This is the accessible text file for GAO report number GAO-10-99 
entitled 'U.S. Labor Force Statistics: Illustrative Simulations of the 
Likely Effects of Underrepresenting Unauthorized Residents' which was 
released on December 7, 2009. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as part 
of a longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

Report to the Chairman, Subcommittee on Income Security and Family 
Support, Committee on Ways and Means, House of Representatives: 

United States Government Accountability Office: 
GAO: 

November 2009: 

U.S. Labor Force Statistics: 

Illustrative Simulations of the Likely Effects of Underrepresenting 
Unauthorized Residents: 

GAO-10-99: 

Contents: 

Letter: 

Results in Brief: 

Agency Comments and Our Evaluation: 

Appendix I: Briefing Slides: 

Appendix II: Comments from the Department of Commerce: 

Abbreviations: 

ACS: American Community Survey: 

BLS: Bureau of Labor Statistics: 

CPS: Current Population Survey: 

DHS: Department of Homeland Security: 

[End of section] 

United States Government Accountability Office: 

Washington, DC 20548: 

November 30, 2009: 

The Honorable Jim McDermott: 
Chairman: 
Subcommittee on Income Security and Family Support: 
Committee on Ways and Means: 
House of Representatives: 

Dear Mr. Chairman: 

In times of economic uncertainty as well as in times of stability, 
policymakers and the public rely on labor force statistics, such as the 
unemployment rate, to provide important information on the current 
state of the economy. These statistics, published by the Department of 
Labor's Bureau of Labor Statistics (BLS), include key figures that are 
based on data obtained from the Current Population Survey (CPS). The 
CPS, a household interview survey administered by the Department of 
Commerce's U.S. Census Bureau, is designed to represent the entire 
United States civilian noninstitutional population. However, certain 
U.S. residents--specifically, foreign-born persons who are not 
authorized to live here, to whom we refer as unauthorized residents in 
this report--may not be represented in CPS data to the same extent as 
the general population. Little research has been done on whether 
potential underrepresentation of the unauthorized population might 
noticeably affect labor statistics, but policy efforts that depend on 
valid and reliable labor force statistics would benefit from such 
information. 

We agreed with your office to examine issues concerning unauthorized 
residents' impact on labor force statistics. We addressed the following 
key questions: 

1. Extent of underrepresentation: What is known about the extent of any 
underrepresentation of unauthorized residents in CPS data used to 
compile labor force statistics? 

2. Labor force status: What is known about the likely labor force 
status of unauthorized residents? 

3. Possible effects: How might CPS underrepresentation of unauthorized 
residents affect key labor force statistics? 

To answer the first two questions, we examined documents and data from 
and interviewed officials from the Census Bureau, the Department of 
Homeland Security, and BLS. We also interviewed and obtained data from 
immigration and statistical experts outside these agencies (see slide 
44 in app. I for a listing of these experts and their roles). 
Additionally, we reviewed previous GAO work and other studies on 
unauthorized residents, government surveys, and labor force statistics 
and interviewed immigration advocates about these topics. To answer the 
third question, we conducted simulations to illustrate how 
undercounting unauthorized residents may have affected the unemployment 
rate, size of the U.S. labor force, and number of employed persons in 
March 2008. We determined that March 2008 was the most recent time 
period for which data were available that were sufficiently reliable 
for the purposes of our main simulations. We created the simulated 
statistics by first using combinations of assumptions obtained from 
experts about the extent of underrepresentation of unauthorized 
residents and their level of unemployment in March 2008. Then, because 
the estimates obtained from experts were uncertain, we used much 
broader assumptions to test the robustness of our results. We compared 
all of the simulated statistics with the margins of error for the 
original statistics. We also updated these simulations using June 2009 
national labor force statistics as a test of whether changing economic 
conditions might affect our results.[Footnote 1] 

We conducted our work from November 2008 to November 2009 in accordance 
with all sections of GAO's Quality Assurance Framework that are 
relevant to our objectives. The framework requires that we plan and 
perform the engagement to obtain sufficient and appropriate evidence to 
meet our stated objectives and to discuss any limitations in our work. 
We believe that the information and data obtained, and the analysis 
conducted, provide a reasonable basis for any findings and conclusions. 

Results in Brief: 

On October 2, 2009, we briefed your staff on the results of our work. 
This report formally conveys the information provided during that 
briefing (see appendix I for the briefing slides). In general, we found 
the following: 

* The extent of CPS underrepresentation of unauthorized residents is 
unknown, but experts we consulted told us that the CPS data might not 
represent approximately 10 to 15 percent of unauthorized residents. 

* Little information is available about the labor force status of 
unauthorized residents, but experts we consulted suggested that their 
approximate unemployment rate in March 2008 may have been 6.5 to 8.5 
percent, compared with 5.2 percent nationally. 

* Little is known about the effect of underrepresentation of 
unauthorized residents on labor force statistics. However, using the 
above information provided by experts, we simulated the likely effects 
of adding unauthorized residents assumed not represented in March 2008 
labor force statistics. Because the expert assumptions were uncertain, 
we tested our results using a broader range of assumptions. Our 
simulations showed that adding unauthorized residents not represented 
in CPS data would likely have a minimal effect on the overall U.S. 
unemployment rate. The simulations did not, however, rule out the 
possibility that adding such residents would increase the estimated 
size of the national labor force and the estimated number of persons 
employed nationally. Updating these simulations using national labor 
force statistics for June 2009 revealed similar effects. 

Agency Comments and Our Evaluation: 

Prior to our October 2, 2009, briefing, we provided the Departments of 
Labor, Homeland Security, and Commerce with a draft of the briefing 
document and incorporated technical comments provided by the Bureau of 
Labor Statistics and the Census Bureau as appropriate. Homeland 
Security had no comments on the slides. Since then, Labor, Homeland 
Security, and Commerce have reviewed a draft of this report. Homeland 
Security had no comments, and BLS informally remarked positively on the 
way that our draft accounted for uncertainty related to unauthorized 
workers. Commerce provided written comments, reproduced in appendix II, 
which did not dispute any of our three main findings concerning the 
possible impacts of underrepresentation of unauthorized residents on 
national labor force statistics. However, Commerce suggested that we 
(1) further explicate how survey data are weighted to improve 
statistical representation and how we conducted our simulations, as 
well as providing--in this letter--more specifics about the experts we 
consulted; (2) remove information on the two-card (or "grouped 
answers") approach for estimating the number of illegal immigrants 
represented in a survey, because including it implies both that this 
method is appropriate for use in the CPS and that it can be used to 
estimate underrepresentation;[Footnote 2] and (3) reduce the range of 
underrepresentation in our illustrative simulations, which we vary from 
0 to 50 percent, because Commerce stated that using this range gives 
the range more plausibility than is supported by research. 

In response to Commerce's first suggestion, we added some additional 
information to the slides. One new footnote cites a Census Bureau 
publication that provides detailed information on how CPS data are 
weighted, and another new footnote provides a numerical illustration of 
how we conducted our simulations. We also added a parenthetical note in 
this letter, indicating the slide that lists the experts we consulted 
and their roles, and clarified that we consulted experts in both 
immigration and statistics. 

