estimating the work life expectancy of undocumented ... · pdf fileestimating the work life...

25
PRELIMINARY DRAFT Please do not quote 1 Estimating the work life expectancy of undocumented Mexican migrant workers Dwight Steward Jean Elliott Amy Raub Econ One Research, Inc. 106 E 6 th St Ste 230 Austin, Texas 78701 [email protected] DRAFT Preliminary and Incomplete – Please do not quote Comments and suggestions welcome

Upload: vantram

Post on 11-Mar-2018

221 views

Category:

Documents


1 download

TRANSCRIPT

PRELIMINARY DRAFT Please do not quote 1

Estimating the work life expectancy of undocumented Mexican migrant workers

Dwight Steward Jean Elliott Amy Raub Econ One Research, Inc. 106 E 6th St Ste 230 Austin, Texas 78701 [email protected] DRAFT

Preliminary and Incomplete – Please do not quote Comments and suggestions welcome

PRELIMINARY DRAFT Please do not quote 2

Abstract

This paper introduces new U.S. work life tables designed specifically for undocumented Mexican migrants based on data collected by the Mexican Migration Project at the University of Pennsylvania. The constant growth of the number of undocumented residents in the U.S. accompanies an increasing number of personal injury and wrongful death lawsuits involving undocumented migrant workers.

This paper examines the inaccuracy of the two current choices of how to calculate Mexican migrant worker work lives. The first option is to assume that the migrant worker would have returned to his country of origin right away and should therefore have no damages associated with his US earnings. This assumption vastly understates damage estimates for migrant workers likely to remain in the U.S. for continued periods of time.

The second option is to rely on existing work life tables of U.S. citizens which assumes the migrant worker will continue to earn money in the U.S. for the remainder of their work life. Similar to other research on the topic, our analysis of migration data shows that the typical undocumented Mexican worker will have a shorter work life than that of native born U.S. worker. This assumption overstates personal injury or wrongful death damage estimates for migrants likely to return to Mexico after working only a short period of time in the U.S.

After illustrating the need for a more accurate third option, this paper presents the results and methodology for calculating the remaining U.S. work life of migrant workers based on previous U.S. work experience found in the Mexican Migration Project data. This new methodology and data enable work life expectancies to be calculated for migrant workers based on how long they are likely to earn wages in the U.S. over their lifetime. The data consists of a yearly survey of households in the United States and Mexico, in which demographic and socioeconomic information is collected. By incorporating a number of new factors such as legal status, work locations in the U.S., home locations in Mexico the length of time and number of trips to the US, a more accurate work life, and therefore, more accurate damages, can be calculated for migrant workers in personal injury and wrongful death cases in the U.S. One of our additional findings is that a Cox proportional hazards model of the probability of returning to Mexico shows that for undocumented Mexican workers the probability of leaving the U.S. labor market and returning to Mexico is highly correlated the individual’s demographics and stock of human and cultural capital.

PRELIMINARY DRAFT Please do not quote 3

Introduction

Similar to native born U.S. workers, many undocumented Mexican migrant workers will come into contact with the U.S. legal system at some point in their working life. As with any group of workers, a portion of the over 6 million undocumented Mexican individuals in the U.S. will be injured or killed while working and living in the U.S. In many instances, the undocumented Mexican migrant worker or the migrant worker’s family will have legal standing to sue, in U.S. courts, the individuals and companies that allegedly caused the harm.

For instance, in 2002, the families of 11 immigrant workers who died crossing into Arizona from Mexico filed a $41 million claim against the Department of Interior and the U.S. Fish and Wildlife Service, alleging the government's refusal to put water out in the desert contributed to the migrants' deaths. In another high profile human trafficking case in Texas involving the suffocation death of 19 persons entering the U.S. from Mexico. The families of the deceased filed a multi-million dollar wrongful death lawsuit against the trucking company after the driver of the truck was convicted in a criminal case. In some instances, migrant workers have become parties in employment and labor lawsuits involving Fair Labor Standard Act (FLSA) violation allegations and Federal employment discrimination claims. Much more common, however, are lawsuits in which undocumented workers have sued employers and contractors for lost past and future wages in cases involving injuries sustained while working.

In these cases, as with U.S citizens, if it is determined the defendant is liable for the injury or death, the undocumented worker or their family can be awarded economic damages to compensate for the loss of earnings, earnings capacity, and other services the injured or deceased person would have provided to the family had the incident not occurred. To date there have been a number of notable economic damage awards in cases involving undocumented Mexican workers. For instance, in contrast to the Supreme Court ruling in Hoffman Plastic v. NLRB, 535 US 137 that held the award of back pay ran counter to immigration national policy, the injured plaintiff in Celi v. 42nd Street Development Project, Inc, was awarded about $1 million dollars in both back pay and future wage losses by a New York state court. In EEOC v. Phase 2 Co., 10 Mexican workers, all of whom had an unknown immigration status, settled a hostile environment employment discrimination case with the employer for approximately $600,000. Similar cases involving the economic damages of undocumented Mexican workers are pending in state and federal courts across the country.

