The Department of Homeland Security (DHS) released a new memo to curtail the Deferred Action for Childhood Arrivals (DACA) program. This program has allowed (previously) unauthorized immigrants who grew up in the United States since at least 2007 to stay and receive work authorization. The memo responds to the Supreme Court’s decision last month to block the government’s September 2017 attempt to wind down the program. Here are important facts to understand about this latest memo.

Same as 2017 rescission: As a practical matter, this memo accomplishes many of the same aspects of the September 2017 memo: stop new enrollees, stop travel authorization, and provide for a limited renewal. In September 2017, DACA enrollees had one month to apply to renew for up to 2 years (as has been the case throughout the program). This time, they may only apply for 1‑year renewals, but there is no explicit cutoff for when the renewals will stop being considered. A cynic might expect a determination about November 4, 2020, which would mean that current participants would be granted a much shorter “wind-down” than the last time.

New messaging: While it’s functionally the same, the new DHS memo does not describe itself as a rescission. It says it’s enacting “changes to the DACA policy” while (possibly temporarily) suspending certain aspects of it (new enrollees/​travel authorization). During this time, the government states that it will consider whether to permanently end the program. Again, a reasonable person would expect that the government will conclude this new evaluation sometime in November 2020.

New justifications: The September 2017 memo focused entirely on legal concerns with the policy. The “concerns” raised in this new memo are entirely policy focused. In other words, it’s claiming that DACA might just be a bad idea, even if it’s not illegal. This is a profound shift from the original DACA memo.

New position toward DACA: The government is now unsure whether it wants to end DACA. The government states that “fully rescinding the policy would be a significant administration decision that warrants additional careful consideration.” This is surprising to say the least. Since September 2017, the government has argued that it desperately wanted to end DACA to abide by the law, avoid litigation, and (later on) to further certain policy goals. Will courts accept this sudden reversal or will they conclude that the government is again trying to hide its real reasons for changing DACA?

New treatment of pending new applicants: In the September 2017 memo, the government stated that it would process all pending initial DACA applications (meaning for people who have never had DACA before). In this latest memo, the government will deny all such applications. This is incredibly unfair, particularly since these applicants will mainly be from immigrants who came at the youngest ages (including as young as one-year-old) because applicants needed to be at least 15-years-old in order to apply, so only the youngest applicants had no chance to apply before September 2017.

Ignoring the Supreme Court: On June 18, the Supreme Court had explicitly affirmed lower court decisions that stated that the government had to accept new DACA enrollees and travel authorizations, which it hadn’t done since September 2017. A Maryland district court then issued a ruling on July 17 requiring the government to abide by the Supreme Court’s decision. Yet the memo admits, “Since the issuance of the Supreme Court’s decision, DHS has, on an interim basis, generally held properly submitted initial requests for DACA in anticipation of potential policy changes.”

The same Maryland district court recently found that in fact, DHS was denying properly submitted new applications and was asserting erroneously on its website that such applications could not be accepted. In other words, it refused to issue DACA to these applicants, despite the Supreme Court and the district court orders. It now says that it will deny all these initial applications, which it claims are based on “the fortuity of when DHS received the request within a short period of uncertainty” (i.e. after the Supreme Court ruled in favor of those applicants). Will courts accept this? We will find out soon.

Considering but not deciding the issue: The government says it is “considering anew the DACA policy” as suggested by the Supreme Court. Yet it hasn’t actually fully considered the question or come to any actual conclusions about it, yet it is still making decisions to curtail or wind down the program. Will courts accept changes where the agency itself is saying it hasn’t fully considered the questions at issue? We will find out.

Suspect concerns: The policy “concerns” for suspending DACA are suspect. To begin with, DHS claims that DACA could cause more illegal immigration, particularly by children. But it provides no evidence for that because there isn’t any evidence. Indeed, as I have written before, the theory that DACA spurred a rush migrants to the border fails to explain any of the facts. DACA applicants are mainly from Mexico. The children came from Central America. DACA started in June 2012. The rush of children started in January 2012.

The memo asserts that DACA “may tilt the scales in deciding which aliens should receive deferred action.” Yet DACA doesn’t discriminate against non-DACA applicants for deferred action. It doesn’t undermine their applications in any way. This explanation is basically like saying that DHS objects to the DACA policy because it doesn’t want Dreamers to receive deferred action. It’s incoherent. Will courts accept a memo based on an incoherent claim?

DHS asserts against all common sense that “rescinding DACA entirely may well create a more pressing need for Congress to decide whether it wants to address this issue.” Not only do we now have ample evidence that ending DACA won’t force Congress to act, DHS’s memo—by laying out policy, not legal arguments against DACA—makes it less likely that Congress will act. This is because DACA opponents in Congress will now have more ammunition to make their case against action.

In the last week, the public has learned that the Department of Homeland Security (DHS) has deployed 2,174 Customs and Border Protection (CBP) agents to Portland, Oregon and other American cities to help combat rioting there. While there are certainly major problems with public safety and property destruction in some of these cities, which should require a robust and legal law enforcement response, there are also many reasons to be worried about the deployment of CBP agents. One of them is that there is very little oversight of CBP law enforcement officers (LEOs) and a long history of disciplinary, performance, and corruption problems in their ranks.

CBP didn’t have an internal affairs department that could investigate criminal misconduct until 2014, 11 years after it was formed in 2003. The old internal affairs department that managed the Immigration and Naturalization Service was attached to Immigration and Customs Enforcement, another sub-agency under DHS, and another wasn’t created for CBP until 2014. Internal affairs officers, Office of Personnel Management occupation code 1811, weren’t hired in CBP until early in the fiscal year of 2015 according to OPM data. Still, they number very few on the ground. 

In 2019, CBP had 60,525 employees, 48,035 of whom were LEOs. In the same year, only 255 of them were internal affairs officers. That’s one internal affairs investigator for every 188 LEOs in CBP. By contrast, the New York Police Department (NYPD) has one internal affairs investigator for every 65 sworn officers, and the older internal affairs division at Customs before the 2003 creation of DHS had one internal affairs investigator for every 136 employees.

Since CBP is now enforcing laws against American citizens in the interior of the United States, it should at least have the same ratio of internal affairs officers to LEOs that the NYPD has. That would mean increasing the numbers of 1811 internal affairs agents to 739, a 2.9 fold increase over the current numbers. CBP should have to hire 484 internal affairs officers to investigate criminal complaints and hold CBP LEO’s accountable for their actions.

The Department of Labor (DOL) requires that businesses that hire workers under the H‑2B program—America’s temporary worker program for seasonal nonagricultural jobs—pay a minimum wage known as the “prevailing wage.” This requirement substantially reduces the number of requests for H‑2B workers. Yet Congress has never explicitly required employers to pay the prevailing wage to temporary nonagricultural workers, and agencies who invented the rule did so initially without publicly explaining it, making its origins mysterious.

