Group Title: Working paper - International Agricultural Trade and Policy Center. University of Florida ; WPTC 05-06
Title: Immigration policy and the agricultural labor market : the effect on job duration
Full Citation
Permanent Link:
 Material Information
Title: Immigration policy and the agricultural labor market : the effect on job duration
Series Title: Working paper - International Agricultural Trade and Policy Center. University of Florida ; WPTC 05-06
Physical Description: Book
Language: English
Creator: Iwai, Nobuyuki
Napasintuwong, Orachos
Emerson, Robert D.
Publisher: International Agricultural Trade and Policy Center. University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: July 2005
 Record Information
Bibliographic ID: UF00089803
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.


This item has the following downloads:

wp050006 ( PDF )

Full Text

WPTC 05-06

I '-ional Agricultural Trade and Policy Center

Nobuyuki Iwai, Orachos Napasintuwong, & Robert D. Emerson

WPTC 05-06 July 2005





Institute of Food and Agricultural Sciences



The International Agricultural Trade and Policy Center (IATPC) was established in 1990
in the Institute of Food and Agriculture Sciences (IFAS) at the University of Florida
(UF). The mission of the Center is to conduct a multi-disciplinary research, education and
outreach program with a major focus on issues that influence competitiveness of specialty
crop agriculture in support of consumers, industry, resource owners and policy makers.
The Center facilitates collaborative research, education and outreach programs across
colleges of the university, with other universities and with state, national and
international organizations. The Center's objectives are to:

* Serve as the University-wide focal point for research on international trade,
domestic and foreign legal and policy issues influencing specialty crop agriculture.
* Support initiatives that enable a better understanding of state, U.S. and international
policy issues impacting the competitiveness of specialty crops locally, nationally,
and internationally.
* Serve as a nation-wide resource for research on public policy issues concerning
specialty crops.
* Disseminate research results to, and interact with, policymakers; research, business,
industry, and resource groups; and state, federal, and international agencies to
facilitate the policy debate on specialty crop issues.


Nobuyuki Iwai
International Agricultural Trade and Policy Center
Food and Resource Economics Department
PO Box 110240
University of Florida
Gainesville, FL 32611

Orachos Napasintuwong
International Agricultural Trade and Policy Center
Food and Resource Economics Department
PO Box 110240
University of Florida
Gainesville, FL 32611

Robert D. Emerson
International Agricultural Trade and Policy Center
Food and Resource Economics Department
PO Box 110240
University of Florida
Gainesville, FL 32611

Selected Paper prepared for presentation at the American Agricultural Economics Association
Annual Meeting, Providence, Rhode Island, July24-27, 2005

Copyright 2005 by Nobuyuki Iwai, Orachos Napasintuwong and Robert D. Emerson. All rights
reserved. Readers may make verbatim copies of this document for non-commercial purposes by
any means, provided that this copyright notice appears on all such copies.



The effects of immigration policy change on the agricultural labor market have received

much attention both economically and politically. The most important immigration policy change

in recent years for the agricultural labor market was the Immigration Reform and Control Act

(IRCA) of 1986. IRCA granted amnesty to a substantial number of undocumented agricultural

workers, entitling them to work legally in the United States. Just before the passage of IRCA,

many farmers and legislators expressed concern about its possible effect on the agricultural labor

market. Their prediction was that undocumented agricultural workers who received amnesty

would leave agriculture for other employment opportunities, which would lead to serious labor

shortages and wage increases in agriculture.1

Limited empirical work has been done on the relationship between legal status and farm

work duration (Hashida and Perloff 1996, Tran and Perloff 2002, and Emerson and

Napasintuwong 2002). Generally, these studies conclude that estimated durations for

documented, in contrast to undocumented, workers are significantly longer. Among these, the

most comprehensive study is Tran and Perloff (2002). Using the National Agricultural Workers

Survey (NAWS) data for the years 1987-91, Tran and Perloff estimate a stationary, first-order

