Transcript Intro to LEWIE

Mateusz Filipski
International Food Policy Research Institute
May 2014






1 - General Equilibrium Large and Small
2 - LEWIE: Local Economy-wide Impact
Evaluation
3 - Application 1: Kenya’s CT-OVC
4 - Other recent applications
5 - Preview of Current work in China
Conclusions
General Equilibrium, Large and
Small
“Once upon a time in Mexico …
…“I don’t get Progresa, but tomorrow buyers will be
lining up here”
Spillovers
Local GE effects
Rest of
the country
Treatment
household
Shock
Market
Rest of World
Rest of
Zimbabwe
Control
household?

Villages can have their own markets (& prices)
◦ Factor markets (land, labor)
◦ Commodities (non-tradables, specific varieties, etc.)

Interventions can have spillovers in the village
◦ From target households to non-target households
◦ From target sector to other sectors

What’s a great way to research such
spillovers?
◦ Computable General Equilibrium methods
◦ Can be applied to economy of any scale



Local Economy-wide Impact Evaluation
Central idea = local economies also
experience general equilibrium effects.
CGE analysis is applicable to an economy of
any size (household, hamlet, village, region,
country, multiple countries)
Book (Forthcoming): Taylor and Filipski (2014): Beyond Experiments in
Development Economics: Local Economy-Wide Impact Evaluation.
Oxford University Press.
Most of this talk is based on material in the book

What are they?
◦ Systems of equations representing economies

What are they not?
◦ Not econometrics = no statistical significance
◦ Not a forecasting tool

How we build them?
◦ Computer code such a GAMS

What do we do with them?
◦ “Laboratory for economic experiments”, “Flight
Simulator”
◦ All about markets and linkages. “General
Equilibrium effects”, “Higher-order effects”,
“Spillover effects”, etc...

Are they a CGE?
◦ “Computable”, “General”, “Equilibrium” => in
essence yes
CGE–
Usually:
LEWIE–
Usually:
Scale
National
Subnational
Data
National Accounts
Household surveys
Uses
Policy Analysis
Macro Shocks
Local Projects
Pilot projects
Rural focus
More similarities than differences. Models
akin in spirit, very similar equations






When you arrive ex-post
When you need results ex-ante
When you cannot randomize your treatment
When outcomes are multifaceted, with
winners and losers
When you need to know why there is an
effect, not just whether (i.e. structure)
When you expect spillovers
LEWIE basics




Start from the household model
Nest up to a village/region/island/[…] model
Calibrate the model, usually from household
data
Perform simulations
FDh, factor 
Ph,good  QPh,good   h, factor
QPh, good  Ah 
Production and
Consumption
behavior
QCh , good 
Yh 

Factors
Ph, factor
  FD 
f Factors
 h , good
Ph , good
h, f
Yh
Endow fac  W fac
g , f
Treate
d
NonTreate
d
FDh, factor 
3x
Ph,good  QPh,good   h, factor
Ph, factor
QPh, good  Ah 
QCh, good 
Yh 

  FD 
f Factors
 h,good
Ph,good
g , f
h, f
Yh
Endow fac  W fac
Factors
• Marketed Surplus (for household)
MSh , g  QPh , g  Eh , g  QCh , g  IDh , g
Produced
Marketed
Surplus
Endowments
Consumed
Used as
Inputs
NonTreate
d
Treate
d
Crops, livestock,
retail, services,
labor
Rest
of
World
Manufactured
goods, purchased
inputs
• Market Closures (for village)
𝑀𝑆ℎ,𝑔 = 0 (𝑁𝑜𝑛𝑡𝑟𝑎𝑑𝑎𝑏𝑙𝑒𝑠)
ℎ
𝑝𝑔 = 𝑝𝑔 (𝑇𝑟𝑎𝑑𝑎𝑏𝑙𝑒𝑠)
FDh, factor 
Ph,good  QPh,good   h, factor
Ph, factor
QPh, good  Ah 
QCh, good 
Yh 

  FD 
f Factors
 h,good
Ph,good
g , f
h, f
Yh
Endow fac  W fac
Factors
MSh , g  QPh , g  Eh , g  QCh , g  IDh , g
𝑀𝑆ℎ,𝑔 = 0
ℎ
𝑝𝑔 = 𝑝𝑔
Indexing allows us to greatly
increase number of variables and
equations without complicating the
model.
Cash Transfer in Zambia


“Standard” cash transfer intervention
Targets the most vulnerable
Treatment
household
Transfer
Market
Rest of World
Rest of
Zimbabwe
Control
household?
From Protection to
Production
• Most evaluations look at the beneficiary
households
• They are a conduit through which cash enters
local economies
• Does the whole local economy, then, become
a beneficiary of the CGP
• …including those who do not get transfers?
Every Kwacha transferred to a poor household generates an
additional 0.79 Kwacha in spillovers
Most Spillovers Go To Nonbeneficiaries
*Real-income multipliers (Kwacha) if land, capital, and
liquidity constraints limit the local supply response



