The 2020 Census: Potential Impacts on California

The 2020 Census: Potential Impacts on California MAC TAYLOR LEGISLATIVE ANALYST DECEMBER 13, 2018 analysis full gutter AN LAO REPORT LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT Executive Summary Federal Government Counts the Number of People in the United States Every Ten Years. The U.S. Constitution requires the federal government to count the total number of people in the United States every ten years. This is the decennial Census. The U.S. Census Bureau (“the Bureau”), a division of the U.S. Department of Commerce, will conduct the next Census in 2020. Census Population Counts Used to Determine Congressional Representation and Certain Federal Funding Levels for States. The results of the Census are used to distribute seats in the House of Representatives and inform the amount of federal funds allocated to states for certain programs. Census results also are used to determine legislative districts for federal, state, and local representatives within each state. Moreover, the Census provides a social, demographic, and economic profile of the country’s residents, informing decisions by policymakers and businesses across the state. California Has Experienced an Undercount in Past Censuses, Particularly in 1990. An undercount of California occurs when there are more people living in the state than are counted by the Census. While the Census has become more accurate over time, some past Censuses have resulted in significant undercounts of California’s population. For example, the 1990 Census undercounted California’s population by 2.74 percent (about 835,000 people), which was disproportionately worse than the national undercount—1.59 percent or about 4 million people. As a result, California gained one fewer seat in Congress than it was entitled to receive and was estimated to have lost over $200 million of federal funds in a single fiscal year. The 2000 and 2010 Censuses were much more accurate. California Could Experience an Undercount in 2020 . . . There are a variety of reasons to believe that California could experience an undercount in 2020. First, the state experienced greater undercounts than the rest of the nation in all of the past three Censuses. Second, the state has a higher share of people who historically have been more difficult to count relative to the rest of the nation (referred to as the hard-to-count population). This hard-to-count population includes immigrants, who may be even harder to count in 2020 than they were in the past. . . . But an Undercount Would Likely Be Smaller Than the 1990 Scenario. That said, we think that a statewide undercount equivalent to the 1990 undercount is not likely. There are a variety of reasons for this, including: (1) improvements in the Census’ methodology, (2) advances in access to statistical and administrative data, and (3) adequate funding levels for current Census preparation activities. (While past funding issues did limit some of the Bureau’s planning activities, Congress has granted the Bureau its full budget request for the most recent fiscal year.) California Unlikely to Lose a Seat Due to an Undercount . . . Other researchers have found that an undercount like the one that occurred in 1990 would not cause California to lose a congressional seat. Because we think even that scenario is unlikely, the risk of California losing a congressional seat is low. . . . Or Much Federal Funding. Some reports have indicated that California has billions of dollars at stake as a result of an undercount similar to 1990. Even in this worst-case scenario, however, we find that an undercount could result in the California state government losing tens of millions of dollars—not billions of dollars—in federal funding. In budgetary terms, this amount www.lao.ca.gov 1 analysis full gutter AN LAO REPORT of money is very small. (Federal funds distributed directly to local governments also could be affected, but the size of this effect is unknown.) Complete and Accurate Count in 2020 Still Important. There are other reasons to ensure a complete and accurate count in 2020. For instance, certain regions in the state run a greater risk of being undercounted than others. As a result, an undercount of some regions relative to others could result in some people within the state having less congressional (and state legislative) representation relative to what they should receive. 2 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT INTRODUCTION The U.S. Constitution requires the federal discusses these potential effects of an undercount government to count the number of people in in California. the United States every ten years—the Census. In this report, we first provide information on The Census is important, in part, because it is what the Census is. Second, we discuss how the used to determine each state’s representation in Census is conducted—including the steps the the House of Representatives and the amount federal government has taken to date and what of federal funds allocated to states for certain it plans to do over the next few years to conduct programs. Generally, a higher population count by the 2020 Census. (We also describe state efforts the Census means that a state will be eligible to to supplement these federal activities.) We then receive more seats in Congress and more funds. discuss the likelihood of an undercount in California In contrast, “undercounting” the population of a in 2020. Finally, we discuss the implications of state could result in fewer seats and less funds an undercount for California both in terms of relative to what the state should receive. This report representation in Congress and federal funding. WHAT IS THE U.S. CENSUS? U.S. Constitution Requires a Count of 2020. For instance, in 2013 the Bureau conducted Each Person Residing in the Country. Article I, a series of studies to examine the feasibility and Section 2 of the U.S. Constitution requires the effects of using new data collection methods in the federal government to count the total number of 2020 Census. In addition, the Bureau’s analysis and people residing in the United States, including publication of the final Census population counts citizens and noncitizens, once every ten years. will continue until 2021. (We discuss 2020 Census This decennial count is known as the Census. preparation and expected future activities in detail (In addition to residents physically located in the later in the report.) United States, the Census also includes military Why Is the Census Important? and civilian federal personnel and their dependents who live overseas.) The U.S. Census Bureau (the Congressional Representation. The U.S. Bureau), which is part of the U.S. Department of Congress—comprised of the Senate and House Commerce, conducts the Census. of Representatives—passes federal laws, raises Next Census Will Occur in 2020. The United revenues for the federal government through taxes, States conducted the first Census in 1790 and has and allocates those funds through the federal done so every ten years since. The next decennial budget process. While each state—regardless of Census—the nation’s 24th—will be conducted in the number of people who live there—has two 2020. In general, the Census collects information seats in the U.S. Senate, seats in the U.S. House on how many people live in a household, as well as of Representatives are apportioned based on their respective sexes, ages, races, and ethnicities. population. There are 435 seats in the House. The Bureau also uses the decennial Census as Under the Constitution, each state is guaranteed the basis for its annual population estimates of the one seat in the House and the remaining 385 seats nation, as described in the box on the next page. are apportioned based on the Census population Census-Related Activities Occur Over Many counts. Based on the 2010 Census, California has Years. While the counting of people in the United the largest delegation of Representatives in the States will occur in 2020, key planning, preparation, House with 53 seats—accounting for 12 percent of and operational activities for the 2020 Census the seats. (The second largest delegation is from began as early as 2013 and will continue past Texas with 36 seats.) As the definitive source of www.lao.ca.gov 3 analysis full gutter AN LAO REPORT Federal and State Population Estimates Affected by the Census The Decennial Census Informs Annual Population Estimates. The U.S. Census Bureau (the Bureau) produces annual population estimates for the nation, states, and counties for the years in between Censuses. In years the decennial Census is conducted, the Bureau’s population counts are “re-benched” to equal the final Census count. For years in-between decennial censuses, the Bureau adjusts the most recent Census counts using administrative records on the number of births, deaths, and net migration (and some other information) to calculate annual population