Description
Hi! Please answer the following questions in 2-4 sentences.
1. Assume you are interested in measuring the impacts of the 1980 arrival of Marielitos on Miami’s labor market. To do so, you decide to compare the labor market outcomes of native-born workers in Miami in 1979 (before the arrival of the Marielitos) to the labor market outcomes of native-born workers in Miami in 1981 (after the arrival of the Marielitos). Explain one potential issue with this research strategy and discuss how you could improve it.
2. In one-two sentences, describe the general consensus of economists on the effects of immigration on the wages of natives.
3. Pick one of the studies we discussed in class on the labor market effects of immigration and in two-three sentences describe the basics of the authors’ empirical strategy. In other words, how do the authors estimate the effects of immigration on natives’ labor market outcomes? (Describe the natural experiment and how they use it to say something about immigration’s effect on natives’ labor market outcomes.)
Please use: Source: Abramitzky et al. (forthcoming) – “The Effect of Immigration Restrictions on Local Labor Markets: Lessons from the 1920s Border Closure” Page 18 in the attached power point
4. Most of the papers we discussed in class on the labor market effects of immigration found small effects of immigration on natives’ labor market outcomes. Provide three examples of explanations you have seen these authors provide as to how/why the increase (or decrease) in immigration had such small effects on natives’ labor market outcomes in their context.
5. Provide one reason why second-generation immigrants do better in the labor market than comparable 3rd-plus generation kids.
6. Provide two reasons why the simple model in which immigration only results in a shift in labor supply is probably wrong in most contexts
Unformatted Attachment Preview
Main methods
1. Difference-in-differences (DID): compare the change in outcome
for treated group (before and after treatment) to change in
outcome for control group
2. Regression discontinuity (RD): treatment based on threshold/cutoff, compare those just above (treated) and just below (control)
3. Instrumental variables: find something else that randomly affects x
but not in any way related to y
1. Difference-in-differences (DID)
• Compare change in outcome ( , − , ) for “treated” (T) to change in outcome
( , − , ) for “control” (C)
Diff-in-diff= ( , − , ) – ( , − , )
In graph on left, treatment group
experienced bigger change in
outcome Y compared to control.
Difference in change in Y between
treatment and control provides
estimated effect!
DID practice
Diff-in-diff= ( , − , ) – ( , − , )
Effect of Mariel Boatlift on white workers’
unemployment rate:
(3.9-5.1) – (4.3-4.4) = -1.2 – – 0.1 = -1.2 + 0.1 = -1.1
⇒Suggests Mariel Boatlift decreased white workers’
unemployment rate
Effect of Mariel Boatlift on Black workers’
unemployment rate:
(9.6-8.3)-(12.6-10.3)=1.3-2.3=-1.0
⇒Suggests Mariel Boatlift decreased Black workers’
unemployment rate
2. Regression Discontinuity
treatment based on some threshold (e.g., test score,
land size limits for credit access, etc.)
In theory, group just below cutoff very similar to
group just above cutoff
Only difference is the assignment of given treatment
Compare outcomes for those just above cut-off for
those just below
Key limitation: can only study causal effects for
those around cutoff
Key assumption: no manipulation of treatment
status
See next slide for example of an example where
there does appear to be some manipulation
Manipulation example
“Manipulation of Social Program
Eligibility” by Adriana Camacho
and Emily Conover (AEJ: Policy
2011)
If below score, eligible for social welfare
programs
Poverty score index algorithm on dwelling
characteristics, demographics, income, and
employment at individual and household level
collected during Census of the Poor.
Targeted Programs
Have Their Issues…
Algorithm released in 1998
Discontinuities are larger in more politically
competitive municipalities -> higher benefits for
incumbent
What’s happening? A few possibilities:
• Enumerators can change answers
• Local politician can instruct someone to
change poverty index score
• Respondents can lie
Can’t use cut-off to identify effects of program
on other outcomes since it appears over time it
is not random.
Example of RD: “Split Decisions: Household Finance When a Policy Discontinuity
Allocates Overseas Work” by Clemens and Tiongson (2017)
Prospective Philippine migrants had to pass a Korean
language test to get high-wage jobs in Korea.
