[Population Index on the Web] [Home] [Browse] [Search]

Copyright ©1995, Office of Population Research, Princeton University
Population Index 61(3):353-6. Fall 1995.

A Comment on Rothman and Espenshade

Julian L. Simon

In their useful 1992 Population Index article titled "Fiscal Impacts of Immigration to the United States", Eric S. Rothman and Thomas J. Espenshade made some incorrect statements about my work on the tax-and-transfer-payment situation of immigrants. These statements may mislead the reader and wrongly give the impression that my work was in error or not soundly done.

1. Rothman and Espenshade say that there was an "omission of older immigrant groups" and that this "gives a sense of incompleteness to Simon's findings" (p. 386). In fact, I did not omit any groups. The data on the oldest group were presented in my original published report to the Select Commission on Immigration and Refugee Policy in 1981, and are stated in full in my 1989 book. These data were omitted from my 1984 article against my will because of lack of space. The editors decided that the data could be safely omitted because I wrote in the article that the data were not relevant to the analysis. I worried then that the omission would lead to confusion, and this has indeed occurred.

As I explained in the article and the book, however, the data were not relevant for two reasons: first, at any reasonable discount factor, economic events 25 years and more in the future have little effect on current present-value-based decisions. Even more decisive is the fact that most older immigrants have produced grown children who pay into the system in the form of taxes, thus balancing the amounts that their parents are withdrawing as transfer retirement payments. Thus, immigrants' excess payments during their working lives represent a onetime windfall to the equilibrium system.

The only alternative to omitting the transactions of older persons with the system (as I did) would be to include the transactions of their children (and their grandchildren, etc.) with the public coffers. This is because the children of immigrants are a part of the stream of effects of immigration. This obviously would be unwieldy at best. Proceeding as I did in my analysis is a more convenient and analytically acceptable substitute.

2. Rothman and Espenshade write that "Simon's method of adjusting for public goods consumption is misleading because it includes immigrants' public goods tax contributions as net benefits to natives without including natives' own contributions as net costs to themselves" (p. 386). Here they repeat an error made by George Borjas in his review of my 1989 book which appeared in the Journal of Economic Literature. (A correction to his review has been accepted for publication in that journal; see also Simon, 1995, a correction in another medium, in which Borjas again made the same erroneous criticism). Borjas assumed that I was simply comparing the expenditures on immigrants from the public coffers, as measured by the survey data, to the contributions of immigrants to the public coffers in the form of their own tax payments. Such a comparison would have been flawed by the omission of public goods from the survey data. But I did not do that, as shown in my original tables. In fact, I compared the net balances in the flows for natives and immigrants against each other, hence allowing for public goods usage on a very generous assumption with respect to immigrants.

3. Rothman and Espenshade fault me for not breaking down the data by country of origin--"failure to separate immigrants by nationality" (p. 386). The nationality variable may be of interest in other contexts. It is not, however, relevant for a policy decision of the sort that arises in immigration legislation affecting the number of persons allowed into the country. There is no necessity for including such a variable, and I therefore do not do so. I feel no need to provide information on ethnicity and race that might distract attention from the variables of policy interest.

4. Rothman and Espenshade write that my "differentiating immigrants by year of entry is an effective method for demonstrating the effects of assimilation" (p. 385). I appreciate the words of praise, but my aim was not to show the effects of assimilation but rather to fulfill my assignment to the Select Commission--to make an appropriate assessment that would facilitate policy decisions about the number of immigrants that are to be admitted to the United States. That assignment influenced all research decisions made in the study.

Julian L. Simon teaches at the University of Maryland, College of Business and Management, Tydings Hall, College Park, MD 20742


Rothman, Eric S. and Thomas J. Espenshade. 1992. Fiscal impacts of immigration to the United States. Population Index 58(3):381-415.

Simon, Julian L. 1981. What immigrants take from, and give to, the public coffers. In U.S. Immigration Policy and the National Interest: Appendix D to the Staff Report of the Select Commission on Immigration and Refugee Policy, Papers on Legal Immigration to the United States, pp. 225-61. Washington, D.C.: U.S. Government Printing Office.

-----. 1984. Immigrants, taxes, and welfare in the United States. Population and Development Review 10(1):55-69.

-----. 1989. The Economic Consequences of Immigration to the U.S. Oxford: Basil Blackwell.

-----. 1995. Tired, poor, on welfare (continued). National Review, April 3:20.

-----. Forthcoming. A correction of a review by Borjas. Journal of Economic Literature.

Reply to Simon

Thomas J. Espenshade

Issues surrounding immigration and the fiscal impact of immigrants are high on the national agenda as we move toward the 1996 U.S. Presidential contests. Therefore, it is timely to revisit some of the matters that Eric Rothman and I first discussed in the review of research on the fiscal impacts of immigration to the United States that we prepared in 1992 (Rothman and Espenshade 1992). That review drew upon a study by Julian Simon (1981) in addition to works by numerous other authors. There are far-reaching conceptual and methodological problems that one confronts in estimating immigrants' fiscal impacts, but I shall confine this reply to the points raised in Simon's comment.

The first issue concerns whether it is appropriate to disregard the fiscal effects of older immigrants. Simon evaluates the fiscal impacts of families by comparing their benefits from public schooling for children and from transfer payment income in 1975 (unemployment compensation, public welfare, supplemental security, aid to dependent children, food stamps, social security, and Medicare and Medicaid) with an estimate of taxes paid to federal, state, and local governments. The typical family headed by a native-born individual paid $3,201 in taxes and received $2,279 in public services for a net gain to government of $922. Families headed by immigrants who came to the United States between 1950 and 1974 were even larger assets to the treasury--in amounts varying from $1,250 to $2,102 per immigrant family, depending on the year of immigration. It is on the basis of calculations like these that Simon concludes, "...immigrants are a remarkably good investment at any conceivable rate of discount" (p. 250).