At present, however, there has been little economic research on the most appropriate manner to measure economic damages for undocumented Mexican workers working in the U.S. In particular, there are a number of unique factors that affect the work life expectancy of Mexican migrant workers that must be incorporated into the calculation of economic damages. Generally speaking, using work life expectancy tables that are primarily based on the work experiences of native born and authorized U.S. workers will tend to over state the economic damages in cases involving undocumented Mexican workers. On the other hand, assuming the Mexican migrant worker would have returned to his country of origin the day the injury or death occurred is unrealistic and

PRELIMINARY DRAFT Please do not quote 4

shortchanges any rightful claim to damages by the worker or the worker’s family. Because the flaw in assuming the undocumented worker would have returned to Mexico immediately rather than continue to work in the U.S. is obvious, being that it yields no damages for the injured party, I will focus for a moment on the flaw of utilizing U.S. work life tables.

For instance, unlike native born U.S. workers, undocumented workers who come to the U.S. to work may, for any number of reasons, decide to leave the U.S. labor market and return to Mexico. In fact, a significant body of research suggests it is not the goal of the typical undocumented Mexican worker to permanently work and live in the U.S. Instead, these studies, as well as empirical data on the average length of stay and total time in the U.S. labor force for an undocumented Mexican worker, strongly suggest the typical migrant worker from Mexico enters the U.S. with a goal of prospering from the relatively greater labor market opportunities in the U.S., remit financial resources to family members in Mexico, and then returning home. Standard measures of work life expectancy used in the calculation of economic damages of U.S. workers are generally inappropriate for the calculation of economic damages in cases involving Mexican migrants.

In this paper we use data from the Mexican Migration Project (MMP) to study the U.S. immigration experiences and analyze the work life expectancy of undocumented Mexican workers. Our key results are as follows. First, like other research on the topic, our research shows that the typical undocumented Mexican worker will have a shorter work life than that of native born U.S. worker. For instance, the average undocumented Mexican worker during his or her most recent migration to the U.S worked in the U.S. labor market for approximately 3.03 years before returning back to Mexico. In terms of total U.S. labor market experience, the typical undocumented Mexican migrant can be expected to work on average 6.08 years in the U.S. which is significantly less than similarly situated native born U.S. workers. Collectively our results strongly suggest that in cases involving injury, wrongful death, and other employment issues, economic damage estimates that use U.S based work tables to calculate economic damages will over estimate the alleged economic loss of earnings in cases involving undocumented Mexican workers.

Second, a Cox proportional hazards model of the probability of returning to Mexico shows that, for undocumented Mexican workers, the probability of leaving the U.S. labor market and returning to Mexico is highly correlated the individual’s demographics and stock of human and cultural capital. For instance, married undocumented Mexican workers, the majority of whom had families back in Mexico and those who did not speak or read English, were significantly more likely to leave the U.S. labor market and return to Mexico than other undocumented workers. In contrast, undocumented workers who held higher wage jobs and had developed stronger social ties to the U.S. were significantly less likely to return to Mexico.

This paper is laid out as follows. In the first section we provide a description of a case study to motivate the discussion of the issues that arise in injury, wrongful death and employment discrimination lawsuits. In the second section, we discuss the factors that

PRELIMINARY DRAFT Please do not quote 5

enter the undocumented worker’s economic decision to leave the U.S. labor market and return to Mexico. In this section we also provide background and a literature review of studies that have previously examined these issues. In the third section of this paper, we provide an introduction to the MMP data files used to study the work life expectancy of undocumented Mexican immigrants in this paper. In the fourth and final section of this paper we use a Cox proportional hazards model to estimate the probability that the Mexican worker will return to Mexico and leave the U.S. labor market. We also discuss the economic damage calculation for the hypothetical migrant workers in our case study.

Case study

To motivate the economic damage methodology, consider the following case study. This case study involves the calculation of economic damages stemming from the loss of earnings associated with the death of two undocumented Mexican workers working in the U.S. The case details are based on facts from an actual case in which we were involved. The names of parties have been changed.

In this case, two undocumented Mexican workers were killed while riding in a 12 passenger van that was traveling from Texas to California. Both workers were killed when the van rolled over after a tire exploded. In response to the accident, the plaintiffs’ attorneys representing the families of the two workers sued a number of parties including the tire manufacturer, the automaker, and the dealership that sold the van.

Worker #1 lived in California and was traveling to Texas to work at a major construction site. This individual was a 32 year old male with approximately 5 years of schooling who emigrated from Mexico approximately 3 years prior to the accident. At the time of his death, he worked as a construction worker and day laborer and earned on average $6.50 per hour and worked an average of 8 months out of each year. The deceased was married, and although he had a family in Mexico he also had relatives and extended family in the U.S. He was active in sports and social activities in the U.S. and spoke and wrote English proficiently.

Worker #2 lived in California and was traveling in the van to a new intrastate location for work at the same location as Worker #1. Worker #2 was also a 32 year old male with approximately 5 years of schooling. He emigrated from Mexico approximately 2 years ago prior to the accident. At the time of his death, he worked as drywall installer and earned approximately $8.75 per hour and was able to find work approximately 10 months out of each year. The deceased worker was married and had a spouse that resided in Mexico as well as relatives who lived in the U.S. According to depositions of the deceased’s family members, he was not active in any sporting organization or social activities in the U.S. and did not speak or read English with any degree of proficiency..

Based on these case facts, the objective is to provide an analysis of the economic value of the potential lost earnings associated with the death of each of the undocumented Mexican workers. Unlike a native U.S. worker, there are a myriad of reasons, such as voluntarily returning home, inability to find a job, and involuntarily being deported, which could hinder an undocumented Mexican worker’s ability to earn a wage in the U.S.

