Women reported working, on average, less than 10 days a month during five months compared to 15 to 21 days per month during the peak season. Although a significant proportion of the individuals surveyed lived in or close to towns, roughly 85% of the jobs reported by this sample of workers were in agriculture. Females had somewhat greater packing shed employment experience than males. Surprisingly, women had higher average daily earnings than did men. Women worked more frequently on a piece rate basis , which paid more than comparable wage employment, and women were employed primarily during the peak season, when earnings were highest.Most workers lived in households with several workers. Twenty-five percent of the females surveyed and half of the males provided more than 50% of their household’s annual income. Only a third of the females who were widows or separated were their household’s major earner . Still, interviews indicated that many women had been able to separate from their husbands and/or live apart from their parents because of income obtained as a temporary fruit laborer. Although female-headed households tended to have lower incomes than male-headed households, many female heads of households spoke with satisfaction that their work allowed them to support themselves. Worker’s household characteristics influenced the number of days employed each year. Figure 3 shows that married men worked the most, especially if they had young children, approximately 275 days per year. Single males worked much less, about 170 days. Men who were separated or widowed worked an amount intermediate between these levels. The significant affect of marriage on the number of days worked suggests that marriage affected the motivation to work and that search effort was an important determinant of employment.

Women averaged significantly fewer days worked per year than men did. Some women worked more than 220 days per year, flower buckets wholesale but no female category had such a high average. Women also showed less variation in the number of days worked with respect to their household situation, at least as here categorized, and the variation shown was directly reversed from that of men. For example, married women with young children worked the least of individuals in the sample, while single women who were not living with their parents worked the most of all female categories. There is thus evidence that married women with young children had a higher reservation wage than other workers. However, women lacking income from a husband or parents worked substantially even when they had young children.Female labor force participation varied greatly by season, declining sharply from February to May, remaining low through September, and then rising steadily to February. Labor force participation was less variable for males. Daily earnings varied seasonally more in agricultural than in non-agricultural jobs, especially for jobs held by women. Women tended to earn more than men in agricultural jobs during the peak season, but less during the slack season, while the situation was reversed for non-agricultural jobs. As agricultural wages declined, a rising proportion of workers was employed in non-agricultural jobs . While female temporary workers face greater wage variation than men and vary their labor participation more, they also suffered substantially more unemployment . The female unemployment rate exceeded 50% during five months. Male unemployment was also high, but averaged only about half as much. 4.1. Labor Market Participation Equation and Expected Earnings Jarvis and Vera Toscano explored adjustment in this market to identify whether seasonal differences in labor force participation was attributable to the existence of specific ‘barriers’ to employment, differences in preferences or differences in observed worker characteristics.

Specifically, they modeled labor force participation for male and female workers by estimating a random effects probit that allowed for unobserved heterogeneity in preferences. Table 5 reports the results. For women, the estimated coefficients on the explanatory variables were generally highly statistically significant and in line with prior expectations. Few of the estimated coefficients were statistically significant for men, a result consistent with the relatively constant male labor force participation rate.13 Women participated in the labor force less than men did. Female labor force participation increased with age. Since rising education was associated with higher daily earnings, education may have altered the preference for work versus leisure. Marriage reduced labor force participation for females, perhaps due to increased household responsibilities and/or a social-cultural bias against work, but did not affect male participation. Female labor participation declined as the number of the worker’s children aged 0-5 years increased, but this effect was reduced if another adult female lived in the household, suggesting that childcare was gender specific and indicating the importance of childcare for female labor force participation. Men and women were more likely to participate during the peak season and less during the slack season as compared to the transition months of April and October through December, a result probably linked to expected earnings. Jarvis and Vera Toscano examined the sensitivity of labor force participation decisions to changes in expected earnings using a probit equation that included the same regressors plus estimated earnings . The coefficient on expected earnings was positive and significant and the other coefficients were closely similar to those obtained cols. 1 and 2. Though labor force participation for men and women responded strongly and positively to the expected wage, the female participation rate varied substantially more because females tended to have a higher reservation wage. Still, female unemployment was generally much higher than male unemployment .