estimates. The Bureau Uses the Decennial Census for Other Products. The annual population estimates from the Bureau serve as the population counts for other data products. For example, the Bureau also produces the American Community Survey (ACS) annually, which surveys demographic and economic characteristics of populations at various geographic levels. These products are commonly used by academics, governmental entities, research and policy institutions, and businesses to conduct studies and implement programs. For example, our office uses the ACS to analyze socio-economic and housing trends in California. State Entities Use the Decennial Census for Annual Estimates. At the state level, the Department of Finance (DOF) also uses Census population counts to re-bench its population estimates every ten years (although DOF has its own population estimates for the intervening years). DOF population estimates are a key input in a variety of state-level spending formulas and allocations. As a result, the Census also affects many aspects of state and local financing in California. population counts for purposes of determining the is entitled to three electors.) By determining the number of House seats allotted to California, the size of each state’s congressional delegation, the accuracy of the Census helps ensure that residents Census determines the level of representation each of California have appropriate representation in the state has in electing the President. federal legislative and budgetary process. (As we Federal Funding to the State. In some cases, discuss later, Census data also informs how these the Census population count affects how much seats are distributed within the state.) federal funds each state receives. That is, many Electors in the Electoral College. The U.S. federal statutes allocate funding to the states President is elected by the Electoral College. based on a formula that uses states’ populations— The Electoral College consists of 538 electors. as determined by the Census—as one variable to A presidential candidate must receive at least a determine their respective shares of the funding. In majority (270) of electoral votes to be elected to most cases, the higher a state’s population count, the Office of the President. Each state is entitled the more federal funding the state is eligible to to a number of electors equal to state’s number of receive. Similarly, certain federal funds provided House and Senate seats, meaning California has directly to local governments and nongovernmental 55 electors. (In addition, the District of Columbia entities are based on Census population counts. HOW WILL THE FEDERAL GOVERNMENT CONDUCT THE 2020 CENSUS? This section describes how the federal we outline recent actions the federal government government will conduct the 2020 Census. First, has taken to fund the Census. Finally, we describe we discuss the overall goals of the Census. Then, the Bureau’s timeline for completing the Census, 4 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT from pre-count activities that have already been unit, it would cost less than the 2010 Census, but completed, to the count of people in 2020, to the still maintain high-quality results. (The 2010 Census activities it completes post-count to finalize the cost the Bureau an estimated $120 per housing Census. unit. The Bureau set out to have the 2020 Census Goal of the Census Is to Get a Complete and cost $107 per unit, for a total estimated cost of Accurate Count. The Bureau’s goal for the 2020 $15.6 billion.) Census is to count everyone once, only once, and Typical Census-Related Spending Cycle. The in the right place. In other words, the Bureau strives federal government typically provides the Bureau to get a complete and accurate count. A “complete with funding during the early years of a Census cycle count” occurs if the Census counts the right (that is, early in the decade) to complete certain number of people in the state. An “accurate count” planning and preparation activities. As shown in occurs if the Census captures the characteristics Figure 1 (see next page), funding grows slowly of people correctly—including where they live and in these early years. It is not until the last three to other demographic information. Ideally, the 2020 four years of a Census cycle—when the Bureau Census will be both complete and accurate. (As we is beginning to open regional offices, hire Census explain later, the Bureau evaluates accuracy and workers, and visit households—that federal funding completeness with post-Census analyses using significantly ramps up. As Figure 1 also shows, to surveys and other methods.) date, funding throughout the 2020 Census cycle The Bureau Conducts Two Types of has, in nearly all years, been slightly higher in inflation Population Counts. The Bureau will complete and population-adjusted terms than for the last three two separate counts of people living in the United Census cycles. States as part of the 2020 Census: the household Recent Actions on Federal Funding to enumeration (a count of people living in houses, Support the 2020 Census. In the past couple apartments, and other dwellings) and a group of years, the funding allocated to 2020 Census quarters enumeration (a count of people living activities has been less than what the Bureau initially in facilities, like dormitories, military bases, and requested. For example, in federal fiscal year (FFY) prisons). This report focuses on the household 2016-17, the amount of federal funding provided enumeration, which accounts for about 97 percent for Census-related activities ($1.47 billion) was of the state’s total population. 10 percent less than what the Bureau requested ($1.63 billion). The 2017-18 federal spending bill Funding for the 2020 Census (passed in March 2018), however, nearly doubled the Bureau’s FFY 2016-17 funding level to a total of Census Has Made a Number of Changes to Improve Cost-Efficiency. Over the last several $2.8 billion. (The majority of this funding is expected decades, conducting the Census has become to support 2020 Census activities, rather than other more expensive on a per-person basis. There activities, like conducting annual population and are a variety of reasons for this cost increase: demographic surveys.) The Bureau has requested (1) self-response rates have declined (which the $3.8 billion for FFY 2018-19, a 37 percent increase Bureau believes could result from distrust of relative to FFY 2017-18 funding levels. Such an government or concerns about sharing information); increase in spending follows past trends of the (2) people move more frequently; and (3) more Bureau’s budget significantly increasing in the years people live in complex living arrangements (such leading up to the Census. What level of funding as multigenerational households or individuals the federal government ultimately will allocate to with ties to multiple residences), which makes it the Bureau in FFY 2018-19 and FFY 2019-20 to a challenge to associate a person with a single support future Census efforts is unknown. (When location. In light of these trends and direction this report was published, the federal government from Congress, the Bureau set out to create cost had yet to approve a final FFY 2018-19 funding level efficiencies for the 2020 Census. The Bureau for the Bureau.) sought to design the Census so that, per housing www.lao.ca.gov 5 analysis full gutter AN LAO REPORT status in the questionnaire. Figure 1 (The California Attorney General U.S. Census Bureau Spending Is and others have challenged Highest in Final Years of the Census Cycle the inclusion of the citizenship Per-Person Spending in Inflation-Adjusted Dollars question on the grounds that it could result in lower response rates $30 2010 Census among immigrants, resulting in an 25 incomplete population count that violates its constitutional purpose.) 2000 Census 20 The nearby box describes the history of the inclusion of a 15 citizenship question in the Census. 