U.S. Bureau of the Census
Washington, D.C. 20233
POPULATION DIVISION TECHNICAL WORKING PAPER NO. 8
The views expressed are attributaable to the author and do not necessarily reflect the views of the U.S. Bureau of the Census
Edward W. Fernandez & J. Gregory Robinson
U.S. Bureau of the Census
The following presents updated indicatorsof the relative distribution of the undocumented immigrant population by State. Previously, we had issued unofficial indicators by State for April 1993 based on a U.S. total of 4,000,000 undocumented immigrants (see Appendix). That illustrative set was often misconstrued as representing the Census Bureau's "official" estimates of the number of undocumented immigrants for States.
Given the nature of the undocumented immigrant population, any estimation of its magnitude and distribution obviously will require some broad assumptions. Frequently, these assumptions are not directly measurable; and we believe that any results generated by analysis should be characterized not as "point estimates" but as indicators of magnitude presented in the form of ranges of estimates (i.e. estimate intervals showing "lower" and "upper" limits or bounds). To advance this view, we have extended our previous technique for estimating the number of undocumented immigrants by State. Our new results - in the form of estimate-ranges - are more reasonable than the single (i.e. point) estimates we presented earlier. We believe that the illustrative technique used here to generate these ranges is clearly more responsive to the inherent uncertainty existent in any single set of "point" estimates of the undocumented immigrant population.
Estimate-ranges or "indicators" of the distribution of undocumented immigrants
Firstly, to derive our ranges of undocumented immigrants by State, we have used two national estimates of the undocumented population for Oct. 1992. The first is a preliminary estimate of 3.4 million undocumented persons updated recently by the Immigration and Naturalization Service (INS), (Warren, 1994). This INS estimate is based partly on analysis of non-visa "overstayers" and IRCA legalization data. The second is based on the Census Bureau's estimate of 3.3 million undocumented immigrants for April 1990 carried forward to Oct. 1992 (giving an estimate of 3.8 million). Based on these estimates and the results of continuing research at both the Census Bureau and the INS, we believe a good up-to-date "working" estimate of the total number of undocumented immigrants in the Nation (as of April 1994) would be between 3.5 and 4.0 million persons. Consequently, we have used this range to establish these national estimates as a mainstay on which to derive our State estimate-ranges of undocumented immigrants, as described below.
Secondly, to apportion the number of undocumented immigrants for each State, we calculated percent distributions based on: (1) an average between the number of undocumented immigrants by State included in the 1980 census and of the number of legalization applicants by State produced by the IRCA law of 1986; (2) the number of undocumented immigrants by State estimated by INS; and (3) the number of foreign born non-citizens by State counted in the 1990 census.
Thirdly, multiplying the three percent distributions noted above to each of these two national estimates, we generated six estimates of undocumenteds for each State. From these, we chose the lowest value as the lowest limit or "bound" and the highest value as the "upper" bound of the estimate-range for the State. For example, based on a national estimate of 3.5 million undocumented immigrants in the U.S., the estimates for California are 1.3 million, 1.5 million, and 1.6 million persons; alternatively, based on a national estimate of 4.0 million undocumented immigrants, the estimate for California includes 1.5 million, to 1.7 million, and 1.8 million persons. Using both national estimates, therefore, the overall range for California shows a "lowest" estimate of 1.3 million undocumented immigrants and a "highest" estimate of 1.8 million of these persons.
Similarly to the nation's foreign born population, as well as to some large racial and ethnic groups in the U.S., undocumented immigrants are concentrated also in specific geographic areas; and their asymmetrical distribution by State is noteworthy (see Table 1 and Figure 1). Regardless of method or process of estimation, California emerges unequivocally as the State with the greatest number of undocumented immigrants. Although we do not have an exact single number of undocumented persons for California, we nevertheless note that its share of these persons is very large relative to that of any other State. Furthermore, only a handful of States in the nation contain the vast majority of undocumented immigrants. For example, other States (besides California) with ranges that average more than 100,000 undocumented immigrants are: New York, with between 462,000 and 539,000 undocumented immigrants; Texas, from 300,000 to 427,000; Florida, with between 243,000 and 385,000; Illinois, from 157,000 to 225,000; and finally, New Jersey, with an estimated range of from 98,000 to 168,000 undocumented immigrants. For most of the remaining States, the estimated number of undocumenteds were substantially lower.
