9

State and Local Fiscal Effects of Immigration

9.1 INTRODUCTION

Of concern to policy makers and the public are not only the net fiscal effects of immigration for the nation as a whole, currently and over time, but also the effects on revenues and expenditures for state and local governments. Immigrants are not distributed equally among the states or localities, and state and local governments differ in their fiscal policies generally and in their policies toward immigrants specifically. Consequently, any examination of the fiscal effects of immigration at the state and local levels and the extent to which immigrants are a net fiscal burden or benefit must consider the individual circumstances of each jurisdiction.

The 1997 National Research Council report, The New Americans, estimated the net state and local government fiscal effects of immigration for only two states: California and New Jersey (National Research Council, 1997). Around that time, based on 2000 Decennial Census long-form sample data, California alone accounted for nearly one-third of the total number of 31 million foreign-born, while California and New Jersey, together with Florida, Illinois, New York, and Texas, accounted for about 70 percent of the foreign-born population. By 2011-2013, American Community Survey (ACS) data indicate that a larger number of immigrants had become more widely dispersed so that California accounted for about one-quarter of the total number of 41 million foreign-born and the same six states accounted for about 65 percent of the foreign-born population. Other states with significant numbers of foreign-born in 2011-2013 included Arizona, Georgia, Maryland, Massachusetts, and Virginia. Relative to the total

population, the foreign-born increased from 11 percent of the U.S. total in 2000 to 13 percent in 2011-2013. By state, as of 2000, the foreign-born accounted for 14 percent or more of the population in only seven states, while by 2011-2013 the foreign-born accounted for 14 percent or more of the population in 12 states; see further discussion in Section 9.3 below.

This chapter examines the state and local government fiscal effects of immigration for each of the 50 states and the District of Columbia for the 3-year period 2011-2013. We focus on the individual as the unit of analysis—more specifically, the independent individual. The panel’s analysis here attributes the fiscal costs of (and taxes received from) dependents to their parents. This independent-person concept best acknowledges that the fiscal costs or benefits of children are due to the decisions of their parents independent of the children’s own immigrant status. In addition, as in portions of Chapter 8, we distinguish among three immigrant generations (first, second, and third-plus, where “third-plus generation” is used as shorthand for all U.S. residents with two native-born parents).

Before proceeding to describe our measurement methods and results, it is worth referencing the extensive discussion of how, theoretically, immigrants’ net costs to state and local governments are treated in The New Americans (National Research Council, 1997, pp. 254-270). That discussion, well worth reviewing, also details the range of simplifying assumptions that are necessary to derive empirical estimates of net costs. For example, for tractability, one must generally assume that immigrants use government services, such as public libraries or highway maintenance, at the same rate as natives (except when there is an explicit eligibility criterion excluding immigrants). Under this assumption, the costs of each service are allocated equally to immigrants and to natives on a per capita basis. In our evaluation, we present results making similar assumptions but then also examine what the relative costs of immigrants would be using different assumptions about whether the overall level of spending on a particular service is likely to change. For example, if the number and staffing of libraries is assumed to be unchanged, we would ask, “What is the relative cost of immigrants, assuming they produce zero marginal costs to state and local governments for library services?” The rationale behind the marginal and average cost choice for allocating the cost of public goods—particularly pure public goods such as national defense—is discussed in detail in Chapters 7 and 8.

A key difference between the fiscal impact study in this chapter and the state-level analyses in The New Americans (National Research Council, 1997) is the unit of analysis. In The New Americans, analysis was done at the household level using the nativity of the household head to determine immigrant status. This panel’s preferred estimates present results based on independent individuals, including the cost of dependent children in the net benefit or burden of their parent(s). This makes our results more com-

parable to those presented in Chapter 8 and better captures revenues and expenditures of all immigrants, independent of who is listed as householder. We do also present results on a household basis (see Table 9-7).

9.2 MEASUREMENT METHODS

We constructed our estimates from the Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC). This nationally representative survey of people in households and group quarters, excepting institutions (e.g., jails, nursing homes), enabled us to identify generation status, including first generation immigrants (individuals who were born abroad who are noncitizens or naturalized citizens), second generation individuals (individuals who were born in the United States with at least one foreign-born parent), and third-plus generation individuals (individuals who were born in the United States with two native-born parents).¹ For each generation, we examined household living arrangements, income from various sources, and estimated taxes paid. It is important to account separately for second generation immigrants; this was not done in the state estimates in The New Americans (National Research Council, 1997), but it is done in the analysis in this chapter. At any point in time, many second generation immigrants are of working age and, when treated as independent individuals as in this report, contribute revenues that exceed costs. However, they may have represented a cost burden for their state and locality as children—costs that would not have been borne if their parent(s) had not entered the United States. Indeed, many second generation immigrants are themselves school-age children, whom we assign to their first generation parent(s) and who will likely represent a net burden for state and local governments for their education.

In order to achieve sufficient sample size for our analysis, we pooled 3 years of the CPS ASEC data covering 2011-2013.² Our sample represents, on a weighted basis and averaging over the 3 years, about 223 million independent people (essentially adults, as described below), of whom 16 percent are first generation and 8 percent are second generation. The remaining 76 percent are third-plus generation individuals, many of whose families have been in the United States for decades or centuries. The sample also repre-

___________________

¹ The second generation also includes those born abroad to an American parent with their other parent foreign-born, and the third-plus generation also includes those who are born abroad to two American parents.

² Were the ACS to include a question on place of birth of parents, it would be possible to carry out an analysis of state-level fiscal effects of immigration, by immigrant generation, with a much larger sample and correspondingly greater reliability than is possible with the CPS ASEC, even pooling over 3 years (see Chapter 10).

sents about 85 million dependent children in each year.³ In our study, as discussed more fully below, revenues and expenditures for dependent children are assigned to their parent(s) for estimation purposes, independent of the child’s own immigrant status.

We used the 3-year average of the 2011-2013 Census of Governments (COG) Annual Survey of State and Local Government Finances⁴ as our source for estimates of state and local revenues and expenditures of various kinds. Because different states provide different services at the state versus local level, we found it most useful to combine state and local revenues and expenditures to provide a complete picture of nonfederal government services. We did not have sufficient sample sizes or information on individuals to provide estimates for most substate areas—indeed, our estimates for many states are highly variable due to small sample sizes. These limited sample sizes also mean we do not estimate differences across places of origin.

We used simulation methods to piece together the data from the CPS ASEC and the 2011-2013 COG to estimate—for each independent person and his or her dependent children in the sample, weighted to be representative of their numbers in their state’s population—the revenues each provides to his or her state and locality of residence and the expenditures incurred by that state and locality on his or her behalf (including expenditures on behalf of any dependent children). Additional description of this method, as well as of differences in this approach from that used in The New Americans (National Research Council, 1997), is provided in the relevant sections below.

We then compared the resulting estimates of net state and local government fiscal benefit (or burden) for independent persons, characterized by immigrant status (first, second, or third-plus generation). We present comparisons among the states on an average-per-independent person unit basis by generation and on an aggregate basis. It can be the case, at one extreme, that a state has not only a high net fiscal burden (expenditures exceed revenues) per first and/or second generation immigrant but also a large number of first and/or second generation immigrants. At the other extreme, a state may have not only a low net fiscal burden (or a net fiscal benefit) per first and/or second generation immigrant but also a small number of immigrants. And there can be any combination in between. We further assessed how differences among and within the states in estimates of

___________________

³ Of these dependent children, considered in their own right, 4 percent are first generation, 21 percent are second generation, and 75 percent are third-plus generation. There are as many second generation dependents as independent adults; their costs in part explain why first generation independent persons are more costly.

⁴ Available: https://www.census.gov//govs/local/historical_data.html [November 2016].

net fiscal burden result from differences in characteristics of first and second generation immigrants, such as their age distribution, education level, and number of dependents as compared to native born individuals.

We also assess how differences among states in net fiscal burden result from differences in taxation and spending policies of individual states and their localities. We finally present some analyses to indicate the sensitivity of our main results to alternative assumptions about some components of revenues and expenditures. The key driver of differences has to do with education costs—the largest single part of state and local expenditures.

As defined in Chapter 7, the approach used in this chapter is a static analysis, producing estimates for a point in time. We did not attempt, for example, to play out, over time, the consequences for a state of its investment in education of first and second generation immigrant children on the skill mix of its labor force at a future date. Such an analysis would be difficult to conduct, not least because of the mobility of the population among the states so that children educated in one state may, to a greater or lesser extent, work as adults in another state.

Constructing Independent Person Units for Analysis

Most people live in households, as opposed to institutions and other similar living situations. Although a significant number of household residents live alone, many others live with relatives or with nonrelatives as in a group house. Many tax and expenditure programs are carried out on a family or household basis (e.g., state income and local property taxes and benefits from many low-income assistance programs, such as school meals). So the household would seem to be a natural unit of analysis, and The New Americans (National Research Council, 1997) carried out its analysis of net fiscal state and local government burdens for California and New Jersey on a per-household basis. Households, however, change in their composition within and across years and also may contain a mix of immigrant generations and a mix of related and unrelated members. For this conceptual reason, we conducted our analysis in terms of persons, which is also the unit for our analysis of fiscal effects over time at the national level in Chapter 8. Specifically, we constructed “independent person” units, consisting of one independent adult plus an assignment of any dependent children in whole or in part, as described below. Box 9-1 repeats the definitions of independent persons and dependent children given in Box 8-2 and also defines “independent person unit.”

