The U.S. 1986 Immigration Reform and Con

Population Research and Policy Review 9: 93-116, 1990.
© 1990 Kluwer Academic Publishers. Printed in the Netherlands.

The U.S. 1986 Immigration Reform and Control Act and
undocumented migration to the United States

M I C H A E L J. W H I T E , 1 F R A N K D . B E A N 2 &
T H O M A S J. E S P E N S H A D E 3
1Department of Sociology and Population Studies and Training Center, Box 1916,
Brown University, Providence, R102912, USA; 2Population Studies Center, The Urban Institute,
2100 M Street, N.W., Washington, D.C. 20037, USA; 3Department of Sociology and Office of
Population Research, Princeton University, 21 Prospect Avenue, Princeton, NJ 08544-2091, USA

Abstract. One of the major goals of the 1986 Immigration Reform and Control Act (IRCA) is to

reduce the number of undocumentedimmigrants coming to and residing in the United States. This
goal is pursued by allocating increased resources to Border Patrol enforcement, imposing penalties on employers for hiring undocumented workers, and offering to legalize the undocumented
population that has resided in the country for a substantial period of time. This paper evaluates the
impact of IRCA on the flow of undocumented migrants across the U.S.-Mexican border by
analyzing a monthly time series of Border Patrol apprehensions from January 1977 to September
1988 within the context of a multivariate statistical model. The model provides a good fit to the

data (R2= 0.94), and our results indicate that INS resources, Mexican population growth,
comparative economic conditions on both sides of the border, and seasonal factors related to the
agricultural planting and harvesting cycle are all determinants of border apprehensions and, by
implication, of the flow of undocumented migrants to the United States. IRCA's impacts on the
number of 'apprehensions averted' operate mainly through changes in INS effort, the SAWs
agricultural legalization program, and other IRCA-related factors. Our analysis concludes that
the effects of IRCA, though perhaps smaller than sometimes alleged, were associated with a
cumulative net reduction in tinewatch apprehensions of nearly 700,000 in the 23-month period
following enactment of the law. The associated reduction over the same period in the number of
illegal border crossings may be as high as 2 million.

Introduction

I n O c t o b e r 1986 the U . S . C o n g r e s s passed the I m m i g r a t i o n R e f o r m a n d
C o n t r o l A c t ( I R C A ) , a n d P r e s i d e n t R e a g a n signed the legislation into law in
N o v e m b e r of t h a t s a m e year. O n e of I R C A ' s m a i n o b j e c t i v e s is to r e d u c e the
n u m b e r of illegal i m m i g r a n t s c o m i n g to a n d residing in the U n i t e d States, an
o b j e c t i v e that follows a r e c o m m e n d a t i o n of the Select C o m m i s s i o n o n Imm i g r a t i o n a n d R e f u g e e Policy (1981). I n a d e p a r t u r e f r o m the r e c o m m e n d a tions of the C o m m i s s i o n , h o w e v e r , I R C A leaves legal i m m i g r a t i o n policy
largely u n a f f e c t e d , a l t h o u g h s o m e of the legislation's provisions, m o s t n o t a b l y
the l e g a l i z a t i o n p r o g r a m s , p r o v i d e for f u t u r e increases in legal i m m i g r a t i o n

94
(Bean et al., 1989). Other provisions of IRCA did not have such a long
gestation period but, rather, were the outcome of last-minute political compromises reached after five years of wrangling in the U.S. Senate and House of
Representatives.
Included among IRCA's major features are new measures to keep employers from hiring undocumented workers, legalization programs to which more
than three million undocumented immigrants applied before the deadlines
expired, and stepped-up enforcement at the southern U.S. border against
future illegal migration. 1These provisions of IRCA are designed to affect the
stock and flow of undocumented migrants in several ways. First, a key feature
of IRCA penalizes employers who knowingly hire undocumented workers.
All employers are subject to these sanctions, the intent of which is to eliminate
employment opportunities forillegal aliens in the United States and thereby
remove a major motivation for workers to enter the country without proper
documentation. 2 Employers are required to verify each new employee's eligibility to work in the United States by inspecting appropriate documents. For
example, various combinations of the following are acceptable: a passport,
resident alien card, birth certificate, social security card, and driver's license.
Those who knowingly hire undocumented workers are subject to civil fines
ranging from $ 250 to $ 2,000 for each undocumented migrant in the case of a
first offense and to criminal penalties of up to six months in prison for a

'pattern and practice' of employing illegal workers.
Second, the new immigration law extended legal temporary resident alien
status to undocumented immigrants who could prove they were residing in the
United States continuously since January 1, 1982. Such persons may later
adjust to permanent resident alien status provided they can demonstrate a
minimal understanding of English and a basic knowledge of U.S. civics and
history. In addition, special agricultural workers (SAWs) will qualify for
legalization in one year if they can show that they worked at least 90 days in
U.S. perishable agriculture in each of the years 1984, 1985 and 1986, or qualify
in two years if they worked at least 90 days in the 12 months preceding May 1,
1986. By the time the application periods for these two programs had ended,
1.77 million applications for temporary resident alien status had been filed
through the general legalization program and an additional 1.3 million through
the SAWs program (U.S. Immigration and Naturalization Service, 1989a).
Third, IRCA contains provisions for increased enforcement capabilities
within the U.S. Immigration and Naturalization Service (INS). The INS was
authorized to receive funding of $ 380 million in fiscal year 1986 for enforcement activities, a figure that jumped to $ 456 million in fiscal year 1987 and to
$ 538 million in fiscal year 1988. 3 Substantial increases were also provided for
the number of Border Patrol personnel (from 3,687 in fiscal year 1986 to 4,669
in fiscal year 1988).

