economic mobility and migration

56
Economic Mobility and Migration Economics 2333 Class 11 Spring 2014 Robert A. Margo

Upload: jadyn

Post on 23-Feb-2016

65 views

Category:

Documents


0 download

DESCRIPTION

Economic Mobility and Migration. Economics 2333 Class 11 Spring 2014 Robert A. Margo. Outline. Background Ferrie and Long Olivetti and Paserman Abramitzky , Boustan , and Eriksson (student presentation) Salisbury (if time). Mobility. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Economic Mobility and Migration

Economic Mobility and Migration

Economics 2333 Class 11 Spring 2014Robert A. Margo

Page 2: Economic Mobility and Migration

Outline

• Background• Ferrie and Long• Olivetti and Paserman• Abramitzky, Boustan, and Eriksson (student

presentation)• Salisbury (if time)

Page 3: Economic Mobility and Migration

Mobility• Life-cycle and across generation Across generation:

intergenerational transmission of inequality. • Ln Y (child) = a + b ln Y (parent). B = intergenerational correlation

coefficient.• Role of geographic mobility: within and across countries• “Assimilation” of immigrants: does second generation converge?• Another important example: black-white income differences• C19: Traditional view is that US offered more upward mobility

than Europe.• Upward mobility: higher wages, cheaper land, (possibly) more

occupational upgrading.

Page 4: Economic Mobility and Migration

Ferrie• Numerous studies by sociologists and historians examining

“occupational mobility” across generations in a particular place.• In history, pioneered by Thernstrom. Link individuals over time

or fathers-sons.• Severe methodological problem: if studying a single community,

outmigration may be selective.• Joe Ferrie (Northwestern): census record linkage. Very popular at

the moment, much cheaper because of computer linkage and ancestry.com

• Ferrie and Long: samples and methods to measure occupation mobility across generations. Compare over time and across countries.

Page 5: Economic Mobility and Migration

Census Linkage• Invented by historians but developed in economic history by Ferrie,

Atack, and others.• Create sample of census in year t. Find the same individuals in year

t+ z, where z = 10 or more. Earliest year is 1850, latest is 1940.• Matching can be by hand (slow) or by computer (fast). Matches

subject to Type #1 and Type #2 error.• Criteria for matching: last name, first name, place of birth, date of

birth, other family members.• Generally cannot be done for women because of name changes at

marriage.• Historical match rates generally low (can be much better for

uncommon names but this may create selection bias).

Page 6: Economic Mobility and Migration

Ferrie and Long• Intergenerational mobility in C19: Was the US exceptional?• Comparison is with England. Similar census-based samples. Base

sample consists of young men living at home with fathers. Locate son in later census. Compare occupations of fathers and sons in an N x N contingency table. Develop statistical method for assessing which tables exhibit greater occupational mobility.

• Focus on occupational mobility comes from sociology and also data limitations (no C19 information on income). Could use IPUMS-type occupation status but presumably the argument is that the contingency table is more informative.

• No formal model, but discussion has the flavor of Becker and Tomes, and Solon. In BT, and Solon, intergenerational correlation reflects various factors:

Page 7: Economic Mobility and Migration

The Data• Samples of males linked across censuses from 1861-1881 &

1881-1901 in Britain, and males linked from 1860-1880 & 1880-1900 in the U.S.

• Linkage based on (i) name, (ii) year of birth, (iii) parish & county (Britain) or state (U.S.) of birth.

• Individuals were 30-39 years old in the terminal year and were observed with their fathers in the initial year.

• Fathers’ & sons’ occupations observed at same life-cycle point.

Page 8: Economic Mobility and Migration

Paper: tinyurl.com/Migration2013 Slides: tinyurl.com/MigrationSlides2013

Measuring Intergenerational Mobility

• The conventional approach:• ln Yi

Son = β ln Yi Father + εi

• where β is intergenerational income elasticity• But we’ve only got occupations, and they’re

difficult to order unambiguously

Page 9: Economic Mobility and Migration

Paper: tinyurl.com/Migration2013 Slides: tinyurl.com/MigrationSlides2013

Measuring Mobility with Occupation Across CountriesOccupational Classification

White Collar Professional etc., occupations(accountant, lawyer, doctor,clergyman) and Intermediate occupations (bookkeeper, manager, farm foreman)

Skilled & Skilled occupations (blacksmith,Semi-Skilled tailor, weaver, craftsman and Semiskilled

occupations (machinist, factory operative)