For comparative purposes, we may think of the magnitude of the undocumented immigrant population of each State as being included in one of three tiers. California alone occupies the first tier, with both the lowest and highest bounds falling above one million. Five States--New York, Texas, Florida, Illinois, and New Jersey--comprise the second tier, which includes States with estimates of undocumented immigrants of less than 1 million but mostly greater than 100,000 persons. Finally, the remaining States (which include the vast majority of States) occupy the third tier; in this group, the estimates of undocumented immigrants are all relatively small.
Although there is scarce information on the undocumented immigrant population, we believe our derived indicators or ranges of the size of this population provide - at the very least - useful illustrations of the current relative distribution of undocumented immigrants by State. Our conceptual basis for developing these estimate-ranges was intuitive; and our procedures straightforward. To generate the ranges, for instance, we assumed that the State distributions of undocumented persons followed a pattern similar to that of foreign born non-citizen immigrants enumerated in the 1990 census by State; and to other available items that pertained specifically to the undocumented population such as the INS estimates and the distributions of IRCA legalizations, etc.
The range of estimates we present here are neither official nor final, and we intend to continue our research on the magnitude and distribution of the undocumented immigrant population at the national and at lower geographic levels. In particular, we are trying to consolidate and improve upon previous work because earlier methods include some assumptions that need re-working and re-thinking. We are also investigating the use of alternative analytic techniques, and reviewing the more recent methods of other researchers. One approach, for which we will soon have results, is the development of "residual" estimates of the number of undocumented immigrants at the national level based on an in-depth comparative analysis of the 1990 census foreign born immigrant population classified by country-of-birth and period-of-entry. We wish to emphasize again, however, that the national and State undocumented "point" estimates recently produced by the Immigration and Naturalization Service (INS) staff seem highly plausible.
Finally, we continue our research to estimate, as accurately as possible, the size and distribution of the undocumented immigrant population in the U.S. and States. And we stress that, though we lack "point" estimates of this population, we believe our "indicator-ranges" are particularly meaningful because they underscore the relative differences in the concentration of these persons by State.
Brief Summary of Procedures for Deriving National
Estimates of Undocumented Immigrants, 1992-1994
Briefly, our previous estimate of 4.0 million undocumented immigrants in the U.S. in 1993 was the result of augmenting an estimate of 3.3 million undocumented persons, calculated for 1990, by an estimate of net annual increase for this population of about 200,000 persons per year. The estimate for 1993 of 4.0 million undocumented persons was, therefore, the sum of the 1990 estimate plus the estimated three-year increase of this population to 1993. Presently, however, recent research from both the Census Bureau and the INS leads us to believe that a reasonable "working" estimate-range of the total number of undocumented immigrants in the U.S. as of April 1994 is between 3.5 and 4.0 million persons. And based on these upper and lower "bounds", we have proceeded to estimate the ranges of the number of undocumented immigrants in each individual State for that date.
We note below the procedures used to estimate the total of 3.3 million undocumented immigrants in the U.S. in 1990, and to calculate the net annual growth of that population. We also describe the rationale behind our preliminary calculations generating the 1994 distributions by State of the undocumented immigrant population.
Estimation of the Number of Undocumented Immigrants in 1990
The April 1990 estimate of 3.3 million undocumented immigrants was derived by applying two techniques. The first technique estimated the number of undocumented immigrants included in the November 1989 CPS by the residual method; this estimate was then increased to allow for undocumented persons not included (i.e. enumerated) in the November 1989 CPS. The November estimate was subsequently carried forward to April 1990 based on assumptions about population change. The second technique used an estimate of undocumented immigrants included in the 1980 census (about 2.0 million persons) and then "extended" this estimate to 1990 by using alternative estimates of change of the undocumented immigrant population during the 1980 to 1990 intercensal period.
Uncertainties in the underlying data used to estimate undocumented immigration by the above techniques required making some broad assumptions about that data; therefore, instead of "point" estimates" the estimates were derived as "estimate-ranges" of the number of undocumented persons. Ultimately, the "estimate-ranges" calculated from both of the above techniques converged to a single "point" estimate of 3.3 million undocumented persons in the U.S. on April 1990.
Exclusion of 'SAWs' in the Preliminary 1990 Estimate
The above preliminary estimate of 3.3 million undocumented immigrants for April 1990 includes a number of Special Agricultural Workers (SAW's) who may have become legalized U.S. residents under the IRCA law of 1986; and some researchers believe this makes the 3.3 million estimate too high. Certainly, a realistic estimate of undocumented population size must clearly consider the SAW population in the U.S. and their legal and non-legal status. Therefore, it is quite possible that the estimate of 3.3 million undocumented persons for 1990 is too high.