95
A considerable debate has arisen in the post-IRCA period over the extent to
which IRCA is working to reduce illegal immigration. There can be little
doubt that the two legalization programs within IRCA have succeeded in
substantially lowering the size of the undocumented population residing in the
United States, simply by reclassifying as legal temporary resident aliens some
persons who were formerly here as illegal aliens. More at issue is whether
IRCA has reduced the flow of undocumented immigrants across the U.S.Mexican border. Representatives of the Immigration and Naturalization Service have argued on the basis of a 42 percent decline between 1986 and 1988 in
the number of Border Patrol apprehensions of undocumented migrants that
the new law is achieving its desired effect (Mathews, 1988).4 Critics charge,
however, that IRCA has had little or no impact on the flow and that apprehension statistics are down because U.S. Border Patrol officers have been diverted
to other assignments, including hunting for drug smugglers. It is also claimed
that the pool of potential undocumented migrants has been reduced because
many persons who previously went back and forth across the border illegally
now do so legally as a result of the SAWs legalization program that is part of
IRCA (Suro, 1989). 5
We analyze apprehensions data in this paper and argue that they can provide
useful insights into some of IRCA's demographic impacts. We also develop a
statistical model to quantify the contributions of IRCA to the decline in the

number of linewatch apprehensions since 1986. The estimated model is then
used to simulate the number of apprehensions one would have observed in the
absence of IRCA and to calculate the number of 'apprehensions averted' as a
result of the recent reform of immigration law. The paper concludes with a
discussion of what inferences can reliably be drawn from analyses of apprehensions data about IRCA's impact on the flow of illegal immigrants into the
United States.

Why analyze apprehensions data?
Apprehensions data are one of the most frequently cited sources of information about undocumented migration to the United States. 6 Such statistics are
generated whenever the U.S. Border Patrol or other INS enforcement personnel apprehends someone who is in the country illegally. 7 The vast majority of
such apprehensions (92 percent between fiscal years 1986 and 1988) are made
by the Border Patrol. Border Patrol apprehensions may further be distinguished by the status in which an alien first enters the country. Persons who enter
with legal visas (tourists and students, for example) and then remain after their
visas expire or take unauthorized employment are referred to as 'visa-abusers'. Those who enter without any sort of legal papers are referred to as

96
'entrants-without-inspection' (EWIs). More than 99 percent of Border Patrol
apprehensions consist of EWIs. 8
Entry without inspection can occur at any land border of the United States
and may involve persons from any country. Because aliens from Mexico

accounted for 96 percent of apprehensions in fiscal year 1987 (U.S. Immigration and Naturalization Service, 1988), the entry of individuals without inspection at the U.S.-Mexican border is of particular interest. It is these EWI
data that we examine in this paper.
Apprehensions data are relevant to the question of whether illegal immigration has risen or fallen over a given period of time because they may in some
sense be an indicator of the number of border crossings and successful entries
into the country by illegal migrants. The number of Border Patrol apprehensions increased, for example, from 736,451 in fiscal year 1982 to 1,025,451 in
fiscal year 1983, following the collapse of the Mexican economy. This increase
was widely interpreted at the time as reflecting an increase in illegal immigration to the United States (The New York Times, 1983).
Reliance on apprehensions data as a proxy for the flow of undocumented
migrants into the United States also has several limitations. Apprehensions
data refer to events, not to people, and the same individual may be apprehended more than once in a given time interval. Moreover, many undocumented
migrants manage to enter the country successfully without detection by the
INS. The extent to which these two errors cancel each other is difficult to judge
(Frisbie, 1975). Finally, the probability that an undocumented migrant is
arrested entering the United States illegally changes over time, which implies a
corresponding time variation in the rate of repetition crossing the border and
in the ratio between apprehensions and illegal border crossings. The model we
develop below addresses these limitations of data.
Since 1983 the number of Border Patrol apprehensions has averaged more
than one million per year, peaking at 1,609,278 in fiscal year 1986. In fiscal
years 1987 and 1988, the numbers declined to 1,115,267 and 936,795, respectively. A more detailed picture of the time trend comes from examining

month-to-month variations in Border Patrol apprehensions. Figure 1 shows
these fluctuations for the number of EWI linewatch apprehensions between
fiscal years 1977 and 1988. Linewatch apprehensions are Border Patrol arrests
that take place right on the southern U.S. border as a direct result of patrolling
and watching for crossings. They constitute the majority of all Border Patrol
apprehensions.
Information on the number of Border Patrol apprehensions and on how
these statistics have varied provides at least a partial basis for assessing
whether the monthly number of illegal border crossings into the United States
(or the 'flow' of undocumented migrants) has changed since the passage of
IRCA. But to be useful for this purpose, it is important to recognize that the

97
Thousands

120
100
80
60
40

20
0
Jan

Jan

I 77 [ 7 8

Jan
Jan
Jail
Jan
Jan
Jan
Jan
Jan
Jan
Jan_
I 79 I 80 I 81 I 8 2 1 83 I 84 I 85 I 86 I 87 1 8 8 1

Month and Year
Fig. 1. Total linewatch apprehensions at border.

level of apprehensions is affected by numerous factors, including many having
nothing to do with IRCA. Before drawing conclusions about whether changes
in apprehensions might be attributed to IRCA, therefore, it is necessary to
control for the influence of these additional factors. Our strategy here is to
account for a wide range of factors that might influence the flow coming into
the country by developing and estimating a statistical model for linewatch
apprehensions data.