1990 Census Bureau Completed One and 10 Only “Dress Rehearsal” for 2020 Census 2018. Before the Census count, 5 the Bureau typically conducts several full dress rehearsals to 1 2 3 4 5 6 7 8 9 Census Year test its methods for reaching Years in a Census Cycle people who do not respond to the Dotted line represents requested, but not yet approved, funding level for federal fiscal year 2018-19. questionnaire on their own. The Bureau initially planned to do a dress rehearsal in three areas of Census Preparation Activities the country. As a result of budget limitations discussed earlier, the Bureau reduced This section discusses the activities the Bureau the number to one (although the questionnaire has conducted to prepare for 2020 Census. used in the dress rehersal excluded the question Bureau Has Compiled the National Address on citizenship). In 2018, the Bureau conducted that List. A couple of years before beginning to count dress rehearsal in Providence, Rhode Island. The people, the Bureau collects addresses for every Bureau has called the test a success: self-response known household in the country. This list is used rates were higher than expected (52.3 percent to mail census forms and follow up with people compared to 49.3 percent expected) and Census who do not respond to those forms. In past years, workers were able to count 1.56 houses per hour the Bureau relied heavily on field workers on the (higher than the rate in the 2010 Census of 1.05 ground to update the national address list. To houses per hour). These results, however, may not reduce costs for the 2020 Census, the Bureau be indicative of how quickly Census workers will has relied primarily on state and local government be able to count people in other cities and towns data, satellite imagery, and other administrative across the nation. Specifically, the dress rehearsal records to update its address list. (Although in was conducted in an urban area and enumerations 2019 the Bureau will conduct some on-the-ground will likely be slower in less densely populated areas. canvassing to finalize the address list.) The Bureau also plans to test response rates with Bureau Has Updated the Questionnaire. the inclusion of the citizenship question in the The Bureau has finished updating the questions questionnaire in 2019. contained in the Census questionnaire—the questions every household will receive as part of Census Count of Households the 2020 Census. As is required by law, the Bureau Bureau Contacts Every Household in Nation submitted to Congress its planned questions for the by Mail. In 2019, the Census will begin developing 2020 Census on March 29, 2018. These questions broad public awareness about the upcoming include topics on age, race, and sex. The Bureau Census, including through paid advertisements, also plans to include a question about citizenship 6 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT social media outreach, and events. In 2020, using In these follow-up visits, staff will go door-to-door its compiled address list, the Bureau will contact with mobile devices and tablets that they can use every household in the country by mail. Specifically, to collect the household’s data. In previous years, the Bureau will mail: to 80 percent of households Census workers visited every nonresponding an invitation to submit responses online and to household up to six times to complete their 20 percent of households a paper questionnaire surveys. In 2020, the number of follow-up visits a to submit through the mail. (The decennial Census household will receive will vary. The default number has never systematically used an online response of visits is still six, but it could be higher or lower option before, but the Bureau has allowed depending on two main factors: households to respond to the American Community • Availability of Administrative Data. If Survey online for several years.) The Bureau will ask the Bureau has complete and reliable households to respond by April 1, 2020: “Census administrative data available for a household Day.” Those who do not respond to the survey that does not respond to the initial mailed online after two mailed reminders will receive a questionnaires and reminders, a Census paper questionnaire. worker will conduct one follow-up visit to that After April 1, 2020, Census Workers Will home. If the household does not respond after Follow-Up With Nonrespondents in Person. that visit, the Bureau will complete the count From mid-May to mid-August 2020, a team of for that household using its administrative about 420,000 Census workers will conduct data. Administrative records that the Bureau in-person follow-up visits to households across the will use include: tax returns from the Internal country that did not respond to the questionnaire. History of the Census Questionnaire and Citizenship Question History of the Census Questionnaires. The Census has been conducted every decade since 1790. For most of this history, there was only one relatively short Census questionnaire that asked basic questions about overall household size and characteristics for each household member (race, gender, and age). Beginning in 1970, the U.S. Census Bureau (the Bureau) began using two questionnaires—a short-form and a long-form—so that it could ask a subset of households more detailed information. Five in six households received the “short-form” questionnaire, which continued to include basic questions on household size and characteristics. One in six households received the “long-form” questionnaire. The long-form Census included many more questions on various topics, such as the value of a person’s home or their monthly rent, number of rooms in a household, or a person’s marital status, occupation, and income. Recent Changes to the Structure of the Census Questionnaire. In 2010, all households received the short-form Census only. Instead of distributing a long-form Census, the Bureau instead collected detailed data on a small percentage of households throughout the decade in the American Community Survey (ACS). Effectively, this means most questions asked on the long-form question are now asked on the ACS instead. The 2020 Census will continue with this approach. History of the Inclusion of a Citizenship Question. Beginning in 1870 and running through 2000, a Census questionnaire included at least one question about citizenship. When the Census began using two separate forms (the short- and long-form) in 1970, a citizenship question was included on the long-form, but not the short-form, questionnaire. When the long-form questionnaire was replaced with the ACS in 2010, a question about citizenship status was included on the ACS instead. As such, a citizenship question was not included in the 2010 decennial Census. www.lao.ca.gov 7 analysis full gutter AN LAO REPORT Revenue Service, Medicare and Medicaid estimate how many people live in a household that enrollment information, Supplemental Nutrition never responded to the Census. Or, if a household Assistance Program records, and property tax completes the survey, but does not indicate the age and deed data. of one of the people living there, the Bureau can use • Whether the Household Is in a imputation to estimate the persons age. The 2010 Hard-to-Count Area. The Bureau has Census counted about 1.2 million people using designated certain areas as hard to imputation (less than 1 percent). count based on how many of that area’s Post-Count Activities households mailed back their 2010 census questionnaire. Households in areas known to Finalize the Count. After conducting the core be hard-to-count may receive more than the count operations described above, the Bureau six default attempts. will combine the results from the household enumeration with the group quarters enumeration. The Bureau may also increase follow-up The Bureau will then tabulate all of the data visits to certain households based on the test and release the results. The Bureau has until of the citizenship question. As they have in the December 1, 2020 to deliver apportionment counts past, after a Census worker follows up with a to the President and until March 31, 2021 to household three times, and no one has completed deliver the counts for redistricting purposes to the questionnaire, Census workers are allowed California and other states. Figure 2 summarizes to complete the count for that household by the main events in the 2020 Census activities, from interviewing a “proxy.” A proxy can be a neighbor, preparation to post-count. landlord, relative, caregiver, utility worker, or other After Finalizing All Enumerations, Bureau Will person who knows the household. Conduct Surveys to Evaluate Accuracy. After Bureau Will Use Statistical Methods to Impute the count is finalized, the Bureau will conduct two Data for Other Households. If Census workers major analyses to measure the accuracy of the are unable to complete the count for a household household enumeration. One approach compares the using all of the above methods, the Bureau can Census counts to data from surveys that the Bureau count the household using statistical imputation. completes after the Census. A second approach Imputation means the Bureau infers a household’s compares the count to administrative records. Both characteristics based on the characteristics of these methods examine how complete the overall of surrounding households. The Bureau uses count was and whether the count was accurate imputation to indicate whether or not a household is for some demographic groups. These surveys occupied; the number of people living there; and the are conducted only for informational purposes; characteristics of those people, like age and race. the Bureau does not update its decennial Census For example, the Bureau could use imputation to population counts based on these surveys. HOW ARE STATE GOVERNMENTS PREPARING FOR THE 2020 CENSUS? In preparation for the 2020 Census, a number of and promotional activities to supplement the states, local governments, and nongovernmental Bureau efforts, (2) established Complete Count entities have launched independent efforts to Committees, or (3) done both. (California has ensure an accurate and complete count. For done both.) Complete Count Committees engage this report, we focus on state-led and funded state and local community partners to distribute efforts. Figure 3 shows the 18 states that educational materials, educate the public on the have (1) allocated state funding for outreach importance of the Census, and conduct community 8 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT Figure 2 Major Events for the 2020 Census 2017 Began working with local governments to compile the national address list. 2018 Preparation Activities Updated the questionnaire. Conducted dress rehearsal. 