PSIA: Key Issues in Trade Policy Reform or How to Measure the Linkages between Trade Growth and Poverty

1
PSIA Key Issues in Trade Policy Reformor How
to Measure the Linkages between Trade Growth and
Poverty
Maurizio Bussolo DECPG The World Bank
March 2006
Based on the PSIA note on Trade Policy reform
authored by Maurizio Bussolo and Alessandro
Nicita.
2
Intro

• Assessing the sign and the strength of the trade
  and poverty links is a complex task
• No clear consensus/generalization has yet
  emerged, thus we should not push one-size-fits
  all policy advice
• However important lessons are being learned
• Distributional effects need to be addressed
• Country specific detailed studies are useful
• Trade policy can only play a part in a
  comprehensive development strategy

3
Trade and Poverty Links an overview
3
4
1
2
4
1st group of studies the growth-inequality-pover
ty regression approach

• Many recent studies de Janvry and Sadoulet
  (1995,2001), Ravallion and Chen (1997), Dollar
  and Kray (2000) focused on the statistical
  relationship between growth and poverty across
  countries and time periods
• Main conclusion growth strongly reduces poverty,
  Ravallion and Chen find an e of 3
• Policy implication if the e is sufficiently
  high, poverty reduction strategies based mainly
  on growth may be justified.

5
The regression approach some refinements

• This approach postulates a constant e across
  countries/time periods
• Such an approach entails a naïve model that does
  not take into account the existence of an
  identity that links poverty reduction with growth
  and changes in the income distribution
• A graphical representation

6
Poverty-growth-income distribution
Z
Reduction due to Growth
Reduction due to distribution change
7
The regression approach some refinements

• Bourguignon (2003) shows that a regression
  derived from the mentioned identity would
  include
• Distributional change (e.g.. ? of inequality)
• Interaction terms of growth with a) the initial
  level of development and with b) initial
  inequality
• Main conclusions a) explanatory power doubles
  (fifty-fifty), b) ? values change
• Lower level of development and higher level of
  inequality reduce the growth ? of poverty
• Policy implications
• Research implications

8
Trade and Poverty Links an overview
3
4
1
2
9
Trade Reforms

• Reduction of protection of domestic markets by
  elimination of quantitative restrictions
  (quotas), by lowering tariff rates, and regulate
  the use of Non Tariff Barriers (NTBs)
• Typology
• Multilateral liberalization (WTO negotiations)
• Unilateral liberalization
• Discriminatory liberalization (regional,
  bilateral agreements)
• Different institutional arrangements and
  different effects

10
Income distribution implications of Multilateral
and Regional liberalizations, an example for
Nicaragua
Kernel Distribution of Gains for Nicaraguan
households for a multilateral (full) and a CAFTA
liberalization scenarios
11
Trade and Poverty Links more details

• goods prices (expenditure channel)
• factor prices, employment (income channel)
• government fiscal position (transfers, tariff
  revenues, other taxes)
• Investment/productivity changes (long term
  growth)
• Short run adjustment costs (volatility, risk)
• Other external shocks (global lib., TOT)
• Standard small country H-O-S model main
  predictions (focus on a. and b., magnification
  effect, type of lib.)

12
2 other groups of studies 1

• partial equilibrium / cost of living analyses
  (detailed micro)
• general equilibrium (macro top-down approach)
• micro-macro synthesis
• These are all quantitative studies, data
  requirements can be a limitation in their
  implementation

13
Partial equilibrium / cost of living analyses
intro

• Based on real households rather than
  representative ones
• Have a central equation of this form
• The change in welfare is function of a) the
  change in output prices, b) inputs prices, c)
  factor prices and d) consumption goods prices

14
Partial equilibrium analyses advantages/disadvant
ages

• the ?s are household specific full
  heterogeneity is included
• Key characteristics of the poor inform the
  analysis
• Examples Ethiopia export/food crops regional
  segmentation, and own consumption in Nicita
  Olarreaga (2003) Madagascar booming textile in
  Nicita Razzaz (2003)
• Problems
• price changes are partial equilibrium
• no substitutability (i.e. no quantity response)
• Extensions (panels, ex-post studies Winters on
  Vietnam, McCulloch on Chinas province)

15
General equilibrium analyses advantages/disadvant
ages

• Based on representative households, disaggregated
  in many groups as to minimize the intra-group
  income heterogeneity
• Generate GE consistent prices, accommodate
  different degrees of resource mobility and
  substitution, perfect identification
• Many studies, starting Adelman and Robinson
  (1978) for Korea, up to a recent one for Brazil
  (Harrison, Rutherford and Tarr 2003)
• Main result poorest household earn 4 times more
  than richer households
• Decomposition exercises test different type of
  liberalization
• Problems
• Within group variance is fixed
• No heterogeneity across the poor
• Perfect price transmissions