What history is known is lengthy, complex, and bizarre. Not only does the rule not exist in statute, DOL itself never shows up in the statute either. Stranger still, despite the H‑2B program tracing its origins to a 1952 law, no “prevailing wage” requirement existed in any formal regulation from any government agency for more than half a century. In 1952, the government proposed a prevailing wage rule, but then dropped it from the final regulations. Two years later, it formally revoked DOL’s regulatory authority over the nonagricultural program, yet the agencies continued to act as if this authority still existed for nearly a decade before the regulation was restored.

For 56 years, DOL determined the H‑2B prevailing wage entirely under informal guidance documents, many of which are not publicly available. Even when the documents were public, they often did not provide enough detail to understand how the wages were determined. In 1998, for example, it implemented a skills-based methodology for the prevailing wage it never explained publicly in any form for more than a decade. Amazingly, for several years in the 2010s, DOL issued prevailing wage determinations using this methodology in violation to court orders to stop.

Even now, DOL still calculates the prevailing wage under a methodology that directly contradicts the plain language of a 2004 law restricting the methodology that DOL can use for prevailing wage determinations for any visa program. Its justifications for using the prevailing wage are not obvious logical outgrowths from any part of the law and contradict congressional intent.

Overview of H‑2B Prevailing Wage History

Congress created the H‑2 program, the predecessor to the H‑2B program, in the Immigration and Nationality Act of 1952.[1] The Department of Labor (DOL) mandates the prevailing wage as part of the H‑2B labor certification process, a process that Congress never expressly created in the statute.[2] The original final Immigration and Naturalization Service (INS) regulations in 1952 ordering the creation of DOL’s H‑2 labor certification dropped a prevailing wage requirement proposed in the draft regulations.[3]

DOL failed to create any regulations at all for the H‑2 nonagricultural program in the 1950s.[4] Two years later, INS revoked the regulation requiring any kind of DOL involvement in the H‑2 program.[5] Yet DOL and INS continued to operate as if the regulation was still in force[6] until INS reintroduced the labor certification regulation in 1964.[7]

In 1966, INS required DOL to consider the effect of H‑2 nonagricultural workers on U.S. workers’ wages,[8] borrowing language from an unrelated statute enacted in 1965 to govern labor certifications for prospective permanent residents.[9] Incredibly, it came up with an entirely different process for H‑2 agricultural worker wages, despite the then-identical statutes. In 1968—16 years after the program’s creation—DOL created its first-ever regulations for H‑2 nonagricultural labor certifications, which also required considering the effect on U.S. worker’s wages but did not explicitly mention the prevailing wage requirement or how it would be administered.[10]

From the 1952 to 2008, DOL H‑2 nonagricultural prevailing wage calculations were determined by administrative guidance documents that were either not public or not widely available. At least since 1984 — [11] and probably since 1967 — [12] DOL required the same process for determining the prevailing wage for permanent labor certifications—again despite completely different statutes. Its methodology relied mostly on Davis-Bacon or Service Contract Act (DBA-SCA) wages based on DOL surveys where “prevailing” was primarily defined as 50 percent or, if not 50 percent, 30 percent of wages being the same identical rate in the area—effectively union wages.[13]

In 1986, Congress divided the H‑2 program into the H‑2B nonagricultural and H‑2A agricultural programs.[14] Only the H‑2A program had a requirement to consider the effects on the wages of U.S. workers,[15] yet INS and DOL continued to treat the statutes the same.[16]

Since probably 1977, whenever DBA-SCA wage was not available in the area, DOL required at least 95 percent of the average wage for jobs requiring “substantially similar skills” using surveys conducted by State Employment Service Agencies (SESA) or other published surveys.[17] In 1995, DOL explicitly required SESA surveys to provide for two wage levels based on skills.[18] In 1998, it decided that the Bureau of Labor Statistics (BLS) Occupational Employment Statistics (OES) surveys provided better data than SESA surveys and requested BLS create two skill levels mathematically, using average wages for the bottom third of the occupation as the lower skill level.[19] Yet DOL never publicly acknowledged this request or methodology until at least 2009.[20] Partly as a result of this change, H‑2B usage grew rapidly.[21]

In 2004, Congress mandated the creation of four wage levels based on education, skills, or experience whenever DOL uses “a governmental survey to determine the prevailing wage.”[22] DOL initially proposed H‑2B regulations to implement this rule, even while strangely asserting that the new law did not apply to the H‑2B program.[23] But before it finalized them, it issued a guidance document that implemented the rule, eliminated the mandatory use of DBA-SCA wages, and required that 100 percent, rather than 95 percent, of the average at a given skill level be paid.[24]

In 2008—56 years after the original H‑2 nonagricultural program was created—DOL finalized its first-ever prevailing wage regulation, but it refused to respond to comments on the wage methodology because it asserted it was just formalizing the existing 2005 guidance.[25] H‑2B labor certification applications peaked in 2007.[26] The refusal to consider comments led to legal challenge, culminating in a 2010 district court decision that vacated DOL’s wage regulation, finding that the 2004 law did not apply to the H‑2B program and that DOL could not lawfully consider H‑2B skill levels.[27]

In 2011, DOL finalized a regulation that eliminated wage levels, reimposed DBA-SCA wages, and banned employer-provided surveys.[28] Congress repeatedly blocked implementation of the rule,[29] and so DOL continued to use the court-vacated 2008 rule until 2013 when, after another court decision,[30] it issued a new rule that only eliminated the use of the four OES wage levels.[31] Employer-provided surveys using skill levels continued to be accepted until late 2014.[32] After yet another court decision,[33] DOL issued a rule in 2015 that created the current structure: 100 percent of the OES average wage for the occupation without skill levels, employer-provided surveys only where the OES wage is unavailable or using a state issued survey, and no use of DBA-SCA wages.[34] Later that year, Congress passed a law that required DOL to allow private wage surveys in any H‑2B prevailing wage determination (without actually requiring the prevailing wage). 