PRELIMINARY DRAFT Please do not quote 6

labor market. In addition, as will be discussed below, the demographics, human capital stock and cultural capital stock will have an effect on the probability that a given individual worker will leave the U.S. labor market and return to Mexico.

Economic damage model and the decision to leave the U.S. labor market

Using the standard labor force participation model, the economic damage model that describes the economic value of the lost U.S. wages for the two workers in the case can be written as follows.

(1) tttt

T

tt ilapWUSWagesFutureLostPV −

=

+××××= ∑ )1()(0

where t is the time period, W are the wages earned by the undocumented worker during period t, p is the probability of remaining in the US during time period t, a is the probability of being active in the US labor force in time period t, and l is the probability of being alive at time period t.

While each of the factors that go into the present value calculation of the lost wages provide a potential challenge in any economic damage analysis in cases involving undocumented workers, the areas where there is the least amount of prior research is regarding the probability of remaining in the U.S. labor market (p) and the probability of being active in the labor market (a). To date, in actual court cases, economic damage studies that have examined the loss of earnings or earnings support for undocumented workers have generally, by assumption, either ruled out future economic losses or have simply used U.S. based work tables to provide an estimate of work life expectancy. Although there have been studies on the likelihood a Mexican citizen will migrate to the U.S., to my knowledge there have been few studies that look at the likelihood that an undocumented worker will remain in the U.S.

Estimating the probably length of time that an undocumented worker will remain in the U.S. is the focus of this paper. Specifically, the decision to remain in the U.S. can be evaluated in the standard cost and benefit framework. The simple framework used in this paper is essentially the same as in other economic studies such as Orrenius and Zavodny (2001) that has examined the decision to migrate to the U.S. In our model, if the perceived utility (benefit) from returning home to Mexico exceeds a certain threshold (cost) then the undocumented worker will return to Mexico. Unlike the decision to migrate to the U.S., there are relatively lower transactional costs associated with returning home. For instance, there is no need to hire a coyote or guide to assist with re-entry into Mexico. However, the worker still faces issues such as the pecuniary cost of relocating, the loss of U.S. wages, and the uncertainty of employment when he or she returns to Mexico. Similarly, an undocumented worker will stay in the U.S. if he or she finds that the cost of relocating back to Mexico exceeds the benefits from staying.

The decision to leave the U.S. labor market can then be written as follows.

PRELIMINARY DRAFT Please do not quote 7

(2) 0)(

1)(

*

*

=<

=>

∑ythenMXUIf

ythenMXUIf

iii

iii

β

β

In practice, the researcher can only see the individual level characteristics that identify the undocumented worker and the outcome variable y, i.e. if the person has returned to Mexico or not. As mentioned above, while there has not been much work done on the rate at which undocumented Mexican workers return to Mexico, the current literature does suggest there are several factors that affect a worker’s decision to return home. These factors include the demographic characteristics of the individual and the stock of human and cultural capital possessed by the individual worker.

In terms of demographics, factors such as a person’s age, gender, and marital status can potentially affect the undocumented workers decision to return to the labor force. Female undocumented workers may be less likely to voluntarily return home to Mexico for a number of reasons. For instance, some researchers have suggested females may be more likely to stay in the U.S. because of the relatively higher potential physical risk associated with undocumented entry into the U.S. while other studies suggest that the occupations female undocumented workers perform may be at a slightly lower risk for deportation. Similarly, older workers may be less likely to want to voluntarily leave the U.S. and may also become more adept at avoiding issues and situations that could lead to deportation. In contrast, undocumented workers who are married and have spouses and families in Mexico may find that separation from their families proves too difficult and may be relatively more likely to leave the U.S. and return to Mexico. See Massey, Durand and Malone (2002), Cerrutti and Massey (2001), Kanaiaupuni (2000) and Massey (1987) for more discussion on the factors that to enter a Mexican migrant worker’s decision to leave the U.S. labor market and return to Mexico.

In addition to the demographic make up of the undocumented worker, the stock of human capital possessed by the individual can also potentially affect a given person’s decision to leave the U.S. labor market and return home. For instance workers who have a higher education level may find they have greater job opportunities and be less likely to leave the U.S. Similarly, undocumented workers who are able to secure employment in relatively higher paying jobs in the U.S. may believe it would be relatively difficult to obtain similar employment in Mexico and will therefore be less likely to return to Mexico. Conventional wisdom would also suggest undocumented workers who are able to speak and write English may have an easier time becoming acclimated to living in the U.S. and may be less likely to voluntarily want to return to Mexico. English speaking ability may also lower the probability or being involuntarily returned to Mexico. Individuals who apply and receive some level of authorization, short of citizenship, should also be less likely to reverse migrate.

The stock of cultural capital possessed by undocumented Mexican workers is another area that can potentially affect an individual’s decision to return home. Cultural capital in the current setting includes factors such as the person’s family structure, social networks, and attachment to ‘U.S. ways’. Undocumented workers who are identify more

PRELIMINARY DRAFT Please do not quote 8

closely with the U.S. may be less likely, all other factors equal, to want to return to Mexico. For instance, even undocumented workers who have immediate family members that reside in Mexico may still be more likely to stay in the U.S. if they have a well established network of extended family members. Undocumented workers who have a greater attachment to ‘U.S. ways’ as demonstrated by factors such as having children in U.S. schools and having U.S. bank accounts should also tend to exhibit a greater tendency to stay in the U.S. The same is true for the existence of a strong non-family social network as exhibited by participation in sports or social organizations or can be described as “transnational.”

