Skilled Immigration and the Employment Structures of US Firms

Transcription

1 Skilled Immigration and the Employment Structures of US Firms Sari Kerr William Kerr William Lincoln 1 / 56

2 Disclaimer: Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent the views of the U.S. Census Bureau. All results have been reviewed to ensure that no confidential information is disclosed. 2 / 56

3 3 / 56

4 4 / 56

5 5 / 56

6 6 / 56

7 Main Objective Hope to build a deeper view of the firm s role in immigration (Was) the first study we know of to consider the effects of immigration using employer-employee data Study how high skilled immigration affects the employment structures of US firms 7 / 56

8 8 / 56

9 9 / 56

10 Overview of Results Total skilled employment expands with the hiring of highly skilled immigrants Employment expansion is larger for young skilled natives relative to older natives 10 / 56

11 Overview Data Conceptual framework OLS Estimations IV Estimations 11 / 56

12 LEHD Employment Data 12 / 56

13 LEHD Data All private firms and their employees Sourced from unemployment insurance filings Combined with information from social security filings 29 participating states with various start years, from 1990 to 2002, and end year of 2008 Information for each firm: LEHD: establishment code, industry, total employment, payroll and exact location Linked to all other Census Bureau operating data 13 / 56

14 14 / 56

15 LEHD Data Information for each employee: Quarterly earnings Age, gender, and race Citizenship status: US citizen, naturalized citizen, non-citizen Place of birth No information on occupation Education is imputed Exact location within state for firm establishments but is imputed for workers 15 / 56

16 Firm Sample Focus: major employers & patenting firms Sample meets one of following criteria: Accounts for >0.05% of patents Top 100 "employer name" in LBD during any year from Top 100 Compustat worldwide sales or employment over full period A Fortune 200 company in / 56

17 Firm Sample Firm selection Consider 18 states present by 1995 Drop firms with <25% employment in LEHD states Final group on average >50% in LEHD states Sample: 319 firms Average employment is 52k workers in 18 LEHD core states 17 / 56

18 Firm Sample Sizeable share of activity: Consistent with highly skewed firm size distribution (Axtell, 2001) 34% of US patenting 10%-20% of total LEHD employment 67 million workers in total Our baseline regressions contain 3,374 observations 18 / 56

19 Sample Group: Employees Skilled definition: Median earnings over $50,000 in real $2008 Calculated over employment spells % of workforce earns $50k+ Aged 18-65, young-old split at 40 yrs 19 / 56

20 Conceptual Framework 20 / 56

21 Conceptual Framework We are interested in looking at how changes in the employment of skilled immigrants affect changes in the employment of other groups "Microsoft has found that for every H-1B hire we make, we add on average four additional employees to support them in various capacities" - Bill Gates in 2008 Congressional Testimony We consider a simple conceptual framework that will allow us to think about these employment patterns in a straightforward way in terms of substitution and complementarity between different types of workers allow us to relate our findings to arguments made in the public debate over high skilled immigration give us guidance for empirical work 21 / 56

22 Conceptual Framework A firm that makes output using two types of labor domestic and immigrant with the concave production function Q = Q (L D, L I ) Positive but diminishing marginal returns to each type of labor The concave revenue function of the firm is R (Q, y), with y representing economic conditions exogenous to the firm 22 / 56

23 Conceptual Framework The firm maximizes R (Q, y) c D L D c I L I where c D is the cost for domestic workers and c I is the cost for immigrant workers This leads to the familiar conditions for profit maximization that R Q = c D and R Q Q L D Q L I = c I 23 / 56

24 Conceptual Framework Denote the change in immigrant employment by dl I, and the change in domestic employment by dl D Totally differentiating the first expression above dc D = Q 2 R L D Q 2 [ 2 Q R Q L 2 D [ Q L D dl D + Q dl D + Q 2 R L D Q y dy L I dl I ] + ] 2 Q dl I + L D L I We assume that dc D /dl I = 0 and that dy/dl I = 0, given that y is assumed exogenous 24 / 56

25 Conceptual Framework We can then rearrange the remaining terms to be dl D = [ Q L D Q L I 2 R Q 2 + R Q [ ( Q L D ) 2 2 R Q 2 + R Q ] 2 Q L D L I 2 Q L 2 D ]dl I Given our assumptions, the denominator is positive The relationship between dl D and dl I will be positive only if 2 Q L D L I > 0 and is suffi ciently large to offset the magnitude of the (negative) first term in the summation of the numerator 25 / 56