In response to Commerce's second suggestion, we did not change our 
presentation of information on the grouped answers method for the 
following reasons: 

* Our slide presentation clearly states that the grouped answers method 
is appropriate for a survey conducted by a private-sector organization. 
Consistent with Commerce's comments, we concluded in our September 2006 
report[Footnote 3] on estimating the unauthorized population that the 
grouped answers method is not appropriate for any existing government- 
conducted survey, including the CPS; the 2006 report therefore raised 
the possibility of a new survey designed for the foreign-born 
population. If the grouped answers method were validated and included 
in a private-sector survey of the foreign-born, the resulting data 
logically could be used to help assess statistics based on a government-
conducted general population survey such as the CPS, as well as serve 
various other policy and evaluation purposes.[Footnote 4] 

* While the grouped answers method is not intended to estimate 
underrepresentation, if validated and used in a new private-sector 
survey, it could provide information on specific groups within the 
foreign-born population (such as the number of employed unauthorized 
workers and the number of unauthorized workers in the labor force). 
Such data would be useful to the task addressed in this report in at 
least two ways. First, in the illustrative simulations we conducted, we 
made assumptions about unemployment levels within the unauthorized 
worker population because there was no survey estimate of unemployment 
for this group. The grouped answers method, applied in a new private- 
sector survey, might provide such an estimate. Second, our illustrative 
simulations used an indirect estimate of the unauthorized worker 
population represented in the CPS; this indirect estimate had been 
produced by a researcher using a subtraction method (see slide 12 of 
app. I) and various assumptions regarding labor force status in this 
group. A private-sector survey using the grouped answers method could 
help researchers using the subtraction method or other indirect methods 
by, for example, providing a survey estimate of the number of 
unauthorized workers represented in a private-sector survey.[Footnote 
5] 

Regarding Commerce's third suggestion, we did not reduce the range of 
the test assumptions for underrepresentation of unauthorized workers in 
the CPS. Our reasons are that this report (1) already presents analyses 
that use fairly narrow ranges of assumptions deemed most plausible by 
experts (for both underrepresentation and unemployment among 
unauthorized workers) and (2) uses much broader ranges of 0 to 50 
percent to test results produced using the fairly narrow ranges. We 
conducted the test simulations because of uncertainty about the 
narrower expert-based ranges. Specifically, using alternative 
assumptions of 0 to 50 percent, we tested how far "off the mark" the 
expert assumptions would have to be for our findings to change. Our 
tests show that the initial expert-based narrow-range results would 
hold up even if the expert-based assumptions were rather far off the 
mark. 

Finally, Commerce provided additional information about correspondence 
between GAO and the Census Bureau concerning an open GAO 
recommendation.[Footnote 6] This recommendation dates from 1998 and 
concerns devising a plan of research for evaluating the quality of 
census and survey data on foreign-born persons. There have been various 
communications in addition to the letter referenced by the Census 
Bureau,[Footnote 7] but the Census Bureau has not proposed a strategy 
or plan of research for evaluating the quality of census and survey 
data on foreign-born persons. Innovative thinking and collaboration may 
be required to make progress on this long-term recommendation. For 
example, one approach may be for the Census Bureau to design research 
and evaluation options that it might pursue in partnership or 
coordination with others (such as other federal agencies, universities, 
or private-sector organizations). 

As agreed with your office, unless you publicly announce its contents 
earlier, we plan no further distribution of this report until 7 days 
from the report date. We will then send copies of this report to 
relevant congressional committees, the Secretary of Labor, the 
Secretary of Commerce, the Secretary of Homeland Security, and other 
interested parties. In addition, this report will be available at no 
charge on GAO's Web site at [hyperlink, http://www.gao.gov]. 

If you or your staff have any questions about this report, please 
contact either Nancy Kingsbury at (202) 512-2700 or kingsburyn@gao.gov 
or Cornelia Ashby at (202) 512-7215 or ashbyc@gao.gov. Contact points 
for our Offices of Congressional Relations and Public Affairs may be 
found on the last page of this report. GAO staff who made key 
contributions to this report are listed on slide 47 of appendix I. 

Sincerely yours, 

Signed by: 

Nancy Kingsbury: 
Managing Director, Applied Research and Methods: 

Signed by: 

Cornelia M. Ashby: 
Director, Education, Workforce, and Income Security Issues: 

[End of section] 

Appendix I: Briefing Slides: 

U.S. Labor Force Statistics: Illustrative Simulations of the Likely 
Effects of Underrepresenting Unauthorized Residents: 

Briefing for Staff of Chairman Jim McDermott: 
Subcommittee on Income Security and Family Support: 
House Committee on Ways and Means: 

October 2, 2009: 

Prepared by GAO’s Applied Research and Methods and Education, 
Workforce, and Income Security Teams: 

Introduction: 

Labor force statistics are economic benchmarks for policymakers and the 
general public. 

* The Labor Department’s Bureau of Labor Statistics (BLS) produces key 
statistics, such as the unemployment rate, the size of the labor force, 
and the number of employed persons. 

* BLS compiles these statistics using data from the Current Population 
Survey (CPS), a monthly household interview survey that is designed and 
weighted to represent the U.S. population as measured by updated census 
counts. 

* The CPS is intended to represent the entire civilian noninstitutional 
population, including those foreign-born persons who are not authorized 
to reside in the United States. 

CPS data may represent many—but likely not all—unauthorized foreign-
born residents.[Footnote 8] 

* CPS data differentiate between native and foreign-born persons, but 
not between the foreign-born who reside here legally and those who 
reside here illegally (referred to as unauthorized residents in this 
briefing). 

* Indirect analyses suggest that millions of unauthorized residents may 
be represented in CPS data.[Footnote 9] 

* Some additional unauthorized residents may not be represented in 
these data. 

[End of section] 

Appendix II: Comments from the Department of Commerce: 

Introduction: Effects of Underrepresentation: 

The effect that underrepresenting unauthorized residents may have on 
CPS-based labor force statistics depends on two key factors: 

* The number of unauthorized residents who are not represented in CPS 
data.For example, if this number is small compared with the general 
population, then omitting data on them would probably have a small 
effect. 

* The labor force status of unauthorized residents who are not 
represented in CPS data. For example, if this group’s employment status 
is similar proportionately to that of the general population, then 
omitting data on them would have little effect on some statistics, such 
as the unemployment rate. 

Key Questions: 

For this briefing, we examined these questions: 

1. Extent Of Underrepresentation: What is known about the extent of any 
underrepresentation of unauthorized residents in CPS data used to 
compile labor force statistics? 

2. Labor Force Status: What is known about the likely labor force 
status of unauthorized residents? 

3. Possible Effects: How might CPS underrepresentation of unauthorized 
residents affect key labor force statistics? 

Scope and Methodology: 

To obtain background information and answer questions 1 and 2, we: 

* Examined documents and data from the Department of Commerce’s Census 
Bureau, the Department of Homeland Security (DHS), and BLS. 

* Reviewed existing GAO work. 

* Interviewed federal officials, immigration experts, and immigrant 
advocates about unauthorized residents and employment. 

* Examined selected academic and policy institute studies of 
unauthorized residents, government surveys, and labor statistics. 
[Footnote 10] 

To answer our third question: 

* We conducted simulations to illustrate how underrepresenting 
unauthorized residents might have affected three BLS statistics in 
March 2008:[Footnote 11] 
- unemployment rate; 
- size of the U.S. labor force; and; 
- number of employed persons. 

* We updated this analysis using the more limited data available for 
June 2009. 

* For these analyses, we obtained ranges of possible effects by varying 
our assumptions about the: 
- extent of underrepresentation of unauthorized residents and; 
- level of unemployment for unauthorized residents who were not 
represented in CPS data.[Footnote 12] 

See appendix A for more information on our methods. 

We did not focus on: 

* statistics for local areas, specific occupations, or economic 
sectors; 

* trends in labor force statistics over time; or; 

* other data sources for labor statistics, besides the CPS. (See app. 
B.) 