Early Pre-H-2B History

Congress created the H‑2 classification—the predecessor to the H‑2B—in the Immigration and Nationality Act (INA) of 1952. The original H‑2 encompassed both agricultural (now known as H‑2A) and nonagricultural temporary low-skilled jobs (now known as H‑2B). It defined an H‑2 worker as any alien “who is coming temporarily to the United States to perform other temporary services or labor, if unemployed persons capable of performing such service or labor cannot be found in this country.”[35]

“Other services or labor” was defined in contrast to the H‑1 category’s “labor or services of an exceptional nature” requiring “distinguished merit or ability.” In other words, H‑2 covered all low-skilled workers—on farms or otherwise. The INA further provided that the “question of importing any alien as a nonimmigrant under section 101 (a) (15) (H) … shall be determined by the Attorney General, after consultation with appropriate agencies of the Government, upon petition of the importing employer.”[36] DOL is not mentioned in the statute. The House report accompanying the legislation stated that the H visa’s purpose was:

to admit temporarily certain alien workers, industrial, agricultural, or otherwise, for the purpose of alleviating labor shortages as they exist or may develop in certain areas or certain branches of American productive enterprises, particularly in periods of intensified production.[37]

The Department of Justice’s Immigration and Naturalization Service (INS)—the predecessor to today’s immigration agencies under the Department of Homeland Security (DHS)—proposed regulations implementing the H‑2 program that would have required certification by a labor union that an H‑2 job is not “in a position to supply persons ready, able and willing at the then prevailing wage to perform labor or services of a like nature.”[38] This was the only reference to the “prevailing wage” in the proposed regulations, and the final regulation dropped the provision. Instead, it retained the following mandates:

If the alien is to enter under clause (ii) of [INA section 101(a)(15)(H)], there shall be… a clearance order bearing a statement from the United States Employment Service, that (i) qualified workers of the kind proposed to be imported are not available within the United States, and (ii) the Employment Service policies have been observed.[39]

Though it initially started in 1907 to help immigrants find jobs upon arrival, the U.S. Employment Service became a component of DOL that coordinates a system of State Employment Service Agencies, now called State Workforce Agencies (SWAs), that help unemployed Americans find jobs.[40] The 1952 INS regulations also specified that Mexican agricultural workers could enter as H‑2s in 1953, but DOL could veto any workers if it found sufficient U.S. workers or if they were “adversely affecting” the wages of U.S. workers—requirements enforced under the separate Bracero guest worker program at the time.[41]

After 1953, Mexican agricultural workers were excluded from the H‑2 program until the separate Bracero program expired in 1965.[42] In 1953, DOL promulgated Employment Service regulations for all agricultural workers—foreign or otherwise—that required employers who recruit through the Employment Service to offer the local prevailing wage.[43] In 1953, INS redrafted its H visa regulations to state that a clearance order was only required from DOL, “if requested by” INS.[44]

In 1954, the INS rewrote its H‑2 regulations again and dropped any recruitment or wage requirements or reference to DOL.[45] In 1958, it again rewrote the section, still excluding all DOL rules.[46] Nonetheless, INS never removed the instructions to seek DOL approval and follow Employment Service policies from its employer petition Form I‑129B,[47] and in 1960, DOL twice provided notices in the Federal Register claiming authority under INS regulations to certify that these two requirements were met, despite the asserted authority no longer existing in the cited regulations.[48] In 1963, DOL explained its H‑2 policies in congressional testimony as follows:

If an employer desires workers, the first thing he does is place an order with a local Employment Service office for domestic workers.… The Employment Service policies require that reasonable efforts to recruit have been made not only by the Employment Service but by the employer himself. They must recruit at the prevailing wage rate, and they must … offer conditions of employment that are prevailing with regard to transportation.[49]

Yet DOL cited as the basis of these requirements only its H‑2 rules pertaining only to agricultural workers.[50]

The End of the Bracero Program

Expecting the expiration of the Bracero guest worker program at the end of 1964, the INS reissued its H‑2 regulations with additional evidentiary requirements:

There shall be submitted a clearance order from the United States Employment Service concerning the availability of like labor in the United States which shall state that its policies have been observed.… A statement shall be furnished describing in detail the situation or conditions which make it necessary to bring the alien to the United States, and whether the need is temporary, seasonal, or permanent; if temporary or seasonal, whether it is expected to be recurrent.[51]

DOL subsequently responded to Bracero’s end by updating its regulations for H‑2 foreign agricultural workers, including the mandating of the recently invented Adverse Effect Wage Rate—an intentionally inflated prevailing wage.[52] Following the Immigration Act of 1965, INS revised its H‑2 regulations in 1966 to state that DOL must certify that qualified U.S. workers “are not available and that the employment of the beneficiary will not adversely affect the wages and working conditions of workers in the United States similarly employed.”[53]

This certification requirement is the same as the one adopted by Congress for permanent immigrant visa petitions in the 1965 act.[54] Congress had evolved this standard from the earlier 1952 standard, which had similar language, but prohibited visas or entry of immigrants only if DOL certified that they would, in fact, “adversely affect” U.S. workers.[55] In 1965, Congress flipped the presumption to prohibit immigrants coming to work until DOL certified that they would not “adversely affect” U.S. workers. Even though Congress made no changes to H visa requirements in the 1965 act, INS still adopted the same standard for the H‑2 visa.

In 1967, DOL revised its regulations to create uniform procedures for determining the prevailing wage for permanent labor certification requests.[56] DOL used the wage determinations or the wage formula borrowed from the Davis-Bacon or Service Contract Act (DBA-SCA), which require that government contractors pay the local “prevailing wage” for certain occupations. DOL defined “prevailing” to mean the rate paid to 50 percent or more workers or, if no such rate exists, the rate paid to 30 percent or more, or the average in the area. Whenever a single rate prevails among such large shares of workers, it is almost always the union wage.[57] To establish DBA-SCA wages, DOL conducted surveys that were routinely attacked as statistically inaccurate and biased in favor of easier-to-obtain union wages.[58]

Following this action, DOL finally introduced its first regulations specifically for H‑2 nonagricultural workers in 1968.[59] These required a finding that “qualified persons in the United States are not available and that the terms of employment will not adversely affect the wages and working conditions of workers in the United States similarly employed.”[60] Again, these are the same standards as for the permanent labor certification required by the Immigration Act of 1965. The agency would consider “the employer’s attempts to recruit workers and the appropriateness of the wages and working conditions offered.”[61] The prevailing wage requirement that was likely still being enforced was not mentioned. These new regulations were imposed to prevent Mexicans from jumping from the cancelled Bracero program to the H‑2 program.

In 1977, DOL updated its prevailing wage determination for permanent labor certification to require the use of DBA-SCA wages only if actually required in the occupation and area.[62] Otherwise, it required the offer be at least 95 percent of the average wage in the area and occupation so long as the job required “substantially similar level of skills”.[63]

In 1978, in regulations primarily focused on agricultural workers, DOL added additional general language applicable to both programs that spelled out how it would implement the prevailing wage:

Before any factual determination can be made concerning the availability of U.S. workers to perform particular job opportunities, two steps must be taken. First, the minimum level of wages, terms, benefits, and conditions for the particular job opportunities, below which similarly employed U.S. workers would be adversely affected, must be established. (The regulations in this part establish such minimum levels for wages, terms, benefits, and conditions of employment.)