26 Conceptual Framework We can then rearrange the remaining terms to be dl D = [ Q L D Q L I 2 R Q 2 + R Q [ ( Q L D ) 2 2 R Q 2 + R Q ] 2 Q L D L I 2 Q L 2 D ]dl I This makes sense intuitively if domestic and immigrant worker employment are complementary and suffi ciently strong to overcome the concavity of the revenue function, then we should see a positive relationship between growth in domestic employment and growth in immigrant employment in the data 26 / 56

27 OLS Estimations 27 / 56

28 Following the results from the conceptual framework we consider the following specification Y f,t = β ln(emp YSI f,t ) + δ X f,t + η i,t + ε f,t, Firm f, sector i, year t ln(empf YSI,t ) is the log number of young skilled immigrants employed in year t by firm f Y f,t is the outcome variable of interest X f,t is a vector of firm-year controls η i,t are sector-year fixed effects 28 / 56

29 Firm-Year Controls X f,t Local area controls calculate firm s initial employment across counties and then weight county trends by these shares: LEHD employment, immigrant share, and share of workers over 40 (Card) "Supply-Push" controls Calculate each firm s initial skilled immigrant distribution across 12 geographic groups (Europe, Latin America, etc.). Then interact this with the growth of skilled immigrants at the national level, weighting by the initial distribution. Do the same for low skilled workers. (Card) Age-education controls calculate firm s initial employment distribution across 6 age-education cells (young, old; HS or less, some college, college or more) and interact this with national growth in skilled immigration in these categories (Borjas) 29 / 56

30 Table: OLS Estimations Log employment of skilled worker group: Older natives Young natives Older immigrants Log employment of young skilled immigrants (0.021) (0.020) (0.045) 30 / 56

31 Table: OLS Estimations Log total Older skilled Older native skilled skilled emp. worker share worker share Log employment of young skilled immigrants (0.020) (0.003) (0.003) 31 / 56

32 IV Estimations 32 / 56

33 Instrumental Variable Estimations While the OLS estimations account for fixed effects and a wide variety of additional controls, there still may be omitted factors driving the results We now turn to an IV approach that uses large changes in national high skilled immigration policy 33 / 56

34 Instrumental Variable Estimations Specifically, we take advantage of changes in the limit on H-1B visas H-1B is a non-immigrant visa Category governing high-skilled immigration Employment in "specialty occupations" Employer is responsible for visa application Three-year visa, renewable once Prevailing wage requirement Cap on visa issuances since 1990 Computer-related and SE occupations ( 60%) Large percentage coming from India ( 40%) or China ( 10%) 34 / 56

35 H-1B National Trends 35 / 56

36 Instrumental Variable Estimations We instrument for ln(emp YSI f,t ) with Dep f,t0 ln (H 1BPop t ) where Dep f,t0 is a measure of how likely they are to find and hire H-1B visa holders (or the firm s "dependency" on high-skilled immigrants) The results we consider here measure the variable Dep f,t0 with the firm s initial share of skilled immigrant workers that were born in India and China. This is similar to Card s (2001) approach except the dependency is at the firm rather than city level. It takes advantage of the fact that high skilled immigrants from these countries are likely to go to firms where there are already high skilled immigrants from their own countries The first stage F statistic is / 56

37 Table: IV Estimations Using the Chinese and Indian Worker Dependency Log employment of skilled worker group: Older natives Young natives Older immigrants Log employment of young skilled immigrants (0.115) (0.083) (0.104) 37 / 56

38 Table: IV Estimations Using the Chinese and Indian Worker Dependency Log total Older skilled Older native skilled skilled emp. worker share worker share Log employment of young skilled immigrants (0.081) (0.022) (0.022) 38 / 56

39 Instrumental Variable Estimations We also considered two alternative instruments, interacting the change in the log national H-1B population with The log ratio of the firm s LCAs (H-1B applications) to its skilled employment in 2001 Share of the firm s workforce in STEM occupations We come to similar conclusions with these instruments We also consider similar IV estimations controlling for changes in medium-skilled employment. This approach is somewhat more robust and yields similar magnitudes. 39 / 56

