Work Migration and Poverty Reduction in Nepal

1
Work Migration and Poverty Reduction in Nepal

• Michael Lokshin, Mikhail Bontch-Osmolovski, Elena
  Glinskaya
• DECRG-PO and SASPR
• The World Bank

2
Country background

• Poorest country of South Asia
• Per capita GDP 240 (1,420 PPP)
• Maoist Insurgency and political instability
• Diminishing export markets, difficult economic
  situation
• But
• Between 1995 and 2004 poverty declined from 42 to
  31 percent
• Per capita expenditure grew 40 percent in real
  terms
• Why?

3
Work migration and remittances

• Increase in work migration and remittances
• About 1M prime age males work outside Nepal
• The proportion of households receiving
  remittances has increased from 24 percent in 1995
  to 32 percent in 2004
• Remittances grew at 30 per year, from 3 of GDP
  in 95 to 15 of GDP in 04
• Official statistics 1 billion comes in Nepal as
  remittances. Unofficial statistics even larger

4
Migration and remittances in Nepal

• The history of Nepal foreign employment dates
  back almost 200 years. Gurkhas.
• The Foreign Employment Act of 1985 officially
  recognized the benefits of overseas migration.
• Labor migration from Nepal extended from India to
  the countries of Southeast and Far East, and to
  Arab countries of the Gulf.
• Internal migration predominantly rural-to-rural
  migration is twice as large as rural-to-urban
  migration. Mid-West and Far-West experienced net
  out-migration. Maoists insurgency. Malaria.

5
Research on effects of migration and remittances
on poverty

• The empirical research of the impacts of work
  migration and remittances on poverty and
  inequality is limited.
• Most studies indicate that increase in
  remittances leads to decline in poverty. Lesotho
  (Gustafson and Makonnen 1993) Egypt (Adams
  1995) Nicaragua (Barham and Boucher 1998)
  Uganda, Bangladaash, Ghana (Adams 2005) China
  (Yang et al 2005)
• No formal studies for Nepal. Few mostly
  descriptive works by the local researchers.
• Our paper provides the first formal evidence of
  the impact of work migration and remittances on
  income distribution in Nepal.

6
Data

• Use two rounds of Nepal Living Standard Survey
  (NLSS).
• First round June 1995- June 1996 3,373
  households in 274 PSUs
• Second round April 2003-April 2004 3912
  households in 326 cross-sectional and 95 panel
  PSUs.
• 2001 Nepal Census data

7
(No Transcript)
8
(No Transcript)
9
Non-parametric regression of incidence of
migration and amount of remittances by lagged
asset index
10
(No Transcript)
11
(No Transcript)
12
Work Migration and Remittances

• Q1 What is the impact work migration?
• Q2 What is the impact of remittances?
• A1 Compare the current expenditure distribution
  with the counterfactual distribution of less(no)
  migration (that also means no remittances).
  Counterfactuals household with a migrants
  similar household with no migrate.
• A2 Compare the current expenditure distribution
  with the counterfactual distribution of no
  remittances. Counterfactuals household with
  migrant who sends remittances household with a
  migrant who sends no remittances. All other
  consequences of migration are disregarded.
• A1 Important policy implications Easily
  affected by the government policies.
• A2 Limited policy implications Hard to measure.
  Not that simple to affect.
• This paper focuses on Q1

13
Theoretical framework (1)

• Effect of work migration on household wellbeing
• Change relative productivity of members of the
  sending households.
• Labor market implications for the household
  members
• Affects health and educational attainments, etc
• Remittances as the most tangible benefits of
  migration
• enable households to overcome credit and risk
  constraints on their ability to engage into the
  modern and more productive activities.
• Investment in housing and schooling
• Direct consumption, etc.
• The observed consumption behavior is the result
  of all these effects.

14
Theoretical framework (2)

• Main assumption
• Three states of migration Migration abroad,
  Internal migration, No migration.
• Every household has a choice to send its member
  to work abroad or inside Nepal.
• Migration has to be planned ahead.
• Two period model of utility maximization
• Period 1 Households compare expected net
  benefits (in period 2) in each state of migration
  and select the state providing highest utility.
• Period 2 Households observe the realized labor
  market outcomes a migrant inform his household
  about his wages and remittances. The household
  decides on the LFP of its members, investment
  decisions, adjusts levels of consumption.

15
Theoretical framework (3)

• Presence of unobserved factors that would
  simultaneously affect the migration decision and
  consumption decisions.
• Selection of household into migration states
  could be non-random.
• The challenge for our empirical strategy is to
  estimate our model controlling for such
  unobserved factors and selectivity.
• Need an instrument to identify the non-random
  selection.

16
Identification strategy

• Theory ? some conditions that affect migration
  decision in period 1 have no effect on
  consumption in period 2.
• Two instruments (standard in the field)
• A proportion of abroad migrants in a ward in 2001
  (based on 2001 Census). A proxy for ward-level
  migration networks, affects cost of abroad
  migration.
• A proportion of internal migrant in a district in
  1995 (NLSS 1995). Affects cost of internal
  migration.
• Theory ? Applied only for households that have a
  choice to send a migrant. This imposes
  restrictions on our sample.