Having classified each individual in the sample as independent or dependent, we then constructed independent person units for analysis. The assignment of dependent children in a household to independent

individual(s) includes all of their revenue and expenditure flows.⁵ Dependent children and their flows are split between parents if they reside in a two-parent household; they are assigned fully to the resident parent if in a single-parent household. They are assigned to the grandparent(s) if their parents are dependents; they are assigned to the householder when parents are not present and the householder is a foster parent or grandparent. Dependents in households with family members other than parents or grandparents are assigned to the highest earning independent relative.

___________________

⁵ In the case when the dependent weights differ from that of the independent person they are assigned to, the dependents and their flow amounts are multiplied by their dependent weight and then divided by the weight of the independent person(s) they are assigned to.

Dependents in households without any family members are assigned to the highest earning independent household member. As with dependent children in two-parent households, nonchild dependents (related and unrelated) are split between married couples in cases where they are assigned to one of the spouses. Ninety-four percent of dependent children in our dataset are assigned to parents, with an additional 5 percent assigned to other family members (remaining dependents are assigned to nonfamily members).

Defining Immigrant Generations for Independent Person Units

The classification of independent person units by immigrant generation was performed on the basis of the independent person’s status as follows: as in the other analyses in this report, independent individuals born abroad who are not citizens or who are naturalized citizens are classified as first generation immigrants. Independent individuals who are native born (including those born in Puerto Rico) and one or both of whose parents are foreign born are classified as second generation, as are those born abroad to an American parent with their other parent foreign-born. As defined above, the third-plus generation includes independent individuals who are native born to two native-born parents, as well as those who are born abroad to two American parents.

It bears repeating that independent person units are classified by the generation of the independent person; children are assigned to one or two independent persons on the basis of relationship and not generation. Thus, first generation independent person units with children may include children born abroad, those who in their own right would be classified as second generation, or both. Similarly, second generation independent person units with children may include children who are second or third generation when considered in their own right.

Estimating State and Local Revenues per Independent Person Unit

After constructing independent person units, the next step was to assign the revenues each such unit provided to its state and locality, using 2011-2013 COG data on taxes and other forms of revenue. Revenues (and expenditures) were assigned to each individual, with flows for dependents then being wrapped up to the independent persons who support them. So, for example, any benefits received by a child living with two parents would be assigned to the child and then half of the value would be pulled into each parent’s independent person unit amounts. For many types of revenue, the amount assigned to each independent person unit depended on the unit’s demographic and economic information. For example, state income taxes paid depended on income and taxes reported in the CPS data. Because CPS

data, on average, underreport income, amounts allocated for income and sales taxes were scaled up to equal COG state aggregates. The following types of revenues were assigned to independent person units (their percentage of all state and local revenues is shown in parentheses, but these average numbers mask the wide variations among states):

• property taxes (14%);
• general sales taxes (10%);
• selective sales taxes and public utilities (5%);
• individual income taxes (9%);
• business taxes (3%);
• higher education charges (tuition, etc.) (3%);
• school lunch sales (less than 1%);
• other education charges (less than 1%);
• insurance trust revenue (15%);
• other revenue (22%); and
• intergovernmental revenue (18%).

Table 9-11, in the Technical Annex to this chapter, provides detailed information on each revenue type and how the revenues for each type were allocated to independent person units.

Estimating State and Local Expenditures per Independent Person Unit

After the assignment of revenues, the next step was to assign state and local expenditures to independent person units using 2011-2013 COG data. Similar to revenues, these amounts often vary with individual characteristics; most notably, education expenses depend on the number and age of dependents. CPS noninstitutional Medicaid and public welfare expenditure amounts were scaled up to equal COG state aggregates. The following types of expenditures were assigned (their percentage of all state and local expenditures is shown in parentheses, and again these average values mask wide variations among states):

• higher education expenditures (7%);
• elementary and secondary education expenditures (16%);
• other education expenditures and libraries (4%);
• Medicaid and public welfare (16%);⁶
• insurance trust expenditures (11%);

___________________

⁶ While it is included in the 16 percent of COG expenditures from Medicaid and public welfare, we do not assign the 2 percent of the total 2011-2013 COG expenditures that went to institutional Medicaid spending.

• other expenditures and capital outlays (45%); and
• intergovernmental expenditures (less than 1%).

Table 9-12 in the Technical Annex provides additional information on each expenditure type and how the expenditures for each type were allocated to independent person units.

Differences from the Approach Used in The New Americans

We followed a similar but not identical approach to that used in The New Americans (National Research Council, 1997, Ch. 6) to estimate the cross-sectional, point-in-time net fiscal effects of immigrants on state and local government budgets. Below we indicate key differences and the reasons for them:

• Coverage. The 1997 report constructed net fiscal effects estimates for just two states—California and New Jersey—using March 1995 CPS data for California and the 1990 Public Use Microdata Sample for New Jersey. By using 3 years of pooled CPS ASEC data in our analysis,⁷ we were able to construct estimates for all 50 states and the District of Columbia, although small sample sizes for many states impair the quality of the estimates.
• Unit of analysis. The 1997 report used households as the unit of analysis on the grounds that most government programs and services are planned on a household basis. As argued above, this panel views households as too heterogeneous in composition. We therefore used an independent person unit of analysis, consisting essentially of an adult and any dependent children (or shares of children if married). This difference in analysis unit means that dollar amounts of net effects per unit are not comparable between the 1997 study and this report (even if one accounted for inflation and other differences). The reason is that there are about twice as many independent person units as there are households. Section 9.6 includes information at the household level and highlights how differences in household size can affect relative costs or benefits.
• Immigrant characteristics. The 1997 report distinguished between households headed by foreign-born individuals (further catego-

___________________

⁷ The CPS ASEC for any one year in 2011-2013 has about twice the sample size of the 1995 March CPS, and the pooling of the CPS ASEC over 3 years increases the CPS ASEC sample size of unique respondents twice again. We keep respondents appearing in 2 consecutive years in our sample for both years so that each of the 3 data years is fully representative of the non-institutionalized population in that year and we capture these respondents’ different revenue and expenditure flows in each of the 2 years.

• rized by region of origin—Europe/Canada, Asia, Latin American, and other) and households headed by native-born individuals. Other household members might include a mix of foreign- and native-born people. This study, in contrast, looked at three groups of independent persons: first, second, and third-plus generation. (Dependent children were assigned to one or two parents or another independent person in their household regardless of their own immigrant generation.) This grouping permitted us to ascertain the contribution of second generation independent persons, which in many states provide a return on the investment made in their education as children through taxes paid when they become working-age adults. We did not look at region of origin for first generation independent persons, in part due to small sample sizes for many states.

• Revenues and expenditures. The 1997 report broke out state from local revenues and expenditures, which we did not do because of differences among states in how functions such as education are allocated between the state and local governments. The 1997 report also looked at revenues and expenditures at the federal level for households living in California. In contrast, our analysis did not attempt to estimate federal fiscal effects for independent person units by state if those effects involved the individual directly rather than flowing through state or local governments. For example, federal funds for primary and secondary (K-12) education are included because the money is directed to the states and then distributed. In contrast, federal Social Security payments are excluded because the funds are directly sent to individuals. Similarly, state income taxes are included but not federal income taxes.

9.3 GEOGRAPHIC AND DEMOGRAPHIC DISTRIBUTION OF IMMIGRANTS

As background for our discussion of state and local fiscal effects, we provide information for two periods, 2011-2013 and 2000, not only on the geographic distribution of immigrants by state in the two time periods but also on variations among states in the demographic composition of their immigrant populations. For comparability when examining changes over time, we look at distributions of the foreign-born (noncitizens or naturalized citizens born abroad), which corresponds to the sum of “independent

persons” in the first generation plus any of their children born abroad.⁸ Comparisons across and within states among different groups are for the three generations of independent persons as defined above.

Geographic Distribution of the Foreign-born, 2000 Compared to 2011-2013

Table 9-1 shows the percentage of foreign-born in each state’s population for the period 2011-2013 compared with 2000, using data from the 2011-2013 ACS and the 2000 Decennial Census long-form sample. The states are ranked from highest to lowest percentage of foreign-born in 2011-2013. Also shown is the percentage point change between 2000 and 2011-2013. Percentages are expressed in whole numbers without decimals to remind the reader that the data are estimates from samples of the resident population.

For the United States as a whole, the foreign-born population as a percentage of the total increased by 2 percentage points over the period—from 11 percent in 2000 to 13 percent in 2011-2013. Percentage point increases by state ranged narrowly from no change in several states to 4 points in Maryland. Accordingly, it is not surprising that the patterns of geographic dispersion of the foreign-born were broadly similar in the two periods. Thus, the seven states with the highest percentages of foreign-born in 2011-2013—California, Florida, Hawaii, Nevada, New Jersey, New York, and Texas—were the states with the highest percentages of foreign-born in 2000, although overall the concentration of foreign-born individuals in these states has declined.