Conceptualizing the process
The purpose of this section is to describe the model and related data used to
analyze the monthly statistics on linewatch apprehensions that are graphed in
Figure 1. Border Patrol linewatch apprehensions are the largest component of
all Border Patrol apprehensions and constituted more than 60 percent of the
total between fiscal years 1977 and 1988 9 They also correspond most closely to
what one normally thinks of as undocumented migrants apprehended 'at the
border'. From January 1977 until September 1988, for example, nearly 98
percent of the 6.9 million linewatch apprehensions of persons who crossed the

border illegally from Mexico into the United States and later were located by
linewatch officers were recorded as apprehended 'at the time of entry'. By
contrast, 24 percent of Border Patrol apprehensions resulting from farm and
ranch checks involved persons who had entered the United States without
inspection and had been in this country for more than 30 days. One may

98
therefore consider Border Patrol personnel assigned to linewatch duty as the
line of first defense against overland undocumented migration across the
U.S.-Mexican border.

The model

We view the process of undocumented migration to the United States leading
to apprehension at the U.S.-Mexican border as depending upon three factors:
the population at risk of migrating illegally to the United States, the propensity
of this population to migrate, and the probability of being apprehended at the
border. The product of the first two factors is the number of border crossings in
undocumented status in a specified time interval. Because some fraction of
these crossings results in apprehension at the point of entry to the United

States, the multiplication of all three elements corresponds to the number of
linewatch apprehensions along the U.S.-Mexican border per time period. Our
analytic strategy is to model the data on linewatch apprehensions by considering the determinants of these three components.I°
Because Mexican mortality declined nearly three decades before fertility
began to fall in the late 1970s, the population in the prime migrating ages has
been increasing in recent years (Alba & Potter, 1986; Massey, 1988). A
convincing measure of this growing population at risk is the size of the young
adult population in Mexico. n An additional factor to consider, however, is the
effect of the SAWs legalization program in IRCA. Special agricultural workers
who have been granted legal temporary resident alien status may now move
back and forth between the two countries without fear of apprehension when
returning to the United States. The effect of IRCA's provisions for SAWs
legalization, therefore, may be to reduce the size of the pool of persons at risk
of being apprehended.
A second factor determining the level of apprehensions at the U.S.-Mexican
border is the rate of undocumented migration from the pool of eligibles. This
propensity is likely to be substantially motivated by an evaluation of comparative economic opportunities in the United States and in Mexico. To the extent
that this is the case, such influences as relative unemployment rates, wage
opportunities, and income levels in the two countries are likely to play important roles in this evaluation. The growth of migrant networks in the United
States, fueled by the accumulation from prior migrant flows, lowers the cost of
migration and also should stimulate the propensity to migrate (Massey, 1988).
In addition, a component of labor demand for low-wage farmworkers in
California and other parts of the American Southwest is highly seasonal,
building during the spring planting months, reaching a peak during the summer harvest, and then tapering off significantly in the winter (Espenshade &

99
Taylor, 1988). One may expect this influence to be reflected in seasonal
variations in the tendency to migrate illegally to the United States. Finally,
other factors associated with IRCA may also affect the outcome. Beliefs of
prospective undocumented migrants that IRCA's employer sanctions provisions have substantially reduced the probability of locating employment in the
United States and/or that the U.S. Border Patrol has increased its enforcement
effort to such a level rendering any attempted border crossing extremely costly
are likely to reduce the incentive to migrate. IRCA may also have dampened
the incentives for undocumented migrants living in the United States prior to
IRCA to circulate back and forth between the United States and Mexico out of
concern that they might have a difficult time locating new unemployment upon
their return. 12 These forces operating in the post-IRCA period may exhibit
both secular and seasonal tendencies as migrants adjust their perceptions to
new realities. 13
Third, the probability that an undocumented migrant will be apprehended
at the U.S.-Mexican border depends directly on the enforcement effort
mounted by the Immigration and Naturalization Service. Prominent factors
affecting INS effort are the human and physical resources devoted to enforcement. These include, among other things, the number of Border Patrol officers stationed along the line and the capital equipment they have at their
disposal in the form of motor vehicles, infrared sensing devices, and the like.
The preceding discussion suggests that monthly variations in the level of
apprehensions of undocumented migrants along the southern U.S. border
may be conceptualized as the outcome of a process very much like that of
economic production in which the volume of output (in this case apprehensions) depends upon inputs of labor, capital, and resources. Labor and capital
inputs correspond to INS enforcement effort and reflect the probability of
being apprehended. The resources or raw materials can be represented by the
number of border crossings by undocumented migrants within a specified time
interval. It is clear that the level of apprehensions will be zero if either the
probability of being apprehended or the number of border crossings is zero.
This reasoning suggests a production function involving multiplicative terms.
In particular, we assume that
Y = P ' m * 7r,

(1)

where
Y = level of monthly linewatch apprehensions along the U.S.-Mexican
border,
P -- population at risk of migrating in undocumented status to the United
States,
m = monthly rate of undocumented migration to the United States, and

100
Table 1. V a r i a b l e n a m e s , d e f i n i t i o n s , s o u r c e s of d a t a , a n d e x p e c t e d signs of e s t i m a t e d coefficients.
Variable name

Definition

Data source

Expected
sign of
coefficient

Dependent variable:
LWAPPS

Natural log of monthly finewatch U.S. Border Patrol reports of
apprehensions along the
field operations
U.S.-Mexican border