Markov model of employment turnover, and calculate the steady-state probability for each

demographic group to work in agriculture. They conclude that "Predictions made when the 1986

* The authors are grateful to Susan Gabbard, Trish Hernandez, Alberto Sandoval and their associates at Aguirre
International for assistance with the NAWS data, and to Daniel Carroll at the U.S. Department of Labor for granting
access and authorization to use the NAWS data. This research has been supported through a partnership agreement
with the Risk Management Agency, U.S. Department of Agriculture; by the Center for International Business
Education and Research at the University of Florida; and by the Florida Agricultural Experiment Station. The authors
alone are responsible for any views expressed in the paper.
1 See Tran and Perloff (427-28) for a detailed discussion of industry and legislative concerns.


Immigration Reform and Control Act was passed that granting people amnesty would induce

most of them to leave agriculture were incorrect," (p. 427) and ". .... the steady state probability of

working in agriculture is higher for someone with amnesty than for an undocumented worker, so

that IRCA increased the long-run probability that people granted amnesty stayed in agriculture."

(p. 437)

However, this conclusion is a little problematic. As the authors mentioned in their work,

the portion of undocumented workers in the agricultural labor force grew substantially in the

1990s. In the sample (1987-91) used by Tran and Perloff, only 7% are undocumented workers.

According to the NAWS data, the portion of undocumented workers rose to 46% for the years

1995-98, and 48% for the years 2002-2004. This implies that there has been a large-scale inflow

of undocumented workers into the agricultural labor market and a large-scale outflow of

documented workers from it. The latter might mean that documented workers tend to leave

agriculture in the long-run: the opposite observation to their conclusion.

There are some concerns that might lead to statistical problems in their work. First, a data

sample (1987-91) is taken in a transitional period in the sense that workers granted amnesty might

not have had enough time to move to other industries. It is also a transitional period in another

sense that the legal status of many workers changed. The study is unable to control for this status

change using the observed status at the time of interview, the only legal status information

available in the NAWS data. As a result of the 1987-91 sample used, the study cannot capture the

major inflow of undocumented workers from foreign countries after IRCA and who have become

a major component of the labor force in agriculture. The most serious problem, however, is that

the study tries to estimate a probability matrix and a steady state for the whole migration process

using data from only a small sector (the agricultural labor market). Most migration for any status

of worker would be from non-agriculture to non-agriculture, and most would not work in

agriculture at all. It may be difficult to estimate the whole migration pattern without data from all

sectors. In this presentation, we present an alternative method (duration model with sample bias

correction) to estimate the effect of the legal status of a worker on duration in farm work. Based on

the existing studies which have used the duration model (Hashida and Perloff 1996, Emerson and

Napasintuwong 2002), we develop the Heckman-type two-stage method, with the ordered probit

model in the first stage and the duration model in the second stage.

The sample selection bias issue should be investigated first. Duration for a worker with a

legal status is observed only if the worker is in that legal status. Each foreign-born worker chooses

his/her legal status, considering conditions such as his/her individual demographic characteristics,

cost of application, and benefit of the status. Without correcting for this selection process, the

duration model will yield biased estimators. Hashida and Perloff (1996) correct selection bias

using Lee's extension of Heckman's two-stage sample selection method (Lee 1983). In the first

stage, the multinomial logit model is run to calculate a correction term assuming the error term has

a Gumbel distribution. The second-stage duration model with this correction term does not

generally yield consistent estimates with the normal distribution assumption of error term in the

duration model.2 We will use the ordered probit model in the first stage for two reasons: (1) this is

consistent with the assumption about the error term in duration model in the second stage and (2)

the multinomial logit does not account for the ordinal nature of the legal status. Considering the

advantages in the labor market, they can be ordered as "citizen, permanent resident, authorized,

and unauthorized workers."3

2 Lee's method yields consistent estimator under very restrictive condition (Bourguignon et al. 2004).
3 The definition of each legal status is given in the Data section.