Table 9-1 also shows the numbers of foreign-born in 2011-2013 and the increase from 2000 (in thousands) by state. Numeric gains are important to keep in mind when considering how immigration may affect states’ fiscal pictures and their policies toward immigrants. Every state has experienced positive net numeric growth in its immigrant population since 2000. California, Florida, and Texas gained between 1 million and 1.4 million immigrants over the period, and New York State gained over 500 thousand. Six states gained between 300 and 500 thousand, five states gained between 200 and 300 thousand, and seven states gained between 100 and 200 thousand immigrants. Of the 22 states that experienced increases in numbers of immigrants of 100 thousand or more, 12 had populations with

___________________

⁸ Our analysis in this chapter is subject to the same caveats about the difficulties of identifying immigrants with existing data that are outlined in the Technical Annex to Chapter 2 above. In addition, as discussed further below, our state-level analysis is compromised by small sample sizes for many states in the CPS ASEC, even after pooling data over 3 years.

TABLE 9-1 Percentage Foreign-born Population by State, 2011-2013 and 2000, Ordered from Highest to Lowest Percentage Foreign-born in 2011-2013

^aEstimate is from the American Community Survey 5-year estimates because 3-year estimates are not available due to small sample size.

SOURCE: Foreign-born in 2000 from 2000 Decennial Census long-form sample, Summary File 4, Table QT-P14, Population Group—Total population, Nativity, Citizenship, see http://www.census.gov [November 2016]. Foreign-born in 2011-2013 from American Community Survey 3-year estimates, Table S0501: Selected Characteristics of the Native and Foreign-Born Populations, see http://www.census.gov [November 2016].

13 percent (the U.S. average) or more foreign-born in 2011-2013, and 10 had populations with smaller percentages of foreign-born.

Geographic Distribution of Independent Persons by Generation, 2011-2013

Turning to our analysis for 2011-2013, we first consider the composition by state of independent person populations from the CPS ASEC, classified by immigrant generation (first, second, or third-plus). Table 9-2 provides estimates of the three immigrant generations as percentages of each state’s population of independent persons, ordered from highest to lowest percentage for the first generation. We use this ordering for subsequent tables as well, to help readers focus on the states with the largest percentages of first generation independent persons, which are also the states

TABLE 9-2 Percentage Independent Persons by Immigrant Generation, by State, 2011-2013, in Order from Highest to Lowest Percentage (first generation independent persons)

NOTE: See text for definitions of independent person and immigrant generation. Rows may not sum to 100% for a state due to rounding error (values of 0.5 to 0.9 percent are rounded up).

SOURCE: Panel tabulations of the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

with the largest sample sizes for the first generation. (Tables 9-13 and 9-14 in the Technical Annex provide, respectively, annualized weighted sample counts and total 3-year unweighted counts of first, second, and third-plus generation independent persons by state in the pooled CPS ASEC data for 2011-2013.)⁹ For the United States as a whole, first generation independent persons are 16 percent of all independent persons, second generation independent persons are 8 percent of all independent persons, and third-plus generation independent persons are 76 percent of all independent persons.¹⁰

By state, West Virginia has the lowest proportion of first generation

___________________

⁹ As noted in Tables 9-13 and 9-14 in the annex, the full 2011-2013 sample does not account for overlap among sample cases due to the rotation group design of the survey.

¹⁰ The estimated percentages of first generation independent persons in 2011-2013 in Table 9-2 are generally higher than the corresponding estimated percentages of foreign-born in Table 9-1 (e.g., 16% versus 13% for the United States). The reason is that the denominator in Table 9-2 is all independent persons (essentially all adults) and not the total population, combined with the fact that, proportionally, the first generation includes more independent persons compared with dependent children (considered in their own right) than does the remaining population. Nonetheless, the ordering of states is not that different between Tables 9-1 and 9-2.

independent persons in the state’s total independent population (1%) and California has the highest proportion (35%). Ten states have first generation independent populations that make up less than 5 percent of the state’s total independent population.¹¹ First generation independent individuals in these 10 states (and the other 26 states below the national average of 16%) are less represented in the first generation independent population nationwide than are all of their independent individuals in the national independent population. Seven states have first generation independent populations that comprise at least 20 percent of their total independent population.¹² First generation individuals in these seven states (and the other seven states and District of Columbia above the national average) are more represented in the first generation independent population nationwide than are all of their independent populations in the national independent population. Consequently, caution should be taken in comparing national averages of state and local revenue and expenditure flows for the first, second, and third-plus generations, due to the differing composition of individuals in each state among the three generations.

Demographic Distributions of Independent Persons, 2011-2013

The three immigrant generations of independent persons that we define differ among themselves within and among states on characteristics that affect the net fiscal benefit or burden they entail for their state (and its localities). Among these characteristics are age, number of dependent children associated with the independent person unit, unit income, and education, for which we provide a broad overview below.

Age

The age of an independent person has an effect on the person’s net fiscal benefit or burden for the state and locality. Working-age people with employment, for example, typically pay significantly more in taxes than they receive from expenditures and therefore provide a net fiscal benefit to their state and locality, other things equal. However, if their independent person unit includes dependent children, these benefits are lessened and often reversed because of costs for the children’s education and other services. We observed these patterns in the national level analyses in Chapter 8 as well. The net fiscal benefit or burden of retirees will depend on a state’s

___________________

¹¹ These 10 states are Alabama, Louisiana, Maine, Mississippi, Missouri, Montana, North Dakota, South Dakota, West Virginia, and Wyoming.

¹² These seven states are California, Florida, Hawaii, Nevada, New Jersey, New York, and Texas.

tax structure and social services for the elderly; for low-income retirees on Medicaid, the net fiscal impact is likely to be negative.¹³

Table 9-3 shows the average age of independent persons by state and generation and the percentage who are 65 and older in our data for 2011-2013. Nationwide, first generation independent persons are 45.8 years on average; second generation independent persons are older, at 46.5 years on average; and third-plus generation independent persons are older still, at 48.5 years on average. Nationwide, the elderly population (age 65+) comprises 14 percent of first generation independent persons, 23 percent of second generation independent persons, and 19 percent of third-plus generation independent persons.

The general patterns evident for the nation hold for states, but there are some significant exceptions. For example, among the seven states with the highest percentages of first generation independent persons, the average age of this generation varies from 44 years in Texas to 51 years in Hawaii. Florida has higher-than-average percentages of people ages 65 and older in all three generations (20%, 27%, and 24%, respectively), while Hawaii has even higher percentages of people ages 65 and older in its first and second generation independent person populations (24% and 32%, respectively) but a lower-than-average percentage in its third-plus generation independent person population (17%).

Number of Dependent Children

The number of children in an independent person unit has an important effect on its net state and local benefit or burden, primarily stemming from the expenditure side of the ledger, for at least two reasons. First, education expenditures, which are allocated to school-age children, are a large item in state and local budgets (23% on average). Second, the more children in an independent unit, the larger the amount the unit is assigned for expenditures that are allocated to all persons (these expenditures total 49% on average—4% on other education and libraries and 45% on all other—see Table 9-12 in the Technical Annex to this chapter). Similarly, revenues that are allocated to all persons also total about half of revenues—see Table 9-11 in the Technical Annex to this chapter.¹⁴

___________________

¹³ As indicated in Table 9-12, Medicaid costs for the institutionalized, who are not represented in the CPS ASEC, are not included in the allocation of expenditures to independent person units.

¹⁴ A few of the expenditures and revenues we include in the group of those allocated to all persons are allocated selectively based on age. Liquor store expenditures, which are part of other expenditures, are allocated to all persons ages 21 and older. Motor fuels and tobacco product sales taxes, which are part of selective sales taxes, and motor vehicle license and motor vehicle operators license revenues, which are part of other revenues, are allocated to

TABLE 9-3 Average Age and Percentage, Ages 65 and Older, Independent Persons by Immigrant Generation by State, 2011-2013

NOTES: See text for definitions of independent person and immigrant generation. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2).

SOURCE: Panel tabulations of the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

In our data for 2011-2013, nationwide, first generation independent persons have an average of 0.52 children per unit, second generation persons have an average of 0.33 children per unit, and third-plus generation persons have an average of 0.36 children per unit (see Table 9-15 in the Technical Annex to this chapter). For most states, independent individuals in the second generation have fewer children than those in the first and third-plus generations, although the second and third-plus generations are quite similar (independent persons of the second generation in California have more children on average than do those in the third-plus generation). Among the seven states with the largest percentages of first generation independent persons, the average number of children per first generation independent person unit ranges from 0.39 in Florida to 0.64 in Texas, while the range per second generation independent person unit is from 0.24 in New Jersey to 0.47 in Texas. The range per third-plus generation independent person unit is from 0.31 in Florida to 0.38 in Texas. For the next seven states and District of Columbia that have between 15 and 20 percent of their independent persons in the first generation, the variation is even greater, with the District of Columbia having the lowest average number of children in each generation. States with smaller shares of immigrants also show wide variation in their average number of children per independent person, with Vermont averaging 0.29 children while Utah averages 0.53 children.