Compare households where member just passed (and
likely deployed) to those where member barely failed
(and likely wasn’t deployed)
Study how migration of a family member impacts
households’ spending, savings, and work decisions
Findings:
Migration triples home household’s expenditure on
education and health, reduces borrowing, and raises
savings
3. Instrumental Variables
An instrument z: (1.) strongly predicts the endogenous variable x, and
(2) only affects the outcome y through its effect on x
Example of IV: “Lifetime Earnings and the Vietnam Draft Lottery: Evidence
from Social Security Administrative Records” Joshua Angrist (1990)
• Research question: What’s the effect of serving in Vietnam on earnings?
• Q: Why can’t you just compare earnings of veterans to non-veterans?
• Those who enlist in military likely different in many ways (different
socioeconomic backgrounds, personal characteristics, etc.) and those differences
might lead to differences in earnings.
• Empirical Strategy: use draft lottery as an instrument for military service
• Video of draft lottery: https://www.youtube.com/watch?v=OkJH6sapQMA
• Lottery over birthdays in the year (195 dates selected at random)
• : having your birthdate drawn
Findings
• For men born in 1950, having your birthdate drawn :
• ↑ the likelihood of being observed as a veteran in 1981 by 27 p.p.
Q: Why only 27 p.p.?
• Some who weren’t draft-eligible volunteered
• Some draft-eligible were exempted
• ↓ earnings by $436 (relative to a mean of $16,461) in 1981
⇒This must mean military service decreased earnings later in life since having
your birthday drawn was random and only impacted earnings by making it
more likely that you served!
• Roughly 1 in 4 men who was had their birthdays drawn was randomly
drafted into service
⇒ effect of military service on earnings ≈ 4 x -$436 = -$1,744
Recall: Wrapping up the theory
• How immigration affects natives’ labor market outcomes depends on their
substitutability/complementarity and labor demand/supply elasticities
• If immigrants and natives are substitutes (and labor demand remains fixed), immigration reduces
wages of natives
• If immigrants and natives are complements and immigration increases the demand for native
labor, immigration increases wages of natives
• When there is more than one type of labor, immigration will hurt some workers and benefit others
• Immigration is not only a labor supply shock!
• Immigrants also consumers which increases labor demand for natives
• Immigrants may also complement natives and increase their productivity => increases labor demand
• Short-run effects likely larger than long-run effects as capital adjusts
Shifting towards empirics
• Effects of immigration far from obvious, but depend on:
• Whether immigrants and natives substitute or complement each other
• Whether immigration increases labor demand
• Skill composition of immigration
• Ambiguity of economic theory -> need empirical evidence
Labor market effects of immigration: Evidence
We are interested in answering:
Does immigration cause the wages of natives to go down?
We could look at employment/wages of natives in areas with lots of
immigrants to others with less.
What might be wrong with this approach?
Problems with this approach?
Immigrants may choose to go to
areas with already higher wages
and aren’t necessarily increasing
natives’ wages
Recall: Some form of
randomization is key to getting
us close to estimates of the
causal effects of immigration on
natives’ wages/employment
Source: “The Economics of Immigration” by
Bansak, Simpson, and Zavodny (2021)
Natural experiments for immigration
• What if we have an episode in which immigrants do not choose when
or where they move to/from?
• In this case, we won’t have the issue that immigrants choose when
and where to move to areas with stronger labor markets
• We could then compare:
• Wages/employment in the areas to which immigrants moved
• Wages/employment in the areas to which immigrants did not move
Recall: Quota System in the U.S. 1921-1965
Emergency Quota Act in 1921
• Annual quota of 360,000 European immigrants (800,000 entrants per year in the early 1910s)
• Slots set to 3% of the foreign-born stock for each nationality living in the U.S. as of 1910
Johnson-Reed Act of 1924
• Slots set to 2% of the foreign-born stock AND shifted base year to 1890
⇒In 1890, immigration levels were much lower, especially for immigrants from Southern and Eastern
Europe as they hadn’t migrated much by that point!
• From 1927 onwards, annual cap of 150,000 European immigrants per year
Abrupt change in policy.
Average of 200,000 Italians per year entered the U.S. (1900-1910) → annual quota ≈ 4,000
Note: No restrictions for migrants from Western hemisphere
(including North, Central, and South America)
Recall: Where were immigrants from at
different points in time?