16
Micro-Macro synthesis

• Top-down sequential approach
• Inclusion of the full household sample in the CGE
  model
• Feed-back full models

17
Top-Down sequential approach an example

• Macro model is a global multi-sectoral CGE model
• Link variables are commodity and factor prices
  (from CGE model to micro data)
• For the incidence analysis we use four household
  surveysBrazil, Chile, Colombia and Mexico
• Two policy experiments FTAA and multilateral
  global liberalization

18
Global trade reform can reduce poverty more than
proportionately(Poverty elasticities generated
by household micro-simulation linked to global
trade reform scenario)
Low labor mobility
Distributionally neutral elasticity
High labor mobility
Source Staff simulations with Linkage model and
country micro-simulation modules.
19
BRAZIL Poverty effects of FTAA and Full lib
Headcount percent changes due to full
multilateral liberalizationc
Full distrib.
Yellow dotted distrib
Bussolo, Lay, van der Mensbrugghe (2003)
20
Mexico Poverty effects of FTAA and Full lib
Headcount percent changes due to FTAA
Full distrib.
Yellow dotted distribt
Bussolo, Lay, van der Mensbrugghe (2003)
21
Other approaches

• CGE more complex micro model Bourguignon,
  Robilliard and Robinson (2002) on Indonesia and
  Bussolo and Lay (2003) on Colombia
• CGEfull household sample Cogneau and Robilliard
  (2000) on Madagascar, Cockburn (2002) on Nepal
• Full feed-back models Savard (2003) on
  Philippines

22
Off-the-shelf methods

• simSIP
• PovStat
• Both the above methods assess the (ex-ante or
  ex-post) effect of aggregate economic growth on
  poverty
• Poverty Analysis Macroeconomic Simulator (PAMS)
• Maps macroeconomic results (differential output,
  employment, labor revenues across sectors) into
  poverty effects
• 123PRSP

23
Difficulties

• As Kanbur (2001) puts it trade and openness is
  the archetypal, emblematic, area around which
  there are deep divisions, and where certainly the
  rhetoric is fiercest
• Level of aggregation
• Time horizon
• Market structure
• Domestic policies (regulation and competition)

24
Dynamic gains from trade Increasing integration
in global markets is associated with faster
growth(Average annual per-capita GDP growth,
1980-2003, percent)
Increasing export share in GDP
Decreasing export share in GDP
Source World Development Indicators and DECVP
staff calculations
25
Much is left out of the analysis(Real income
gains for developing countries in billion)
Merchandise trade vs. services, GTAP5 (1997)
Merchandise trade and productivity, GTAP 6 (2005)
Source Left panel, GEP 2002. Right panel,
Anderson, Martin and van der Mensbrugghe (2006).
26
Chile wage gap (relative poverty) with different
labor market settings
Wage gap (Skilled / unskilled) percent changes
Bussolo, Mizala, Romaguera (2001)
27
Other elements in a good PSIA Stakeholders
analysis

• Trade policy is an economic inefficient but
  politically efficient re-distributive tool
• Trade policy normally favors import-competing
  sectors (and factors) rather than export-oriented
  ones (for historical reasons ? revenues
  collection)
• Predictive analysis of the distributional effects
  is crucial for a successful and sustainable trade
  policy reform
• Approach 1 the factor endowments model
• Approach 2 the sector specific model

28
Institutions

• Globalization and governance
• WTO commitments
• Effects on domestic institutions Bonaglia, Braga,
  Bussolo (2002)
• Coordination across different public institutions
• Example trade liberalization in Colombia
• Trade reform as part of a larger development
  agenda

29
Conclusions

• Poverty impacts are different across countries
  and scenarios
• We show that at least for one case, Mexico, a
  simple approach is detecting a reduction in
  poverty, in contrast to an increase, as shown by
  an approach considering distribution
• Strong distribution effects can dominate the
  average growth effects
• Useful information for the design of trade
  reforms (capture, rent seeking) and compensatory
  measures (targeting) different emphasis for the
  short run and the long run
• Volatility and risk (Loyaza on Managing
  Volatility Handbook)
• MDG context Halving poverty requires about 5
  yearly reductions, with full trade lib, in the
  good cases, we get about 2 (static gains only)