Income and Education

Income levels affect taxes paid and benefits received for independent persons and their children (see Table 9-16 in the Technical Annex to this chapter). Nationwide, average adjusted gross income (AGI) is lowest among first generation independent person units at about $29,450 per unit, considerably higher among second generation independent person units at $34,900 per unit, and higher still among third-plus generation units at $35,900 per unit. Among the seven states with the highest percentages of first generation independent person units, average AGI for the first generation varies from $26,100 per unit in Texas to $35,700 per unit in New Jersey. Average AGI for the second generation in these seven states varies from $28,250 per unit in Nevada to $37,900 per unit in New Jersey. For third-plus generation independent person units, New Jersey again has the highest average income of $47,250, in contrast to the lowest average income for third-plus generation independent person units of $33,800 in

___________________

all persons ages 18 and older. Alcoholic beverage sales taxes, which are part of selective sales taxes, and liquor store revenues, which are part of other revenues, are allocated to all persons ages 21 and up.

Florida. Among states with over one-quarter of their independent persons in the first or second generation (or at least 15 percent of independent individuals in the first generation), Arizona has the lowest average income for first generation independent person units ($25,100).

Income levels relate to education levels, and education levels differ significantly across generations. A larger percentage of first generation independent persons have not received a high school degree (28%) than is the case for the second and third-plus generations (10% and 9%, respectively). However, the percentage of first generation immigrants with advanced degrees beyond a bachelor’s degree is comparable to that of the other two generations (all between 10% and 12%; Table 9-17 in the Technical Annex to this chapter presents the state-by-state figures). The percentage with at least a bachelor’s degree is likewise comparable between the first and third-plus generations (28% and 29%, respectively). However, these statewide averages in part mask differences in higher education both across and within states. For example, in California, 8 percent of first generation independent persons have more than a bachelor’s degree, compared with 10 percent for second generation and 12 percent for third-plus generation independent persons. The District of Columbia has the highest share of individuals with more than a bachelor’s degree (29%), but 45 percent of second generation independent persons have more than a bachelor’s degree compared to 27 and 28 percent, respectively, of first and third-plus generation individuals. Part of the difference comes about because many of the states with small immigrant populations also have lower numbers of residents with more than a bachelor’s degree.

9.4 FISCAL VARIATION AMONG STATES, 2011-2013

We next look at state and local government revenues and expenditures by state and immigrant generation. States must generally balance their budgets year by year, but they vary greatly in the types and amounts of taxes they levy and the level of services they provide. Given the large differences in population size among states, it is important when examining state and local government fiscal data to convert the information to an appropriate population base. While we have calculated revenues, expenditures, and net fiscal effects for all states, the discussion below focuses on the 14 states and the District of Columbia with at least one-quarter of independent persons in the first or second generation (these states and the District of Columbia also have the 15 highest percentages of first generation individuals). Calculations are available, and presented in the tables, for all states, but caution must be exercised when examining differences for other states, especially those near the bottom of the tables, due to limited sample sizes. We also round all dollar amounts to the nearest $50 to emphasize that the basis for

our estimates is a relatively small sample. In the remainder of the chapter, we present estimates of revenues, expenditures, and net fiscal effects on a per-independent person unit basis (where dependent children and their revenues and expenditures are assigned to their parents).

State and Local Government Revenues

While most state governments rely on general sales and income taxes and local governments rely primarily on property taxes, the composition of state and local revenues varies substantially. Nine states (including Florida, Nevada, Texas, and Washington) do not levy a broad-based personal income tax, and five states do not levy a general sales tax.

Table 9-4 provides population-based revenue estimates by state for all independent person units and those in each generation with per-unit amounts derived using the allocation process described in Table 9-11 in the Technical Annex to this chapter. For the United States as a whole, 2011-2013 annualized state and local government revenue averaged $14,700 per independent person unit. This amount masks considerable variation by state, particularly at the higher end. Thus, 17 states averaged between $11,650 and $12,950 per independent person unit; 13 states averaged between $13,000 and $14,500; 16 states averaged between $14,550 and $17,800; and five states exceeded $17,800 per independent person unit. The five jurisdictions with the highest average state and local government estimated revenue per independent person unit were Alaska ($36,400), the District of Columbia ($27,600), Wyoming ($24,150), New York ($22,400), and North Dakota ($20,300), while the five states with the lowest state and local government estimated revenue per independent person unit were Idaho ($11,650), Florida ($11,800), New Hampshire ($11,850), Georgia ($11,900), and Arizona ($11,900). If we limit our analysis to the 15 jurisdictions with the largest shares of first and second generation immigrants, the average revenue per independent person unit is $15,750 and varies from $11,800 in Florida to $27,600 in the District of Columbia.

By generation nationwide, state and local government revenue averaged about the same amount per first generation and third-plus generation independent person unit: $14,350 and $14,700, respectively. Revenue was higher for the second generation, averaging $15,500 per second generation independent person unit. However, these national similarities among generations mask large differences across states among generations. For the 15 jurisdictions with the largest shares of first and second generation immigrants, the average revenue for an independent person unit in the third-plus generation exceeds that of a unit in the first generation by $1,450 ($16,100 versus $14,650) and is only slightly lower than that of a unit in the second generation ($16,200).

TABLE 9-4 State and Local Revenues per Independent Person Unit (rounded to nearest $50), by Immigrant Generation by State, 2011-2013

NOTES: See text for construction of revenues by state and generation. Because the difference between first and third-plus generation revenue amounts is taken from the unrounded estimates and then rounded to the nearest $50, the value may differ from the first generation column less the third-plus due to rounding in some cases. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2). Caution should be taken when examining the state-level estimates, especially those near the bottom of the table, because of small first (and second) generation populations for many states.

SOURCE: Panel estimates implemented on the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

State and Local Government Expenditures

Spending varies across states, with some states raising and spending more money than others. Note that, due to balanced budget rules, the states that raised more revenues almost always spent more funds. Table 9-5 provides population-based expenditure estimates by state for all independent person units and by generation with per-unit amounts derived using the allocation process documented in Table 9-12 in the Technical Annex to this chapter. For the United States as a whole, 2011-2013 annualized state and local government expenditures averaged $13,850 per independent person unit, or about $900 less than was raised in revenue. Sixteen states had average expenditures between $10,450 and $11,950 per independent unit; 17 states were between $12,000 and $14,000; 13 states were between

$14,050 and $16,700; and five states exceeded $16,700 of expenditures per independent unit. The five states with the highest average state and local government expenditures per independent unit were Alaska ($29,950), California ($16,750); the District of Columbia ($28,500), New York ($20,700); and Wyoming ($20,750); the lowest were Arizona ($10,900); Arkansas ($10,900), Florida ($10,850), Idaho ($10,450, and Indiana ($11,250). If the analysis is limited to the 15 states with the largest state share in the first generation, the average spending per independent person unit is $14,950 (which is higher than the national average) and ranges from $10,850 in Florida to $28,500 in the District of Columbia: the same two lowest and highest jurisdictions among these 15 for average revenue per independent person unit.

TABLE 9-5 State and Local Expenditures per Independent Person Unit (rounded to nearest $50), by Immigrant Generation by State, 2011-2013

NOTE: See text for construction of expenditures by state and generation. Because the difference between first and third-plus generation expenditure amounts is taken from the unrounded estimates and then rounded to the nearest $50, the value may differ from the first generation column less the third-plus due to rounding in some cases. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2). Caution should be exercised when examining the state-level estimates, especially those near the bottom of the table, because of small first (and second) generation populations for many states.

SOURCE: Panel estimates implemented Current Population Survey Annual Social and Economic Supplement for 2011-2013.

By generation, for the United States as a whole, annualized state and local government expenditures for the 2011-2013 period were considerably higher for first generation independent person units ($15,950) than for second generation ($13,800) or third-plus generation ($13,400) independent person units. This was due to greater program participation (including public education). For the 15 states with the largest share of their independent population in the first generation, average expenditures for each immigrant generation were higher than the national averages by generation, but the gap between the generations was smaller (with average expenditures of $16,350 for the first generation versus $14,600 for the second and $14,450 for the third-plus generation).

9.5 AGGREGATE FISCAL EFFECTS BY STATE

Total state and local government revenues averaged $3.3 trillion per year in 2011-2013, while total state and local government expenditures averaged $3.17 trillion, nearly balancing out. In theory, when one looks across the amount of revenues contributed by each generation and the expenditures received by each generation, and if balanced budget rules held, the net total across generations in each state should be zero. In fact, because certain state and local funds run surpluses and deficits, no state actually has state and local revenues precisely equal to state and local expenditures. California, the state with the largest population and the largest number and percentage of first generation independent person immigrants, had the largest positive net difference in dollars between total average annual state and local revenue and expenditure flows in 2011-2013 ($22.9 billion) out of all 50 states and the District of Columbia. With spending exceeding revenues by $2.4 billion, Pennsylvania had the largest negative net difference in dollars. The District of Columbia had the largest negative net difference as a percentage of state and local revenues (−6%), while North Dakota had the largest positive net difference (+23%).