Although wages varied greatly by season, Jarvis and Vera Toscano found they did not vary sufficiently to fully equate the supply and demand of labor and achieve zero unemployment. Four factors were advanced to explain this high unemployment. First, frictional unemployment was high as a result of individuals entering and/or leaving the labor force, changing jobs, and searching for employment in a spatially dispersed market where jobs were relatively short lived and search costs relatively high. Second, many or all firms may have paid an efficiency wage or piece rate to motivate workers, thereby causing the unemployment rate to remain above zero even during periods when labor demand is high. Third, the average reported wage in agriculture lay above the average reported wage in the non-agricultural sector throughout the year. Thus, waiting for an agricultural job could easily have been the better strategy for most workers even when few agricultural jobs were available. Fourth, some workers, especially females, may incorrectly report having been in the labor force and actively seeking work. Alternatively, they may have considered themselves in the labor force, but searched only within a small, local area, where there were few jobs.The average wage rose by about 50% from the slack season to the peak season, a surprisingly large variation. To understand the determinants of changes in daily earnings over the one-year period, Jarvis and Vera Toscano estimated an earnings equation where the dependent variable was the log of average daily earnings and the regressors included both supply and demand side factors. Human capital variables such as education and experience were hypothesized to influence worker productivity and earnings, while monthly dummies reflected the net influence of seasonal fluctuations in agricultural labor supply and demand. Wages were hypothesized to vary in response to the worker’s decision to seek either piece rate or wage employment, and either non-agricultural or agricultural employment. Such choices were assumed dependent on a worker’s willingness to supply effort and preference for factors such as work environment and a shorter commute time to work. Since dummy variables were used to measure the effect of working at a piece rate as opposed to a wage, flower harvest buckets the other coefficients measured the effect of the respective independent variables on the daily wage. Consistent estimates of the earnings function were obtained using the two-step estimator proposed by Vella and Verbeek. The results for both men and women are reported in Table 6. The earnings of both men and women increased with schooling, suggesting that education significantly increased labor productivity in agricultural work, although the higher return was probably partly due to the innate ability that allowed individuals to successfully complete additional schooling. Experience had a significant positive impact on female daily earnings in jobs throughout the year; the analogous coefficient was not significant for males. The square of experience had a significant negative coefficient, indicating that rising experience had a non-linear effect.

A dummy variable was also used to measure the earnings effect of working on a piece rate basis. A piece rate system was frequently used to motivate and remunerate temporary agricultural workers in the fruit sector and a substantial theoretical literature indicates that the piece rate system should increase worker’s productivity and workers’ incomes . There have been few empirical studies. The estimated coefficient on the piece rate dummy indicates that piece rate jobs in this case earned a daily premium of about 12 percent relative to wage jobs. A dummy variable was also used to measure the effect of working in the agricultural as opposed to the non-agricultural sector. Agricultural work paid substantially more, particularly for women . Men’s wages in this sample were about 18 percent higher when working in agriculture, while women’s wages were about 37 percent higher. Agricultural jobs were probably even more attractive than shown for women since there were few piece rate jobs available in non-agricultural work. As earlier noted, women’s average daily earnings were higher than men’s average daily earnings . Women working as temporary agricultural laborers were thought to earn relatively high wages in the Chilean fruit sector , and the results in Jarvis and Vera-Toscano supported that view. Nonetheless, women earned substantially less than men did in wage employment once earnings were adjusted for observed and unobserved characteristics. The estimated gender wage differential was about 25 percent. Although females had higher average daily earnings than men, women earned less than men when working for a wage, but not when working for a piece rate. Jarvis and Vera-Toscano suggested that these results indicated discrimination in the wage market. There may be less possibility of discrimination when workers are employed at piece rate since pay is directly linked to productivity. The large magnitude of the gender wage differential suggests an area for further analysis.Newman and Jarvis found that women were highly informed about many aspects of the packing shed jobs that they accepted, e.g., shed-related characteristics that affected workers’ productivity, fringe benefits, and the expected duration of the job. Women’s willingness to accept work at a specific piece rate was strongly influenced by these characteristics. Piece rates for the same tasks were found to vary by as much as 100%among different packing sheds and these differentials were well explained econometrically by the observed heterogeneity among workers and firms. For example, most processing sheds provided workers with some combination of fringe benefits that included meals, snacks, transportation to and from work, childcare, interest-free loans, and higher quality bathrooms. Supervisors and managers in different sheds treated created different quality work environments. According to the theory of equalizing wage differentials, sheds that provide more and better fringe benefits and/or a better work environment should have paid lower piece rates. This hypothesis was supported by the data. Similarly, Newman and Jarvis hypothesized that firms’ investments in technology, improved plant organization, or the ability to process grapes that were in better condition would raise worker productivity. Further, so long as workers were aware of firm-influenced productivity differences, such higher productivity should lead to lower, not higher piece rates. To the extent that firms possessed improved technology that allowed their workers to achieve higher productivity or were better organized and could provide a constant flow of good quality grapes to workers, allowing workers to process more boxes per time period, the firm should pay a lower piece rate. This followed from the assumption that each worker should earn an income consonant with her opportunity cost in equilibrium. If a firm’s characteristics allowed its workers to produce more output, ceteris paribus, worker competition for the jobs at the firm should have caused the piece rate to decline until its workers’ incomes were equal to what they would earn elsewhere. This hypothesis was also supported by the econometric results. Workers could easily ascertain the piece rates paid by different firms, but the effect of firm characteristics on a worker’s productivity should have been harder topredict.