|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 THREE ESSAYS ON IMMIGRATION REFO RM, WORKER SELF-SELECTIVITY AND EARNINGS IN THE U.S. FARM LABOR MARKET By LURLEEN M. WALTERS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008
2 2008 Lurleen M. Walters
3 To my family for their unconditi onal love, guidance and support and to the memory of my fath er, the late Ivor L. Walters An educator by profession and at heart, he was equally my toughest critic and my staunchest supporter I deeply regret that he passed aw ay before he saw the fruits of his labor for he would have been extremely proud
4 ACKNOWLEDGMENTS I give thanks to Alm ighty God for His many bl essings and for the strength and patience to persevere when faced with lifes challenges. Se veral persons have been instrumental in my being able to complete this research and I take this opportunity to thank them all. I extend sincere thanks to th e members of my supervisory committee for their assistance throughout my program and with this research. I am especially grateful for the guidance and support of my committee chair, Dr. Robert Emerson, who patiently encouraged my interest in agricultural labor and immigra tion issues (alongside pre-existing interests in trade and development). Through his high st andards I truly gained newfound appreciation of the rigors of research, and I sincerely thank him for providing me with seve ral professional opportunities in the process of completing this st udy. I also wish to extend my si ncere gratitude to my co-chair Dr. Carlton Davis for his sage advice on various matters over the years; in true West Indian fashion, he always emphasized the importan ce of academic excellence. I thank him for encouraging me to pursue a PhD degree at the Un iversity of Florida and for his strong support throughout my program. I must also express very sincere thanks to Dr. Edward Evans for his friendship and unwavering encouragement of my academic and professional pursuits. When times got tough on occasion, his advice (generously sprinkled with hilarious but very practical West Indian proverbs) served as much needed d oses of commonsense that kept me focused. I am also very grateful to Dr. Richard Kilmer for his advice and assistance in various capacities over the years. Finally, I wish to thank Dr. Da vid Denslow for his constructive comments on my work and for encouraging me to broaden my perspectives on my research. My appointment as a graduate research assi stant to Dr. Emersons farm labor research team allowed me to benefit immensely from th e advice of Dr. Nobuyuki Iwai. I sincerely thank him for his assistance during those times when the challenges seemed insurmountable,
5 particularly during our visits to Aguirre International in Bu rlingame, CA and during the wee hours of the morning at the office! I am also grat eful to Dr. Susan Gabbard and her associates at Aguirre International for assistance with the National Agricultural Workers Survey (NAWS) data, and to Daniel Carroll at th e U.S. Department of Labor for granting access and authorization to use the data. This research has been supporte d through a partnership ag reement with the Risk Management Agency, U.S. Department of Agricu lture; by the Center for International Business Education and Research at the University of Florida; by the Florida Agricultural Experiment Station; and the International Agri cultural Trade and Policy Center at the University of Florida. I have had the pleasure of meeting people from far and wide while at the University of Florida. I thank my core-mates, particular ly Shiferaw Feleke and Mariana Varese, for friendships that grew strong over the shared ex perience of the core and many fun experiences thereafter. I also thank S ophia Glenn, Raphael Pierre, Brian Francis, Ronald Gordon, Orachos Napasintuwong, Mihaela Marcu, Garfield & Marsha Lowe, Keith & Nicole Duncan and my Caribbean liming partners for their friendship and countless hours of comic relief. I am also very appreciative of the assistance I received from Vicki Jenkins and Jessica Herman throughout my program, and extremely grateful for the strong support of the Wright family of Gainesville. I would like to extend heartfelt thanks to so me very special friends Alexander Wooten, Evol Byron-Alexander, Don Nisb ett, Elburna Prentice, Racquel Morrison and Thelma Mills for their unwavering loyalty and st rong moral support. They sto od by me through thick and thin, particularly during recent mont hs when my temperament was anything but rosy. Upon introspection, I admit that I may have been a tad bratty at times and I th ank them for allowing me to draw on their respective streng ths when it was most needed.
6 My family has been my rock throughout my academic sojourn in the US. They took every step with me: they celebrated my achievement s and, in equal measure, offered constructive criticism when they felt I needed to do better. I wish to express my love and gratitude to my mother: Mrs. Millicent Morton-Walters, my siblings: Gillian Laps ey, Kaye Bartlette (posthumously), Barbara Morton, Chris Morton and Karen Evelyn, and my niece: Lesa Tyson, for their unwavering love and support of my end eavors. I extend extra special thanks to my cousin Oselyn Griffin, who has been my surrogate mo ther since I migrated from the West Indies. She has been a source of strength and calm during the most challenging moments of my private life and I am forever indebted to her. I sincer ely thank Janice Walters, Wycliffe Walters and the rest of the Walters clan for their love and suppor t. I will never forget how they closed ranks behind me after my fathers untimely passing. Th ey kept me focused on my education for they knew it is what their brother (my father) would have wanted. I extend sincere thanks to my cousins Edris Liburd, (the re st of) the Liburd family, and Yo landa & Monty Davis for their support and prayers when they were most needed. Finally, I extend heartfelt thanks to my aunts Annette Manners, Emerald Amory, Shirley Jeff ers, Violet Hanley their families and the extended Amory clan for their many words of encouragement over the years.
7 TABLE OF CONTENTS Page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ..........9LIST OF FIGURES.......................................................................................................................11ABSTRACT...................................................................................................................................13CHAPTER 1 INTRODUCTION..................................................................................................................15Overview....................................................................................................................... ..........15Data.........................................................................................................................................182 U.S. IMMIGRATION POLICY AND FARM LABOR M ARKETS.................................... 19U.S. Immigration Policy Review............................................................................................ 21The Bracero Program......................................................................................................22Amendments to the Bracero Program............................................................................. 23The British West Indies (BWI) Temporary Alien Labor Program.................................. 24The H-2 Program............................................................................................................. 25The Immigration Reform & Control Act of 1986...........................................................26Employer sanctions.................................................................................................. 26Legalization..............................................................................................................28The H-2A temporary worker program.....................................................................31Specialty Crop Agriculture and Farm Labor.......................................................................... 32The Immigration Reform and Control Act and Farm Labor Market Outcomes..................... 35US and Florida Farm Labor Market Outcomes, 1989-2004................................................... 35Workforce Characteristics............................................................................................... 35Demographics...........................................................................................................35Legal status............................................................................................................... 38Labor Market Characteristics.......................................................................................... 39Employer type and work experience........................................................................ 39Task at the time of interview.................................................................................... 40Seasonal employment............................................................................................... 41Compensation by Task, Employer Type, Contract Length and Legal Status................. 42Employment Patterns.......................................................................................................43Duration of labor force activity................................................................................ 43Post 9/11 labor force activity...................................................................................45Duration of labor force activity by employer type................................................... 46Concluding Remarks............................................................................................................. .47
8 3 WORKER SELF-SELECTIVITY AND FARM WAGES..................................................... 61Introduction................................................................................................................... ..........61Theoretical and Analytical Framework.................................................................................. 63Earnings Functions..........................................................................................................63Sample Selection Bias.....................................................................................................64Empirical Remedies.........................................................................................................66Multiple Selection Rules................................................................................................. 68Data..................................................................................................................................73Determinants of the Legal Status Decision..................................................................... 74Determinants of the Job Type Decision..........................................................................74Determinants of Farm Earnings....................................................................................... 75Results and Discussion......................................................................................................... ..75Descriptive Statistics....................................................................................................... 75Determinants of Legal Status and Job Type Decisions................................................... 76Selectivity Corrected Wage Equation Models................................................................ 79Concluding Remarks............................................................................................................. .824 PROPOSED IMMIGRATION POLICY RE FORM & FARMWORKER OUTCOMES ..... 93Overview....................................................................................................................... ..........93Immigration Policy Reform Proposals................................................................................... 96Analytical Framework........................................................................................................... .99Treatment Effects Approach.......................................................................................... 100Homogeneous and heterogeneous treatment effects.............................................. 104Parametric model with hetero geneous treatment effects........................................ 105Treatment effect parameters................................................................................... 107Data........................................................................................................................108Results and Discussion......................................................................................................... 108Policy Implications...............................................................................................................113Concluding Remarks............................................................................................................ 1145 CONCLUSIONS.................................................................................................................. 127Summary...............................................................................................................................127Research Issues and Suggestions for Future Research......................................................... 131APPENDIX BETA COEFFICIENTS FOR POLYNOMIA L AND NONPARAMETRIC METHODS ......... 133LIST OF REFERENCES.............................................................................................................134BIOGRAPHICAL SKETCH.......................................................................................................143
9 LIST OF TABLES Table Page 2-1 Crop worker characteristics by lega l status, United States and F lorida, 1989-2004.......... 55 2-2 Farm labor market characteristics by le gal status, United States and F lorida, 19892004....................................................................................................................................56 2-3 Workforce composition by legal status a nd em ployer type across time, United States.... 56 2-4 Workforce composition by legal status & em ployer type over time, Florida.................... 57 2-5 Legal status by task at time of inte rview, United States and Florida, 1989-2004 .............57 2-7 Compensation method by task type, Florida, 1989-2004 .................................................. 58 2-8 Compensation method by employer type, United States and Florida, 1989-2004 ............ 59 2-9 Average real hourly earnings by legal status, em ployer type task and contract length, United States and Florida, 1989-2004...............................................................................59 2-10 Activity duration (days) in the last ye ar: United States and F lorida farm labor markets, preand post 2001............................................................................................... 60 2-11 Total days of employment in the last year by legal status, em pl oyer type and contract type, United States and Florida.......................................................................................... 60 3-1 Explanatory variables for models based on the Na tional Agricultural Workers Survey for 1993-2002.....................................................................................................................85 3-2 Descriptive statistics for explanatory variables ................................................................. 87 3-3 Bivariate probit model estimates for forei gn-born workers legal status and job type selections ............................................................................................................................88 3-4 Marginal effects of bivariate probit selec tion estim ates into lega l status & job type........ 89 3-5 Selectivity corrected wage models for each worker group................................................ 90 3-6 Average predicted conditional wage fo r each leg al status & job type group.................... 92 4-1 Explanatory variables of the choice and param etric wage regression models................. 117 4-2 Summary statistics of foreign farm workforce, NAWS, 1989-2006............................... 119 4-3 Probit model estimates for legal status treatment............................................................ 120
10 4-4 Estimated parameters from parametric wa ge regressions for treated and untreated groups ...............................................................................................................................120 4-5 Treatment effects of legalization..................................................................................... 121 A-1 Beta coefficients and standard erro rs for the outcom e equations estimated by polynomial, nonparametric I and nonparametric II methods........................................... 133
11 LIST OF FIGURES Figure Page 2-1 Temporary foreign worker admissi ons under the H-2A program 1989-2005................... 51 2-2 2002 Florida agricultura l labor expenditures ..................................................................... 51 2-3 Proportion of authorized to unauthorized crop workers in the United States farm labor market, 1989-2004.................................................................................................... 52 2-4 Proportion of authorized to unauthorized crop workers in the Florida farm labor market, 1989-2004............................................................................................................. 52 2-5 Legal status by task at ti m e of interview over time, United States farm labor market, 1989-1998..........................................................................................................................53 2-6 Legal status and task at time of intervie w over time, United States farm labor market, 1999-2001..........................................................................................................................53 2-7 Legal status and task at time of intervie w over time, United States farm labor market, 2002-2004..........................................................................................................................54 4-1 Frequency of Propensity Score by Legal Status ..............................................................122 4-2 Marginal treatment effect (MTE) of lega lization for foreign fa rm workers (with 95% confidence intervals), parametric method........................................................................ 123 4-3 Marginal treatment effect (MTE) of lega lization for foreign fa rm workers (with 95% confidence intervals), polynomial method....................................................................... 124 4-4 Marginal treatment effect (MTE) of lega lization for foreign fa rm workers (with 95% confidence intervals), nonparametric method I............................................................... 125 4-5 Marginal treatment effect (MTE) of lega lization for foreign fa rm workers (with 95% confidence intervals), nonparametric method II.............................................................. 126
12 LIST OF ABBREVIATIONS ATE Average Treatment Effect ATET Average Treatment Effect on the Treated ATEU Average Treatment Effect on the Untreated FLC(s) Farm Labor Contractor(s) IRCA Immigration Reform and Control Act LAW Legally Authorized Worker MTE(s) Marginal Treatment Effect(s) NAWS National Agricultural Workers Survey RAW Replenishment Agricultural Worker SAW Special Agricultural Worker TE Treatment Effects
13 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THREE ESSAYS ON IMMIGRATION REFO RM, WORKER SELF-SELECTIVITY AND EARNINGS IN THE U.S. FARM LABOR MARKET By Lurleen Michelle Walters May 2008 Chair: Robert D. Emerson Cochair: Carlton G. Davis Major: Food and Resource Economics The purpose of this study is to examine contemporary issues in US farm labor markets and immigration policy via three stated objectiv es. Specifically, the study evaluates how farm labor market outcomes have changed with the increasing presence of foreign workers and in the wake of past immigration policies, it assesses the implications of legal status for unauthorized workers wages and employment, and it evaluate s the potential impact of immigration policy reform for farm workers earnings. The first essay evaluates the historical linkages between U.S. immigration policy and U.S. farm labor markets, and specifically how ma rket outcomes have evolved following previous legislation such as the Immigra tion Reform and Control Act (IRCA). This is accomplished with a review of previous resear ch on immigration policies from 1917 through 1986, and with an evaluation of detailed descriptive statistics on fa rm worker and labor market characteristics from the National Agricultural Workers Survey (NAWS). The descriptive statistics are used to characterize the US and Florida farm labor markets in the post IRCA period. The implications of legal status for unau thorized workers wages and employment are assessed in the second essay. Foreign farm worker s are found to jointly select into U.S. farm
14 employment in an authorized or unauthorized status and into skill ed or unskilled jobs, and these choices are found to have certain earnings imp lications. The essay makes a contribution to the literature by analyzing workers joint selections into authoriz ed and unauthorized status and skilled and unskilled employment in the contex t of a double selectivity framework. Previous studies have dealt with both of th ese issues but separately. The issue of legalization for unauthorized work ers is addressed in the final essay. The analytical approach uses a treatment effects appr oach which casts legalization as a treatment (or policy intervention) under the assumption of heterogeneity. The results show an overall positive impact of legalization on farm worker wage out comes, and with the expected positive sorting on the gains from legal status. The evaluation of im migration policy implications for the farm labor market via the treatment effects framework is a va luable contribution to th e literature since this approach has not been used in the context of farm labor before. Given the current strong nati onal and political interest in immigration reform and attendant issues for the agricultura l sector, the study is a timely cont ribution. It should also be of considerable interest to agricultural economists, particularly those working in areas of labor intensive agriculture where labor issues are prime concerns for growers.
15 CHAPTER 1 INTRODUCTION Overview Throughout much of the last century, the par ticipation of foreign born workers in US agriculture steadily increased. This is particul arly evident in the specialty crop sector where growers are the largest users of hired and contract workers on a per-farm basis (Oliveira et al, 1993), due to the heavy reliance on manual labor for seasonal tasks. Recent statistics from the National Agricultural Workers Survey (NAWS) estimate that foreign born workers comprise about 78% of the crop farm workforce (Carroll et al. 2005). However, the problematic issue with the use of foreign labor in agriculture stems from the lack of legal status for about 53% of the workforce (Carroll et al. 2005). The marked increase in the proportion of una uthorized workers in the agricultural labor market and other low-skilled occupations has renewed national and pol itical interest in immigration reform, leading to intense political debate. Numerous le gislative proposals were introduced in the 109th US Congress, but two, in particular have provoked much debate across America and currently serve to frame U.S. immigration issues. Legisl ative proposal H.R. 4437 (the Border Protection, Antiterrorism, and Ille gal Immigration Control Act of 2005 ) was passed by the House of Representatives in December 2005. It was arguably one of the more restrictive proposals introduced for consideration in Congress. It took a pro-enforcement stance on illegal immigration and employment, advocating criminal penalties for unauthorized immigrants and significant fines for the U.S. employers who may ha ve hired them. It made no modifications for legal immigration nor did it a dvocate legalization for illegal immi grants. In contrast, S. 2611 (the Comprehensive Immigration Reform Act of 2006 ) passed by the U.S. Senate in May 2006 proposed earned legalization for unauthorized immi grants and favored modifications to existing
16 laws on legal immigration. It was not as severe as H.R. 4437 but favored a strict approach to enforcement. S.2611 also contained specific provisions for agriculture under AgJOBS ( Agricultural Job Opportunity, Bene fits and Security Act of 2005 (S. 359/H.R. 884; S.2611 Subtitle B), which sought to improve the admini strative process for the admittance of foreign farm workers, to improve wages and benefits, and to establish a pilot program for earned legalization for qualified unaut horized workers. The 109th Congress concluded without resolution on comprehensive immigration reform and neither bill passed. Similar bills have been introduced in the 110th Congress, both in the Senate ( the Secure Borders, Economic Opportunity and Immigration Reform Act of 2007 S. 1348) and the House of Representatives (the STRIVE1 Act of 2007 H.R.1645), but neither has been passed. Arguing that citizen workers are unwilling to do farm work, farm employers have repeatedly expressed a preference for increased access to immigrant labo r and favor using guest worker programs to offset possible labor shortages during harvest time. Employers have expressed concern about immigration reform that may curtail their access to foreign labor and increase wage rates. Employers fear that the s ubsequent wage increases may lead to significant crop losses in the short-run that may eventually threaten their livelihoods. Producers of laborintensive specialty crops may be especially vulnerable, given the lack of mechanization. On the other hand, farm worker advocates favor legalizat ion, as it is viewed as a means of improving workers economic opportunities. Against this backdrop, the purpose of this dissertation is to examine the issues that are germane to US farm labor markets and immigr ation policy. Given this overall theme, the following questions are relevant and comprise the specific objectives of the dissertation: 1 This is an acronym for Security through Regularized Immigration and a Vibrant Economy
17 How have farm labor market outcomes cha nged with the increasing presence of foreign workers? What are the implications of legal stat us for unauthorized workers wages and employment? What are the implications of immigration reform for the farm labor market? These questions are evaluated through three distinct essays In addressing the first objective, the first essay reviews the historical linkages between U.S. immigration policy and U.S. farm labor markets, and specifically how mark et outcomes have evolved in the wake of past policies. A substantial portion of the chapter focuses on the set of changes in the farm labor market after the 1986 Immigration Reform and Control Act (IRCA) and uses detailed descriptive statistics on farm worker characteristics, employm ent, benefits and compensation to characterize the US and Florida farm labor markets. In keeping with the second obj ective, the second essay makes the argument that foreign born workers jointly select into U.S. farm employ ment in an authorized or unauthorized status and into skilled or unskilled jobs, and that their decisions may ha ve certain earnings implications. Previous studies have dealt with both of thes e issues but separately. The essay makes a contribution to the literature by analyzing work ers joint selections into authorized or unauthorized status and skilled or unskilled employment in th e context of a double selectivity model framework. The final objective is addressed in the third essay which evaluates the effects of proposed legalization for earnings outcomes of unauthorized farm workers. The issue is addressed from a treatment effects (TE) perspective in which le galization is cast as a treatment (or policy intervention) and the earnings outcomes for treated and non-treat ed farm worker groups are noted. The TE framework has not been used in pr evious analyses of immigration policy impacts on farm labor markets.
18 Data The NAWS was established by a m andate of the 1986 IRCA which required the U.S. Secretaries of Labor and Agriculture to m easure the farm labor supply between 1990 and 1993 for possible annual labor shortages.2 The NAWS is an employment-based random survey of the U.S. crop workforce which gathers and analyz es data on demographi c characteristics, employment, wages and working conditions, health, safety and educational, social service and housing issues (US-DOL, 2007). Respondents are hi red workers employed in perishable crops (fruits and vegetables, nursery crops, field crops cash grains, silage and animal fodder), and there are four sampling levels (region, crop reporting district, county and employer). A multistage sampling process is used to overcome the challenges associated with obtaining a nationally representative sample due to seasonality of farm work and mobility of the hired workforce. Workers are surveyed in three cy cles annually, and each cy cle lasts about twelve weeks. Interviews are allocated according to proportional allotment across twelve regions, and the number of interviews per cycle is proportional to the amount of seasonal agricultural service activity at the time of year (Perloff, Lynch and Gabbard, 1998). The data used in this study encompasses the entire survey period (1989-2006) that is currently available. With the exception of the first essay, subsamples of foreign workers (with complete data) are used in the analysis. 2 Had shortages been detected, foreign workers would have been granted authorization for temporary U.S. employment under the Replenishment Agricultural Worker (RAW) program.
19 CHAPTER 2 U.S. IMMIGRATION POLICY AND F ARM LABOR MARKETS Overview U.S. immigration policy has long been intert wined with the labor needs of the U.S. agricultural industry (Levine, 2004) which tend to be driven by th e specialty crop sector. The farm labor force has become increasingly relia nt on foreign workers over time, with most workers originating from Mexico. The most conten tious aspect of foreign workers participation in low-skilled occupations such as agriculture stems from the undocumented status of a large percentage of these workers. Indeed, it creates a conundrum for lawmakers who aim to establish and enforce immigration policy goals but who are cognizant of the potential economic implications. Immigration reform has taken cen ter stage once more in the political arena, and agriculture has drawn scrutiny due to the significant proportion of unauthoriz ed workers in the crop workforce. 1 The participation of unauthori zed workers in agriculture was first formally addressed during the third phase of the Bracero program when tighter rest rictions were placed on employers use of foreign labor.2 The Immigration Reform and Control Act (IRCA) of 1986 was crafted in a similar vein and sought to discour age illegal immigration and employment through several provisions. First, sanctions were put in place to penalize empl oyers who knowingly hired unauthorized workers and to remove the commercial incentives associated with illegal 1 The term is used in the literature to describe foreign na tionals who lack authorization for U.S. employment. They are not citizens or permanent residents and are not categorized under any of th e temporary authorized statuses that would allow longer-term residence or employment (Passel, 2005). It is also used to describe persons who enter the US without proper inspection at the borders, who overstay their visas and/or work in violation of nonimmigrant visas that were lawfully obtain ed (Chiswick, 1988). 2 The program allowed for admission of Mexican workers (1942-1964) for employment on US farms. The term bracero originates from the Spanish term brazo (meaning arm) thus the designation bracero meant one who works with his arms (Briggs, 2004).
20 immigration. Second, unauthorized workers present in the U.S. at the time of IRCAs passage were adjusted to legal status under the Le gally Authorized Worker (LAW) and Special Agricultural Worker (SAW) programs, provide d that they met certain qualifications. Approximately 3 million unauthorized immigrants gain ed legal status in this manner. Third, the Replenishment Agricultural Worker (RAW) program was implemented to facilitate employment of foreign guest workers if needed. 3 This was done to assuage growers concerns of labor shortages if the newly converted SAWs left ag riculture following legalization. However, these shortages never materialized and the program wa s allowed to expire in 1993. Fourth, the H-2 program was separated into the H-2A and H-2B programs in order to create a more efficient process for hiring foreign guest workers for te mporary agricultural (H-2 A) and non-agricultural (H-2B) U.S. employment. Studies documenting IRCAs effect on the national farm labor market suggest that its core objectives have not been achieved. Not only ha s illegal immigration dramatically increased, with most illegal immigrants arriving after 1990 (Passel, 2005 ; Passel and Suro, 2005), but the overall proportion of unauthori zed workers in some low-skilled occupations has increased markedly since IRCA (Passel, 2006; Mines, Ga bbard and Steirman, 1997; Carroll et al. 2005). In the case of agriculture, statistics from th e National Agricultural Workers Survey (NAWS) show that the percentage of unauthorized crop workers has grow n consistently since the mid 1990s. Whereas unauthorized workers comprise d only 7% of the workforce in 1989, by 19941995 and 2001-2002, this proportion had expanded to 32% and 53%, respectively (Mines, Gabbard and Steirman, 1997; Carroll et al. 2005). Declining and lax enforcement of employer 3 A guest worker is typically a foreign low-skilled worker, hired on a temporary basis for seasonal work, usually in agriculture (Bruno, 2006).