Farmer Farm Operators

Unskilled Unskilled Laborers (general laborer, servant, farm laborer)

Page 10: Economic Mobility and Migration

Paper: tinyurl.com/Migration2013 Slides: tinyurl.com/MigrationSlides2013

Measuring Intergenerational Mobility

MP = 3/8 MQ = 7/10

MP = 3/8 MQ/ = 5/8

Page 11: Economic Mobility and Migration

Paper: tinyurl.com/Migration2013 Slides: tinyurl.com/MigrationSlides2013

Measuring Intergenerational Mobility

Cross-Product Ratios: (3 x 2) / (2 x 1) = 3 for P (2 x 1) / (6 x 1) = 1/3 for Q

Page 12: Economic Mobility and Migration

Paper: tinyurl.com/Migration2013 Slides: tinyurl.com/MigrationSlides2013

Measuring Intergenerational Mobility

Cross-Product Ratio for Q = CPR for Q' = 1/3

Page 13: Economic Mobility and Migration

Paper: tinyurl.com/Migration2013 Slides: tinyurl.com/MigrationSlides2013

Measuring Intergenerational MobilityFor tables > 2 × 2, use the “Altham statistic,”which uses all of the cross-product ratios:

Measures distance between mobility in P andmobility in QAlso calculate d(P,J), d(Q,J): distance between P/Q and J, a matrix of 1’s representing perfect mobility

Page 14: Economic Mobility and Migration
Page 15: Economic Mobility and Migration
Page 16: Economic Mobility and Migration
Page 17: Economic Mobility and Migration
Page 18: Economic Mobility and Migration
Page 19: Economic Mobility and Migration
Page 20: Economic Mobility and Migration
Page 21: Economic Mobility and Migration
Page 22: Economic Mobility and Migration
Page 23: Economic Mobility and Migration
Page 24: Economic Mobility and Migration

Follow Up

• Ferrie-Long does NOT take account of British migrants to the US. More recent work does.

• Consider British migrants to the US. How much intergenerational mobility did this group experience? How did their mobility experience compare with that of non-migrants in both countries?

• Use 2 cohorts of British movers and stayers (1861-1880 & 1881-1900), observing (1) migrants before & after departure and (2) non-migrants before & after the migrants left

Page 25: Economic Mobility and Migration

The Context

• Migration was completely unrestricted at this time (before the Quota System of the 1920s)

• Driven not by desperation (c.f. Irish Famine migrants) but by “normal” forces (e.g. relative wages)

• The British were a large fraction of the migrant stream (close to 40% in some years), but their share moved opposite the total volume of migration

Page 26: Economic Mobility and Migration

The Context

• The Britain each cohort left behind was a decade or more ahead of the U.S. in its industrialization

• More opportunity in the U.S. for those squeezed out by changes (consolidation in farming, displacement of craft workers by factories and machines)

Page 27: Economic Mobility and Migration

Additional Data

• For comparable data on migrants from Britain to the U.S., Ferrie-Long generated 2 new samples:

• British-born males age 30-39 in the 1880 U.S. Census of Population linked back to the 1861 British Census

• British-born males age 30-39 in the 1900 U.S. Census of Population linked back to the 1881 British Census

Page 28: Economic Mobility and Migration

The Data• Main challenge: Lack of specific birthplace info for migrants in U.S.

censuses• Requirements/Checks (1861 1880):

– Unique record (name, age birthplace) in 1880 U.S. census and 1861 Br census

– Not present in British 1881 census

– Not present in U.S. 1860 census index

– If they were present in the 1870 U.S. census index, they were not also present in the 1871 British census index, and if they were present in the 1871 British census index, they were not also present in the 1870 U.S. census index.

– Oldest U.S.-born child in 1880 was born after 1860

– Youngest Britain-born child in 1880 was born before 1862

Page 29: Economic Mobility and Migration

The Data

• U.S. samples: 4,138 (1860-1880) &• 3,919 (1880-1900)• British samples: 2,039 (1861-1881) &• 4,071 (1881-1901)• Migrant samples: 1,176* (1861-1880) &• 1,144 (1881-1900)

• * 2,174 linked; remainder awaiting occupational transcription

Page 30: Economic Mobility and Migration

British Migrants vs Non-Migrants

Non-migrants (“stayers”) Migrants (“movers”)

Page 31: Economic Mobility and Migration

Mobility Measures

Page 32: Economic Mobility and Migration

Visualizing Mobility Differences

Page 33: Economic Mobility and Migration

Paserman and Olivetti• Traditional estimates of intergenerational correlation coefficient (or

contingency table) rely on fathers and sons. Daughters too difficult to link.• PO advocate a new (and creative) methodology: use first names.