Second, the wages, terms, benefits, and conditions offered and afforded to the aliens-must be compared to the established minimum levels. If it is concluded that adverse effect would result, the ultimate determination of availability within the meaning of the INA cannot be made since U.S. workers cannot be expected to accept employment under conditions below the established minimum levels. …

Once a determination of no adverse effect has been made, the availability of U.S. workers can be tested only if U.S. workers are actively recruited through the offer of wages, terms, benefits, and conditions at least at the minimum level or the level offered to the aliens, whichever is higher.[64]

The 1978 DOL regulations also interpreted the INS regulations as “adopt[ing] the same admission standards for nonimmigrant temporary foreign workers that the INA (8 U.S.C. 1182(a)(14)(A) and (B)) requires for certain immigrants who seek to perform permanent work in the United States.”[65] Again, this is the closest to a DOL regulation requiring the use of the permanent labor certification prevailing wage methodology.

In 1981, DOL clarified in publicly available training materials that for permanent labor certifications, adjudicators must base the prevailing wage, in the absence of DBA-SCA or union wages on the arithmetic mean of the wages paid other workers similarly employed in the area, using a local office survey or any current survey.[66]

In 1984, DOL published an administrative letter outlining its procedures for the H‑2 nonagricultural program in which it explicitly stated, for the first time, that DOL would deny job offers below “the prevailing wage for such employment in the local area” and further stated that adjudicators should use the permanent labor certification methodology for determining the prevailing wage.[67] Nothing in the document suggests that this was a new rule or method.[68]

It appears that neither DOL nor INS regulations or policies regarding H‑2 nonagricultural workers were ever contested in court, and indirect challenges to the H‑2 agricultural wage regulations focused on the reasonableness of their computation, not the underlying authority,.[69]

Congress Creates the H‑2B Program

In 1986, Congress passed the Immigration Reform and Control Act (IRCA), which separated the H‑2 program into the H‑2A agricultural program and the H‑2B nonagricultural program.[70] The law incorporated the existing H‑2 INS labor certification requirements into statute, prohibiting workers when U.S. workers are available or when H‑2As would “adversely affect” the wages of U.S. workers.[71]

The H‑2B language, however, was not touched: “aliens … coming temporarily to the United States to perform other temporary service or labor if unemployed persons capable of performing such service or labor cannot be found in this country.”[72] As a result, DOL just readopted the same regulations and policies as the H‑2 nonagricultural program, despite the fact that Congress did not adopt the “adversely affect” language for the new H‑2B program.[73]

In May 1995, DOL released an administrative letter regarding H‑2B prevailing wage determinations.[74] It stated that it is “well settled” that employers must pay DBA-SCA wages even if the job isn’t specifically subject to those laws’ requirements (e.g. a government contract), and it specifically barred wage surveys that rely on the median or modal wage or that focus on an average within an industry or certain type of employer.[75] But it allowed adjudicators to consider wage surveys furnished by the employer, so long as they were verified.[76] Importantly, the administrative letter specifically delineated two skill levels within each occupation for prevailing wage determinations: entry level or experienced level.[77]

In 1997, DOL decided, in the absence of DBA-SCA, or collectively bargained (i.e. union) wages, the expanded Bureau of Labor Statistics (BLS) Occupational Employment Statistics (OES) provided the most efficient and accurate source for permanent, H‑1B, and H‑2B prevailing wage surveys and created special procedures to allow DOL to quickly process its existing cases until the new system was ready.[78] In 1998, it issued another administrative letter, mandating the use of the OES in the absence of DBA-SCA, or collectively bargained (i.e. union) wages, unless the employer provided a different survey, for permanent, H‑1B, and H‑2B wage determinations.[79]

The most important decision was that starting with the 1998 administrative letter, DOL instructed BLS to calculate “entry level” wages as the average of the lowest paid 33 percent of workers since the OES didn’t provide wage levels based on skills for each occupation.[80] This instruction was not publicly disclosed at the time. Starting in the year 2000 through 2005, DOL also issued memos that clarified that most landscaping jobs were not subject to the Service Contract Act wage determinations, but rather OES wages.[81]

On December 8, 2004, the president signed the H‑1B Visa Reform Act of 2004, which provided:

(3)The prevailing wage required to be paid pursuant to subsections (a)(5)(A) [permanent], (n)(1)(A)(i)(II) [H‑1B visas], and (t)(1)(A)(i)(II) [E visas] shall be 100 percent of the wage determined pursuant to those sections.

(4) Where the Secretary of Labor uses, or makes available to employers, a governmental survey to determine the prevailing wage, such survey shall provide at least 4 levels of wages commensurate with experience, education, and the level of supervision. Where an existing government survey has only 2 levels, 2 intermediate levels may be created by dividing by 3 the difference between the two levels offered, adding the quotient thus obtained to the first level, and subtracting that quotient from the second level.[82]

On December 27, 2004, DOL updated its permanent labor certification regulations to eliminate the mandatory use of DBA-SCA wage determinations and provide for four wage levels without the earlier flexibility to pay as little as 95 percent of the calculated wage.[83]

A month later, DOL proposed a regulation to eliminate DOL labor certification and instead require employers to attest that they will pay the prevailing wage, enforcing the attestation with audits.[84] It also proposed to use the new DOL permanent labor certification prevailing wage standards, citing the H‑1B Visa Reform Act but asserting that the law actually did not apply to the H‑2B program (despite its general terms).[85] Instead, it based its decision on the “practice to treat prevailing wage determinations the same under the H‑2B program as under the permanent labor certification program.”[86]

On March 1, 2005, without finalizing the proposed rule, DOL issued guidance mandating the new permanent labor certification standards be used for all nonagricultural programs including H‑2B.[87]

H‑2B Prevailing Wage Regulations

On December 19, 2008, DOL finalized its first-ever regulation mandating the prevailing wage for the H‑2B program.[88] It used the same process that it was already using through the 2005 guidance. It allowed the optional use of DBA-SCA wages, mandated calculation of OES wages at different skill levels (implicitly four), pay the highest prevailing wage of all locations where a worker may go, use non-OES surveys at employer request, and use the median wage if that was all the employer-provided survey made available.[89] The four wage levels were based on the average of the bottom third of the wage distribution for Level 1, the average of the top two thirds for Level 4, and the formula described in the H‑1B Visa Reform Act to arrive at Levels 2 and 3.[90]

In 2009, an advocacy organization sued DOL. In an August 2010 decision, a Pennsylvania district court agreed that DOL violated the Administrative Procedures Act by failing to explain the basis of the four skill levels and failing to respond to comments on its use of the OES wage surveys, not DBA-SCA.[91] With respect to skill levels, the court—in a footnote—asserted that the 2005 law mandating the use of four wage levels in “a governmental survey” did not apply to the H‑2B program because the section header stated, “H‑1B Prevailing Wage Level,” even though nothing in the text of the statute limits it to H‑1Bs (and it still applies also to permanent labor certifications as well).[92]