We exclude the institutional portion of Medicaid spending ($72 billion) from our estimates due to missing this population in our data, which widens the gap between aggregate U.S. revenues and expenditures in 2011-2013. After we take out institutional Medicaid spending, all but two states have positive budget balances (compared with seven negative-balance states when all expenditure flows are included).

Nationwide, the fact that the state and local government revenues we allocated exceeded expenditures by $197 billion, after excluding institutional Medicaid spending, means that an average net difference of $900 was assigned per independent person unit. By state, average net differences resulting from fiscal imbalances that were assigned at the unit level varied from -$850 in the District of Columbia to $6,450 in Alaska (see the “All”

column of Table 9-6). With net differences in revenues and expenditures ranging from positive to negative across states, when comparing net differences per independent person unit for different immigrant generations, it can be difficult to disentangle how much variation is from across-generation cost differences versus net cost differences among states.

Our analysis is for a specific time period for which state fiscal balances may not be typical. For example, the difference between state and local total revenues and total expenditures was positive in 2011 ($281 billion, or $353 billion after excluding Medicaid institutional spending). In 2012, this

TABLE 9-6 Net Difference between State and Local Revenues and Expenditures per Independent Person Unit (rounded to nearest $50), by Immigrant Generation by State, 2011-2013

NOTES: See text for construction of revenues and expenditures by state and generation. Because the difference between first and third-plus generation net difference (revenue less expenditure) amounts is taken from the unrounded estimates and then rounded to the nearest $50, the value may differ from the first generation column less the third-plus due to rounding in some cases. Similarly, because differences between revenues and expenditures are calculated on unrounded numbers and then the difference is rounded, these values may differ from calculated differences between Tables 9-4 and 9-5. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2). Caution should be taken when examining the state-level estimates, especially those near the bottom of the table, because of small first (and second) generation populations for many states.

SOURCE: Panel estimates implemented on the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

switched to a negative difference, largely due to a very significant decline in insurance trust revenue (government employee retirement revenue fell from $554.3 billion to $169.9 billion) between the 2 years, reflecting changes due to delayed payments during the recession. In 2013, the net fiscal state balance became positive again. To smooth out these cycles, we averaged revenues and expenditures for 2011-2013. If 2011 rather than 2011-2013 state and local revenue and expenditure amounts were assigned to our sample, the average net difference per independent person unit would become even more positive, going from $900 to approximately $1,600 per independent person unit. If, instead, we had used 2012 amounts, the individual unit average net difference would turn negative (−$200). If we had eliminated all insurance trust contributions and payments along with excluding the institutional portion of Medicaid spending, averaged over the 2011-2013 period, 11 states would be estimated to have higher expenditures than revenues, while, on average per independent person unit, there would be $200 more in revenues raised than spent in the country as a whole.

9.6 NET EFFECTS OF IMMIGRATION ON STATE AND LOCAL BUDGETS

Estimated Differences in Net Fiscal Effects per Independent Person Unit by Generation

Table 9-6 shows the estimated net differences between state and local revenues and expenditures, by generation per independent person unit. Our estimates are derived using the replicate weights in the 2011-2013 CPS ASEC, whereby calculations of net differences are run many times in order to estimate an appropriate standard error and coefficient of variation, or CV (the standard error as a percentage of the estimate). We used replicate weights and show CVs in this part of the analysis (see Table 9-18 in the Technical Annex to this chapter) because we have reached the primary question of interest: How much do first and second generation units cost their states and localities? Also, the net differences are the result of balancing revenue and expenditure assignments, thereby magnifying the errors in each.¹⁵

As seen in Table 9-6, for the United States as a whole, first generation independent person units (which include first and second generation children assigned to independent first generation persons) cost the states on net about $1,600 each. In contrast, second generation independent

___________________

¹⁵ Not only does the CPS ASEC have sampling error, which is large for many states even pooled over 3 years, but also both the CPS ASEC and the COG have other sources of error, such as response error, imputation error, and the like.

person units (which include second and third generation children assigned to independent second generation persons) contribute on net to state and local budgets about $1,700 each, and third-plus generation independent person units (which include their children) contribute on net to state and local budgets about $1,300 each.¹⁶ These estimates of the fiscal impact imply that the total annual aggregate impact of the first generation and their dependents, averaged across 2011-2013, is a cost of $57.4 billion, while the second and third-plus generation individuals (and their children) create benefits of $30.5 billion and $223.8 billion, respectively. Note that the surplus revenues raised amount to $197 billion, which equals the surplus across all 50 states.¹⁷This overall pattern is largely driven by the larger education costs for first generation independent person units, which include more children on average than units of the other two generations. By the second generation, immigrants are a net win for the states as a whole, given that they have fewer children on average than first generation units and are contributing in revenues more than they cost in expenditures.¹⁸ State by state, however, there are wide variations in net gains or losses, although the panel is unable to make claims as precisely as we would like for many of our state and local estimates because of the large sampling errors.

The demographic differences between the first, second, and third-plus generation independent person units serve as the drivers of the differences

___________________

¹⁶ The finding here, that the first generation generates higher net fiscal costs at the state and local level, is consistent with that from the parallel analysis in Chapter 8 (Table 8-1). The numbers for the second and third-plus generations here and for the second and third-plus generation in Chapter 8 do not map exactly, due to slight methodological differences. The Chapter 8 analysis is for 2013, while the analysis here averages over the 2011-2013 period. More importantly, the two analyses treat grant-in-aid spending from the federal government (which pays for programs like Medicaid and some welfare programs) differently. Chapter 9 includes these revenue transfers to states in state revenues (with the exception of the institutional portion of Medicaid spending), while Chapter 8 does not. Also, in Chapter 8, the funding raised by the federal government to pay for the grants-in-aid is treated as either federal taxes or federal deficit spending, which leads to both lower spending and lower revenue estimates at the state and local level. Finally, in the Chapter 8 analysis, there is no balanced budget assumption—the aggregates are as reported in the National Income and Product Accounts. If grants-in-aid are taken off both sides of the state/local ledger, the net fiscal balance becomes more negative.

¹⁷ The $197 billion aggregate surplus here is calculated by totaling the unrounded estimates of net fiscal effects by state multiplied by the average number of independent persons in each year (see Table 9-13 in the Technical Annex to this chapter); this will differ somewhat from the total if the rounded estimates in Table 9-6 are used instead.

¹⁸ These results are driven by the fact that the costs of dependents are assigned to their parents. If, instead, taxes paid and services received were assessed at the individual level, with dependent children considered in their own right, the relative costs would shift across groups. Because half of all second generation individuals are dependents, allocating all costs and benefits to each person (rather than wrapping up dependent children to independents) would cause the average net fiscal cost for first generation individuals to decline and reverse sign in many states, while the costs for second generation individuals would increase.

in revenue and expenditure flows across them. First generation independent person units generate higher costs due to the presence of more dependent children, on average, in those units. Much of this comes from the assignment of K-12 education spending, which accounts for 16 percent of all state and local spending. If we remove all K-12 education expenditures from our estimates, rather than assigning school spending to the students themselves (which in most cases means wrapping it up to the independent parents/individuals who support them), the spending difference between the first and third-plus generations decreases from $2,900 to $1,950, a 32 percent decrease.

Beyond education spending, the 45 percent of spending flows that are classified as “other” (hospitals, health, veterans’ services, etc.—see Table 9-12 in the Technical Annex to this chapter) is allocated to all people, both dependent and independent, evenly. Thus, first generation independent individuals on average have higher total expenditure amounts from these flows allocated to all because they have more flows wrapped up from more dependent children. However, if a portion of these spending costs were treated as fixed expenditure flows, irrespective of population numbers (analogous to the treatment of national defense in some scenarios in Chapter 8), the addition of first generation independent person units to the population base would reduce these average costs for the second and third-plus generation units; the dollar amounts would be spread across a larger population (which is what we did in our baseline estimates), but the marginal cost to the state would not change. Additionally, although it is not evident in cross-sectional estimates, the majority of the dependent children of first generation immigrants who are second generation and whose costs are assigned to their parents will go on to become net contributors once they reach working ages.

Although per unit spending on the second generation independent person units is slightly more than it is on the third-plus generation units, the per unit net difference between revenues and expenditures is the most positive for second generation independent person units. With a positive net difference of $1,700, second generation independent person units contribute $400 more on average than third-plus generation units. This corroborates findings reported in Chapter 8. However, this is largely due to the distribution of second generation independent person units across states, rather than the relative contribution of second versus third-plus generation units within a state. The third-plus generation independent person units contribute less in taxes and other revenue flows on average than the second generation, despite having the highest average income of all three generations. Looking at specific tax flows, the second generation units contribute the most in both state income tax and general sales tax on average, followed by the third-plus generation units and then the first

generation independent person units, reflecting the lower average AGI of first generation independent persons in the sample. However, this is driven by differences in tax structures in place across different states, rather than differences in the characteristics of the independent person units.