Assumption is that first names carry economic information.• Use two cross sections. In first cross section, regress income of fathers on

full set of dummy variables of children’s names. • In later cross section, regress incomes of adults with name j on predicted

income of fathers who name their children j from the first regression.• Advantages (a) no record linkage (b) can be done for daughters , sons-in-

law, across multiple generations.• Does this really work? Amazingly, YES.• PO sketch simple model of intergenerational transmission that includes

marriage. Yields estimating equation for son’s earnings and sons-in-law.

Page 34: Economic Mobility and Migration
Page 35: Economic Mobility and Migration
Page 36: Economic Mobility and Migration
Page 37: Economic Mobility and Migration
Page 38: Economic Mobility and Migration
Page 39: Economic Mobility and Migration
Page 40: Economic Mobility and Migration
Page 41: Economic Mobility and Migration

Data Analysis

• Rely on IPUMS, use OCCSCORE (1950 median income is the occupation weight). Assess sensitivity to alternative measures.

• Coding of names: benchmark ignores middle initials AND also treats nicknames as separate names (William vs. “Bill”). Again, extensive sensitivity analysis.

Page 42: Economic Mobility and Migration
Page 43: Economic Mobility and Migration
Page 44: Economic Mobility and Migration
Page 45: Economic Mobility and Migration
Page 46: Economic Mobility and Migration

Salisbury• Context: relationship between migration across US states in C19 and initial level of wages. • Lebergott (1964) observed that migration tended to go in the direction of higher wages, but

correlation was low. His conclusion: migrants “sub-optimized”.• Simple problems (a) nominal, not real wages (b) population growth, not migration BUT there

are other similar migration anomalies in US econ history.• Salisbury: local labor markets offered two paths to higher income. Path #1: higher income.

Path #2: greater occupation mobility.• Migrants selected based on various factors not observable to the econometrician. One of

these may be “ambition” or “ability” that is complementary to upward occupation mobility. Individuals with this characteristic will leave places where upward mobility is less likely and move to places where it is more likely.

• Hypothesis: skill premium (ratio of skilled to unskilled wages) is a useful, measureable signal. If skill differential is high, relative supply of skill is low and upward mobility may be more likely. So, high ambition/ability individuals are less likely to leave such places but more likely to move to them.

• Construct linked sample from IPUMS and match to data on unskilled wages and skilled (carpenter) wages from Margo (2000); this is from the census of social statistics.

Page 47: Economic Mobility and Migration

Data

• Linked census samples 1850-60 and 1860-70.• Basic data: white males ages 15-60 who

worked as an unskilled laborer or farm laborer in year 0.

• S1: wage data in home and destination counties. Includes migrants and non-migrants.

• S2: cross-county migrants only, must have wage data in destination county.

Page 48: Economic Mobility and Migration
Page 49: Economic Mobility and Migration
Page 50: Economic Mobility and Migration

Basic Results• Basic results in Tables 3-5.• Table 3: people who left counties with low skill premiums

and moved to counties with high skilled premiums tended to experience occupational upgrading. NOTE: on average, places with high skill premiums tended to have lower average real wages (more rural).

• Table 4: conditional on migrating, occupational upgrading is increasing in the skill premium of the county of destination.

• Table 5: similar pattern of results, although not very significant, if farmers are excluded.

Page 51: Economic Mobility and Migration
Page 52: Economic Mobility and Migration
Page 53: Economic Mobility and Migration
Page 54: Economic Mobility and Migration
Page 55: Economic Mobility and Migration

Table 9• Table 9 runs the previous regression in reverse. Dependent variable is

now the destination skill premium. Key independent variable is expected occupational upgrade. This is the actual upgrade, but 2SLS using demographic variables as instruments. Idea is that these help predict upgrading but are not obviously correlated with skill premium in destination.

• Results suggest that selection bias was present.• Bottom line: some unskilled migrants went were the unskilled real

wage was highest but others went where the possibility of occupational upgrading was highest. These places generally had a high skill premium but not necessarily a high unskilled wage. In the aggregate we get the pattern noted by Lebergott.

• VERY likely relevant to international migration, but not studied thus far.

Page 56: Economic Mobility and Migration