“Transnationalism” is utilized to describe social, economic and cultural activities that flow across social and geographic borders. It is central to a new conception of migration in which migrants conduct portions of their lives in the host country and in the country of origin. Why is this considered a new view of migration? One answer is that improved technology (phones, planes, money wires, etc.) makes it easier for migrants to maintain contacts with their home (Portes et al. 1999). Additionally, the global economy encourages migration in two ways—by creating the “push” and “pull” of migration. First, there is a labor shortage in countries that are benefiting from global capitalism. Increasing economic success in First World nations creates the need for so many workers that an international labor force is needed. Second, the changing conditions of a global economy have disrupted Third World economies, such as Mexico’s, spurring migration to other countries for economic survival. It is important to acknowledge these two results of a global economy simultaneously. The “push-pull” forces that create Mexican migration to the U.S. have the same cause—the global economy that helped to produce the flourishing U.S. economy is also a cause of Mexico’s economic hardship.

In conjunction with transnationalism, anthropologist Arjun Appadurai coined the phrase “deterritorialization” which refers to the cultural implications of transnational capital. In “Modernity at Large” (1996) he argues that “-scapes” (ethnoscapes, mediascapes, technoscapes, finacescapes and ideoscapes) constantly shape society. The most significant of these scapes are the ethnoscapes that “make up the shifting world in which we live; tourists, immigrants, refugees, exiles, guest-workers and other groups and persons constitute an essential feature of the world and appear to affect the politics of and between nations to an unprecedented degree.” Cultural practice is connected to transnationalism in the production of imagination (as a social practice). The new technology mentioned above allows migrants to imagine in and between cultural and geographic boundaries. Migrants’ identities are not constructed by a single country; they are shaped by the migrants’ real and imagined experiences in his country of origin and his host country. If undocumented workers in the U.S. experience their new country of residence as part of their identity and have a growing sense of transnationalism, their desire to return to Mexico may decrease.

Regarding transnational social networks within the U.S., one factor that affects which community a Mexican migrants will settle and work in the U.S. is the location of Mexican social networks. This factor exists in response to the geographic distribution of jobs. Rarely does an Mexican migrate to the U.S. without knowing other Mexicans in the host country. Communication with Mexican migrants already working in the U.S. guide

PRELIMINARY DRAFT Please do not quote 9

the future migrant to pick a location based on available housing, work, family and friends. Geographical, institutional, and economy wide factors may also affect the undocumented workers decision to stay in the U.S. labor market. For instance, Mexican migrants that migrate to locations that have a relatively higher Mexican/Mexican-American population or areas with more lenient and accommodating attitudes towards migrant work issues may also be relatively less likely to leave.

Also, changes in economic conditions in either the U.S or Mexico could change the relative labor market opportunities that are available to undocumented workers and may either encourage or discourage undocumented Mexican workers to leave the U.S. work force. Wide sweeping changes in U.S. stances on immigration such as those that occurred in the 1980’s when the Immigration Reform and Control Act (1986) was passed can also potentially indirectly affect the undocumented workers decision in a number of ways. For instance, a more stringent immigration policy that focuses on curtailing entry in the U.S. may result in an increase in undocumented workers desire to stay since the cost associated with re-entry into the U.S. will be relatively higher. All else equal, a more restrictive immigration policy may also reduce the supply of migrant labor and hence result in a positive wage effect that further increases the undocumented Mexican workers desire to stay in the U.S.

Economic damage calculations in actual cases involving undocumented Mexican workers

As discussed earlier, undocumented Mexican workers have been, and continue to be, parties in a number of different types of legal cases such as those involving products liability, medical malpractice, labor disputes involving allegations of FLSA violations and employment discrimination. In some instances, the plaintiffs and/or their families have been awarded some relatively sizable economic damages for both back and front pay losses. In both the initial proceedings and the appeals associated with a number of these cases, two main types of questions arise in the calculation of economic damages in cases involved undocumented Mexican workers.

The first question involves the legality of awarding back pay and past damages to undocumented workers. It is frequently argued in cases involving undocumented workers that the landmark Supreme Court ruling in Hoffman Plastics Compounds voids these workers claims to past or back pay losses. In the Hoffman case, the National Labor Relation Board (NLRB) found that four workers, including an undocumented worker who went by the name Jose Castro, were wrongfully terminated because they engaged in union related activities and the plaintiffs were awarded back pay. In this case, the NLRB awarded back pay to Mr. Castro from the date of his termination to the date that Hoffman Plastic Compounds first learned of his immigration status. The Court’s majority found the Immigration Reform and Control Act (1986) limited the NLRB’s ability to award back pay to undocumented workers because awarding “backpay to illegal aliens would unduly trench upon explicit statutory prohibitions critical to federal immigration policy, as expressed in IRCA”.

PRELIMINARY DRAFT Please do not quote 10

According to some legal observers, this Court ruling has had a significant chilling effect on cases, such as FLSA violation and employment cases, where a significant amount of the potential economic loss allegations involves lost back pay. For instance, after the ruling the EEOC rescinded the agencies previous guidance concerning the appropriateness of back pay to undocumented workers.

In other cases that concern the award of back pay to undocumented workers it has been argued, successfully in some instances, that Hoffman is not appropriate and therefore back pay should be awarded. For instance, in Celi v. 42nd Street Development Project, Inc, it was argued Hoffman did not apply to state court in a New York action and the facts in the case were distinguishable from the Hoffman case. In this particular case the state court judge agreed with the argument and allowed both back and front pay to be awarded to the injured undocumented worker.