We provided a draft of this briefing to the Departments of Labor, 
Commerce, and Homeland Security for review. Within these departments, 
BLS and the Census Bureau provided technical comments that we 
incorporated as appropriate. DHS had no comments. 

We conducted our work from November 2008 to October 2009 in accordance 
with all sections of GAO’s Quality Assurance Framework that are 
relevant to our objectives. The framework requires that we plan and 
perform the engagement to obtain sufficient and appropriate evidence to 
meet our stated objectives and to discuss any limitations in our work. 
We believe that the information and data obtained, and the analysis 
conducted, provide a reasonable basis for any findings and conclusions. 

Summary of Findings: 

1. Underrepresentation: The extent to which unauthorized residents are 
underrepresented in CPS data is unknown, but experts we consulted 
suggested this might range from 10 to 15 percent. 

2. Labor Force Status: Little is known about unauthorized residents’
labor force status, but experts we consulted suggested that their 
unemployment rate may have been between 6.5 and 8.5 percent in March 
2008, versus a national rate of about 5.2 percent. 

3. Possible Effects: Our simulations for March 2008 and update for June 
2009 indicated that, despite uncertainties, 

* adding unauthorized residents not represented in CPS data would 
likely have a minimal effect on the U.S. unemployment rate, but, 

* increases to the estimated size of the U.S. labor force or the 
estimated number of employed persons cannot be ruled out. 

Background: Labor Force Statistics: 

BLS is the source of government labor force statistics, which include 
but are not limited to the following: 

* Unemployment Rate: Percentage of the labor force that is not working 
and is looking for work. 

* Labor Force Size: Number of persons who are working or looking for 
work. 

* Employed Persons: Number of persons who are working.[Footnote 13] 

BLS bases these three statistics on CPS data. BLS reports statistics at 
the national and state levels. 

The Current Population Survey (CPS): 

* The Census Bureau administers the CPS monthly to approximately 60,000 
households. 

* The CPS asks a household member to report characteristics of each 
individual in the household, including country of birth, as well as 
overall household characteristics. BLS bases its labor force statistics 
on CPS data for those aged 16 and older. 

* The CPS does not ask about authorization to reside or work in the 
United States. BLS officials told us that asking about legal status 
could discourage cooperation with the CPS. 

* The Census Bureau takes several steps to minimize underrepresentation 
in the CPS and increase its accuracy. See appendix C for more 
information. 

Almost 12 Million Unauthorized Residents Are Estimated to Reside in the 
United States: 

DHS and the Pew Hispanic Center have estimated the number of 
unauthorized residents by using existing data on foreign-born U.S. 
residents. Essentially, DHS and Pew have: 

* begun with an estimate of the number of all foreign-born persons 
residing here (including both those who are and those who are not 
authorized to live in the United States);[Footnote 14] 

* subtracted the number who are authorized to reside here;[Footnote 15] 

* made various assumptions about, for example, the effects of 
underrepresentation; and; 

* concluded that in early 2008, almost 12 million residents—
approximately one-third of all foreign-born—were unauthorized. 

Previous GAO Work: 

In 1998, we reported that policymakers need reliable information on 
foreign-born persons for policy-related activities and decision making, 
but such data were not available (GAO/GGD-98-164). 

* We recommended that the Census Bureau and DHS publish a plan of 
research to evaluate census and survey data on the foreign-born. 

* Given continued interest by Members of Congress in related issues, we 
continue to follow up on agency efforts in this area. 

* We are working with Census Bureau officials to plan a meeting to 
discuss letters we exchanged in 2008 and 2009 concerning difficulties 
that the Census Bureau anticipates in implementing some approaches to 
evaluating representation of the foreign-born. 

* In 2004, we reported that lack of data precluded us from estimating 
the cost of educating children who are not authorized to reside in the 
United States (GAO-04-733). 

* In 2006, we reported that foreign-born respondents appear to accept 
an innovative method of asking about legal status in personal 
interviews conducted by a nongovernment entity (GAO-06-775).[Footnote 
16] 

[End of Background section] 

Finding 1: Underrepresentation: 

The Extent of Underrepresentation Is Unknown, but Some Experts Suggest 
10 to 15 Percent: 

* Information on the number of unauthorized residents who are not 
represented in CPS data is unavailable. 

* Three experts told us that 10 to 15 percent of unauthorized residents 
might not be represented in CPS data, but these figures are uncertain. 
[Footnote 17] 

* One of the experts, Jeffrey S. Passel of the Pew Hispanic Center, 
used analytic procedures to estimate that the underrepresentation of 
unauthorized residents in the March 2008 CPS was approximately 12.5 
percent. 
- Passel applied Census Bureau estimates of undercounts for the overall 
U.S. population in 2000, by race, age, and gender, to foreign-born CPS 
respondents. He assumed higher undercount rates for recent and 
unauthorized immigrants.[Footnote 18] 

* The figures for the underrepresentation of unauthorized residents 
cited in this briefing are uncertain, because they rely on inferences 
based on demographic characteristics and/or a single study with limited 
scope. For example, the three experts we consulted on 
underrepresentation all referred to a key study that attempted to 
measure the size of the census undercount of unauthorized Mexicans in 
Los Angeles County.[Footnote 19] 

* In our analyses, we first simulated underrepresentation as ranging 
from 10 to 15 percent. Then, because the expert figures were uncertain, 
we tested the robustness of our results by broadening this range to 0 
to 50 percent. 

Note: The experts estimated the 10 to 15 percent range for the entire 
unauthorized population. To conduct our simulations, we applied this 
range to unauthorized residents in the labor force. 

[End of Finding 1] 

Finding 2: Labor Force Status: 

Little Is Known about the Labor Force Status of Unauthorized Residents: 

Labor force status includes: 

* having a job (employed); 

* having no job and looking for work (unemployed); and; 

* having no job and not looking for work (out of the labor force), such 
as homemakers, retirees, and “discouraged workers.” 

Limited information is available on whether the labor force status of 
unauthorized residents who are not represented in CPS data differs from 
that of: 

* unauthorized residents represented in the CPS or: 

* the overall, general population of the United States. 

Passel estimated that in March 2008, there may have been about 8.3 
million unauthorized residents in the U.S. labor force.[Footnote 20] 

Passel and two other experts who provided input on unemployment among 
unauthorized residents told us the following: 

* The unemployment rate of unauthorized residents is likely to vary 
over time and be affected by changes in the economy and government 
policies. 

* The unemployment rate for unauthorized residents might have been 
between 6.5 and 8.5 percent in March 2008, higher than BLS’s reported 
rate of about 5.2 percent for the general population. However, these 
figures are uncertain.[Footnote 21] 

In our simulations of the unemployment rate and number employed,we 
first assumed March 2008 unemployment rates for unauthorized residents 
of 6.5 to 8.5 percent. Then, because the expert figures were uncertain, 
we broadened this range to 0 to 50 percent to test the robustness of 
our results. 

[End of Finding 2] 

Finding 3: Possible Effects: 

Experts Do Not Know the Effect of Under-representation; Some Suggest It 
May Be Minimal: 

Some experts suggested to us that the effect of underrepresenting 
unauthorized residents on overall statistics would likely be small. 
However, due to a lack of data, experts ultimately do not know the 
effect of underrepresentation on labor force statistics. 

Our simulations focused primarily on three March 2008 statistics:
[Footnote 22] 

* unemployment rate (U.S. reported figure: 5.2 percent), 

* size of the labor force (U.S. reported figure: 153.1 million), and, 

* number of employed persons (U.S. reported figure: 145.1 million). 