2019 Open Census field offices, conduct in-field address canvassing, and launch advertising campaign. 2020 Contact every household by mail. Count April 1, 2020: Census Day Activities Households to return or submit their Census questionnaires by this date. Follow up with nonrespondents in person. Deliver apportionment counts to the President by December 1. Post-Count 2021 Activities Deliver counts for redistricting purposes to California and other states by March 31. Figure 3 State-Level Efforts To Ensure Complete and Accurate 2020 Census Count Established Complete Count Commissions Appropriations ME Both AK No Action VT NH WA MT ND MN WI MIa NY MA RI ID WY SD IA IL IN OH PA NJ CT OR NV CO KS MO KY WV VA MD DE CA UT NM KS AK TN NCb VA MD AZ OK LA MS AL GA HI TX FL a Michigan's Complete Count Campaign is run by a group of nonprofits, foundations, and other local community partners. b North Carolina’s budget proposal has yet to be approved. As of fall 2018. www.lao.ca.gov 9 analysis full gutter AN LAO REPORT outreach. (Some states have discussed—but California Has Set Aside $100 Million for 2020 have not adopted—funding proposals to support Census Activities. Initially, as a part of the 2017-18 the 2020 Census. Additionally, some states have Budget Act, the state provided $7 million in incentive expressed interest in establishing Complete grants for local governments to assist the Bureau Count Committees but have yet to do so.) in updating its address list and $3 million for other Figure 3 includes only state-level funding for the state preparation activities. In the 2018-19 Budget Census and excludes funding provided by local Act, the state provided an additional $90.3 million to governments or nonprofits. Below, we discuss past support the Complete Count Committee for outreach and current support California has provided for the activities related to the 2020 Census through decennial Census. 2020-21. On a per-person basis, the $100 million in California Has Provided Some Outreach funding California has provided for Census outreach in the Past. Beginning with the 2000 Census, is much higher than the funding level provided by California provided some level of funding to its any other state so far, as Figure 4 shows. Complete Count Committee. Similar to other states, California’s Figure 4 Complete Count Committee is California Has So Far Spent Much More on appointed by the Governor and 2020 Census Outreach Than Any Other State coordinates the state’s outreach (Dollars in Thousands) efforts (that supplement federal Total 2020 Population Spending outreach efforts). These outreach Amount Projection Per Person efforts focus on the hard-to-count population. For the 2000 Census, California $100,300 40,556,804 $2.47 the state provided $28.8 million Georgia 2,251 10,723,849 0.21 Illinois 1,500 12,802,052 0.12 to support the Complete Count Maryland 5,000 6,163,346 0.81 Committee (in today’s dollars). Michigan 500 10,007,985 0.05 For the 2010 Census, the state Minnesota 380 5,687,915 0.07 provided $2.3 million in today’s North Carolinaa 1,500 10,560,813 0.14 dollars. (The Complete Count Oregon 231 4,268,014 0.05 Committee also raised roughly Virginia 115 8,655,193 0.01 $10 million in private funding to Washington 464 7,675,985 0.06 augment its efforts for the 2010 a North Carolina’s budget proposal has yet to be approved. As of fall 2018. Census.) POSSIBILITY OF UNDERCOUNT IN 2020 CENSUS The possibility of an undercount occurring this section discusses the reasons an undercount in California has been widely discussed by the might or might not occur in 2020. Later in the Legislature, media, and broader public. (An report, we discuss in detail how an undercount undercount occurs when there are more people could affect the number of congressional seats living in an area than are counted by the Census.) the state is apportioned and the amount of money An undercount could negatively affect the state’s California receives from the federal government. share of certain federal funds and the number of California Can Experience an Absolute or congressional seats the state is apportioned. (An Relative Undercount in 2020. There are two types overcount, conversely, occurs when there are fewer of undercounts that California can experience that people living in an area than are counted by the are relevant. Either of these outcomes could have Census—which primarily occurs when the same implications for congressional seats and/or federal people are counted more than once.) The rest of funds. In particular, California can experience an: 10 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT • Absolute undercount if the Census count • Non-Hispanic Black residents were of the state’s total population is lower than undercounted by 2.07 percent. actual (meaning not everyone in the state is • American Indian people living on reservations counted). were undercounted by 4.88 percent. • Relative undercount if a larger share of • Renters were undercounted by 1.09 percent. California’s population is not counted relative • Very young children also were undercounted, to other states. (This could occur, for example, although to a lesser extent than these if all states experience an undercount, but the other groups. Children ages 0 to 4 were undercount in California is larger than average. undercounted by an estimated 0.72 percent. Or, it could occur even if the national count is (One reason very young children may largely accurate, but California experiences an be undercounted is confusion—people undercount.) completing the survey may not realize that the form should be completed for infants Past Census Undercounts and toddlers, as well as adults.) That said, in 2010, children ages 10 to 17 were California Experienced Relative Undercounts overcounted by 0.97 percent and on net all in 1990, 2000, and 2010. Figure 5 shows the children (people aged 0-17) were overcounted results of the post-Census surveys from 1990 to by 0.33 percent. 2010 and their level of accuracy. In each of the last three Censuses these surveys showed California Other Groups Are Also Difficult to Count. experienced both an absolute and relative There are other groups that are more difficult to undercount. In 1990, the U.S. experienced an count but have not been tracked in undercount undercount of 1.59 percent (considered large surveys specifically. These groups include: people by Census standards). In that year, however, who do not speak English fluently, lower-income the undercount of California was even larger at people, people who were born outside of the 2.74 percent. In 2000, the U.S. experienced an United States, and LGBTQ+ people. Additionally, estimated 0.49 percent overcount, but California’s people living in complex living situations—such population was still undercounted by an estimated as individuals with ties to multiple residences 1.52 percent. In 2010, the national count was roughly Figure 5 complete, but the count of California’s population was too California Experienced Relative low by 0.26 percent. Undercounts in Past Three Censuses Hard-to-Count Groups Have 1.0% Been Historically Undercounted Net Overcount at Higher Rates. Even when 0.5 the overall count has been largely accurate, the Census has -0.5 undercounted particular groups of people. These populations -1.0 are often called “hard-to-count.” -1.5 The Bureau’s post-Census Net Undercount -2.0 analyses have measured the likely United States undercount of some different -2.5 California groups. In 2010—when there was -3.0 no national undercount overall— 1990 2000 2010 the Bureau found: www.lao.ca.gov 11 analysis full gutter AN LAO REPORT and multigenerational households—can be California’s population is considered hard to count. particularly difficult to count. People experiencing Almost two-thirds (63 percent) of the state’s homelessness are also considered hard to population is Hispanic or non-white, which is close count, although they are generally counted in the to twice the national rate (39 percent). Additionally, group quarter enumeration, not the household a greater share of households in California live enumeration. The