The second question revolves around the base earnings by which to calculate value of the undocumented workers alleged future lost earnings. In particular, across the country the case law is not completely clear on whether future earnings loss should be based on the wages that the undocumented worker could have earned in the U.S. or in their home country. In some cases, courts have ruled that under certain conditions, such as when the deportation of the undocumented worker is imminent, economic damages should be based on the wages that the illegal worker would have been earned in their home country and not those that would have been earned in the U.S. In other cases, courts have ruled economic damages should be based on the wages the worker would have been able to earn in the U.S. in situations where the workers deportation was not imminent. In still other cases (cites), courts have ruled that U.S. based economic damages are applicable in cases where the employer knew that the worker was in the country illegally.

Mexican Migration Project (MMP) Data

We use data from the Mexican Migration Project (MMP) to study the work life expectancy and the factors related to work life for undocumented workers. The MMP is a collaborative research project based at Princeton University and the University of Guadalajara that chronicles the U.S. immigration experiences and intra-Mexico migration experience for a sample of Mexican citizens. The MMP survey data is constructed as follows. Each year, during the winter months, which is when Mexican seasonal migrants are home, the Mexican Migration Project randomly samples households in communities located throughout Mexico. After gathering social, demographic, and economic information on the household and its members, interviewers collect basic information on each person's first and last trip to the United States. From household heads, they compile a year-by-year history of U.S. migration and administer a detailed series of questions about the last trip northward, focusing on employment, earnings, and use of U.S. social services. Following completion of the Mexican surveys, interviewers administer identical questionnaires to migrants in the U.S. that are from the same communities sampled in Mexico.

In this study, we use the ‘MIG’ data file that is a person-level file containing detailed information about the head of household’s first and last visit to the U.S. The detailed

PRELIMINARY DRAFT Please do not quote 11

questions include measures of economic and social activity during the last U.S. visit. Data concerning the person’s last U.S. visit include the year of crossing, social network, information on social relationships, type of employment, wages earned and information on family finances, and remittances to Mexico. In addition there is also information concerning the geographical location where the individual crossed into the U.S., documentation used, costs, method and whether deported for each border crossing.

The MMP data presents a particularly interesting and insightful picture regarding the working life of a typical Mexican immigrant in the U.S. First, it is clear from the MMP data that the Mexicans who have immigrated more recently have tended to settle in to different places than those individuals who immigrated in more recent years. Table 1 shows the destination of a typical Mexican immigrant worker. It is broken down by decade in which the person last immigrated to the U.S. and the destination of the immigrant. As this table shows, in recent decades the current trend is for Mexican immigrants to immigrate to areas other than California and Texas. In the 1970’s for instance nearly 80% of Mexican immigrants emigrated to either Texas (13.76%) or California (65.06%). During the same time period, the immigration rate to other states, such as Iowa, Nebraska, and Arkansas, was about 10%. In contrast, in the 1990’s the percentage of Mexicans illegally immigrating to California and Texas fell to about 59% while the percentage of illegal immigration to the other states increased to nearly 30% of all immigration activity from Mexico. One often cited reason for the change is the development and expansion of labor intensive manufacturing plants in states such as Arkansas.

Second, once in the U.S., social networks such as relatives, friends, and social organizations, have become more important in finding employment for Mexicans who have immigrated in more recent decades. Tables 2 and 3 show the method of job search used by Mexican immigrants and the occupational distribution for those workers. For instance as shown in Table 2, only about 18.6% of workers who first immigrated before 1960 used a social network to find employment while in the 1990’s about 68.5% of Mexican immigrants relied on a social network to find employment in the U.S. Although not shown in the table, it should also be noted that nearly half of the supervisors of the Mexican immigrants in their U.S. employment were also Latino. Collectively, this highly suggests that the degree of social and cultural capital possessed by the individual immigrant will more than likely have an effect on the likelihood that the immigrant will stay in the U.S.

Third, as is frequently reported, the MMP data shows that Mexican migrants tend to be congregated into agriculture and manufacturing jobs. As shown in Table 3, the degree at which Mexican migrants work in agriculture and manufacturing has increased while the percent of migrants who work in personal services has decreased for individuals who immigrated in later decades. For instance, about 50% of the individuals in the MMP data that migrated in the 1950s worked in manufacturing while about 12% of Mexican migrants worked in personal service related jobs. In contrast, about 70% of individuals who migrated in the 1990s worked in manufacturing and agriculture and only about 6% of Mexican migrants worked in personal service related jobs.

PRELIMINARY DRAFT Please do not quote 12

Fourth, the MMP data shows most undocumented Mexicans in the U.S. ultimately return to Mexico. In fact, the average duration of stay for a Mexican migrant who comes to the U.S. to work is actually relatively short. Table 4 shows the rate by the decade at which Mexican migrants return back to Mexico. As shown in Table 4, the MMP data shows that over the total dataset about 76.3% of the Mexican migrants returned to Mexico and have since not returned to the U.S. For Mexicans who immigrated before 1960 96.2% ultimately returned to Mexico and have not returned to the U.S.