British Isles were the main source of
immigrants in first half of 19th-century
Germans started arriving in 1840s
Starting in 1880, “new immigrants“ (Southern
and Eastern Europeans) represent majority of
inflows
1850: 90% of migrants from N/W Europe
(mostly from Great Britain, Ireland, and
Germany)
1920: 45% from N/W Europe and 41% from
S/E Europe (mostly from Italy, Poland, Russia,
Spain, and Portugal)
Source: Abramitzky, Ran and Leah Boustan. (2017). “Immigration in
American Economic History.” Journal of Economic Literature
Recall: Immigrant inflows to U.S. by level of
restriction 1900-1930 (high, low, and no restriction)
High restriction:
southern and
eastern Europe
Low restrictions:
Northern and
Western Europe
Source: Abramitzky et al. (forthcoming) – “The Effect of Immigration Restrictions on
Local Labor Markets: Lessons from the 1920s Border Closure”
“The Effects of Immigration on the Local Labor
Markets: Lessons from the 1920s Border Closure”
Abramitzky, Ager, Boustan, Cohen, and Hansen (2023)
Some cities were home to large populations of Southern/Eastern Europe
• example: Cleveland
Others were not and instead only had immigrants from Northern/Western Europe
• example: Cincinnati
Some cities thus more exposed to quota
⇒more exposed cities like Cleveland who received large inflows in the 1900s lost more
immigrant workers after restrictions
⇒cities like Cincinnati had steady (and lower) immigrant inflows over the period
Quotas -> foreign-born share ↓
14% in 1920, 5% in 1970
Exposure to quotas across cities in the U.S.
Effects of 1920s Border Closure
Empirical strategy: Compare cities like Cleveland that were strongly affected by the restrictions to cities
like Cincinnati that were not, before and after the policy change
Effects of restrictions:
• small positive effect on workers’ earnings initially (what simple theory would predict)
⇒eventually avoided paying higher wages, replaced European immigrants with immigrants from Mexico and
Canada and U.S.-born rural-urban migrants, Black Southern migrants
• In rural areas, less immigrants didn’t result in hiring of more U.S.-born farmers, but instead adoption of machines
(e.g., tractors). Farmers also shifted toward less labor-intensive crops (e.g., wheat)
Key takeaway: Effects smaller than what simple theory would predict as employers are highly adaptive.
‘They don’t necessarily turn to the “worker next door” to offer that worker a job when immigration
slows down, but instead search for other pools of labor.’
(Streets of Gold, pg 148)
Main methods
Q: which method is this paper most closely using?
1. Difference-in-differences (DID): compare the change in outcome
for treated group (before and after treatment) to change in
outcome for control group
2. Regression discontinuity (RD): treatment based on threshold/cutoff, compare those just above (treated) and just below (control)
3. Instrumental variables: find something else that randomly affects x
but not in any way related to y
1. Difference-in-differences (DID)
• Compare change in outcome ( , − , ) for “treated” (T) to change in outcome
( , − , ) for “control” (C)
Diff-in-diff= ( , − , ) – ( , − , )
In graph on left, treatment group
experienced bigger change in
outcome Y compared to control.
Difference in change in Y between
treatment and control provides
estimated effect!
Application of Difference-in-differences
Treated group: cities w/ many Southern and Eastern European migrants
(more exposed to quotas)
Control group: cities w/ less Southern and Eastern European migrants and
more Northern and Western European migrants (less exposed to quotas)
Before: Before quotas
After: After quotas
Mexican Bracero farmworkers program
• Agreements between U.S. and Mexico to
regulate flows of labor from 1942-1964
• Provided temporary contracts to Mexican
farm workers
• 1951-1954, 3000k Mexican workers
authorized to enter U.S. each year,
primarily to plant and harvest food in
California and neighboring states
• Lyndon B. Johnson ended program with
aim to benefit native workers in 1965
“…We are now embarked on a major
recruiting effort to attract unemployed
American workers to fill seasonal farm jobs.”
Mexican workers arriving by train in 1942 as
part of the Bracero program.
“Immigration Restrictions as Active Labor Market
Policy: Evidence from the Mexican Bracero Exclusion”
Clemens, Lewis, and Postel (2018)
• Ending of program and repatriations, led to large decreases in # of Mexican
immigrant farmworkers
• Compare states that formerly relied heavily on Bracero workers (e.g., Arizona
and California) to states that did not (e.g., Pennsylvania) before and during the
phaseout of the Bracero program
hourly wages for farmworkers
inched up steadily during the
1960s in all states, with no
notable acceleration in states
that had been reliant on Bracero
workers
Main findings
• Effects of removing Bracero farmworkers:
• hourly wages for farmworkers inched up steadily during the 1960s in all states, with no
notable acceleration in states that had been reliant on Bracero workers
• So, what happened?