17
Empirical model

• Indirect utility function for state s
• Choice of migration state
• Consumption in a particular state
• Five-variate normal distribution

18
Empirical specification

• Households with a working migrants are identified
  by the reported remittances. We have no
  information on characteristics of the migrants
  except their age.
• Among abroad destination we cannot differential
  between India and other countries (small sample).
• Explanatory variables can only include variables
  that are not affected by migration household
  productive characteristics, demographics,
  education of females.
• Set of characteristics of the localities labor
  market, literacy. To control for
  community-specific characteristics. See Table 2.

19
Empirical results choice of migration state
(Table 3)

• Households living in wards with historically
  higher proportion of external migrants are
  significantly more likely to migrate abroad.
• Households from the districts with larger shares
  of internal migrants are more likely to send
  their members to work inside Nepal.
• Large households, households with a higher
  proportion of adult males and households with
  less educated females are more likely to have
  their member working outside Nepal.
• Newars are more likely to migrate within Nepal.
  Dalits prefer to send their members abroad.
• Households with large plots are more likely to
  have a migrant.
• The probability to have a migrant is lower among
  the low-wealth households.
• Individuals residing in Katmandu are less likely
  to migrate relative to respondents living in
  other areas of Nepal. Households from Rural
  Western Mountains and Hills are more likely to
  migrate.
• Households living in wards with higher proportion
  of the illiterates are less likely to migrate.
  Households from the wards with a large share of
  wage employment are less likely to send members
  to work inside Nepal.

20
Empirical results consumption equation(Table 4)

• The observed household characteristics play more
  import role in determining the level of
  consumption in no-migrant households compared
  with households with a migrant.
• Households with larger shares of children 0 to 8
  years old have lower per capita consumption
  relative to households with older or no children.
  
• Household with larger shares of all groups other
  than adult males are correlated with lower levels
  of per capita consumption.
• Households with better-educated females enjoy
  higher levels of per capita consumption.
• The size of a land plot has a positive and
  significant impact on consumption of households
  with no migrants and with internal migrants. For
  households with external migrants, those
  possessing more than two hectares of land have
  significantly higher per capita consumption
  relative to landless households.
• Households from the upper percentiles of the
  index have higher per-capita expenditure
  regardless of their members migration status.
• Households receiving pensions are better off in
  all three migration groups.

21
Simulations (1)

• Common mistakes in estimation of the effect of
  migration and remittances on poverty, inequality,
  etc.
• Estimate consumption regressions for different
  state of migration independently from migration
  choice equation.
• Use conditional mean expenditure in order to
  cacluate counterfactual poverty and inequality
  rates.
• Problems with this approach
• Errors in the consumption regression come from
  multivariate distribution.
• We need to estimate the expectation of the error
  term conditional on the migration choice.
• We need to recover the whole distribtuion in
  order to predict poverty and inequality changes.

22
Simulations (2)

• Generate counterfactual distributions based on
  estimated parameters of our model. For each
  observation we generate 1000 sets of 5 error
  terms drawn from estimated multivariate
  distribution (3,912,000 simulated households)
• Different levels of internal and external
  migration are simulated through the changes in
  values of our two instruments.
• The household size is adjusted for the presence
  of would be a migrant, all variables constructed
  using the household size, shares of various
  age-gender groups, are adjusted.
• All parameters of the counterfactual distribution
  could be recovered from the simulated sample
  poverty and inequality rates, mean expenditure by
  groups, etc.

23
(No Transcript)
24
(No Transcript)
25
Main findings

• About 20 percent of total poverty reduction in
  Nepal between 1995 and 2004 could be explained by
  the increase in work migration and remittances.
• Out of total 2 percentage points poverty
  reduction because of the migration
• Increase in external migration and remittances
  contributed 53 percent.
• Increase in internal migration contributed 38
  percent.
• The rest (9 percent) could attributed to the
  interaction effects of internal and abroad
  migration and remittances.
• In the absence of migration the poverty rate in
  Nepal would increase from the currently observed
  29.9 percent to 33.5 percent and the mean per
  capita expenditure would decline from 16,400 NRP
  to 15,000 NPR.
• Work migration and remittances increase income
  inequality in the country.

26
Caveats

• Our estimations neglect the general equilibrium
  effects of increased work migration and
  remittances. We measure only direct impact on
  household consumption. No spill-over.
• We might have misclassified households with a
  migrant without remittances.
• We neglect household- and community specific
  effects.
• We cannot differentiate between India and the
  rest of the world as migration destination.

27
Conclusion

• Our estimates provide only low bounds for the
  effect of work migration and remittances in
  Nepal.
• The proper estimation of these effects is a very
  data demanding and econometrically challenging
  task.
• New, preferably panel surveys with better
  developed remittances and migration modules are
  necessary.
• We need to know about the characteristics and
  living standards of the migrants surveys in the
  host countries.
• Further research is necessary to understand
  better the impact of migration and remittances on
  wellbeing of Nepali households.