Note that the average U.S. spending and revenues raised per independent person unit hide differences across states. Thus, because many of the states with small numbers of first generation independent person units also have lower spending and taxes, the spending per third-plus generation independent person unit is lower for these states than for the states where immigrants often settle. Focusing on the 14 states and the District of Columbia with the highest share of first generation independent person units, one can see differences across generations as well. Moreover, as noted above, these differences can vary from year to year depending on whether a state is running a surplus or deficit. While first generation independent person units have more spent on them than revenues contributed in these 15 jurisdictions, this amount varies from a net cost of $100 in Maryland to a net cost of $3,050 in Washington State. For the second generation independent person units, while on average across states more revenues are raised than money spent on them, the differences vary from a net cost of about $400 per second generation independent person unit in Texas to a net contribution of $7,100 per such unit to the District of Columbia’s budget. Similarly, whether an average third-plus generation independent person unit costs or contributes to a state (and local) budget varies from a net cost of $1,300 in the District of Columbia to a net contribution of $3,100 in California. These differences are largely driven both by different demographic and economic characteristics of individuals and by fiscal choices made by state and local governments.

Estimated Differences in Net Fiscal Effects at the Household Level

If, instead of independent person units, one were to use self-identified households, very similar patterns result across generations (as defined by the generation of the designated head of household), albeit the estimates are often about double the estimates for independent person units. That is, one finds that first generation households in general have higher state and local net costs or smaller contributions than do the second or third-plus generation households (see Table 9-7). Again, this pattern varies across states, with second generation households often, but not always, contributing more to a state and local surplus than either first or third-plus generation households. The estimated amounts are higher because, typically, average household size includes more than one independent person. Differences in household size and composition will affect the relative size of net contribution or burden. Table 9-19 in the Technical Annex to this chapter presents

TABLE 9-7 Net Difference between State and Local Revenues and Expenditures per Household Unit (rounded to nearest $50), by Immigrant Generation by State, 2011-2013

NOTE: See text for construction of revenues and expenditures by state and generation and for definitions of household immigrant generation. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2). Caution should be taken when examining the state-level estimates, especially those near the bottom of the table, because of small first (and second) generation populations for many states.

SOURCE: Panel estimates implemented on the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

information on average household size by generation (of head of household) by state. Table 9-20 in the Technical Annex to this chapter provides annualized weighted sample counts of first, second, and third-plus generation households by state in the pooled CPS ASEC 2011-2013 data.¹⁹

An advantage of using households as the unit of analysis is that assumptions do not have to be made about allocation of income, dependents, or receipt of social services among independent persons in the household—they can be assigned to the whole household. However, a single generation status must be assigned to the entire household, even for cases in which different independent persons within the household are of different genera-

___________________

¹⁹ As noted in Table 9-20, the full 2011-2013 household sample does not account for overlap among sample cases due to the rotation group design of the survey.

tions. Thus, for the estimates presented here, the panel assigned generation status at the household level using the generational status self-reported by the householder. For mixed cases, if the nonimmigrant is more likely to be the householder (say, because of more facility with English), then our estimates will be muddied. While not the standard procedure, which would use either all individuals, all households, or all families, presenting results according to independent person units (and assigning dependent children to their parents to form a unit) provides a cleaner comparison, as it avoids inconsistencies caused by differences in household size or composition.

Decomposing Cross-Generation Differences

To highlight the relationship between the demographic and economic differences across generations of independent person units and the variations in state and local revenues and expenditures across these generations, the panel examined how differences in characteristics like age structure and education levels of the first, second, and third-plus generations impact the average net contribution (or burden) of independent person units of each generation. Table 9-8 shows results from multiple regression analyses that follow closely those conducted in Chapter 8 and reported in Table 8-3, in which net fiscal impact at the state and local level is regressed on generation as defined by independent person units.²⁰ The third-plus generation of independent person units is used as the reference category so that coefficients can be reported on indicators for the first and second generations. We present six models, with each subsequent model adding more control variables to account for cross-generational differences. For brevity, we only report the regression coefficients for immigrant generation, which represent, in dollars, the net fiscal impacts associated with being a first or second generation independent person unit, compared to third-plus generation independent person units.

In almost all cases, the regression coefficients for immigration generation are statistically significant, with first generation independent person units having a net fiscal cost relative to the third-plus generation, while the opposite is true for second generation independent person units. Note that, because time since arrival is not accounted for, the net fiscal burden

___________________

²⁰ Some of the methodological differences between Chapter 8 and Chapter 9 in how net fiscal impact estimates at the state and local level are generated are detailed in footnote 16. In addition, the sample for the estimates and regression analyses here in Chapter 9 differs from the sample used in Chapter 8. The Chapter 8 regression analysis uses observations of independent individuals from a pooled CPS 1994-2013 sample that has adjusted population weights to represent the total residential population (including institutionalized residents). Chapter 9 uses observations of independent individuals from a pooled CPS 2011-2013 sample that is representative of the noninstitutionalized population.

TABLE 9-8 Regression Analysis of Net Fiscal Impact at the State and Local Level per Independent Person Unit, by Immigrant Generation, 2011-2013

of a first generation independent person unit is not the same as that for a new first-generation immigrant (as just 14% of our first generation sample arrived in the U.S. after 2006). On average, recently arrived first generation independent person units (since 2006) have small net fiscal burdens relative to first generation units that have been in the United States longer because the new first generation immigrants heading the unit tend to be younger, have more education, and have fewer dependent children.

Following the same order as in Chapter 8, we add control variables that typically explain (statistically account for) demographic and economic differences; an additional model directly controls for income so that the importance of that factor can be discussed. Note that the order in which the control variables are added matters, so decreases in the difference from the comparison group (third-plus generation independent person units) with each additional variable can be seen as the additional marginal effect of including that variable. For example, the effects we find on age in part

NOTES: Each column presents coefficients and significance levels from a separate ordinary least squares (OLS) regression of net fiscal impact at the state and local level (dependent variable) on indicators for immigrant status (x variables) and indicators for the other characteristics listed. Coefficients are the marginal effects in terms of dollars per independent person unit that are associated with the given immigrant status, relative to third-plus generation independent person units. A positive number is an improvement or savings in net fiscal impact; a negative number is a reduction or deficit. Thus a coefficient on a “1st generation” independent person unit equal to +1,000 implies that, compared to a third-plus generation unit, a first generation unit has a more positive net fiscal impact by $1,000 at the state and local level. Age groups are measured in 5-year intervals. Asterisks denote statistical significance at the 1 percent (***), 5 percent (**), or 10 percent (*) level.

SOURCE: Panel estimates implemented on the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

are related to the likelihood of having different numbers of dependents at different points in an independent person’s life cycle.

Model 1, which does not include any control variables, reports the difference in net fiscal impacts of the first and second generation independent person units relative to the third-plus generation units. Controlling for no other factors, a first generation independent person unit on average costs state and local governments $2,913 more than an additional third-plus generation independent person unit, while an additional second generation independent person unit contributes $384 more than an additional third-plus generation unit. If the regression analysis controls for average state spending and taxes by introducing state fixed effects, as is shown in the right-hand column of Table 9-8, first and second generation independent person units are on average about $200 more costly, compared to a third-plus generation independent person unit, than with no state fixed effects (middle column of Table 9-8, labeled “OLS Regression”). Table 9-8

includes regression model runs with and without state fixed effects, but the discussion below will focus on the coefficients for the model runs without the state fixed effects.

Model 2 adds a set of basic controls for age of the independent person, calendar year, and gender. As discussed in Section 9.3, the first generation independent individuals are on average the youngest of the three generations. With more independent persons concentrated in child-raising ages, the first generation units have, on average, more dependent children and consequently have higher state and local expenditures on education. Because we are limiting our estimates to a 3-year period that is at a similar point in the economic cycle, the year controls make little difference and the coefficients on calendar year are not significant in this model (although they do have small but statistically significant effects in later models).²¹ Similarly, the gender makeup of each generation does not affect the relative fiscal impact. While male independent person units appear to contribute more than females in this model, the difference in income between the two genders is driving this and there is no significant difference between men and women in our later model that controls for income.²² Thus, after controlling for age group (as well as year and gender), the fiscal impact of the first generation units (−$2,429) becomes less negative relative to the third-plus generation independent person units by about $500. The second generation units have the highest share of elderly independent persons in them (age 65 and older), relative to the units of the other two generations, leading to additional public costs. When this is controlled for under Model 2, the second generation’s fiscal impact (+$762) becomes more positive, relative to the third-plus generation by about $400.

Because the independent persons in the first generation units have less education on average than those in second and third-plus generation units, controlling for education (Model 3) shrinks the negative net fiscal impact for first generation independent person units to −$1,478 (a decrease of about 40 percent from the Model 2 net fiscal impact). Conversely, controlling for education lowers the positive net fiscal impact for second generation units due to the higher educational attainment of second generation independent persons compared to third-plus generation independent persons.

Model 4 incorporates controls for race and ethnicity in addition to the controls already included in Model 3. As noted in the discussion of the Chapter 8 regression analyses, race and ethnicity may proxy for differences

___________________

²¹ The coefficients on calendar years 2012 and 2013 (relative to the comparison group for calendar year 2011) in Model 2 are –34 and −72, respectively, and are not statistically significant.

²² The coefficient on male (relative to the female comparison group) in Model 2 is 1,689 and is statistically significant at the 1 percent level. However, when we introduce a control for income in our final model the coefficient on male is −36 and no longer statistically significant.

in treatment and opportunity, affecting earnings opportunities and possibly labor force participation. Under Model 4, the first-to-third-plus generation gap in net fiscal impact closes further (to just −$1,166) and the independent person units in the second generation show a small increase (going from +$422 to +$565) in their net fiscal impact relative to units in the third-plus generation.