21 sanctions and limited enforcement at worksites an d U.S. borders are some of the factors which may have contributed to these trends (T hompson and Martin, 1991; Brownell, 2005). The broad objective of this introductory chapter is to review the historical linkages between U.S. immigration policy and U.S. farm labor markets, with specific focus on how market outcomes have evolved in the wake of pa st policies. The first section of the chapter briefly reviews the various policy measures that were authorized by immigration law from 1917 onwards, and the implications for farm labor. The second section of the chapter provides more direct insight into the market outcomes unde r the 1986 Immigration Re form and Control Act (IRCA) and characterizes the U.S. and Florida farm labor markets based on data from the National Agricultural Workers Survey (NAWS) for 1989 to 2004. The NAWS is a rich nationally representative data set on farm worker s in crops in the U.S, including approximately 2,500 workers each year. It is well-suited for this exercise as it allows for a fairly extensive overview of crop farm worker demographi cs, legal status, employment and wages.4 Much of the discussion in this section centers on the legal status of the workfor ce, particularly in the context of distinctions between authorized and unauthorized workers. The final section of the chapter summarizes the key findings and the main conclusions. U.S. Immigration Policy Review Agricultures linkage with imm igration polic y can be traced to the 1870s when Western growers successfully lobbied the United States Congress for the admission of immigrant workers from Asia and Mexico on the grounds that they we re . needed to establish and maintain a viable agriculture . (Martin, 1990, pg. 71); sin ce then, immigration legislation has explicitly considered the purported labor needs of the agricultural i ndustry. The origin and historical role 4 Data on working conditions, farm work er households, health and health care access, income and assets are also available through the Survey but are excl uded since they are less important to the overall objective of the chapter.
22 of guest worker programs in the United States has been as national emergency programs during military conflicts (Briggs, 2004), whereby guest workers are permitted entry when efforts to obtain U.S. workers have failed (Oliveira, 1989). The first of such programs was authorized by the Ninth Proviso to Section 3 of the 1917 Im migration Act in res ponse to requests from Southwest growers. The order suspended the cont ract labor prohibition, the head tax and literacy test stipulation on aliens and new immigrants to allow for the entry of Mexican farm workers to alleviate labor shortages caused by World War I (Heppel and Papademetriou (1999); Briggs (2004)). The program was deemed a remarkable achievement by the agricultural and railroad sectors since the Act explicitly forbade the entry of foreign workers for employment in the United States (Briggs, 1983). It was discontinued in 1922 after it could no longer be justified as a national defense policy (Briggs, 2004). The Bracero Program The Mexican Labor Program more commonly known as the Bracero program, was established in 1942 following formal negotiations between the U.S. and Mexican governments. Authorized by Public Law 45 (P.L. 45), the objec tive of the labor program was to mitigate labor shortages in U.S. agriculture that were brought about by World War II. Workers were to be employed in the agricultural sector only; they were subject to immedi ate deportation if found working in any other industry (Briggs, 2004). The Bracero Program was implemented in three phases. The first phase (1942-1947) was execu ted during wartime and authorized via the omnibus appropriations bill Public Law 45 which permitted entry of foreign workers for agricultural employment only (B riggs, 2004). The second phase of the program (1948-1951) was regulated by a joint agreement with Mexico and according to the provisions of the 1917 Immigration Act. Growers became more directly involved in r ecruitment and employment after their successful lobbying of the US Congress, which resulted in significant revision to the
23 original agreement. According to Heppel and Papademetriou (1999), intergovernmental contracts were replaced by grower-tobracero contracts wherein farm em ployers were allowed to recruit at the border, after which workers were admitted by the Immigration and Naturalization Service (INS). The Mexican government strong ly protested these arrangements, citing lax enforcement and the potential for breach of contract by U.S. employers (Heppel and Papademetriou, 1999). It is estimated that 65,000 workers were admitted annually on average between 1945 and 1950 for employment on Southwestern farms (Oliveira, 1989; Briggs, 1983) Amendments to the Br acero Program The Bracero program was amended by P.L. 78 in 1951, mainly to assuage concerns over labor shortages caused by the Korean War (O liveira, 1989; Heppel and Papademetriou, 1999). The program was expanded and tighter employment restrictions were put in place to discourage increased illegal immigration from Mexico. U. S. employers were permitted to hire Mexican workers only if domestic (U.S) workers were unav ailable and certain guarantees could be made that their employment would not adversely affect the wages and working conditions of domestic workers. To lessen the likelihood of worker expl oitation, the Mexican gove rnment insisted that workers be guaranteed employment for specific periods according to contract and that they be paid the same wages as domestic workers (Olive ira, 1989). In comparison to the earlier phases, the number of foreign nationals admitted for employment purposes increased dramatically: for example, the number of foreign temporary farm workers grew from 204,000 in 1951, to 460,000 in 1956. It decreased to about 200,000 workers when the program was finally terminated in 1964. It is estimated that Mexican workers co mprised 93% of the farm workforce for the duration of the Bracero Program (Oliveira, 1989). Whether or not the Bracero program was effective in curb ing illegal immigration remains a point of contention. On one hand, the program provisions were somewhat contradictory: for
24 example, the employment restrictions of the th ird phase were undermined in some respects by a 1949 amendment that granted legal status (via bracero contracts) to unauthorized Mexican workers who resided in the U.S (prior to th e amendment). Heppel and Papademetriou (1999) argued that these strategies may have actually increased illegal immigr ation as unauthorized Mexicans figured that the best way to obtai n legal status in the United States (as a bracero) was to migrate illegally to the U.S. On the other ha nd, reports based on apprehensions data show that illegal immigration decreased significantly during the progr am (CRS, 1980; Anderson, 2003), but increased after its termination (A nderson, 2003) as thousands of former braceros continued to migrate to the U.S. for work (Briggs, 2004) Given this overall context, guest worker programs have been sharply criticized as inhere ntly flawed policy instruments that create immigrant networks and ultimat ely facilitate future illega l immigration (Briggs, 1986, 2004; Krikorian, 2001; Martin, 2000). The British West Indies (BWI) Temporary Alien Labor Program Foreign workers were also granted legal entr y to the U.S. via the British West Indies (BW I) Temporary Alien Labor program, which operated concurrently with the Bracero Program. The BWI Labor program was authorized by Public Law 45 and established in April 1943 by joint agreement with governments of the British West Indi es in response to Eastern growers claims of wartime labor shortages (Briggs, 2004). The notable difference between the BWI and Bracero programs was that BWI contract workers were permitted to work in the nonagricultural sector (Heppel and Papademetriou, 1999; Briggs, 2004) During its initial phase (1943-1947), an estimated 19,000 workers per year were admitted for farm work, mostly in Florida (Briggs, 2004). Though seemingly insign ificant relative to the Bracero Program, the influx of workers
25 had a substantial impact on the farm labor mark ets in which the foreign workers were employed (U.S. Senate Committee on the Judiciary, 1978). Between 1947 and 1952, the program was converted to a temporary worker program in which tripartite contracts were established between foreign work ers, their governments and U.S. employers only the U.S. government played no formal role in the process other than to authorize the workers entry to the U.S (H eppel and Papademetriou, 1999; Briggs, 2004). Following the passage of the Immigration and Nationality Act in 1952, contract workers from the BWI were employed through the H-2 program (Heppel and Papademetriou, 1999) The H-2 Program The H-2 program was authorized by secti ons 101(a)(15)(H)(ii) and 214(c) of the 1952 Immigration and Nationality Act and differed from its precursors in the sense that it was a permanent part of immigration law and not geared towards specific labor shortages per se (Oliveira, 1989). Though open to all U.S. employer s, it was most widely used by agricultural employers. Employers were required to pay fo reign and domestic worker s the higher of the adverse effect wage rate (AEW R), the Federal or State minimu m wage, or the prevailing wage rate in the area of employment. 5 They also had to offer id entical benefits, such as free transportation and housing and the minimum wage, to domestic and foreign workers (Heppel and Papademetriou, 1999). The relatively strict hiring requirements of the program caused fewer foreign workers to be admitted for U.S. employ ment, in comparison to the other programs: on average, only 18,000 foreign workers were admitted annually between 1965 and 1987 under the H-2 program, compared to the admission of an annual average of 242,000 workers between 1945 5 The AEWR was established by the U.S. Department of Labor in 1961. It is the wage rate which must be offered and paid, as a minimum, to every H-2A worker and every U.S. worker for a particular occupation and/or area in which an employer employs or seeks to employ an H-2A worker so that the wages of similarly employed U.S. workers will not be adversely affected (U.S. Dept. of Labor, 2006).
26 and 1964 under the Bracero program (Oliveira, 1989). These workers were employed mainly by growers in the Eastern States since growers in the Southwest had an ample of supply of unauthorized workers, who were either former braceros or newcomers (Oliveira, 1989). The Immigration Reform & Control Act of 1986 In the m id 1980s, concern over increased ille gal immigration gave way to legislative reform via the Immigration Reform and Contro l Act (IRCA), P.L. 99-603, which passed in 1986. Lawmakers hoped to reduce illegal immigration a nd discourage the employment of unauthorized labor by U.S. employers through specific policy instruments such as employer sanctions, industry-specific legalization programs and guest worker pr ograms (the RAW and H-2A programs). Based on the findings of previous studies, the specific objectives and impacts of these policy instruments are examin ed in the following sections. Employer sanctions These were intended to rem ove the commercial or employment incentives associated with illegal immigration and were auth orized by Title I, Part (A), S ection 101 of IRCA. According to the law, employers who knowingly hired unautho rized workers could be fined up to $10,000 per unauthorized worker and imprisoned for six mo nths (Library of Congress, 1986; LSC-OIG, 2007). Given the risks and penalties associated with the employment of illegals, it was widely assumed that employers would improve wages a nd working conditions in order to attract and retain legal workers. Employer sanctions were phased in more slowly in agriculture than the rest of the economy, ironically beca use growers successfully made the case that . they had become dependent on unauthorized immigrant workers because the U.S. government had not prohibited hiring such workers in the past (Taylor and Thilmany, 1993, pg. 350). Agricultural employers were thus exempt from sanctions co mpliance and enforcement until December 1988.
27 According to Taylor and Thilmany (1993), em ployer sanctions did little to reduce the inflow of illegal newcomers to the farm labor market during the transi tion period or to reduce worker turnover. Research by Thompson and Martin (1991), Martin (1994) and Brownell (2005) show that sanctions enforcement has waned over time and that enforcement has been lax in the years since IRCAs passage. Inadequate fundi ng, limited personnel, limited intraand interagency coordination, competing priorities among ag encies and an overall inefficient bureaucratic system for ensuring employer compliance have been cited as primary factors in this respect (Thompson and Martin, 1991; Browne ll, 2005). Martin and Taylor (1990) have argued that employers have seemingly made little effort to improve working conditions and wages. According to Taylor and Thilmany (1993), the ag ricultural labor market has continued to be characterized by high turnover and with a large proportion of unaut horized workers. Another factor that may have indirectly affected enforcement is the problem of documented illegals workers with counterfeit work authorization documents in farm labor markets in the U.S. Such documents have prolif erated quite easily since federal law does not allow employers to question the employment elig ibility of workers who present the requisite documentation (Taylor and Thilmany, 1993; Rural Mi gration News, 2000). In short, employers are required to verify legality of status only, and not the authenticity of the documents that are presented as proof of legality. IRCAs enforcement provisions may have en couraged some growers to shift their management responsibilities to farm labor cont ractors (FLCs), arguably to lessen the risk of penalties associated with the employment of unauthorized workers. FLC operations have increased in major specialty crop producing stat es such as California and Florida (Martin and Taylor, 1990; Polopolus and Em erson, 1991; Thilmany, 1996). FLCs tend to be more adept at
28 dealing with the threat of sanctions and recruiting unauthorized immigrants for seasonal farm work than growers, and generally have extensive contacts with farm worker communities and migration networks (Polopolus and Emerson, 1991). Consequently, there exists considerable speculation as to whether growers have used FLCs to stabilize employment or to merely circumvent the law without actua lly reducing their reliance on unauthorized labor (Taylor and Thilmany, 1992). Legalization The Special Agricultural W orker (S AW) program legalized roughly 1.3 million unauthorized crop workers who did at least 90 days of farm work in 1985-86 (Duffield, Morehart and Coltrane, 1989; Thompson and Martin, 1991). The overall goal of legalization was to improve the economic circumstances for unauthorized workers based on the assumption that employers would be encouraged to improve wage s and working conditions to attract and retain legal workers (Martin, 1990; Taylor, 1992). This was a valid assumption since lack of legal status has been shown to affect wages and occupa tional mobility of agricultural workers (Taylor, 1992; Is and Perloff, 1995; Iwai, Emerson and Walters, 2006a). The first of these studies on the agricultura l sector was done by Taylor (1992), who examined the relationship between legal status, wages and workers self -selection into primary (skilled) and secondary (unskilled ) agricultural jobs based on da ta from a 1983 survey of male farm workers in California. Selection bias in to skilled or unskilled employment was accounted for via the Heckman two-stage procedure, and legal status was assumed as having an exogenous influence on wages. His employment results indicated that unauthorized workers were less likely than authorized workers to be observed in primary (skilled) farm jobs and that they were far more likely to be hired for low-skill jobs than were authorized workers. At the very least, this finding suggests that unauthorized status eith er hinders workers mobility into skilled
29 employment, or that it causes workers to ear n lower primary wages which discourages them from seeking skilled employment. The wage re sults indicated that un authorized workers in skilled (unskilled) jobs earned 33 % (5%) less on average than legal workers in skilled (unskilled) jobs. Taylor concluded that their wages and economic mobility would likely improve with legalization but cautioned that th ese benefits would be tempered by an influx of unauthorized newcomers seeking farm employment. Is and Perloff (1995) estimated models for legal status, farm wages and hours of farm employment based on NAWS data for 1989-1991. A multinomial logit model was used to estimate workers selection into native citizen, naturalized citizen, amnest y recipient, permanent resident, and unauthorized status categories. Selection bias into the non-native categories was addressed using Lees extension of Heckmans two stage procedure. On average, authorized workers were found to earn 15% more per hour and per w eek than their unauthorized cohorts. Using NAWS data for 1989-2004, Iwai, Emer son and Walters (2006a) estimated an ordered probit model for self selection into diffe rent legal statuses (u nauthorized, authorized, permanent resident and citizen) and simulated how unauthorized workers wages would change with an adjustment to legal status. Self-selectio n into employment (job) type was not considered. In almost 100% of the scenario s simulated, unauthorized worker s wages increased after legal status was acquired. Unauthorized workers who selected into te mporary authorized status had wage increases between 6% and 31% after 2001. Th e findings in this study compared relatively well with previous work by Taylor ( 1992) and Is and Perloff (1995). Much of the controversy surrounding legali zation under IRCA centered on whether it would cause workers to shun agriculture for ot her occupations. Lawmakers sought to assuage these concerns by establishing the Replenishm ent Agricultural Worker (RAW) program to
30 legally admit foreign guest workers for agricu ltural employment duri ng labor shortages. However, the RAW program was never used as the annual calculations of labor supply and demand made by the U.S. Departments of Agricu lture and Labor for fiscal years 1990 through 1993 found no national farm labor shortages (Lev ine, 2004), and the program was allowed to expire in 1993. Several studies examining the employment e ffects of legalization found that it had no particularly adverse effects on employment duration (Hashida and Perloff, 1996; Emerson and Napasintuwong, 2002; Tran and Perloff, 2002; Iwai, Napasintuwong and Emerson, 2005; Iwai, Emerson and Walters, 2006b). Hashida and Perloff examined the impact of legalization on farm work duration with multinomial logit and dur ation models based on 1989-1991 data from the NAWS. They corrected for selection bias in le gal status using Lees extension of Heckmans two-stage method and used completed employme nt spells. Their findings indicated that expected employment duration increased when wo rkers were employed unde r a legal status. In a different approach based on uncomplet ed spells, Emerson and Napasintuwong (2002) examined employment duration in the context of the number of years reported working in U.S. agriculture as opposed to individual job length. The expected duration of employment was found to be larger for authorized than unauthorized workers. Tr an and Perloff (2002) estimated a Markovian model of migration between agricu lture, nonagricultural work and unemployment based on 1989-1991 NAWS data. Migration rates were conditioned on workers demographic characteristics and legal stat us, and completed employment spells were used. Though job mobility patterns were found to be significantly different by legal stat us, Tran and Perloff determined that newly legalized farm worker s did not leave agricultu re as predicted, and
31 concluded that IRCAs legalizat ion provision seemed to increase the long-run probability that legalized workers would rema in in agriculture. Similarly, Iwai, Napasintuwong and Emers on (2005) and Iwai, Emerson and Walters (2006b) found that there was a gr eater likelihood of longer empl oyment duration once workers obtained legal status. The 2005 st udy utilized an ordered probit model with selection on legal status, and a duration model based on completed employment spells. Unauthorized workers employment duration was found to be no shorter th an authorized workers employment duration, particularly if the workers were converted to authorized or pe rmanent resident status. The 2006 study utilized a similar approach to Tran a nd Perloff (2002) but extended the Markov chain model by accounting for selection bias on legal st atus. The likelihood of workers remaining in agriculture post-IRCA was shown to incr ease modestly with legalization. The H-2A temporary worker program A revision of the H-2 temporary worker prog ram was also m andated by IRCA. Separate H-2A and H-2B programs were created to le gally admit foreign workers for temporary agricultural and non-agri cultural employment provided that qualified U.S. workers are unavailable. 6 Title III, part (A), section 301(c) of the law states that employers must be certified through the U.S. Department of Labor (DOL) befo re they are allowed to hire workers from overseas, so as to (1) certify that U.S. workers are unavailable and that (2) the employment of foreign workers will not adversely affect the wa ges or working conditions of similarly employed U.S. workers (Library of Congress, 1986; LS C-OIG, 2007). The Department of State (DOS) issues the H-2A visas and temporary worker admissions are recorded by the Department of Homeland Security (DHS). 6 Visas are typically issued for a period of up to one year in duration, but ex tensions totaling three consecutive years may be granted (Wasem and Collver, 2001).
32 As shown by Figure 2-1 however, the H-2A program has yet to be used on a large scale. Following an initial admission of a little over 30,000 workers in 1989, admissions fell by nearly 40% the following year and trended steadily downward for much of the early to mid 1990s, supposedly as a result of mechanization of the Florida sugarcane harvest and an overall lack of demand for H-2A workers in other sectors (Rural Migration Ne ws, 2001). Admissions rebounded dramatically between 1998 and 2000, with an all-time high of 33,982 workers in 2000. Most of these workers were employed on tobacco, vegetable and peach farms in North Carolina and Georgia (Rural Migr ation News, 2001). The erratic pattern in the admissions data between 2003 and 2005 may be reflecting a simila r usage pattern by growers. Growers have often cited the programs cost and cumbersome pr ocedures as the main reasons for their limited usage of H-2A workers (Effland and R unyan, 1998; Bruno, 2006). No doubt, the ready availability of unauthorized workers may have also been a factor in this respect. Specialty Crop Agriculture and Farm Labor Specialty crop agriculture invo lves the production of non-program crops such as citrus, vegetables, fruits, nuts, berries, tobacco and hor ticultural and greenhouse commodities. It is the largest sector by value of the U.S. agricultur al sector with approximate 2002 farm-gate and export values of $58.7 billion and $9.3 billion, resp ectively (The California Institute for Federal Policy Research, 2003). Califor nia and Florida are the top pr oducers in the nation. Crop cash receipts comprised more than 70% of the total cash receipts from commodity marketings in 2003, totaling $27.8 billion and $6.4 billion for Ca lifornia and Florida, respectively (USDANASS, 2002). Floridas 2003 crop cash marketings were valued at roughly $5.2 billion, with greenhouse and nursery products (24.8%), orange s (15.3%), sugar cane (8.7%), and tomatoes (8%) as the principal commodities. Florida also ranked second nati onally in 2003 vegetable receipts, and accounted for 8.4% and 21.8% of tota l receipts at the national and state levels,
33 respectively. It ranked third in receipts for fruits and nuts, accounting for 11.1% and 22.5% of total receipts at the national and state levels, respectively. Though Floridas c ontribution (9.8%) to national strawberry cash receipts was signifi cantly less than Californias (84.7%) contribution, it ranked second nationally. Strawberry receipts comprised 2% of the total for all commodities marketed at the state level (ERS, 2005). Markets for fresh produce and horticultural prod ucts have expanded rapidly over the last twenty years, fueled in part by increased cons umer demand and technological developments in transportation and storage facili ties (Rural Migration News, 2000). Strong market growth has increased crop acreage and farm labor demand, part icularly in Californi a and Florida. An estimated 20-25% of vegetable acreage and 40-45 % of fruit tree acreag e is harvested by hand (Sarig, Thompson and Brown, 2000) and large amounts of labor are typically required on seasonal basis. Labor needs are particularly cr ucial during the harvest periods such that the produce may be quickly harvested to preserve quality. This labor intensity causes labor expenditure in the specialty crop s ector to exceed labor expenditures in other agricultural sectors, and more so for the agricultural sector as a w hole. In 2002 for example, labor expenditures in the fruit, vegetable, and horticultural sectors comprise d 37% of total produc tion expenditures in comparison to 13% for the agricultural sector as a whole (USDA-NASS, 2002). Californias total labor expenditures were 29% of total fa rm production expenses whereas Floridas were 33% nearly triple the percentage for the entire U.S. agricultural sector (USDA-NASS, 2002). Floridas hired and contract labor expend itures were 24.4% and 8.9%, respectively, of overall 2002 production expenditure in comparis on to 10.7% and 2%, respectively, for the entire U.S. agricultural sector. On the basis of farm type, labor expenditures comprised 40% of all production expenses reported by Fl oridas vegetable and melon farms, and 38% for fruit and tree
34 nuts farms. Figure 2-2 illustrates the contribution of each farm type to Floridas total agricultural labor expenditure for 2002. Forty six percent of the overall labor expense was attributed to vegetable, melon, fruit and tree nuts farms co mbined. Greenhouse, nursery and floriculture farms accounted for 36% of labor expenditure, an increase of 19% over 1978 expenditures in this category (USDA-NASS, 1978; 2002). The specialty crop sector has increasingly faced challenges from foreign competition, environmental regulations, consumer concerns about health and product quality, and rising production costs in recent years (USDA, 2007). At present, there is growing interest in applications for the fresh mark et that would reduce costs, in crease on-farm productivity, and reduce the strenuous labor requireme nts and reliance on foreign labor.7 In this respect, the Bush Administration has recommended the establishment of a Specialty Crop Research Initiative in the 2007 Farm Bill, which would be supporte d by $100 million in annual mandatory funding (USDA, 2007). The large supply of farm workers, many of wh om are unauthorized, is widely viewed as a factor that stymied technological advancement in U.S. agriculture over the last several years (Sarig, Thompson and Brown, 2000; Krikorian, 2001 and 2004; Martin; 2001). Napasintuwong (2004) and Napasintuwong and Em erson (2004) by way of an induced innovation approach analyzed the impacts of cha nges in immigration policies and labor markets on the rate and direction of technological ch ange in Florida and the U.S. between 1960 and 1999. Comparing technological changes before and after IRCA, th ey determined that technology had been laborsaving prior to IRCA but had become labor neut ral after IRCA due to the large labor supply. 7 Suitable applications for the fresh market have been slow to develop due to lack of public funding for mechanization research. Sarig, Thompson and Brown (2 000) attribute this to the policy stance taken by Bob Bergland, former U.S. Secretar y of Agriculture in the Carter Administration, who argued against public funding of mechanization R&D that would result in displacement of farm labor.
35 These results imply that immigration policies that restrict employers access to foreign labor may cause mechanization to increase in agricu lture (Napasintuwong, 2004; Napasintuwong and Emerson, 2004). The Immigration Reform and Control Act and Farm Labor Market Outcomes The aim of IRCA was to control illegal im migration and employment, primarily through employer sanctions and legalization programs. In order to gauge the success of the legislation, it is instructive to assess the historical data on la bor market outcomes. In this respect, summary statistics based on data from the National Agri cultural Worker Survey (NAWS) for 1989-2004 is evaluated. Toward the end of the chapter, these are discussed and compared in the context of IRCAs stated objectives. The findings summarized in the following sec tions are based on the responses of about 42,000 crop workers in the U.S. and 5,000 crop worker s in Florida who were interviewed for the NAWS between 1989 and 2004. The findings are or ganized according to workforce and labor market characteristics, compensation and employme nt patterns, with legal status used as a common frame of reference. Florida is singled out for comparison as a major specialty crop producer with high labor intensity re lative to the rest of the United States. Florida also has a large immigrant workforce much of which is unauthori zed for US employment. California is similar in these respects but will not be evaluated in this chapter. US and Florida Farm Labo r Market Outcomes, 1989-2004 Workforce Characteristics Demographics At the national level, crop far m worker s hired between 1989 and 2004 were mostly foreign born (73%), Hispanic (80%) and male ( 78%). Only a quarter (25.4%) of the workforce had been born in the United States. Most work ers were Mexican (68.2%), and to a lesser extent,
36 of other nationalities: Central America (2.88%) and Puerto Rico, South America, the Caribbean and Pacific islands, Asia and Southeast Asia combined (<4%). In comparison, Florida had a larger proportion of foreign born workers (82%), most of whom were Hi spanic (86%) and male (74%). Participation of citizen workers in the Florida labor mark et was significantly less than at the national level (
37 another primary language.8 In Florida, 79% of the work force identified Spanish as a primary language whereas less than one tenth identified English as their primary language. Approximately 12% of all worker s identified with another lan guage. These characteristics carried over to workers command of the English language. At th e national level, fewer than a quarter (22%) indicated that they could speak Eng lish well; the vast majority either could not speak it at all (41%), or could speak only a little (29%) or could speak it somewhat (8%). The trends were more pronounced in Florida: only 19% indicated that they had decent English speaking ability, 33% indicat ed that they spoke only a little and 47% could not speak English at all. 9 In sum, these findings suggest that most of these workers would be constrained in their ability to participate in the mainstream economy (Emerson, 2000). Immigrant workers in agriculture are known to be low-skilled and this view was reinforced by the data on both labor markets. The average worker interviewed between 1989 and 2004 had completed only up to the 6th grade in formal education. The outliers in this respect were citizen workers who had completed up to the 10th grade and Central American workers who had completed up to the 4th grade only. In addition, few workers had undergone formal training since migrating to the U.S. Prior to th e time of the interview, only 30% of U.S. crop workers and 28% of Florida crop workers had taken English/ESL, literacy, citizenship, job training, GED/High School Equivale ncy, Migrant Education, Adult Basic education classes or college/university courses. 8 In addition to Spanish and English, the NAWS allows for selection on the following languages: French, Creole, Laotian, Hmong, Vietnamese, Cambodian, Tagalog/Ilocano, Mixtec, Kanjobal (Codebook for NAWS data, 2005). These were grouped as other during the analysis to avoid cell sizes of less than 50 observations. 9 Workers who spoke English somewhat and well were grouped into one category to avoid cell sizes of less than 50 observations.