In 2011, DOL finalized new regulations in which it re-mandated SCA/DBA wages in cases where they are higher than the mean OES wage and eliminated the four wage levels.[93] It asserted that the mean would be the only “skill level” because “there are no significant skill-based wage differences in the occupations that predominate in the H–2B program.” It stated that there was no basis for mathematically creating them, despite Congress’s explicit 2005 requirement to do so. The rule only cites the Pennsylvania district court’s decision to explain why it would ignore this law. DOL dismissed evidence that H‑2B industries have not, in fact, seen wage depression.[94]

Congress intervened to stop the rule from taking effect by defunding its implementation in DOL funding legislation four times, causing DOL to repeatedly delay the effective date of the rule.[95] DOL continued to use the 2008 rule. Then in March 2013, the same Pennsylvania district court that had vacated the 2008 rule implementing the four wage levels ordered that DOL cease using the levels within 30 days, despite the congressional action.[96] In April 2013, it issued a final rule that dropped the mandate to use DBA-SCA wages from the 2011 rule and required the average of the OES wages in the occupation without regard to skills, but continued to permit skill-based wage levels in other employer-provided surveys and allowed the optional use of SCA or DBA wage determinations. [97] DOL then delayed indefinitely the implementation of the 2011 rule.[98]

In 2014, however, Congress finally dropped DOL funding restriction on the 2011 rule. In December 2014, the Third Circuit Court of Appeals vacated the 2013 prevailing wage rule on the grounds that it allowed skill-based wages to be used in employer-provided surveys and that it allowed employer-provided surveys in violation of DOL’s stated conclusions.[99] DOL immediately stopped accepting employer-provided surveys.

The Third Circuit also acknowledged that the 2004 law “suggests that the four-tier methodology should be used” for H‑2B, but stated that “the L‑1 Visa and H‑1B Visa Reform Act amended only provisions of U.S. immigration statutes dealing with the L‑1 and H‑1B visa programs,” based on section headers. Again, this was untrue. The law explicitly reformed E visas and permanent visas as well, and the fact that it removed references to specific visa types in the provisions requiring four wage levels suggests that it should not be read to limit its scope even to those categories. The court concluded—in a footnote—that, despite its own stated view that the language “suggests” it should be used, the statute is actually “unambiguous” in not applying to the H‑2B program and not subject to agency discretion.

In March 2015, a federal district court in Florida determined that DOL entirely lacked the authority under the INA to issue H‑2B regulations independently from DHS, but it stayed the ruling pending new regulations from DHS.[100] In April 2015, DHS and DOL jointly finalized a new wage rule that used the OES average as the basis for prevailing wage and banned the use of SCA or DBA prevailing wage determinations—whether they were higher or lower.[101] Employers had increased their use of SCA or DBA prevailing wage determinations from just over 1 percent to 14 percent of cases from 2010 to 2015.[102] It also eliminated the use of employer-provided wage surveys not conducted by a state, except in areas where the OES did not report a local wage.[103]

DHS justified the removal the skill-based tiers by stating, in part:

Market signals such as labor shortages that would normally drive wages up may become distorted by the availability of foreign workers for certain occupations, thus preventing the optimal allocation of labor in the market and dampening increased compensation that should result from the shortage.[104]

Setting aside the simplistic and flawed economic analysis, this argument admits that DHS is using the prevailing wage to force labor shortages on U.S. employers. Yet labor shortages are exactly what the H‑2B program was intended to prevent. Indeed, the House Judiciary Committee report on the INA in 1952 literally stated that the H‑2 program was needed “for the purpose of alleviating labor shortages.”[105] DHS further asserted that it received no evidence that there are skill-based differences in pay in H‑2B occupations.

For the first time, DHS and DOL justified their creation of the prevailing wage in the 2015 rule by stating:

In order to meet the statutory obligations required under … INA section 101(a)(15)(H)(ii)(b), and to determine whether “unemployed persons capable of performing such service or labor cannot be found in this country,” an adequate testing of the U.S. labor market is necessary. Any meaningful test of the U.S. labor market requires that H‑2B petitioning employers must attempt to recruit U.S. workers at the prevailing wage and pay H‑2B beneficiaries such prevailing wages.[106]

This reasoning requires multiple leaps of logic. First, because a higher wage would marginally reduce the number of job offers, it requires reading into the statute the view that Congress only wanted unemployed Americans hired at a certain wage. Otherwise, it wanted them to remain unemployed. Nothing in the statute hints at that possibility. Congress indicated its concern about finding jobs for unemployed workers, not jobs at certain rates.

Second, even if this were the correct interpretation of the statute, there would still be no basis in the statute for imposing that wage on H‑2B workers. DHS reads into the statute the requirement that if unemployed Americans don’t want jobs at the prevailing wage, employers cannot hire H‑2B workers at any other wage. If U.S. workers reject, say, 100 jobs at the prevailing wage, the government could reasonable allow the employers to hire maybe 150 H‑2B workers at whatever wage the workers would accept. This would allow those firms to increase their production, hire more Americans in complementary positions, and benefit U.S. consumers with lower prices.

Nonetheless, a Florida federal district court upheld the 2015 prevailing wage rules in 2016.[107] H‑2B employers contended that because the law required DOL to consider “adverse effect” on U.S. workers’ wages only for the H‑2A program, while mandating DHS consider only the effect on the availability of jobs for U.S. workers for the H‑2B program, the prevailing wage rule was contrary to congressional intent. The court found that despite this difference, the long history of the H‑2 prevailing wage, the failure of Congress to stop its use, and agency discretion allowed it to implement the rule.

In December 2015, Congress included in the DOL appropriations bill a provision that required that H‑2B prevailing wage determinations be the higher of:

(1) the actual wage level paid by the employer to other employees with similar experience and qualifications for such position in the same location; or (2) the prevailing wage level for the occupational classification of the position in the geographic area in which the H–2B nonimmigrant will be employed, based on the best information available at the time of filing the petition.[108]

The law defines best available information to include statistically valid private wage surveys, even if an OES wage level is available. Beyond requiring the actual wage level, DOL issued a guidance document that interpreted the provision as essentially only affecting the 2015 rule’s prohibition on private wage surveys, not requiring them to accept private wage surveys based on skills. Private wage surveys can allow employers to pay the median wage if that was the only wage calculated by the survey.[109] Congress still did not mandate the use of the prevailing wage, but merely set requirements for any determinations that DOL decided to make.

Conclusion

The strange and complex regulatory history of the H‑2B program’s prevailing wage requirement highlights the importance of Congress offering clearer immigration statutes, and courts not allowing as much discretion to the administration to invent immigration requirements that Congress did not impose. However, a future administration will still have substantial latitude to improve the prevailing wage requirement for the H‑2B program.