Next, treatment of completed and uncompleted employment spells of workers should be

considered. Hashida and Perloff (1996) and Tran and Perloff (2002) use only completed spells,

while Emerson and Napasintuwong (2002) use only uncompleted spells. There are further

distinctions in how spells have been defined in the literature. Hashida and Perloff (1996) define

the duration variable as the average duration of completed spells of farm employment by a

worker. Tran and Perloff (2002) work with employment transitions among three types of spells:

agricultural employment, nonagricultural employment, and unemployed or abroad. They

recorded a transition on a monthly basis over a two-year work history among the three above

types of spells without regard to employer. Emerson and Napasintuwong (2002) define the

duration variable as the number of years reported working in U.S. agriculture. At this point our

estimation uses multiple completed spells per worker of agricultural employment at a single task.

Our current definition is closest to the one used by Hashida and Perloff (2002), and specifically

addresses variations in individual job duration by farm workers.


The basic structure of the Heckman-type two-stage method is specified with the ordered

probit model for the first stage and the duration model for the second stage. The ordered probit

model is used to explain the legal status of worker i as a function of the individuals' demographic

and policy variables (denoted as vector x ). A foreign-born worker's legal status (J,) takes on four

values: 0=unauthorized, authorized, 2= permanent resident (green card holder), 3=citizen.

With the familiar argument of latent regression (Greene 2003), we can assume that an unobserved

variable J* is censored as follows:

J, =0 if J, J, =1 if /0 J, =2 if /l < J, J, =3 if2,
where J,* = xla + e,; x, is a vector of exogenous characteristics of individual i; and 8, is a

disturbance term. The characteristics include gender, marital status, English speaking ability, race

(black, white, and other), ethnicity (Hispanic and other), age, age squared, education, education

squared, US farm experience, US farm experience squared, and the year of interview (before 1993,

after 2001, and in-between).4 We assume that E, is normally distributed with a mean of zero and a

standard deviation of ac. Then the likelihood function can be expressed as

FH W xPaO ] WIi [ If l )x- I aO 1
L(a, uc, |data)= (1)
fL [ K cj C' KfP1 x-nc P2j_ j
J,-2 (7, J, -3 (I ,

where O(.) indicates the cumulative distribution for the standard normal.

Suppose the cumulative distribution function of farm work duration (t,) for person i with

legal status is given as

Fj (t) = Pr(t, < t) .

We denote its density function as fj (t). The probability for the spell to be of length of at least t,

usually called the survival function, is given as

S(t) = 1 (t) .

4 See the Data section below for additional detail.

Suppose that the log of the spell is normally distributed with mean In r, and variance o- Then,

the survival function is expressed as

S' (t)= 1 n-ln

The hazard rate, the rate at which the spell is completed after duration t, is

In t _- In z,
h, (t) =
to-, S, (t)

where 0(-) is the probability density for the standard normal distribution. Next, we assume that

the mean duration of a spell (In ) depends on independent variables z, (gender, marital status,

age, age squared, education, US farm experience, English speaking ability, race, ethnicity,

availability of free housing, task, region (California, Florida, and other), the year of the interview

(after 2001 or not), dummy variable for seasonal workers) so that

Inr, = z3, 1.

Then, the duration can be expressed as Int, = zl + u, where u, N(0, o-). However, duration

t, is observed only if person i has legal status. This is a typical case for selection bias. Assuming

e, and u, are bivariately normally distributed with correlation coefficient p, the mean of the log

of the duration conditioned on the legal status of person i is corrected as

E[ln t, In t is observed] = z,/3 + po-,A,

where A, is the correction term for the selection bias which is given as'

5 Correction term is set to zero for native-born citizen.

P-xj j -x;cj

Note that we can use the result of the ordered probit model in the first stage for xa and

it, xya"
---x Also note that _, = -o,/ 3 = oc from the assumption of normal distribution. In the

second stage of this Heckman type two-stage method, we estimate equation (2) below by ordinary

least squares with only completed spells.