38 Legal status Legal status is self-reported in the N AWS During the interview, workers are asked whether or not they are citizens and if not, what form of work authorization, if any, they have. Workers are categorized as unauthorized if th ey lack employment authorization, as having other work authorization if they have some form of temporary authorization for U.S. employment, as permanent residents if they have gr een cards, or as citizens if they were born in the US or naturalized. Between 1989 and 2004, roughly 42% of crop worker s at the national level self-identified as unauthorized for U.S. employment. Of thos e workers who reported being in a legal status, citizens comprised 29%, and permanent residents and other authorized workers comprised 22% and 7%, respectively, of the workforce. In Florid a, 53% of the crop workforce was unauthorized. Of those workers who reported being in a legal status, 20% were citizens, 18% were permanent residents and 9% were otherwise authorized. For the remainder of this chapter, workers are broadly categorized as author ized if they reported bei ng in a legal status. Figures 2-3 and 2-4 chart the proportion of au thorized and unauthorized crop workers in the U.S. and Florida farm labor markets, resp ectively, for specific periods between 1989 and 2004. The impact of IRCA in the short term is reflected in the early periods (1989-1992) where a majority of the workforce was authorized fo r employment due to legalization under the SAW program. Past this point however, the data sugg est that IRCAs enforcement provisions were ineffectual with regard to illegal immigration and employment. The unauthorized portion of the workforce grew dramatically after 1992, comprisi ng as much as 55% of the workforce between 1999 and 2001. The decline in the number of unaut horized workers that o ccurred in the later period (2002-2004) was likely in response to the he ightened enforcement measures that were set in place in the months following the 2001 terrorist attacks. In the case of Florida however, the
39 summary statistics on Florida reve al a different pattern, indicati ng growth of about 10% in the unauthorized worker population between 1999-2001 and 2002-2004. A comparison of crop workers characteristics by legal status for the U.S. and Florida between 1989 and 2004 is given in Table 2-1. Relative to the US, the participation of unauthorized workers in the Florida labor market is more pronounced across all categories. Male workers were more frequently observed in una uthorized status than female workers. Most of the unauthorized workers were Mexican. With respect to adult edu cation training undertaken by the workforce, the summary statistics indicate that authorized workers were better able to take advantage of these opportunities th an unauthorized workers. Labor Market Characteristics Employer type and work experience Farm workers may be either directly hi red by growers or employed by farm labor contractors (FLCs). Studies by Martin and Taylor (1990), Polopolus and Emerson (1991), Taylor and Thilmany (1992), and Thilmany (1996) chart the emergence of FLC operations since IRCA and discuss their role in farm labor market s. Some of the more interesting points raised are that FLCs may have facilitated growers ci rcumvention of immigration laws (allowing for continued employment of unauthor ized workers with reduced risk of penalty) and that workers employed with FLCs are paid lower wages. At the national level, growers directly hired 80% of the workers who were interviewed between 1989 and 2004. Although FLCs in Florida di d slightly more hiring than at the national level (25%), the trend was sim ilar in that the bulk of the hi ring was done by growers (75%). Table 2-2 examines these trends further in the co ntext of legal status. An immediate observation is that the Florida labor force had a higher per centage of unauthorized workers than existed for the US labor market, and among all employer types. In both markets, authorized workers had
40 more farm work experience. In Florida, the vast majority of workers with less than a year of experience were unauthorized. FLCs generally hired larger proportions of unauthorized to authorized workers than growers. Workforce composition trends over time for both markets are shown in Table 2-3 and Table 2-4, respectively. On average, growers hi red a larger proportion of authorized workers than FLCs, although both employer types hired larg er proportions of unauthorized workers each consecutive period except the last. This pattern is more evident for Florida, and although a direct comparison cannot be made for the final period, it is worth noting that unauthorized workers comprised well over 50% of the workforce between 1998 and 2004. 10 However, an important caveat is that these findings do not suggest that employers kn owingly hired unauthorized workers. The proliferation of fraudulent work authorization documents is a well known problem in farm labor markets with immigrant work ers (Taylor and Thilma ny, 1993); Rural Migration, News, 2000) and by law, employers are not required to authenticate documents that are presented as proof of legality. Task at the time of interview It is instruc tive to examine the various tasks held by farm workers at the time of the NAWS interview, given that remunerative premia are associated with increasing skill levels and workers with authorized status (Taylor, 1992). Th e NAWS lists six task categories: pre-harvest, harvest, post-harvest, semi-skilled, supervisory and other.11 The data reveal that most workers were employed in unskilled tasks. Between 1989 and 2004, 19% of all crop workers at the national level were engaged in pre-harvest activ ities, 35% and 12% performed harvest and post 10 The difference in time periods arises from grouping to avoid small cell sizes in the Florida sample. 11 For convenience, the first three tasks may be considered unskilled and the last three skilled.
41 harvest work, respectively. Semi-skilled workers comprised almost 21% of the workforce and less than one% of the workfor ce (0.32%) had supervisory roles. The remaining 13% of the workforce were assigned to other miscellaneous du ties. In Florida, a larger proportion of the workforce did harvest work (43%), whereas 20 %, 8.9% and 7.6%, respectively, did pre-harvest, post-harvest and semi-skilled work. Compared with the national level data, more workers (28.4%) were assigned supervisory and other mis cellaneous duties combined. The proportion of workers by legal status employed in each task at the time of interview between 1989 and 2004 for both markets is shown in Table 2-5. In both labor markets, th e most obvious characteristic is that unauthorized workers consti tuted significantly larger propor tions of the pre-harvest and harvest work crews than the other task categories. This undersco res the vulnerability of the crop sector to changes in immigrati on policy that would be stricter on legal status requirements for employment and enforcement. Seasonal employment Much of specialty crop agricu lture, such as tree crop and vegetable production is characterized by large labor re quirements over short time spans. Nursery and greenhouse production also utilizes manual labor, but more so on a yearly basis (Emerson, 2007). Most crop workers between 1989 and 2004 were hired on a seasonal basis (74% and 61% in the U.S. and Florida, respectively), with little over 50% reporting between f our and twenty years of U.S. farm work experience. For the US as a whole, approximately 56% of all seasonal workers were authorized for employment in the United States compared to 43% of seasonal workers in Florida. In both markets, the seasonal workforce was pr edominately Hispanic (85% in U.S; 90% in Florida) and Mexican-born (72% in U.S; 67% in Florida). Cons istent with the hiring patterns reported in previous sections, most seasonal workers were directly hired by growers (76% in US; 65% in Florida). Florida FLCs did more hi ring (35%) compared to US FLCs (24%).
42 Figures 2-5 through 2-7 illustrate the cha nges in the composition of the U.S. farm workforce across different time periods by legal st atus and task at the time of the interview.12 An interesting observation is that unauthorized workers have increasingly been hired for lowskilled tasks (pre-harvest and harvest) over time relative to the other ta sk types (post-harvest, semi-skilled and other) where authorized workers ha ve comprised the majority of the workforce. Whether this reflects deliberate action on the part of employers is debatable, and has long been an argument advanced by farm labor advocates: that unauthorized status relegates unauthorized workers to unskilled jobs and limits their econom ic opportunities. A comparison of Figures 2-6 and 2-7 indicates how the proportion of unauthorized workers in pre-harvest and harvest work grew from 1999 onwards exceeding 50% in ei ther case. Although larger proportions of authorized workers were hired in the post-harvest and semi-skill ed categories, the proportion of unauthorized workers employed in those task cate gories increased significantly as well coming within 6 percentage points or less of the authorized proportion over the 1999 and 2001 period. It was only during the 2002-2004 peri od that the percentage of unauthorized workers in all task categories declined. Compensation by Task, Employer Type, Cont ract Length and Legal Status Tables 2-6 and 2-7 document the compensation methods that were used in the US and Florida labor markets. Most workers were pa id an hourly wage and salary and combination methods were not used to a large extent. The ex ception to this trend appears for harvest workers in both labor markets, where ove r 40% of the harvest workers were paid by piece rate particularly in Florida. Tabl e 2-8 shows employers preferences for the hourly rate over other methods. The sole exception in th is case is among FLCs in Florida, who paid more than half of 12 Only five of the six task categories are represented; the supervisory category is omitted as it has too few observations to be meaningful.
43 the workforce by piece rate. Both employer types at the national level used an hourly wage to compensate more than 60% of their work crews. In Florida, over 70% of directly hired workers were paid an hourly wage whereas only 39% of labor-contracted workers were compensated via this method. Table 2-9 shows the average re al hourly earnings for the U.S. and Florida by employer type and legal status, task, a nd contract length for 1989-2004. US farm employers paid wages that generally exceeded those paid by employers in Florida; this was the case across all categories. That legal status matters for rem uneration is evident with a comparison of workers by legal status for both markets and employer type s, where authorized workers wages are shown to exceed those of unauthorized workers. In th e remaining categories that show average wages by task and contract type, workers involved in pre-, harvest and post-ha rvest activities were generally paid lower wages, than semi skilled workers. Seasonal workers also reported lower wage rates relative to year-round workers. Employment Patterns Duration of labor force activity Prior to IRCAs passage, agricultural em ploye rs expressed significant concern regarding the potential adverse effects of immigration reform, as it was wide ly assumed that farm workers with legal status would shun ag riculture for other occupations resulting in significant labor shortages and higher wages. Previous work by Hashida and Perloff (1996), Emerson and Napasintuwong (2002), Tran and Perloff (2002), Iwai, Napasint uwong and Emerson (2005), and Iwai, Emerson and Walters (2006b) provide economic rationale for legaliza tion of unauthorized farm workers and show its positive influence on employment duration. The remaining sections of this study analyze labor for ce activity across legal status and employer type, focusing on how activity patterns have shifted over time.
44 Labor force activity is organized by farm and non-farm employment spells, unemployment spells and time spent abroad for fore ign born workers by legal status. In general, authorized workers had longer empl oyment spells than unauthorized workers. At the national level, they spent 15 years on average in the Un ited States, with roughly 13 years of farm employment and 1.5 years of nonf arm employment. In contrast unauthorized workers spent considerably less time in the U.S. and had fewer years of employment: the average U.S. stay was 4.8 years, with about 4.3 years of farm employme nt and one year of nonfarm employment. The average authorized worker in Florida spent a bout 12.67 years in the U.S, had 10 years of farm employment and slightly more than a year of nonfarm employment. In contrast, the average unauthorized worker spent 4.6 years in the U.S, had 4.3 years of farm employment and less than a year of nonfarm employment. Legal status notwithstanding, th e data show certain interes ting patterns in foreign-born workers labor force activities over the sample period. In general, they have opted for longer stays in the United States and have increased (d ecreased) their tenure in farm work (non-farm work), respectively. Prior to 2002, unauthorized workers typically spent fewer than 5 years on average in the United States. After this period (2002 -2004), their average US stays increased to more than 5 years on average, and they reported longer farm em ployment spells. Average farm and nonfarm work spells both increased for unauthor ized workers but only at the national level; in Florida there was a tendency for unauthorized workers to do more farm work but less nonfarm work. Authorized workers in the U.S. labor market had longer farm and nonfarm employment spells, whereas those in Florida reported sh orter spells of nonfarm work. Arguably, the heightened enforcement and the increase in im migration restrictions following the September
45 2001 terrorist event is like ly to have brought about these cha nges particularly in the case of unauthorized workers. Summary statistics were also generated to determine the number of consecutive days in an average spell of activity.13 Over the sample period, workers who reported non-farm work generally had longer non-far m work spells, with a difference of up to 35 days for both authorized and unauthorized workers. The average period of unemployment for unauthorized workers was markedly less than that for authorized workers, a difference of 33.5 days on average. Unauthorized workers with time spent abroad reported almost 3.7 months abroad on average, roughly 26 days more than authorized workers who also reported spending time overseas. The spells reported for workers in the Florida farm labor market were comparable with the US, except that farm work spells exceeded the nation al average by 60 days or more. Authorized workers in Florida also had shorter unemployment spells than their national cohorts, but there was virtually no difference between average unemployment spells for unauthorized workers in either labor market over the sample period. Post 9/11 labor force activity The stricter immigration rules that ensued immed iately fo llowing 9/11 created cause for concern in farm labor markets, given the sizeable percentage of unauthorized workers. To gain some insight into how this may have affected wo rker behavior, work spells were generated for the preand post 9/11 periods (Tab le 2-10). Except for unemploym ent spells, the trends in the two labor markets tend to counter each other. At the nationa l level, workers with farm employment reported longer farm work durat ion post 9/11, those with nonfarm employment reported fewer days post 9/11, and those with time spent abroad reported longer spells post 9/11. 13 A spell in the NAWS data is a continuous period of activity with the same employer and task (if employed). Averages pertain only to those individuals who participated in each activity.
46 The opposite occurred in the Florida farm labor market. The more significant changes for Florida were in terms of non-farm employment and time spent abroad. For workers with nonfarm employment, their nonfarm employment spells lengthened by 3 weeks on average, whereas those workers who reported time spent abroad curtailed thei r stays by 55 days on average. On the latter, workers may have been trying to avoid increased scrutiny from immigration authorities, which is not surprising given the signif icant presence of unauthorized workers in the Florida labor market. Duration of labor force ac tiv ity by employer type Workers with nonfarm employment who were employed primarily with FLCs generally had more consecutive days of nonfarm employment relative to farm em ployment. In contrast, workers employed with growers had more cons ecutive days of farm employment though they also reported longer unemployment sp ells. In general, workers with authorized status were able to secure more consecutive days of farm employment with both t ypes of employers, though average tenure was longer with grower s, and particularly in Florida. Table 2-11 shows employment duration (day counts ) in the last year prior to the interview by legal status, employer type a nd contract type for workers who were involved in farm and nonfarm work. The extent of Florid as labor intensity relative to the US is reflected in farm work employment duration, which does not vary dram atically by employer type. Only among those workers with nonfarm work are some differences apparent. On average, US workers reported more days of nonfarm employment if they were authorized and employed with growers; the direct opposite occurred in Flor ida where workers reported more days of nonfarm employment if they were unauthorized and employed with FLCs In general, seasona l workers also reported
47 more days of nonfarm employment relative to their cohorts employe d on year-round contracts.14 That seasonal workers would have more days of nonfarm employment is not surprising; in their case, farm employment is contingent about tree crop and vegeta ble production that is highly seasonal and with labor demand highest during ha rvest. Year-round workers, on the other hand, are mostly employed with greenhouse and nursery ope rations that are far less subject to seasonal shifts in labor demand. Concluding Remarks The purpose of this chapter was to provide a historical context of the link between U.S. imm igration policy and U.S. farm labor markets, focusing specifically on how market outcomes have evolved in the wake of past policies such as IRCA. The first section of the chapter summarized the implications of past policies (1917-1986) based on previous studies, whereas the second section focused more on farm labor market outcomes as revealed by the NAWS data for 1989-2004. As IRCA was passed in 1986, the NAWS dataset is useful for characterizing the market since its passage. Based on the preceding findings, what infere nces can be drawn as per the outcomes following the 1986 IRCA? Recalling that IRCAs primary objectives were to curtail illegal immigration and employment, the summary statistics from the NAWS data suggest that the legislation has not been effective in these respects. Lax enforcement, particularly in the area of sanctions enforcement, has clearly undermined th e intended efficacy of the legislation. Over time, employers have hired more unauthorized worker s. It is difficult to establish whether their actions have been deliberate given that they are not required to authenticate work authorization documents presented by workers, and further, the proliferation of false documents among 14 Estimates are unavailable for the Florida labor market.
48 workers is a known problem. These observations s eem to point to flaws in the legislation itself that future reform legislation would have to address in tandem with increased enforcement efforts in order to effectively discourage employment of unauthorized workers. An important finding was that the farm workforce is predominately foreign born and unauthorized, particularly in key areas of producti on such as harvesting. The farm labor market is driven primarily by the specialty crop sector and has increased concomitantly with expansion of the sector. Technology in agriculture has be come labor neutral as opposed to labor saving given the ready availability of immigrant labor implying that the technologies did not cause employers to shift away from labor (Napas intuwong and Emerson, 2004). Also, lack of mechanical applications for the fresh market undoubt edly plays a role in this respect. The data also indicated that a vast majority of the workers were low-skilled, had poor English speaking skills and if unauthorized, they we re paid lower farm wages on average than their authorized cohorts. According to the Pew Hispanic Center (2005; 2006) a nd Lewis (2007), many unauthorized workers readily seek employment in other low skilled occupations besides agriculture. 15 Coupled with the preceding observa tions, it would seem to suggest that employers reliance on this type of labor is prob lematic. Further, if enforcement efforts were increased through stricter legisla tion, there may be significant ra mifications for the specialty crop sector on various levels. As discussed in the first section of the chap ter, the emergence of FLCs in markets with high immigrant worker populations has often led to speculation th at they act as a convenient medium for growers circumvention of immigr ation laws. Although FLCs do considerably less hiring than growers, the summary statistics indicate that more unauthorized workers are 15 These workers are mostly Mexican immigrants.
49 employed with FLCs, average hourly wages are lower for workers employed with FLCs, and that workers have fewer consecutive days of farm em ployment if employed with FLCs. While these findings are not proof that FLCs permit growers to circumvent immigration laws, they offer some insight about the stability of farm employment under FLCs. The studies reviewed at the beginning of the chapter sugge st that IRCAs legalization provisions did not adversely affect farm empl oyment duration in f act it increased once unauthorized workers had been adjusted to lega l status. This was reflected in the summary statistics on duration which indica ted that authorized workers had more consecutive days of farm employment on average than unauthorized worker s. That authorized workers earned higher average hourly wages and tended to be employed in task categories other than harvesting further points to the overall significance of legal st atus for workers economic opportunities. Temporary guest worker programs have traditionally been used to alleviate labor shortages during national emergencie s (for example, the Bracero and British West Indies Labor programs). Studies have shown that the H-2A program that was authorized under IRCA has not been utilized on a large scale, which is not surpri sing given the ready avai lability of unauthorized immigrant labor. Growers have cited the cost and cumbersome nature of the applications process as reasons for lack of usage. The debate on immigration is usually c ouched in terms of whether citizens are disadvantaged (or not) by the pr esence of skilled or unskilled immigrants. Although addressing this issue specifically is beyond the scope of this paper, it is somewhat difficult to make the case that they are disadvantaged by the increasing presence of immigrant farm workers. The differences in skills and legal status between the average citizen and immigrant farm worker suggest that they are not likely to compete for the same types of j obs; further, citizens participate
50 only minimally in the crop farm sector. Lewis (2 007) expressed a similar view, arguing that the adverse distributional consequences of immigr ation were likely small, and that American consumers and businesses would benefit due to the associated multiplier effects. Overall, the findings of this study suggest that immigration polic ies have influenced certain changes in the farm labor market over time, whethe r directly or indirectly Since the NAWS was only established in 1986 by mandate, it is not possibl e to establish patterns in the unauthorized farm worker populations before and after IRCAs passage. However, the fact that the proportion of unauthorized workers increased following IRCA particularly after legalization under the SAW program points to its overall ineffec tiveness with respect to controlling illegal immigration. Further, the fact that unauthorized workers continued to gain employment in agriculture after the law was passe d also suggests signifi cant flaws in its en forcement provisions and how they have been execu ted up to this point. Low-skilled sectors such as agri culture that have employed fore ign labor, much of which is unauthorized, are clearly vulnerable to future ch anges in immigration policy that may restrict employers access to foreign labor. Although reform proposals were introduced in the 109th and 110th Congress, lack of agreement has stymied the re form process thus far. The shortcomings of IRCA as revealed by the findings of this st udy suggest that improvement in enforcement measures would be needed for future legislati on to be more effective in curtailing illegal immigration and employment.
51 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 19891990199119921993199419951996199719981999200020012002200320042005 Fiscal Year (No data available for FY1997)Number of Temporary Workers Figure 2-1. Temporary foreign worker admi ssions under the H-2A program 1989-2005 (Source: US Dept. of Homeland Security, Ye arbook of Immigration Statistics) Other Crop & Livestock Farms 18% Greenhouse, Nursery & Floriculture Farms 36% Vegetable & Melon Farms 18% Fruit & Nut Farms 28% Other Crop & Livestock Farms Vegetable & Melon Farms Fruit & Nut Farms Greenhouse, Nursery & Floriculture Farms Figure 2-2. 2002 Florida agricultu ral labor expenditures (Source: 2002 USDA-NASS, Census of Agriculture)
52 Figure 2-3. Proportion of authori zed to unauthorized crop worker s in the United States farm labor market, 1989-2004 Figure 2-4. Proportion of authori zed to unauthorized crop worker s in the Florida farm labor market, 1989-2004 0 10 20 30 40 50 60 70 80 1989-1992 1993-1995 1996-1998 1999-2001 2002-2004Time PeriodWorkers (%) Unauthorized A uthorized 0 10 20 30 40 50 60 70 80 90 1989-1992 1993-1995 1996-1998 1999-2001 2002-2004Time PeriodWorkers (%) Unauthorized A uthorized
53 1989-1998 0%10%20%30%40%50%60%70%80% Preharvest Harvest Post Harvest Semi-skilled Other Unauthorized Authorized Figure 2-5. Legal status by task at time of interview over time, United States farm labor market, 1989-1998 1999-2001 0%10%20%30%40%50%60%70%80% Preharvest Harvest Post Harvest Semi-skilled Other Unauthorized Authorized Figure 2-6. Legal status and ta sk at time of interview over time, United States farm labor market, 1999-2001
54 2002-2004 0%10%20%30%40%50%60%70%80% Preharvest Harvest Post Harvest Semi-skilled Other Unauthorized Authorized Figure 2-7. Legal status and ta sk at time of interview over time, United States farm labor market, 2002-2004
55 Table 2-1. Crop worker characteristics by lega l status, United States and Florida, 1989-2004 U.S. Crop Workers (%) FL Crop Workers (%) Characteristic Authorized Unauthorized Authorized Unauthorized All Workers 58 42 47 53 Demographic Male 55 45 43 57 Female 72 28 60 40 Single 51 49 43 57 Married 64 36 52 48 Ethnicity Hispanic 49 51 43 57 Non-Hispanic 96 4 N/A N/A Nationality Mexican 43 57 38 62 Other 91 9 61 39 U.S. Adult Education 78 22 63 37
56 Table 2-2. Farm labor market characteristics by legal status, United States and Florida, 19892004 U.S. Crop Workers (%) Florida Crop Workers (%) Characteristic Authorized Unauthorized Authorized Unauthorized All Workers 58 42 47 53 Employer Type Grower 63 38 52 48 Farm Labor Contractor 42 58 32 68 U.S. Farm Work Experience (years) 1 year 23 77 10 90 >1 year & 4 years 44 56 31 69 > 4 years & 10 years 64 36 54 46 > 10 years & 20 years 88 12 84 16 Table 2-3. Workforce composition by legal status and employer t ype across time, United States 1989-1992 1993-1997 1998-2001 2002-2004 % U.S. Workforce Grower FLC Grower FLC Grower FLC Grower FLC All Workers 84 16 81 19 74 26 81 19 Unauthorized 15 36 42 58 50 68 46 63 Authorized 85 64 59 42 50 32 54 38
57 Table 2-4. Workforce compos ition by legal status & employe r type over time, Florida 1989-1992 1993-1997 1998-2004 % FL Workforce Grower FLC Grower FLC Grower FLC All Workers 73 27 79 21 74 26 Unauthorized 32 56 46 65 61 79 Authorized 68 44 55 35 39 21 Table 2-5. Legal status by task at time of interview, Unite d States and Florida, 1989-2004 U.S. Crop Workers (%) Florida Crop Workers (%) Task at Time of Interview Authorized Unauthorized Authorized Unauthorized All Workers 58 42 47 53 Pre-harvest 57 43 50 50 Harvest 51 50 35 65 Post-harvest 71 29 62 38 Semi-skilled 65 35 72 29 Supervisory 94 6 86 14 Other 63 37 59 41
58 Table 2-6. Compensation method by task type, United States, 1989-2004 Percentage of workforce Task at the Time of Interview Hourly Piece RateCombinationSalary TOTAL All Workers 77 1922 100 Pre-harvest 93 502 100 Harvest 53 4241 100 Post-harvest 87 1021 100 Semi-skilled 85 1203 100 Supervisory 80 9110 100 Other 93 214 100 Table 2-7. Compensation method by task type, Florida, 1989-2004 Percentage of workforce Task at the Time of Interview Hourly Piece RateCombinationSalary TOTAL All Workers 65 3311 100 Pre-harvest 92 701 100 Harvest 27 7020 100 Post-harvest 84 1330 100 Semi-skilled 93 601 100 Supervisory 74 1629 100 Other 96 202 100
59 Table 2-8. Compensation method by employer type, United States and Florida, 1989-2004 Workforce Percentage by Employer Type US Employer Florida Employer Compensation Method Grower FLC Grower FLC All Workers 80 20 76 24 Hourly 80 65 73 39 Piece Rate 16 32 25 59 Combination 2 2 1 2 Salary 2 1 1 0 Table 2-9. Average real hourly earnings by legal status, employer type, task and contract length, United States and Florida, 1989-2004 Wage Rate ($) by Employer Type US Grower US FLC FL Grower FL FLC Legal Status Unauthorized 6.17 5.81 Authorized 6.65 7.21 6.42 6.60 6.85 5.91 Task at Time of Interviewa Pre-harvest 6.35 6.33 5.57 Harvest 6.41 6.18 5.89 Post-harvest 6.61 6.35 5.65 Semi-skilled 6.71 7.07 6.56 Other 6.80 6.79 6.89 7.13 7.47 6.63 6.74 5.94 Contract Length Year-round 6.43 6.70 6.00 Seasonal 7.48 6.71 6.42 6.09 5.79 a The supervisory category has too few observations to be meaningful and is therefore omitted.