We also conducted rough simulations to determine whether similar 
patterns of results might hold for June 2009.[Footnote 23] 

To test whether including unauthorized residents who were not 
represented in CPS data would change the three statistics listed above, 
we recalculated them using various assumptions about: 

(1) the underrepresentation of unauthorized residents and; 

(2) the unemployment rate of those who were not represented. 

Finding 3: Possible Effects (March 2008): 

Unemployment Rate: Likely Effect of Underrepresentation Was Minimal: 

According to BLS data, the March 2008 national unemployment rate was 
5.24 percent. The margin of error for this statistic is +/-0.17 
percent.[Footnote 24] 

We conducted two sets of simulations for March 2008: 

* In the first set of simulations, we varied underrepresentation from 
10 to 15 percent and varied unemployment rates for the unauthorized 
residents not represented in the data from 6.5 to 8.5 percent. 

* In the next set of simulations, we used a broader range for 
underrepresentation (0 to 50 percent) and unemployment rates (0 to 50 
percent) to test the robustness of the initial results. 

The first set produced simulated unemployment rates that fell within 
the margin of error of BLS’s reported rate (5.24 percent, +/-0.17 
percent), which we regard as a minimal effect. 

In the second set, only assumptions considerably more extreme than the 
ranges suggested by experts produced simulated unemployment rates that 
fell outside BLS’s margin of error. See figure 1. 

Figure 1: Simulated U.S. Unemployment Rates after Including 
Unauthorized Residents Assumed Not Represented in CPS Data, March 2008: 

[Refer to PDF for image: multiple line graph] 

Assumed underrepresentation of unauthorized residents (percent) plotted 
versus simulated U.S. unemployment rate (percent). 

UUR: Assumed unemployment rate among unauthorized residents. 

Lines representing the following are depicted: 

UUR: 0%; 
UUR: 6.5%; 
UUR: 8.5%; 
UUR: 6.5%; 
UUR: 15%; 
UUR: 20%; 
UUR: 30%; 
UUR: 40%; 
UUR: 50%. 

Additionally, the following are depicted: 

Simulations resulting from assumed underrepresentation and unemployment 
rates among the unauthorized that are within the range provided by 
experts (first set of simulations); 

Simulations resulting from assumptions more extreme than experts 
suggested (second set of assumptions); 

The 90% confidence interval around the BLS unemployment rate of 5.25%: 
5.08% to 5.41%. 

Source: GAO simulations based on reported BLS statistics and expert 
assumptions. 

[End of figure] 

Labor Force: Potential Increases Cannot Be Ruled Out: 

* BLS estimated that the March 2008 national labor force size was 153 
million. The margin of error for this figure is about +/-506,000. 

* For our simulations for March 2008, we assumed that the BLS estimate 
of 153 million includes 7.2 million unauthorized members of the labor 
force. 

* For our first set of simulations for this time period, we further 
assumed that the 7.2 million figure underrepresented unauthorized 
members of the labor force by 10 to 15 percent. The resulting 
simulations added 800,000 to 1.3 million more unauthorized members of 
the labor force to the BLS figure of 153 million, fora simulated total 
of roughly 154 million. 

* Our second set of simulations for this time period explored a broader 
range of assumptions for underrepresentation (0 to 50 percent). These 
simulations added 0 to 7.2 million to the BLS figure of 153 million, 
for simulated totals ranging from 153 to 160 million. 

* Most simulations produced simulated national labor force sizes that 
were above the margin of error of BLS’s reported labor force size.
[Footnote 25] 

Employed Persons: Potential Increases Cannot Be Ruled Out: 

In March 2008, BLS estimated that there were 145 million employed 
persons. The margin of error was about +/-541,000. 

Our simulations for March 2008 assumed, as on the previous slide, that 
the BLS labor force estimate of 153 million persons included 7.2 
million unauthorized residents. 

In the first set of simulations: 

* As on the previous slide, we varied underrepresentation from 10 to 15 
percent, adding from 800,000 to 1.3 million unauthorized residents to 
the labor force. 

* We then assumed that from 6.5 to 8.5 percent of the added 
unauthorized residents were unemployed, meaning that 91.5 to 93.5 
percent were employed; we applied these percentages to the 800,000 to 
1.3 million added to the labor force. 

* Resulting simulations added 700,000 to 1.2 million employed persons 
to the BLS figure of 145 million, for a simulated total of roughly 146 
million. 

In the next set of simulations: 

* As before, we used a broader range of assumptions for 
underrepresentation (0 to 50 percent), adding from 0 to 7.2 million to 
the labor force. 

* We then assumed that 0 to 50 percent of the added unauthorized 
residents were unemployed, or that 50 to 100 percent were employed, and 
applied these employed percentages to the 0 to 7.2 million added to the 
labor force. 

* These simulations added from 0 to 7.2 million employed persons to the 
BLS figure, for simulated totals ranging from 145to 152 million. 

Most simulations produced simulated numbers of employed persons that 
were above the margin of error of BLS’s reported national figure. 
[Footnote 26] See figure 2. 

Figure 2: Simulated U.S. Employed Persons after Including Unauthorized 
Residents Assumed Not Represented in CPS Data, March 2008: 

[Refer to PDF for image: multiple line graph] 

Assumed underrepresentation of unauthorized residents (percent) plotted 
versus simulated U.S. employed persons (in millions). 

UUR: Assumed unemployment rate among unauthorized residents. 

Lines representing the following are depicted: 

UUR: 0%; 
UUR: 6.5%; 
UUR: 8.5%; 
UUR: 6.5%; 
UUR: 15%; 
UUR: 20%; 
UUR: 30%; 
UUR: 40%; 
UUR: 50%. 

Additionally, the following are depicted: 

Simulations resulting from assumed underrepresentation and unemployment 
rates among the unauthorized that are within the range provided by 
experts (first set of simulations); 

Simulations resulting from assumptions more extreme than experts 
suggested (second set of assumptions); 

The 90% confidence interval around the BLS estimate of 145.1 million: 
144.6 to 145.6 million. 

Source: GAO simulations based on reported BLS statistics and expert 
assumptions. 

[End of figure] 

Effects of Underrepresentation May Be Similar under More Recent 
Economic Conditions: 

After March 2008, the month of our detailed simulations, the U.S. 
unemployment rate rose, reaching 9.5 percent in June 2009, and two 
experts told us that the rate for unauthorized residents may have risen 
more steeply over this time period, perhaps to 11 or even 15 percent. 

While more recent estimates of size of the overall unauthorized 
population are not available, one expert believes this population may 
have decreased due to declining economic conditions, and another thinks 
its size has stayed roughly the same.[Footnote 27] 

We performed simulations of the effects of underrepresentation on June 
2009 labor force statistics for which we assumed no change since March 
2008 in the size of the unauthorized labor force represented in CPS 
data. 

* As a check, we also performed sensitivity analyses in which we 
increased and decreased the size of the unauthorized labor force 
(represented in the CPS data) by up to 30 percent. The sensitivity 
analyses indicated that changes of this magnitude would not change our 
simulation findings. 

Our simulations for June 2009 were approximate, or rough, in that we 
used the broad assumptions of 0 to 50 percent underrepresentation of 
unauthorized residents and 0 to 50 percent unemployment for this group 
without identifying a narrower range of potentially more likely values 
based on expert opinion. 

We found patterns in our June 2009 simulations similar to our 2008 
results. 