• Some Mexicans began crossing border “illegally” on their own (many former Bracero
contract workers)
• Some farmers shifted away from crops like lettuce, asparagus, and strawberries (crops
that could only be picked by hand) towards crops like cotton and tomatoes for which
harvesting technology was available
• Farmers already producing crops like tomatoes and cotton replaced labor with
machines
Substitute labor with machines
Harvesting machines were
available since late 1950s (before
Braceros were repatriated)
States like California that were
dependent on Braceros saw large
increases in mechanization
Bracero exclusion and the variety of food
• Not surprisingly, food was pretty bland
in the 1960s and 70s (see frozen dinner
on the right)
• Decades later with growth of the H-2A
visa program, a temporary ag. visa
(similar to Bracero program), the 2000s
was an era of food discovery.
(avocados, organic berries,
microgreens)
Mariel boatlift
• Scarface opening scene:
https://www.youtube.com/watch?v=rPBss
c1epXw
• Fidel Castro encouraged 125k to emigrate
from Cuban to US
• Cuba was closed country. In April of 1980,
Castro decided to open port of Mariel to
emigration. Anybody that wanted to leave
could. Castro also let 3k prisoners go.
• Most didn’t have a HS deg. and worked in
fields like construction or domestic service
• Marielitos arrived (and most stayed) in
Miami
⇒Increased Miami labor force by 7%
“The Impact of the Mariel Boatlift on the Miami
Labor Market”
Card (1990)
Empirical strategy: compares wages/employment of natives in Miami
before and after Mariel Boatlift to what happened in other similar cities
(Tampa, Atlanta, Houston, and Los Angeles)
Effects of Mariel Boatlift on unemployment rates
Results suggest that influx of Cuban refugees had almost no effect on
employment rates for natives (if anything it decreased unemployment rates)
Findings from Card’s Mariel Boatlift Paper
Influx of Mariel immigrants had:
• no effect on unemployment rates of natives, including the less-skilled
(HS deg. or less)
• no effect on wage rates for natives, including the less-skilled (HS deg. or
less), or Cuban migrants
• slight increase in unemployment among Cuban immigrants (can’t
distinguish between Marielitos and previously arrived Cuban migrants)
• Might just be catching higher unemployment rates of Marielitos
• OR previous Cuban migrants lost jobs because of new arrivals
How did the influx of so many refugees not
affect natives, especially the less-skilled?
• Some workers decided to leave Miami to avoid facing more competition
• Miami had experience absorbing migrants from Cuba and Central America
• High concentration of textile and apparel industries ⇒ employ lots of
Marielitos quickly
• High concentration of Hispanics ⇒ lack of English less of an issue
• Miami was slower to adopt new automation technologies like self-checkout
at grocery stores or ATMs at banks. They had plenty of workers, so it wasn’t
necessary. In Miami native workers had to compete with Marielitos, while
in other cities they had to compete with machines.
Back and forth on Mariel boatlift
• Borjas (2017): The Wage Impact of the Marielitos: A Reappraisal
• finds negative impacts for those w/o HS deg. (Card defines low-skill as HS
deg. or less)
• Peri and Yasenov (2019): The Labor Market Effects of a Refugee
Wage: Synthetic Control method Meets the Mariel Boatlift
• dispute Borjas and re-confirm Card’s findings
• point out issues with sample selection in Borjas’s study
• Clemens and Hunt (2019): The Labor Market Effects of Refugees:
Reconciling Conflicting Results
• try to reconcile and conclude that Borjas’s findings were based on a small
sample and results not representative of low-skilled population
Next class: additional evidence on labor
market effects of immigration
Read:
BSZ Chapter 8: Labor market effects of immigration: evidence
Might also help to read introductions of the following:
• “The Supply Side of Innovation: H-1B Visa Reforms and U.S. Ethnic Invention”
Kerr and Lincoln (2010)’
• “The Collapse of the Soviet Union and the Productivity of American Mathematicians”
Borjas and Doran (2012)
• “STEM Workers, H-1B Visas, and Productivity in U.S. Cities” Peri et al. (2015)
Also, if you’re interested, this paper provides a nice review:
“Immigrants, Productivity, and Labor Markets” by Giovanni Peri (2016)
Purchase answer to see full
attachment
