Case Studies
Only a few recent studies take a comprehensive inventory of the fiscal effects of immigration, primarily for local areas and states. Six of them were selected for review at the workshop; see Table 1.1
Each study is reviewed in this section, with a focus on issues of the study's funding source and purpose, sources of data, population estimates, methods used to create the estimates, cost and revenue items, results, and the strengths and limitations of the study. One study considered the United States as a whole, two looked at specific counties, and three assessed the fiscal effects of illegal immigration on specific states.
LOS ANGELES COUNTY
The Los Angeles County Internal Services Department (1992) prepared a study for the County Board of Supervisors assessing the fiscal impact of immigration on Los Angeles County and for school districts within the county. The study estimated costs and revenues for four groups: recent legal immigrants, amnesty immigrants, undocumented immigrants, and citizen children of undocumented immigrants.2
TABLE 1 Case Studies of the Fiscal Impact of Illegal Immigration
Author, Date |
Title |
Sponsor |
Geographic Area |
Period of Analysis |
Los Angeles County Internal Services Department, 1992 |
Impact of Undocumented Persons and other Immigrants on Costs, Revenues and Services in Los Angeles County |
Los Angeles County Board of Supervisors |
Los Angeles County |
Fiscal 1991–1992 |
Huddle, 1993 |
The Costs of Immigration |
Carrying Capacity Network |
United States |
Fiscal 1991 and 1993 (covers immigrants arriving since 1970) |
Huddle, 1994a |
The Net Costs of Immigration to Texas |
Carrying Capacity Network |
Texas |
Fiscal 1992 (covers immigrants arriving since 1970) |
Parker and Rea, 1993 |
Illegal Immigration in San Diego County: An Analysis of Costs and Revenues |
California State Senate Special Committee on Border Issues |
San Diego County |
Fiscal 1991–1992 |
Romero, Chang, and Parker, 1994 |
Shifting the Cost of a Failed Federal Policy |
Governor's Office on Planning and Research |
California |
Fiscal 1994–1995 |
Clark et al., 1994 |
Fiscal Impacts of Undocumented Aliens: Selected Estimates for Seven States |
U.S. Office of Management and Budget, U.S. Department of Justice, Urban Institute's Program for Research on Immigration Policy (funded by the Ford Foundation) and Immigrant Policy Program (funded by the Andrew W. Mellon Foundation) |
Arizona, California, Florida, Illinois, New Jersey, New York, and Texas |
1992 |
Using data from the 1980 census, the Current Population Survey (CPS), the Immigration and Naturalization Service (INS), and information from the California Department of Finance, the study estimated an undocumented immigrant population of 700,000 in Los Angeles County in January 1992. The estimate was derived by extrapolation of the 1980 Census Bureau estimate of the number of enumerated undocumented immigrants with amnesty data and assumptions for net undocumented immigration.
Cost estimates for fiscal 1991–1992 covered criminal justice programs and services, health and mental health services, social services, and elementary and secondary education. Cost data were based on administrative records and case records from county departments and a survey of the amnesty population by the California State Health Services Department and Mental Health Department. Costs for quasi-public goods and some other services were a pro rata share of total costs. Other costs were per client costs based on case data. Cost estimates were based on the assumption that immigrants behave like the rest of the population, after adjusting for income. Available information on illegal immigrants, such as those who applied for legalization in 1986, suggest that they are likely to be of Mexican origin, male, in their early labor force years, and less educated than the average U.S. resident.
Data on revenue sources included Federal Insurance Contributions Act (FICA), federal income, unemployment insurance, and excise taxes; state income, excise, gas, and sales taxes; lottery ticket sales; and local sales and property taxes (with the assumption that one-half of the property tax is paid by the tenant). Effective tax rates were based on data from the Internal Revenue Service, the State Board of Equalization, and the California Franchise Tax Board. Data on household structure and income distribution were based on data for aliens legalized under the general 1986 amnesty, as surveyed in the Westat Legalization Population Survey. Taxes paid were estimated using a matrix of age, income, and household characteristics multiplied by the applicable effective tax rate.