nearby box describes this in rental units compared to the rest of the nation process in more detail. (45 percent and 36 percent, respectively). Should there be an undercount of these hard-to-count Potential Reasons for an populations in 2020, the relative undercount in Undercount in 2020 for California California could be larger due to the demographics of the state. Moreover, because some areas of Below, we discuss key reasons why California the state have larger shares of hard-to-count could experience a different (likely higher) populations, an undercount of certain groups would undercount rate than most other states in 2020. impact some regions of the state differently. A Higher Share of California’s Population Is Administrative Data May Not Be Reliable for Hard to Count. States with a large hard-to-count Certain Hard-to-Count Populations. As previously population face a greater risk of an undercount. mentioned, one way the Bureau plans to reduce Relative to other states, a larger share of Group Quarters Enumeration The Census counts people who live in group quarters differently than people who live in households. Group quarters include jails, prisons, college dormitories, long-term care facilities (like nursing homes), and military bases. People who are experiencing homelessness at the time of the Census are counted in the group quarter enumeration either at a place they are receiving services or in non-shelter outdoor locations. About 1 Million People Counted in Group Quarter Enumeration in California in 2010. In 2010, about 8 million people were counted nationally in the group quarter enumeration—just over 10 percent of them, or nearly 1 million people, lived in California. Of these people, about half lived in institutions (in a correctional facility or nursing home, for example) and about half lived in noninstitutional settings like college dormitories and military bases. This population living in noninstitutional settings included people experiencing homelessness or those living in group homes. According to a separate report by the U.S. Department of Housing and Urban Development, about 133,000 people in California were experiencing homelessness on a single night in 2010. How the Group Quarter Enumeration Works. Rather than being counted individually, people are counted in group quarters collectively. There are three major ways the U.S. Census Bureau (the Bureau) will count people living in group quarters in 2020: (1) through in-field visits, (2) using administrative records, and (3) with responses from administrators of facilities. For example, to count some people experiencing homelessness, Census workers visit emergency shelters and non-shelter outdoor locations and conduct interviews with people staying there. State-Level Efforts for the Group Quarter Enumeration. In California, the Department of Finance, not the Complete Count Committee, will work with the Bureau on the group quarter enumeration. (However, the Complete Count Committee’s Housing Working Group is working on strategies for integrating these groups into outreach efforts.) Moreover, while states may coordinate information and assist the Bureau in identifying group quarter facilities, the actual in-field enumeration is done by federal Census workers. 12 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT nonresponse rates is by using administrative data. data, to the extent that they are households One challenge, however, is that administrative with more complex living situations than data is more accurate (and available) for certain surrounding households (such as larger in size or populations than others. For example, property multigenerational) they may still be undercounted. tax and deed records can help the Bureau The Census Bureau and Complete Count improve its information for homeowners. These Committee May Face Staffing Issues. Both records, however, would not capture renters. the federal and state governments will conduct Additionally, the Bureau acknowledges that while major Census-related operations and outreach in administrative records may accurately capture California in 2020. These efforts require hiring tens some hard-to-count groups (such as low-income of thousands of federal Census workers to help households receiving government assistance), it count people, as well as thousands of state and may underrepresent others (including noncitizens nonprofit staff to conduct outreach and community and recent immigrants who are less likely to be education. If the economy is still expanding, the low eligible for government assistance programs). With unemployment rate may make it difficult for both the Bureau’s more extensive use of administrative federal and state governments to hire enough staff. data in the 2020 Census for nonresponses, the In fact, a report by the Government Accountability Bureau may end up counting some groups very Office released in mid-2018 noted that the accurately, but still undercounting others. To the Bureau is already seeing signs it will face staffing extent that California has more of these populations challenges in a tighter labor market. The report than other states, there could be a relative notes that “in early hiring for 2020, Bureau officials undercount in the state. reported smaller than expected applicant pools, Immigrant Households May Be Even Harder declined offers, and turnover.” to Count in 2020. Immigrants are considered Possible Undercount Scenarios for a hard-to-count group, but in 2020, immigrant households may be even harder to count than they 2020 were in the past. In particular, there is concern In October 2018, the Public Policy Institute that households with undocumented immigrants of California (PPIC) published a report looking may be less likely to respond to the Census as at two undercount scenarios for the state in a result of the added citizenship question and/or 2020. In the first scenario, based on population due to concerns about confidentiality. These projections of different demographic groups, the changes could also affect response rates among researchers estimate the effect of an undercount immigrant households more broadly. Additionally, in 2020 similar to the undercount in 1990. (That as mentioned earlier, the Bureau has found that is, PPIC assumes the groups undercounted in immigrants are less likely to be represented in 1990 would be undercounted at similar rates in administrative data. While the Bureau can use 2020.) In this scenario, as shown in Figure 6, statistical imputation to infer characteristics of California would experience an undercount of immigrant households that do not respond to the 1.1 million people (2.8 percent). Under a second Census and for which there is no administrative scenario, the researchers estimate the size of a Figure 6 Estimates on Different Undercount Scenarios Baseline Undercount Similar to 1990 Immigration-Related Undercount Projected 2020 Number Percent Number Percent Population Population Undercount Undercount Population Undercount Undercount California 40,556,804 39,421,783 1,135,021 2.80% 40,040,275 516,529 1.27% United States 331,928,837 324,911,205 7,017,632 2.11 329,672,486 2,256,351 0.68 Note: Estimates from Public Policy Institute of California. www.lao.ca.gov 13 analysis full gutter AN LAO REPORT potential undercount if households with at least Court has ruled that the Bureau can use imputation one undocumented immigrant are undercounted by to complete the count.) The use of administrative 10 percent across the nation. Under this scenario, records and, potentially, imputation, would allow California would experience an undercount of about the Bureau to be aware of, and correct, any 500,000 people (1.3 percent). PPIC does not draw serious miscounts. As such, it seems unlikely an any conclusions about the relative likelihood of undercount of over 1 million Californians would either of these scenarios occurring in 2020. occur. That said, a scenario in which a subset of Size of 1990 Undercount Unlikely to Occur California’s immigrants and other hard-to-count in 2020. While we think a relative undercount of populations are undercounted is plausible. California in 2020 is plausible, it seems unlikely that it would Figure 7 be as large as the undercount Estimated Undercount in in 1990. First, researchers U.S. Census Has Declined Over Time have estimated that, nationally, the Census has become more 1.0% accurate over time, as shown Net Overcount in Figure 7. Second, while the Census experienced funding -1.0 issues that limited some of its -2.0 planning activities, the Census has now received its requested -3.0 Net Undercount funding for 2018 and has resumed -4.0 normal operations. Finally, technology, survey methods, and -5.0 statistical imputation techniques -6.0 have progressed substantially 1940 1950 1960 1970 1980 1990 2000 2010 since 1990, resulting in far more accuracy in annual population estimates. (The U.S. Supreme IMPLICATIONS OF THE COUNT FOR CONGRESSIONAL REPRESENTATION