60 Table 2-10. Activity duration (days) in the last year: United States and Florida farm labor markets, preand post 2001 US Florida Type of Activity Spella Pre-2001 Post 2001 Pre-2001 Post 2001 Farm Employment 64 83 83 78 Non-farm Employment 110 94 94 115 Unemployment 48 39 39 33 Abroad 100 133 133 78 aThe duration averages specified pertai n only to individuals who participated in each activity, respectively. Table 2-11. Total days of employment in the last year by legal stat us, employer type and contract type, United States and Florida U.S. Florida Farm Work Non-Farm Worka Farm Work Non-Farm Workb Legal Status Authorized 149 130 167 111 Unauthorized 144 117 164 118 Employer Type Grower 148 126 167 112 Contractor 147 123 164 123 Contract Type Seasonal 135 129 N/A N/A Year-round 189 117 N/A N/A a,b The non-farm work values reflect the averages only of those individuals who had non-farm work. Consequently, the values for farm work and non-farm work cannot be added together for an estimate of total days employed in any type of work for the year.
61 CHAPTER 3 WORKER SELF-SELECTIVITY AND FARM WAGES Introduction Are unauthorized farm workers more likely to be observed in unskilled farm jobs than authorized farm workers? Further, do they earn less than authorized farm workers for the same type of work? Proponents of legalization may re spond to these questions in the affirmative given that they have long maintained that lack of lega l status limits unauthorized workers labor market opportunities, and that their situ ation would improve with adjust ment to legal status. The significance of legal status for fa rm employment and wages was fi rst analyzed by Taylor (1992), who found that unauthorized farm workers are more likely to be selected into low-skilled, lowpaying farm jobs, and that they tend to earn less th an their authorized cohorts for the same type of work. Taylor argued that there was self-selectivity into primary (skilled) and secondary (unskilled) farm jobs and estimated separate earn ings functions for each job type. Legal status entered the equations as an exogenous influence, argued to affect earnings differently for each job type. This paper is motivated in part by Taylor s findings but utilizes a different modeling approach. A major distinction is that workers selections into legal status categories are assumed to be endogenous as are their sel ections into job type (skilled or unskilled). Workers decisions are modeled as a joint process i nvolving two potential sources of se lection bias legal status and job type. Specifying one of these choices as exog enous, or ignoring it, may result in biased and inconsistent wage parameter estimates if there are two underlying decisions guiding farm workers into various groups for farm work in the United States. It is in this respect that this paper also differs from previous work by Is and Perloff (1995) and Iwai, Emerson and Walters
62 (2006a), which evaluated the farm wage implications of selection into legal status but not job type. The main argument set forth in this paper is that foreign born workers opt to jointly select into U.S. farm employment in an authorized or unauthorized status and into skilled or unskilled jobs, and that these decisions may have certain ear nings implications. As such, there may be two decision criteria and not a singl e criterion function as assumed in previous studies. It is conceivable that foreign born workers would sort themselves in this fashion, arguably because they have considered the net benefits of worki ng in the United States in a particular employment status (legal status-skill combin ation). Specifically, this study examines whether the decisions are interrelated and whether th ey result in earnings differentia ls among groups of foreign born workers. The analysis is based on a sample of foreign born workers who were interviewed between 1993 and 2002 for the National Agricult ural Workers Survey (NAWS). U.S. born workers are excluded from the sample since, as ci tizens, they would not select into legal status for employment in the United States. The paper is organized as follows. Th e second section reviews the theoretical underpinnings of the earnings function as an empiri cal tool in earnings determination. Context for the later sections is provide d with a brief discussion of the Roy (1951) model. The third section discusses the problem of selectivity bias and its implications for earnings. The fourth section describes the econometric model used in the study and gi ves a brief discussion of the data. The study findings are discussed in the fifth section of the paper. The concluding remarks that are given in the final section serve to evaluate the findings in the context of the current immigration debate.
63 Theoretical and Analytical Framework Earnings Functions Earnings functions, as developed by Mincer (1974), are the standard approach for characterizing individuals earni ngs profiles. Individuals hum a n capital stocks are assumed to be homogeneous, and individuals are assumed to undertake investments to maximize their present value earnings. They are also assumed as having the same rate of return to human capital investment, and investing the same fraction of th eir earnings capacities. With these assumptions, log earnings are specified as: uEXP EXPSyln 2 3 210 (3-1) where EXP,S,y and denote earnings, years of schooling and post-schooling (onthe-job training) investments, respectively, and uis a normally distributed homoscedastic error term 0EXP,S|uE The percentage increase in earnin gs generated by an additional year of schooling is given by1 ; it is interpreted as the rate of return to educational investment and is assumed to be the same across all schooling leve ls. The coefficients on the experience terms estimate the growth rate of earnings resulting from an additional year of labor market experience. These post-schooling investments decrease over time relative to potential earnings and as experience increases (Willis, 1986; Polachek a nd Siebert, 1993; Chiswick, 2003). Mincers earnings function emphasizes the life cycle dynami cs of earnings and th e relationship between observed earnings, potential earnings and human cap ital investment for both formal schooling and on-the-job training (Heckma n, Lochner and Todd, 2003). According to Willis (1986), a key assump tion underlying the use of the human capital earnings function is that it accura tely represents the opportunity se t faced by a typical individual. However, this assumption presents two sets of empirical difficulties as it is not possible to
64 observe the earnings of the same individual w ho makes alternative c hoices on human capital investment, nor is it possible to observe all th e characteristics that determine his earnings opportunities (Willis, 1986). This characterizes a se lf selection problem, implying that data used in wage determination would be censored based on the individuals optimal choice. The wage equation estimates would also be bi ased unless the selection bias is explicitly accounted for. The seminal work of Roy (1951) was the first to bring the implications of self selectivity to bear on earnings distributions. The model has since been formalized and applied to different issues of interest in labor economics. These include the effects of unionization on wages (Lee (1978); Abowd and Farber (1982), education (Will is and Rosen (1979) and migration (Sjaastad (1962); Chiswick (1999)), among other topics. In its simplest context, the Roy model looks at the earnings implications of individuals occupational choices; based on their skill endowment and their comparative advantag e in one activity versus anot her, individuals choose the occupation that provides the highest earnings. In this respec t, the pursuit of comparative advantage in a competitive market results in di stributions that reflect less earnings inequality than if individuals were randomly a ssigned (Heckman and Honor, 1990). Sample Selection Bias1 The main concern with sample selection bias is that it results in nonrandom samples of an underlying population. This may arise from th e assortment of individuals of their own choice into specific groups (Heckman, 1979), commonly referre d to as self-selectivity in the literature. Sample selection bias presents certain empi rical problems which, if ignored, may lead to incorrect inferences about the eff ects of explanatory variables on a de pendent variable of interest. The significance of sample selection bias for empirical research was first addressed in research 1 The terms sample selection bias and selectivity bias are used interchangeably throughout the paper, as is the case in the body of literature.
65 on earnings determination and labor supply beha vior of women by Grona u (1974) and Heckman (1974). The decision to work was shown to be nonrandom and as having significant implications for womens earnings based on the differences in characteristics between working and nonworking women. An example of self selectivity in the overall context of this dissertation would be the decision made by foreign born workers to wo rk in the United States in a particular legal status (whether authorized or una uthorized). Since workers are likely to sort themselves into legal status categories that woul d seem most advantageous given their characteristics, each group would exhibit different ch aracteristics. Wage ijwwould therefore be observed only if worker i has legal status j and bias would arise because the lega l status decision is correlated with earnings determination. Based on a random sample of Nobservations, the conventional model for the ith individual takes the form: N,...,1i;xYii i ; (3-2) N,...,1i;vzJii i ; (3-3) 0 if 0 0 if 1* i i i iJJ JJ )'(1)0(Prob );'()1(Prob i i i iz J zJ 0vz iff 0Y 0vz iff YYii i ii ii ; (3-4) where iY is an endogenous variable (l og earnings) to be estimate d in the outcome equation, ix and iz are vectors of exogenous variables, and are vectors of unknown parameters and the error terms i and iv are joint normally distributed with zero mean and variance 2. Inclusion
66 in a particular group is determined by a latent selection rule (Equation 3-3), which has an index functioniJ that determines observability. With the thiindividuals selection into a group, the observations on *iYare censored for the observed range of Yiwhere 0J if 1J* i i The expected value of Yiis determined as: '|' 0'|'0;,|* iii i iii i iiiizvEx vzxEJzxYE (3-5) The conditional mean of i would be nonzero if the error terms i and iv are correlated. Estimation of iY by ordinary least squares over the subsample of n observations would result in biased and inconsistent estimates of unless the selection bias is addressed. The inconsistency arises due to the correlation between ix andi operating through the co rrelation between the two error terms, i andiv Empirical Remedies Sa mple selection bias is usually resolved using maximum likelihood estimation (MLE) or two-step estimation procedures, either of which requires distributional assumptions on the error terms. Heckman (1974; 1979) is credited with the initial parametric formulat ions for both sets of procedures. These have been extended to incl ude semiand non-parametric methods that relax the distributional assumptions; seve ral of these are discussed in detail by Vella (1998). The first empirical solution for sample selec tion bias was the maximum likelihood estimator proposed by Heckman (1974). This form ulation assumes that the error terms iiv have zero mean and are identically and independently no rmally distributed with variance-covariance matrix 2 2 vv v The errors are also assumed to be independent of the explanatory
67 variables in the selection equation (Equati on 3.3) (Heckman, 1974; Vella, 1998). Given the normality assumption, the maximum likelihood estim ates are efficient and consistent. If normality is violated however, consistency will fail. Within the class of two-step estimation pro cedures, the parametric version developed by Heckman (1976; 1979) is the most popular. The errors are assumed to be joint normally distributed, and the misspecificati on created by the censoring patte rn is resolved by including a correction term to account for selectivity. Fo llowing Heckman (1979) and Greene (2000), joint normality of the errors allows for the specification of the conditional mean of iY as follows: i i iii i iii i iiix zvEx vzxEJxYE '|' 0'|'0;|* (3-6) where v v is the correlation between the two errors, and vi vi i'z 'z is the inverse Mills ratio, with and as the density and distribution f unctions of the standard normal distribution, respectively.2 In practice, the first stage of Heckmans two step procedure involves estimation of the parameters of the selection equation over the entire sample N with a probit model. The first stage probit estimates are used to construct inverse Mills ratios that are included as selectivity correction terms in the second stage estimation, which entails ordina ry least squares estimation of iY by over the subsample n : 0J x'Y* i iiii for (3-7) 2 Joint normality of the errors implies vwhere and are independent and 2 v v Using the mean of the truncated normal distribution where0 and manipulating Equation (3-6) yields i ii v v iiv v v i iiix x xJxYE '0;|2 *.
68 The second stage estimation yields estimates th at are consistent estimates of the population parameters. The heteroscedastic error term,i has variance i i2 212 where v iiiiz ;2 The parameters 2 ii and are generated from the two-stage estimation and are used to derive consistent estimates of 2and Except for the case where no sample selection bias exists, that is where0 the standard errors reported from the OLS regression are inappropriate for tests of significance of the estimated coefficients (Heckman, 1979). Fo llowing Greene (1981; 2000), Equation (3.7) may be revised to show an additional s ource of variation, brought about by using an estimate of the inverse Mills ratio as oppos ed to its true value: 0J x'Y* i iiiiii for (3-8) In this case, the appropriate asymptot ic covariance matrix takes the form: 1 2' 1 2XXQX XXX I Var (3-9) where 2 XWVWXQ with Wcontaining vectors of derivatives v j'z In this covariance matrix, *Xis a larger NxK matrix comprising rows ii* ,xx, is a diagonal NxN matrix with i on the diagonal, and V is the asymptotic va riance corresponding to the probit model coefficients generated in the first stage (Greene, 2000). Multiple Selection Rules Models with multiple decision criteria have been used to evaluate the implications of self selectivity on various topics including employment and occupati onal choice, college and student financial aid choices, migration, a nd maternity. In contrast, studies with relevance to agricultural
69 employment and earnings have focused on a single criterion function only (Taylor, 1992; Is and Perloff, 1995; Iwai, Emerson and Walters, 2006a). However, if farm employment or earnings are actually determined by multiple decision rules, sample selection bias will remain unresolved if one of the rules is ignored or assumed to be exogenous. Incorporation of additional criteria into the analytical framework first requires a determination of whether the decision rules are si multaneous or sequential, as it may affect the selectivity correction procedur es employed during estimation (L ee and Maddala, 1985). In the simultaneous case, the latent selection rules are de fined over the entire set of observations. In the sequential case however, the second selection rule is defined only over a subsample of observations for which the first rule is obser ved (Maddala, 1983). A student financial aid scenario may be used to illustrate this: students decide whether to attend college (1st rule), then decide whether to apply fo r financial aid or not (2nd rule). Clearly, a student is ineligible for financial aid if he decides to forego college. As noted in Lee and Maddala (1985), the error term of the first rule would be defined over the en tire population of high school graduates whereas the error term corresponding to the second rule would be define d for the college student subsample only. In this paper, selectivity bias is assumed to arise from two simultaneous decisions and the econometric model employed for the analysis is based on the double selection model proposed by Tunali (1986). As the term implies, a double se lection process involves two rules that may arise from the related decisions of independent parties or from simultaneous or sequential decisions involving one party (Tunal i, 1986). If the rules are corre lated, joint estimation of the decision functions would generate asymptotically more efficien t estimators (Tunali, 1986). The framework is as follows:
70i iiuxy11 1 1 decision status Legal (3-10) i iiuxy22 2 2 decision type (skill) Job (3-11) i333 i3i3uxWln equation Wage (3-12) In this system, x and are explanatory variable s and unknown coefficients, respectively, and 3 is an unknown scale parameter. The joint normal error terms i3i2i1u,u,u have zero mean and covariance matrix 1 1 12313 23 13 Decisions 1 iyand* 2 iyare unobserved, but the outcomes of th e dichotomous variables D1 (indicating whether the farm worker is authorized or unauthorized for U.S. employment) and D2 (indicating whether he is employed for a skilled or unskilled job) are: 0 1 D1 if if 0y 0y* 1 1 0 1 D2 if if 0y 0y* 2 2 (3-13) This censoring process generates four groupsjG 1,...,4j the elements of which are combinations of D1 and D2: 1,1G and 1,0G,0,1G ,0,0G4 3 2 1 Group G1 denotes foreign-born farm workers who are unauthorized & unskilled, and G2, G3 and G4 are foreign-born farm workers who are unauthorized & skilled, authorized & unskilled, and authorized & skilled respectively. Per the NAWS da ta set, pre-harvest, harvest and post-harvest jobs are classified as unskilled positions, whereas semi-skilled, supervisory and jobs in the other categ ory are classified as sk illed. Provided that the four groups are distinct and completely classified, the probability Sj that an individual is assigned to the jth group is: );C,C( )Cu,CuPr( )0y,0yPr()0D,0DPr(S212 2i21i1 i2 i1 21 1 (3-14)
71 );C,C( )Cu,CuPr( )0y,0yPr()1D,0DPr(S212 2i21i1 i2 i1 212 (3-15) );C,C( )Cu,CuPr( )0y,0yPr()0D,1DPr( S212 2i21i1 i2 i1 213 (3-16) );C,C( )Cu,CuPr( )0y,0yPr()1D,1DPr(S212 2i21i1 i2 i1 214 (3-17) In Equations (3-14) through (3-17), 1i11'xC and2i22'xC 2 is the standard bivariate normal distribution function and is the correlation coefficient between u1 and u2. With outcomes observed for all workers in all groups, these probabilities may be estimated using a bivariate probit model with the likelihood function: 4 S 212 3 S 212 2 S 212 1 S 212,C,C.,C,C. ,C,C.,C,C L (3-18) The inverse Mills ra tios corresponding to each group are constructed from the probit estimates and are used for selectiv ity correction in stage two. In Equations (3-19) through (3-22) below, symbols and denote the standard univariate normal density and distribution functions, respectively. (i) For 1Gi (i.e. D1=D2=0) : 122311132 21 13,| CuCuuEi ii 2 12 2 2 21 1 1 1 2 12 1 2 1 111 CC C ; 1 CC C where S CC S CC (3-19) (ii) For 2Gi (i.e. D1=0; D2=1): 222321132 21 13,| CuCuuEi ii
72 S CC S CC2 1 2 22 2 21 21 (3-20) (iii) For 3Gi (i.e. D1=1; D2=0):322331132i21i1i3Cu,Cu|uE 3 12 32 3 2 1 31S CC S CC (3-21) (iv) For 4Gi (i.e. D1=1; D2=1):422341132i21i1i3Cu,Cu|uE 4 12 42 4 21 41S CC S CC (3-22) The mean log earnings function for the ith individual conditioned on the explanatory variables and the joint outcome of se lection is expressed as: ,x|uEx,x|WlnEi3i33i i3i3i3 (3-23) where denotes the joint outcome of th e double selection process. Sele ctivity bias arises if the expected value of the disturbance term is nonzero 0,x|uEi3i3 (Heckman, 1979; Tunali, 1986; Vella, 1998). The equation is estimat ed by ordinary least squares over n observations: 3122111 31223311133 'x 'xWln (3-24) where1331 ,2332 and 12231113i3u Provided that the estimates from the first stage are consistent, this stage pr ovides consistent estimates of the wage equation parameters for each legal status-skill category. The estimated coefficients will be inefficient due to the heteroscedastic natu re of the error term (v) in the regression equatio n. As with the single selection case discussed in the previous secti on, the standard errors obtained from the OLS regression are not therefore a ppropriate for hypothesis testing. Following Tunali (1986), the error term vhas variance: j j1Si|vVar (3-25)
73 Thus, j 2 jSi|vVarSi|WlnVar (3-26) where 4,...,1j, S g 2 1 C Cj 2 23 2 13 2313 2 2j23 2 D 1 D 1j132j2 2 231j1 2 13j (3-27) where gis a bivariate density, jS is the probability that an in dividual is assigned to group j and all other parameters are as defined previously Using the sum of squared residuals for the thiobservation given by Equation (3-28), a consistent estimator of the variance of the regression is derived as shown in Equation (3-29): W N N1T 2 N 1i j 22 N 1i j 2 (3-28) W T N 1 2 (3-29) The asymptotic covariance matrix for th e parameter estimates is expressed as: 1 ** 223113 223113 2 2' 1 XXX Var vVarXXXVar (3-30) where is the parameter vector from th e first stage bivariate probit and 1 and 2 are the gradient vectors for the derivatives of the lambda terms with respect to the parameters of the first stage bivariate probit. Intere sted readers may refer to Tunali (1986) for specifics on the derivations. Data The data consist of 12,851 foreign born work ers with complete data from the National Agricultural Workers Survey (NAWS) for 1993 to 2002. The NAWS is an employment based random survey of the U.S. crop farm workforce wh ich contains an extensive set of variables on workers demographic characteristics, empl oyment, wages and working conditions, health,
74 safety and educational, social service and housing issues (US-DOL, 2007). The Survey is administered nationally three times per year (J anuary, April/May, and October). Table 3-1 defines the explanatory variable s that were used in the biva riate probit and wage equation models. They reflect the demographic characteris tics of the crop farm workforce, as well as certain farm labor market characteristics. Du mmy variables reflecting the location (state) and time period of interview are also included. Determinants of the Legal Status Decision Demographic characteristics th at influence foreign born workers choice of legal status are: nationality (Mexican), gender (female), mar ital status (married), English speaking ability (English), and age. Relevant human capital variables include edu cation and experience, thought to positively affect productivity and wages. Squared terms are included on the age and experience variables to allow for nonlinear effects. Foreign farm work experience and adult education are included as additional indicators of foreign bor n workers human capital and may have similar directions of influence on earnings as the conventional human capital variables. After 2001 is a dummy variable that is used to distinguish betw een the years that precede and follow the September 2001 event. Determinants of the Job Type Decision The job type decision is influenced prim arily by the basic human capital variables defined above and certain labor market vari ables (seasonal worker, foreign farm work experience, years with employer, farm work week s, piece rate, grower, specialty crop), English and female. One would expect education to posi tively affect a workers decision to seek or accept skilled employment. Workers with good English speaking skills are expected to be better able to capitalize on opportun ities as they arise, and are e xpected to seek skilled farm employment in the U.S. more readily than their cohorts with limited Engl ish speaking skills.
