5 Million Illegal Immigrants: An Analysis of New INS Numbers

By Steven A. Camarota July 1997

pp. 1-4 in Immigration Review no. 28, Spring 1997

On February 7, the Immigration and Naturalization Service (INS) released its latest estimate for the size and growth of the illegal alien population in the United States, updating its 1994 report. The INS estimates that as of October 1996 there were five million illegal aliens living in the United States, with the number growing by 275,000 each year. These new numbers are for the "long-term," illegal population — those who have been in the United States for at least one year. Although the rate of increase is 25,000 less than the 1994 estimate, the new national total is higher than INS previously thought. The change in the growth does not represent a reduction in the flow of illegal aliens, but instead reflects a change in the methodology used to calculate the numbers.

This article will review the basic findings and methodology of the estimate, plus raises questions about how the estimates were arrived at.


  • The number of illegal aliens who remained after the completion of the Immigration Reform and Control Act (IRCA) legalization in October of 1988 was 2,775,000 — well above the 2.18 million previously estimated.

  • The number of new illegal aliens joining the long-term population each year is 420,000. This number is offset by deaths, emigration and adjustment to legal status, so that the increase in the illegal population is 275,000 annually.

  • There were 170,000 new overstayers each year between 1982-1992 and 181,000 between 1992-1996. The number of new illegals who joined the illegal population by Entering Without Inspection (EWI) was 250,000 from 1982 to 1988 and 242,000 from 1988 to 1996.

  • The INS estimates that 41 percent of the illegal population are overstayers and 59 percent are EWIs. This is a change from the estimated 50-50 split in its previous study.

  • The illegal population is 54 percent Mexican.

  • 40 percent of the total illegal population lives in California.

Summary of Methodology

The INS has yet to release a detailed methodological statement on exactly how it arrived at its current estimate. However, based on the press release that accompanied the figures and discussions with Robert Warren, the INS demographer who developed the estimate, it is possible to summarize the report's methodology. (The method used to arrive at these numbers is quite complex and space does not permit a detailed explanation of all the assumptions, calculations, sources and adjustments used to create these estimates).

The estimate for the size of the illegal population in October 1988 is based partly on an estimate of the flow of illegals as indicated by non-immigrant overstay data and the results of the IRCA legalization.1 Additionally, analysis of the November 1989 Current Population Survey (CPS) is used to estimate the Mexican portion of the 2,775,000 figure for the illegal population in October 1988. The number of overstayers is calculated by matching I-94 forms (collected by airlines and Customs upon departure) with an INS database: an allowance is made for errors made by airline staff in collecting departure forms. Persons for whom there is no record of departure one year after they should have are assumed to have joined the long-term illegal population.

Calculating the flow of EWIs is more complex. Mexican EWIs, who account for about 80 percent of all EWIs, are calculated using CPS and INS admissions data for the period after 1990. Legal Mexican immigration numbers are compared with the number of Mexicans counted in the March 1994, 1995 and 1996 CPS. The difference is assumed to represent the illegal Mexican population who entered after 1990. Subtracting the number of Mexican overstayers, as indicated by I-94 form analysis, reveals the number of Mexican EWIs between 1990 and 1996. The flow of EWIs for all nationalities, with the exception of Mexicans in the 1990s, is based on information from the IRCA and SAW legalizations, apprehensions, and a few other sources.2

Reductions in the illegal population come from three sources: death, emigration and adjustment to legal status. Emigration figures are based on Census Bureau estimates of the return rate of migration for immigrants indicated by comparing 1980 and 1990 Census data (see Census Bureau Technical Paper number 9). The Census Bureau's rate of return migration is then increased by 10 percent in order to reflect higher emigration rates for illegals. Return migration rates are assigned to illegals based on year of entry. The death rate for illegal population is assumed to be only 3.9 per 1,000, reflecting the young age and almost non-existent infant mortality rate in the illegal population.3

Those illegals who become legal immigrants each year are controlled for in two ways. First, overstayers who adjust to legal status are removed from the overstay population by matching immigrant data with the overstayer database. EWIs from Mexico who legalize each year are accounted for because the method used to estimate the size of the illegal Mexican population is derived by comparing the number of Mexicans in the CPS with the number of Mexicans who enter legally. For EWIs from countries other than Mexico, the INS does not make any allowance for legalizations.

State Distribution

The previous INS estimates for individual states were based on the state distribution of IRCA beneficiaries. In the new study the portion of the illegal population that arrived before 1982 is again assigned by state based on the distribution of IRCA amnesty recipients. The state distribution of the illegal population that arrived between 1982 and 1988 is based on the residence pattern of SAW beneficiaries. For EWIs that arrived between 1988 and 1996, state distributions are assigned based on the state distribution of dependents legalized under IRCA. The assignment of overstayers by state for 1982 to 1996 is based on a study of the geographical distribution of overstayers between 1986 and 1989 (see Warren in Undocumented Migration to the United States, edited by Frank Bean, Barry Edmonston, and Jeffrey S. Passel. RAND Corporation and Urban Institute Press, Washington, D.C., 1990).

Again it should be noted that the above outline is only a brief summary; the actual methodology is significantly more complex. The INS expects to release a detailed methodological description of the study soon. Even without a detailed methodological statement, it is clear that the estimation procedure is the product of careful and thoughtful analysis. However, despite the rigorous nature of the estimating procedures, there remain several important methodological issues surrounding the numbers.