Table 4 shows the distribution for the duration of stay in the U.S. for the last trip to the U.S. and for the total time spent in the U.S. As shown in the table the average time spent in the U.S. on the last trip taken by the Mexican migrant to the U.S. was about 3 years while the median was less than 1 year. The average total amount of time, i.e. adding up all the time spent on each of the individual’s trips to the U.S., was about 6 years while the median total time spent in the U.S. was about 3 years. Table 5 provides a breakdown by age of time spent and the average number of trips taken by Mexican migrants to the U.S. As the tables show the average amount of time spent by Mexican migrants in the U.S. increases with the age of the individual surveyed but still suggests that Mexican migrants will have a shorter work life than native born U.S. workers.

Estimating the probability of leaving the U.S. work force and returning home

A duration analysis of the time spent in the U.S. during the Mexican migrant’s last stay in the U.S. is performed to estimate the probability that a Mexican migrant will leave the U.S. work force and return to Mexico. The duration of the immigrant’s last stay is used because in an actual lawsuit involving an undocumented worker the injury, death or discrimination would have by design occurred during the person’s last stay in the U.S. Furthermore, the relevant economic damage period for lost U.S. wages in the case is interpreted to begin at the time of the incident and continue over the period in which the undocumented worker could have been expected remain in the U.S. had the incident not occurred. If it is required to estimate the economic value of the U.S. wages that the undocumented worker could have earned from subsequent illegal entries into the U.S., the model discussed below would need to be augmented to account for the possibility of future re-entry into the U.S.

Specifically for our analysis, we use the MMP data to estimate the likelihood that an undocumented Mexican migrant will leave the U.S. labor market and return home to Mexico. For our study we are interested in the instantaneous hazard rate h(t) which by definition is the potential (probability) at time t that the undocumented worker will return to Mexico given that they have already spent t years in the U.S. That is,

)()()3( tTttTtPth ≥∆+≤≤=

The Cox proportional hazards model, which incorporates a time dependent baseline hazard rate ho(t) and time independent the explanatory variables X, and is written as follows.

PRELIMINARY DRAFT Please do not quote 13

∑= i

ii X

o ethXthβ

)(),()4(

where ho(t) is the baseline hazard, X are the explanatory variables related to the person’s likelihood of leaving the U.S. labor force and β are the model coefficients. In the Cox hazard model, the baseline hazard function is not specified but is estimated using kernel smoothing methodology. The hazard model treats all persons who have not returned to Mexico as right censored.

To estimate our model of the migrant’s decision to leave the U.S. labor force, we consider a set of independent explanatory variables that measure the Mexican migrants’ stock of human capital, cultural capital, economic conditions, and demographic profile. A description of the variables included in the Cox model is shown in Table 6. A statistical summary of the variables is found in Table 7. A list of the Cox model coefficients and standard errors is shown in Table 8.

Overall the model fits the data well and the coefficients have the expected signs. In terms of demographics profile, the model suggests once a female migrates to the U.S. she is less likely than a similarly situated male to leave the U.S. workforce. A similar result is found for older workers, as the coefficient on the age variable is negative, but the effect diminishes over time as demonstrated by the squared age term. In terms of human capital variables, the wage level, education, and English speaking ability variables display their expected signs. Mexican migrants with higher levels of education and those who are able to earn a higher than median wage are also less likely to leave the U.S. workforce which is in line with conventional thinking that suggests that the more educated and skilled workers may have greater job opportunities all other factors equal. The ability to speak and read English is a highly significant and important determinant of a person’s decision to leave the U.S. worker. Converting the Cox model coefficient estimates shown in Table 8 to hazard rates by taking the exponential of the estimates, shows that a person who is not able to speak or read English is about 1.5 times more likely to leave the U.S. labor market than a Mexican migrant who is able to speak and read English.

The measures of cultural capital included in the model are also important determinants in the Mexican migrant’s decision to return home and leave the U.S. labor market. Individuals who possess a U.S. bank account, have children in U.S. schools, and are active in social and sports organizations are more likely to stay in the U.S. than those Mexican migrants who do not possess that cultural capital. For instance, individuals with either a bank account or who have children in U.S. schools are about half as likely as those who do not have a bank account or children in school to stay in the U.S. These results collectively suggest Mexican migrants who have a growing sense of transnationalism and become more attached and accustomed to the social and cultural mechanisms in the U.S. may in part have better job and labor market opportunities. Mexican migrants who are married, most of whose spouses are in Mexico, or have extended family in the U.S. are more likely to leave the U.S. labor market.

In addition to the human capital, cultural capital, and demographics, other important factors include the legal status of the worker, the immigration environment, and the

PRELIMINARY DRAFT Please do not quote 14

immigration destination. Not surprisingly, workers, although a very small percentage of undocumented workers, who are able to obtain legal status are less likely to leave the U.S. labor force than those who do not become legally authorized workers. All other factors equal, individuals who migrate to California or Illinois are significantly less likely to leave the U.S. work force. Mexican immigrants who migrate to Texas do not exhibit this tendency. This result may in part be explained by more permissive attitudes toward illegal immigration, relatively better access to living or job resources, and better developed social networks in California and Illinois.

Evaluating the economic damages for the case study

For this case study, we will use the standard LPE economic damage approach as shown in equation 1. The major difference in the LPE approach in the current case is the incorporation of the factor for the probability of remaining in the U.S. To estimate the probability of leaving in the U.S. labor force, equation 4 is evaluated using the information contained in the case study. The probability of remaining in the work force is equal to 1 minus the results from equation 4. The probability that an undocumented Mexican migrant worker, with the characteristics discussed in the case study, would remain in the U.S. workforce, given that the person has remained (survived) in the U.S. workforce t years, is shown in Table 9.