In this section, we discuss the potential requires the use of population counts from each implications of an undercount for California’s decennial Census to distribute (apportion) House representation in Congress. First, we discuss seats among the states. Historically, five different the implications of an undercount for the total methodologies have been used to determine how number of congressional representatives from to use the Census data to apportion congressional California. Second, we discuss the implications of seats. Since 1940, a mathematical formula has an undercount for the distribution of congressional been used to determine the priority of awarding an representation within the state. additional seat to each state. This methodology is referred to as the “Method of Equal Proportions.” Number of Congressional Essentially, the formula determines which state has Representatives From California the highest priority to receive an additional seat in the House until all seats have been assigned. Number of Congressional Seats Per State Determined by Formula. The U.S. Constitution 14 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT A State’s Population Growth Relative to received at least one more seat if the state’s count Others Determines Its Priority to Receive a was as accurate as the national count. Seat. With a state’s population—determined California Likely to Maintain 53 Seats if by the Census—being the only input into the Population Count Is Accurate in 2020. Whether formula, states effectively compete against one California loses, maintains, or gains a congressional another for seats based on their population. A seat depends on two factors: (1) the relative growth state’s population growth relative to other states of California compared to other states and (2) how determines whether a state gains or loses a seat. accurately that growth is reflected in the Census. Consequently, a state’s population growth in Estimates suggest that California’s population isolation—without considering the growth in other has increased since the 2010 Census—although states—cannot predict whether a state will lose, at a lower rate than other states since 2010. In maintain, or gain a seat. the PPIC’s analysis referred to previously, PPIC States Can Receive Fewer Seats Than They concludes that California will likely maintain its Should. Because apportionment is based on current number of seats (53), assuming that current relative population growth determined by the population trends continue and the 2020 Census Census, an undercount in one state relative to count is accurate. The states most likely to gain at other states can result in that state losing or not least one seat include Oregon, Arizona, Colorado, gaining seats that actually should be allocated to Texas (likely to gain three seats), North Carolina, it. For example, if the Census undercounts one and Florida (likely to gain two seats). state by more than others, it could fail to gain Scenario in Which California Loses a a seat that it should have gained. A state also Congressional Seat Seems Relatively Unlikely. could inappropriately gain—or fail to lose—a seat Based on the PPIC’s analysis, California would if the count is accurate there but undercounted have to experience a very significant undercount in elsewhere. Overall, an accurate population count order to lose a seat. PPIC estimates that California for all states is the best way for a state to avoid would lose one seat if it both experienced a similar potentially losing or not gaining seats that actually undercount to the 1990 Census and households should be allocated to it. with at least one undocumented immigrant were California Has Never Lost a Seat. For most of undercounted by 10 percent (the scenarios its history, California’s population has grown faster from Figure 6). In each of these scenarios alone, than other states, which has been reflected in California would maintain its current number of Census population counts. This relative population seats. This means that even a significantly large growth resulted in California’s representation in undercount, like the one that occurred in 1990, the House growing from 2 seats (out of 234 total would be insufficient for California to lose a seat. seats) in the 1850s to 53 seats following the 2000 While certain immigrant households may be Census. California has never lost a seat. In the undercounted at a higher rate in 2020 relative to case of the past apportionment (following 2010), past Censuses, whether the undercount rate will be California’s population had grown, but the state’s higher or lower than PPIC’s estimate is unknown. population growth was not sufficient relative to the However, as discussed earlier, we think an growth in other states for California to gain a seat. undercount in 2020 similar to the 1990 Census is Past Census Undercount Likely Resulted in unlikely. Because both scenarios would be required, California Not Gaining a Seat. The 1990 Census we think California losing a seat as a result of an was the least accurate of the past three Censuses— undercount is quite unlikely. it resulted in a net average undercount of Representation Across Regions 1.59 percent across all the states. The undercount Within State in California—at 2.74 percent—was worse than the national undercount. Although California gained Division of Congressional Seats Within seats following the 1990 Census (increasing from State Important for Representation. In 45 seats to 52 seats), California likely would have California, the boundary lines for the U.S. House www.lao.ca.gov 15 analysis full gutter AN LAO REPORT of Representatives districts, along with the state Commission Not Explicitly Barred From Legislature (Assembly and Senate) and Board Using Other Population Data for Redistricting. of Equalization districts, must be adjusted every During the last redistricting process in 2010, ten years following the decennial Census. This the Commission relied exclusively on Census process is known as “redistricting.” Because population counts to determine population levels in people move between Census years, the primary each district, which are intended to be reasonably purpose of redistricting is to reestablish districts equal in population. However, the Commission to be “reasonably equal” in population. California’s also solicited significant input from the public districts traditionally have been established by to gather information to ensure that the district using population counts from the Census. While boundaries adhered to other state constitutional it is important that California be apportioned requirements (such as ensuring communities of an appropriate number of seats in the House, interest were kept together). Although the U.S. distributing congressional seats within the state Constitution is clear that a state’s total number of in such a way that Californians have equal House seats can only be based on the Census, representation relative to other Californians also is there is no explicit requirement for states to use the important. Census population counts for redistricting. That is, Citizens Redistricting Commission if there were evidence some regions of the state Establishes Boundaries of Congressional were undercounted, the Commission would not and State Districts. In 2008, the voters passed be prohibited from using other population data to Proposition 11, establishing the Citizens inform the redistricting process. Redistricting Commission (the Commission). A Commission Would Need to Use Very High subsequent measure, Proposition 20 in 2010, Quality Data to Replace Census Data. That expanded the duties of the Commission. The being said, we are not sure what alternative data Commission is an independent, bipartisan would be reliable and comprehensive enough to commission that determines the boundaries for supplant or supplement the population counts the state’s districts. The Commission has only from the Census. (There are no publicly accessible drawn district boundaries once—following the population estimates that are as geographically 2010 Census. The Commission members who detailed and more reliable than the decennial will determine district boundaries following the Census.) One possible alternative source is the 2020 Census will be new members (the 2020 data from the Bureau’s detailed post-count surveys, Commission will be selected by August 15, 2020). some of which are public and some are not. We do Regional Under/Over Counts Can Skew not know what internal survey data the Bureau will Congressional Representation Within the collect after 2020 beyond those that are planned State. Whether or not California experiences a net to be released publically. Ultimately, it will be up undercount, the accuracy of the Census will vary to the Bureau to determine what level of detail to across regions within the state. Specifically, regions release. Once the 2020 Commission is selected (by with a larger share of hard-to-count populations August 15, 2020), if there were evidence that some likely will experience a higher undercount relative regions within the state were undercounted relative to regions with a smaller share of hard-to-count to others, the Commission would need to decide populations. (Meanwhile, some regions could how it would address the undercount and what experience overcounts.) In its report, PPIC alternative data source, if any, to use to inform the estimates that large undercounts are most likely to redistricting process. occur in portions of Los Angeles, the San Joaquin Valley, and the Bay Area. If these regions are significantly undercounted, they could have less congressional (and state legislative) representation relative to what they should receive. 