Controlling for the number of dependent children (Model 5) has a dramatic effect on the relative costs of an average unit in the first and second generations, relative to an average third-plus generation unit. Because first generation independent person units have more dependent children on average compared with third-plus generation units (0.52 versus 0.36), they incur higher public education costs when education expenditures are assigned fully to school-aged children rather than a portion being considered a public good. Controlling for the number of dependents decreases the negative net fiscal impact of a unit in the first generation relative to third-plus generation units by close to $500 (going from −$1,166 to −$706). In contrast, due to having fewer dependent children as compared to third-plus generation independent individuals, the fiscal benefit of second generation units relative to third-plus generation independent person units declines by about half (to +$258), compared to the fiscal benefit before controlling for dependents (Model 4). The coefficient on number of dependents indicates that, for each additional dependent child, an independent person unit’s net fiscal impact is decreased by almost $9,750.²³

Finally, Model 6 in Table 9-8 shows how the net impact changes when AGI is controlled for in the regression. With average incomes for first generation independent person units being the lowest of the three generations (see Table 9-16 in the Technical Annex to this chapter), they contribute less to state and local tax revenues and are more likely to receive government benefits. Adding income to the control variables already included in Model 5 further diminishes the difference in net fiscal impact between independent person units in the first and third-plus generations to just −$421, and the difference between independent person units in the second and third-plus generations is not statistically significant. The Model 6 coefficient on AGI indicates that for each additional $100 of income, a unit’s net fiscal impact is made more positive by about $11.²⁴ Thus, after adding controls for age group, year, sex, education, race and ethnicity, number of dependents, and income, the average negative net fiscal impact of the first generation units relative to independent person units in the third-plus generation is

___________________

²³ The coefficient on the number of dependents in Model 5 is −9,739 and is significant at the 1 percent level.

²⁴ The coefficient on AGI in Model 6 is 0.107 and is statistically significant at the 1 percent level.

significantly diminished. Demographic and economic characteristics of first generation independent person units account for close to −$2,500 of the original −$2,931 gap relative to third-plus generation units. These characteristics also account for all of the positive contribution of second generation independent person units relative to the third-plus generation units.

When the regression analysis sample is limited to independent person units living in the 14 states and the District of Columbia in which at least one-quarter of all independent persons belong to the first or second generation, the results for the first generation are similar to those in the sample that includes all states. Demographic and economic characteristics of first generation independent person units in these jurisdictions account for close to $3,100 of their original $3,383 net fiscal cost relative to a unit in the third-plus generation. For second generation units in these jurisdictions, the initial difference in fiscal impact compared with third-plus generation independent person units is statistically insignificant, but after controlling for demographic and economic characteristics, a second generation unit would contribute $150 more than an average third-plus generation unit.

9.7 ALTERNATIVE TREATMENTS OF EDUCATION COSTS

As noted in Section 9.6, much of the differential expenditure burden for first generation independent person units comes from the cost of educating the dependent children in the unit. However, these children will grow up to be higher contributing second generation adults. In our baseline estimates, the panel assigned the cost of education to families that include children attending school. This means K-12 costs are assigned based on the presence of school-age children and public higher education payments are assigned to independent persons who are either attending, or have a dependent attending, an institution of higher education. This allocation ignores the future public benefit of education to those with and without children and the benefit to society of a better educated population. On average, K-12 spending per student is almost $9,000 in the United States as a whole but varies from $5,400 per pupil in Utah and $5,550 in Arizona to $26,950 in the District of Columbia.

To examine the possible public benefit spillovers, the panel re-ran the baseline estimates with various alternative assumptions about who receives the benefit (or would be responsible for the cost) of K-12 and public higher education. We first assigned half of the cost of K-12 education accruing to state and local governments to everyone within the state (including all independent and dependent persons) on a per capita basis. The remaining half was assigned to students as in the baseline scenario. This approach recognizes a level of public value to others of school spending. Table 9-9 shows how these differences in assigning education expenses affect estimates of

the relative costs for the United States as a whole and for specific states. Allocating half of K-12 expenses per capita in this manner, the net fiscal burden of first generation independent units declines by about $250 per independent unit (a change in net fiscal impact from −$1,600 to −$1,350). The costs borne by second generation independent person units increase by about $150 per unit, reflecting the lower number of dependents for second generation independent units overall; costs remain about the same for the third-plus generation units under this alternative scenario.

If we instead allocate half of the K-12 expenditures to just the independent persons, rather than to all persons, and the remaining half to students, the net fiscal impact of first generation units becomes −$1,250 and the second and third-plus generations have small increases in the costs they bear. This reduces the difference in net costs between first and third-plus generation independent person units from $2,900 to $2,500. Row 4 of Table 9-9 shows the results of assigning half of state and local spending on both K-12 and higher education to just the independent persons. Including higher education spending in this assignment approach has little effect on relative revenues and expenditures as reflected in the U.S. averages for independent person units in the first and third-plus generations.

The lower three panels of Table 9-9 illustrate, for specific states, how independent person units in the generations fare when education expenses are allocated differently. Most of the changes are small, but when we allocated half of the K-12 education benefits in California to independent persons, the net cost of first generation units declined by $300, with a similar decline in the net benefit from units in the second and third-plus generations. Interestingly, second generation Californian independent person units have increased fiscal contributions to the state under the scenario in which half of K-12 and higher education costs are attributed to all independent persons. This reflects higher-than-average usage of higher education by second generation Californians. Similarly, how educational expenses are allocated in New Jersey affects the relative costs and benefits between units in the first and second generations, with the relative benefits for third-plus generation units staying fairly constant.

9.8 MARGINAL VERSUS AVERAGE FIXED COSTS

The New Americans (National Research Council, 1997) included a theoretical discussion of the relative cost of a new immigrant family in terms of its marginal cost to governments. However, most of that report’s estimates of household level state and local finances were based on allocating revenues and expenditures across existing immigrant and nonimmigrant households on an average cost basis; the same was true for that report’s treatment of federal spending (with the exception of national defense).

TABLE 9-9 Net Difference between State and Local Revenues and Expenditures per Independent Person Unit with Alternative Assignment of Education Expenditures (rounded to nearest $50), by Immigrant Generation, 2011-2013

NOTE: See text for construction of revenues and expenditures by state and generation.

SOURCE: Panel estimates implemented on the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

The evidence on the public versus private nature of government-provided services is mixed. Whereas total public spending no doubt increases with the size of the population, some categories of spending are likely to be unaffected, at least for a small increase in immigrant population and in the short run.

For the analyses in this chapter, about half of all spending (and revenues) is allocated based on personal or family attributes. But for many spending categories such as public safety, hospitals, and libraries, the costs have been allocated across all persons (both independent and dependent). Similarly, some revenue sources—such as transfers from the federal government for roads and those from natural resource extraction, which would be the same even if there were more new immigrants—are allocated on a per capita basis. While the panel did not specify which particular expenditures are public goods, it is important to highlight that some of these fixed costs are not higher due to the presence of immigrants.²⁵ The amounts of these fixed costs assigned to second and third-plus generation persons are lower than they otherwise would be due to the presence of more first generation arrivals as these costs become spread across a larger population. Not surprisingly, the implicit savings to nonimmigrants created by spreading fixed costs across a larger population varies with the population share in the first generation. For some communities, especially those facing declining populations, the influx of new immigrants can help lower their fixed costs. Indeed, for some costs, notably capital expenditures, bond repayments, and public pension obligations, the benefits of the government spending may have been received by earlier generations so having a larger population to pay off these debts benefits the existing population.

While not definitive, Table 9-10 highlights the difference in fiscal gaps that results from changing from an approach in which the fixed revenues

___________________

²⁵ Fixed costs are the part of expenses that do not change with the addition of another individual.

and fixed costs for public goods are allocated to all individuals to a marginal allocation in which they are allocated only to second and third-plus generation independent persons and their dependents. When these fixed revenues and expenditures are assigned only to second and third-plus generation independents and their dependents on a per-person basis, instead of being assigned evenly to persons from all generations—thus assuming a marginal amount of zero to first generation independents and their dependents—the negative gap in net fiscal impact between first and third-plus generation independent person units decreases (in absolute terms) from −$2,900 to −$450 (Table 9-10). Thus, part of the higher fiscal costs for first generation independent units found in most of the analyses in this chapter are from these fixed costs. Under the assumption that first generation independent person units do not bear these costs, net positive fiscal impacts decrease or turn negative for second and third-plus generation independent person units—and these cost increases are highest in the states with more immigrants. For the 15 jurisdictions with the largest percentage of their populations in the first generation, the fiscal cost gap between first and third-plus generation independent person units closes from −$3,400 to −$150. In terms of overall fiscal impact, in California, for example, if these fixed costs (and revenues) were only allocated to second and third-plus generation independent persons and their dependents, the state’s first generation independent person units change from generating a large net negative burden for the state to making a net positive contribution (going from generating a net cost of $2,050 to a net fiscal benefit of $1,050—about $400 less than that of third-plus generation independent units under a marginal allocation). As the share of the population that is composed of first generation independent person units declines, the impact of shifting from an average to a marginal allocation of these fixed revenues and expenditures diminishes.