Barring extreme assumptions or other differences we were not able to 
measure, underrepresentation may have had similar effects on 2008 and 
2009 statistics.[Footnote 28] 

Effects of Underrepresentation May Be Larger in Some States: 

We did not simulate state-level labor force statistics, but our 
research indicates the following: 

* Unauthorized residents probably represent a higher percentage of the 
population in certain states. 

* In some cases (e.g., Arizona, California, and Nevada), the percentage 
may be about twice as high as the national figure. 

* Additionally, some experts believe that the underrepresentation of 
unauthorized residents is greater in certain states. 

* Effects on state labor force statistics could be greater in states 
with more unauthorized residents or higher underrepresentation, 
especially if, for example, the unemployment rate of unauthorized 
residents who are not represented in CPS data differs from estimates 
for the rest of the state population. 

[End of Finding 3] 

Implications: 

Although the overall unemployment rate appears not to be sensitive to 
possible underrepresentation of unauthorized residents, other labor 
force statistics (such as the size of the labor force) could be. 
Therefore: 

* Fuller representation of unauthorized residents in the CPS and other 
surveys would improve the accuracy of some labor force statistics. 

* Fuller representation might be especially important in states with 
high concentrations of unauthorized residents, where 
underrepresentation may produce pronounced effects on labor force 
statistics.[Footnote 29] 

In general, better information about the unauthorized population would 
help policymakers implement laws and evaluate policies that apply 
differently to foreign-born persons who reside here legally and to 
unauthorized residents. We are continuing to monitor federal efforts 
related to possible approaches for evaluating representation of the 
foreign-born population in data collected. 

[End of section] 

Appendix A: Methods for Simulating National Labor Force Statistics: 

For our March 2008 detailed simulations, we used publicly available CPS 
estimates of the U.S. unemployment rate, size of the labor force, and 
number of employed persons aged 16 and older.[Footnote 30] 

* March 2008 was the most recent time for which we were able to obtain 
published estimates of the number of unauthorized residents in the 
labor force (Passel and Cohn, 2009(a)). 

We also used Jeffrey S. Passel’s unpublished estimate that 7.2 million 
unauthorized residents aged 16 and older in the U.S. labor force were 
represented in the March 2008 CPS.[Footnote 31] 

* Passel made this estimate before he adjusted CPS data for 
underrepresentation. 

* While we did not vary this figure in our simulations, we performed 
sensitivity analyses and determined that our findings would not change 
if we increased or decreased Passel’s figure by up to 20 percent. 

We initially simulated national labor force statistics using a range of 
assumptions based on the opinions of one government and three private 
sector experts, who suggested that: 

* the underrepresentation of unauthorized residents in CPS data might 
be between 10 and 15 percent and; 

* unemployment among unauthorized residents, as of March 2008, might 
range from 6.5 to 8.5 percent.[Footnote 32] 

Using these values, we simulated revised versions of: 

* the national unemployment rate, 

* the number of persons in the U.S. labor force, and, 

* the number employed in the United States. 

To test whether more extreme assumptions would change our results, we 
also calculated additional simulated values for March 2008 using values 
of 0 to 50 percent for both underrepresentation and the unemployment 
rate of unauthorized residents. 

We examined whether the resulting simulated labor force statistics fell 
within the margins of error for reported BLS statistics.[Footnote 33] 

Table 1: March 2008 Labor Force Statistics and Margins of Error 
(Civilian Noninstitutional Population Aged 16 and Older): 

Statistic: Unemployment Rate; 
CPS Estimate: 5.24%; 
Lower 90% Confidence Interval: 5.08%; 
Upper 90% Confidence Interval: 5.41%. 

Statistic: Labor Force; 
CPS Estimate: 153,135,000; 
Lower 90% Confidence Interval: 152,629,000; 
Upper 90% Confidence Interval: 153,641,000. 

Statistic: Employed Persons; 
CPS Estimate: 145,108,000; 
Lower 90% Confidence Interval: 144,567,000; 
Upper 90% Confidence Interval: 145,649,000. 

Source: Calculated by GAO using published CPS data. 

[End of table] 

To determine whether our results would likely be similar under more 
recent economic conditions, we updated our analysis using June 2009 CPS 
data on the unemployment rate (8.7%), labor force (155 million), and 
employed persons(141 million). 

We assumed that the number of unauthorized residents in the labor force 
represented in the CPS was unchanged from March 2008 (7.2 million). 

We used the same broad range of assumptions for underrepresentation (0 
to 50 percent) and unemployment rate (0 to 50 percent) and compared the 
results with the 90 percent confidence interval for the June 2009 
statistics. 

While this study focused on possible underrepresentation of one 
specific group, unauthorized residents, it is also possible that CPS 
data under-or overrepresent the general population by a small amount 
(based on factors such as net undercounting and overcounting in the 
2000 census).[Footnote 34] 

Our simulations did not adjust for any overall under or 
overrepresentation in the CPS, but we tested our results to determine 
whether they would be affected by an overall error of less than one-
half of 1 percent in either direction. 

This sensitivity analysis showed that adjusting for a small overall 
error would not substantially change our results. 

[End of Appendix A] 

Appendix B: 

Data Sources for Labor Force Statistics We Examined: 

The Current Population Survey (CPS),which is the primary data source 
for the statistics examined here, is a nationally representative 
monthly survey of approximately 60,000 households administered by the 
Census Bureau. 

* BLS uses CPS data to calculate statistics such as unemployment rates, 
the size of the labor force, and the number of employed persons. 

The decennial census (which is not a survey, but a comprehensive 
household-based count) and the American Community Survey (ACS)are not 
used directly in calculating official labor force statistics. 

* The Census Bureau uses them in calculating the national population 
estimates, which are used to weight CPS data.[Footnote 35] 

None of these data sources collects information on legal status. 

Data Sources for Labor Force Statistics We Did Not Examine: 

The Current Employment Statistics survey is a monthly survey of 
approximately 400,000 employers. The results are used in calculating 
the number of nonfarm jobs, known as payroll employment. Those surveyed 
are drawn from businesses making unemployment insurance payments. 

* This survey does not collect data on legal status. 

Unemployment Insurance claims data are collected by states and used by 
BLS as supplementary information in calculating some state and local 
labor force statistics. 

* Unemployment insurance claims data may exclude most unauthorized 
residents, who are ineligible to collect unemployment benefits. 

[End of Appendix B] 

Appendix C: 

CPS Data, National Population Estimates, and Underrepresentation: 

The Census Bureau weights CPS data to make them consistent with 
national population estimates. 

* The Census Bureau produces the national population estimates (1)using 
counts from the decennial census and (2) updating these counts based on 
birth and death records as well as data on net migration from the ACS, 
a large survey that it conducts every year.[Footnote 36] 

* According to the Census Bureau, the purpose of weighting CPS data is 
to improve accuracy and help minimize underrepresentation.[Footnote 37] 

CPS data distributions—such as Hispanics in defined age groups—thus 
match the distributions from the national population estimates. 
However, problems affecting the national population estimates 
logically “carry over”to the weighted CPS data, including any 
underrepresentation of unauthorized residents. 

National population estimates may underrepresent a group if, for 
example, that group is undercounted in the census or partially composed 
of foreign-born residents who may be underrepresented in the ACS. 

Undercounting in the decennial census is thought to be higher for 
unauthorized residents than for others. 

* Undercounting in the decennial census occurs when a household is 
missed or an individual is omitted from the list of persons residing in 
a covered household.[Footnote 38] 

* To illustrate, if a subpopulation (such as the unauthorized 
population) numbers 12 million and the census fails to count 1.2 
million of them, the undercount would be 1.2 million or 10 percent. 