It seems that everywhere you turn these days you’ll find someone in Washington lamenting the collapse of American innovation and industrial output, and, naturally, proposing his own industrial policy to solve the alleged problems. This includes both President Trump and Democratic challenger Joe Biden, both of whom have promised all sorts of subsidies, protectionism and procurement mandates intended to reinvigorate the American industrial base and restore U.S. innovation supremacy. What these plans have mostly failed to emphasize, however, is how freer markets – especially the liberalization of U.S. trade and immigration restrictions – might help to achieve key industrial innovation objectives (without messy and costly central planning) or how they’ve been harmed by past U.S. government restrictions.

A new NBER Working Paper from Wharton’s Britta Glennon adds to a growing body of literature showing just how wrong-headed the candidates’ omission of these free market policies might very well be. In particular, Glennon shows that past U.S. restrictions on high-skilled immigration (implemented through caps on H1‑B nonimmigrant visas) resulted not in an increase in hiring American workers but instead in a substantial offshoring of multinational corporation (MNC) jobs and R&D activities to these companies’ affiliates in more welcoming countries. Perhaps even more concerning, given recent events at home and abroad, Glennon shows that one of the biggest beneficiaries of these U.S. immigration restrictions was China, and that U.S. firms doing the most R&D offshored the most jobs.

Glennon’s conclusions are worth quoting at length (emphasis mine and citations omitted):

[F]oreign affiliate employment increased as a direct response to increasingly stringent restrictions on H‑1B visas. This effect is driven on the extensive and intensive margins; firms were more likely to open foreign affiliates in new countries in response, and employment increased at existing foreign affiliates. The effect is strongest among R&D‑intensive firms in industries where services could more easily be offshored. The effect was somewhat geographically concentrated: foreign affiliate employment increased both in countries like India and China with large quantities of high-skilled human capital and in countries like Canada with more relaxed high-skilled immigration policies and closer geographic proximity. These empirical results also are supported by interviews with US multinational firms and an immigration lawyer.

Despite the outsized role that multinational firms play in the economy – for example, US multinational firms are responsible for 80% of US R&D and employ about ¼ of US private employees – policy debates surrounding immigration have largely overlooked the fact that multinational companies faced with decreased access to visas for skilled workers have an offshoring option, namely, hiring the foreign labor they need at their foreign affiliates. This is the first paper to provide evidence that multinational firms do in fact utilize this option – both at the extensive and the intensive margin – and to examine the relationship between foreign affiliate employees and immigrants, in contrast to the relationship between immigrants and native-born workers.

The results have important implications for understanding how multinational firms respond to artificial constraints on resources and how they globally re-distribute those resources. The results also have important policy implications; the offshoring of jobs appears to be an unforeseen consequence of restricting skilled immigration flows. Even if H‑1B immigrants displace some native workers, any policies that are motivated by concerns about the loss of native jobs should consider that policies aimed at reducing immigration have the unintended consequence of encouraging firms to offshore jobs abroad.

The finding that skilled foreign-born workers will be hired at foreign affiliates rather than in the US also has important implications for the innovative capacity of the US. Skilled immigrants have been shown to have outsized impacts on innovation in the host country through spillovers. The spatial diffusion of these spillovers disappears with distance since innovative spillovers are geographically localized. From a nationalistic perspective, this is problematic; if skilled foreign-born workers are at a US firm’s foreign affiliate instead of in the US, the innovative spillovers that they generate will go to another country instead. Furthermore, the finding that immigrants often are not equally innovative outside the United States has even wider welfare implications. In short, restrictive H‑1B policies could not only be exporting more jobs and businesses to countries like Canada [me: and China], but they also could be causing the U.S.’s innovative capacity to fall behind….

Vice President Biden, to his credit, has elsewhere supported “expanding the number of high-skilled visas and eliminating the limits on employment-based visas by country” in order to boost “American innovation and competitiveness.” Hopefully the Team Biden folks writing the campaign’s immigration plans can talk some sense into their colleagues writing the industrial policy plans, and perhaps even explain how freer access to all global resources – whether high-skilled labor or essential goods like steel and machinery – can boost American companies’ innovation, output and global competitiveness while simultaneously denying potential adversaries those same advantages. As Glennon notes, multinational corporations that drive American R&D have other production options abroad, and they’ll use those options when misguided U.S. policies push them to do so.

Unfortunately, President Trump’s longstanding aversion to increased immigration or freer trade – and recent Trump administration efforts to restrict those things even further – provide little opportunity for similar liberalization hopes in the coming weeks or during any second Trump term. To paraphrase a great American poet, they’ll do anything to boost American innovation (or to counter China’s rise), but they won’t do that.

President Trump banned nearly all permanent immigration and temporary workers last month for the remainder of the year. His order will affect as many as half a million workers and immigrants if consulates reopen, other COVID-19 travel restrictions are lifted, and migration resumes to last year’s levels. It will also affect about 20,000 U.S. employers seeking to hire foreign workers over the next six months.

Tables 1 through 5 show the number of petitions that the government has approved for individual employers to sponsor workers for visas or status, the share of workers who fit into those categories who are outside the United States (and subject to the ban), the number of employers who received labor certification or labor condition application approvals from the Department of Labor (DOL) allowing them to petition for workers, and the number of employers who received approved petitions from U.S. Citizenship and Immigration Services (USCIS). The years reflect the most recent data publicly available.

The H‑1B category for high skilled workers represents the largest number of employers with 59,427 employers receiving labor condition application approvals and 29,012 employers receiving petition approvals for initial or new workers in 2019 (Table 1). Only 38 percent of their H‑1B initial hires were abroad. The H‑2B nonagricultural category had about 5,000 employers receive labor certification application approvals in recent years. The most recent year’s data for 2017 show 3,788 employers who received petition approvals (Table 2). Nearly all H‑2B workers came from abroad.

The employment-based second and third preference categories saw 924 employers receive labor certification approvals for workers abroad (Table 3), and 776 of them were eligible to petition for their employees to receive immigrant visas to travel here. The L‑1 category data are older but show 7,020 employers in 2018 receiving approvals for initial L‑1 petitions (Table 4). Almost 100 percent of L‑1s are abroad because the program exists to allow multinationals to transfer foreign employees to U.S. jobsites. The J‑1 Summer Work and Travel Program is even older—from 2015—but the size of the program has been similar in recent years. In 2015, nearly 16,000 employers hired J‑1s under the Summer Work and Travel Program (Table 5). Nearly all hires appear to have occurred abroad, but no public data on the share is available.

Using the employment data from the most recent years shows about 55,000 unique employers across all these programs. Assuming that the share of hires abroad reflects the share of employers who exclusively hiring workers abroad, a little more than 6 months would normally see a little less than 20,000 distinct employers submit petitions, sponsor, or hire for foreign workers in these categories.[1] 

Note that this doesn’t include a small number of employers of EB‑1 category for the highest skilled immigrant visa applicants. It also excludes the employers of a large number of J‑1 visa holders in programs other than the Summer Work and Travel Program. 