N/A

Explanatory variables:
MEXPOP
SAWS
UNEMPRATIO

WAGERATIO

JAN - NOV

POSTIRCA

YEAR2
SPRING

Mexican population 15-34 years
of age (in millions)
Cumulative number of SAW's
applications (in millions)
Ratio of U.S. male
unemployment rate (percent)
to Mexican male unemployment
rate (percent)
Ratio of hourly wage rate in
U.S. non-agricultural sector
(in Mexican pesos) to hourly
earnings in Mexican
manufacturing sector (in pesos)
Monthly dummy variables for
seasonal labor demand
(December is omitted month)
Deterrent efficacy of IRCA

Altered IRCA effect in second
year of post-IRCA period
Spring effect in post-IRCA
period

SUMMER

Summer effect in post-IRCA
period

SPRING2

Altered spring effect in year 2
of post-IRCA period
Altered summer effect in year 2
of post-IRCA period
Reciprocal of U.S. Border
Patrol officer hours devoted
to linewatch activities (hours in
millions)
Reciprocal of the sum of INS
capital investment for
enforcement purposes in
months 7-18 prior to
observation month (capital
expenditures in millions of
constant 1988 U.S. dollars)

SUMMER2
LWHRS

CAPINV

International Division, U.S.
Bureau of the Census
U.S. Immigration and
Naturalization Service (INS)
U.S. Bureau of Labor Statistics,
Employment and Earnings;
International Labor Office,
Bulletin of Labor Statistics;
World Bank
U.S. Bureau of Labor Statistics,
Employment and Earnings;
International Labor Office,
Bulletin of Labor Statistics;
World Bank
Each monthly dummy = 1 for
that month and = 0 elsewhere

+

-

+

+

Dummy variable ( = 1 for
NOV 1986 and all subsequent
months)
Dummy variable (= 1 for NOV
1987 and all subsequent months)
Dummy variable (= 1 in
M A R - JUN months of postIRCA period)
Dummy variable (= 1 in
J U L - OCT months of postIRCA period)
SPRING * YEAR2

?

SUMMER * YEAR2

?

Border Patrol sector reports of
field operations

-

Office of Management and
Budget, Budget of the United
States Government, Appendix
material for INS

?

?

101
7r = probability that an illegal immigrant crossing the U.S.-Mexican border is
apprehended by a linewatch officer.
If we postulate that the natural logarithm of each term on the right side of
equation (1) is linearly related to a set of underlying behavioral determinants
(X,), then we have
lnY = 13o+ 13~X~+ 1~2X2 -[- . , . -°r- [~kXk jr. U,

(2)

where u is a stochastic error term. With this formulation, each coefficient on an
X variable represents the estimated relative or proportional change in Y
corresponding to a one unit change in X. This is the equation we will use for
estimation purposes, a4

Variables and data

The variables used in the regression analysis together with the hypothesized
signs of their regression coefficients are listed in Table 1. MEXPOP and SAWS
affect the population at risk of migrating illegally to the United States in
different ways. Because the great majority of undocumented migrants across
the southern U.S. border are born in Mexico, we have used the size of the
young adult population in Mexico (ages 15-34) as a rough proxy for the
expanding at-risk population. At the same time, a growing number of legalized
special agricultural workers who receive legal temporary resident alien status
and who continue to circulate between Mexico and the United States should
reduce the number of undocumented travelers to this country.
Several factors are assumed to be related to the rate of undocumented
migration from Mexico to the United States. Principal among these is a
comparison of relative economic opportunities in Mexico and the United
States, captured by relative wage and unemployment ratios (WAGERATIO
and UNEMPRATIO, respectively). Improving conditions on the U.S. side of
the border and/or worsening circumstances in Mexico (reflected in unemployment rates or in wage rates) should accelerate the rate of migration. Second,
we hypothesize that seasonal variations in demand for low-wage agricultural
workers in the American Southwest are related to incentives for undocumented migration. These demands are typically least during the winter months and
accelerate as the planting and harvest seasons unfold. We incorporate these
seasonal effects through a set of monthly dummy variables, JAN to NOV,
using December as the reference month.
Third, if IRCA is successful in reducing the flow of undocumented immigrants into the United States, there should be fewer apprehensions along
the U.S.-Mexican border in the post-IRCA period than during pre-IRCA

102
months, holding all other things constant. We attribute this hypothesized
negative effect, measured by the variable POSTIRCA, to the possibility that
employer sanctions and stepped-up border enforcement may have altered
perceptions of the likelihood of being able to enter the United States illegally
and locate subsequent employment. Because perceptions and, therefore, the
incentives and risks associated with undocumented migration may change as
employer sanctions and other provisions of IRCA become more fully implemented, this effect of IRCA may be strengthened or weakened in the second
year of the legislation. We include the variable YEAR2 to allow for this
possibility. In addition to a secular trend, IRCA's success in affecting undocumented migration may also possess seasonal elements (SPRING and
SUMMER), especially if there is a seasonal component to agricultural labor
demand. We expect these seasonal IRCA effects to be greatest during the
spring months when agricultural labor demand is building. The variables
SPRING2 and SUMMER2 are included to test whether seasonal effects in the
post-IRCA period are altered in the second year of IRCA.
The final pair of explanatory variables in Table 1, LWHRS and CAPINV, is
related to the probability of being apprehended along the southern U.S.
border. Increases in both the number of U.S. Border Patrol officer linewatch
hours devoted to this activity and the amount of physical capital (for example,
unattended electronic ground sensors, lighting, a variety of imaging devices,
and transportation vehicles) employed in the enforcement effort are assumed
to raise the probability of apprehension.15 Because of the logarithmic nature of
equation (2) and the requirement that probabilities lie between zero and one,
we have used reciprocals of these labor and capital variables. Negative coefficients imply that the probability of being apprehended grows as labor and
capital inputs increase.
Altogether the model in equation (2) is estimated from 141 monthly observations from January 1977 to September 1988. The estimation results are
discussed in the next section.