16 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT IMPLICATIONS OF THE COUNT FOR FEDERAL FUNDING TO CALIFORNIA As Figure 8 shows, federal funds reach by population estimates. Some money reaches Californians through a variety of pathways. In Californians by passing through state and local 2017, our office estimated that the total amount governments or other entities—a portion of which of federal spending in California several years ago could be affected by population estimates. This was very roughly $375 billion. Most of this federal section of the report focuses only on federal funds spending (nearly 80 percent) is distributed directly that are distributed first to state government to individuals and would therefore be unaffected (and excludes funding distributed directly to local Figure 8 How Federal Expenditures Reach Californians Federal Government Payments to the Payments to Direct Payments to State Government Local Governments Individuals, Private Entities, and Universities State Governmenta Local Governmentsa,b Public (Cities, Counties, Universities and Special Districts) Private Entities Other Entities (For example, businesses, healthcare intermediaries, (For example, school districts, universities) Local Housing Authorities) Individuals a The state and local governments retain some federal money for program administration and operation. b Local governments remit some federal funds to private entities. www.lao.ca.gov 17 analysis full gutter AN LAO REPORT governments). In particular, we discuss why an $93 billion in federal funds for all state programs. undercount in California is not likely to have a large Most of this total, about $50 billion, is allocated impact on federal funding. to the state without reference to population How Federal Formulas Typically Use State estimates, and would therefore be unaffected by an Population to Determine Funding Levels. Some undercount in 2020. These funds are sometimes programs are distributed to the state based on a allocated to states on a fixed level defined in federal formula that considers population, among other rules or based on performance or cost-sharing. factors. (Other program funding is distributed For example, the federal government provides based on a competitive grant process or through California with a fixed annual amount of $3.7 billion performance metrics. For these programs, changes for the federal Temporary Assistance for Needy in estimated population would not have a direct Families program regardless of how population effect on federal grants to the state.) Most of these changes. Or, when California faces a natural formulas are set in federal statute. Consequently, disaster, the federal government provides hundreds this means that congressional action would be of millions or billions of dollars for emergency needed for the federal government to change these assistance and remediation efforts based on the funding formulas or allow the use of alternative costs to respond to the disaster, not the state’s population estimates. population. The remaining $43 billion in federal funds—for an estimated 51 federal programs—are Some Programs Rely on Counts of Specific based on formulas that take population estimates Populations. While some federal funding formulas into account, among many other factors. Although use total state population, others rely on regional these programs have federal funding formulas that or population counts of certain groups to inform factor in population, nearly all of this funding would funding levels. These funding formulas are typically not be affected by an undercount in the 2020 used for programs that target certain types of Census, as we explain below. areas, such as urban and rural areas, or certain populations, such as low-income families or Several Reasons Remaining Federal Funding children. For example, the Child Abuse Prevention Would Be Almost Entirely Unaffected by and Treatment Act funding is primarily allocated an Undercount. An undercount of the state’s to each state or territory based on its share of the population would leave the $43 billion in federal child population (under age 18) relative to other funding almost entirely unaffected. The reasons for states and territories. For some programs, these this are: targeted areas or populations are considered hard • An Undercount Would Have No Effect to count. on Federal Medicaid Funds Provided to Implications of Undercount for California. The state’s Medicaid program, Medi-Cal, represents roughly $35 billion— Overall Federal Funding nearly 80 percent—of the total 2016-17 After the undercount in 1990, California lost federal funding potentially affected by an over $200 million in federal funds in a single fiscal undercount. (This is the estimated amount of year. (For comparison, the state budget received federal funding for Medi-Cal that is informed roughly $25 billion in federal funds in state fiscal by the state’s population count—the state year 1990-91, so this loss would represent slightly receives additional funding for Medi-Cal that less than 1 percent of the total.) However, this is not informed by the state’s population.) loss of funds would be unlikely to occur again in As described in the nearby box, because 2020 even if an undercount were as large. Below, California already receives the minimum we explain why. amount under the current statutory formulas, Vast Majority of Federal Funding to the State it would be impossible for an undercount of Would Be Unaffected by an Undercount in California’s population to result in less federal 2020. In 2016-17, the state budget received about Medi-Cal funding. 18 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT • Population Is One Factor Among Many expenditure levels, and California’s funding in Federal Allocation Formula. For many level is already at that minimum level. For programs, population is one factor among example, after Medi-Cal and Title I education many in a complicated formula used to grants, the next largest federal funding program determine the amount of federal funding based on population is the special education provided to states. In these cases, changes in funding through the Individuals With Disabilities state population estimates have little effect on Education Act (IDEA). States’ IDEA grants the overall funding amount. For example, the are based, in part, on their respective shares second largest federal funding amount based of children ages 3-21. IDEA also includes in part on population is Title I education grants minimum thresholds, including a hold harmless (the state received $1.8 billion in these grants provision that prevents a state’s allocation in FFY 2016-17). These grants are distributed from declining year over year. In FFY 2017-18, in part based on the number of low-income California received $1.3 billion in IDEA grants, children that live in various parts of the already at a minimum allocation floor. As a state. As described in the box on page 20, result of this provision and the hold harmless a significant undercount of children could provision, even a significant undercount of indirectly affect this funding level, but the California’s children would be unlikely to result magnitude of the effect would likely be very in less IDEA funding for the state. low (perhaps a few millions of dollars). • Program Is Relatively Small. In other • Federal Allocation Formula Has Minimum cases, population may be an important Levels. In other cases, programs have determinant of formula-driven spending, but “allocation floors,” usually based on past the programmatic spending total is small. For An Undercount Would Not Affect Federal Medi-Cal Funding Medi-Cal Funding Based on the FMAP Formula. Medicaid—or Medi-Cal, as it is known in California—is a federal match program, meaning that program costs are shared between the state and the federal government. The federal share of costs for Medicaid is commonly referred to as the federal medical assistance percentage (FMAP) and varies by state. The FMAP cannot be less than 50 percent or more than 83 percent. The FMAP is calculated from a formula that takes into account the average personal income divided by the Bureau’s estimate of its total