Intj = zP,, + pojl A + e, = z,j, + ,l A, + ej (2)

We can also show that the conditional variance of the log of the duration would be

var[lnt ln t is observed] = [1- pcS6] ,


fU 1 /C a i f)l. / ^1 /U t / ^1 a
7, 7 )0 7, ) + 7u, x 7,

I2 2 J{ /1-j e an o i te C

Then a consistent estimator of o-2 is given by r2 = j+ + / We can obtain the

estimator of the asymptotic covariance matrix for [/Pj,P,8j ] by substituting these results in the

formulation in Greene (2003).

The difficulty in the farm worker duration model is that it has two sources of

inconsistency. The observations are censored in the sense that the duration of a person with a

particular legal status is observed only if the person has that status. Some observations are also

censored in the sense that they are uncompleted. On the other hand, the legal status model

(ordered probit) does not have restriction on observations, so that it should be a consistent

estimator. The above method takes care of the selection bias by using correction terms for the

mean duration. The current estimation approach drops uncompleted spells from the data set,

introducing an unknown extent of bias in the estimation. However, given the size of the data set,

the bias is believed to be minimal.


The data used in this study are obtained from the National Agricultural Workers Survey

(NAWS) (Office of Assistant Secretary for Policy). We used the study period from 1989, when

the NAWS was first available, to the most recent year, 2004. This section will describe the

definitions of each variable we used in our model.

Legal status is a discrete variable ranging from 0 to 3. Status 0 = "unauthorized" workers

means that the worker is undocumented (did not apply to any legal status or application was

denied) and also includes those who had no work authorization even if they were documented.

Status 1 = "authorized" workers or documented workers; these workers must have a work

authorization and may fall into any of the following status: having border crossing card/commuter

card, with pending status, or temporary residents holding a non-immigrant visa. Status 2 =

"permanent residents or green card holders" who have the right to reside and work in the U.S., and

status 3 = "citizens" who are a citizen by birth or a naturalized foreign born citizen.

The variable English measures the capability to speak English, and does not include

English reading skills. The variable is a discrete variable ranging from 1 to 4, where 1= not

speaking English at all, 2 = speak a little English, 3 = somewhat able to speak English, and 4 =

speaking good English.

Hispanic is a dummy variable for Hispanic which includes Mexican-American, Mexican,

Chicano, Puerto Rican, and other Hispanic ethnic groups. Black (or African American) and White

are also dummy variables derived from a question regarding their race which may also be

American Indian/Alaka Native, Indigenous, Asian, Native Hawaiian or Pacific Islander, or others.

Age was calculated from the difference between the date of interview and the date of birth, except

for the questionnaire in the earlier years when age was asked directly.

Education is the highest grade level for education, and it ranges from 0 to 20. Experience

is the number of years of doing farm work in the U.S. (not including farm work experience

abroad). Task is the task at the time of interview. Although task is also asked for each period of

work in the past two years, we use only the task at the time of interview for each duration.

Although the original questions have over 100 task codes, tasks are grouped into six categories as

follows: 1 = pre-harvest, 2 = harvest, 3 = post-harvest, 4 = semi-skilled, 5 = supervisor, and 6 =

others. They are argued to be ordered by increasing skill requirements. Seasonal Worker is a

dummy for workers who were working on a seasonal basis for the employer at the time of

interview. Free housing is a dummy variable for workers (or workers and their family) who

receive free housing from their current employer. It does not include those who own the house or

live for free with friends or relatives. It also excludes those who pay for housing provided by

employers or by the government or charity.

The dummies for Florida and California are the state for each work duration, and not

necessarily the state at the time of interview. Before 1993 dummy variable is for all the years prior

to 1993 when the majority of IRCA legalization was granted, and After 2001 is the years

post-September 11, 2001 event.

Duration or farm work spells is a variable created from the work grid in the questionnaire.