Workshop participants noted several strengths of the Los Angeles County study. First, unlike the other fiscal impact studies the workshop examined, the cost estimates took into account revenues from federal and state governments, thus including the offsetting effect of intergovernmental transfers. The study also has a reasonably clear presentation of data sources, methodology and results, and could likely be replicated by other analysts.
Workshop participants also pointed out several weaknesses of the Los Angeles County study. The assumption that the behavior of undocumented immigrants is the same as legal U.S. residents, regarding program use and costs, is questionable. Moreover, in a large geographic area like Los Angeles County, the uneven distribution of the undocumented immigrant population, coupled with geographical variations in cost and spending, casts doubt on cost estimates, particularly those for education. One workshop participant suggested that average per pupil expenditures may underestimate per pupil costs by excluding the cost of bilingual and English-as-a-second-language education. Finally, this study does not include an identifiable cost for capital improvements, and capital costs may not have been included in calculations for those improvements: such costs may be important because very high levels of illegal immigration may result in a need for school districts to construct new schools, for example.
UNITED STATES AND TEXAS
Two studies on the effect of immigration by Huddle (1993, 1994a), a study of the United States and a study of Texas, respectively, were commissioned by the Carrying Capacity Network. The Carrying Capacity Network is a national advocacy group with environmental and population concerns that urges reductions of population growth and a dramatic diminution of U.S. immigration levels. Cost and revenue estimates were presented for legal and illegal immigrants, with separate estimates for amnesty immigrants. Workshop participants focused on the findings regarding illegal immigrants.
Huddle estimates costs for 22 programs (14 programs for illegal immigrants) but concentrated his analysis on the three broad categories of public assistance, criminal justice, and education. Revenue estimates included federal income and excise taxes; state income, sales, and excise taxes; the state lottery (for Texas); and county sales and property taxes. It is not clear if federal taxes were included in the revenues paid by the immigrant groups.
The author used many different data sources. Cost estimates were based on a variety of state and county data; taxes paid were based on data cited from the 1992 Statistical Abstract of the United States , but without table or page references, making it difficult to replicate his estimates. Census Bureau data were used to calculate public assistance participation rates. Primary and secondary education costs were based on a 1990 estimate of participation rates from the Center for Immigration Studies. Fiscal 1991 data are used for total national program costs, inflated to 1992 dollars.
Huddle used an average cost approach extrapolated from national program costs per capita, adjusted for the participation rate of each group of immigrants—legal, amnesty, and illegal. For example, for public assistance costs, he calculated the following:
national cost per participant = program cost per participant
national participation rate = number of participants/national population
immigration participation rate = national participation rate multiplied by (immigrant poverty rate/native-born poverty rate)
number of immigrant participants = immigrant participation rate multiplied by total immigrant population
total cost = number of immigrant participants multiplied by program cost per participant
Taxes paid are estimated similarly, using Los Angeles County revenue data (for both the national and Texas studies) for each tax divided by the total immigrant population with adjustment factors to obtain national estimates.
These two studies use a current labor displacement rate of 25 percent. The costs of displacement, principally public assistance costs, are based on increasing displacement rates and are projected into the future. The estimate of a 25 percent displacement rate is based on Huddle's own small-scale studies in the Houston metropolitan area and the work of Altonji and Card (1991). It is based on the percentage of unemployed native workers Huddle surveyed in his 1982–1983, 1985, and 1989–1990 "microstudies of job displacement" who were still unemployed after some period of time. Critics argue that these figures cannot be construed as measures of displacement by illegal aliens because the studies did not show that the unemployed natives lost their jobs to illegal aliens or were unable to find work because of the presence of illegal aliens in the Houston labor market. In other words, Huddle's microstudies of job displacement assumed a correlation between the employment of illegal aliens and the unemployment of native workers for which there was no evidence. Finally, even if the studies had accurately measured the level of job displacement in Houston in the 1980s, the phenomenon of job displacement is so locally sensitive that the studies' results for Texas cannot be generalized to the nation (see U.S. General Accounting Office, 1995, for a complete discussion).