The United States government is harming its own industries, making the country less competitive and attractive for businesses who employ millions of Americans. This policy choice makes no sense and cannot be justified economically.

Last week, Immigration and Customs Enforcement (ICE) announced that it will terminate the visas of foreign students enrolled at U.S. universities that do not offer in‐​person classes. Prior to the COVID-19 emergency, ICE required foreigners on F‑1 student visas to be enrolled in a “full course of study” at a U.S. school where online classes could not account for more than a quarter of the total required courses.

On March 13, 2020, ICE responded to the COVID-19 pandemic by allowing foreigners on student visas to temporarily count all online classes toward a full course of study. This change in the rules was supposed to last for the “duration of the [COVID-19] emergency” according to ICE. On July 6, ICE rescinded the temporary emergency policy for foreign student even though they haven’t suspended the other so-called temporary immigration policy changes associated with the COVID-19 emergency, such as a near-total ban on immigration. See David Bier’s blog post for a clear description of the rule, how it was changed, and how it’s changing back.

In response to the rule change on July 6^th, over 200 universities are suing ICE. In response, ICE filed a brief that laid out of the government’s defenses and justifications. Most of ICE’s defense are uninteresting legal claims amounting to “we can do what we want,” but a few passages caught my attention. ICE claims that national security is jeopardized by allowing foreign students to attend online classes. Their brief reads:

A solely online program of study provides a nonimmigrant student with enormous flexibility to be present anywhere in the United States for up to an entire academic term, whether that location has been reported to the government, which raises significant national security concerns.

And

Over the past two decades, Congress has repeatedly stressed the importance of monitoring nonimmigrant students noting national security concerns.

The “past two decades” reference is to regulatory and legal changes in response to the 9/11 attacks in 2001. Although there are other national security concerns from foreign-born students such as espionage, the main concern is terrorism. What is the historical terrorist threat from foreign-born students?

From 1975–2017, 21 foreigners who entered on student visas murdered a total of about 159 people in attacks on U.S. soil. They accounted for 10.9 percent of all foreign-born terrorists during that time and 5.2 percent of people murdered in attacks committed by foreign-born terrorists. The annual chance of being murdered by a foreign-born terrorist who entered on a student visa is 1 in 72,838,750 per year during the 43-year period.

Of those 21 foreign-born terrorists who entered on student visas, only three managed to kill people in terrorist attacks on U.S. soil. Hani Hanjour was a 9/11 hijacker and is responsible for 156.8 murders (his share of the victims from those attacks). Eyad Ismoil and Rashid Najib Baz each murdered one person in attacks in 1993 and 1994, respectively. Hanjour is responsible for 98.7 percent of all the people murdered by terrorists who entered on student visas from 1975–2017. All three were motivated by Islamism.

The terrorist threat from foreign-born students is small. Furthermore, a regulation requiring them to take more classes in person rather than online won’t prevent a determined terrorist from committing an attack. If the goal of the rule is to keep track of students, there are cheaper ways to do that than to saddle students and universities with onerous regulations that are especially costly during a pandemic. Whatever the legal merits of ICE’s rule changes and change-backs are and the response by the universities, there is no good national security justification for restricting online instruction for those on student visas.

Economists Paolo Giuliano and Marco Tabellini wrote a new working paper titled “The Seeds of Ideology: Historical Immigration and Political Preferences in the United States.” Helen Andrews at the American Conservative summarizes the paper here with a blog titled: “How Early Immigration Shifted Our Politics Permanently To The Left.” The paper’s authors use a shift-share instrument to see how European immigrants in the early 20^th century affected public support for the Democratic Party and redistribution on the county level during the 2006–2018 period. They found that more European immigrants in the past led to more support for the Democratic Party and for redistribution on the county level today.

This new working paper complements Tabellini’s earlier paper in the Review of Economic Studies where he found that European immigrant populations on the city level, especially if they were from Eastern and Southern Europe, resulted in lower municipal tax rates, less tax revenue, and less government supply of services during the 1910 to 1930 period. According to this earlier paper, a one-standard deviation increase in the immigrant population (about five percentage points) reduced per capita government spending by 5 percent and property tax rates by 7.5 percent on the city level. This is especially important back then because most government services were supplied locally.

Looking at the papers together presents the full picture. Giuliano and Tabellini wrote:

While there is no statistically significant relationship between the 1910–1930 fraction of immigrants and the Democratic vote share until 1924 (included), the coefficient abruptly spikes in 1928, when it becomes strongly positive and highly statistically significant.

We posit that 1928 was the first national election when immigrant-heavy counties swung to support Democrats because it was the first national election after the National Origins Quota Act permanently cut legal immigration from Europe (the Emergency Quota Act of 1921 limited immigration, but many thought it was a temporary post-war emergency measure). Closing the border removed the politically-effective counter-argument that more welfare would attract immigrants who will take advantage of new government programs. Closing the border thus freed both natives and immigrants to vote for more welfare without the fear that immigrants would come from Europe to consume the new benefits. A relatively open immigration system reduced support for redistribution and big government, but the removal of that political check resulted in increased public support for those programs.

Tabellini’s two papers are not contradictory, but the second seems like a very serious challenge to the research we’ve conducted on how immigrants don’t increase the size of government (not Tabellini’s goal, by the way). However, the two papers are perfectly consistent with the theory that Powell and Nowrasteh lay out in their forthcoming book: Government doesn’t grow as fast when the borders are open and government grows more when the borders are closed.

Many prominent economists who support more redistribution believe that such programs are only politically possible after the borders are closed. They have a lot of history on their side. The role of government in American domestic life surged only after Congress enacted the restrictive immigration laws of the 1920s. To be sure, the Great Depression and other ideological shifts in public opinion were necessary for big changes in American policy during the New Deal, but they were politically possible because closed immigration removed the politically-effective immigrant welfare-queen argument.

Prior to the closure of the border in the 1920s, American progressives and other supporters of a large welfare state near-uniformly supported restricting immigration because they saw it as a political impediment to their goals (and for other less-savory reasons like eugenics). Even Karl Marx, Friedrich Engels, and their American communist followers warned that immigrant-induced diversity reduced worker solidarity and that that would slow Marxist efforts to stoke a socialist revolution in the United States or achieve middle-ground policies like the creation of a large welfare state and mass unionization.[1] Across the developed world today, support for welfare, redistribution, and government provision of public goods is inversely correlated with the share of the population that is foreign-born and diverse. The early Marxists worry was well-founded.

There is a big difference between Tabellini’s papers that bears pointing out. Tabellini’s first paper looks at the effect of immigration on actual spending and taxation policies on the city level. His second paper looks at how immigrant populations in the past influenced public opinion decades later. That is an important difference. The decline in local government provision of services and rise of state and federal spending since the New Deal means that we can’t replicate Tabellini’s first paper today and get meaningful results. Looking at public opinion is worthwhile and interesting, but it’s a different measurement.

Below, we briefly look at the relationship between real federal outlays per capita and the immigrant share of the population at the national and state level. Although useful, looking at the national-level relationship can only tell us so much. In the next section, we look at the relationship between changes in real state-level outlays per capita and the immigrant share of the population to see if there is a relationship between growth in immigration and growth in government spending.

Real Federal Outlays Per Capita and Immigration

Figure 1 tracks real federal outlays as a percent of GDP and the immigrant share of the population. There is no statistically significant relationship between the variables, but the coefficient is negative. There simply isn’t a positive relationship between government outlays and immigration from 1928–2017. If one includes 1900–2017, the relationship becomes strongly and significantly negative whereby a 1 percent increase in the immigrant share of the population is correlated with a 1.3 percent decline in government spending as a share of GDP.

More importantly, what happened to federal outlays during the Golden Age of Immigration Restrictions? From 1921–1968, U.S. immigration law radically reduced annual inflows such that the percentage of the population that was foreign-born fell from 13 percent in 1920 to 4.8 percent in 1968 (Figure 2). The book end dates are the year of the enactment of the Emergency Quota Act of 1921 and the year the Immigration Act of 1965 went into effect. When immigration was most restricted, federal outlays as a percent of GDP increased from 6.9 percent to 18.9 percent, an increase of almost 174 percent. Congress created the New Deal and Great Society programs after the reduction in legal immigration caused the share of the foreign-born population to decline. In other words, the argument that immigrants would come in to take welfare benefits was not politically salient when the borders were closed. From 1968 to 2017, when immigration was slightly liberalized, federal outlays as a share of GDP rose from 18.9 percent to 20.4 percent, an increase of a mere 7.9 percent.

Real State Outlays Per Capita and Immigration

As mentioned above, it wouldn’t be useful to extend Tabellini’s earlier work on immigrants and city-level spending in the early 20^th century to today because the state and federal governments supply goods that used to be supplied by cities. In the previous section, we looked at the correlation between federal spending as a percent of GDP and immigrant shares of the population. The national-level analysis is only suggestive because there is no United States with a different immigration policy that we can use for comparison. Thus, we turn to the relationship between changes in state-level spending and the immigrant population shares over time.

We look at state-level decadal panel data for state budgets and immigrant population shares over the period 1960–2010. Data on state government expenditures come from the Government Finance Database (GFD), which compiles government financial data collected by the U.S. Census Bureau and harmonizes the data for state, county, and local governments.[2] GFD data include detailed public financial data on revenues and expenditures broken down by type and function. Specifically, we focus on four variables of interest:

1. General revenues: State revenues from all sources, including federal grants and any other intergovernmental revenue sources.
2. General revenue from own sources: Revenues from state taxes, fees, returns on state-run insurance schemes, and revenues from state-owned/regulated enterprises like utilities and liquor stores.
3. Total expenditures: All state expenditures such as wages, capital outlays, construction expenditures, and intergovernmental transfers.
4. Direct expenditures: All state expenditures such as wages, capital outlays, and construction expenditures.

Using intercensal data, we convert the expenditures into per capita terms and adjust for inflation using the personal consumption expenditures (PCE) price index.[3] Data on the foreign-born share by state are taken from the 1960–2000 decennial Censuses and the 2008–2012 American Community Survey from IPUMS.[4] These data include the total population and foreign-born stock for each decade for each state for 1960–2012.[5]

Our objective is to test the correlation between changes in states’ foreign-born population and the growth in state government spending. Leveraging the panel structure of our data, we estimate fixed effects regressions that regress the log difference in per capita expenditures on the foreign-born share. In each model we control for state and region-by-year fixed effects. This implies that the correlation between spending and the foreign-born share is estimated from changes in the foreign-born shares within a state over time relative to other states in the same region and year.

Tabellini and others note that immigrant settlement patterns may reflect other local factors that are uncaptured by fixed effects. For instance, immigrants may settle in states that provide larger safety nets and that might appear to show that they grow the size of government while, in fact, they might just settle where it is already biggest. Similar to the approach used by Tabellini, we use a shift-share instrumental variable (IV) to partially address the endogeneity concerns.[6] In particular, we construct an instrument that apportions national immigrant inflows into the United States to states according to the shares that settled in each state in 1960 using this method.

Government Spending and Immigration, 1970–2010

Figure 3 shows the correlation between changes in real state-level per capita total expenditures and changes in the state-level foreign-born population share. Figure 4 shows the correlation between changes in real state-level per capita direct expenditures and changes in state-level foreign-born population share. Both figures show a negative correlation between expenditure growth and growth in the foreign-born share of the population.

To test whether these negative relationships are statistically significant, we run a series of fixed effects IV regressions. In each regression we instrument a state’s foreign-born share using a shift-share instrument and find negative correlations between state government spending growth and the foreign-born share (Table 1). The first column of Table 1 shows the first stage of our IV regressions. As a measure of our instrument’s strength we report a first stage F‑statistic of 29.8, which is a strong instrument.[7] Columns 2 through 5 show correlations between state revenue growth, spending growth, and foreign-born shares.

The dependent variable in Columns 3 through 5 (Table 1) is the log difference of each per capita measure. The coefficient for each specification relates a percentage point increase in a state’s foreign-born share to a percentage change in the level of per capita revenues or expenditures. Our simple regressions show no significant correlation between both growth in general revenue and revenue from the state’s own sources, controlling for fixed effects. Only the estimate for total expenditure growth is significant at the 5 percent level. Column 4 shows a significant, negative correlation between a state’s foreign-born share and total government expenditure growth of about 1.3 percent. The estimate for direct expenditures in Column 5 is negative and only marginally significant.

Taken together, our regression results suggest that a larger share of immigrants at the state level is correlated with slower growth in state per-capita government expenditures. While these regressions attempt to control for endogeneity and other unobserved factors, the results are more suggestive than causal. Establishing a causal link between immigration and government spending or revenues would require additional empirical work outside of a blog post.

Conclusion

How immigrants affect the size and growth of government is a complex issue with many moving parts. Immigrants have tended to vote for the Democratic Party or its predecessor since the 1790s. When the Democratic Party was the laissez-faire party, immigrants voted for it. When it became the interventionist party, immigrants continued to vote for it. They did so because the Democratic Party has been more pro-immigration than its competitors during most of American history.

With some exceptions, people tend to choose a political party first and then change their opinions to match that party’s platform. Public opinion certainly has a huge influence on public policy as our system is designed to turn voter preference into policies, but the empirical work above indicates that more immigration is not related to growth in the size of state governments. The reactions of native-born voters to immigrants complicate the story significantly as immigrants don’t merely shift the median voter by being present but also by influencing the opinions of natives on myriad policy issues. Much of the work on this topic shows no relationship or a negative relationship between the two.