It is the difference between the ending dates and starting dates for each "farm work" spell, and

only includes the completed spells (all spells completed at the time of interview).

Ordered Probit Model for Legal Status

Here we estimate the ordered probit model for legal status for foreign-born farm workers

using NAWS data. Table 1 shows estimates for parameters and asymptotic standard errors (given

in the parentheses) using 30912 observations of foreign-born farm workers. Using a 0.05

significance criterion, we find that all coefficients except education squared are statistically


The third column of Table 1 shows the marginal effect of each variable on the probability

of a worker being legal. The probability of worker i being legal is given by

Prob(J* > po) = 1 ((/po x,'a). Then the marginal effect of variable k evaluated at the mean is

((/0o x'a)a, for the continuous variables6 and (/u0 - ~(/k x' ak_ a k) for the

dummy variables, where x' k and a k are variables and coefficients excluding k. Females,

married, workers with higher English speaking ability, non-black, white, non-hispanic are

statistically significantly more likely to have more advantageous legal status all else being the

6 Marginal effect for variables with squared term is given by (/uo k)(k + 2ak ,,sqk) where ck,,, is
coefficient for the squared variable. Also, we treated English speaking ability as a continuous variable.


same. We also find that both age and US farm experience have a significant nonlinear effect on

legal status. US farm experience has positive effect on legal status up to thirty-five years. Age has

positive effect on legal status up to eighty years. Education has a significantly positive linear

effect on legal status. We find that the greatest positive marginal effect is from the female dummy

followed by English speaking ability and before 1993 dummy. The greatest negative marginal

effect is from the Hispanic dummy followed by the after 2001 dummy. Note that, holding all other

characteristics the same, the workers interviewed before 1993 are eleven percent more likely and

those interviewed after 2001 are fourteen percent less likely to be legal compared to those

interviewed between these periods.

Finally, Table 2 shows actual-predicted legal status table. A worker is predicted to be

status 0 (unauthorized) if x, 'd < ,/[, and is predicted to be status 1 (authorized) worker

if/i0 < x, < /, and so on. Table 2 shows that 80 percent of unauthorized workers are correctly

predicted to be unauthorized. In the same way, 21 percent of authorized workers, 70 percent of

permanent resident and 26 percent of citizens are correctly predicted in their legal status. Our

ordered probit model does a very good job in distinguishing type 0 workers from legal workers,

but many of type 1 workers and type 3 (citizen) workers are mistakenly predicted to be type 2

(permanent resident) workers.

Duration Model with Selection Bias Correction

Here we estimate the duration model with selection bias correction using the results from

the ordered probit legal status model in the first stage. Table 3 shows estimates for parameters and

asymptotic standard errors (given in the parentheses) for farm workers with each legal status.

Status 0 (unauthorized) workers have 33,865 observations, status 1 (authorized) workers have

12,560 observations, status 2 (permanent resident) workers have 30,240 observations, and status 3

(citizen) workers have 18,307 observations. Based on asymptotic standard errors using a 0.05

significance criterion, the coefficients on the selectivity variable, X, are all significant except for

citizen workers. That is, using ordinary least squares without correcting for selectivity would lead

to bias in all equations except for citizen workers. Actually, the selection bias correction term is

set to zero for majority of citizen workers, because they are native born. So, the selection bias does

not have a significant effect for this equation as it does for the other legal status equations.

Many variables have a statistically significant effect on duration in a common direction

for all equations. Regardless of the legal status, workers in tasks requiring higher skill,

non-seasonal workers, workers without free housing from employers, workers in California,

workers in Florida, and workers interviewed after 2001 are statistically significantly more likely

to have a longer duration farm job. Most of the signs of these coefficients are reasonable, except

for the availability of free housing offered by the employer, which we expected to have a positive

effect on duration. This may be because workers offered free housing are often migratory,

seasonal workers with low skill and whose length of contract is generally short.