Empirical results
In this section of the paper we present the results of estimating the statistical
model. We then use these results to calculate the proportion of the decline in
linewatch apprehensions following IRCA that is attributable to each of three
factors: (1) changes in INS effort, (2) the operation of the agricultural legalization program; and (3) the remaining effects of IRCA. Because we express
the dependent variable, linewatch apprehensions, in its logarithm, the estimated coefficients represent the percentage change in apprehensions due to a
one unit change in the independent variable of interest.

103

Regression effects
Because ordinary least squares estimates with time-series data typically exhibit
serial correlation in the residuals, we estimate the model with a two-step
autoregressive procedure. 16 A first-order autoregressive term is sufficient to
capture the serial correlation; higher-order terms were not significant. The
estimated regression equation appears in Table 2.17For all of the variables that
are related to our hypotheses, the estimated coefficients are in the expected
direction, and with the exception of the SAWs variable, the coefficients are all
statistically significant. Our model was also able to account for a substantial
amount of the variation in the dependent series (here 94.3 percent of the
variance in linewatch apprehensions). Although high R 2 values are often
found in time-series regressions, the good fit here is especially gratifying
because of the strong seasonal patterns apparent in Figure 1. A test of the
equivalence of the model between the first and second halves of the series was
accepted, indicating that the coefficients on the regressors are statistically
equivalent in the 1977-83 period and the 1983-88 period.
The two groups of persons that are hypothesized to affect the population at
risk of migrating illegally to the United States have the expected impact on the
pattern of apprehensions. First, the Mexican population 15-34 years old
totaled roughly 23 million persons in 1980 and was growing at about 90,000
persons per month in 1986. An increase of one million Mexicans aged 15-34
translates into nearly a 5 percent increase in monthly apprehensions. Thus, the
effect of Mexican population growth in the mid-1980s was to raise the number
of linewatch apprehensions by about 4,200 over the same month in the previous year. Because the time trend in population is smoothly monotonic, however, its inclusion also may capture other similarly shaped trends, such as the
growth of social networks (Massey & Espafia, 1987) and the spread of information.
Second, the legalization program for special agricultural workers may have
removed from the at-risk population a group of seasonal farm workers having
a disproportionately high tendency to circulate back and forth between Mexico and the United States. Prior to IRCA these individuals would return to the
United States in undocumented status; now with the legalization program they
may come and go lawfully. Hence, the model includes the cumulative number
of SAWs who have filed for legalization by the end of each month. As
expected, the SAWs variable's coefficient is negative but not statistically
significant at customary levels, because this measure is correlated with others
in the model. The SAWs coefficient implies a 28 percent reduction in apprehensions for every million SAWs applications. By the end of our data series in
September 1988, at which time more than 870,000 applications had been
received, this 'SAWs effect' was large enough to produce a 24 percent decline

104
in m o n t h l y a p p r e h e n s i o n s . S t a t e d a n o t h e r w a y , t h e results a r e c o n s i s t e n t with
t h e n o t i o n t h a t t h e n u m b e r o f a p p r e h e n s i o n s d e c l i n e d b y o n e for e v e r y 65
persons who had applied for the SAWs program.
E c o n o m i c c o n d i t i o n s in t h e U n i t e d S t a t e s a n d M e x i c o h a v e an i m p o r t a n t
i n f l u e n c e o n t h e b o r d e r a p p r e h e n s i o n series. F o r e a c h o n e unit i n c r e a s e in t h e
r a t i o b e t w e e n U . S . a n d M e x i c a n h o u r l y w a g e s ( W A G E R A T I O ) , for e x a m p l e ,
m o n t h l y a p p r e h e n s i o n s a r e p r e d i c t e d to i n c r e a s e b y 4.5 p e r c e n t . I n t h e l a t e
1970s this w a g e r a t i o v a r i e d b e t w e e n 2.5 a n d 3.5, b u t following t h e e c o n o m i c
crisis in 1982, M e x i c a n w a g e s d e t e r i o r a t e d s u b s t a n t i a l l y . B y t h e t i m e o f
I R C A ' s p a s s a g e in N o v e m b e r 1986 t h e U . S . w a g e s t o o d at r o u g h l y e i g h t t i m e s
t h e M e x i c a n w a g e , a c h a n g e since t h e l a t e 1970s large e n o u g h to b r i n g a b o u t a

Table 2. Regression results for the logarithm of monthly linewatch apprehensions, 1977-88.

Variable

b

t

MEXPOP
SAWS
UNEMPRATIO
WAGERATIO
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
POSTIRCA
YEAR2
SPRING
SUMMER
SPRING2
SUMMER2
LWHRS
CAPINV
INTERCEPT
RHO(1)
R2
N

0.048**
- 0.276
- 0.123 **
0.045 *
0.476**
0.557**
0.723"*
0.670"*
0.625**
0.549**
0.615"*
0.587**
0.426**
0.273**
0.116"*
- 0.270"*
0.072
0.229 **
0.014
0.070
- 0.066
0.143"*
- 0.934"*
10.082 **
- 0.351"*
0.943
141

(3.948)
( - 0.837)
( - 3.220)
(2.041 )
(13.976)
(14.159)
(16.729)
(16.237)
(14.876)
(13.396)
(14.825)
(14.104)
(10.299)
(6.566)
(3.517)
( - 3.532)
(0.568)
( - 2.829)
( - 0.145)
(0.530)
( - 0.371)
( - 5.446)
( - 2.163)
(34.790)
( - 4.043)

* p -< 0.05, two-tail test.
**p-< 0.01.
t-statistics in parentheses.