While the Census Bureau uses census data and the national population 
estimates to improve the accuracy of CPS data, it does not make 
statistical adjustments to correct for undercounting in the census or 
underrepresentation in the national population estimates. 

Regarding the issue of statistical adjustment of the census,[Footnote 
39] a topic of discussion among policymakers over the years, the 
director oft he Census Bureau said in a May 15, 2009, testimony before 
the Senate Committee on Homeland Security and Governmental Affairs: 

“I agree fully…that statistical adjustment of the census is eliminated 
as an option for reapportionment...[or] redistricting. The 2003 
decision of [the then Census Bureau] director…assured that no 
implementation infrastructure for adjustment was put in place for [the] 
2010 [census].” 

Underrepresentation of unauthorized residents in CPS data may be caused 
by factors additional to undercounting in the census, including: 

* possible underrepresentation of unauthorized residents in the ACS or 
miscategorization of some members of key groups (e.g., Hispanics)—which 
would affect how CPS data are weighted, and; 

* nonresponse in the CPS, which occurs, for example, when no one in a 
selected household is available to take the survey.[Footnote 40] 

[End of Appendix C] 

Appendix D: Grouped Answers Approach to Estimating the Unauthorized 
Resident Population: 

The "grouped answers" approach is a questionnaire method that might be 
used in private-sector, personal-interview surveys to obtain 
information about the overall population of unauthorized residents, 
while ensuring the anonymity of individual survey responses.[Footnote 
41] 

This personal-interview approach: 

* shows each respondent a flash card with answers grouped in three 
boxes, 

* asks the respondent to "just pick the box" that contains his or her 
current legal status (along with other statuses), and thus, 

* avoids questioning respondents about whether they are unauthorized. 

Estimation for an overall population is possible because the grouped 
answers approach (two card version): 

* divides respondents into two subsamples; 

* shows subsamples 1 and 2 slightly different cards (see the following 
slide);[Footnote 42] and; 

* compares the percentages that picked key boxes, as indicated by the 
arrow on the following slide. 

When the grouped answers approach is applied in a personal interview 
survey conducted by a university or private-sector organization, it 
appears to be acceptable to foreign-born respondents and, with certain 
provisions, to immigrant advocates. 

Figure 3: Legal Status Cards 1 and 2 Compared: 

[Refer to PDF for image: illustration] 

Subsample 1, Card 1: 

A: United States Citizen: 

Student, work, business, or tourist visa: I am not in violation of 
admission period limits or work restrictions. 

B: Legal permanent resident: with a valid and official green card 
issued to me by the U.S. Government. 

Currently "undocumented:" Right now I do not have a currently valid, 
legal U.S. immigration status. 

Refugee or asylee: (approved, not applicant). 

C: TPS*, parolee, or some other category (*: Temporary Protected 
Status). 

Subsample 2, Card 2: 

A: Legal permanent resident: with a valid and official green card 
issued to me by the U.S. Government. 

Refugee or asylee: (approved, not applicant). 

B: United States Citizen: 

Student, work, business, or tourist visa: I am not in violation of 
admission period limits or work restrictions. 

Currently "undocumented:" Right now I do not have a currently valid, 
legal U.S. immigration status. 

C: TPS*, parolee, or some other category; Not in Box A or Box B (*: 
Temporary Protected Status). 

Source: GAO; Corel Draw (flag and suitcase); DHS (resident alien 
cards). (The actual size of each card is 8 1/2 by 11".) 

[End of figure] 

[End of Appendix D] 

Appendix E: List of Experts Consulted: 

Experts who suggested parameters for our simulations:[Footnote 43] 

* Steven A. Camarota, Center for Immigration Studies; 

* Randy Capps, Migration Policy Institute; 

* Michael D. Hoefer, U.S. Department of Homeland Security; 

* Jeffrey S. Passel, Pew Hispanic Center. 

Experts who reviewed our work: 

* Douglas S. Massey, Office of Population Research, Princeton 
University; 

* Fritz J. Scheuren, National Opinion Research Center, University of 
Chicago; 

* Robert E. Warren, former Director, U.S. Immigration Statistics 
Division, Immigration and Naturalization Service. 

[End of Appendix E] 

Selected References: 

Camarota, Steven A. and Karen Jensenius. “A Shifting Tide: Recent 
Trends in the Illegal Immigrant Population.”Washington, D.C.: Center 
for Immigration Studies, July 2009. 

Groves, Robert M. et al. Survey Methodology, 2ndedition. Hoboken, N.J.: 
John Wiley & Sons, Inc, 2009. 

Hoefer, Michael, Nancy Rytina, and Bryan C. Baker. “Estimates of the 
Unauthorized Immigrant Population Residing in the United States: 
January 2008.”Office of Immigration Statistics, U.S. Department of 
Homeland Security. 

Marcelli, Enrico A. and Paul M. Ong. “Estimating the Sources of the 
2000 Census Undercount among Foreign-born Mexicans in Los Angeles 
County.”Prepared for the Annual Population Association of America 
Meetings. Atlanta, Georgia: May 10, 2002. 

Mulry, Mary. “Summary of Accuracy and Coverage Evaluation for Census 
2000.”Research Report Series, Statistics #2006-3. Washington, D.C.: 
U.S. Census Bureau, Feb. 28, 2006. 

Passel, Jeffrey S. and D’Vera Cohn. “A Portrait of Unauthorized 
Immigrants in the United States.”Washington, D.C.: Pew Hispanic Center, 
April 14, 2009 (a). 

Passel, Jeffrey S. and D’Vera Cohn, “Mexican Immigrants: How Many Come? 
How Many Leave?” Washington, D.C.: Pew Hispanic Center, July 22, 2009 
(b). 

Slater, Courtenay M. “The Impact of Census Undercoverage on Federal 
Programs.”Conference on Census Undercount: Proceedings of the 1980 
Conference. Washington, D.C.: U.S. Bureau of the Census, July 1980. 
Pages 107-111. 

U.S. Census Bureau. Current Population Survey: Design and Methodology. 
Technical Paper 66. Washington, D.C.: October 2006. 

[End of Selected References] 

GAO Contacts and Staff Acknowledgments: 

GAO Contacts: 

Nancy Kingsbury, Managing Director, Applied Research and Methods, (202) 
512-2700, kingsburyn@gao.gov. 

Cornelia M. Ashby, Director, Education, Workforce, and Income Security, 
(202) 512-7215, ashbyc@gao.gov. 

Staff Acknowledgments: 

Gale C. Harris and Judith A. Droitcour, Assistant Directors; Brittni 
Milam, Analyst-in-Charge; and Susan Aschoff, Eric M. Larson, Rhiannon 
Patterson, Meredith Trauner, and Monique B. Williams made significant 
contributions to this report. 

Related GAO Reports: 

Estimating the Undocumented Population: A “Grouped Answers”Approach to 
Surveying Foreign-Born Respondents. [hyperlink, 
http://www.gao.gov/products/GAO-06-775]. Washington, D.C.: September 
29, 2006. 

Illegal Alien Schoolchildren: Issues in Estimating State-by-State 
Costs. [hyperlink, http://www.gao.gov/products/GAO-04-733]. Washington, 
D.C.: June 21, 2004. 

Overstay Tracking: A Key Component of Homeland Security and a Layered 
Defense. [hyperlink, http://www.gao.gov/products/GAO-04-82]. 
Washington, D.C.: May 21, 2004. 