An interesting result is for English speaking ability. For unauthorized workers, higher

English speaking ability is more likely to lengthen the duration in farm work. However, English

speaking ability tends to shorten the duration in farm work for authorized and permanent resident

workers. That is, legal workers leave agricultural work earlier as their English speaking ability

improves, all else being the same. This variable does not have a significant effect on duration of

citizen workers most of whom (77 percent) can speak English well, so that the variable has little


Demographic variables tend to have various directions of influence on farm work

duration for each legal status. Being female has a significantly positive effect on duration for

authorized and citizen workers, while it has a significantly negative effect for permanent resident

workers and no significant effect for unauthorized workers. Marriage has a significantly positive

effect on duration for authorized and permanent resident workers, while it has a significantly

negative effect for citizen workers. Permanent resident and citizen Hispanic workers tend to have

shorter farm work duration than non-Hispanic workers, while unauthorized and authorized

Hispanic workers tend to have longer farm work duration than non-Hispanic workers. Education

has a significantly positive effect on the duration for all legal status, and experience has a

significantly positive effect on the duration for unauthorized and authorized workers, but no

significant effect on permanent resident or citizen workers. Age has a significant nonlinear effect

on duration for all equations. The effect is positive up to an age of 87 years for unauthorized, and

up to 105 years for citizen workers. On the other hand, the effect is negative through 39 years for

authorized, and continuously for permanent resident workers (not turning positive until age 199


Next, using estimates of each equation, we calculate the predicted durations of farm work

by legal status by averaging the predictions over all observations for each equation (Table 4). The

results indicate that the average predicted duration for unauthorized workers is not necessarily

shorter than those for legal workers (authorized, permanent resident, or citizen). Actually, its

average predicted duration is the second longest, and longer than for permanent resident and

citizen workers.

Finally, we implement a simulation to test how farm work duration of a typical

unauthorized worker would be expected to change with a change in legal status. This approach

isolates the effect of legal status of the worker from differing characteristics of workers by holding

the characteristics constant across different legal status. We fix each continuous variable at the

mean of unauthorized worker observations, and fix each discrete variable at the category with the

maximum number of observations of unauthorized workers, except for the "After 2001" dummy

variable. Although observations after 2001 are approximately 24 percent of all unauthorized

worker observations, the post-2001 period is more relevant for current policy purposes. The

profile of the "typical" unauthorized worker is illustrated in Table 5.

The expected duration for this "typical" unauthorized worker is shown in Table 6 using the

equation estimates for each legal status, conditionally upon being an unauthorized worker. The

first row of Table 6 shows the typical unauthorized worker's expected duration under each legal

status; the second row shows the percentage change from the unauthorized status. The result

indicates that the duration of the "typical" unauthorized worker would be 4.4 percent longer if he

were working as an authorized worker, and 3.9 percent larger if working as a permanent resident.

Expected duration would decline by 4.7 percent were he to be working as a citizen, although this

result is based on a statistically insignificant parameter estimate. In contrast to the results in Table

4, the Table 6 results hold worker characteristics constant across status whereas they vary across

status in Table 4.

Setting aside the result for citizens that is based on a statistically insignificant estimate for

the coefficient on the Mills ratio (k), our estimated effect of a change in legal status from

unauthorized to a legal status (either temporary authorization or permanent resident) is largely

consistent with Tran and Perloff s result. In our case, expected duration is somewhat longer when

working under a legal status; they report that "... IRCA increased the long-run probability that

people granted amnesty stayed in agriculture." (p. 437) Hashida and Perloff s result is in the

same direction, but larger. Emerson and Napasintuwong's result similarly suggested a longer

duration for authorized rather than unauthorized workers. Their result referred to the number of

years working in U.S. agriculture, rather than individual jobs as the above three analyses do.

Their model did not directly address the sample selection issue, as the other three analyses do.