What Simon omits from his calculation is the fact that immigrant families who migrated before 1950 place a substantial burden on other taxpayers. Migrants arriving between 1920 and 1949 received $834 more in services per family than they paid in taxes. The fiscal deficit was nearly $3,000 for families headed by pre-1920 migrants. These estimates are conservative because they exclude educational benefits to children in immigrant-headed households. They are also difficult to ignore because pre-1950 migrant households comprise one-half of all immigrant-headed families in the 1976 Survey of Income and Education (SIE). Simon (1981) argues that the fiscal circumstances of older migrants are irrelevant because "[in retirement]...their own children are supporting them through the Social Security and Medicare system" (pp. 249-50).This claim raises unresolved issues of intergenerational accounting, but it is not obvious without first doing the calculations that the fiscal contributions of children of immigrants do "balance" withdrawals by their parents in retirement. Moreover, by this reasoning one should also set aside the deficits attributable to elderly native-born families, which would have the effect of reducing the fiscal attractiveness of immigrants relative to natives. Either way, Simon's disregard of the revenue and expenditure impacts of immigrant families whose heads migrated prior to 1950 has the effect of creating an overly optimistic picture.

Second, the 1976 SIE data contain no direct information on consumption of public goods--goods characterized by the fact that a greater consumption by one individual does not reduce consumption by everyone else. Simon estimated that public goods expenditures represented 20 percent of the average tax dollar in 1975. Therefore, to incorporate public goods into comparisons between the fiscal impacts of immigrants and those of natives, he reduced native tax contributions by 20 percent, leaving the impression that the fiscal benefits of immigrants relative to natives are even larger after taking public goods into account. In 1992 Rothman and I argued that Simon's approach "implicitly assumes that natives pay nothing for their own public goods usage" (p. 386). This assessment still seems accurate. The assumption in this calculation is that native families are being forgiven 20 percent of their taxes and that all public goods are financed from the 20 percent of taxes paid by immigrants. This would be insufficient to sustain the 1975 level of public goods expenditure. Instead of trying to adjust for public goods consumption on the revenue side of the ledger, it would be more satisfactory to prorate public goods expenditures across all families using an average cost method.

Third, it is not entirely correct that immigration policy is concerned only with the "number of persons that will be allowed into the country". Migrant characteristics matter too, as illustrated by the Immigration Act of 1990 which attempts to infuse more skilled foreign workers into the U.S. labor market. In work that Deborah Garvey and I are doing to examine the state and local fiscal impacts of immigrants, we disaggregate immigrants by region of birth and show that differences among immigrant households as well as those between natives and immigrants are due largely to the effects of household income, family size, and place of residence, with nativity playing only a minor role after controlling for these other household characteristics. This approach seems preferable to deciding beforehand to ignore differences among immigrant groups out of concern that the results might be misused.

Fourth, Simon gives his data on year of entry a cohort interpretation and argues, for example, that "somewhere between 2-6 years after entry, the average immigrant family comes to earn about as much as the average native family, and after that earns more" (p. 243). However, work by George Borjas that is summarized in Borjas (1994) questions the utility of inferring life-cycle behavior from cross-sectional data. In particular, because the average education and other human capital characteristics of recent immigrant cohorts have been declining relative to native-born males in the labor force, the wage trajectory of immigrants after their arrival in the United States is actually much flatter than cross-sectional patterns would indicate. This, together with the fact that the SIE data are now 20 years old, also suggests that the fiscal impacts of immigrants are less positive than those estimated by Simon.

Finally, and more generally, one has to question the starting point of the exercise. Apart from educational benefits and pure public goods, the governmental benefits and services that families are assumed to receive are those that can be measured with the SIE data. These are essentially limited to monetary payments, so that much is being left out on the benefit side. This helps to explain why Simon finds that the typical native family in 1975 as well as the average immigrant-headed family in all immigrant cohorts from 1950 to 1974 paid more in taxes than they received in benefits. These figures imply that governments at all levels were running sizeable surpluses in 1975. But they were not. State and local budgets are often required by law to balance. And the federal government ran a $53 billion deficit in fiscal year 1975 (U.S. Bureau of the Census 1993). To avoid these problems, it would be preferable (albeit more tedious) to begin with public budgets as the primary data source, and then attribute government revenues and expenditures back to their sources and to their target populations, respectively.

Thomas J. Espenshade is affiliated with Princeton University, Office of Population Research, 21 Prospect Avenue, Princeton, NJ 08544-2091.


Borjas, George J. 1994. The economics of immigration. Journal of Economic Literature 32(4):1,667-717.

Rothman, Eric S. and Thomas J. Espenshade. 1992. Fiscal impacts of immigration to the United States. Population Index 58(3):381-415.

Simon, Julian L. 1981. What immigrants take from, and give to, the public coffers. In U.S. Immigration Policy and the National Interest: Appendix D to the Staff Report of the Select Commission on Immigration and Refugee Policy, Papers on Legal Immigration to the United States, pp. 225-61. Washington, D.C.: U.S. Government Printing Office.

U.S. Bureau of the Census. 1993. Statistical Abstract of the United States: 1993. 113th edition. Washington, D.C.: U.S. Government Printing Office.