Although the Huddle studies did attempt to be comprehensive, to measure net costs, and to estimate prospective public sector costs of legal and illegal immigration, workshop participants noted several problems. 3 First, the population estimates—which differed dramatically from all other studies and were 50 to 300 percent higher than available INS estimates—were not explained or compared with other estimates. 4 For example, Huddle estimates 4.8 million undocu-
mented immigrants in the United States in 1992, while INS estimates 2.0 million in 1990. Estimates of the illegal immigrant population are complicated by the legalization of about 3.2 million persons in the 1980s. Workshop participants discussed the problems of estimates of the legal foreign-born population and amnesty immigrants, which would affect the estimates used in various studies, including these.
Second, prospective immigration estimates were weak, especially for illegal immigration. Some workshop participants suggested that there is little basis for projecting illegal immigration flows in the rapidly changing economic environment in Mexico. Third, neither of the studies was grounded in the existing research literature, and the assumptions and methods were sometimes obscure and, as a result, not reproducible. Finally, participants remarked that the analysis of labor displacement did not explain the implicitly assumed, but unusually high, labor force participation rate of immigrants, did not take dynamic adjustment into account, and did not differentiate between transitory and permanent effects.
SAN DIEGO COUNTY
The California Senate Special Committee on Border Issues requested this study in February 1993 to gauge the costs to state and local governments of providing services to undocumented immigrants in San Diego County. Estimates used in the study covered either fiscal 1992–1993 or calendar 1993. Based on a combination of INS amnesty data and Border Patrol apprehensions, as well as numerous assumptions about the inflow and outflow of illegal immigrants and deaths for resident illegal aliens, Parker and Rea (1993:21) estimated that there were about 223,000 undocumented immigrants in San Diego County in June 1993.
Much of the study relied on findings from a survey of 308 self-identified undocumented immigrants. From all indications, survey participants were not a randomly selected group, which is understandable since the survey was used to produce behavioral assumptions and was not intended to accurately represent the total population of undocumented immigrants in San Diego County. The study also used INS data on employer sanctions in 1991–1992;5 state and local data on tax collections and tax rates; interviews with agency personnel and administration records on caseloads; and information from the San Diego Association of Governments (such as population estimates for the county and estimates of undocumented immigrants).
The study provided cost estimates for criminal justice programs, health ser-
vices, public education, and social and other public services. Revenue estimates included state and local sales, income, payroll, excise and gas taxes; the state lottery; and vehicle registration fees. Most of the study's cost estimates were based on average cost per recipient. Labor force participation rates and consumption patterns were derived from the survey noted above. The study also included a multiplier to capture the full effect of additional spending by undocumented immigrants on the local economy. Spending by illegal immigrants is believed to have a multiplier effect because they spend a considerable portion of their earnings in the local area, thus generating additional employment (Borjas, 1994).
Workshop participants noted several strengths of the study, including its effort to measure revenues comprehensively, particularly through the inclusion of the estimated amount of money that is spent locally. The study also adjusted for the estimated amount of income that illegal immigrants remit to their countries of origin. The presentation of costs and revenue estimates was fairly clear. Finally, the authors used the administrative records of local agencies.
Workshop participants, however, also had criticisms of the study. First, the study relied on a small, nonrandom sample for many of its behavioral assumptions: a survey with unknown biases raises concern about the arbitrary nature of estimates for earning multiplier effects. Second, per capita cost figures led the authors to attribute one-third of the costs of the district attorney's office to the prosecution of undocumented immigrants, a figure rejected by the district attorney's office as unrealistic. Finally, on the revenue side, the study assumed that the statutory and economic incidences of the property tax are the same, that is, that tenants pay no property taxes—a questionable assumption when there is little empirical evidence about the housing arrangements of illegal immigrants, including their property taxes.
CALIFORNIA
The Governor's Office of Planning and the California Department of Finance prepared a study in 1994 to support its argument for federal reimbursement of the state costs of providing services to illegal immigrants (see Romero et al., 1994). The study calculated the fiscal impact on state government of federal mandates and other services provided to illegal immigrants.
On the basis of INS information, the study estimated 1.7 million undocumented immigrants in California in September 1994—an extrapolation of the October 1992 INS estimate, with an assumption of continued net undocumented immigration for 2 years. To estimate the number of school-age children, the study used data from the California Department of Finance, which extrapolated the number of undocumented school-age children from the total population on the basis of an Urban Institute estimate of the age structure of the undocumented immigrant population.
This study included estimates for the cost of elementary and secondary edu-
cation, incarceration, health care, and general services. Education cost estimates are based on state average per pupil expenditures and school participation rates developed by the Urban Institute.6 Data on incarceration of adults and juveniles were from the California Department of Corrections. Health costs are estimated using the state's ordinary least squares regression model developed to calculate Omnibus Budget Reconciliation Act of 1986 costs. Other state costs are pro rata shares of total operating costs.
Revenue estimates included state income, sales, excise, and gas taxes; vehicle fees; lottery ticket sales, and a portion of local property taxes.7 Combining these estimates, the study calculates a per capita state tax rate.
In workshop discussion, participants noted a strong advocacy tone in this study. This study included health care cost estimates for services—prenatal care and long-term care, for instance—that are not federally mandated. Finally, the study did not include data on the effective tax rates, and the information on tax estimates was not current for the period of the study.
SEVEN STATES
In 1994 the U.S. Office of Management and Budget (OMB) and the U.S. Department of Justice jointly asked the Urban Institute to estimate the fiscal impact of undocumented immigration in seven states—Arizona, California, Florida, Illinois, New Jersey, New York, and Texas. Those states were selected because they had the largest number of undocumented aliens, according to INS estimates for October 1992. OMB sponsored the study because several states were filing lawsuits requesting federal reimbursement for local state services for illegal immigrants.
Using the INS data, the Urban Institute estimated that there were 2.9 million undocumented immigrants in the seven states in October 1992—accounting for 86 percent of the total undocumented population in the United States (Clark et al., 1994). The Urban Institute study also examined a limited number of cost and revenue items. Three costs were estimated: incarceration of undocumented immigrants in state prisons, elementary and secondary education for undocumented immigrant children, and emergency medical services provided to un-
documented immigrants as mandated under Medicaid. Revenue estimates were limited to state income and sales taxes and state and local property taxes. The limited scope of the study precluded estimating net costs.
Most data were drawn from federal agencies, thus ensuring uniform data for each state. The state administrative record data differed in quality of record keeping, certification of eligibility rules, and specific information collected. Workshop participants noted that state comparisons would be more consistent if they relied on federal data using similar definitions. Prison costs were obtained from the Census of State Prisons. Education costs were drawn from the National Center for Educational Statistics. Emergency medical care cost "benchmarks" (see below) were based on costs for eligible legalized immigrants reported to the federal government, as required for states to receive State Legalization Impact Assistance Grants. Revenue estimates were derived from INS data, the 1990 census, and a 1992 study of Los Angeles County. Cost and revenue estimates were pro rata shares that were sometimes adjusted for state differences. Overall, the estimates were aggregates and rest on the assumption that undocumented immigrants behave like the rest of the population, except in terms of paying taxes.
The participants commented on three noteworthy strengths of the study. First, it used a consistent methodology across the seven states, which allowed comparisons among states and across categories. Second, the method was clearly explicated; workshop participants believed that it would be possible to replicate the study's results. Third, the study compared its estimates with the states' own estimates and explained any significant differences between them.
Workshop participants also noted three limitations of the study. First, because of the small number of cost items, overall cost estimates for public services could not be produced. Second, the reported cost estimates rely on specific estimates for the number of illegal immigrants, by age and sex. Workshop participants noted that it would be helpful to see studies that use a range of estimates for illegal immigrants, yielding a range of cost estimates. Third, the study did not exploit all the available state data. The variation of costs within states—for example, differences in per pupil spending across school districts—was not examined.