75Determinants of Farm Earnings Farm earnings are influenced by the hu man capital characteristics (education, experience), other workforce demographic characte ristics and labor market characteristics as defined previously. A location dum my variable (California) is in cluded to control for workers location at the time of the intervie w. Selectivity correction terms ij are also included in the earnings equations for each group to acc ount for the effect of the decision criteria on earnings. Results and Discussion Descriptive Statistics Table 3-2 reports the means and standard deviations for each group (authorized and skilled, authorized and unskill ed, unauthorized and skilled, unauthorized and unskilled) identified from a sample of 12,851 foreign bor n farm workers who were interviewed between 1993 and 2002. Over this period, more than half of the workforce wa s unauthorized: 37.9% were unauthorized & unskilled workers whereas 17.9% were unauthorized & skilled workers. Within the same time frame, a little over 44% of all foreign born farm workers identified as authorized workers: 25.3% were authorized & unskilled, and the remaining 18.9% were authorized & skilled. The descriptive statistics for the different groups reveal that the average foreign born farm worker employed between 1993 and 2002 was a Mexican male with limited English speaking ability. The worker had completed only 6 years of schooling on average, and had not taken any adult education courses si nce migrating to the U.S. Regardless of job type, author ized workers tended to be older than their unauthorized cohorts. The average authorized worker was about 38 years compared to 29 years for the average unauthorized worker. Workers were employed on a seasonal basis with growers, and
76 were compensated by methods other than the piece rate. They reported having foreign farm work experience and five years or more of U.S. farm work experience. Unauthorized workers (skilled and unskilled) had about five years of U.S. farm work on average in comparison to authorized workers (skilled and unskilled) who had 14 or more years of U.S. farm work experience. Authorized workers also reported more years of employment with their current employers than did unauthorized workers. On average, the workers reported completing 30 weeks of farm work in the year preceding the NAWS interviews. According to Table 3-2, the average worker in terviewed in California was authorized. As noted in Walters, Emerson and Iwai (2007), this result is a conse quence of restricting the sample to foreign born workers. Most authorized workers on the East Coast are native born so that when native born citizen workers are excluded, virtua lly all workers are unauthorized, for example, there tend to be few green card farm workers on th e East Coast. By contrast, on the West Coast most authorized workers were foreign born (green card, naturalized citi zen, or other form of authorization). As such, the excl usion of native born ci tizens on the East Coast results in a very large change in the proportions of authorized and unauthorized wo rkers from what they would be over all types of workers, including native bo rn citizens (Walters, Emerson and Iwai, 2007). Determinants of Legal Status and Job Type Decisions The results of the bivariate prob it model are reported in Table 3-3. The rho coefficient is positive and statistically significant at the 1% sign ificance level. This indicates that the two selection rules are interrelated, in that foreign bo rn workers jointly consid er legal status and job type when seeking U.S. farm employment. That the rho coefficient is positive indicates that foreign born workers who choose to work in the U. S. as authorized workers are also more likely to be observed in a skilled farm jobs. It is likewise suggestive of the opposite case that foreign
77 born workers who opt to work in the U.S. in an unauthorized status ar e more likely to hold unskilled farm jobs. Most of the estimated coefficients are st atistically significant at the 10% significance level or better and have the expected signs. All else being the same, the characteristics associated with increased probability of authorized status are if the worker is female, married, has several years of U.S. farm work experien ce and speaks English well Workers with higher schooling levels and who have ta ken additional training (adult education) since migrating to the U.S also have a higher probability of being authorized. Conversely, workers from Mexico are less likely to be authorized, as are workers who were interv iewed after 2001. Based on the results for the j ob type equation, foreign born workers with U.S and foreign farm work experience have a higher probability of being employed in skilled farm jobs, all else being the same. This was also the case for wo rkers who did farm work the previous year, and who had worked for several years with thei r current employers, and for workers with good English speaking skills. The latter characteristic is logical as it implies that such workers may be better able to capitalize on job oppor tunities, particularly if Englis h proficiency is required. On the other hand, the characteristic s that increase the probability that workers are observed in unskilled jobs are if the workers are female, if th ey are employed in the specialty crop sector and on a seasonal basis, or compensated by piece rate. The statistical insignificance of the education coefficient suggests th at it has practically no effect on the probability that a worker acquire s a skilled farm job, holding all other factors constant. This finding runs counter to human capita l theory, from which edu cation is expected to improve a workers skill set and, l ogically, the probability that he is employed in a skilled job. It also contrasts with that reported by Taylor (1992), where the edu cation coefficient was
78 significantly positive and shown to increase th e probability of selection into skilled employment.1 The marginal effects corresponding to the biva riate probit model are reported in Table 35. Holding all other characteris tics constant, the marginal effect indicates the change in the observed outcome given a change in th e independent variable of interest.2 For each group, the marginal effects for the after 2001 dummy variable implies that foreign born workers who were interviewed after 2001 were 7% le ss (more) likely to be authori zed & skilled (unauthorized & skilled), and 12% less (more) likely to be auth orized and unskilled (unauthorized & unskilled) than if they were interviewed prior to that peri od. Holding all other char acteristics constant, the results also suggest a stronger te ndency for Mexican workers to choose to work as unauthorized workers: they were 2% less (more) likely to be authorized & skilled ( unauthorized & skilled) and 4% less (more) authorized & unskilled (unauthorized & unskilled), respectively. The significance of English speaking ability fo r authorized employment is evident from the positive marginal effects for those groups with authorized status. Regardless of skill level and holding all other characteristics constant, workers are 6% more likely to choose to work in authorized statuses if they are ab le to converse in English. In th is respect, such workers are also 9% less likely to be unauthorized & unskilled Among those characteristics that directly impact workers choices, the piece rate and specialty crop variables have the largest marginal effects. Holding all el se constant, piece rate workers are 11% more (less) likely to be unauthorized & unskilled (unauthorized & skilled) and 8% more (less) likely to be author ized & unskilled (authorized & skilled). The direction of 1 However, Taylors study focused on a considerably smaller sample (<600) of male farm workers in California. 2 Note that for any variable included in one equation but not the other, the marginal ef fects are the mirror image for the opposite group in most cases.
79 influence is similar if the workers are employed in the specialty crop sector, where they are 3% more (less) likely to select into U.S. empl oyment as authorized & unskilled (authorized & skilled). Selectivity Corrected Wage Equation Models The estimated coefficients and asymptotic standard errors of the selectivity corrected wage models for each group are reported in Tabl e 3-5. With few exceptions, most parameter estimates are statistically significant at 10% or better and have the expected signs. The selectivity variables for legal status and job type, 1 and 2, account for potential selection bias arising from foreign born workers selections into U.S. employment in authorized or unauthorized status and skilled or unskilled employment, respectively. The estimated coefficients for both of these variables are statis tically significant at the 5% significance level or better across all worker groups, suggesting the presence of selectivity bias in the system. The estimated coefficients in the wage equations would therefore be bias ed if the selectivity correction variables were excluded from the wage equation models. The conventional variables of interest in ear nings models, educati on and experience, are shown to have significantly positive effects on ear nings across all worker groups. The education coefficient points to the fact that schooling le vels have a limited impact on mean earnings of farm workers in general; holding all other charac teristics constant, education increased mean earnings by less than 1% across a ll groups. Farm workers typically have low levels of formal education and from previous studies it would appear that the overa ll implication of education for farm workers is unclear (Taylor, 1992; Is and Perloff, 1995; Iwai, Emerson and Walters, 2006a).3 Dummy variables for adult education were included in the models to proxy additional 3 Across a broader range of occupations (agricultural versus non-agricultural occupations), the magnitude of the education impact is expected to be larger.
80 educational and training undertak en by farm workers, and are shown to have significantly positive effects on earnings across the board. The earnings effects are less than 2%, holding all other characteristics constant, for unauthor ized workers in general. Demographic variables denoting gender, age and English competency are also statistically significant across all groups, and shown to positively affect earnings. The exception in the latter respect is the fe male coefficient in the unauthorized & unskilled group which implies a penalty of about 1.8% on earni ngs for female workers in that group, holding all other factors constant. This finding is more in keeping w ith those reported by previous studies by Is and Perloff (1995) and Iwai, Emers on and Walters (2006a), where fe male farm workers reportedly earn less than male farm workers. Age has a significant positive nonlin ear effect on earnings except for unauthorized & unskilled workers. Sp eech competency in English has a significantly positive impact on earnings, and particularly among skilled workers; for authorized & skilled and unauthorized & skilled workers, re spectively, mean earnings increase by 3% or more if workers can speak English, holding all other factors constant. Though positive, the magnitude of influence on mean earnings is sm aller for unskilled workers. Among the labor market variables reported, the piece rate dummy is most dominant and has significantly positive effects on earnings acro ss all the worker groups. Holding all other characteristics constant, payment by piece rate in creased the mean earnings by 25% or more in all groups except the unauthorized & skilled group. Regard less of job (skill) type, the magnitude of influence is greater for the authoriz ed workers than for unauthorized workers. The remaining explanatory variables denoting employment with a grow er, employment in California, and seasonal employment have si gnificantly positive effects on earnings across the board, the exception to this being for s easonal workers in the unauthori zed & unskilled group. Iwai,
81 Emerson and Walters (2006a) reported a statistically ne gative impact of seasonal work on farm worker earnings in their study.4 The average predicted earnings for each worker group are reported in Table 3-6. Again, the importance of legal status is evident from th e fact that authorized workers earn more than unauthorized workers. Specifically, average predicted earnings authorized & skilled and authorized & unskilled workers ex ceed those for workers of the other groups. Perhaps the most interesting finding is that even unauthorized & un skilled workers earn more than unauthorized & skilled workers. It clearly implies that for fo reign born workers lacking legal status, selection into skilled employment is not advantageous; if anything, employment in the U.S. in an unauthorized status penalizes skilled workers. According to Emerson (2007), the wage penalty conferred by unauthorized status may be inte rpreted as a risk premium to employers to compensate for potential penalties and production losses that may arise if unauthorized workers were apprehended by the authorities. This finding is broadly cons istent with employment imp act noted by Taylor (1992) and lends support to his argument that unauthorized status may lessen workers chances of acquiring skilled jobs, or at least may discourage workers from seeking skilled jobs given the low earnings potential. From an employers perspective, if a worker is suspected or known to be unauthorized for US employment, placing that worker in a skilled job position increases risks considerably if he were apprehended by the authorit ies. Following Taylors argument, ceteris paribus, the apprehension of a worker employed at a skilled position implies an associated loss of human capital which translates to larger production lo sses for the employer than if the worker were 4 Other farm worker earnings studies with a legal status or job type selection focus (Taylor (1992) and Is and Perloff (1995)) did not include seasonal dummies in the wage models.
82 employed at an unskilled position (Taylor, 1992; p. 890).5 The punitive effect of unauthorized status on earnings is consistent with Is a nd Perloff (1995) and (Iwai, Emerson and Walters, 2006a) as well. Legal status is evidently the dominant factor affecting employment opportunities and earnings potential of forei gn born farm workers; whether or not they possess valuable job skills appears less important for employment in the U.S. crop farm sector. Concluding Remarks This study sought to address two key points that are commonly raised in the debate over legalization for unauthorized worker s: (1) whether unauthorized farm workers are more likely to be observed in unskilled farm jobs than their au thorized cohorts and, (2 ) whether they earn less than authorized farm workers for the same type of work. Previous stud ies have dealt with both of these issues but separately: selection bias was modeled as or iginating from a single criterion function, as opposed to the multiple decisi on criteria assumed in this study. The significantly positive bivariate probit m odel results reveal that the two decision criteria (legal status and job type) jointly affect foreign farm workers choices on U.S. farm employment, and that unauthorized workers are more likely to be observed in unskilled farm jobs. In addition to education, age and farm work experience, ma rital status, gender (female) and English speaking ability are shown to influence employment as authorized workers. In terms of job type, the work experience (U.S. and fore ign), English speaking skills, education and employment with a grower are the characterist ics associated with sk illed employment. Among the key variables of interest in the wa ge models, consistent significantly positive effects are shown for education and experience on earnings for all workers. Similar effects are 5 Taylor (1992) proposed a theoretical model that shows how wages of authorized and unauthorized workers may differ for the same job position, and how potential penalties and production and human capital losses may factor into the wage rates that are offered to workers.
83 noted for employment with a grow er, employment in California, English speaking ability, and the piece rate. The latter is shown to have a particularly strong effect relative to the other explanatory variables. In mo st instances, variables denoting seasonal employment and female gender had significantly positive effects on farm worker earnings. However, these findings contrast directly with estimates reported in past studies. The predicted earnings results hi ghlight the fact that legal status is most important for foreign born workers. Assuming that foreign wo rkers view U.S. employment as an optimal economic choice, employment in an illegal stat us is ill advised sin ce it would not increase earnings. Even if the workers were skilled, they would still earn less than if they chose to work as unauthorized & unskilled workers. Consequently, the lack of legal status may be correctly viewed as a barrier, in the sens e that unauthorized workers woul d be discouraged from seeking and holding skilled farm jobs due to the wage penalty. In the context of the ongoing immigration refo rm debate, the result s of this study would seem to suggest that if unauthorized workers we re able to gain legal status, they could earn higher wages and possibly move into better paying jobs. Under the proposed Comprehensive Immigration Reform Act of 2007 (S.1348), the proposed Agricultural Job Opportunities, Benefits and Security Act of 2007 (AgJOBS) (S.237, S.340, H.R.371) w ould establish a pilot program for adjustment to permanent resident status of qua lifying agricultural worker s who have worked in the United States during the two-year peri od ending December 31, 2005, and have been employed for specified periods of time subsequent to enactment of the Act (Library of Congress, 2007). To date, however, S.1348 has not been passed into law as it has been met with considerable opposition from segments of the American public that oppose its legalization provisions in particular.
84 Legalization must be addressed if the labor market opportunities ar e to improve for the majority of farm workforce, most of whom ar e unauthorized foreign born workers. That more than half of the current workforce fits this profile is an untenable position for not only the workers, but also for agricultural employers, given the financial risks associated with apprehensions.
85 Table 3-1. Explanator y variables for models based on the National Agricultural Workers Survey for 1993-2002 Variable a Definition LnWage Natural logarithm of the real farm wage in 2002 dollars. Conversions from the nominal wage were made using the consumer price index for all urban consumers Authorized =1 if farm worker is authorized for U.S. employment (citizen, permanent resident, or has other work authorization) = 0 if otherwise (i.e. unauthorized) Skill =1 if task is semi-skilled, supervisory, or other =0 if otherwise (pre-harvest harvest, post harvest jobs) Piece Rate = 1 if worker is paid by piece rate = 0 if otherwise (by the hour, hour/piece combination, or salary) Seasonal Worker =1 if worker is employed on a seasonal basis = 0 if otherwise (year-round) Female =1 if female =0 if male Mexican = 1 if worker is of Mexican nationality =0 if otherwise Education Highest grade level of educati on completed by the farm worker, ranging from 0 to 16 Adult Educationb =1 if worker had attended any adu lt education classes or school in the U.S. =0 if otherwise After 2001 Dummy variable reflecting the interview years following September 2001 California (CA) Dummy variable reflecti ng employment in California at the time of the interview English (speaking ability) = 1 if none at all = 2 if a little = 3 if somewhat = 4 if well
86 Table 3-1. Continued. Variable Definition Married = 1 if married/living together =0 if otherwise Years with current employer Number of years of employment wo rker has completed with current employer. One year is measured as one or more days per year (NAWS) Farmwork weeks Farmwork weeks completed in the last year Foreign Farm Work Experience =1 if worker had been employed in agriculture, either full-time or part-time, while living in native (foreign) country =0 if worker had been employed in non-agricultural sector or had never worked while living in native (foreign) country Grower = 1 if employed by a grower = 0 if employed by a farm labor contractor Specialty Crop = 1 if worker was employed in sp ecialty crop production at the time of the interview =0 if otherwise Age Respondent age in years Age2 Age squared Experience Years of U.S. farm work Experience2 Experience squared 1 Selectivity correction term from the legal status (authorized) decision equation 2 Selectivity correction term from the job type (skilled) decision equation a Data were sourced from the National Agricultural Workers Survey. Definitions enclosed in quotation marks are as they appear in th e NAWS Codebook. b This would include English/ESL, citizenship, literacy, job training and Adult Basic Education classes, GED/high school equivalency classes, college or university classes, and Even Start and Migrant Education classes.
87Table 3-2. Descriptive statistics for explanatory variable s (N =12,851 foreign workers) Authorized & Skilled Group (n=2428) Authorized & Unskilled Group (n=3251) Unauthorized & Skilled Group (n=2300) Unauthorized & Unskilled Group (n=4872) Variable Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Real wage 7.30 1.987.262.616.511.546.772.22 Adult education 0.27 0.450.260.440.150.360.140.34 After 2001 0.31 0.460.200.400.420.490.240.43 Female 0.14 0.350.220.420.120.320.140.35 Married 0.80 0.400.760.430.510.500.480.50 English 2.12 0.921.910.911.560.701.520.72 Mexican 0.87 0.330.900.300.890.320.910.29 Education 5.69 3.385.723.4220.127.116.113.25 Experience 16.32 8.8314.468.515.685.845.105.70 Experience squared 344.25 365.38281.45338.9266.34183.6258.52176.06 Age 39.45 11.1937.9211.1628.469.6628.4710.25 Age squared 1681.28 954.271562.21927.95903.38685.67915.79756.23 Seasonal worker 0.63 0.480.780.410.690.460.830.38 Foreign farm work Experience 0.70 0.460.630.480.700.460.690.46 Years with current Employer 6.74 5.805.224.812.792.792.322.26 Farm work weeks 39.14 12.2934.7113.4835.2615.4331.6716.19 Piece rate 0.11 0.310.240.430.110.320.290.45 Grower 0.84 0.370.780.420.810.390.740.44 Specialty crop 0.80 0.400.860.340.720.450.780.42 California 0.58 0.490.460.500.330.470.240.43
88Table 3-3. Bivariate probit model estimates for forei gn-born workers legal status and job type selections Authorized Skilled Variablea Parameter Estimate Standard Error Parameter Estimate Standard Error Adult education 0.1803*** 0.0360 ----After 2001 -0.5173*** 0.0325 ----Female 0.4586*** 0.0377 -0.1491*** 0.0353 Married 0.2305*** 0.0318 ----English 0.2987*** 0.0188 0.0886*** 0.0152 Mexican -0.1596** 0.0460 Education 0.0351*** 0.0048 0.0043 0.0040 Experience 0.2074*** 0.0051 0.0095*** 0.0016 Experience squared -0.0036*** 0.0001 ----Age 0.0531*** 0.0077 ----Age squared -0.0005*** 0.0001 ----Seasonal worker -----0.2252*** 0.0297 Foreign farm work experience ----0.0619** 0.0283 Years with employer ----0.0160*** 0.0032 Farm work weeks ----0.0061*** 0.0009 Piece rate -----0.5303*** 0.0313 Grower ----0.0655** 0.0294 Specialty crop -----0.1758*** 0.0289 Constant -3.5134*** 0.1459 -0.5545*** 0.0673 Sample size = 12851; Log lik elihood = -13172.59; Rho ( ) = 0.095*** a Triple and double asterisks (***, **) indicate statistical signi ficance at 1% and 5% levels of significance, respectively.
89 Table 3-4. Marginal effects of bivariate probit selection estimate s into legal status & job type Marginal Effecta Variable Authorized & Skilled Authorized & Unskilled Unauthorized & Skilled Unauthorized & Unskilled Adult education 0.0257 0.0447-0.0257 -0.0447 After 2001 -0.0710-0.1203 0.0710 0.1203 Female 0.0357 0.1449-0.0903 -0.0903 Married 0.0325 0.0558-0.0325 -0.0558 English 0.0562 0.0594-0.0231 -0.0925 Mexican -0.0228-0.0397 0.0228 0.0397 Education 0.0056 0.0079-0.0040 -0.0095 Experience 0.0453 0.0592-0.0453 -0.0592 Age 0.0020 0.0033-0.0020 -0.0033 Seasonal worker -0.0353 0.0353-0.0503 0.0503 Foreign farm work Experience 0.0096-0.0096 0.0135 -0.0135 Years with current Employer 0.0025-0.0025 0.0035 -0.0035 Farm work weeks 0.0009-0.0009 0.0013 -0.0013 Piece rate -0.0772 0.0772-0.1066 0.1066 Grower 0.0101-0.0101 0.0142 -0.0142 Specialty crop -0.0276 0.0276-0.0392 0.0392 a All marginal effects are statistically significant at the 5% level or better.
90Table 3-5. Selectivity corrected wage models for each worker group Authorized & Skilled (n=2428) Authorized & Unskilled (n=3251) Unauthorized & Skilled (n=2300) Unauthorized & Unskilled (n=4872) Variablea Parameter Estimate Corrected S. Error Parameter Estimate Corrected S. Error Parameter Estimate Corrected S. Error Parameter Estimate Corrected S. Error Adult Education 0.05212*** 0.011100.02067***0.004170.01635***0.006170.01712**0.00724 California 0.03884*** 0.002140.07742***0.001290.03523***0.001170.05191***0.00178 Grower 0.04725*** 0.010700.06010***0.006550.04405***0.003520.06167***0.00179 Piece Rate 0.28992*** 0.023810.39628***0.012420.17457***0.010160.25901***0.00719 Seasonal Worker 0.05484*** 0.013630.04477***0.01010-0.01700***0.003230.01050***0.00310 Female 0.07900*** 0.018720.02021**0.009210.01377**0.00662-0.01860***0.00647 Education 0.00558*** 0.001920.00588***0.001010.00593***0.000520.00564***0.00059 English 0.04105*** 0.008050.01584***0.004800.03391***0.004540.01616***0.00498 Age 0.01764*** 0.002010.01022***0.000700.00862***0.003300.001320.00129 Age squared -0.00021*** 0.00003-0.00012***0.00001-0.00011**0.00005-0.000010.00002 Experience 0.02801*** 0.002360.01416***0.001180.02643***0.004810.01875***0.00355 Experience squared -0.00052*** 0.00005-0.00030***0.00002-0.00048***0.00012-0.00038***0.00007
91Table 3-5. Continued. Authorized & Skilled Authorized & Unskilled Un authorized & Skilled Un authorized & Unskilled Variablea Parameter Estimate Corrected S. Error Parameter Estimate Corrected S. Error Parameter Estimate Corrected S. Error Parameter Estimate Corrected S. Error 1 0.19174*** 0.019790.08284***0.010010.13674***0.052680.09735**0.04468 2 -0.47152*** 0.05304-0.41924***0.03725-0.10748***0.02129-0.13658***0.03013 Constant 1.38847** 0.661971.01502**0.430461.61350***0.356181.53875***0.30150 a Triple and double asterisks (***, **) indicate statistical si gnificance at 1% and 5% levels of significance, respectively.
92 Table 3-6. Average predicte d conditional wage for each legal status & job type group Legal Status & Job Type Groups Wagea ($) Authorized & skilled (G11) 8.42 Authorized & unskilled (G10) 7.76 Unauthorized & skilled G(01) 6.62 Unauthorized & unskilled G(00) 6.78 a Average wages are conditioned on the selectivity variables for legal status and skill type.
93 CHAPTER 4 PROPOSED IMMIGRATION POLICY REFORM & FARMWORKER OUTCOMES Overview Immigration reform has generated much poli tical debate in recent years. The last substantial revision of immigra tion law occurred in 1986 with th e passage of the Immigration Reform and Control Act (IRCA), which authorized several policy instruments to discourage illegal immigration and employme nt. In the twenty years sin ce however, it is apparent that IRCA has failed in its stated objectives for not only has illegal immigration increased significantly, but unauthorized immi grants have continued to ga in employment in the U.S. particularly in the low-skilled, low-wage sectors of the economy (Passel, 2005; Passel and Suro, 2005; Passel, 2006; Mines, Gabba rd and Steirman, 1997; Carro ll et al. 2005). Statistics published by the Pew Hispanic Center provide am ple evidence in these re spects. In 2004, for example: There were 10.3 million unauthorized immigrants residing in the United States. They accounted for 29% of the foreign born population; Most unauthorized immigrants had arrived since 1990; 66% arri ved between 1994 and 2004, and 30% migrated to the U.S. between 2000 and 2004; Most unauthorized immigrants were Latin American (81%), with Mexicans comprising the largest subgroup (57%); Occupations with the highest percentage of unauthorized workers included agriculture (19%), cleaning (17%) and cons truction (12%) (Passel, 2005). Unsurprisingly, media attention on these trends has refocused national interest on immigration reform. Judging from the intensity of the debate in political and national fora, issues attendant to policy reform (border and worksite enforcemen t and legalization of una uthorized immigrants) will feature prominently in the 2008 presidential elections.
94 The political debate began in earnest with the passage of two earlier proposals in the 109th US Congress. Legislativ e proposal H.R. 4437 (the Border Protection, Antiterrorism, and Illegal Immigration Control Act of 2005 ) was passed by the US Hous e of Representatives in December 2005. It is arguably one of the more restrictive proposals introduced for consideration in the 109th Congress in that it contai ned no provisions for legaliza tion of unauthorized workers or for a guest worker program. H.R. 4437 em phasized a pro-enforcement stance on immigration reform; it advocated criminal penalties for unaut horized immigrants and significant fines for the U.S. employers who would hire them. The proposal also argued for I-9 document reform and for increased worksite/interior/border enforcement, but made no mention of modifications to existing laws on legal immigration. In contrast, S. 2611 (the Comprehensive Immigration Reform Act of 2006 ) passed by the U.S. Senate in May 2006 proposed earned legalization for unauthorized immigrants and modifications to existing laws on legal immigration. Though it fa vored stricter enforcement and I-9 reform, overall, it was not as severe as H.R. 4437 in the overall approach to illegal immigration. Specific provisions for the agricultural sector were proposed under AgJOBS ( Agricultural Job Opportunity, Bene fits and Security Act of 2005 (S. 359/H.R. 884; S.2611 Subtitle B), which would streamline the H-2A program to improve wages, working conditions and minimum benefits (housing and transportatio n) for farm workers and establish a pilot program for earned legalization of eligible unauthorized workers.1 Neither S.2611 nor H.R. 4437 was passed since Congress failed to reach a compromise between the two sets of views on immigration. The failure to achieve compromise can be linked directly to the competing interests that lawmak ers had to contend with: disagreements on policy 1 Unauthorized workers would have to meet certain work-related requirements to qualify for legalization under the AgJOBS.