We have proposed and estimated a Heckman-type two stage model with legal status

ordered probit model in the first stage and a duration model in the second stage. This methodology

aims at overcoming two sources of inconsistency of farm work duration study: selection bias and

the censoring problem. Our first methodology deals with the former problem adequately, but it

takes only a rudimentary measure on the second problem: we have used only completed spells.

Our current estimation result is based on this method.

The current estimation has significant coefficients on the selection bias correction term

for all legal status equations except for that of citizen workers. That is, using ordinary least

squares would lead to inconsistent estimates in all equations except for citizen workers. The most

important finding from our estimation is that unauthorized workers do not necessarily have

shorter farm work duration than legal workers. This is supported by two statistics. First, average

predicted farm work duration for unauthorized workers is second longest. Second, the simulation

analysis shows that the duration of the "typical" unauthorized worker will be longer when

working under an authorized or permanent resident status.

Table 1. Orderd Probit Model for Legal Status for
Foreign-Born Farm Workers
Parameter Marginal Effect
Female 0.463 0.162
Married 0.206 0.078
English Speaking 0.374 0.140
Black -0.172 -0.066
White 0.151 0.056
Hispanic -0.616 -0.206
Age 0.035 0.007
Age2 -0.0002
Education 0.033 0.014
Education2 0.0003
Experience 0.151 0.038
Experience2 -0.002
Before 1993 0.313 0.113
After 2001 -0.356 -0.137
So 2.487
J1 2.912
12 5.074

Table 2. Actual-Predicted Legal Status Table
Predicted Legal Status Total
Actual Legal 0 1 2 3
0 80% 9% 11% 0% 100%
1 43% 21% 36% 0% 100%
2 14% 15% 70% 1% 100%
3 7% 7% 60% 26% 100%

Table 3. Duration Model for Farm Workers with Each Legal Status
Unauthorized Authorized Permanent Citizen




English Speaking







Seasonal Worker

Free Housing



After 2001





Table 4. Average Predicted Duration for
Each Legal Status (Days)

Unauthorized 56.2

Authorized 59.3

Permanent Resident 54.1

Citizen 53.6

Table 5. Profile of the "Typical" Unauthorized Worker
Constant 1
Female 0
Married 0
English Speaking 1.470
Hispanic 1
Age 28.201
Age2 795.296
Education 6.073
Experience 5.075
Task 2
Seasonal Worker 1
Free Housing 0
California 1
Florida 0
After 2001 1

Table 6. Change in Duration for the "Typical" Unauthorized Worker
Unauthorized Authorized Permanent Citizen
Expected Duration (Days) 62.6 65.3 65.0 59.6
Percent Change 4.4% 3.9% -4.7%


Bourguignon, F., M. Fournier, and M. Gurgand. "Selection Bias Corrections Based on

the Multinomial Logit Model: Monte-Carlo Comparisons," DELTA Working Papers 2004-20,

DELTA (Ecole normal superieure) (September 2004).

Emerson, R. D., and 0. Napasintuwong. "Foreign Workers in Southern Agriculture."

Selected paper, Southern Agricultural Economics Association Annual Meeting, Orlando, Florida

(February 2002).

Greene, W. H. Econometric Analysis, 5th Edition.Upper Saddle River, N.J.: Prentice

Hall, 2003.

Hashida, E., and J. M. Perloff. "Duration of Agricultural Employment." Working Paper

No. 779, Department of Agricultural and Resource Economics (CUDARE), University of

California, Berkeley (February 1996).

Lee, L. F. "Generalized Econometric Models with Selectivity," Econometrica, 51(2)

(March 1983): 507-12.

Office of the Assistant Secretary for Policy, U.S. Department of Labor. "The National

Agricultural Workers Survey: What is the National Agricultural Workers Survey (NAWS)?"

(May 2005). Available at

Tran, L. H., and J. M. Perloff "Turnover in U.S. Agricultural Labor Markets." American

Journal ofAgricultural Economics. 84 (2) (May 2002): 427-37.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs