulare County farm workers,” sponsored by the Tulare County Department of Health. Interviewers chosen to administer the questionnaire were fluent in colloquial Spanish and either had farm work backgrounds or had extensive familiarity with farm workers.

This farm worker population largely consists of Mexican-born immigrants with varying degrees of experience with and assimilation into American society. While a large segment of the population — the long-term settled immigrants — have relatively stable living and employment conditions,vertical garden indoor system many of the more recent immigrants do not. The recent immigrants are primarily young Mexican families cr “lone Mexican males” . These workers are usually hired by crew leaders or foremen Who work for several growers, associations, or packing houses. As a result, the immigrants frequently change from job to job on a daily or weekly basis. Many workers frequently switch crew leaders as well during the season. These mercurial employment conditions are often associated with informal housing arrangements including make-shift shacks, public and private labor camps, and overcrowded apartments in small towns. Many such residences provide inadequate sanitation and food preservation facilities.Many of the survey population are foreign nationals without visas. The threat of apprehension by the Immigration and Naturalization Service induces these workers to be wary of government agencies. Thus, even when such workers are located, they are reluctant to provide comprehensive information to government officials about their employment or legal status. Moreover, most county and other government officials these immigrants meet are non-Hispanic and do not speak Spanish . As a result, more general government surveys often overlook this farm worker population, which is probably exposed to greater health risks than other groups. This study is restricted to the 367 farm workers who are the reported head of their household for whom no data are missing on key variables . Table 1 presents the means and standard deviations and formal definitions for the variables used in the analysis. The average worker is a 34 year old male, has lived in Tulare County for nearly 9 years, has access to a refrigerator and water at home, consumes nearly 8 beers a week and 5 cigarettes, has travelled to Mexico to visit his family 1.3 times in the last 5 years, has an observed family of 4 people, has a 1 in 5 chance of having been deported in the last year, is probably a harvester of grapes or citrus, and has a 30% chance that he lives in either a field or a public or private camp. Of these workers, 57% do piece work, 25% receive unemployment compensation, and 17% of their families receive welfare payments. Workers reported whether or not they exhibited various acute or chronic health problems at least once a month, and these self-reported illness are not separately confirmed. These problems are coded as binary dummy variables. As a result, ~ach of these health variables captures both serious and relatively minor problems.

The probability that a worker reports a GI problem is 17%; a respiratory problem. 26%; and a muscular problem. 50%. Although the survey only recorded the presense or absence of a job site toilet, this variable probably represents the effects of the lack of toilets, fresh drinking water, and water for washing hands. That is, the lack of toilets is believed to be highly correlated with the lack of water for drinking and washing. Other statistically significant variables also have substantial effects on the probability of having a 01 disorder. Compared to the typical worker, a female worker’s probability of having a 01 disorder is 127% higher than a male’s . Interviewers reported, however, that females were more likely to complain about both major and minor illnesses than men, so that this difference may be due to reporting difference rather than difference in health. Similar results were found in Wisconsin . Not having a refrigerator tripled the probability . An individual who lives in a public camp has a 325% higher probability of 01 disorders. A worker who lived in Mexico six months ago has a 136% higher probability of disease. The likelihood-ratio test statistic that none of the household amenities matter equals 8.46 and hence that hypothesis is rejected at the 0.05 level. Since there are only 35 households headed by a female or lacking a refrigerator and these variables have large coefficients, the health equations were re-estimated dropping those families. The resulting equations were virtually identical in terms of the effects of on the remaining variables on the probability of health problems and the asymptotic t-statistics. Based on this weak robustness test, including these two variables and the entire sample does not qualitatively alter the probit estimates. The elasticity of the probability with respect to the number of times an individual has been deported in the last year, at the sample means, is -0.16. The sign of this variable is puzzling. Other variables that are significant at the 0.10 level include the number of times one visited his or her family in Mexico in the last five years, which has the expected positive effect, and whether one is a non-Mexican foreigner, which has a positive effect. This equation correctly predicts the health of 84% of the sample, but is over-likely to predict that one does not have the disorder. This over-prediction of health is not surprising since only 17% of the sample have GI problems, and probits typically have difficulty predicting relatively rare eventsthat is, events on the tail of the distribution. Four pseudo-R2 measures and Hensher and Johnson, which range from 0.10 to 0.17, are reported in Table 2. least squares interpretation. McFadden has suggested an alternative measure of goodness of fit for an estimated dichotomous model called a prediction success index. This index compares the proportion successfully predicted for an alternative compared to that which would be predicted by chance.

This model’s prediction success index is 0.12. These results suggest that being exposed to a bacteria, parasite, or virus in lexico; lacking sanitation at work; lacking refrigeration at home; other living and working conditions; and gender are the primary factors Only two factors appear to explain respiratory problems. First. and most statistically significant , is whether the individual is a lone Mexican male worker . Nearly half of the lone Mexican male workers, who comprise 29% of the sample, reported respiratory problems, compared to 20% of the rest of the sample. The corresponding figures for GI problems are 22% versus 15%; and for muscular problems, the figures are 60% versus 47%. These lone males are the workers most likely to have recently immigrated from Mexico. They have lived in Tulare County for an average of only 3.4 years compared to 10.5 years for the rest of the sample. Controlling for other factors, a lone Mexican male has a 46.8% probability of having a respiratory problem compared to 15.4% for other males . The second factor that is statistically significant is whether the individual lives in a public camp. Compared to a worker with average characteristics,mobile vertical grow racks someone who lives in a public camp is 83% more likely to have respiratory problems.It was not a statistically significant determinant of respiratory problems, however. The pseudo-R2 measures vary between 0.11 and 0.18. The percentage of correct predictions is 73%. while McFadden’s prediction success index is0.13.As an experiment, we added to the basic specification crop and occupation variables. The coefficient on spraying is positive with an asymptotic tstatistic of 1.86, so that it is statistically significantly different from 0 at the 0.10, but not the 0.05 level. No other occupational or crop coefficient had an asymptotic t-statistic higher than 0.9. The explanatory power of that probit was about the same as the basic specification. Since this extended model produces similar results to the basic model, none of the crop and occupational variables have asymptotic t-statistics that are different from zero at even the 0.10 level in the other equations, and these variables may be endogenous, only the basic equations are reported.Respiratory problems, then, are primarily associated with lone Mexican males, but not with any particular living or working condition except, possibly, spraying and public camps. The factors that put lone Mexican males at greater risk of respiratory problems than others are unknown. Muscular Problems The results indicate that muscular problems have six statistically significant determinants. The number of deportations has an elasticity at the means of 0.05, while the number of trips to visit relatives in Mexico has an elasticity at the means of 0.08.

Presumably these variables are correlated with being a worker who changes employers frequently and who lives in rough conditions, not otherwise measured. The same explanation of frequent employment changes can be applied to the lone Mexican male variable , whether one lived in Mexico six months previously , and the public camp variable as well.Finally, males are 41% less likely to have muscular problems. This variable may reflect physiological differences, since males are more likely to have jobs involving heavier lifting. Females may do jobs that involve more bending over and may suffer from muscular problems relating to giving birth to and raising children or they may report problems more frequently than men. Again, the sanitary work conditions variable was included as a proxy for other dangers at the workplace. However, it did not have a statistically significant effect. The pseudo-R2 measures range between 0.10 and 0.17. The percentage correctly predicted is 64.6, while McFadden’s prediction success index is 0.13. Apparently workers who change jobs often suffer from more muscular problems, although that factor is only indirectly measured in our sample. Presumably they work at jobs that involve more muscular strain or live in worse conditions that are not measured explicitly by the sample questions. Again, no particular crop or activity is statistically significantly related to muscular problems. Thus, individual characteristics and home and job site conditions have statistically significant effects on three health problems. It is possible, however, that these health problems do not have a significant impact on an individual’s ability to work productively. If these health problems are debilitating, individuals suffering from them should be more likely to be partially or totally unemployed or to be less productive on the job. These effects should be reflected in higher probabilities of being on welfare or unemployment compensation or to have lower earnings.We first test the hypothesis that ill-health contributes to higher participation in welfare programs and then the earnings effects are considered. Both welfare and unemployment compensation are modeled as functions of personal characteristics and the three health problems. The sample includes a disproportionate number of employed agricultural workers, so the following results probably underestimate the full effect of ill-health for the population at large. Further, since only three health problems are studied, all ill-health effects are not captured. Indeed, severe health problems were excluded because their effects are self-evident. Since our database does not contain information about the eligibility of individuals or families for the programs, the participation rates examined in the following equations reflect the combined effects of being eligible and applying to the programs.