The economic impacts of social cash
transfers are likely to significantly exceed
the amount transferred
There may be less of a tradeoff between
protection and production than we once
thought
Non-beneficiaries should be interested in
seeing the transfer programs continue—
and expand
Positive spillovers depend on having
a good supply response
 Interventions may be needed to
make sure this happens

◦ Micro-credit, extension, etc.
From Protection to Production Project
http://www.fao.org/economic/PtoP/en/
The Transfer Project
http://www.cpc.unc.edu/projects/transfer
FAO Report:
Impact of the CGP program on productive activities and
labour activities. Benjamin Davis, Silvio Daidone, Josh
Dewbre and Mario Gonzalez
Other applications
Mostly from “Beyond Experiments” book
(Taylor and Filipski, forthcoming 2014)
5
Production Effects by district
4
Kilombero
Mvomero
Other Districts
3
2
1
0
-1
-2
Rice
Processed Rice
Local crops
Export crops
Livestock
Ressources
Food Proc.
Trade
Services




Irrigation increases yields in the target zone…
but creates spillovers through the region.
Ultimately affects all consumers (+), affects
non-irrigated producers (-), affects food
processors (+), livestock producers (+)
Milling capacity outside of the irrigated
region => regional spillovers
Urban households may be the biggest
winners
Reference: Filipski, M., Manning, D., Taylor, J. E.,
Diao, X., & Pradesha, A. (2013). Evaluating the
Local Economywide Impacts of Irrigation
Projects: Feed the future in Tanzania. IFPRI
publications.

Mexico’s leaky Pro-Campo program
◦ Payments are proportional to land ownership
◦ Two databases: payments due / payments received
 => there exist discrepancies


Reverse of a Cash Transfer
Creates negative spillovers: each $1 not
received by a supposed beneficiary means
$1.2 dollars of real income foregone in the
economy




Tourism on the Galapagos islands
Construction ban supposed to control
tourism and environmental degradation
Small share of tourist expenditures
Can we assume a small impact?
=> No, because of local migration

LEWIE model with migration
◦ Labor comes from the mainland of Ecuador

Increased demand for tourism services
triggered increases:
◦ 58% increase in labor migration to the islands
◦ 77% increase in income from fishing activities
◦ 67% in income from agriculture on the islands
Full economic impact much larger than tourist
expenditures alone suggest



Roatan Corral Reef (Honduras, Caribbean)
Many aspects to value: use value (fishing),
non-use value (“existence”), potential value
(future scientific knowledge?) etc…
We value is only by tourist expenditures =
conservative lower bound

We value is only by tourist expenditures…
◦ Accounting for spillovers
Yearly tourist expenditures = $80 million
 Net Present Value over 30 years =
between$1.3 billion and $4.5 billion
(more than the country’s national debt)

Preview of work in China
With Dr. Yumei Zhang from CAAS-AIRI
(张玉梅博士，中国农业科学院农业信息研究所)

Figure 1:Rural household income source (%)
50
40
30
20
10
0
43
33
30
35
2004
9
11
15
6
3
4
Local off farm
income share
increased
from 30% to
35% during
2004-2011
The transfer income share in rural HH increased from 3.7% to 8.7%
during 2003-2012.
Low income HH: the transfer income share reached 14%.
2002 Grain for
green subsidies
2003 Rural health
insurance
2003 Abolish
agricultural tax
2006 Agricultural
subsidies
2011 Subsidy for
dilapidated housing
2009 Pension
insurance
2007 Rural
subsistence
allowance
About 20% of laborers worked local odd jobs in 2011.
The wage rate increased from 10~15 yuan to 80~100 yuan per
day between 2004 and 2011.
The per capita local odd job income increased from 258 yuan to
926 yuan between 2004 and 2011 with annual real growth rate of
20%.
926
1000
800
520
600
400
259
338
200
0
2004
2006
2009
2011
Fig. Per capita local odd job
income (yuan at 2004 constant
price

Reveal the hidden impacts of rural China’s
safety-nets, and understand how they have
participated to the dramatic evolution of the
country-side.

How have the different safety-nets influenced
the growth of rural activities?

How did they impact the supply of labor and the
shift towards urban employment?

Look forward to those results!
Conclusions

LEWIE modeling:
◦ Applies GE methodology at the local scale
◦ Uncovers spillover impacts of programs and policies
◦ Provides a flexible framework for a variety of
situations

Why not Econometrics?
◦ Modeling and Econometrics are complements, not
substitutes
◦ Ideally, both… if data is available!

This methodology is not difficult – maybe can
be applied in your research
Thank you
Contact: [email protected]