However, there are two factors that may tend to counteract this SAW effect and suggest that our 1990 estimate is too low. One factor is the assumption about the 1980 census "undercount" of undocumented immigrants. Although 2.0 million undocumented immigrants were estimated as counted (i.e. enumerated) in that census, it was assumed that another 1.0 million were not counted - an implicit undercount-rate of 33 percent. Given the elusive character of the undocumented population, we may contend that the undercount was larger, and that a national estimate of 3.0 million in 1980 was really too low. Similar reasoning applies to the national estimate of undocumented immigrants derived via the foreign born immigrant population enumerated in the November 1989 CPS. A second factor suggesting that our 3.3 million estimate for 1990 is too low is the possibility that we may have understated the growth - assumed at 200,000 persons a year, as noted below - of the undocumented population since 1990.
Net Annual Growth of the Undocumented Population
Determination of the annual growth of the undocumented population during the 1980-1990 period was based on Current Population Survey (CPS) estimates of the foreign born population. Comparisons of the estimates of the undocumented population determined from the 1979, 1986, and 1988 CPS' generated estimates of net undocumented population change for the 1979-1986; 1979-1988, and 1986-1988 periods. These indicators suggested a net annual undocumented immigration rate of increase of about 200,000 persons per year. Comparisons between the November 1989 CPS and the 1979 and 1986 CPS' supported this rate (Woodrow, 1991). Recent work by Warren (1994), however, suggests that the net annual growth of the undocumented population is more than 200,000 persons and probably closer to 300,000 persons per year. For our own "working" upper bound estimate of approximately 4.0 million undocumented immigrants in the U.S. for 1994, we simply extended our 1990 estimate of 3.3 million persons at a rate of increase of about 200,000 undocumented persons per year. Our lower bound of 3.5 million was based on recent estimates of the undocumented population by the INS.
Undocumented Immigrants by State: 1994
As noted above, to estimate the range of the number of undocumented immigrants by State, we applied each of the following State percent-distributions of undocumenteds to the two national estimates of the total number of undocumented immigrants in the U.S. in April 1994 (i.e. 3.5 and 4.0 million, respectively): (1) the average of the State percent distribution of the number of undocumented immigrants included (i.e. counted) in the 1980 census (Passel and Woodrow, 1984) and the percent distribution of the number of undocumented immigrants applying for legal resident status under the 1986 IRCA law; (2) the percent of the number of undocumented immigrants by State estimated by the INS; and (3) the percent distribution of the number of foreign born non-citizen immigrants counted in the 1990 census. For each State, therefore, we derived six individual estimates of the number of undocumenteds in the State. To derive a single range for each State, we chose the lowest and highest values among the six estimates: one representing a "lower" bound and the other an "upper" bound of the number of undocumented immigrants in each State as of April 1994.
Apart from currently presenting our estimates of undocumented immigrants as "indicator-ranges" instead of as "point" estimates, we believe we have also strengthened and increased the plausibility of our State estimates by using country-specific breakdowns in getting our national estimate (3.5 or 4.0 million persons) distributed to State levels; specifically, we calculated separate distributions by State for persons born in Mexico, El Salvador, and Guatemala and fof persons born in all other countries. We expect this refinement enhances our State range-estimates of undocumented immigrants.
Finally, our latest preliminary research results on the number of undocumented persons counted (i.e. enumerated) in the 1990 census are consistent with the national estimates used here. Presently, our research perseveres in the attempt to estimate the total number of undocumented immigrants for the U.S., States, and for smaller geographic levels.
(1) Warren, Robert and Jeffrey Passel. 1987. "A Count of the Uncountable: Estimates of Undocumented Immigrants Counted in the 1980 Census," Demography, 24:375-393.
(2) Woodrow, Karen. 1991. Demographic Analysis Evaluation Project D2, "Preliminary Estimates of Undocumented Residents in 1990." 1990 Decennial Census; Preliminary Research and Evaluation Memorandum No. 75. U.S. Bureau of the Census, Wash. D.C.
(3) Passel, Jeffrey S. and Woodrow, Karen, "Geographic Distribution of Undocumented Immigrants: Estimates of Undocumented Aliens Counted in the 1980 Census by State", International Migration Review, 18:642-671, 1994.
(4) Warren, Robert, "Estimates of the Unauthorized Immigrant Population Residing in the United States, by Country of Origin and State of Residence: October 1992" Unpublished report, INS, April 1994.
Illustrative Ranges of Estimates of Undoc. Immigrants for Selected States
Table 1: Illustrative Ranges of Estimates of the Undocumented Resident Populatioin by State: 1994