For instance for a Mexican migrant who has resided continuously in the U.S. for 10 years, is active in social organizations, and speaks English, there is a 0.6678 probability that he will remain in the U.S. labor force throughout that year. It is important to remember that the table does not give the probability that the person will be active in the work force, that likelihood must be estimated separately. The table provides estimates for the likelihood that a Mexican migrant will be in the U.S. labor market given years of residency between 0 and 28 years in the U.S.1 The probabilities can be used in a LPE model or the total expected time spent in the U.S. labor market could be estimated by summing the relevant time periods. For instance, a Mexican migrant who has resided in the U.S. for 15 years can be expected to remain in the U.S. for approximately another 7.91 years. Using the same reasoning a Mexican migrant, fitting the case study characteristics of course, who just arrived in the U.S. from Mexico, can be expected to stay in the U.S. for approximately 16.63 years. A similar methodology can be employed to calculate the economic damages for a person with any given set of characteristics.

To complete the analysis of the case study, the parameters for economic damages are used to calculate the damages in the case study are shown in Table 10.

(Discussion)

1 Observations where a time greater than 28 years, which was approximately the 99th percentile, was not included because the quality of the data for these observations was suspect and they also had a disproportionate effect on the parameter estimates.

PRELIMINARY DRAFT Please do not quote 15

Table 1: Destination of Mexican Immigrant Worker

Decade of Immigration AZ CA IL TX Other States Total

1959 and before 28 285 20 167 108 608 4.61 46.88 3.29 27.47 17.76 100.00

1960 to 1969 30 321 43 110 62 566 5.30 56.71 7.60 19.43 10.95 100.00

1970 to 1979 9 676 109 143 102 1,039 0.87 65.06 10.49 13.76 9.82 100.00

1980 to 1989 21 1,245 123 256 244 1,889 1.11 65.91 6.51 13.55 12.92 100.00

1990 to 1999 59 865 169 288 543 1,924 3.07 44.96 8.78 14.97 28.22 100.00

2000 to 2005 8 49 22 44 169 292 2.74 16.78 7.53 15.07 57.88 100.00

Total 155 3,441 486 1,008 1,228 6,318 2.45 54.46 7.69 15.95 19.44 100.00

PRELIMINARY DRAFT Please do not quote 16

Table 2: Method of job search used by immigrant worker by decade Decade of Immigration

Job search method Before 1960

1960 to 1969

1970 to 1979

1980 to 1989

1990 to 1999

2000 to 2005 Total

Self Obtained 24.63 20.85 26.70 27.24 26.65 29.50 26.29 Social network/Contac 18.58 33.62 69.22 68.89 68.53 61.15 60.81 Contracted 56.58 44.89 3.83 3.74 4.24 7.91 12.49 Other 0.21 0.64 0.25 0.12 0.59 1.44 0.41 100.00 100.00 100.00 100.00 100.00 100.00 100.00

PRELIMINARY DRAFT Please do not quote 17

Table 3: Occupational Distribution for immigrant Mexican workers Decade of Immigration

Labor force status Before 1960

1960 to 1969

1970 to 1979

1980 to 1989

1990 to 1999

2000 to 2005 Total

Not in Labor Force 31.09 17.31 11.93 7.04 5.87 5.48 10.65 Professional/Arts 1.15 4.77 4.43 4.34 4.68 5.48 4.24 Agriculture and Manufacturing 50.82 59.19 59.87 64.64 69.23 71.23 63.74 Sales/Personal Services 13.98 13.60 17.13 14.29 10.76 6.16 13.22 Other 2.96 5.12 6.64 9.69 9.46 11.64 8.15 100.00 100.00 100.00 100.00 100.00 100.00 100.00

PRELIMINARY DRAFT Please do not quote 18

Table 4: Rate of return to Mexico by decade

Decade of immigration

Percentage that have not returned to Mexico since last migration

Percentage that have returned to Mexico since last migration

1959 and before 3.78 96.22 1960 to 1969 12.72 87.28 1970 to 1979 17.32 82.68 1980 to 1989 27.32 72.68 1990 to 1999 28.79 71.21 2000 to 2005 51.71 48.29

Total 23.68 76.32

PRELIMINARY DRAFT Please do not quote 19

Table 5: Average US Work Experience and Number of Trips by Age Group Age Group Average Time Spent

Working in US Average Number of Trips to US

20-29 4.62 years 1.36

30-39 7.76 years 1.91

40-49 11.17 years 2.26

50-59 11.12 years 2.25

PRELIMINARY DRAFT Please do not quote 20

Table 6: Distribution of selected variables

Percentile Duration of the last stay in U.S. (in years)

Total U.S. Labor Market Experience (in years)