Questions Arising from The Methods

It is important to keep in mind that the INS estimates are only for the long-term illegal population and do not include those illegal aliens who stay in the United States for less than one year. These "short-term" illegal aliens have still broken the law. Moreover, by holding low-skilled jobs, they are competing with natives and legal immigrants in a section of the economy where wages are already low. And even though their stay is brief, they may still use costly taxpayer-provided services such as public hospitals and the criminal justice system. The number of illegal aliens who stay less than one year must be well over one million. For a discussion of short-term illegal immigration read Bean, Espenshade, White and Dymowski in Undocumented Migration to the United States, 1990.

Probably the most important question involves the use of data from the IRCA and SAW amnesties to estimate the net addition of EWIs for illegals from countries other than Mexico in the 1990s. The nativity of amnesty beneficiaries reflects the EWI flow of an earlier period. By the 1990s it is very possible that the level of EWIs and the source countries may have changed. Basing Mexican EWIs on the results of the CPS partly alleviates this problem because Mexican EWIs represent the vast majority (80 percent according to the INS) of EWIs each year. However, for illegals from countries such as China, the use of IRCA data seems questionable. Emigration pressures increased significantly in the 1990s as the pace of modernization accelerated in that country. Thus, to assume that the number of EWIs from China in the 1990s is similar to the 1980s may significantly underestimate EWIs. This may account for the seemingly low estimate of 25,000 chinese illegal aliens.

Using CPS data to determine Mexican EWIs may also present some problems, possibly resulting in an estimate for the illegal population that is too low, because there is some question as to whether the CPS does in fact count the total illegal population. It is true that starting in January of 1994 the CPS numbers reflect new weighting procedures designed to control for the undercount of illegals. However, the Census Bureau does not claim that the new adjustments count the entire illegal population. The corrections done to the CPS reflect the estimated size of the illegal population enumerated in both the 1980 and 1990 Census. Thus, the number of immigrants found in the CPS is increased to reflect the flow of illegals between 1980 and 1990 as indicated by the Census. However, the Census Bureau recognizes that neither the 1980 nor 1990 Census counted the entire illegal population. Thus, the adjustments made to the CPS do not consider the Census undercount. Moreover, the flow of illegals may have changed since 1990. Since the adjustments to the CPS are based on the Census, they cannot reflect this change. Therefore, there is reason to suggest that the INS's use of CPS may understate the size of the illegal Mexican population.

In response to this situation the INS states that their illegal figures are still valid because their estimate is for the long-term illegal population, which is much more likely to show up in CPS data. Moreover, the size of the long-term illegal population is partly inflated by those illegals counted in the CPS who stay less than one year. Though these may be reasonable arguments they are not based on systematic evidence. There is still the very real possibility that a significant portion of the long-term illegal population is missed even if "long-term" illegals are more likely to show up in the CPS than are "short-term" illegals. The Census Bureau does not make any claim about the percentage of the illegal population, "long-term" or otherwise, that shows up in the CPS.

The method used in the new study for emigration is also an important issue. The INS uses a return migration rate for illegals determined by the Census Bureau. The Census Bureau's return migration rate is based on a comparison of the 1980 and 1990 Censuses. The INS takes this return rate and adds 10 percent to it in order to reflect what it believes to be higher rates of return migration for illegals. The method assumes that overstayers and EWIs have the same return migration rates. Given the high cost of air travel and the fact that most overstayers are not from Mexico or Central America, this assumption may not be correct. The high cost of returning home for nationals from distant countries may reduce their propensity to return home. Also, there is no systematic rationale for assuming that return migration for illegals is dramatically higher than for immigrants in general.

A final methodological question surrounding these numbers concerns the state distribution of EWIs and overstayers in the 1990s. Using the state distribution of legalized dependents to determine the state distribution of EWIs in the 1990s creates state estimates that reflect illegal immigration of an earlier time, because legalized dependents are simply joining those who were granted amnesty in the 1980s. The distribution of overstayers in the 1990s is based on a study of their state distribution done in 1990 for the years 1985 to 1988. Like the distribution of EWIs, this method may produce results that reflect the 1980s and not the 1990s.

There are some minor issues associated with the new numbers as well. Because the INS used a uniform method to calculate the number of illegal aliens for each country, there are anomalous results for a few nationalities. For example, the illegal population from the Bahamas is estimated at 70,000, making it the 8th largest source country, even though the total population of that country is only 273,000. Data from the non-immigrant information system for that country is unreliable, which makes it appear as if the number of overstayers is dramatically higher than is possible. The 32,000 figure for the Caribbean island nation of Dominica is also grossly inflated (the total population of the country is 88,000). The INS estimates that roughly 25,000 of these individuals are really from the Dominican Republic, bringing the total illegal aliens from that country to 75,000.

Even with the issues outlined above, there can be no doubt that the new estimates represent a good faith effort on the part of the INS to deal with an extremely complex and difficult question. The INS has solicited input from the GAO, Census Bureau, the U.S. Commission on Immigration Reform, and elsewhere. Thus, these numbers have been well reviewed. Finally, the INS does not claim that these number are "set in stone." They anticipate a process of continual revision and refinement. Perhaps when better data and estimating procedures become available future estimates will deal with some of the methodological issues discussed above.

1 While most overstayers are visa holders, the INS estimates include non-visa holders as well.

2 Persons who legalized under IRCA and SAW were required to report how they entered, and these data are used to estimate the level of EWIs for many nationalities before, during and after the legalizations.

3 Illegal aliens have an almost nonexistent infant mortality rate because children born to illegals are automatically citizens.