population of each state relative to the national average. States with per capita incomes above the national level receive a lower FMAP rate, while states with per capita incomes below the national level receive a higher FMAP rate. In addition to Medi-Cal, the FMAP informs the federal share of cost for other social services programs, including foster care and adoption assistance payments. California’s FMAP Expected to Remain at Federally Required Minimum Level. Using current data projections and the statutory formula, California’s raw federal fiscal year (FFY) 2019-20 FMAP would be 40.49 percent. Given that the FMAP cannot fall below 50 percent, California’s FMAP in FFY 2019-20 would remain at the minimum level. As a result, it would be impossible for an undercount to result in less federal Medi-Cal funding. For the state to receive an increase in its FMAP, it must experience a decline in its average per capita income, meaning either its population increases or total personal income decreases relative to the nation. For California’s FMAP to increase the Census would need to show that the state’s current population (39.8 million residents in 2018) increased by several millions of people in 2020, but personal income remained constant. Based on historic population growth trends (which have averaged in the hundreds of thousands per year) this scenario is nearly impossible. www.lao.ca.gov 19 analysis full gutter AN LAO REPORT example, 75 percent of state and community Figure 9 shows the likely effects of an highway safety grants are distributed based undercount on federal funding for the largest on population. California received $23 million ten programs we identified. Together, these ten for these grants in 2016-17, meaning a programs account for nearly 95 percent of the relative undercount of the state of 2 percent estimated total federal funds ($43 billion) that are could result in a few hundred thousand distributed to the state using population estimates. dollars less in funding—a very small amount in As the figure shows, most funding in these budgetary terms. programs would be unaffected—or only very slightly • Federal Allocation Formula Uses Annual affected—by a significant undercount. Population Estimates. Most programs that The State Could Lose Tens of Millions of receive federal funding based on population Dollars in Federal Funds From a Significant either use annual population estimates or Undercount. Other reports cite the entire annual demographic surveys, rather than the amount of federal funding distributed to the decennial Census, as population inputs in state from programs that allocate funds based their funding formulas. For annual population on population-related formulas as the amount estimates, the Bureau adjusts the most “at stake” as a result of a Census undercount. recent decennial Census counts based on the Only a very small portion of federal funding, number of births, deaths, and net migration, however, would be affected by an undercount. In as well as some other information. Through fact, we find that a significant undercount in the these adjustments, the Bureau has some 2020 Census (such as the one that occurred in flexibility to adjust its annual population 1990) could reduce federal funding to the state estimates to reflect new information about government by tens of millions, not billions, of underlying population counts. So, an dollars. This represents about 0.04 percent of total undercount could affect federal spending federal funding to the state government. Given that right after the decennial Census, but these we view a 1990 undercount as unlikely to occur effects would not likely persist throughout the again in 2020, this is likely a worst case scenario. decade. Title I Education Grants How the Title I Grants Work. Title I of the Elementary and Secondary Act of 1965 provides four main grants to local education agencies (LEAs). In federal fiscal year 2016-17, California received $1.8 billion in these four Title I grants—representing nearly 12 percent of the total funds distributed. The formulas that determine the amount of a state’s allocation are complicated, but one important contributing factor is each LEA’s count of children ages 5-17 who live in families with incomes at or below the poverty level. These grants also include hold harmless provisions that prevent a LEA from losing more than 15 percent of its preceding year’s Title I funds as a result of a drop in these populations. How an Undercount Could Affect Title I Grant Funding to California. The number of children living below the poverty level is estimated by the U.S. Census Bureau (the Bureau) in an annual survey. The Bureau will eventually use the decennial Census to inform these annual surveys—for example, to know how many people to survey from different areas of the country. As a result, an undercount of children in the decennial Census could indirectly—but probably minimally—affect the estimate of children living under the poverty level in each LEA. A large undercount of children in California relative to the United States could affect the Title I calculation, but from a state perspective, the effects would be very small in dollar terms. 20 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT Figure 9 Effect of an Undercount on Largest Ten Programs Distributed Using Population Estimates Funding at Risk Program in an Undercount Reason(s) Medi-Cal None State is funded at the minimum base FMAP. Children’s Health Insurance Program (CHIP) None State is funded at the lowest possible FMAP for CHIP. Title I Education Grants Low millions of dollars Population is one factor among many others. Special Education Grants to States Likely none Formula has minimum levels and California’s funding level is at one of those minimums. Child Care and Development Block Grant Low millions of dollars Population is one factor among many others. Vocational Rehabilitation State Grants Low millions of dollars Population is one factor among many others. Substance Abuse Block Grant Low millions of dollars Population is one factor among many others. Improving Teacher Quality State Grants (Title II) Low millions of dollars Population is one factor among many others. Social Services Block Grant Low millions of dollars Total federal funding for the program is relatively low. Low Income Home Energy Assistance Program Less than $1 million Population is one factor among many others. FMAP = federal medical assistance percentage. CONCLUSION California Unlikely to Lose a Seat or Much • A Misallocation of Representation Within Federal Funding Due to an Undercount. the State. If some regions within the state are California could experience an undercount in the undercounted relative to others, they could 2020 Census. For various reasons, an undercount end up with less congressional (and state as large as 1990, however, is unlikely. Past legislative) representation relative to what they researchers have found that an undercount of should receive. this magnitude is insufficient, alone, for California • A Misallocation of Funding Within the to lose a congressional seat. As a result, the risk State. Many local governments directly of California losing a congressional seat is low. receive federal and state funds that are Moreover, in a worst case scenario, an undercount based on population counts, which we do not could result in the state losing tens of millions of estimate here. As a result, the potential loss dollars—not billions of dollars—in federal funding. of funds due to an undercount may have a In budgetary terms, this amount of money is very greater impact on certain localities and their small. budgets relative to our state-level analysis. Complete and Accurate Count in 2020 Although, given that this issue has not been Still Very Important. There are other reasons investigated in depth, the potential effect on to minimize the possibility of an undercount and local governments is unknown. ensure an accurate count in 2020. As discussed earlier, certain regions in the state run a greater risk of being undercounted than others. As a result, an undercount of these regions could mean: www.lao.ca.gov 21 analysis full gutter AN LAO REPORT 22 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT www.lao.ca.gov 23 analysis full gutter AN LAO REPORT 24 LEGISLATIVE ANALYST’S OFFICE analysis full gutter AN LAO REPORT www.lao.ca.gov 25 analysis full gutter AN LAO REPORT LAO PUBLICATIONS This report was prepared by Jackie Barocio, Ann Hollingshead, and Nick Schroeder, and reviewed by Ginni Bella Navarre and Carolyn Chu. The Legislative Analyst’s Office (LAO) is a nonpartisan office that provides fiscal and policy information and advice to the Legislature. To request publications call (916) 445-4656. This report and others, as well as an e-mail subscription service, are available on the LAO’s website at www.lao.ca.gov. The LAO is located at 925 L Street, Suite 1000, Sacramento, CA 95814. 26 LEGISLATIVE ANALYST’S OFFICE