40 Interpretation If high skilled immigrants are unique inputs (especially at the high end), then being able to hire more could expand firm market share and lead to greater use of citizen workers (relation to trade literature, innovation). It could be that immigrants and citizen workers are substitutes within occupation categories but are complements across categories. 40 / 56

41 Conclusions Total skilled employment expands with the hiring of highly skilled immigrants Employment expansion is larger for young skilled natives relative to older natives 41 / 56

42 Thank You 42 / 56

43 Appendix 43 / 56

44 Diffi culties in Constructing Firms The primary basis in the LEHD for identifying employer-employee linkages is the state employer identification number (SEIN) that identifies individual establishments. The BRB includes for each SEIN the associated federal EIN and Census Bureau firm identifier by year. From the BRB, we collect the SEINs that are associated with our firms at any point in time. 44 / 56

45 Diffi culties in Constructing Firms This collection of complete SEIN records is important as firms occasionally change SEINs for reasons unrelated to our interests, and these legal adjustments could otherwise be confused with actual changes in the company s employment dynamics. With the collected SEINs, we then prepare the employment records for our firm sample. We need each SEIN to be uniquely associated with a firm, and therefore we research any overlapping identifiers and assign them to the appropriate company. As many of our firms are multi-establishment companies, on average our composite firms contain roughly 200 SEINs. 45 / 56

46 Firm Sample Sample: 319 firms Older natives are 50% of skilled group Younger natives are 31% of skilled group Immigrants are 19% of skilled group Hiring and departing rate of 13-14% per year 46 / 56

47 Firm Sample Sector distribution within LEHD: Manufacturing: 30% Wholesale and retail trade: 25% FIRE and services: 30% Other sectors: 15% 47 / 56

48 OLS Robustness Similar results when Controlling for changes in medium skilled immigration When considering the subsample of just top patenting firms Considering different weighting strategies Using a firm-state approach using all 29 states Raising the threshold to 66% employment in LEHD states Splitting the sample by the long-term growth rates of the firms Setting minimum employment thresholds for companies Using alternative definitions of skilled workers 48 / 56

49 IV Robustness Similar results when When considering the subsample of just top patenting firms Using a balanced panel Dropping major M&A firms Dropping firms that lobby about immigration Splitting the sample across industries 49 / 56

50 STEM Match CPS collects employment data from a random group of workers in the US every year A bridge between the CPS and LEHD has been established Ascertain the occupations of over 25k workers in our firm sample at the time of their inclusion in the CPS survey Share of the firm s workforce in STEM occupations measured in the first three years where matched employees are observed, which may be later than the typical initial period. Winsorize these shares at the 5% and 95% values. 50 / 56

51 Table: IV Estimations Using STEM Occupation Share Dependency Log employment of skilled worker group: Older natives Young natives Older immigrants Log employment of young skilled immigrants (0.261) (0.170) (0.297) 51 / 56

52 Table: IV Estimations Using STEM Occupation Share Dependency Log total Older skilled Older native skilled skilled emp. worker share worker share Log employment of young skilled immigrants (0.167) (0.057) (0.049) 52 / 56

53 Table: IV Estimations Using the Chinese and Indian Worker Dependency with Medium Skilled Workforce Control Log employment of skilled worker group: Older natives Young natives Older immigrants Log employment of young skilled immigrants (0.098) (0.077) (0.098) 53 / 56

54 Table: IV Estimations Using the Chinese and Indian Worker Dependency with Medium Skilled Workforce Control Log total Older skilled Older native skilled skilled emp. worker share worker share Log employment of young skilled immigrants (0.071) (0.018) (0.020) 54 / 56

55 Table: IV Estimations Using the Chinese and Indian Worker Dependency with Medium-Skilled Workforce Control and H-1B Cap Summations Log employment of skilled worker group: Older natives Young natives Older immigrants Log employment of young skilled immigrants (0.109) (0.091) (0.135) 55 / 56

56 Table: IV Estimations Using the Chinese and Indian Worker Dependency with Medium Skilled Workforce Control and H-1B Cap Summations Log total Older skilled Older native skilled skilled emp. worker share worker share Log employment of young skilled immigrants (0.078) (0.024) (0.026) 56 / 56