Note that, if one were to only shift the fixed costs (and revenues) currently being borne by new immigrants who have arrived since 2006 (rather than all first generation individuals) to the remaining population (including other first generation individuals previously resident), the fixed costs for the rest of the population (both independent and dependent) would increase by about $50 per independent person unit; and, in most states, recent immigrants would provide a net fiscal benefit. Again, the size of the shift in costs depends on the number and makeup of recent immigrant families. For example, the increase in net fiscal costs for nonrecent first generation independent person units in California would be about $100. This alternative approach recognizes that, in many states, first generation independent persons are long-term residents of this country.

9.9 CONCLUSIONS

While previous chapters have highlighted the role immigrants play in affecting federal budgets and in their impact across state and local governments combined, it is important to recognize that the burdens and contributions to fiscal balance sheets vary tremendously across states.²⁶ Under the strictest set of assumptions, in which all costs of public education fall on the parents of those being educated and in which the cost of public goods are shared across the population equally, first generation independent person units are estimated to be the most costly relative to second and third-plus generation units. For the 2011-2013 period, first generation independent person units incurred a net cost on average of $1,600 per unit per year, compared to a net benefit of $1,700 for second generation independent person units and $1,300 for third-plus generation units.

Most states follow the national pattern in which units in the second generation contribute the most per unit due to slightly higher incomes and fewer average dependents, but this is not the case in California. Additionally, among the 15 states with the most first and second generation independent individuals, California has the largest difference, $5,150, between the fiscal shortfall of independent person units in the first generation (−$2,050) and the fiscal benefit of units in the third-plus generation ($3,100), while Maryland has the smallest difference at $650. In Maryland, independent person units in the second generation generate an even higher level of per-unit fiscal benefit ($2,050) than do units in the third-plus generation ($550), while in California, the positive fiscal impact of units in the second generation, at $1,550, falls short of that for units in the third-plus generation. Both states have progressive income taxes, and some of these differences appear to be related to Maryland having a larger percentage of first and second generation independent persons with more than a bachelor’s degree. In many of the states with the fewest first generation independent person units, the difference in relative contribution between units in the first and third-plus generations is negligible, while units in the second generation contribute more to a state’s bottom line.

The relative contribution or burden of any independent person unit is driven largely by that unit’s demographic and economic characteristics—most notably the number of dependents in the unit and the unit’s income levels. Because first generation units tend to have less income and more dependents than units in the second or third-plus generation, they are more costly to state and local governments. However, the children of

___________________

²⁶ Fiscal impacts also vary widely at substate levels. Ideally, our analysis would estimate impacts at city and county levels, as insights about local jurisdictional responsibilities and benefits are of great interest to those governments. However, for the kinds of analyses done here, it is not possible to analyze at the local level with the available data, due to sample size limitations.

TABLE 9-10 Net Difference between State and Local Revenues and Expenditures per Independent Person Unit with a Marginal Allocation of Fixed Revenues and Expenditures^a (rounded to nearest $50), by Immigrant Generation by State, 2011-2013

NOTES: Fixed revenue flows include other revenues and intergovernmental revenues (see Table 9-11 in the Technical Annex to this chapter for more information). Fixed expenditure flows include expenditures on other education and libraries, public welfare vendor payments to private vendors and administration expenditures, and other expenditures and capital outlays (see Table 9-12 in the Technical Annex to this chapter for more information). See text for more detail on the construction of revenues and expenditures by state and generation. Because the difference between first and third-plus generation net difference (revenue less expenditure) amounts is taken from the unrounded estimates and then rounded to the nearest $50, the value may differ from the first generation column less the third-plus due to rounding in some cases. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2). Caution should be taken when examining the state-level estimates, especially those near the bottom of the table, because of small first (and second) generation populations for many states.

^aThe marginal cost allocation of fixed expenditures in these estimates reassigns fixed revenues and expenditures to second and third-plus generation independents and their dependent children, rather than assigning them to all individuals (both independent and dependent) in all generations as in the average cost allocation in the baseline estimates (see Table 9-6).

SOURCE: Panel estimates implemented on the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

immigrants who are being educated grow up to become second generation adults, the group that, in general (but not always), contributes the most, when assessed in terms of independent person units, to a given state’s fiscal health. In addition, the age distribution of independent persons also affects the relative contribution they make as a unit (with their dependents) to a state’s budget. The share of the population that is elderly increases costs and decreases tax revenues to states. While not as costly as dependent children, the smaller share of first generation independent persons who are ages 65 and older offsets some of the costs for states, most notably in the form of Medicaid payments.

While the characteristics of individuals within an independent person unit affect the relative contribution or burden made by that unit, decisions made by the state and local governments about the level and structure of taxes and services provided also affect the relative burden or contribution of the unit. In places with higher spending on K-12 schools, for example, the relative cost of units in the first generation is higher than for units in the second or third-plus generation because the first generation units include more dependents.

The differences in contributions or burdens across generations and states also depend on whether fixed costs are allocated to all persons equally. The cost of an additional independent person unit in the first generation (or for that matter, an additional unit in any generation) is dampened to the extent that many of the costs that accrue to state and local governments are not sensitive to a small increase in the population. Using a marginal cost allocation, under which an additional immigrant is presumed not to add to the costs of administering the subset of state and local government services categorized as public goods, leads to more similar estimates of per-unit fiscal impacts across the three generations. The reason is that expenditures for the second and third-plus generation units increase, while those for first generation units decrease. In this respect, the cross-generation fiscal patterns are quite similar to those presented in Chapter 8 for the national level.

9.10 TECHNICAL ANNEX: SUPPLEMENTAL TABLES

This annex includes tables referenced in the text of this chapter but that are not included at the point of reference.

TABLE 9-11 Census of Governments (COG) State and Local Revenue Flow Types and Allocation Methods

^aVariable names reflect CPS data variable names used in the Integrated Public Use Microdata Series.

^bState sales tax amounts (prior to scaling to COG totals) come from the IRS Optional State and Certain Local Sales Tax Tables. We do not explicitly account for additional local sales taxes but expect them to be captured in scaling to COG totals. The one exception to this is Alaska, which has a statewide local sales tax but no state sales tax; in this case we use the IRS Optional Local Sales Tax Tables for Certain Local Jurisdictions.

TABLE 9-12 Census of Governments (COG) State and Local Expenditure Flow Types and Allocation Methods

^aVariable names reflect CPS data variable names used in the Integrated Public Use Microdata Series.

^bTotal Medicaid institutional spending (2% of total 2011-2013 COG expenditures) is taken from the Centers for Medicare & Medicaid Services 2015 report Medicaid Expenditures for Long-Term Services and Supports in FY 2013, appendix Table D. See http://www.medicaid.gov/medicaid-chip-program-information/by-topics/long-term-services-and-supports/downloads/ltss-expenditures-fy2013.pdf [November 2016].

TABLE 9-13 Annualized Weighted Sample Cases of Independent Persons by Immigrant Generation by State, Current Population Survey Annual Social and Economic Supplement for 2011-2013

NOTES: See text for definitions of independent person and immigrant generation. These sample counts are the average number of weighted cases classified as independent persons per year in the Current Population Survey Annual Social and Economic Supplement (ASEC) for 2011-2013. Note that these counts are not representative of the annualized total U.S. population in these years because they do not include dependent children. The ASEC includes cases in February, March, and April of each year. Because of the rotation group design, by which addresses are in the sample for 4 months, out for 8 months, and in again for 4 months, the total 3-year sample double-counts individuals who are in the sample in pairs of years (2011-2012 or 2012-2013). States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2).

SOURCE: Panel tabulations of the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

TABLE 9-14 Sum of Unweighted Sample Cases of Independent Persons by Immigrant Generation by State, Current Population Survey Annual Social and Economic Supplement for 2011-2013 Total

NOTES: See text for definitions of independent person and immigrant generation. These sample counts are the total number of cases classified as independent persons in each Current Population Survey Annual Social and Economic Supplement (ASEC) for 2011-2013. The annual observations for each state and immigrant generation are approximately one-third of the counts listed above. The ASEC includes cases in February, March, and April of each year. Because of the rotation group design, by which addresses are in the sample for 4 months, out for 8 months, and in again for 4 months, the three-year sample counts shown above double count individuals who are in the sample in pairs of years (2011-2012 or 2012-2013). States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2).

SOURCE: Panel tabulations of the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

TABLE 9-15 Average Number of Children (dependents) per Independent Person Unit, by Immigrant Generation by State, 2011-2013

NOTE: See text for definitions of independent person unit, dependent person or child, and immigrant generation. States are listed from highest to lowest percentage of first generation independent persons in the state’s population of independent persons (see Table 9-2).

SOURCE: Panel tabulations of the Current Population Survey Annual Social and Economic Supplement for 2011-2013.

TABLE 9-16 Average Adjusted Gross Income (AGI) per Independent Person Unit (rounded to nearest $50), by Immigrant Generation by State, 2011-2013