95th 15.5 22.5 75th 2.50 8.50 50th 0.83 3.00 25th 0.42 1.00

Mean 3.03 6.08

PRELIMINARY DRAFT Please do not quote 21

Table 7: Description of variables in model Variable Description Mean

female Equals 1 if female 0.0461

age Age as of the date of last migration to the U.S. 35.73

edyrs Number of years of education 5.15 married Equals 1 if the individual is married 0.1844

familyinUS Equals 1 if the individual has extended family in the U.S. 0.5457

social Equals 1 if the individual is involved in sports or social organization 0.1203

noenglish Equals 1 if the individual does not speak or read English 0.4468

highwage Equals 1 if the the person earns over the median income ($5 per hour) 0.3183

bank Equals 1 if the person has a bank account in the U.S. 0.1294

legalmigration Equals 1 if the person has become a legal or authorized worker or citizen 0.2421

schoolkid Equals 1 if the person has a child in school 0.1441

texas Equals 1 if the person's last migration was to Texas 0.1605

ca Equals 1 if the person's last migration was to California 0.5431

il Equals 1 if the person's last migration was to Illinois 0.0755

preIRCA Equals 1 if the year of the migration was before the IRCA 0.5143

yr70s Equals 1 if the year is between 1970 and 1979 0.1193

yr80s Equals 1 if the year is between 1980 and 1989 0.2561

yr90s Equals 1 if the year is between 1990 and 1999 0.3808

yr00s Equals 1 if the year is between 2000 and 2004 0.0867

PRELIMINARY DRAFT Please do not quote 22

Table 8: Cox model coefficients

Variable Coefficient Standard Error Z-stat

female -0.15432 0.069609 -2.22 age -0.01338 0.006433 -2.08 age2 0.000139 7.76E-05 1.79 EDYRS -0.01083 0.004686 -2.31 married 0.174784 0.042412 4.12 familyinUS 0.146534 0.030695 4.77 social -0.17611 0.053717 -3.28 noenglish 0.481799 0.033069 14.57 highwage -0.23251 0.039834 -5.84 bank -0.68157 0.068354 -9.97 legalmigra~n -0.2871 0.045946 -6.25 schoolkid -0.6302 0.058841 -10.71 texas 0.02874 0.046197 0.62 ca -0.21513 0.037723 -5.7 il -0.41781 0.064336 -6.49 preIRCA -1.028 0.05014 -20.5 yr70s -0.17646 0.050788 -3.47 yr80s -0.53522 0.049913 -10.72 yr90s -1.45408 0.068363 -21.27 yr00s -2.31148 0.099593 -23.21

PRELIMINARY DRAFT Please do not quote 23

Table 9: Probability of remaining in the U.S. labor force Time spent already spent in U.S. No Social, No English Speaker Social, English 0 0.781197 0.838827 1 0.588385 0.696799 2 0.528994 0.653051 3 0.519997 0.646424 4 0.592118 0.699549 5 0.588400 0.696810 6 0.641132 0.735653 7 0.707776 0.784744 8 0.652843 0.744280 9 0.682289 0.765970 10 0.549052 0.667826 11 0.634764 0.730962 12 0.632303 0.729150 13 0.605070 0.709090 14 0.738788 0.807588 15 0.555195 0.672351 16 0.646915 0.739913 17 0.539794 0.661006 18 0.633715 0.730190 19 0.666836 0.754587 20 0.377279 0.541296 21 0.575562 0.687354 22 0.723740 0.796503 23 0.405889 0.562370 24 0.476665 0.614505 25 0.370966 0.536646 26 0.450357 0.595126 27 0.39776 0.556382 28 0.370894 0.536593

PRELIMINARY DRAFT Please do not quote 24

Table 10: Case study economic damage results [TO BE INSERTED HERE]

PRELIMINARY DRAFT Please do not quote 25

Bibliography

Orrin Baird (20030, “Undocumented Workers and the NLRA: Hoffman Plastics and Beyond”, The Labor Lawyer 19, Pages 153-169

George Borjas (1992), “Ethnic Capital and Intergenerational Mobility”, Quarterly Journal of Economics, Vol. 107, Issue 1, Page 123-50.

Michael R. Brown (2003), “Hoffman Plastic Compounds v. NLRB: The First Step”, The Labor Lawyer 19, Pages 169 – 185.

Cerrutti and Massey (2001), “On the Auspices of Female Migration Between Mexico and the United States”, Demography 38: 187-200

Ciecka, James, Thomas Donley, and Jerry Goldman (2000). “A Markov Process Model of Work-Life Expectancies Based on Labor Market Activity in 1997-1998,” Journal of Legal Economics, 9(3), 33-66.

Catalina Amuedo-Dorantees and Kusum Mundra (2004), “Social Networks and Their Impact on the Employment and Earnings of Mexican Imigrants, Working Paper.

Massey, Douglas, Durand, Jorge and Malone, Nolan (2002), “Beyond Smoke and Mirrors”, Russell Sage Foundation.

Pia M. Orrenius and Madeline Zavodny (2001(, “Self-selection among Undocumented Immigrants from Mexico, Federal Reserve Bank of Atlanta, Working Paper Series.

Jeffery Passel (2005), “Estimates of the Size and Characteristics of the Undocumented Population”, Pew Hispanic Center.

Mexican Migrant Project, The (2004). “MMP93 (pers file)”, mmp.opr.princeon.edu, 2003.

Millimet, Daniel L., Michael Nieswiadomy, Hang Ryu, and Daniel Slottje (2002). “Estimating Worklife Expectancy: An Econometric Approach,” Journal of Econometrics, vol. 113(1), 83-113.

Skoog, Gary and James Ciecka (2001). “A Markov (Increment-Decrement) Model of Labor Force Activity: New Results Beyond Work-like Expectancies,” Journal of Forensic Economics, 11(1), 1-21.

Smith, Shirley (1985). “Revised Worklife Tables Reflect 1979-80 Experience,” Monthly Labor Review.

U.S. Bureau of Labor Statistics (1982). “Tables of Working Life: The Increment-Decrement Model,” Bulletin 2254.