economic impact of refugees - proceedings of the …10.1073/pnas.1604566113/-/dc... · supporting...
TRANSCRIPT
Supporting Information for
Economic Impact of Refugees
J. Edward Taylor, Mateusz Filipski, Mohamad Alloush, Anubhab Gupta,
Irvin Ruben Rojas, and Ernesto Gonzalez
Correspondence to: [email protected]
This file includes:
Table S1
Table S2
Table S3
Table S4
Table S5
Table S6
Table S7
Table S8
Table S9
Table S10
Fig. S1
Datasets for this manuscript include the following:
Dataset S1_LEWIE_Model_Kigeme(in-kind)_forTable1.gms.txt
Dataset S2_LEWIE_Model_Gihembe(cash)_forTable1.gms.txt
Dataset S3_LEWIE_Model_Nyabiheke(cash)_forTable1.gms.txt
Dataset_S4_LEWIE_data_input_sheet_new.xlsx
Tables
Table S1: Simulated Impacts of an Additional Adult Refugee in Three Congolese
Refugee Camps in Rwanda
Refugee Impacts (US$ per
additional refugee)
A B C
Gihembe
(Cash)
Nyabiheke
(Cash)
Kigeme (in-
kind)
Real income (Inflation-adjusted) $205.71 $252.86 $145.71
[166, 260] [194, 320] [133, 161]
WFP transfer $126.31 $120.11 $113.75
Other tranfers $9.57 $6.75 $7.33
Total Spillover $69.83 $126.00 $24.64
To Refugees $28.40 $56.00 $37.50
To Host-country Households $41.43 $68.57 -$12.86
Production By Sector (Nominal)
Crop $102.86 $144.29 $48.57
Livestock $4.29 $2.86 $1.43
Retail $80.00 $82.86 $52.86
Other $51.43 $67.14 $40.00
Trade with Rest of Rwanda $54.89 $48.67 $35.39
The numbers in the table are impacts of an additional refugee within a 10 km radius of each camp and on
trade with the rest of Rwanda. 95% confidence intervals on total real-income impacts (in brackets) were
constructed by making random draws from all parameter distributions (Tables S8-S9), recalibrating the
base model, and repeating each simulation 1000 times. Dollar values calculated using an exchange rate of
700 RWF/US$.
Table S2. Set, Subset and Mapping Names Used in Model Statement
SETS
Subsets
g commodities gtv Goods locally tradable
f factors gtz Goods traded in outside markets
h or hh households gp Locally produced goods
gag Agricultural goods
gnag Nonagricultural goods
v camps fk Fixed factors
ft Locally tradable factors
Mappings ftw Factors traded in outside markets
maphv(h,v) Mapping of households to camps fpurch Purchased variable inputs
Table S3. Commodities, Factors, Households, and Camps
Commodities
Crop Local crops: produced and consumed within the cluster
Livestock Local livestock, produced and consumed within the cluster
Retail Local retailers in the cluster
Other Services and other production
Outside good Any commodity purchased outside the local economy
Factors
Labor Labor (family and hired receiving wage in cash or kind)
Land Land
Capital Capital
Input Purchased inputs
Households
Refugee Refugee households in camps
Host Host-country households within 10 km radius of camp
Camps Gihembe (cash), Nyabiheke (cash), Kigeme (in-kind)
Table S4. Variable Names Used in Model Statement
VARIABLES
Values
Consumption and income
PV(g,v) price of a good at the cluster level QC(g,h) quantity of g consumed by h
PZ(g) price of a good at the regional level Y(h) nominal household income
PH(g,h) price as seen by household h (=PV or PZ) RY(h) real household income
PVA(g,h)
price of value added net of intermediate
inputs CPI(h) consumer price index
R(g,f,h) rent for fixed factors TROUT(h)
transfers given by a household of
others
WV(f,v) wage at the cluster level SAV(h) household savings
WZ(f) wage at the regional level EXPROC(h)
household expenditures out of the
region
Production Trade
QP(g,h) quantity produced of a good by a household HMS(g,h)
household marketed surplus of good
g
FD(g,f,h) factor demand of f in production of g VMS(g,v) cluster marketed surplus of good g