-

-

-

105
20 percent increase in apprehensions net of other factors. After mid-1987,
relative conditions in Mexico improved, bringing down the ratio to just under
six by the end of the series in September 1988. We constructed a similar ratio of
the unemployment rates in the United States and Mexico (UNEMPRATIO).18
The coefficient takes on the expected negative sign and indicates that a rise in
the ratio of one unit is associated with a 12 percent decline in apprehensions.
The seasonal pattern that was observed in Figure I is clearly reflected in the
set of monthly coefficients. Each coefficient represents the average percentage
increase in apprehensions in that month in relation to December. For example, the volume of apprehensions in March is typically 72 percent above the
December level. Apprehensions are usually low in October and November
and lowest in December. This pattern is consistent with the hypothesis that
seasonal factors, including the seasonal nature of agricultural farmwork and
the desire among many non-agricultural workers to spend the winter in Mexico, provide much of the motivation to cross the border at certain times. This
basic monthly pattern persists in the post-IRCA period, although it is modified
in a way we will discuss below.
We now turn our attention to the measures of resources and enforcement
effort by the Immigration and Naturalization Service. As expected, an increase in the number of Border Patrol officer linewatch hours (LWHRS) leads
to greater numbers of apprehensions, an effect that statistically is highly
significant. To see the effect more concretely, consider an increase of linewatch hours from 200,000 (approximately the monthly average during FY
1986) to 210,000. This five percent increase leads to a 3.4 percent increase in
apprehensions, or almost 3,000 per month, for the levels of effort that were
being realized in the period immediately preceding IRCA. Except for the
monthly dummy variables, linewatch hours provide the largest single t-statistic
among variables in the equation. The effect of capital budget expenditures
(CAPINV) is statistically significant and of the expected sign. Using budget
and apprehensions values current at the time of IRCA's passage, we find that
each additional million dollars of capital expenditures (1988 dollars) means an
increase in monthly apprehensions of about 8 percent.
Together with its impacts operating through the special agricultural worker
legalization program and through changes in INS effort, IRCA doubtless had
the additional effect of balkanizing circulatory flows of undocumented workers. Because of employer sanctions and other widely advertised features of the
new legislation, prospective undocumented workers in Mexico might have
been less willing to come north as undocumented migrants. Illegal immigrants
with jobs in the United States might have been less willing to give them up in
exchange for a return trip home, especially workers who were hired prior to
IRCA and who are not subject to 1-9 requirements. The construction of a set of
dummy variables to capture these potential mobility-inhibiting features of

106
IRCA enables us to observe how this pattern might have changed over time.
The underlying effect, which we have labeled POSTIRCA in Table 2, indicates that monthly apprehensions are 27 percent lower in the post-IRCA
period than in the pre-IRCA period, everything else being held constant. This
basic post-IRCA effect is, of course, susceptible to modification. In the second
year following 1RCA's passage, for example, it weakens slightly to a 20
percent reduction in apprehensions, but the t-statistic on YEAR2 indicates
that this apparent relaxation in IRCA's influence is not significant.
We find a very strong negative 'spring effect' in the post-IRCA period. For
the months from March through June, apprehensions are down an additional
23 percent, net of all the other variables in the model. When the POSTIRCA
and SPRING effects are combined, we find that linewatch apprehensions in
the period from March to June 1987 are 50 percent below comparable levels in
the pre-IRCA period. Because the coefficient on SPRING2 is positive (though
not statistically significant), there is a slight suggestion of a weaker 'spring
effect' in 1988. Thus, in those months during which the volume of apprehensions is usually the greatest (due, one suspects, to the influx of agricultural
workers), the fall-off in apprehensions is pronounced, although not so large
that it completely offsets the seasonal pattern we observed in Fig. 1. Summer
effects are generally negligible.

Apprehensions averted
As a step toward assessing IRCA's impact on undocumented immigration to
the United States, we first calculate the influence of the new legislation on
recent levels of linewatch apprehensions of illegal immigrants. This estimate
depends on how many apprehensions one might reasonably expect to have
observed in the post-IRCA period if the 1986 reforms had not been adopted.
IRCA's combined effects on apprehensions can then be gauged by subtracting
the monthly number of linewatch apprehensions that were observed from the
number anticipated in the absence of IRCA. In addition, it is of considerable
policy interest to know the relative contributions of each of IRCA's several
dimensions to this aggregate difference.
To maintain internal consistency, we use the estimated regression equation
reported in Table 2 to calculate the portion of the monthly decline in apprehensions in the post-IRCA period attributable to each of three sources: (1)
changes in INS effort; (2) the SAWs legalization program; and (3) the remaining effects of IRCA. We make four separate calculations. First, to estimate the
monthly number of linewatch apprehensions in the absence of IRCA, we set
the SAWs variable and all six IRCA-related variables (POSTIRCA, SPRING,
SPRING2, SUMMER, SUMMER2, and YEAR2) equal to zero and assume

107

that the two INS effort variables (capital investment and linewatch hours) stay
constant at their average values during fiscal year 1986. The remaining variables in Table 2 are assumed to take on their actual values. Then, beginning
with November 1986, the expected number of linewatch apprehensions is
calculated month by month to September 1988. Second, to isolate the effect of
changes in INS effort under IRCA, the first calculation is repeated by substituting the actual values of linewatch hours (LWHRS) and capital budget
expenditures (CAPINV) for their average fiscal year 1986 values. The difference between the predicted values of linewatch apprehensions in steps i and 2
is the partial effect of IRCA due to changes in INS effort.
Third, to separate the effect of the SAWs legalization program, the calculation in step 2 is repeated by substituting the observed values of the SAWs
Table 3. Reduction in linewatch apprehensions due to IRCA's provisions, including changes in
INS enforcement, SAWs legalization, and mobility effects.

