RuralUrban Migration and Poverty

1
Rural-Urban Migration and Poverty

• Bruce Weber
• Oregon State University
• West Coast Poverty Center Seminar Series
• January 28, 2008

2
Acknowledgments

• Co-authors
• Alexander Marré and Monica Fisher
• Oregon State University
• Financial Support
• USDA ERS Cooperative Agreement 58-5
• Agricultural Experiment Station, Oregon State
  University
• Rural Studies Program, Oregon State University

3
Motivating question

• What is the best strategy for enhancing income
  and reducing poverty of rural families ?
• Investing in education of youth
• Improving local economic conditions
• Encouraging rural to urban migration
• Previous research focus on local economic
  conditions.
• Migration is new focus for us
• Concern about brain drain, dooming rural poor
  by trapping them in rural places

4
Previous Studies

• Education increases income and reduces poverty
• Improved local economic conditions increase
  income and reduce poverty (Haynie and Gorman)
• Rural- to- urban migration increases income and
  reduces poverty (Mills and Hazarika, Rodgers and
  Rodgers)

5
But.

• Migration may be endogenous to income and poverty
  (unobserved factors may affect both migration and
  income/poverty)
• If migration is endogenous, a correction is
  needed to produce consistent estimates

6
Test for endogeneity of migration

• Endogeneity if the unobserved factors that
  influence migration also influence income
• Rivers Vuong approach
• Residuals from the migration model are
  observations of the unobserved factors
• If the residuals explain income, then there is
  some evidence that migration is endogenous
• Statistical significance of the residuals
  p-value 0.006 in the real income model

7
Endogeneity correction Instrumental Variable

• We need a variable that is highly correlated with
  the migration decision and uncorrelated with the
  error term in the income model growing up in a
  rural area is expected to be correlated with
  migration by not with income/poverty
• Household head Did you grow up on a farm, in a
  small town, in a large city, or what?
• Grew up Rural statistical significance
• Migration model p-value 0.005
• Income models p-values 0.281, 0.224

8
Education, Migration, Poverty
9
Data Panel Study of Income Dynamics

• Panel Study of Income Dynamics (PSID)
  longitudinal survey that follows a representative
  sample of 5,000 families since 1968
• We use the 1993 and 1999 waves
• Public data include variables for nonmetro/metro
  residence of household and county unemployment
  rate in 1993
• Need confidential data for better model
  specification

10
Sample Working-age Rural Household Heads

• N 701 household heads who are working age (25
  to 64 years) and reside in nonmetro counties in
  1993
• Average age 39 years old in 1993
• Mostly White (76) and male (81.9)
• 70 are married, with 12 changing marital status
  between 1993 and 1999 through marriage or divorce
• Average family size 3

11
Educational Attainment
12
Poverty, Income, Migration

• Poverty 11 are poor in 1993 and 8.8 are poor
  in 1999,
• Average household income 49,922.87
• Cost of living is likely lower in rural areas
• Housing costs are ¼ of household expenditures
  thus ¼ of income is adjusted by nonmetro/metro
  differences in cost-of-housing (Fair Market Rent
  Index at state level)
• Average real income 53,529.68
• Migration 18.8 are rural to urban migrants
  between 1993 and 1999.

13
Migration by Educational Attainment
14
Migration Model

• Probit Model of Nonmetropolitan to Metropolitan
  Migration
• Includes individual, family, and regional
  characteristics and instrument (grewuprural)
• Migration a0 a1highschool a2college
  a3postgrad a4age a5agesq a6gender a7race
  a8chgmarital a9familysize a10unemployment
  a11region a12grewuprural e
• In an alternative specification, education is
  defined as years of formal education

15
Poverty Model

• Probit Model of Poverty status of household poor
  v. nonpoor
• Includes individual, family, regional
  characteristics and predicted migration
• poverty ß0 ß1education ß2age ß3agesq
  ß4gender ß5race ß6chgmaritalstatus
  ß7familysze ß8region ß91993poor
  ß10pmigration ?

16
Income Model

• Ordinary Least Squares Model of 1999 Household
  Income and real income
• Includes individual, family, regional
  characteristics and predicted migration
• Income a0 a1highschool a2college
  a3postgrad a4pmigration a5age a6agesq
  a7gender a8race a9married a10familysize
  a11region e

17
Results Migration Model (1 of 2)
18
Results Migration Model (2 of 2)
19
Results Poverty Model (1 of 2)
20
Results Poverty Model (2 of 2)
21
Results Income Model (1 of 2)
22
Results Income Model (2 of 2)
23
Conclusions

• Education of the household head influences the
  decision to migrate to an urban area more
  educated heads are more likely to move
• The migration decision of household heads is
  endogenous to income
• Observed migration influences income, but
  correcting for endogeneity results in a
  statistically nonsignificant effect of migration
  on income

24
Conclusions on financial returns to education and
migration

• Returns to education are significant and large
  rural household heads with a high school diploma,
  college, or postgraduate education all have
  higher household incomes than those without a
  high school diploma
• These returns are higher whether they move to
  urban areas or not
• Returns to rural-urban migration are not
  significant once the endogeneity of migration is
  controlled for people appear to move for other
  reasons than higher incomes.

25
Future Research

• Confidential geocodes will make possible research
  on
• Greater number of migration choices by type
• Rural to rural rural to urban
• Urban to urban urban to rural
• More specific community characteristics
• Labor market and industry structure
• Community indicators
• Environmental amenities

26
Future Research

• Confidential geocodes will make possible research
  on
• Longer and shorter time spans
• Role of rural and urban economic differences
• Short-term versus long-term moves
• Consider different time periods with different
  macroeconomic conditions it may be that the less
  stark differences in rural and urban economic
  opportunity in 1990s is responsible for the weak
  association between migration and poverty during
  this era

27
Results Alternative Migration Model (1 of 2)
28
Results Alternative Migration Model (2 of 2)
29
Migration and Poverty

• 45 people (6.4) moved out of poverty between
  1993 and 1999 only 5 out of the 45 were
  rural-to-urban migrants
• 29 people (4.1) moved into poverty between 1993
  and 1999 only 2 of the 29 were rural-to-urban
  migrants