Survey Methodology: An Innovative Technique for Estimating Sensitive 
Survey Items. [hyperlink, http://www.gao.gov/products/GAO/GGD-00-30]. 
Washington, D.C.: November 1999. 

Immigration Statistics: Information Gaps, Quality Issues Limit Utility 
of Federal Data to Policymakers. [hyperlink, 
http://www.gao.gov/products/GAO/GGD-98-164]. Washington, D.C.: July 31, 
1998. 

[End of Appendix I, Briefing slides] 

Appendix II: Comments from the Department of Commerce: 

United States Department Of Commerce: 
The Under Secretary for Economic Affairs: 
Washington, D.C. 20230: 

November 10, 2000: 

Cornelia M. Ashby: 
Director: 
Education, Workforce, and Income Security Issues: 
United States Government Accountability Office: 
Washington, DC 20548: 

Dear Ms. Ashby: 

The U.S. Department of Commerce appreciates the opportunity to comment 
on the United States Government Accountability Office's draft report 
entitled: Unauthorized Residents and U.S. Labor Force Statistics: 
Illustrative Simulations of the Likely Effects of Undercounting (GAO-10-
99). The Department's comments on this report are enclosed. 

Sincerely, 

Signed by: 

Rebecca M. Blank: 

Enclosure: 

[End of letter] 

U.S. Department of Commerce: 

Comments on the United States Government Accountability Office Draft 
Report Entitled "Unauthorized Residents and U.S. Labor Force 
Statistics: Illustrative Simulations of the Likely Effects of 
Undercounting" (GAO-10-99): 
November 2009: 

* The Population Division provided comments on an earlier draft of this 
report on August 26, 2009. In general, the GAO incorporated 
corrections, edits, and additional text that addressed many of these 
comments. However, the GAO did not fully address three comments: 

- First, the presentation still lacks a basic explanation of how 
surveys in general adjust for statistical representation and how the 
intercensal population estimates produced by the Census Bureau are used 
as a final weight to implicitly account for and partially improve 
coverage. An additional slide with relevant citations would be 
appropriate. 
- Second, the benefits of the "two-card method" remain overstated. 
Including Appendix D implies the "two-card method" is a way to estimate 
the foreign-born population and, by extension, will provide a measure 
of coverage, which it will not. The method only addresses possible 
response error, not coverage. The Census Bureau cannot use the two-card 
method in the Current Population Survey (CPS) for two main reasons. 
First, many of the surveys administered by the Census Bureau use 
multiple modes to collect data, including the CPS (phone and in person) 
and the American Community Survey (mail, phone, and in person), not 
just face-to-face. Second, even if it were possible, incorporating the 
two-card method in a government setting would likely generate survey 
participation issues. Including a detailed description of the two-card 
method in this report implies it can and should be used by the CPS, 
which it cannot. Given its lack of applicability to the CPS, we 
recommend removing Appendix D. 
- Third, the presentation does not provide a thorough enough 
explanation of the methodology used for the simulation, making it hard 
to comment on the conclusions. While Appendix A discusses the 
methodology used, it is quite "high level." A detailed discussion of 
the methodology is even more important now that this presentation is to 
be published by the GAO as a stand-alone report. An additional section 
for the report, detailing the methodology, would be appropriate. 

* Some additional comments:
- The letter to Representative Jim McDermott mentions that information 
was obtained from "experts" about 1) under-representation of the 
unauthorized immigrants in the CPS and 2) the unemployment rates of 
unauthorized immigrants. However, the letter does not identify who 
these experts are, what fields of expertise they represent, what 
institutions they are associated with, and why they were chosen. 
Rather, that information is provided in an appendix of the 
presentation. Given that so many of the assumptions were determined by 
the analyses (and, often, "educated guesses") of these experts, it 
would be appropriate to talk about them in the letter. Also, given that 
the focus of this report is on statistical representation and coverage, 
it would have been better to have included experts in survey 
methodology, sampling, and weighting, as well as experts on 
international migration. 
- Including 0 to 50 percent as a range for rates of under-
representation of the unauthorized immigrants in the CPS in a published 
GAO report gives the extreme values of this range plausibility not 
supported by any research. It invites taking these results out of 
context and using them as legitimate measures of undercoverage. We 
recommend reducing this range. 
- We would be happy to work with GAO to identify additional survey data 
on the foreign born that is required to implement laws and evaluate 
policies, as referred to on slide 29. 
- The letter referred to on slide 13 was sent to GAO on February 26, 
2008 and explained why the Census Bureau does not estimate coverage of 
the foreign born in decennial Censuses and why a census post-
enumeration survey is not a vehicle that can be used to estimate 
coverage of the foreign born. 

[End of section] 

Footnotes: 

[1] After March 2008, the month of our detailed simulations, the U.S. 
unemployment rate rose, reaching 9.5 percent in June 2009. Declining 
economic conditions may affect the number of unauthorized residents in 
the United States, although recent estimates of the size of this 
population are not available. With this in mind, we performed 
sensitivity analyses that included decreasing the size of the 
unauthorized labor force (represented in the CPS data) by up to 30 
percent. The sensitivity analyses indicated that a decrease of this 
magnitude would not change our simulation findings. 

[2] The grouped answers method is described in slides 41 to 43 in 
appendix I of this report. 

[3] GAO, Estimating the Undocumented Population: A "Grouped Answers" 
Approach to Surveying Foreign-Born Respondents, [hyperlink, 
http://www.gao.gov/products/GAO-06-775] (Washington, D.C.: Sept. 29, 
2006). 

[4] For example, data on a respondent's legal status (such as legal 
permanent resident or refugee) are not collected by government- 
conducted general-purpose surveys such as the CPS--but potentially 
could be collected by a private-sector survey using an approach such as 
the grouped answers method. Such data are relevant to implementing 
various laws and evaluating various immigration policies. 

[5] Additionally, because the subtraction method estimates the 
unauthorized population by subtracting estimates of immigrants in 
specific legal status categories from an overall estimate of the total 
foreign-born population, researchers using this approach would be 
helped by grouped answers estimates of foreign-born persons in specific 
legal status categories (such as the number of legal permanent 
residents). The grouped answers method could provide direct estimates 
of these categories. 

[6] GAO, Immigration Statistics: Information Gaps, Quality Issues Limit 
Utility of Federal Data to Policymakers, [hyperlink, 
http://www.gao.gov/products/GAO/GGD-98-164] (Washington, D.C.: July 31, 
1998). 

[7] Perhaps most notably, in addition to the letter mentioned by the 
Census Bureau, we outlined various possible approaches that Census 
might take in estimating undercoverage of foreign-born persons, 
including a new record linkage (data-matching) approach. We shared this 
information with Census Bureau staff at a meeting and subsequently 
discussed this and other possible methods with them during 2008. We 
also made additional suggestions in a February 2009 letter to the 
Census Bureau. 

[8] Foreign-born residents include naturalized citizens, other 
authorized persons, and unauthorized residents. 

[9] See, for example, Jeffrey S. Passel and D’Vera Cohn, A Portrait of 
Unauthorized Immigrants in the United States, (Washington, D.C.: Pew 
Hispanic Center, Apr. 14, 2009). 

[10] We selected studies on the basis of their relevance to our work. 

[11] March 2008 is the most recent time period for which an expert has 
published estimates of not only the unauthorized population, but also 
the number of unauthorized workers. 

[12] We determined that the data we used were sufficiently reliable for 
the purposes of our simulations. 