95 provisions between anti-immig ration/pro-enforcement groups and pro-immigration groups, disagreements on specific reform measures between and within political parties in Congress, and intense lobbying from employer and worker advocacy groups for certain concessions. In many respects, the most divisive i ssue has been proposed legalization for unauthorized immigrants. There are segments of the Amer ican public that strongly oppose legalization on the grounds that it would reward illegal behavior a nd encourage future illegal immigra tion, as there are others that view legalization as the only viab le means of bringing unauthorized immigrants into mainstream U.S. society, that is, in lieu of mass deportations. Amidst these divergent views, employers of low skilled foreign labor particularly farm employers have expressed preference for increa sed access to immigrant labor to offset labor shortages. This issue is particularly importa nt to farm employers that have high demand for manual labor over short periods during harvest ti me. Immigrant workers presently comprise a significant proportion of the crop farm workforc e (78%), an estimated 53% of which is unauthorized for US employment (Carroll et al. 2005). Farm employers are justifiably concerned since these statistics clearly highlight their vulnerability to changes in immigration policy that may curtail their access to foreign labor. Given this context, the purpose of this pape r is to evaluate the implications of U.S. immigration policy reform for U.S. farm la bor market outcomes, focusing specifically on proposed legalization for unauthor ized immigrant workers. The study uses a treatment effects (TE) framework in which legalization is modeled as a treatment or (policy) intervention. The TE framework is a novel approach to immigration policy evaluation that ha s not been used in previous studies that have eval uated the potential impact of legalization for farm outcomes.
96 The paper is organized as follows. Following th is introduction, the second section reviews the immigration policy proposals of the 109th Congress (S.2611 and H.R. 4437). Much of the discussion is devoted to the Ag JOBS proposal. The proposals th at were introduced in the 110th Congress (S.1348 and H.R. 1645) are only briefly mentioned since they have not been debated or passed by Congress. The third section comprise s the analytical framework employed in the study, and the fourth section pres ents the study findings. The final sections of the paper present the policy implications and some concl uding remarks. Immigration Policy Reform Proposals The reform measures proposed in S.2611 and H.R. 4437 may be broadl y categorized into six core issues: worksite enforcement, documen t reform, enforcement provisions, guest worker programs, legal immigration and adjustment to legal status for unauthorized immigrants. In general, H.R. 4437 took a more restrictive st ance on immigration reform than S.2611 and proposed stricter penalties for employers and una uthorized workers. The similarities between the two proposals concerned issues pertaining to enforcement (wor ksite and border) and document reform. Their major differences were primarily on legalization measures, specifically adjustment to legal status fo r unauthorized immigrants and guest worker programs. Whereas S.2611 set forth various conditions for legalization, H.R. 4437 was notably silent on this issue except for a single stipulation on sanctions for unau thorized presence in the US. The sanctions stipulation would have changed th e criminal penalty for illegal entry and unauthorized presence (from a misdemeanor to a felony) (Lib rary of Congress, 2007). S.2611 addressed matters of work authorization and legalization of unauthorized immigrants under Title VI, Subtitle A, Section 601. Unauthori zed immigrants could qualify for earned adjustment status provi ded that they were able to give proof of unauthorized US
97 presence on or before April 5, 2001.2 Except for brief absences, they would be required to show that they had not departed the United States be fore that date. Unauthorized immigrants would have been required to demonstrate that they were employed in the United States for at least three out of the five years prior to application for lega l status and show employ ment for six years after the date of enactment of the Act. Immigrants would be required to pa y fines, taxes and meet other conditions for admissibility (Library of Congress, 2007). Section 245c indicates similar unauthorized presence and employment conditions for unauthorized immigrants seeking to apply for deferred mandatory departure status. However, except for brief departures, the unauthorized immigrant would have been required to give proof of continuous US presence on January 7, 2004 as well as employme nt before and after that date.3 Earned adjustment would allow for automatic legal permanent resident status whereas deferred mandatory departure status would allow immigrants to apply for visas while in the US Both statuses have specific ineligibility conditions under wh ich unauthorized immigrants may have been disqualified for authorized US residence (Lib rary of Congress, 2007). The main proposals of the 110th Congress the Comprehensive Immigration Reform Act of 2007 or the Secure Borders, Economic Opportunity and Immigration Re form Act of 2007 (S.1348) and the Security through Regularized Immigration and a Vibrant Economy (STRIVE) Act of 2007 (H.R. 1645) were introduced in the US Senate and House of Representatives, respectively. As they exist currently, they a ppear to have more in common with proposal S.2611 on certain core issues. To date, neith er proposal has been passed by Congress. 2 This section is also cited as the Immigrant Accountability Act of 2006. 3 This particular status is granted to unauthorized immigrants to allow them time to depart the US and apply for readmission as a nonimmigrant or immigrant alien (Library of Congress, 2007).
98 Legalization conditions for agricultural workers are addressed via the AgJOBS ( Agricultural Job Opportunity, Benefits and Security Act of 2005 (S. 359/H.R. 884; S.2611 Subtitle B) proposal. 4 It first gained bipartisan support in Congress in 2000 and was adopted by the Senate in May 2006 under Senate Bill S.2611. AgJOBS has two major objectives, the first of which would involve revision of the existing H-2A temporary worker program. Title II of the proposal stipulates: Revision of the administrative procedures, in cluding elimination of the labor certification process; Reformation of the hiring requirements fo r H-2A employers, including an immediate reduction and gradual elimination of the Adverse Effect Wage Rate; Streamlining of the process for admission of H-2A aliens. There is also a provision which seeks to eliminate, on a one-time basis, the statutory bar preventing aliens not currently in the program from acquiring H-2A status (Libra ry of Congress, 2007; Craig, 2006). The second major objective focuses on earned le galization. This would allow agricultural workers to obtain temporary legal status based on past work experience with the possibility of adjusting to permanent status thr ough future/continued work in agricu lture. Title I, Subtitle A, Section 101 of S. 237 stipulates the following conditi ons for blue card qualification. The worker must: Perform agricultural employment in the Unite d States for at least 863 hours or 150 work days during the 24-month period ending on December 31, 2006; Apply for blue card status during the 18-month application period beginning on the first day of the seventh month that begins after th e date of enactment of the legislation. Immediate family members would acquire derivative legal status and spouses would be able to apply for work permits; Must be otherwise admissible to the United States and have no history of felony or misdemeanor convictions involving bodily injury threat of serious bodily injury, or harm to property in excess of $500 (Library of Congress, 2007). 4 AgJOBS is referenced as S.237/S.340/H.R.371 in the 110th Congress (see Library of Congress, 2007).
99 Title I Subtitle B of the proposal also allows for amendment of Social Security records to permit exemption of blue card aliens from prosecution for improper conduct (related to identity or payment false statements) if they committed such conduct prior to acquiring blue card status (Library of Congress, 2007). Transition to permanent (green card) status is addressed in Titl e I Section 103 of the proposal. In order to qualify for legal permanent resident status, workers must complete additional agricultural work of either 100 days annually for each of the five years beginning on the enactment date of AgJOBS or 100 days for one year and 150 days annually for three years during a 4-year period begi nning on the enactment date or 150 days annually for each of the three years during the 3-year period beginning on the enactment date. Their immediate families would be granted legal status c ontingent upon their fulfillment of these requirements (Library of Congress, 2007). The last major legalization program in recent history was the Special Agricultural Worker (SAW) program of the 1986 Immigration Reform and Control Act (IRCA), which granted legal status to an estimated 1.1 million farm workers. Analyzing the wage effects of this legalization program could provide significan t insight into the potential wage impact of the proposed AgJOBS for different segments of the workforce. It would also be instructive to look at the factors that may affect worker s decisions on obtaining legal status for US farm work. Analytical Framework According to Heckman and Vytlacil (2005), the structural econometric (SE) approach and the treatment effects approach (TE) are two competing paradigms in the area of empirical policy evaluation. Of the three centr al tasks of economics (p. 669) described (policy evaluation, forecasting of policy effects in new settings, and prediction of new policies), the SE approach is noted for its applicability to al l three tasks provided that certai n assumptions are satisfied. The
100 TE approach, as currently develo ped, differs markedly in that it concentrates more on policy evaluation and compares outcomes for treatment pa rticipants and non-partic ipants. It therefore invokes fewer functional form and exogeneity assumptions and its linkages with economic theory are more loosely specified (Heckman and Vytlacil, 2005). The differences between the two approaches were outlined in an earlier paper by Heckman (2001). First, whereas the SE approach seeks to answer many counterfactual questions via well defined economic parameters, the TE approach focu ses on a more limited range of questions and analyzes them under weaker conditions. Second, whereas the parameters generated under the SE approach compare well across studies, those ge nerated via the TE appr oach do not. Third, whereas the SE approach lends well to extrapola tion, the TE approach allows for evaluation of a single program in a single environment only. Fourth, whereas the SE approach allows for partial as well as general equilibrium applications, th e TE approach reflects a partial equilibrium viewpoint only and cannot be used for evaluation of interventions that apply universally within an economy (Heckman, 2001). These differences not withstanding, the TE approach has become more popular in recent years because, as noted by Cameron and Trivedi (2005): Policy relevance of treatment evaluation is direct because successful treatments can be linked to desirable social pr ograms, or improvements in existing programs to attain objectives of social policy (p. 861). Treatment Effects Approach The treatment effects (TE) approach measures the impact of treatment on outcomes of interest. In this context, tr eatment may refer to medical treatments, public programs or social interventions (Basu et al. 2007), a nd the causal effect of the treatment on the outcome is defined as the treatment effect. The methodology originated with the medical sciences where interventions are primarily concer ned with patients responses to treatment regimes relative to specific benchmarks that may include alterna tive treatment regimes or no treatment. The
101 analogy to be drawn with economics is that treatme nt is akin to participation in a program and outcomes are concerned with changes in the economic status or environment on the economic outcomes of individuals who elect to participate in the progr am (Cameron and Trivedi, 2005). 5 The contrast between the two fields rests primarily on the type of data used. Medical studies tend to rely more on experimental data where non-participants ad equately serve as controls to treatment, whereas economic studies rely more on non-experimental data where the use of control groups may be highly uneth ical (Lee, 2005). In medical st udies therefore, the process is randomized whereas in economic studies, there is underlying self-selectio n into treatment. Economic issues that have utili zed the TE approach include enroll ment and evaluation of social programs (welfare and Job Corps programs), the im pacts of changes in regulations for financial transactions and of changes in economic incentives. The standard problem involves the inference of a causal connection between participation (treatment) ( D) and the potential outcome (Y) where the potential outcomes for the participant (treated) 1Y and non-particip ant (non-treated) 0Ystates are compared for the ith individual to evaluate how his average economic outcome would ch ange if he were to pa rticipate in a program or not. Following the latent variable framework of Heckman, Tobias and Vytlacil (2001; 2003) and Blundell and Costa Dias ( 2002), the potential outcomes based on observable characteristics ( x) and the participation decision for a program may be defined as: otherwise 0D D if 1 D where treatment) in eparticipat to (decision ZD group) ipant /nonpartic (untreated uXuxgY group) articipant (treated/p uXuxgY* i i i ;0 '0 00 00 1 11 11 (4-1) 5 Treatment and participation are used interchangeably throughout the chapter.
102 In this setup,xg ,xg01represent the relationship between the observable characteristics and the potential outcomes and ,u,u01 Z and x are unobserved and observed random variables, respectively. The errors are assumed to be independent of x and Z. Ceteris paribus, the treatment or causal effect is defined as shown by equation (4.2), and is the difference between the potential outcomes: N 1,...,i YYi0i1i (4-2) This effect is not directly estimable as it is im possible to simultaneously observe an individual in both states. The observed outcome is actually: i0i i1iiYD1YDY (4-3) where the unobservable portion of th e effect is referred to as the counterfactual outcome. (For those individuals receiving treatment 0Y is the counterfactual out come; for those who do not, 1Y is the counterfactual outcome.) The treatment effect of each person is independent of the treatment of other individuals, implying that an individuals potential ou tcomes are affected by his participation decision only and not the de cisions of other indivi duals (Wooldridge, 2002; Caliendo, 2006). Gains from treatment are typically define d as population averages. Some relevant parameters include: Average Treatment Effect (ATE). This is th e expected gain from participating in a program for a randomly chosen individua l (Heckman, Tobias and Vytlacil, 2001), calculated as the differences in expect ed outcomes before and after treatment: 0 1 ATEYEYEE (4-4) Average Treatment Effect on the Treated (ATET). This is the average gain from treatment for those who select into the treatment (Heckman, Tobias and Vytlacil, 2001): 1D|YE1D|YE1D|E0 1 ATET (4-5)
103 Average Treatment Effect on the Untreated (ATEU). This is the effect for nonparticipants which may be useful for future policy decisions on extending treatment to groups that were excluded from treatment (Caliendo, 2006): 0D|YE0D|YE0D|E0 1 ATEU (4-6) Marginal Treatment Effect (MTE). 6 This is the expected eff ect of treatment conditional on observed (X) and unobserved (Ud) characteristics of participants (Heckman and Vytlacil, 2005).7 One interpretation is that it is th e mean gain for an individual with characteristics X and unobservables Ud such that he is indifferent between treatment or not given a set of Z values, z, where ( z)=ud. It is defined as: ddii0i1 01 d i d d i d 01 dd duU|uuE XuU,xX|E uU,xX|YYEuU,xX|EU,XMTE (4-7) The challenge posed by selection bias is evident in Equation (4 -6), which shows a hypothetical outcome in the absence of treatment for those individuals who received treatment (Caliendo, 2006). With non-experimental data, this outcome is not equivalent to the outcome of non-participants: 0|1|0 0 DYEDYE (4-8) Selection bias may arise since participants and non-participants may be deliberately selected groups with different outcomes, even in the ab sence of treatment, due to observable and unobservable factors that may determ ine participation (Caliendo, 2006): bias Selection DYEDYE ATET DYYEDYEDYE 0|1|1| 0|1|0 0 01 0 1 (4-9) If selection bias is id entified as arising from observed f actors, matching methods, propensity scores and linear regression techniques are suit able estimation methods. Additionally, in cases where unobserved sources of selection bias are suspected, instrumental variable techniques, 6 Bjorklund and Moffitt (1987) are credited with introducing this concept to the literature. 7 The unobserved characteristics are introduced into the model by the decision rule described by equation (4-2).
104 difference-in-differences methods and selecti on models are more common though not in the case of models that assume essential heterogeneity. Homogeneous and heterogeneous treatment effects Much of the previous literature on treatment effects assumed homogeneous responses to treatment, meaning that based on certain observable characteristi cs, effects are constant across individuals and that they would derive identical benefits from treatment. Recent studies have given more attention to heteroge neous responses where the effects vary across individuals due to their observable or unobservable characteristics. Much of the focus is now on the role of unobservable characteristics in de termining outcomes particularly in cases where individuals are otherwise identical in their observed characteristics (Bas u et al. 2007; Caliendo, 2006). Basu et al. (2007) describe two instances in which heterogeneity (arising from unobservable characteristics) may fa ctor into treatment evaluation. The first instance is where individuals with id entical observable characte ristics respond differently to treatment but do not opt for treatment based on their idiosyncratic bene fits or gains (non-esse ntial heterogeneity). The second instance is where individuals have identical observable char acteristics and respond differently to treatment and are aware of the benefits to be derived from treatment. In this latter case, their treatment choices are influenced by an ticipation of idiosyncratic gains (Basu et al. 2007). Basu et al. (2007) and Heckman, Urzu a and Vytlacil (2006a; 2006b ) refer to the second instance as essential heterogeneity. In the context of this paper, heterogene ity of foreign farm worker responses to legalization is maintained and subjected to a statis tical test. In the presence of heterogeneity, it is assumed that they would seek legalization thr ough AgJOBS (or other le galization mechanisms) because of individually perceived wage benefits The analysis follows a parametric approach developed by Heckman, Urzua and Vytlacil (2006 b) to estimate the choice and outcome models,
105 and the treatment effects of legaliz ation; alternative non-parametric es timates are also evaluated. Their MTE algorithm was used for the analysis.8 The overall models and estimation procedure are described in the following sections, and draw heavily on the theoretical expositions of Heckman, Urzua and Vy tlacil (2006a; 2006b). Parametric model with heterogeneous treatment effects The parametric model with essential heterogeneity adopts the familiar latent variable framework shown: )(otherwise 0D status)legal for opts workerthe (if D if 1 D rule sion model/deci Choice ZZD* 0 (4-10) group) untreated for outcome (Wage uXY group) treated for outcome (Wage uXYii ii 0 00 1 11'ln 'ln (4-11) where the Z and X are vectors of observa ble characteristics and iiuu01,, are error terms that encapsulate the unobservable characteristics of individuals. The decision to accept treatment (legal status) is defined by a choice model that allows for two separate log wage outcomes01ln,ln YY .9 The choice model may be interpreted as a net utility for individuals with the characteristics Z and Similarly, the (log wage) outcomes are functions of the ith workers characteristics denoted by Xi and uji 1,0 j, respectively. The error of the choice model ( ) is assumed to be independent of Z given X The parametric model assumes joint normality of the errorsiiuu01,, which are assumed to be independent of the observable characteristics (Z and 8 Information on the MTE is available at http://jenni.uchicago.edu/underiv/ (Cited as Heckman, Urzua and Vytlacil, 2006c in reference list). Also, see Heck man, Urzua and Vytlacil (2006a; 2006b). 9 Parameters Z and are assumed to be additively separable as is the predominant specification in the literature.
106 X ). Based on this assumption, the expectations on the errors of the outcome equations reflect the differences in legal status choice ( D=1 if legalized/treated, D=0 if not legalized/untreated): ZP Z pZPDxXuE ,1,|1 1 (4-12) ZP Z pZPDxXuE 1 ,0,|0 0 (4-13) where 0, 0 1, 1; are the correlations between the disturbances of the respective outcome equations and the choice equation, and (.) denotes the standard normal density function (Heckman, Urzua and Vytlacil, 2006b). The probability of becoming legalized is defined as: Z ZPrzZ|1DPrzPr (4-14) where .is the cumulative distribution of Heckman, Urzua and Vytlacil (2006a) refer to this function as a propensity score, taken as a monotonic function of the mean utility of treatment (legal status). Equation (4-10) is revi sed to reflect the acceptance decision as: dUZP1 Z1D (4-15) where Ud denotes the unobserved charac teristics of individuals. The algorithm estimates the propensity score using a probit m odel, from which the predicte d values for the treated and untreated groups are used to define values over which the marginal treatment effect (MTE) of legalization may be identified (Heck man, Urzua and Vytlacil, 2006b). Since it is impossible to obs erve an individual in the tr eated and untreated states simultaneously, Equations (4-3) and (4-10) are used to derive the actual outcome to be estimated:
107 ii ii ii ii ii ii iiiuXD uXuu XD YDYDY0 0 0 00101 0 11lnln (4-16) where ii iiuu X0101 is the heterogeneous return to legal status for the ith foreign farm worker (i.e. the effect varies across all fa rm workers). If the heterogeneous effect were from a differential between the j terms only, this would be observed heterogeneity ; if it were to arise as a consequence of differences between the uji terms, it would be unobserved heterogeneity. In either case, this parameter w ould imply different wage effects of legalization across foreign workers in the farm workforce even if they have identical observable characteristics. Note that for individuals who gain legal status ( D=1 ), i captures the benefit of legal status. Treatment effect parameters The literature on marginal treatment eff ects (MTE) spearheaded by Heckman provides several interpretations of the MTE which are equivalent under certain assu mptions that apply in this analysis (see Heckman and Vytlacil (2007b) and references ther ein). One interpretation of the MTE presents it as a measurement of the marg inal return to individua ls who are indifferent between foregoing ( D=0 ) or accepting treatment ( D=1 ) when their mean utility ( d(Z )) is equivalent to Ud If jY ln are defined as value outcomes, it may be interpreted as a willingness to pay measure for individuals with certain observable characteristics ( X ) and unobserved heterogeneity (Ud ) at a specified margin of indiffere nce (Heckman and Li, 2004; Heckman, Urzua and Vytlacil, 2006a). The other treatment effect estimators the average treatment effect (ATE), the average treatment effect on the treat ed (ATET) and the average treatment effect on
108 the untreated (ATEU) are generated as weight ed averages of the MTE (Heckman, Urzua and Vytlacil; 2006a; 2006b): dd MTE dd iduux uUxXEATE ,|1 0 (4-17) dd ATET MTE iduux DxXEATET 1,|1 0 (4-18) dd ATEU MTE iduux DxXEATEU 0,|1 0 (4-19) where the applicable weights based on the propensity score (P ) are: PE and PEi P ATEU i P ATET 1 1 (4-20) Data The data consist of 19,152 foreign worker s with complete data from the National Agricultural Workers Survey (NAWS) for 1989 to 2006. Subsamples reflecting the treated (those who have obtained legal status; N1=8097) and untreated (thos e without legal status; N0=11055) worker groups are specified. Table 4-1 defines the variables that were used in the analysis. The variables reflect the demographic characteristics of the crop farm workforce and certain characteristics of the farm labor market Dummy variables reflecting the location, time period of interview, and time period when the fore ign workers would have entered the US to live or work are also included. Results and Discussion Table 4-2 reports the summary st atistics for the variables that were used in the analysis. The treated group comprised 8,097 foreign work ers whereas the untreated group comprised 11,055 foreign workers. One of the more intere sting findings to emerge from the data is the difference in time spent abroad by workers of the two groups: workers who are not legalized have much longer overseas stays than their cohorts who are legalized (5 weeks on average). In
109 addition, workers who gained legal status reporte d more months and weeks of farm work in the previous year on average than their cohorts who did not gain legal stat us. Although workers had similar foreign farm work experience, there was a sizeable difference in US farm work experience. On average, legalized workers repor ted 16.7 years of US farm work compared to 6 years for workers who were not legalized. Not su rprisingly, legalized wo rkers had migrated to the US much earlier (~12 years more) and had worked with their current employers for much longer periods (~4 years) in comparison to their cohorts who did not gain legal status. On average, 62% of the foreign farm workforce had mi grated to the US to live and work after 1986. Among legalized workers, only 26% had migrated to US after 1986, whereas approximately 88% of unauthorized workers had entere d the US since that period. Tables 4-3 reports the estimated choice m odel results from the MTE algorithm. The instruments included in this model are farm wo rk weeks, years with employer, years since immigration, after 1986 and weeks spent abroad. The characteristics that significantly increase the likelihood of legalization (t reatment) are years with employe r, English, and years since immigration; those that decrease the likelihood of treatment are farm work weeks, after 2001 and weeks spent abroad. The after 2001 dummy variab le was included to distinguish between the preand post9/11 periods, and the after 1986 dummy was included to distinguish between the periods when workers first entered the US to live or work. The latter reflects the broad legalization through the SAWs progr am for those who were in the US and working prior to the passage of IRCA in 1986, and the relative difficulty of acquiring legal status since 1986. The direction of influence signaled by the after 2001 coefficient suggest s that foreign farm workers were less likely to gain legal status following the September 2001 terrorist attacks; this makes sense given that enforcement efforts and secur ity were heightened in the US following that
110 event. Arguably legal status would have been more difficult to attain with the additional checks and safeguards that were put in place. That is no t to say that the tightening on legal status would have necessarily had a significant adverse effect on foreign farm workers; if anything, these workers are more likely to have migrated ille gally across the US border with Mexico. The magnitude of the after 1986 dummy suggests that le gal status has been diff icult to acquire since the last major legalizatio n in 1986 (the SAWs program). The farm work weeks effect indicates that more weeks of farm work reduce th e likelihood of having legal status. Table 4-4 presents the parametric model wage results for the treated and untreated worker groups.10 All parameter estimates have the expected direction of influence on the wage results, and the parameters are statistically significant at the 1% level of significance; the exceptions are the foreign farm work experience and age vari ables in the treated and non-treated groups, respectively. For both groups, the magnitude a nd statistical significance of the piece rate and after 2001 estimates suggest dominant influences on fa rm wages relative to the other variables of the model. Table 4-5 reports the estimated treatment effect s of legalization which are all positive. The average treatment effect (ATE) re flects the expected gain for a random foreign farm worker who became legalized, the average trea tment effect on the treated (ATET) indicates the return to those workers who became legalized, and the av erage treatment effect on the untreated (ATEU) indicates the potential return fo r those who were not legalized. The order of magnitude of the estimates generated by each method indicates po sitive sorting on the gains associated with legalization (ATET>ATE>ATEU), wherein those foreign workers who were most likely to 10 Wage results for the nonpar ametric methods (polynomial, nonparametric I, nonparametri c II) are reported in the Appendix.
111 participate in the legalization program benefited the most from it more so than the average person and more so than their c ohorts who were not legalized. A comparison of the different estimation met hods shows a striking difference between the parametric and nonparametric methods in the magnit ude of sorting gains from legalization. The sorting gains are the difference between the ATET and ATE estimates the average gains for the worker who opts for treatment (legalization) versus the worker who randomly selects into treatment. The parametric method reports the smallest sorting gain of the four estimation methods: the average earnings ga in for the legalized (treated) foreign worker was 0.0023, implying that the average earnings gain to le galization was 0.23% grea ter than the average earnings gain for the average foreign worker who randomly selected into legalization. The gains for the nonparametric methods (polynomial, nonpa rametric I, and nonparametric II) range from 3.16% to 6.57%. A comparison of the estimates reported within shows significant differences in the magnitude of average returns to legalization for the treated (legali zed) and untreated (nonlegalized) groups. The parametric method estimat es are the exception in this respect: earnings gains average 10% across the board irrespective of treatment status. The differentials are largest for the nonparametric I method, followed by the nonparametric II and polynomial methods, respectively. The average earnings gain for the untreated (ATEU) range between 15% and 18%, and those for the treated (ATET) range between 24% and 26%. The ATEU are particularly informative as they suggest the potential gains of a future legalization for workers, most of whom would have entered the US after the SAWs program. The relevant support for each marginal treatment effect (MTE) is given by the propensity score frequencies for foreign workers who were tr eated (legalized) and un treated (not legalized)
112 shown in Figure 4-1. The marginal treatmen t effects generated by each method are shown in Figure 4-2 through Figure 4-5. The MTE is evaluated at values at which the propensity score zP and unobservable factors (du ) are equivalent (Heckman, Urzua and Vytlacil, 2006a; 2006b). Heckman and Vytlacil (2005) emphasize the role of the unobservable characteristics in the interpretation of the MTE: for sm aller values of the unobservables ( ud) (points closer to zero on the x axis), the MTE is the expect ed benefit for individuals who are more likely to participate in treatment and who would participat e even if the mean scale utility ( d(Z )) were small. Conversely, for larger values of ud the mean scale utility ( d(Z )) would have to be much larger to induce individuals participati on in treatment and they are less likely to participate. The MTE may also be interpreted as the mean gain for persons with observable characteristics ( X ) who would be indifferent between acquiring legal status or not, and may be viewed as a willingness to pay (WTP) measure if the outcomes are value ou tcomes (Heckman and Vytlacil, 2007a; 2007b). The latter interpretation of the MTE is useful given the findings depicted in Figures 4-2 through 4-5 which seem to conf lict with the positive sorting on the gains indicated by the average treatment effect parame ters. Figure 4-2, which is ba sed on the parametric method, suggests that the worker who became legalized (on account of having a low ud ) benefited less than the worker who was not le galized (on account of having a high ud). Although this is difficult to reconcile with the positive sorting on the gains indicated by the average treatment effect parameters, the WTP interpretation of th e MTE may offer reasonable explanation. The upward slope of Figure 4-2 would therefore suggest an increasing willingness to pay by workers who have larger unobservables that usually would make them less eligible fo r participation in the program. The increasing ud values may be indicative of idio syncratic enhanced productivity, and a larger willingness to pay for legal status in order to permit more options to earn better returns.
113 The MTEs generated by the nonparametric methods shown in Figures 4-3 through 4-5 are quite different from the parametric MTE. Again, the WTP interpretation may offer some explanation for the three different segm ents exhibited by the MTEs along the ud range. On the downward sloping segments, individuals who have lower unobservables are more likely to participate in a legalization program and exhibi t a large willingness to pay for legal status acquisition. Toward the middle segment of th e MTEs (~0.51) however, it is possible that workers are more difficult to categorize in terms of legal status on the basis of their observable and unobservable characteristics and therefore are the least willing to pay for legal status relative to other individuals. Individua ls with high unobservables fall within the upper segment of the MTE. They exhibit high willingness to pay for legal status, arguably because they have a lower likelihood of gaining legal status due to unfavorable unobservable characteristics. On account of the lower likelihood of becoming lega lized, these individuals are like ly to have fewer options for employment in other sectors but may possibly be more productive than their cohorts who have legal status for US employment. Policy Implications Much of the political warring over immigrati on reform stems from proposed legalization of unauthorized immigrants wh ether it would reward illegal behavior and encourage future illegal immigration or whether it would serve the nations interests better to adjust unauthorized workers to legal status to prevent shocks to the labor intensive industries th at mostly hire them. In addressing these issues, AgJOBS seeks to stabilize the crop farm workforce that is extremely vulnerable to immigration reform that may af fect labor supply, wages and labor costs. The average treatment effect on the untreated (ATEU) parameter offers key insight as to how earnings may be potentially affected by a lega lization program such as AgJOBS. Table 4-5 shows average earnings gains ranging from 0. 1002 to 0.1784, suggesting potential earnings
114 increases between 10% and 18% for unauthorized work ers that become adjusted to legal status. The findings are broadly consistent with previous work that assesse d the earnings implications of legalization. Is and Perloff (1995) estimated average wage increases of about 15% for unauthorized workers that are granted amnesty; if they were to become permanent residents however, their wages would increase by about 12 %. Iwai, Emerson and Walters (2006a) found that unauthorized workers who gain ed legal status would, in gene ral, earn higher wages. Wages would increase by as much as 31%: for exampl e, unauthorized workers who selected into temporary authorized status had wage increases between 6% and 31% after 2001.11 Such results suggest cost incr eases for farm employers of unauthorized workers. Given the large percentage of the farm workforce that is currently un authorized for US employment, the increased cost may be substantial for em ployers with large propor tions of unauthorized workers among their crews, and for whom labor co sts comprise significant portion of total cost. Employers may respond by using other producti on factors more intensively; Napasintuwong (2004) suggested that the degree of intensity to which capital and labor are used in agriculture have been affected by the availability of im migrant labor, which is in turn affected by immigration policy. Concluding Remarks This study sought to analyze the potential im pact of proposed legalization on the wage outcomes of foreign farm workers. The results provi de some insight as to how future legalization could impact farm wages and, by extension, la bor costs for employers. The key distinction between this study and previous work is analyt ical framework: this study approached the problem from a treatment effects perspective wi th legalization modeled as a policy intervention 11 These are based on specific simulations that account for lo cation, time, payment type, etc. See Iwai, Emerson and Walters (2006a) for details.
115 or treatment. The application of this analytical framework is a signifi cant contribution to the literature on immigration policy eval uation and farm labor markets as it had not been previously applied in this context. The study also assumed that essential heterogeneity existed, meaning that workers would not only displa y different responses to treatmen t but would also select into treatment based on idiosyncratic gains. The te st of essential heterogeneity suggested by Heckman, Urzua and Vytlacil ( 2006a) was used to show that this assumption was indeed supported by the data. The results show an overall positive impact of legalization on farm worker outcomes. There is positive sorting on the gains from lega lization, implying that foreign workers who specifically sought legalization benefited more than the average worker and even more so than their cohorts who had not been legalized. The ma gnitude of gain is sensitive to the method of estimation used, with modest increases noted for the parametric method relative to the nonparametric methods. The findings from the marg inal treatment effects are not entirely clear, and seem to conflict in most respects with the av erage treatment effect results. Given the stark differences between the parametric and nonpara metric methods, it would appear that the assumption of joint normality (on which the parame tric method is based) is not supported by the data. However, this is not to imply th at the nonparametric MTEs offer less ambiguous interpretations. As they are presently, they ar e somewhat difficult to reconcile with the findings suggested by the positive sorting gains suggested by the average treatment effect parameters. If the MTEs are viewed in the c ontext of willingness to pay measur es however, the interpretations seem reasonable. Clearly, these findings suggest a need for future research. A likely starting point would be additional refinement of estim ates based on the nonparametric methods as the normality assumption appears problematic.
116 Most importantly, the results show that una uthorized workers may potentially gain from future legalization, with wage increases by as much as 18%. In this respect, the cost implications for farm employers are clear in that labor costs would increase if amnesty were to be granted to workers who are currently unauthorized. Whethe r this may encourage employers to shift to more capital intensive methods of production ov er time would depend on the magnitude of the cost increase and the degree of stringency and e ffectiveness of future legislation in controlling illegal immigration and employment.
117 Table 4-1. Explanatory variables of the c hoice and parametric wage regression models Variable a Definition LnWage Natural logarithm of the real farm wage in 2006 dollars. Conversions from the nominal wage were made using the consumer price index for all urban consumers Legal status =1 if farm worker is authorized for U.S. employment (citizen, permanent resident, or has other work authorization) = 0 if otherwise (i.e. unauthorized) Piece rate = 1 if worker is paid by piece rate = 0 if otherwise (by the hour, hour/piece combination, or salary) Seasonal worker =1 if worker is employed on a seasonal basis = 0 if otherwise (year-round) Female =1 if female =0 if male Mexican = 1 if worker is of Mexican nationality =0 if otherwise Education Highest grade level of educati on completed by the farm worker, ranging from 0 to 16 Adult educationb =1 if worker had attended any adu lt education classes or school in the U.S. =0 if otherwise After 1986 Dummy variable reflecting years before and after 1986 when foreign workers entered the United States for the first time to live or work After 2001 Dummy variable reflecting the interview years following September 2001 California (CA) Dummy variable reflecting employment in California at the time of the interview English (speaking ability) = 1 if none at all = 2 if a little = 3 if somewhat = 4 if well Married = 1 if married/living together =0 if otherwise
118 Table 4-1. Continued. Variable Definition Years with current employer Number of years of employment wo rker has completed with current employer. One year is measured as one or more days per year (NAWS) Farm work weeks Farm work weeks completed in the last year Foreign farm work experience =1 if worker had been employed in agriculture, either full-time or part-time, while living in native (foreign) country =0 if worker had been employed in non-agricultural sector or had never worked while living in native (foreign) country Grower = 1 if employed by a grower = 0 if employed by a farm labor contractor Age Respondent age in years Age2 Age squared Experience Years of U.S. farm work Experience2 Experience squared Farm work in the last year Months of US farm work in the previous year (prior to work grid estimate) Weeks spent abroad Number of weeks abroad last year a Data were sourced from the National Agricultural Workers Survey (1989-2006). Definitions enclosed in quotation marks are as they appear in the NAWS Codebook. b This would include English/ ESL, citizenship, literacy, job training and Adult Basic Education classes, GED/high school equivalency classes, college or university classes, and Even Start and Migrant Education classes.
119 Table 4-2. Summary statistics of fo reign farm workforce, NAWS, 1989-2006 N =19152 workers N1=8097 workers N0=11055 workers Variable Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Lnwage 2.0351 0.25132.08760.26031.9966 0.2372 Weeks Spent Abroad 6.6531 12.28873.79748.44208.7447 14.1097 Specialty Crop 0.7986 0.40110.81760.38620.7846 0.4111 Adult Education 0.2007 0.40050.28120.44960.1417 0.3487 After 1986 0.6148 0.48670.25770.43740.8763 0.3293 After 2001 0.5198 0.49960.47630.49950.5517 0.4973 Female 0.1615 0.36800.18600.38910.1436 0.3507 Married 0.6314 0.48240.78660.40970.5177 0.4997 English 1.7231 0.82782.00420.90551.5172 0.6974 Mexican 0.8873 0.31630.86970.33660.9001 0.2998 Education 5.9630 3.28265.75993.40466.1118 3.1821 Experience 10.4756 9.069716.69729.15725.9188 5.6548 Age 33.7474 11.815239.980911.351329.1819 9.9084 Farm Work Done Last Year 7.4226 4.31968.62843.33346.5394 4.7268 Years Since Immigration 11.9724 10.098419.12509.47636.7336 6.7813 Year With Employer 4.5438 4.85796.80926.11752.8846 2.6388 California 0.3957 0.48900.50810.50000.3134 0.4639 Grower 0.8003 0.39980.81850.38550.7871 0.4094 Piece Rate 0.1839 0.38750.16010.36670.2014 0.4011 Seasonal Worker 0.6873 0.46360.66980.47030.7002 0.4582 Farm Work Weeks 36.4671 14.554137.799812.860235.4910 15.6077 Foreign Farm Work Experience 0.6830 0.46530.66840.47080.6936 0.4610
120 Table 4-3. Probit model estimates for legal status treatment Variable Parameter Estimatea Standard Error Constant -0.5119*** 0.0733 Farm Work Weeks -0.0093*** 0.0010 Years with Employer 0.0680*** 0.0056 English 0.2464*** 0.0134 Years since Immigration 0.0563*** 0.0026 After 2001 -0.3544*** 0.0276 Weeks Spent Abroad -0.0192*** 0.0013 After 1986 -0.7906*** 0.0402 a Triple asterisks (***) indicate st atistical significance at 1% level. Table 4-4. Estimated parameters from parametric wage regressions for treated and untreated groups Authorized Status (Treated)U nauthorized Status (Untreated) Variable Parameter a Estimate Standard Error Parameter b Estimate Standard Error Constant 1.68363***0.044381.73827*** 0.01594 Age 0.00821***0.002050.00122 0.00083 Age sq. -0.00012***0.00002-0.00003** 0.00001 Farm Work Last Year 0.00530***0.000860.00386*** 0.00055 Experience 0.00328***0.001110.00704*** 0.00108 Experience sq. -0.00005**0.00002-0.00019*** 0.00004 English 0.03114***0.004670.01498*** 0.00417 Female -0.06034***0.00785-0.04470*** 0.00555 Piece Rate 0.21334***0.009920.18298*** 0.00746 Grower 0.07502***0.006510.05197*** 0.00487 Seasonal Worker -0.04230***0.00551-0.01224*** 0.00359 Education 0.00455***0.000950.00534*** 0.00059 Foreign Farm Work Experience 0.001890.005690.01565*** 0.00436 After 2001 0.11875***0.004960.06282*** 0.00319 California 0.04137***0.004470.04523*** 0.00408 Rho 0.04599***0.007990.03842*** 0.00799 a, bTriple and double asterisks (***, **) indicate statistical significance at 1% and 5% levels of significance, respectively.
121 Table 4-5. Treatment effects of legalization Parametera Parametricb Method Polynomial Methodc Nonparametric Method Id Nonparametric Method IIe ATET 0.1043 0.23850.2635 0.2538 ATEU 0.1002 0.17840.1459 0.1616 ATE 0.1020 0.20690.1978 0.2031 Sorting gain (ATET-ATE) 0.0023 0.03160.0657 0.0507 a A test for essential heterogeneity in the treatment effect s yielded an F-statistic (p value) of 18.19 (0.0000), indicating self-selection arising from heterogeneous and unobserved gains for individuals in the sample (See Heckman, Urzua and Vytlacil, 2006). bThe extent of selection bias is gaug ed with a comparison of the OLS and parametric model results: selection bias = OLS-ATET= 0.0359-0.1043= -0.0684. It show s that the OLS estimate of the average effect of legalization on earnings is downwar d biased, indicating a 3.6% av erage earnings gain relative to the 10% average gain suggested by the ATET estimate in the parametric method. Overall bias (OLS-ATE) is 0.0661. cThe outcome equation is estimated as a polynomial in the propensity score (Hec kman, Urzua and Vytlacil, 2006). dThis is the LIV estimator from Heckman and Vytlacil (2001, 2005). eThis method combines the nonparametric I and the polynomial approach. See Appendix for the estimated parameters for the polynomial and nonparametric methods.
122 0 2 4 6 8 10 120.010.110.210.310.410.510.610.710.810.91 pDensity Untreated ( D=0 ) Treated ( D=1 ) Figure 4-1. Frequency of Pr opensity Score by Legal Status
123 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.010.110.210.310.410.510.610.710.810.91UdMTE MTE Lower CI Upper CI Figure 4-2. Marginal treatment effect (MTE) of legalization for foreign farm workers (with 95% confidence intervals), parametric method
124 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 0.010.110.210.310.410.510.610.710.810.91UdMTE MTE Lower CI Upper CI Figure 4-3. Marginal treatment effect (MTE) of legalization for foreign farm workers (with 95% confidence intervals), polynomial method
125 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.010.110.210.310.410.510.610.710.810.91UdMTE MTE Lower CI Upper CI Figure 4-4. Marginal treatment effect (MTE) of legalization for foreign farm workers (with 95% confidence intervals), nonparametric method I
126 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.010.110.210.310.410.510.610.710.810.91UdMTE MTE Lower CI Upper CI Figure 4-5. Marginal treatment effect (MTE) of legalization for foreign farm workers (with 95% confidence intervals), nonparametric method II
127 CHAPTER 5 CONCLUSIONS Summary Several contemporary issues on US farm la bor markets and immi gration policy were explored in this study. Specifically, the study ev aluated how farm labor market outcomes have changed with the increasing presence of foreign workers, it assessed the implications of legal status for unauthorized workers wages and empl oyment, and it evaluated the potential impact of immigration policy reform specifically legalizati on for farm workers earnings. On the latter issues, the study departs from previous work primarily on account of the empirical methods used, and in so doing contributes significantly to the existing literature on farm labor markets. This chapter summarizes the major findings that emer ged from the different essays. It briefly discusses the research issues th at were encountered during the re spective analyses and sets forth certain points that may warrant additional research in the future. The first essay provided an historical contex t of the linkages between U.S. immigration policy and U.S. farm labor markets, and focu sed on how labor market outcomes had evolved following the passage of immigration policies an d programs, most importantly the Immigration Reform and Control Act (IRCA) of 1986. Revi ews of previous research and findings from descriptive statistics based on the National Agricultural Workers Survey (NAWS) were used to characterize the farm labor market between 1989 and 2006. Although IRCAs stated objectives were to curtail illegal immigration and unauthorized employ ment, the evidence suggests that it has fallen short in these respects. The literature identifies lax enforcement of employer sanctions and the proliferation of fraudulent work authoriz ation documents as key factors that undermined the efficacy of the legislation. The significance of these factors for effective immigration reform is not lost on lawmakers; in fact, these issu es are addressed extensively in current reform
128 proposals that seek to improve sanctions enforcement and cu rtail the use of counterfeit documents in the farm labor market. The first essay highlighted the critical linkages between specialty crop agriculture, the farm labor market and US immigration policy. In some respects, the US specialty crop sector is unique in the sense that a majority of the work force is predominately foreign and unauthorized for US employment, yet are employed in critical ar eas of production such as harvesting. Clearly this underscores the vulnerability of the sector to immigration policy changes that may decrease labor availability, particularly in light of the f act that there are few mech anical options available for application to fresh market production. An interesting finding to emerge from this essay was the impact of IRCAs legalization programs on farm employment. Critics of th e program had argued that legalization would adversely impact employment duration, yet this di d not occur. Several st udies reported increased duration of farm employment, and this was also confirmed by the summary duration statistics. Legal status was shown to positively affect employment and earnings, wherein authorized workers had more consecutive days of farm em ployment, were more likely to be employed at skilled tasks, and earned higher hourly wages than their unauthorized cohorts. In sum, the findings of the fi rst essay highlight the influent ial role of US immigration policy on the US farm labor market and its de velopment over time. Guest worker programs authorized by the various policies made it possi ble for employers to source farm labor from overseas, arguably because domestic laborers were unavailable. In the case of IRCA however, lack of synergism with the enforcement-oriented policy instruments has allowed for the development of a farm labor market that is mostly foreign born and unauthorized for US
129 employment. An unfortunate cons equence is that the market is therefore very vulnerable to changes in immigration policy that may affect labor availability and labor costs for employers. The second essay focused on two issues that are often raised in support of legalization for unauthorized workers: (1) whether unauthorized farm workers are more likely to be observed in unskilled farm jobs than their authorized cohorts and, (2) whether they earn less than authorized farm workers for the same type of work. This study offered a novel empirical approach in that selection bias was assumed to arise from multiple decision criteria and not a single criterion set forth by previous studies. This is an important methodological contributi on to the literature. Overall, the findings support the aforementioned hypotheses. The results indicated that indeed the multiple decision criteria specified (leg al status and job type) jointly affect foreign workers choices on U.S. farm employment. Foreign workers were likely to select into US employment as unauthorized (authorized) workers in unskilled (skilled) employment. Education, age and farm work experience, marital status, ge nder (female) and English speaking ability were reported as key characteristics that influenced employment in an authorized status, whereas US and foreign farm work experience, English spea king skills, education and employment with a grower were some of the key characteristic s associated with skilled employment. The wage regression models found significan tly positive effects for education and experience on earnings for all workers. Similar effects were noted for employment with a grower, employment in California, English speaki ng ability, and the piece rate. The latter was reported as having a particularly st rong effect relative to the other ex planatory variables. In most instances, variables denoting seasonal employment and female gender had significantly positive effects on farm worker earnings.
130 The predicted earnings results indicated that legal status was the most important consideration for foreign workers who chose US employment. Assuming that foreign workers view U.S. employment as an optimal econom ic choice relative to employment opportunities elsewhere, employment in an una uthorized status adversely impact s earnings. It is particularly disadvantageous if the unauthorized workers were employed at skilled positions since they would earn substantially less than their authorized cohorts at the same positions. This is because unauthorized workers employed in skilled positi ons are viewed as considerable risks for employers. If these workers were apprehended by the immigration authorities, employers could not only incur significant financ ial penalties, but also expe rience major production losses, particularly if the workers held key positions at th e farm operation. Interestingly, the results also indicated that unauthorized sk illed workers would earn less than other workers who were similarly unauthorized but employed at unskilled positions. Legalization is arguably the most contentious is sue of the reform process; as a matter of fact, it is perhaps the political wrangling and natio nal dissension on this particular issue that has stymied passage of comprehensive immigration reform to date. The potential effects of legalization on the wage outcomes fo r foreign farm workers were addressed in the third essay. A treatment effects framework was used. Essentia l heterogeneity was found to exist in the data, implying that foreign farm work ers wage outcomes were non-cons tant and that their selection into legal status was influenced by perceive d idiosyncratic gains. The overall methodology employed is a significant contribution to th e literature on the earni ngs implications of immigration reform in light of the fact that it has not been used in the context of farm labor before. The results are intended to provide some in sight as to the potential effect of legalization under AgJOBS, if it were passed.
131 Legalization was shown to increase earnings of foreign farm workers, with positive sorting on the average gains from legalization. This is consistent with theoretical expectations where the individuals that specifically opt for legaliza tion benefit more than the average worker and unauthorized workers. In the context of this st udy, the average treatment effect on the untreated (ATEU) offers significant insight as to the pote ntial gains for foreign wo rkers who are currently unauthorized, but who may acquire legal status in the future. Unauthorized workers may experience wage increases of as much as 18%. The cost implications ar e evident in that an amnesty in the future would definitely increase labor costs for farm employers. The extent to which the increased labor cost is offset by reduce d risk to the employer was not evaluated in this essay. In lieu of current m echanical applications for the fresh market, legalization of unauthorized workers could pose significant financia l challenges for farm employers in the short run. Research Issues and Suggestions for Future Research Construction of the variable denoting skilled employment status posed some challenges during the analysis. The data shows the vast ma jority of farm workers as being employed at unskilled tasks (pre-harvest, harvest, post-harvest) with few observations on the supervisory category, which is typically assumed as denoting skilled status. The semi-skilled task category was taken to reflect skilled empl oyment. It was difficult to obtain reliable predictions for the skill categories (unskilled/skilled) based on the bivariate probit model used in the first stage of the analysis in essay two, and as such wage simulations of different scenarios could not be completed. Further inspection of the data indi cated myriad categories of farm work (broadly categorized in the NAWS as pre-harvest, harves t, post-harvest, semi-skilled, supervisory and other), some of which could not be clearly assigned to either the skilled or unskilled category based on given descriptions. It would seem that certain improvements are necessary in this
132 regard such that future analysts may gain bett er appreciation for the distinctions between the tasks. In essay three, the sizeable differentials in the average gains to legalization and the dissimilarities in the MTE cu rves between the parametric and nonparametric methods would seem to suggest that the normality assumption (assumed in the parametric method) is not supported by the data. The interpretations suggested by the MTEs are also not entirely consistent with the positive sorting suggested by the average treatment effect parameters. This may suggest need for additional refinement on the underlying choice model, in which context the issue of freedom of choice on legalizati on by foreign workers must be looke d at more carefully. It may present some limitations on the specifications th at are reasonable for the choice model. Finally, it would be useful if data sets with legalization information were available on other sectors besides agriculture. This would permit future analysts to better gauge comparative advantage of legal status, particularly as worker s move between sectors of the economy in search of lucrative employment opportuniti es. Since agricultural employment is viewed as an option of last resort for workers with legal status, any in formation that would enab le analysts to better appreciate the significance of legal status would be valuable for the policy process.
133 APPENDIX POLYNOMIAL AND NONPARAMETR IC WAGE RE GRESSIONS Table A-1. Beta coefficients and standard errors for the outcome equations estimated by polynomial, nonparametric I and nonparametric II methods Polynomial Method Nonparametric Method I Nonparametric Method II Variablea CoefficientS.Error CoefficientS.Error Coefficient S.Error Constant 1.791820.02072------------Age -0.001150.00117-0.001040.00118-0.00115 0.00117 Age sq. 0.000010.000020.000010.000020.00001 0.00002 Farm work last year 0.001920.000720.001800.000740.00192 0.00072 Experience 0.011890.001860.012820.001890.01189 0.00186 Experience sq -0.000590.00010-0.000650.00010-0.00059 0.00010 English -0.003920.00530-0.004180.00533-0.00392 0.00530 Female -0.039700.00842-0.039970.00827-0.03970 0.00842 Piece rate 0.177840.009010.177860.009000.17784 0.00901 Grower 0.041050.006330.040850.006320.04105 0.00633 Seasonal worker 0.000110.004980.000330.004990.00011 0.00498 Education 0.006000.000840.005970.000840.00600 0.00084 Foreign farm work Experience 0.006200.005350.006300.005360.00620 0.00535 After 2001 0.060190.006830.059840.006500.06019 0.00683 California 0.049730.003760.049650.003760.04973 0.00376 Age*pscore 0.011880.002850.011670.002860.01188 0.00285 Age sq*pscore -0.000160.00003-0.000160.00003-0.00016 0.00003 Farm work last year*pscore 0.005210.001640.005320.001680.00521 0.00164 Experience*pscore -0.005840.00251-0.006520.00242-0.00584 0.00251 Experience sq*pscore 0.000470.000100.000540.000100.00047 0.00010 English*pscore 0.037460.007460.039620.007450.03746 0.00746 Female*pscore -0.030080.01768-0.029010.01753-0.03008 0.01768 Piece rate*pscore 0.042760.019600.042780.019530.04276 0.01960 Grower*pscore 0.045550.014800.046660.014750.04555 0.01480 Seasonal Worker*pscore -0.047030.01385-0.048180.01388-0.04703 0.01385 Education*pscore -0.002540.00172-0.002410.00174-0.00254 0.00172 Foreign farm work Experience*pscore 0.008370.011510.007400.011520.00837 0.01151 After 2001*pscore 0.069550.015410.070200.015070.06955 0.01541 California*pscore -0.011120.00848-0.011160.00844-0.01112 0.00848 Pscore -0.150460.10896------------Pscore2 0.736970.44991------------Pscore3 -2.209580.61096------------Pscore4 1.479090.28288------------aPscore denotes the propensity score, which is the probability of becoming legalized.
134 LIST OF REFERENCES Abowd, J., and H. Farber. 1982. Job Queues and the Union Status of Workers. Industrial and Labor Relations Review 35: 354. Anderson, S. 2003. The Impact of Agri cultural Guest Worker Programs on Illegal Immigration. The National Foundation fo r American Policy. Available at: http://www.nfap.net/researchactivities/studies/Nov_study1.pdf (Accessed August 2007). Basu A, Heckm an J, Navarro-Lozano S, Urzua S. 2007. Use of Instrumental Variables in The Presence of Heterogeneity and Self-Sel ection: An Application to Treatments of Breast Cancer Patients, Health Economics 16: 1133-1157. Bjrklund, A. and R. Moffitt. 1987. The Estima tion of Wage Gains and Welfare Gains in Self Selection Models. The Review of Econom ics and Statistics 69(1): 42-49 Blundell, R. and M. Costa Dias. 2002. Altern ative Approaches to Evaluation in Empirical Microeconomics. Portuguese Economic Journal 1: 91-115. Briggs, V.M. 1983. "Non-Immigrant La bor Policy in the United States." Journal of Economic Issues 17(3): 609-630. Briggs, V. M. 1986. "The Albatross of I mmigration Reform Temporary Worker Policy In the United-States." International Migration Review 20(4): 995-1019. Briggs, V. M. 2004. "Guestworker Programs: Lessons from the Past and Warnings for The Future." Backgrounder, Center for Immigration Studies. Available at: http://www.cis.org/articles/2004/back304.pdf (Accessed August 2007). Bruno, A. 2006. Imm igration: Policy Cons iderations Related to Guest Worker Programs. Domestic Social Policy Division, Congressional Research Service, Library of Congress. Available at: http://trac.syr.edu/immi gration/library/P333.pdf (Accessed September 2007). Brownell, P. 2005. "The Declining Enforcement of Employer Sanctions." Migration Information Source. Available at: http://www.migrationinformation. org/Feature/display.cfm ?id=332 (Accessed August 2007). Caliendo, M. 2006. Microeconometric Evaluation of Labor Market Policies Berlin: Springer. Cameron, A. and P. Trivedi. 2005. Microeconometrics: Me thods and Applications New York: Cambridge University Press.
135 Carroll, D., R. Samardick, S. Bernard, S. Gabbard, T. Hernandez. 2005. Findings from The National Agricultural Workers Survey (NAWS) 2001 2002: A Demographic and Employment Profile of United St ates Farm Workers. U.S. De partment of Labor, Office of the Assistant Secretary for Policy, Office of Programmatic Policy, Research Report No. 9. Available at: http://www.doleta.gov/agworker/report9/naws_rpt9.pdf (Accessed June 2006). Chiswick, B. R. 1988. "Illegal Im m igration and Immigration Control." Journal of Economic Perspectives 2(3): 101-115. Chiswick, B.R. 1999. Are Immigrants Favorably Self Selected? American Economic Review 89(2): 181-185. Chiswick, B.R. 2003. Jacob Mincer, Experien ce and the Distribution of Earnings. Review of Economics of the Household 1: 343-361. Codebook for NAWS Data. 2005. Cycles 1-49, Release 3.0, April. Congressional Research Service (CRS). 1980. Temporary Worker Programs: Background and Issues. A Report prepared at the request of Sen. Edward M. Kennedy, Chairman, Committee on the Judiciary, U.S. Senate, for the use of the Select Commission on Immigration and Refugee Policy. U.S. Government Printi ng Office, Washington D.C: Library of Congress, 96th Congress, 2nd Session, February. Craig, L. 2006. Summary of the Provisions of the Agricultura l Job Opportunities, Benefits and Security Act of 2006 (AgJOBS) Included in the Bill Reported Out of the Senate Judiciary Committee on March 27, 2006. Available at: http://craig.senate. gov/tpt_agjobs2006.pdf (Accessed December 2007). Duffield, J.A., M.J. Morehart, and R. Coltr ane. 1989. Labor Expenditures Help Determ ine Farms Affected by Immigration Refo rm. U.S. Department of Agriculture, Economic Research Service, Agriculture Information Bulletin, No. 557. April. Economic Research Service. 2005. Available at: http://www.ers.usda.gov/ (Access ed November 2005). Effland, A. and J. Runyan. 1998. "Hired Farm Labor in U.S. Agriculture." Agricultural Outlook. Economic Research Service, U.S. Department of Agricu lture, Washington D.C: 19-22, October. Emerson, R.D. 2000. Changes in the Southeastern Farm Labor Market. Food and Resource Economics Department, University of Florida. Paper prepared for The NAWS at 10: A Research Seminar at the University of California-Davis, Oc tober. Available at: http://migration.ucdavis.edu/rmn/ ChangingFace/cf_nawsat10/Em erson.pdf (Accessed December 2006).
136 Emerson, R.D., & O. Napasintuwong. 2002. F oreign Workers in Southern Agriculture. Paper presented at the SAEA Annual Mee ting, Orlando, FL, February 3-6. Emerson, R.D. 2007. Agricultural Labor Markets and Immigration. Choices 22(1): 57-66. Greene, W. 1981. "Sample Selection Bias as a Specification Error: Comment." Econometrica 49: 795-798. Greene, W. 2000. Econometric Analysis 4th Edition. New Jersey: Prentice-Hall Inc. Gronau, R. 1974. Wage Compar isons: A Selectivity Bias. Journal of Political Economy 82: 1119-1149. Hashida, E., & Perloff, J.M. 1996. Duration of Agricultural Employme nt. Department of Agricultural and Resource Economics. Work ing Paper 779. University of California, Berkeley. Available at: http://agecon.lib.umn.edu/cgi-bin/pdf_view.pl?paperid=72 (Accessed March 2007). Heckm an, J. 1974. ShadowPrices MarketWages and Labor Supply. Econometrica 42: 679. Heckman, J. 1976. The Common Structure of Statistical Models of Truncation, Sample Selection and Limited Dependent Variables and a Simple Estimator for Such Models. Annals of Economic and Social Measurement 5: 475. Heckman, J. 1979. "Sample Selection Bias as a Specification Error." Econometrica 47(1): 153161. Heckman, J. 2001. Micro Data, Heterogeneity and the Evaluation of Public Policy: Nobel Lecture. Journal of Political Economy 109: 673-748. Heckman, J. and B. Honor. 1990. The Empirical Content of the Roy Model. Econometrica 58(5): 1121-1149. Heckman, J and X. Li. 2004. Selection Bias Comparative Advantage and Heterogeneous Returns to Education: Evidence from China in 2000. Pacific Economic Review 9(3): 155-171. Heckman, J. and E. Vytlacil. 2005. S tructural Equations, Treatment Effects and Econometric Policy Evaluation. Econometrica 73(3): 669. Heckman, J. and E. Vytlacil. 2007a. Econometric Evaluation of Social Progr ams, Part I: Causal Models, Structural Models and Econometric Po licy Evaluation. In J. Heckman and E. Leamer, eds. Handbook of Econometrics Vol 6, Part 2, Chapter 70. New York : NorthHolland Publishing Co.
137 Heckman, J. and E. Vytlacil. 2007b. Econometric Evaluation of Social Programs, Part II: Using the Marginal Treatment Effect to Organize Alternative Econometric Estimators to Evaluate Social Programs and to Forecast Th eir Effects in New Environments. In J. Heckman and E. Leamer, eds. Handbook of Econometrics Vol 6, Part 2, Chapter 71. New York : North-Holland Publishing Co. Heckman, J., L. Lochner, and P. Todd 2003. "Fifty Years of Mincer Earnings Regressions. IZA Discussion Paper No. 775. Available at: ftp://repec.iza.org/RePEc/Discussionpaper/dp775.pdf (Accessed July 2007). Heckman, J., J. Tobias and E. Vytlacil. 2001. F our Parameters of Interest in the Evaluation of Social Programs. Southern Economic Journal 68: 210-223. Heckman, J., J. Tobias, and E. Vytlacil. 2003. Simple Estimators for Treatment Parameters in A Latent Variable Framework. Review of Econom ics and Statistics 85:748-755. Heckman, J, S. Urzua and E. Vytlacil. 2006a. U nderstanding Instrumental Variables in Models with Essential Heterogeneity. The Review of Econom ics and Statistics 88(3): 389-432. Heckman, J, S. Urzua and E. Vytlacil. 2006b. Estimation of Treatment Effects under Essential Heterogeneity. Available at: http://jenni.uchicago.ed u/underiv/docum entation_2006_03_20.pdf (Accessed March 2007). Heckman, J, S. Urzua and E. Vytlacil. 2006c. U nderstanding Instrumental Variables in Models with Essential Heterogeneity. Available at: http://jenni.uchicago.edu/underiv/ Heppel, M. and D. Papadem etriou. 1999. Government Intervention and the Farm Labor Market: How Past Policies Shape Future Options. Forum for Transnational Employment, California Institute for Ru ral Studies, Davis. Available at: http://www.cirsinc.org/fte/Chapters3.pdf (Accessed August 2007). Is, S., and J. M. Perloff. 1995. "Legal Status and Earnings of Agricultural-W orkers." American Journal of Agricultural Economics 77(2): 375-386. Iwai, N., O. Napasintuwong, & R.D. Emerson. 2005. Immigration Policy And The Agricultural Labor Market: The Effect on Job Duration. Paper presented at the AAEA Annual Meeting, Providence, RI July 24-27. Available at: http://agecon.lib.umn.e du/cgibin/pdf_view.pl? paperid=18193&ftype=.pdf (Accessed May 2007). Iwai, N., R.D. Emerson and L.M. Walters. 2006a. Legal Status and U.S. Farm Wages. Paper presented at the SAEA A nnual Meeting, Orlando, FL, February 5-8. Available at: http://agecon.lib.umn.edu/cgi-in/ pdf _view.pl?paperid=19740&ftype=.pdf (Accessed May 2007).
138 Iwai, N., R.D. Emerson and L.M. Walters. 2006b. Farm Employme nt Transitions: A Markov Chain Analysis with Self-Selectivity. Paper presented at the AAEA Annual Meeting, Long Beach, CA, July 23-26. Available at: http://agecon.lib.umn.edu/cgibin/pdf_view.pl? paperid=23093&ftype=.pdf (Accessed June 2007). Krikorian, M. 2001. "Guest Worker Programs: A Threat to American Agriculture." Backgrounder. Center for Immigr ation Studies, Washington D.C. Available at: http://www.cis.org/ar ticles/2001/back801.pdf (Accessed Decem ber 2006). Krikorian, M. 2004. "Flawed Assumpti ons Underlying Guest Worker Programs." Backgrounder. Center for Immigration Studies, Washington D.C. Available at: http://www.cis.org/articles/2004/back204.pdf (Accessed Decem ber 2006). Lee, L. 1978. Unionism and Wage Rates: A Simultaneous Equations Model with Qualitative And Limited Dependent Variables. International Economic Review 19(2): 415-433. Lee, M. 2005. Micro-Econometrics for Policy, Program and Treatment Effects New York: Oxford University Press Inc. Lee, L.F. and G.S. Maddala. 1994. Sequential Selection Rules and Selectivity in Discrete Choice Econometric Models. In G.S. Maddala, ed. Econometric Methods and Applications Volume II England: Edward Elgar Publishing Limited, pp. 311-329. Legal Services Corporation Office of Inspector General (LSC-OIG). 2007. Immigration Reform And Control Act (IRCA). Available at: http://www.oig.lsc.gov/legis/irca86.htm (Accessed June 2007). Levine, L. 2004. Farm Labor Shortages and Immigration Policy. Congressional Research Service Report, No. RL30395, 2004. Available at: http://www.ncseonline.org/NLE/CRSreports/04Feb/RL30395.pdf (Accessed Ma rch 2007). Lewis, E. 2007. The Impact of Immigr ation on American Workers and Businesses. Choices 22(1): 49-54. Library of Congress. 1986. S.1200: Title : A Bill to Amend the Immigration And Nationality Act to Effectively Control Unaut horized Immigration into the United States, And For Other Purposes. Available at: http://thomas.loc.gov/cgibin/bdquery/z? d099:SN01200:@@@L&summ2=m&|TOM:/bss/d099query.html| (Accessed August 2007). Library of Congress. 2007. S.1348: Comprehe nsive Immigration Reform Act of 2007. Available at: http://thomas.loc. gov/cgi-bin/thom as (Accessed December 2007).
139 Maddala, G.S. 1983. Limited Dependent and Qualitative Variables in Econometrics Econometric Society Monographs Se ries, No. 3. New York: Cambridge University Press. Martin, P. 1990. "Harvest of Confusion: Im migration Reform and California Agriculture." International Migration Review 24(1): 69-95. Martin, P. 1994. "Good Intentions Gone Awry: IRCA and U.S. Agriculture." Annals of the American Academy of Political & Social Science 534: 44-57. Martin, P. 2000. "Guest Worker Programs for the 21st Century." Backgrounder. Center for Immigration Studies, Washington D.C. Available at: http://www.cis.org/articles/2000/back400.pdf (Accessed March 2007). Martin, P. 2001. "There is Nothing More Perm anent than Temporary Foreign Workers." Backgrounder, Center for Immigration Studi es. Washington D.C. Available at: http://www.cis.org/articles/2001/back501.pdf (Accessed March 2007). Martin, P. L., and J. E. Taylor. 1990. "The Initial Effects of Immigration Refor m on FarmLabor in California." Population Research and Policy Review 9(3): 255-283. Mincer, J. 1974. Schooling, Experience, and Earnings Human Behavior and Social Institutions, 2. New York: Columbia University Press. Mines, R., S. Gabbard, and A. Steirman. 1997. A Profile of U.S. Farm Workers. Demographics, Household Composition, Income and Use of Services. U.S. Department of Labor, Office of the Assistant Secretar y for Policy, Office of Program Economics Research Report #6. Washington DC, April. Napasintuwong, O. 2004. Immigrant Workers and Technological Change: An Induced Innovation Perspective on Florida and United St ates Agriculture. Ph.D Dissertation, University of Florida. Available at: http://plaza.ufl.edu/onapasi/napasintuwong_o.pdf (Accessed March 2007). Napasintuw ong, O., & Emerson, R.D. 2004. Labor Substitutability in Labor Intensive Agriculture and Technological Change In The Presence of Foreign Labor. Paper presented at the AAEA Annual Meeting, De nver, CO, August 1-4. Available at: http://agecon.lib.umn.edu/cgi-bin/pd f_view.pl? paperid=14330&ftype=.pdf (Accessed June 2007). Oliveira, V. J. 1989. Trends in the Hired Farm Work Force, 1945-1987 U.S. Department of Agriculture, Economic Research Service, Agriculture Information Bulletin, No. 561, Washington DC, April. Oliveira, V. J., A. Effland, J. Runyan, and S. Hamm. 1993. Hired Farm Labor Use on Fruit, Vegetable and Horticultural Specialty Farm s." U.S. Department of Agriculture, Economic Research Service, Agricult ural Economic Report, No. 676.
140 Perloff, J. M., L. Lynch, and S. M. Gabbard 1998. "Migration of Seasonal Agricultural Workers." American Journal of Agricultural Economics 80(1): 154-164. Passel, J. 2005. Unauthorized Immigrants: Numbers and Characteristics, Pew Hispanic Center. Washington, D.C. Available at: http://pewhispanic.org/files/reports/46.pdf (Accessed May 2007). Passel, J. 2006. The Size and Charact eristics of the Unauthorized Migrant Population in the U.S. Pew Hispanic Ce nter. Washington, D.C. Available at: http://pewhispanic.org/files/reports/61.pdf (Accessed May 2007). Passel, J. and R. Suro. 2005. Rise, Peak and Decline: Trends in U.S. Imm igration: 19922004. Pew Hispanic Center. Available at: http://www.ilw.com/artic les/2005,1205-passel.pdf (Accessed May 2007). Polachek, S. W and W.S. Siebert. 1993. The Economics of Earnings New York, NY: Cambridge University Press. Polopolus, L. and R. D. Emerson. 1991. "Immigration and Employment Policy Impacts The Florida Citrus Case." Choices (4th quarter): 16-18. Roy, A.D. 1951. Some Thoughts on the Distribution of Earnings. Oxford Economic Papers 3 (2): 135-146. Rural Migration News. 2000. U.S. Departme nt of Labor Report to Congress: The Agricultural Labor Market Status and Recommendations. December. Available at: http://migration.ucdavis.edu/ rm n/word-etc/dec_2000_labor.htm (Accessed April 2007). Rural Migration News. 2001. H-2A, H-2B. Available at: http://migration.ucdavis. edu/rm n/more.php?id=528_0_4_0 (Accessed May 2007). Sarig, Y, J. Thompson, and G. Brown. 2000. "A lternatives to Immigrant Labor? The Status of Fruit and Vegetable Harvest Mechanization in the United States." Backgrounder, Center for Immigration Studie s, Washington D.C. Available at: http://cis.org/artic les/2000/back1200.pdf (Accessed May 2007). Sjaastad, L. 1962. Costs and Returns of Hum an Migration. Journal of Political Economy 70(5):80. Taylor, J. E. 1992. "Earnings and Mobility of Legal and Illegal Immigrant Workers in Agriculture." American Journal of Ag ricultural Economics 74(4): 889-896. Taylor, J. E., and D. Thilmany. 1992. "Calif ornia Farmers Still Rely on New Immigrants for Field Labor." California Agriculture 46 (September-October): 4-6.
141 Taylor, J. E., and D. Thilmany. 1993. "Worker Turnover, Farm-Labor Co ntractors, and IRCA's Impact on the California Farm-Labor Market." American Journal of Agricultural Economics 75(2): 350-360. The California Institute for Federal Polic y Research. 2003. House Agriculture Subcommittee Considers Specialty Crop Issu es. California Capitol Hill Bulletin Volume 10, Bulletin 33, November. Available at: http://www.calinst.o rg/bulletins/b1033.pdf (Accessed August 2007). Thilm any, D. D. 1996. "FLC Usage among Ca lifornia Growers under IRCA: An Empirical Analysis of Farm Labor Market Risk Management." American Journal of Agricultural Economics 78 (4): 946-960. Thompson, G. D., and P. L. Martin. 1991. "I mmigration Reform and the Agricultural LaborForce." Labor Law Journal 42 (8): 528-536. Tran, L. H., and J. M. Perloff. 2002. "T urnover in US Agricultural Labor Markets." American Journal of Agricultural Economics 84(2): 427-437. Tunali, I. 1986. "A General Structure for Models of Double-Selection and an Application to a Joint Migrat ion/Earnings Process with Re -Migration." In Ronald G. Ehrenberg, ed. Research in Labor Economics, Vol. 8 (Part B). Greenwich: JAI Press, pp. 235-84. U.S. Department of Agriculture. 2007. C ooperative State Research, Education and Extension Service. Newsroom, News Re lease: USDA Hosts Specialty Crop Research Workshop. Available at: http://www.csrees.usda.gov/newsroom /news/2007news/specialty_crops.htm l (Accessed August 2007). U.S. Department of AgricultureNational Agricultural Statistics Service (USDA-NASS). 1978. 1978 Census of Agriculture. U.S. Department of Commerce, Bureau of the Census, Washington D.C. U.S. Department of AgricultureNational Agricultural Statistics Service (USDA-NASS). 2002. 2002 Census of Agriculture. Available at: www.nass.usda.gov/Census_of_Agriculture/index.asp (Accessed July 2007). U.S. Departm ent of Labor (US-DOL). 2006. Code of Federal Regulations Pertaining to ESA. Labor, Wage and Hour Division, Enfor cement of Contractual Obligations for Temporary Alien Agricultural Workers Admitted Under Section 21. Available at: http://www.dol.gov/dol/allcfr/ESA/ Title_29/Part_501/29CFR501.10.htm (Accessed March 2006).
142 U.S. Department of Labor (US-DOL). 2007. The National Agricultural Workers Survey. Available at: http://www.doleta.gov/agworker/naws.cfm (Acc essed January 2007). U.S. Senate Committee on the Judiciary, Subcommittee on Immigration. 1978. The West Indies (BWI) Temporary Alie n Labor Program, 1943-1977. U.S. Government Printing Office, Washington DC. Vella, F. 1998. "Estimating Models with Sample Selection Bias: A Survey." Journal Of Human Resources 33(1): 127-169. Walters, L.M., R.D. Emerson and N. Iwai. 2007. Implications of Proposed Immigration Reform for the U.S. Farm Labor Market. Paper pr esented at the SAEA A nnual Meeting, Mobile, AL, February 4-7, 2007. Available at: http://agecon.lib.umn.edu/cgibin/pdf_view.pl? paperid=25284&ftype=.pdf (Accessed May 2007). Wasem, R. E., and G. K. Collver. 2001. RL30852: Immigration of Agricultural Guest Workers: Policy, Trends, and Legislation. Congressi onal Research Service, Washington, DC: Available at: http://digital.l ibrary.unt.edu/govdocs/crs/perm alink/meta-crs-1382:1 (Accessed May 2007). Willis, R. J. 1986. "Wage Determinants: A Survey and Reinterpretation of Human Capital Earnings Functions." In O. Ashenfelter and R. Layard Eds. Handbook of Labor Economics, Vol. 1 The Netherlands: Elsevier Scien ce Publishers BV, pp. 525-602. Willis, R.J., and S. Rosen. 1979. Education and Self Selection. The Journal of Political Economy 87(5): S7-S36. Wooldridge, J. 2002. Econometric Analysis of Cr oss Section and Panel Data Massachusetts: The MIT Press.
143 BIOGRAPHICAL SKETCH Lurleen M. Walters was born on February 6, 1974 in Nevis, West Indies. Following several years teaching with the Nevis Island Adm inistration, she enrolled at Alabama A&M University in 1994 and graduated with a Bachelor of Science degree in agribusiness management in 1997. She earned her Master of Science degree in agribusiness from Alabama A&M University in 1999. She was employed as a Research Associate with the Department of Agribusiness at Alabama A&M University for several years. A graduate research assistantship at the University of Florida pr ovided her with several opportunities for research in the areas of agricultural labor, immigr ation policy, and international trad e and development, in addition to other professional experiences that served to enrich her academic training. Lurleen graduated with a Doctor of Philosophy degree in food and resource economics from the University of Florida in 2008.