INS enforcement

Mobility effects

SAWs

Total

Month

Monthly

Cumulative Monthly Cumulative Monthly Cumulative Monthly Cumulative

(1)

(2)

(3)

Nov-86
Dec-86
Jan-87
Feb-87
Mar-87
Apr-87
May-87
Jun-87
Jul-87
Aug-87
Sep-87
Oct-87
Nov-87
Dec-87
Jan-88
Feb-88
Mar-88
Apr-88
May-88
Jun-88
Jul-88
Aug-88
Sep-88

(1,471) (1,471)
(1,138) (2,609)
(5,324) (7,933)
911
(7,022)
(2,961) (9,984)
5,898
(4,086)
(592) (4,678)
392
(4,286)
1,696
(2,590)
(671) (3,261)
(470) (3,731)
(724) (4,456)
2,457
(1,998)
2,527
529
2,835
3,364
9,867
13,232
5 , 4 8 8 18,720
7,653 26,372
6,083 32,456
10,910 43,366
12,678 56,043
11,532 6 7 , 5 7 5
13,338 80,913

(4)

(5)

(6)

0
0 14,191
0
0 12,831
0
0 21,502
0
0 22,398
0
0 44,346
0
0 38,442
0
0 39,687
402
402 37,180
1,398
1,800 24,359
2,423
4,223 24,052
3,028
7,252 20,001
3 , 3 0 7 10,559 16,330
3,114 13,673 9,377
3,172 16,845
8,379
5,477 22,322 13,117
6,169
28,491 13,128
9,180 37,671 26,950
10,998 48,670 24,130
12,755 61,425 22,570
12,826 74,251 19,172
15,203 89,455 16,664
16,240 105,694 16,072
14,535 120,230 13,011

(7)

(8)

(9)

14,191
27,022
48,524
70,922
115,267
153,709
193,396
230,577
254,936
278,988
298,989
315,319
324,696
333,075
346,192
359,320
386,270
410,400
432,970
452,14t
468,805
484,877
497,888

12,721
11,693
16,178
23,308
41,384
44,340
39,095
37,974
27,453
25,804
22,559
18,913
14,949
14,078
21,429
29,164
41,619
42,781
41,408
42,908
44,545
43,843
40,884

12,721
24,414
40,591
63,899
105,284
149,623
188,718
226,693
254,146
279,950
302,510
321,422
336,372
350,450
371,879
401,042
442,661
485,442
526,850
569,758
614,303
658,146
699,031

Note: ( ) indicates an increase in apprehensions due to INS enforcement in comparison with no
change in policy

108
variable for zeros. The difference between the predicted number of linewatch
apprehensions in steps 2 and 3 is attributable to IRCA's SAWs program. 19
Fourth, the calculation in step 3 is repeated by giving all six IRCA-related
variables their actual values. Given our model's good fit to the data, the
predicted values of linewatch apprehensions from step 4 come close to those
for observed apprehensions in each month. The difference between steps 3
and 4 in the predicted linewatch apprehension values is a measure of the
non-SAWs-related and non-INS-effort-related effects of IRCA. 2° Finally, the
difference between the first and fourth sets of estimates provides an indication
of the amount of decline in apprehensions occurring in each month of the
post-IRCA period (that is, of the 'total' IRCA effect). This total IRCA effect
and its three separate components are displayed month by month in Table 3.
Numbers in Table 3 that are in parentheses represent negative values and
imply that the associated IRCA component lifted the level of linewatch
apprehensions in that month above the level expected in the absence of IRCA.
This outcome occurs only in the case of IRCA's influence on changes in INS
effort in the first year following IRCA. Border Patrol linewatch hours increased through the latter part of 1986 and on into early 1987, and then
declined through the remainder of 1987 and 1988. Capital expenditures devoted to enforcement, however, decreased until April 1987, after which they
substantially increased. Because values of the two INS effort variables moved
in opposite directions in the post-IRCA period, they tended to have somewhat
offsetting effects. Their combined net effect raised linewatch apprehensions
for most months in the year following IRCA's passage above the level anticipated without IRCA. In November 1986, for example, the 8 percent increase
in linewatch hours over the average fiscal year 1986 value combined with a 27
percent decline in capital expenditures to produce a rise of 1,471 linewatch
apprehensions.
In the second year, however, IRCA had the apparent effect of lowering
linewatch apprehensions. In June 1988, for example, the number of linewatch
hours logged by Border Patrol agents was 20 percent below fiscal year 1986
levels so that, despite a near doubling of relevant capital budget inputs, there
was a net reduction in apprehensions of nearly 11,000. These results suggest
that trends in Border Patrol linewatch hours are more influential predictors of
variations in apprehensions than trends in the captital budget measure. This
conclusion is consistent with the relative magnitudes of their coefficients'
respective t-statistics in Table 2.
Both the SAWs effect and the remaining IRCA effects consistently work to
lower linewatch apprehensions, either by reducing the size of the population at
risk of migrating illegally or by slowing the migration rate. The SAWs effect in
Table 3 is zero prior to June 1987 when applications for the legalization
program began to be received. Then as the SAWs program got underway, its