[13] BLS publishes two estimates of employment: (1) the number of 
persons employed, measured by the CPS, and (2) the number of jobs, 
measured by the Current Employment Statistics survey. We focus here on 
the CPS estimate only. 

[14] The Pew Hispanic Center estimates this using data from the CPS, 
while DHS uses data from another national survey, the American 
Community Survey. 

[15] This number, estimated using primarily DHS data, includes persons 
such as legal permanent residents and refugees. 

[16] This survey approach groups legal statuses in boxes on a flash 
card. Each foreign-born respondent chooses the box containing his or 
her legal status, along with other statuses. No one specifically 
chooses the unauthorized status. See appendix D for more details. 

[17] These experts were Steven A. Camarota, Michael D. Hoefer, and 
Jeffrey S. Passel. See appendix E. 

[18] We use the term "undercount" when referring to census counts of 
the population and the term "underrepresentation" when referring to 
statistics based on data from a sample survey, such as CPS. 

[19] See Enrico A. Marcelli and Paul M. Ong, Estimating the Sources of 
the 2000 Census Undercount among Foreign-born Mexicans in Los Angeles 
County (prepared for the Annual Population Association of America 
meetings, Atlanta, Georgia, May 10, 2002). 

[20] This estimate includes an upward adjustment for assumed 
underrepresentation. Passel’s figure before adjusting for 
underrepresentation was 7.2 million. 

[21] These experts were Steven A. Camarota, Randy Capps, and Jeffrey S. 
Passel. The figures they provided are marked by varying degrees of 
uncertainty because they rely on assumptions and inferences, such as 
using CPS data on all immigrants from Mexico and Central America as a 
proxy population for unauthorized residents. 

[22] In this briefing, we use labor force statistics that are not 
seasonally adjusted. 

[23] Our simulations cannot be generalized to other time frames without 
further analysis. 

[24] BLS' published estimate is 5.2 percent, but we are providing a 
more precise figure for purposes of comparison with simulated values 
and margins of error. The margins of error used in this briefing 
correspond to 90 percent confidence intervals around BLS’s reported 
statistics. 

[25] The simulated labor force size fell within the margin of error 
only in simulations with assumed underrepresentation of less than 7 
percent. We provide this information as an observation; we have not 
conducted formal tests of significance. 

[26] The simulated number of employed persons fell within the margin of 
error of the reported statistic for a fairly narrow range of 
assumptions, including (1) underrepresentation of less than 8 percent 
or (2) underrepresentation of less than 9 percent and an unemployment 
rate of more than 13 percent. This is provided as an observation; we 
did not conduct formal tests of significance. 

[27] See Jeffrey S. Passel and D’Vera Cohn, Mexican Immigrants: How 
Many Come? How Many Leave? (Washington, D.C.: Pew Hispanic Center, July 
22, 2009), and Steven A. Camarota and Karen Jensenius, A Shifting Tide: 
Recent Trends in the Illegal Immigrant Population (Washington, D.C.: 
Center for Immigration Studies, July 2009). 

[28] Our findings for both years are consistent with much earlier 
analyses that adjusted for census undercoverage for the overall 
population. See Courtenay M. Slater, “The Impact of Census 
Undercoverage on Federal Programs,”in Conference on Census Undercount: 
Proceedings of the 1980 Conference (Washington, D.C.: U.S. Bureau of 
the Census, 1980), 107-111). 

[29] The same may be true for certain occupational categories, 
industrial sectors, or local areas, but examinations of the effects for 
these groups are beyond the scope of this briefing. 

[30] CPS data, which we and Passel used as starting points in our 
analyses, are weighted to reflect the Census Bureau’s national 
population estimates and to correct for survey nonresponse. 

[31] Passel estimated the March 2008 labor force participation rate of 
represented unauthorized residents to be 76.5 percent. 

[32] For example, if 7.2 million of the unauthorized resident labor 
force are represented in the CPS but 10 percent of this population are 
not represented, the total numbers 8 million, with 800,000 not 
represented. If the 800,000 have an unemployment rate of 8 percent, 
64,000 are unemployed. To recalculate the existing unemployment rate, 
we add 64,000 to its numerator (number unemployed) and 800,000 to its 
denominator (total in labor force). 

[33] The margins of error correspond to 90 percent confidence intervals 
around the reported statistics. We use labor force statistics that are 
not seasonally adjusted. 

[34] For more information on possible overall counting error in the 
2000 census, see Mary Mulry, “Summary of Accuracy and Coverage 
Evaluation for Census 2000,” Research Report Series, Statistics #2006-3 
(Washington, D.C.: U.S. Census Bureau, Feb. 28, 2006). 

[35] See appendix C. 

[36] For more information on national population estimates, see U.S. 
Bureau of the Census, Methodology for the United States Resident 
Population Estimates by Age, Sex, Race, and Hispanic Origin (Vintage 
2008): April 1, 2000, to July 1, 2008, [hyperlink, 
http://www.census.gov/popest/topics/methodology/2008-nat-meth.pdf]. 

[37] U.S. Census Bureau, Current Population Survey: Design and 
Methodology, Technical Paper 66, October 2006. See chapter 10, 
especially the section on national coverage adjustment. 

[38] For a discussion of undercoverage and related issues, see Groves 
et al., Survey Methodology, 2nd Edition (Hoboken, N.J.: John Wiley & 
Sons, Inc., 2009). 

[39] The U.S. Supreme Court ruled in 1999 that the use of statistical 
sampling was statutorily prohibited for purposes of apportioning seats 
in the House of Representatives, but did not specifically address its 
use for other purposes. Department of Commerce v. U.S. House of 
Representatives, 525 U.S. 316 (1999). 

[40] To correct CPS data for nonresponse, the Census Bureau increases 
the weights of the answers of respondents who might be similar to 
nonrespondents, but this may not fully address nonresponse on the part 
of unauthorized residents. 

[41] See GAO, Estimating the Undocumented Population: A “Grouped 
Answers” Approach to Surveying Foreign-Born Respondents, [hyperlink, 
http://www.gao.gov/products/GAO-06-775] (Washington, D.C.: Sept. 29, 
2006). 

[42] No one is asked to respond to both cards. This method (1) has thus 
far not been developed for mail surveys and (2) is not appropriate for 
telephone interviewing because respondents must view the flash cards. 

[43] We chose these four experts on the basis of their recent 
experience in estimating the unauthorized population. Camarota, Hoefer, 
and Passel suggested possible underrepresentation levels, and Camarota, 
Capps, and Passel commented on unemployment levels. 

[End of section] 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the performance 
and accountability of the federal government for the American people. 
GAO examines the use of public funds; evaluates federal programs and 
policies; and provides analyses, recommendations, and other assistance 
to help Congress make informed oversight, policy, and funding 
decisions. GAO's commitment to good government is reflected in its core 
values of accountability, integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's Web site [hyperlink, http://www.gao.gov]. Each 
weekday, GAO posts newly released reports, testimony, and 
correspondence on its Web site. To have GAO e-mail you a list of newly 
posted products every afternoon, go to [hyperlink, http://www.gao.gov] 
and select "E-mail Updates." 

Order by Phone: 

The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site, 
[hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional 
information. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]: 
E-mail: fraudnet@gao.gov: 
Automated answering system: (800) 424-5454 or (202) 512-7470: 

Congressional Relations: 

Ralph Dawn, Managing Director, dawnr@gao.gov: 
(202) 512-4400: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7125: 
Washington, D.C. 20548: 

Public Affairs: 

Chuck Young, Managing Director, youngc1@gao.gov: 
(202) 512-4800: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7149: 
Washington, D.C. 20548: