## _wp08244

## Source details

**Canonical URL:** [_wp08244](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08244.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08244.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08244.pdf.json)

---

### Research question and motivation
- Primary question: What determines U.S. immigration policy today, and specifically, are political-economy factors (industry-specific interest groups) important in shaping immigration to the United States?
- Motivation: migration literature focused largely on supply factors; demand side (host-country migration policy and political determinants) has received little systematic empirical attention.
- Timeframe and scope: industry-level analysis covering the period between 2001 and 2005.

### Data and identification
- New U.S. industry-level dataset combining:
  - Number of visas across sectors (2001–2005).
  - Measures of political organization:
    - Workers’ union membership rates across sectors from the Current Population Survey.
    - Firms’ lobbying expenditures targeted to immigration from Center for Responsive Politics (CRP) semi-annual reports (1998–2005 compiled; analysis uses 2001–05 averages).
- Lobbying measurement (three-step procedure):
  1. Consider only firms that list “immigration” as an issue.
  2. Split each such firm's total expenditure equally among the issues they lobbied for.
  3. Aggregate firm-level expenditures on immigration across firms in the industry.
- Alternative/robustness lobbying measure: total lobbying expenditure of firms who list migration as an issue (no issue-splitting).
- Dataset scope and coverage:
  - 2001–05 unbalanced panel of 12,376 firms/associations; 481 list immigration as an issue in at least one year.
  - Industry coverage: 120 three-digit industries (1950 Census Bureau classification).
  - Visas covered: H1A, H1B, H1C, H2A, J1, O1, O2, P1, P2, P3, R1; L1 and H2B not available by sector.
  - Time coverage used: averages over the five years between 2001 and 2005 (cross-sectional variation exploited).

### Theoretical framework and prior evidence
- Modeling approach:
  - “Protection formation function” linking lobbying expenditures by pro- and anti-migration groups to policy restrictiveness (quota q_i).
  - Two lobbies per sector: pro-migration (capital owners) and anti-migration (workers/unions).
- Comparative static predictions:
  - Higher organized labor expenditures/union organization in a sector → higher protection (more restrictive policy) → fewer immigrants in that sector (ceteris paribus).
  - Higher organized business expenditures in a sector → less restrictive migration policy → more immigrants in that sector (ceteris paribus).
- Historical/anecdotal evidence cited:
  - Organized labor historically engaged to limit foreign inflows (Chinese Exclusion Act of 1882; AFL involvement in 1917 Act; AFL-CIO and 1986 IRCA).
  - Business lobbying examples: Technet and H2R visa creation after Maryland seafood industry lobbying (amendment to Tsunami Relief Act (P.L. 109–13) of May 11, 2005).

### Main empirical findings (industry-level, 2001–05 averages)
- Dependent variable: number of visas divided by number of native workers in the sector (visas/native workers).
- Key explanatory variables:
  - log(lobbying expenditure on migration / native workers) — political organization of capital.
  - union membership rate — proxy for political organization of labor.
- Preferred point estimates (industry-level elasticities and effects):
  - A 10 percent increase in industry's lobbying expenditures on migration per native worker is associated with a 2.9–4.4 percent larger number of visas per native worker (core IV/robust estimates).
  - A 1 percentage point increase in the union membership rate is associated with a 2.6–5.5 percent lower number of visas per native worker (core IV/robust estimates).
  - OLS illustrative magnitudes reported: a 10 percent increase in lobbying expenditures per native worker raises visas per native worker by 3.6 percent; a 1 percentage point increase in union density reduces visas per native worker by 2.6 percent.
- Statistical significance and explanatory power:
  - Lobbying coefficient significant at 1 percent in key OLS specifications.
  - Union membership coefficient significant at 10 percent (OLS) and stronger in IV.
  - The two key variables explain 14 percent of cross-industry variation in visas per native worker in one OLS specification; lobbying alone explains 11 percent.

### Endogeneity, instruments, and IV results
- Endogeneity concerns: potential reverse causality and measurement error for lobbying and union variables; IV strategy employed.
- Instruments used:
  - For lobbying on migration:
    - Industry-level lobbying expenditures by firms that do not list migration as an issue (log(lobbying exp on other issues/native workers)).
    - Sector concentration: variance of firm size (annual payroll) within a sector (U.S. Census County Business Patterns).
  - For union membership rate:
    - Union densities across industries from the United Kingdom (cross-country correlation exploited).
- First-stage performance:
  - First-stage F values reported: 62.66 and 40.61 (reported in regression (1) first stages); other reported first-stage F values include 47.22 and 16.38 depending on specification.
  - Table 4b: log(lobbying exp on other issues/native workers) coefficients ~1.017*** [0.077] and 1.036*** [0.090]; variance of firm size jointly significant (p-value = 0.00).
  - Hansen test for overidentifying restrictions satisfied (e.g., p-values reported 0.97 and 0.73 in Table 4a).
- IV estimates (high-level):
  - IV magnitudes on lobbying and union membership increase relative to OLS (possible reasons: attenuation bias, negative correlation between lobbying and unobservables, or positive correlation between unionization and unobservables).
  - Core IV summary: 10 percent more business lobbying per native worker → 2.9–4.4 percent more visas per native worker; 1 percentage point higher union membership rate → 2.6–5.5 percent fewer visas per native worker.

### Robustness checks and heterogeneity
- Robustness exercises reported (appendix tables):
  - Controls included: output, industry-specific unemployment rate, price of good produced, domestic capital, foreign direct investment, negative origin-country shocks, lagged U.S. wages.
  - Results robust to inclusion of these industry-level controls; coefficients on lobbying and union membership remain similar in sign and significance.
  - Alternative dependent variables:
    - Excluding J1 visas, restricting to visas with quotas, and focusing on H1B approvals produce similar results.
    - Example: log number of H1B visas approved regression: 10 percent higher lobbying per native worker associated with 1.8 percent larger number of H1B visas per native worker; 1 percentage point higher union membership associated with 3.6 percent lower number of H1B visas per native worker.
  - Alternative lobbying measure (upper-bound): total lobbying expenditures of firms that list “immigration” as an issue yields similar estimated impacts.
  - PAC contributions vs lobbying:
    - PAC contributions account for about 10 percent of targeted political activity; lobbying expenditures about 90 percent.
    - PAC-based political-organization coefficients are weak or not significant; when both are included, lobbying expenditures (not PAC contributions) drive the positive effect on visas.
  - Additional robustness:
    - Dropping agriculture, pooled vs averaged data with year fixed effects, restricting observations, including zeros in lobbying (log(0) replaced by minimum positive value) — findings remain very similar.
  - Skilled- vs unskilled-intensive split:
    - Absolute magnitudes larger for skilled-intensive sectors, though differences not statistically significant.

### Mechanisms, policy instruments, and interpretation
- Channels of influence:
  - Congressional lobbying for statutory changes (e.g., H1B cap changes in 1998, creation/expansion of visa categories).
  - Administrative/regulatory lobbying to influence labor certification, wage rules, and hiring procedures at executive agencies (e.g., Department of Labor).
- “Visible” vs “invisible” barriers:
  - Visible: explicit congressional quotas (H1A, H1B, H2B historically).
  - Invisible: regulatory requirements that reduce program utilization (e.g., residency requirements for foreign medical doctors; advertising and Adverse Effect Wage Rate requirements for H2A).
- Policy interpretation:
  - Political-organization forces at the sector level materially affect allocation of immigrants across sectors.
  - Targeted lobbying can create new visa categories or expand access for specific industries (example: H2R introduction after seafood industry lobbying).
  - Using PAC contributions alone risks misleading inference because PACs are a small share of targeted political activity and weakly correlated with immigration-specific lobbying.

### Key descriptive statistics and correlations (selected, 2001–05 averages unless noted)
- Non-immigrant visas (2001–05 averages):
  - Total non-immigrant visas issued on average per year: 5,735,577.
  - Work and related visas average per year: 835,294.
    - Temporary workers classified: 315,372.
  - Selected category averages (2001–05):
    - Exchange visitors and spouses/children J1, J2: 291,241
    - Workers with extraordinary ability O1, O2: 8,865
    - Internationally recognized athletes or entertainers P1, P2, P3: 32,762
    - Cultural Exchange and Religious Workers Q1, Q2, R1: 10,414
    - Treaty traders/investors and their children E: 35,282
    - Spouses/children of certain foreign workers O3, P4, Q3, R2, I: 21,469
    - NAFTA Professionals and spouses/children TN, TD: 2,124
    - Intracompany transferees and spouses/children L1, L2: 117,765
    - Temporary workers total: H1B 130,051; H1A, H1C 122; H2A 31,322; H2B 72,684; H3 1,518; H4 79,675
    - Other admissions total: 4,900,283
      - Temporary Visitors B1, B2, B1/B2: 4,154,485
      - Official representatives and transitional family members A, G, K: 165,141
      - Students and spouses/children F1, F2, M1, M2: 266,099
      - Other non work visas: 314,558
    - Total non immigrant visa issuances: 5,735,577
- Lobbying and political-activity magnitudes:
  - Lobbying expenditures for immigration (mean): 99,811 (US$) per industry per year (when split among issues).
  - Total expenditures by firms in a sector that lobby for immigration: 1,084,469 (US$) per year (mean).
  - Top-spending industries on immigration lobbying (2001–05): Engineering and computer services; Educational services; Hospitals; Food and related products; Office machines and computer manufacturing; Agriculture.
  - Lobbying accounts for close to 90 percent of targeted political activity; PAC contributions represent about 10 percent.
  - Aggregate political activity figure cited: 3.8 billion U.S. dollars per political cycle on targeted political activity, which includes PAC.
- Cross-industry variation and correlations:
  - 97.4 percent of variation in log(visas) is across industries (little within-industry time variation).
  - Correlation statistics:
    - Corr(log lobbying for immigration, log number of visas) = 0.316 (robust s.e. = 0.080; p-value = 0.000).
    - Corr(union membership rate, log number of visas) = -2.283 (robust s.e. = 1.220; p-value = 0.063).
    - Corr(log lobbying for immigration, log number of H1B visas) = 0.255 (robust s.e. = 0.068; p-value = 0.001).
    - Corr(union membership rate, log number of H1B visas) = -2.180 (robust s.e. = 1.161; p-value = 0.062).
- Summary statistics (selected):
  - Log(Number of visas as a ratio of native workforce): Obs 120, Mean -7.86, Std. Dev. 1.84, Min -11.87, Max -1.34.
  - Log(Lobbying expenditures for immigration as a ratio of native workforce): Obs 120, Mean -3.03, Std. Dev. 2.05, Min -9.41, Max 1.56.
  - Union membership rate in the US: Obs 120, Mean 0.12, Std. Dev. 0.12, Min 0.00, Max 0.47.
  - Number of visas: Obs 120, Mean 2,101, Std. Dev. 10,593, Min 2, Max 102,836.
  - Lobbying expenditures for immigration (US$): Mean 99,811; Std. Dev. 215,278; Min 0; Max 1,513,108.
  - Total lobbying expenditures by all firms which lobby for immigration (US$): Mean 1,084,469; Std. Dev. 2,363,724; Min 14; Max 17,800,000.
  - Contributions by PACs (US$): Obs 127, Mean 6,741,956; Std. Dev. 4,856,364; Min 174,543; Max 28,700,000.

### Policy-relevant implications and suggested extensions
- Political organization matters for sectoral immigration outcomes:
  - Business lobbying targeted to immigration expands sectoral access to immigrant labor.
  - Union strength and worker organization reduce sectoral access to immigrant labor.
- Policy design and reform debates should account for industry-level political-economy forces when evaluating changes to visa rules and quotas (targeted lobbying can create new visa categories or expand access in specific industries, as illustrated by H2R).
- Suggested extensions and data improvements:
  - Explore alternative measures of immigration policy (legal vs illegal, temporary vs permanent).
  - Examine variation by occupation and geography (e.g., across U.S. states).
  - Use firm-level lobbying data to study political-economy effects on other policies (trade, environment, taxes).

*Source: IMF working paper content (section and appendix listings and excerpts) covering 2001–2005 industry-level analysis of lobbying, unionization, and visa issuances.*

### 1.   Number and Types of Non-Immigrant Visa Issuances, 2001–05......................................11

### 1.   Number and Types of Non-Immigrant Visa Issuances, 2001–05......................................11

### Research question and motivation
- Primary question: What determines U.S. immigration policy today, and specifically, are political-economy factors (industry-specific interest groups) important in shaping immigration to the United States?
- Motivation: The migration literature has focused largely on supply factors; the demand side (host-country migration policy and its political determinants) has received little systematic empirical attention.
- Timeframe and scope: Industry-level analysis covering the period between 2001 and 2005.

### Data and identification
- New U.S. industry-level dataset constructed by combining:
  - Number of visas across sectors (2001–2005).
  - Measures of political organization:
    - Workers’ union membership rates across sectors from the Current Population Survey.
    - Firms’ lobbying expenditures targeted to immigration policy from a dataset developed by the Center for Responsive Politics (allows identification of lobbying by targeted policy area).
- Rationale for data choices:
  - Lobbying expenditures constitute 90 percent of targeted political activity; PAC contributions represent only 10 percent and cannot be disaggregated by issue.
  - Ability to link lobbying expenditures to immigration policy is presented as a substantial improvement over prior work relying on PAC contributions.

### Theoretical framework and prior evidence
- Theoretical prediction (summarized):
  - In a given sector, more politically organized labor (higher union organization / expenditures) leads to higher protection and therefore fewer immigrants.
  - Conversely, more politically organized business (higher lobbying expenditures for immigration) leads to less restrictive migration policy and therefore more immigrants in that sector.
- Historical and anecdotal evidence cited:
  - Organized labor historically engaged to limit inflows of foreign workers (e.g., Chinese Exclusion Act of 1882; AFL in 1917 Immigration Act; AFL-CIO on illegal immigration culminating in 1986 Immigration Reform and Control Act).
  - Business lobbying examples: Technet advocating relaxation of migration policy; creation of H2R visas after industry lobbying (example: seafood industry lobbying Senator Barbara A. Mikulski leading to amendment to Tsunami Relief Act (P.L. 109–13) of May 11, 2005; H2R visas have the same requirements as H2B visas but no quota if individual held H2B in one of previous three fiscal years).

### Main empirical findings
- Both pro-migration (business) and anti-migration (labor) interest groups play a statistically significant and economically relevant role in shaping migration across sectors.
- Preferred point estimates (industry-level effects):
  - A 10 percent increase in the size of lobbying expenditures by business groups per native worker is associated with a 2.9 percent larger number of visas per native worker.
  - A 1 percentage point increase in union density (example: moving from 10 to 11 percentage points, which amounts to a 10 percent increase in union membership rate) reduces the number of visas per native worker by 3.2 percent.
- Robustness:
  - Results are robust to inclusion of industry-level control variables (e.g., output, prices, origin country effects).
  - Endogeneity concerns addressed using an instrumental-variable estimation strategy; robustness checks are reported (appendix references).

### Policy-relevant implications
- Political organization matters for sectoral immigration outcomes:
  - Business lobbying targeted to immigration expands sectoral access to immigrant labor.
  - Union strength and worker organization reduce sectoral access to immigrant labor.
- Policy design and reform debates should account for industry-level political-economy forces when evaluating changes to visa rules and quotas (e.g., targeted lobbying can create new visa categories or expand access in specific industries, as illustrated by the H2R case).

*Source: IMF working paper content (section and appendix listings and excerpts) covering 2001–2005 industry-level analysis of lobbying, unionization, and visa issuances.*

### Section VI. Finally, Section VII concludes the paper.

### _wp08244 - Section VI. Finally, Section VII concludes the paper.

### II. Literature
- Few studies analyze the politics of distortions in international factor movements; unified framework for migration policy formation has yet to emerge.
- Key theoretical contributions:
  - Benhabib (1996):
    - Median voter under majority voting chooses immigrants complementary to her endowment.
    - If median voter unskilled → lower bound on skill level (only skilled admitted).
    - If median voter highly educated → upper bound on skill level (only low-education admitted).
    - Shortcoming: optimal policy does not identify the actual size of inflows (no quota determination).
  - Ortega (2005):
    - Extends Benhabib to a dynamic setting; accounts for descendants of migrants gaining the vote.
    - Skilled natives prefer admitting unskilled workers for short-run wage complementarities; political channel can reverse preferences over time.
    - Characterizes conditions under which an equilibrium migration quota might arise (prediction on size of inflows).
  - Facchini and Willmann (2005):
    - Menu auction framework with organized groups and an elected politician.
    - Policies depend on whether a production factor is represented and on substitutability/complementarity across factors.
    - Present paper differs by linking equilibrium policies explicitly to lobbying expenditures and by using a multi-sector environment to analyze industry lobbying effects.
- Empirical literature is scarce:
  - Historical accounts document political-economy of immigration restrictions (end XIX–beginning XX century); Goldin (1994) documents capital owners opposing the 1917 literacy test while AFL and Knights of Labor supported it.
  - Hanson and Spilimbergo (2001) on U.S. border enforcement: enforcement softens when sectors using illegal immigrants expand—suggests sectoral lobbying affects enforcement.
  - Empirical work directly linking lobbying to migration policy remains limited.

### III. Migration policy in the United States (institutional facts and patterns)
- Legal entry channels:
  - Permanent (immigrant) admission → Lawful Permanent Residents (LPR), green card; allowed to work and may apply for U.S. citizenship.
  - Temporary (non-immigrant) admission → generally not allowed to work except under specific categories.
- Historical notes:
  - Distinction introduced in Steerage Act of 1819; Immigration Act of 1907 required declaration as permanent or temporary; Immigration Act of 1924 introduced temporary admission classes.
  - Immigration and Naturalization Act (INA) of 1990 and subsequent modifications currently discipline policy.
- Annual quotas and preferences for LPR acquisition:
  - Annual flexible quota for family-sponsored preferences, employment preferences and diversity immigrants: 416,000 to 675,000.
  - Immediate relatives are exempt from annual numeric limits; immediate relatives of U.S. citizens account for over 40 percent of annual LPR inflows (CBO (2006)).
  - Refugees: cap set by the U.S. President; for the period 2003–07, the cap fixed at 70,000 admissions per year.
  - No numeric limit for asylum seekers.
- Non-immigrant visas (2001–05 averages):
  - Total non-immigrant visas issued on average per year: 5,735,577.
  - “Other admissions” (temporary visitors, official representatives, transitional family members, students plus spouses/children) represent approximately 85 percent of non-immigrant visas in 2001–05.
  - Work and related visas average per year: 835,294.
    - Of these, 315,372 are classified as “Temporary workers” (includes H1B, H1A & H1C, H2A, H2B, H3, H4).
  - Selected category averages 2001–05 (from Table 1):
    - Exchange visitors and spouses/children J1, J2: 291,241
    - Workers with extraordinary ability O1, O2: 8,865
    - Internationally recognized athletes or entertainers P1, P2, P3: 32,762
    - Cultural Exchange and Religious Workers Q1, Q2, R1: 10,414
    - Treaty traders/investors and their children E: 35,282
    - Spouses/children of certain foreign workers O3, P4, Q3, R2, I: 21,469
    - NAFTA Professionals and spouses/children TN, TD: 2,124
    - Intracompany transferees and spouses/children L1, L2: 117,765
    - Temporary workers total: H1B 130,051; H1A, H1C 122; H2A 31,322; H2B 72,684; H3 1,518; H4 79,675
    - Other admissions total: 4,900,283
      - Temporary Visitors B1, B2, B1/B2: 4,154,485
      - Official representatives and transitional family members A, G, K: 165,141
      - Students and spouses/children F1, F2, M1, M2: 266,099
      - Other non work visas: 314,558
    - Total non immigrant visa issuances: 5,735,577
- Visa quotas and sector specificity:
  - Many work visa categories are subject to explicit quotas set by Congress (e.g., H1A, H1B, and until 2005 H2B).
  - Universities and government research laboratories obtained a permanent exemption from the overall H1B quota starting in 2000.
  - Introduction in 2005 of the H2R visa category (result of Maryland seafood industry lobbying) effectively increased quota for non-agricultural temporary workers (H2B).
  - Some visa categories are sector- or occupation-specific (H1A & H1C for nurses, H2A for agricultural workers, R1 for religious workers, P for performing artists/athletes).
  - Other categories (H1B, L1, H2B) are not immediately linked to a specific sector; sectoral allocation under these programs is likely influenced by lobbying.
- Employment-based green cards:
  - Employment-preference green cards represent a small fraction of total LPR admissions.
  - Example: in 2001, out of 1,064,318 individuals granted permanent resident status, 179,195 (16.8 percent of the total) were admitted under the employment-preference category (this number includes spouses and children).
  - Data on employment-based green cards by sector were not available for the authors' sector-level analysis.

### IV. Theoretical framework (model structure and predictions)
- Modeling approach:
  - Use of a "protection formation function" approach rather than explicitly modeling quid-pro-quo with campaign contributions (Grossman and Helpman (1994)) to allow a more general role for lobbies (informational and financial channels).
  - Government policy is modeled as a function of expenditures by pro- and anti-migration groups.
- Economic environment:
  - Small open economy in goods and factors with 1 + n sectors.
  - Short-run view with sector-specific factors (labor markets segmented by industry).
  - Numeraire sector uses only sector-specific labor; all other sectors use sector-specific labor (assumed internationally mobile) and a fixed factor (capital).
  - International goods and factor prices are given.
  - Consumers have separable, quasi-linear utility.
  - Restrictions to physical relocation of people modeled as a binding quota (policy tool).
- Lobbying game and comparative statics:
  - Two lobbies per sector: pro-migration lobby (capital owners) and anti-migration lobby (workers).
  - The protection function maps lobbying expenditures into policy restrictiveness for labor inflows.
  - Comparative predictions:
    - Higher organized labor contributions in a sector → higher protection from foreign inflows → lower equilibrium number of immigrants in that sector (ceteris paribus).
    - Higher organized business (capital owners) expenditures in a sector → less restrictive migration policy → higher number of immigrants in that sector (ceteris paribus).

### V. Data
- Focus and period:
  - Analysis restricted to temporary non-immigrant work visas (sector-level focus).
  - Time period for empirical analysis: 2001–05 (due to visa data availability).
- Lobbying expenditures (institutional and measurement notes):
  - Lobbying Disclosure Act of 1995 requires semi-annual reports starting in 1996 from lobbyists and firms with in-house lobbying departments.
  - Reports list clients and total income received and require disclosure of issues lobbied (76 general issues available for selection).
  - Center for Responsive Politics (CRP) compiles lobbying expenditure data from semi-annual reports; CRP reports cover 1998 through 2005.
  - For empirical work, annual lobbying expenditures and incomes are calculated by adding mid-year and year-end totals.
  - CRP matches each firm to an industry.
- Measures constructed:
  - “Overall” or “total” lobbying expenditures in an industry = sum of lobbying expenditures by all firms in that industry on any issue.
  - Lobbying expenditures for immigration in an industry calculated via three-step procedure:
    1. Consider only firms that list “immigration” as an issue in their lobbying report.
    2. Split each such firm's total expenditure equally among the issues they lobbied for.
    3. Aggregate these firm-level expenditures on immigration across firms in the industry.
  - Robustness measure: alternative measure based on total lobbying expenditure of firms who list migration as an issue (no issue-splitting).
- Additional empirical and descriptive points:
  - Lobbying expenditures are of “... an order of magnitude greater than total PAC expenditure” (Milyo, Primo, and Groseclose (2000)).
  - Lobbying Disclosure Act definitions:
    - “Lobbying activities” includes lobbying contacts and efforts supporting such contacts, including preparation, research, and coordination.
    - “Lobbying contact” defined as communication to covered officials on legislation, rules, administration of programs, or nominations/confirmations.
  - An individual is a “lobbyist” for a client if he/she makes more than one “lobbying contact” and lobbying activities constitute at least 20 percent of the individual's time for that client over any six-month period.
  - Example: Morrison Public Affairs Group reported for O'Grady Peyton Intl (AMN Health Care Services) listing only “immigration” as the issue for January–June 2004.
  - Example: Microsoft’s January–June 2005 report lists immigration among other issues.
  - American Hospital Association contributed about 10 percent of the lobbying expenditures for immigration in 2005.
- Data limitations cited:
  - Lack of Department of Homeland Security data on employment-based green cards by sector constrained analysis to non-immigrant work visas.
  - Lobbying reports list issues at a general level (one of 76 options), necessitating allocation rules for attributing expenditures to immigration.

*Source: _wp08244 - Section VI. Finally, Section VII concludes the paper.*

### 3.8 billion U.S. dollars per political cycle on targeted political activity, which includes PAC

### _wp08244 - 3.8 billion U.S. dollars per political cycle on targeted political activity, which includes PAC

### Dataset and scope
- Lobbying dataset: 2001–05 unbalanced panel of a total of 12,376 firms/associations of firms; 481 list immigration as an issue in at least one year.
- Time coverage used in empirical analysis: averages over the five years between 2001 and 2005 (cross-sectional variation exploited).
- Industry coverage: dataset that covers 120 three-digit industries that follow the 1950 Census Bureau industrial classification (CPS classification).
- Visas covered: H1A, H1B, H1C, H2A, J1, O1, O2, P1, P2, P3, R1. L1 and H2B not available by sector.
- Additional data sources: USCIS (H1B by sector), ‘Report of the Visa Office’ for other visa types, IPUMS-CPS (2001–2005), Bureau of Economic Analysis (output, price, FDI), ACES (domestic capital), Heidelberg Institute / World Bank (wars), Ramcharan (2007) (other shocks).

### Key descriptive facts and measurements
- Lobbying expenditures represent close to 90 percent of all interest groups' targeted political activity; PAC contributions represent 10 percent.
- Average industry spending on immigration-related lobbying (2001–05):
  - about $100,000 per year (when split equally among various issues).
  - total expenditures by firms in a sector that lobby for immigration: about $1.1 mn per year.
- Top spenders on immigration lobbying (2001–05) include: Engineering and computer services, Educational services, Hospitals, Food and related products, Office machines and computer manufacturing, Agriculture.
- Four industries with very high immigration lobbying expenditures are also among those receiving the highest number of visas (Educational services and Engineering and computer services among the top).
- 97.4 percent of the variation in log(visas) is across industries (little within-industry time variation).
- There are 25 worker unions during 2001–05 that lobby for immigration (some national, some sector-specific).

### Empirical strategy and main variables
- Dependent variable: number of visas divided by the number of native workers in the same sector (visas/native workers).
- Key explanatory variables:
  - log of industry's lobbying expenditure on migration divided by number of native workers in the sector — measures political organization of capital.
  - union membership rate (fraction of natives who are union members in each industry) — proxy for political organization of labor.
- Variables scaled by number of native workers to control for industry size differences; log number of native workers included as additional control.
- Control variables included in robustness checks: output, industry-specific unemployment rate, price of the good produced, domestic capital, foreign direct investment (FDI), negative origin-country shocks, lagged U.S. wages.

### Main empirical findings
- Coefficients and explanatory power:
  - Positive and significant coefficient on industry's lobbying (significant at the 1 percent level in regressions (1)-(2)).
  - Negative and significant coefficient on union membership rate (significant at the 10 percent level in regressions (1)-(2)).
  - The two key variables explain 14 percent of the variation in the number of visas per native worker across sectors (regression (2)).
  - The lobbying variable alone explains 11 percent of the variation.
- Magnitudes (from regression results and illustrative elasticities):
  - A 10 percent increase in the size of the industry's lobbying expenditures on migration per native worker raises the number of visas to that industry, per native worker, by 3.6 percent.
  - A 1 percentage point increase in union density (for example, from 10 to 11 percentage points, which amounts to a 10 percent increase in the union membership rate) reduces the number of visas per native worker by 2.6 percent.
- Additional results:
  - Output, the unemployment rate, prices, domestic and foreign capital have an insignificant effect on the number of visas per native worker in regression (3).
  - Negative origin-country shocks have a negative and significant coefficient (interpretations: decreased ability to migrate despite higher willingness; or increased asylum flows reducing work visas in affected sectors).
  - Lagged U.S. wages have a positive and significant impact on the number of visas issued in a given sector.

### Robustness and interpretation
- Results are robust to inclusion of industry-level controls (output, unemployment rate, prices, domestic and foreign capital) in regression (3); coefficients on lobbying and union membership remain similar in sign and significance.
- Use of visas issued as the measure of migration restrictions is defended as an ex post quota-related measure; H1B quotas are often binding (filled within the first few days of each fiscal year).
- Acknowledged limitation: absence of direct data on lobbying expenditures by unions at the industry level forces use of union density as a proxy for labor lobbying activity; this prevents a structural interpretation of coefficient estimates.

### Policy-relevant analysis and implications
- Sectoral political organization matters for migration policy outcomes:
  - Stronger capital-side lobbying (higher industry lobbying expenditures per native worker) is associated with more visas per native worker (lower barriers).
  - Stronger labor-side organization (higher union membership rates) is associated with fewer visas per native worker (higher barriers).
- Using PAC contributions alone to study lobby influence on migration policy could be misleading because PAC contributions are only 10 percent of targeted political activity and correlate weakly with immigration-specific lobbying.
- The availability of issue-specific lobbying data (immigration-tagged expenditures) is crucial to accurately identify the effect of lobbying on migration outcomes.

*Source: _wp08244 - 3.8 billion U.S. dollars per political cycle on targeted political activity, which includes PAC*

### introduction of these additional regressors.

### introduction of these additional regressors.

### Endogeneity concerns and identification strategy
- The authors are concerned that lobbying expenditures by capital and labor (their two key variables) are endogenous and potentially subject to reverse causality.
- Two opposing directions of bias from reverse causality are discussed:
  - Sectors with more migrants may be near optimal levels and thus invest less in lobbying, biasing estimates toward zero.
  - Sectors receiving more visas might increase lobbying to address immigrant-related issues, biasing estimates upward (true effect lower than estimated).
- To address endogeneity, an instrumental-variable estimation strategy is used.

### Instruments for lobbying expenditures and union membership
- Instruments for lobbying on migration:
  - Measure of lobbying expenditures by firms in each sector that do not list migration as an issue (out of 12,376 firms in the lobbying dataset, 96 percent do not list immigration as an issue).
    - Exclusion assumption: these firms’ lobbying on other issues does not directly affect migration.
    - First-stage relevance argument: industry-level factors affecting all firms’ lobbying make this instrument correlated with lobbying on migration.
  - Measure of sector concentration: variance of firm size (proxied by annual payroll) within a sector, from the U.S. Census, County Business Patterns (http//www.census.gov/csd/susb/defterm.html).
    - Theoretical rationale: higher concentration (higher variance) eases collective action and increases lobbying (Olson (1965)).
- Instrument for union membership rate:
  - Union densities across industries from the United Kingdom.
    - Rationale: sector-specific union membership rates are positively correlated across industrialized countries (Riley (1997), Blanchflower (2007)); U.K. rates plausibly do not directly affect U.S. visas (exclusion restriction).

### First-stage strength and overidentification
- Reported first-stage statistics:
  - In regression (1), first-stage F value for excluded instruments = 62.66.
  - In first stage of the other endogenous variable, first-stage F value = 40.61.
- Table 4b first-stage findings:
  - Lobbying expenditures on immigration are positively and significantly correlated with lobbying on other issues and with sector concentration.
  - In Table 4b, columns (1)-(2), log(lobbying exp on other issues/native workers) is highly significant; log(variance of firm size) is jointly significant with log(lobbying exp on other issues/native workers) (p-value for joint F-test = 0.00).
- The Hansen test for overidentifying restrictions is satisfied at the 1 percent significance level (cannot reject null of zero correlation between residuals and excluded instruments).

### Main IV estimates and interpretation
- IV regression results (Table 4a and related):
  - Number of visas per native worker is higher in sectors where business lobbies are more active.
  - Number of visas per native worker is lower in sectors where labor unions are more important.
  - Magnitudes of coefficients on both lobbying expenditures and union membership rates increase relative to OLS (Table 3).
    - Possible reasons for larger IV magnitudes:
      - Negative correlation between lobbying expenditures on migration and the unobserved component of the number of visas (sectors with more visas contribute less because closer to ideal number).
      - Positive correlation between union membership rates and the unobserved component of the number of visas (natives feel stronger threat where visas higher, increasing unionization).
      - Measurement error in key explanatory variables causing attenuation bias in OLS.
- Caution on IV interpretation:
  - If lobbying expenditures on non-immigration issues draw resources/policymakers’ attention away from migration and directly reduce visas, IV estimate would be biased toward zero and represent a lower bound.
  - Results should be interpreted with due caution given absence of a clean natural experiment.

### Key quantitative estimates (summary)
- Core IV/robust estimates:
  - A 10 percent increase in the size of lobbying expenditures by business groups, per native worker, is associated with a 2.9–4.4 percent larger number of visas per native worker.
  - A 1 percentage point increase in the union membership rate is associated with a 2.6–5.5 percent lower number of visas per native worker.
- Alternative dependent-variable results (Table 5):
  - Excluding J1 visas, restricting to visas with quotas, and focusing on H1B approvals produce similar results.
  - Regression (3) (log number of H1B visas approved): sectors with 10 percent higher lobbying expenditures per native worker are associated with a 1.8 percent larger number of H1B visas approved per native worker.
  - In the same regression, a 1 percentage point increase in union membership rate is associated with a 3.6 percent lower number of H1B visas per native worker.
- Alternative lobbying measure (Table 6):
  - Using total lobbying expenditures of firms that list “immigration” as an issue (upper bound measure) yields estimated impacts very similar and not statistically different from Table 3.
- Political-organization proxy comparison (Table 7):
  - PAC campaign contributions yield coefficients on political-organization variables that are either not significant or marginally significant.
  - When both PAC contributions and lobbying expenditures are included, lobbying expenditures on migration (not PAC contributions) positively affect number of visas.
  - PAC data averaged over 2001–02 and 2003–04 election cycles; lobbying expenditures averaged over 2001–04 for comparison.

### Robustness checks and heterogeneity
- Robustness analyses (Appendix, Table A6 and A7) confirm Table 3 findings:
  - Dropping agriculture (columns (1)-(2), Table A6).
  - Using pooled rather than averaged data with year fixed effects (columns (3)-(4), Table A6).
  - Constraining observations to be the same across regressions (columns (5)-(6), Table A6).
  - Including sectors with zero lobbying expenditures (columns (7)-(8), Table A6); in the log specification zeros replaced by minimum positive value.
  - Results remain very similar to Table 3.
- Skilled- vs. unskilled-intensive sector split:
  - Absolute magnitudes of coefficients on log(lobbying exp/native workers) and union membership rates are larger for skilled-intensive sectors relative to unskilled-intensive ones.
  - However, the two sets of coefficients are not significantly different (results available upon request).
- Controls:
  - Results robust when controlling for capital/labor ratio and skilled/unskilled labor ratio in each sector (see columns (1)-(2), Table A7).
- Functional form:
  - Data best fits a log specification rather than levels; no evidence of non-linear effects in lobbying and union membership (results available upon request).

### Policy-relevant findings and mechanisms
- Both pro- and anti-immigration interest groups significantly and economically shape migration outcomes across sectors:
  - Business lobbying increases sectoral visas; labor unions decrease them.
  - Quantified effects: 10 percent more business lobbying associated with 2.9–4.4 percent more visas; 1 percentage point higher union membership associated with 2.6–5.5 percent fewer visas.
- Policymakers use a variety of instruments beyond explicit quotas (“visible” restrictions) to manage sectoral access:
  - “Invisible” barriers include rules restricting foreign medical doctors (residency requirements) and complex procedures for H2A seasonal agricultural worker hiring (advertising requirements, Adverse Effect Wage Rate obligations), which can substantially reduce program utilization.
- Channels of lobbying:
  - For statutory changes (e.g., increasing a visa cap), interest groups lobby Congress (example: H1B increase in 1998 as part of Omnibus Appropriations Bill HR 4328; agricultural interests and HR 371).
  - For regulatory changes (e.g., labor certification, H2A wage rates), interest groups lobby executive agencies such as the Department of Labor.
- Implication: political-economy forces play a quantitatively important role in determining cross-sectoral allocation of immigrants; policymakers target allocation across sectors.

### Suggested extensions and data improvements
- Further work could:
  - Explore alternative measures of immigration policy (legal vs illegal, temporary vs permanent).
  - Examine variation in policy outcomes by occupation and geography (e.g., across U.S. states).
  - Use firm-level lobbying data to study political-economy effects on other policies (trade, environment, taxes).

*Source: _wp08244 - introduction of these additional regressors.*

### 45. Cambridge.

### 45. Cambridge.

### Theoretical framework — small open economy with sector-specific factors and migration policy
- Economy: 1+n sectors; sector zero is the numeraire produced with sector-specific labor; other sectors use sector-specific labor that is internationally mobile.
- Production: diminishing returns to labor in non-numeraire sectors; domestic return to labor in sector i denoted ω_i; sector-specific fixed factor (called capital) returns denoted π_i.
- Prices: international prices normalized to one; international return to each type of labor set equal to one.
- Preferences: consumers have separable, quasi-linear utility:
  - Utility: u_0 + sum_{i=1}^n u_i(x_i) with indirect utility expressed as V = I + sum_i p_i d_i(p_i) + consumer surplus terms as in the source.
- Labor supply and quotas:
  - Total domestic supply of labor of type i denoted L_i; demand denoted l_i.
  - Binding quotas q_i in sectors i ∈ {1,...,n} may be accompanied by a tax; wage under a binding quota denoted ω_i^q and employment level l_i^q.
  - Fiscal revenues from quota cum tax (rebated lump-sum) given by equation (2): γ_i q_i L_i (ω_i - ω_i^q) + (1-γ_i) q_i L_i (ω_i - ω_i^q) [expressed in the source as the relevant fiscal revenue expression with parameters γ_i and migrant share].
  - γ_i ∈ [0,1] denotes government share of quota rent; migrants keep fraction (1-γ_i) of wage premium.
- Welfare of domestic agents:
  - Welfare of domestic workers in sector i (supplying labor): V_{iL} = α_i [1 + ω_i + sum p_i s_i + T(q)] + q l_i — as given in equation (3) format in the source.
  - Welfare of owners of sector-specific capital: V_{iK} = α_{iK} [1 + π_i + sum p_i s_i + T(q)] + q π_i — as given in equation (4) format.
- First-best policy:
  - Maximizing native welfare W = sum_i (V_{iL} + V_{iK}) yields free labor mobility as first-best (i.e., admit all foreign workers firms are willing to hire).
  - First-best quota q_i^* satisfies condition (6) as stated: q_i^* such that p_i L_i - p_i m_i(l_i) ≥ ... (expressions as in source).
  - q_i^* is ceteris paribus increasing in sectoral capital stock k_i and in the relative price p_i via outward shifts in labor demand.

### Political economy of migration policy — lobbying model and equilibrium quota
- Lobbies:
  - Pro-migration lobby composed of capital owners; anti-migration lobby composed of workers.
  - Protection function for sector i assumed to be increasing in worker expenditures and decreasing in capital expenditures: q_i = 1 - λ + (1-λ) ... (source defines 22))((1)(=1)( iKiLii EEqλλω−−− with λ weight of labor).
  - Increasing returns to lobbying (larger donors have disproportionately greater influence).
- Lobbying payoff:
  - Net welfare for lobby groups: V_{group}(q) - Ω(E_group) with E the expenditure.
  - First-order conditions (assuming γ_i = 1 for all i for simplification) given by equations (7) and (8) in the source. These equate marginal benefit of lobbying on factor returns to marginal cost (equal to 1).
- Linear labor demand example:
  - Domestic labor demand: ω_i = b_i L_i - ... (source uses linear form ii bLLω−= ).
  - Assuming concentrated capital ownership (α_{iK} = 0 for all i in the example), solving first-order conditions yields equilibrium quota (equation (10)):
    q_i = [ ( ... expression involving E_{iL}, E_{iK}, λ, α_{iL}, α_{iK}, b_i, L_i ... ) ] as shown in the source.
  - Comparative statics: sectors with higher union activity/expenditures have smaller quotas (higher protection); sectors with higher capital expenditures have larger quotas (less restrictive migration policies).

### Supply shocks vs policy determination of migrant inflows
- Consider an international supply shock that raises world wage from 1 to ω_S' (source notation):
  - Two scenarios illustrated in Figure A1:
    - Panel (a): quota remains binding after shock; wage determined by quota w_q remains above world wage ω_S' and the number of migrants is set by the host country's policy.
    - Panel (b): shock is large enough such that world wage exceeds quota wage; quota no longer binding, migrants admitted are determined by market intersection of domestic labor demand and international labor supply; political-economy forces become irrelevant for volume of migrants.
- Conclusion: for supply-side shocks to fully determine migrant inflows and render political-economy forces irrelevant, the shock must be very large (as per the source discussion).

### Empirical evidence — summary of key statistics, correlations, and regression results (data averaged mostly over 2001-2005)
- Correlations and descriptive:
  - Correlation between (log) contributions from PACs and (log) overall lobbying expenditures: 0.328 (robust standard error = 0.099; p-value = 0.000).
  - Correlation between (log) contributions from PACs and (log) lobbying expenditures for immigration: 0.074 (robust standard error = 0.132; p-value = 0.580).
  - Correlation between (log) lobbying expenditures for immigration and (log) number of visas: 0.316 (robust standard error = 0.080; p-value = 0.000).
  - Correlation between union membership rates and (log) number of visas: -2.283 (robust standard error = 1.220; p-value = 0.063).
  - Correlation between (log) lobbying expenditures for immigration and (log) number of H1B visas: 0.255 (robust standard error = 0.068; p-value = 0.001).
  - Correlation between union membership rates and (log) number of H1B visas: -2.180 (robust standard error = 1.161; p-value = 0.062).
- Regression highlights (tables report coefficients with robust standard errors in brackets; significance: ***, **, * at 1, 5, 10 percent):
  - Table 2 (OLS, avg 2001-2005): Dependent variable log(visas/native workers)
    - log(lobbying exp/native workers): 0.316*** [0.076] in column [1]; 0.356*** [0.080] in [2]; 0.294*** [0.084] in [3].
    - union membership rate: -2.594* [1.430] in [2]; -3.232** [1.455] in [3].
    - shocks: -6.834** [2.811] in [3].
    - log(lag US wages): 10.318*** [3.329] in [3].
    - N = 126 in columns [1] and [2]; N = 120 in [3]. R-squared 0.11, 0.14, 0.27 respectively.
  - Table 4a (Instrumental Variables):
    - IV estimates show log(lobbying exp/native workers) 0.439*** [0.126] in [1]; 0.325** [0.124] in [2].
    - union membership rate -3.671 [2.253] in [1]; -5.495** [2.193] in [2].
    - First-stage F for log(lobbying exp/native workers): 62.66 and 47.22 reported.
    - First-stage F for union membership: 40.61 and 16.38 reported.
    - Hansen's J-statistic (p-value): 0.97 and 0.73 in two specifications.
    - N = 109 and 106; R-squared 0.11 and 0.24.
  - Table 5 (alternative dependent variables — Visas excluding J1; Visas with quota; H1B visas):
    - log(lobbying exp/native workers): 0.287*** [0.082], 0.253*** [0.080], 0.182*** [0.065] across columns [1]-[3].
    - union membership rate: -3.516** [1.350], -3.299** [1.376], -3.623*** [1.292].
    - shocks: -6.589** [2.748], -5.389** [2.594], -4.892** [2.444].
    - log(lag US wages): 10.265*** [3.295], 10.193*** [3.173], 9.834*** [2.951].
    - N = 120 for each; R-squared 0.29, 0.31, 0.34.
  - Table 6 (alternative measure of lobbying expenditures):
    - log(lobbying exp_upper bound/native workers): 0.321*** [0.082] in [1]; 0.252*** [0.085] in [2].
    - union membership rate: -2.224 [1.423] in [1]; -3.087** [1.458] in [2].
    - N = 126 and 120; R-squared 0.11 and 0.25.
  - Table 7 (PAC contributions vs lobbying expenditures):
    - log(PAC contribution / native workers): coefficients vary and are not uniformly significant (e.g., 0.208* [0.119] in [1]; -0.131 [0.150] in [2]).
    - log(lobbying exp/native workers): 0.326*** [0.068] in [2]; 0.243*** [0.080] in [3]; 0.305*** [0.069] in [4]; 0.237*** [0.079] in [5].
    - union membership rate consistently negative and often significant across specifications (e.g., -1.801* [1.056] to -4.068*** [1.485]).
    - N ranges from 112 to 133; R-squared ranges 0.06 to 0.32.
- Robustness and first-stage evidence:
  - Table 4b (first-stage for IV): instruments include lobbying expenditures on other issues/native workers, variance of firm size, and union membership rate in the UK. Coefficients:
    - log(lobbying exp on other issues/native workers): 1.017*** [0.077] in [1]; 1.036*** [0.090] in [2].
    - union membership rate in the UK: -1.620* [0.853] in [1]; -1.314* [0.790] in [2]; also enters positively and significantly in columns [3] and [4] for alternative first-stage specifications with coefficients 0.514*** [0.104] and 0.459*** [0.106].
  - Table A1a / A1b: summary statistics (selected, averaged 2001-2005 unless noted)
    - Log(Number of visas as a ratio of native workforce): Obs 120, Mean -7.86, Std. Dev. 1.84, Min -11.87, Max -1.34.
    - Log(Lobbying expenditures for immigration as a ratio of native workforce): Obs 120, Mean -3.03, Std. Dev. 2.05, Min -9.41, Max 1.56.
    - Union membership rate in the US: Obs 120, Mean 0.12, Std. Dev. 0.12, Min 0.00, Max 0.47.
    - Number of visas: Obs 120, Mean 2,101, Std. Dev. 10,593, Min 2, Max 102,836.
    - Number of H1B visas: Obs 120, Mean 955, Std. Dev. 4,807, Min 2, Max 48,824.
    - Lobbying expenditures for immigration (in US$): Mean 99,811; Std. Dev. 215,278; Min 0; Max 1,513,108.
    - Total lobbying expenditures by all firms which lobby for immigration (in US$): Mean 1,084,469; Std. Dev. 2,363,724; Min 14; Max 17,800,000.
    - Contributions by PACs (in US$): Obs 127, Mean 6,741,956; Std. Dev. 4,856,364; Min 174,543; Max 28,700,000.
- Robustness checks (Table A6, A7):
  - Results robust to dropping agriculture (industry code = 105), inclusion of year fixed effects, restricting observations, including industries with zero lobbying expenditures (handled by replacing log(0) with log of minimum value).
  - Controlling for capital-labor and skilled-unskilled labor intensity does not overturn the positive effect of lobbying expenditures and the negative effect of union membership on visas per native worker. Example (Table A7):
    - log(lobbying exp/native workers): 0.294*** [0.084] in [1]; 0.235*** [0.072] in [2].
    - union membership rate: -3.232** [1.455] in [1]; -1.139 [1.255] in [2].
    - In [2], log(skilled-unskilled labor intensity): 1.469*** [0.266].

### Empirical interpretation and key takeaways (as presented in the source)
- Political activity measured by sectoral lobbying expenditures for immigration is positively and significantly associated with larger numbers of visas per native worker across sectors.
- Union membership rates at the sector level are negatively associated with visas per native worker, often statistically significant at conventional levels.
- Instrumental variables and robustness checks (including alternative dependent variables such as H1B visas and quotas-excluded visas) support the interpretation that political economy forces (lobbying and union strength) materially influence sectoral immigration outcomes.
- Large international supply-side shocks can, in principle, override domestically set quotas; however, such shocks must be sufficiently large for market forces to dominate political-economy determinants in the model.

*Source: _wp08244 - 45. Cambridge. (PDF chapter/section content provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08244.pdf_