Agent-Based Modeling Simulations for Solving Pakistan's Urban Challenges by D...
Population Mobility and Monsoon Anomalies in Pakistan by Katrina Kosec, IFPRI
1. Population Mobility and Monsoon Anomalies in Pakistan
Presented by
Katrina Kosec
December 13, 2012
2. Main Research Questions
• What individual and household characteristics predict
migration?
– The opportunity costs of migrating vary across types of
people and households
– Ability to leave home also varies (e.g., security concerns,
gender norms)
– Who is migrating, who is not, and what predicts migration?
• How do climate shocks in particular affect the
prevalence of migration?
– Recent climate shocks (e.g. 2010 and 2011 floods) and
global warming may be changing where and how
Pakistanis live and work. In what ways?
3. Motivation: Migration Is An Important
Tool for Improving Household Welfare
• Migration can help smooth income and consumption risk
(Rosenzweig and Stark 1989)
• Migration can better match individuals with work
opportunities and motivate human capital investments
(Schultz 1961)
• Migrants generate positive income shocks that lead to
enhanced human capital accumulation and entrepreneurship
in origin households (Edwards and Ureta 2003; Yang 2005)
• This can be especially important in settings with variable
incomes (e.g., rural areas highly dependent on agriculture)
4. Rural Households in Pakistan are Highly
Dependent on Agriculture
• Only 30% of households
in rural Pakistan are
Rural Non- completely non-
Agricultural agricultural (Rural
Households Landowners
30% 38% Household Panel Survey,
2012)
• Thus, natural disasters
Agricultural and monsoon anomalies
Waged
Labor Tenants have the potential to
21% 11% have a major impact on
rural livelihoods, and to
motivate migration
5. Example of Vulnerability: Severe Floods
of 2010 and 2011
• In 2010, floods affected over 20 million people
(Pakistan Ministry of Finance 2011)
– 14 million people displaced, 3.3 million living in camps or
roadside settlements 2 months afterward (D. Walsh, The
Guardian, 2011)
– Estimated 1 billion USD of crop value destroyed (IFRC, 2011)
– Estimated 10 billion USD in total damages (Ministry of Finance,
2011)
• In 2011, floods affected 9.6 million people (Ministry
of Finance 2012)
– Almost 4 billion USD in total damages (Ministry of Finance 2012)
6. Literature on the Impacts of Climate on
Labor and Migration Patterns
• Rosenzweig and Stark (1989) show that Indian HHs exposed to higher
agricultural income risks tend to have longer-distance marriages
• Halliday (2006) shows that adverse agricultural conditions in El Salvador
increase migration
• Gray (2009) finds that international migration in rural Ecuador
decreased with agricultural and rainfall shocks, while local mobility and
internal migration increased with variation in rainfall
• Gray and Mueller (2012) find that men’s labor migration in Ethiopia
increases with drought; women’s migration decreases (revealing gender
differences in responses)
• Jayachandran (2006) finds that landless individuals experiencing a small
loss of production are more inclined to migrate in response to a shock
than are those with land (poverty level differences in responses)
7. Data Sources
• Survey Data: households are tracked over 21 years
– 1991: Data collected fpr Round 14 of IFPRI’s Pakistan Rural
Household Survey
– 2001 and 2012: Same households tracked by PIDE (2001)
and IDS/ IFPRI (2012)
– We create a person-year dataset, using all individuals ages
15-40 (“at risk for migration”)
• Weather station data from the Pakistan Metrological
Department
– Total rainfall during the monsoon (in 100s of mm)
– Date of monsoon onset (1 = June 1st, 2= June 2nd, …)
8. Migration Rates (Ages: 15-40)
Men Women
Left household, but stayed in village 1.51 2.13
Left household and village 1.34 2.25
TOTAL 2.85 4.39
9. Reasons for Migration (Ages: 15-40)
• Marriage or setting up a new household are the most
common reasons for both genders
• Men are much more likely to migrate for employment
Men Women
1%
11% Employment
20% 21%
Marriage/ new
household
Other
59% 88%
10.
11. Factors That Predict Migration – Linear Probability
Model Analysis
Dep. variable: Individual
migrated (mean = 0.0362) Coeff. Sig. S.E. Coeff. Sig. S.E.
Male -0.0163 *** 0.0020 -0.0168 *** 0.0017
Age 0.0007 *** 0.0002 0.0009 *** 0.0002
Head or spouse -0.0386 *** 0.0045 -0.0371 *** 0.0049
Female head 0.0025 0.0069
Age of head -0.0001 0.0001
Years of education of head -0.0007 0.0005
# Children 0.0022 *** 0.0005
Owned land (10’s of hectares) -0.0008 * 0.0004
Total assets 0.0001 0.0000
% of owned land irrigated -0.0072 * 0.0040
Annual monsoon rainfall (100s mm), t-1 0.0032 *** 0.0010 0.0031 *** 9E-04
Monsoon start date (1=June 1st), t-1 0.0022 *** 0.0008 0.0021 ** 8E-04
Household FE? No Yes
Individuals 4,574 4,574
Notes: Standard errors are clustered at the village level. *** p<0.01, ** p<0.05, * p<0.1.
12. How does the Probability of Migration Vary
with Individual and HH Characteristics?
• Being female: 45%↑
• Owning an additional 10 hectares of land: 2%↓
• One year older: 2%↑
• One more dependent child in the household: 6%↑
• Not being the household head: twice as likely to migrate
• Having all land irrigated (as opposed to none): 20%↓
• Does not affect migration: Age, gender, and education level of
household head in an individual’s household; total value of
assets
13. Main Findings: Effects of Monsoon
Anomalies on Migration
• Higher rainfall during the monsoon increases migration
– 1 S.D. increase in monsoon rainfall last year (i.e. 271 mm
more rain during Jun.–Sept.) 0.9 percentage point
increase in the probability of migration (24% increase)
• A delayed monsoon onset increases migration
– 1 S.D. increase in the start date of the monsoon last year
(i.e. a 25 day delay) 0.5 percentage point increase in the
probability of migration (15% increase)
• Not shown: Income is decreasing in monsoon rainfall,
using data from 1986-1991 IFPRI Panel
– Consistent with use of migration to mitigate income risk
14. Conclusions
• There are real impacts of negative climate shocks on
migration; monsoon anomalies (more rainfall or delayed
monsoon) increase migration
• Being female, older, having less land, and having more
dependents is associated with increased migration
• Implications?
– Policymakers should view migration as a coping mechanism for
negative weather shocks
– Land/ assets may reduce access to/ use of this coping
mechanism
• Next Steps:
– More systematic analysis of migration patterns (by gender, by
distance of move, and by motivation of move)
– Incorporating more and better climate data
– Analysis of underlying factors associated with too little migration