109
Thousands
200 t

162,800

150

Q-..~Q Q
\\\\
\\\\
\\\\

100
63,900

(172,400)*

168,700

129,300"
94,700

\\\\
\\\\
\ \ \ \

0
Fiscal Year 1987
-50

,
Nov-Feb

,
~-Jun
I

"Jul-Sep 1988

INS

Fiscal Year 1988
I

I

Jul-Oct
SAWS

I

Nov-Feb
~

t ProjectedJ~-O~ 1988

Mar-Jtm

I

Jul-Oct

Mobitity Effects
Source:Table 3

Fig. 2. Reductions in apprehensions due to IRCA.

monthly impact on linewatch apprehensions also grew. IRCA's remaining
effects reduce apprehensions from the outset; the largest impacts tend to be
registered in the spring months from March through June.
By September 1988 the cumulative net total reduction in apprehensions due
to IRCA amounted to nearly 700,000. This quantity represents a drop to
approximately 1.3 million from the 2.0 million linewatch apprehensions we
estimate would have materialized between November 1986 and September
1988 in IRCA's absence. The portion of the reduction attributable to the net
shift in INS resources and effort amounted to 80,900, or to 11.6 percent of the
total. Agricultural legalization accounted for another 17.2 percent. The residual effect of IRCA accounted for the remaining 71.2 percent, or 498,000
apprehensions.
Figure 2 summarizes these values by four-month intervals and displays their
total simulated impact. Because for many months in the first year of IRCA,
changes in INS effort increase the expected number of apprehensions, this
component's contribution to reducing apprehensions is not appreciable until
later in the series. The pronounced decline in linewatch hours later in 1987 and
during 1988 may have represented the diversion of Border Patrol officers'
efforts away from linewatch duties and towards such other activities as informing employers about their new responsibilities under IRCA. Figure 2 also
shows the steady growth of the impact of the SAWs program. IRCA's remain-

110
ing effects reduce apprehensions by about 20,000 per month, although the
proportionate effects are greatest in the first year after IRCA.
Taken together our results point to a number of IRCA-dependent mechanisms operating in the post-IRCA period. The introduction of employer
sanctions is likely to have made the prospect of undocumented movement to
the United States less attractive. Fears, even if not well-founded, of more
serious punishments than had been imposed in the past would also be likely to
decrease the number of attempts to cross. A general decline in circulatory
migration, including any increased tendency to remain in the United States
among those who had once entered illegally, would have a similar effect.
The differential reduction of apprehensions across seasons in the postIRCA period is also worth noting. It may represent a reduction in attempts
concentrated among seasonal agricultural workers who would be most likely
to enter the United States in the spring. One may also speculate that it
represents a decline in first-time migrations or in 'new starts' among individuals who would normally seek work in agriculture, although Massey, Donato,
and Liang (forthcoming) present data suggesting new starts have not decreased. Similarly, it may represent the dwindling of annual circulatory migration or, in other words, a general decline in return migration after the winter
holiday season.

Discussion and conclusions

This paper has developed and estimated a statistical model to analyze the
determinants of linewatch apprehensions at the southern U.S. border over the
period 1977-1988. The first part of the time span includes the late 1970s and
early 1980s when the number of Border Patrol apprehensions was relatively
stable and the Mexican economy underwent a period of relative prosperity.
The period also includes the mid-1980s when apprehensions increased sharply
as the Mexican economy deteriorated owing to the global collapse of oil prices
in 1982. Finally, the series encompasses a two-year period after IRCA was
signed into law on November 6, 1986. Our research devoted special attention
to a quantitative assessment of the components of changes in apprehensions
during this IRCA period and develops estimates of the amount of change
associated with (1) enrollment in the SAWs program; (2) changes in INS
enforcement strategies; and (3) several dummy variables designed to capture
the remaining effects of IRCA.
The results show a substantial decline in linewatch apprehensions during the
post-IRCA period resulting from the implementation of immigration reform.
We estimate an overall net decline in apprehensions between November 1986
and September 1988 of nearly 700,000, or an amount about 35 percent below

111
the level that would be anticipated in the absence of IRCA. But it is impossible
to tell from this overall total alone whether IRCA is having its intended effect.
Because IRCA aims to reduce the number of illegal border crossings and
simultaneously to raise the probability of being apprehended at the border, it
is in principle possible for IRCA to be highly effective and for there to be at the
same time a relatively small change in the aggregate number of linewatch
apprehensions due to IRCA. 21For this reason, it is necessary to disaggregate
the 'total IRCA effect' into its three separate components. As already noted,
about 12 percent of this decline is due to reductions in INS enforcement effort,
about 17 percent is due to the agricultural legalization program, and about 71
percent is due to the remaining effects of IRCA. Because INS enforcement
strategies have had the perverse effect of lowering linewatch apprehensions in
the post-IRCA period, one may conclude that IRCA has been considerably
more successful in reducing illegal border crossings that it has in increasing the
probability of being apprehended by the Border Patrol.
These results considerably refine our understanding of the reasons for the
post-IRCA decline in linewatch apprehensions. In particular, they suggest
that IRCA has not exerted a simple or uniform influence on apprehensions
but, rather, has resulted in several effects sometimes working in offsetting
directions. They also have a number of policy implications concerning the
likely future impact of IRCA's provisions on the course of apprehensions.
Because the enrollment period for the SAWs program ex