ID(g,gg,h) intermediate demand for production of g ZMS(g) Regional marketed surplus of a good
QVA(g,h) quantity of value added created HFMS(f,h)
factor marketed surplus from the
household
HFD(f,h) factor demand in the household VFMS(f,v)
factor marketed surplus out of the
cluster
HFSUP(f,h)
labor supply from the household (elastic
endowment) ZFMS(f)
factor marketed surplus out of the
region
Table S5. Parameter Names Used in Model Statement (GAMS)
PARAMETERS
Production Consumption
a(g,h)
Shift parameter in CD
production function alpha(g,h) consumption share parameters in the LES
beta(g,f,h)
Factor share parameters (CD
exponents) cmin(g,h) minimal consumption in the LES
vash(g,h) Value-added share of output exinc(h) exogenous income of household
idsh(gg,g,h) Intermediate input share vmsfix(g,v) fixed marketed surplus at the village level
fixfac(g,f,h) Fixed factor endowments Transfers
vfmsfix(f,v)
Factors fixed at the local level
(family, hired labor) troutsh(h) share of transfers in household expenditures
exprocsh(h)
share of expenditures outside 10 km radius
of camp
endow(f,h) Household factor endowments savsh(h) share of income saved
hfsupzero(f,h) Initial labor supply trinsh(h)
share of total transfers received by a given
household
hfsupel(f,h) Factor supply elasticity For Experiments
transfer(h) WFP transfer to household
pibudget(g,h) Liquidity constraint on inputs hfsnewref(ft,h) Refugee labor supply
pibsh(g,h) Share of pibudget to good g packsold(g) In-kind transfer sold on market
Table S6. Equation Definitions
Equation Name Description
* prices
EQ_PVA(g,h) prive value added equation
EQ_PH(g,h) market price as seen from household h
* production
EQ_FDCOBB(g,f,h) factor demands cobb douglas
EQ_FDPURCH(g,f,h) factor demands for purchased inputs - constrained or not
EQ_QVACOBB(g,h) quantity VA produced cobb douglas
EQ_QP(g,h) quantity produced from QVA and ID
EQ_ID(gg,g,h) quantity of ID needed for QP
* consumption
EQ_QC(g,h) quantity consumed
* income
EQ_Y(h) full income constraint for the household
EQ_CPI(h) consumer price index equation
EQ_RY(h) real household income equation
* transfers
EQ_TRIN(h) inter household transfers in (received)
EQ_TROUT(h) interhousehold transfers out (given)
* exogenous expenditures
EQ_SAV(h) savings (exogenous rate)
EQ_EXPROC(h) expenditures outside of the zoi (exogenous rate)
* goods market clearing
EQ_HMKT(g,h) qty clearing in each household
EQ_VMKT(g,v) market clearing in the village
EQ_ZMKT(g) market clearing in the zoi
EQ_VMKTfix(g,v) price definition in the village (camp+sourroundings)
EQ_ZMKTfix(g) price definition in the zoi
* factor market
clearing
EQ_HFD(f,h) total household demand for a given factor
EQ_FCSTR(g,f,h) fixed factors constraint
EQ_HFSUP(f,h) household elastic supply
EQ_HFMKT(f,h) tradable factor clearing in the household
EQ_VFMKT(f,v) tradable factors clearing in the village
EQ_ZFMKT(f) tradable factor clearing in the zoi
EQ_VFMKTfix(f,v)
wage determination for tradable factors clearing in the
village
EQ_ZFMKTfix(f) wage determination for tradable factors clearing in the zoi
* In case of nlp solve
EQ_USELESS trick to make gams think it's maximizing something
Table S7. Equations in the Model
Name Equation
1) HOUSEHOLD EQUATIONS
Price Block
EQ_PH(g,h).. 𝑃𝐻𝑔,ℎ = [𝑃𝑍𝑔]
𝑔∈𝑔𝑡𝑧 ∪𝑔𝑡𝑤+ [∑ 𝑃𝑉𝑔,𝑣
𝑣|𝑚𝑎𝑝ℎ𝑣(ℎ,𝑣) ]
𝑔∈𝑔𝑡𝑣
EQ_PVA(g,h).. 𝑃𝑉𝐴𝑔,ℎ = 𝑃𝐻𝑔,ℎ − ∑ 𝑖𝑑𝑠ℎ𝑔𝑎,𝑔,ℎ × 𝑃𝐻𝑔𝑎,ℎ𝑔𝑎
Production Block
EQ_QVACOBB(g,h).. 𝑄𝑉𝐴𝑔,ℎ = 𝑎𝑔,ℎ × ∏(𝐹𝐷𝑔,𝑓,ℎ)𝛽𝑔,𝑓,ℎ
𝑓
EQ_FDCOBB(g,f,h) [𝑅𝑔,𝑓,ℎ]
𝑓∈𝑓𝑘+ [𝑊𝑍𝑓]
𝑓∈𝑓𝑡𝑧+ [∑ 𝑊𝑉𝑓,𝑣
𝑣|𝑚𝑎𝑝ℎ𝑣(ℎ,𝑣)]
𝑓∈𝑓𝑡𝑣
=𝑃𝑉𝐴𝑔,ℎ × 𝑄𝑃𝑔,ℎ × 𝛽𝑔,𝑓,ℎ
𝐹𝐷𝑔,𝑓,ℎ
EQ_QP(g,h) 𝑄𝑃𝑔,ℎ = 𝑄𝑉𝐴𝑔,ℎ/𝑣𝑎𝑠ℎ𝑔,ℎ
EQ_ID(gg,g,h).. 𝐼𝐷𝑔𝑎,𝑔,ℎ = 𝑄𝑃𝑔,ℎ × 𝑖𝑑𝑠ℎ𝑔𝑎,𝑔,ℎ
Consumption and income block
EQ_QC(g,h).. 𝑄𝐶𝑔,ℎ =
𝛼𝑔,ℎ
𝑃𝐻𝑔,ℎ× (𝑌ℎ − 𝑇𝑅𝑂𝑈𝑇ℎ − 𝑆𝐴𝑉ℎ − 𝐸𝑋𝑃𝑅𝑂𝐶ℎ − ∑ 𝑃𝐻𝑔𝑎,ℎ × 𝑐𝑚𝑖𝑛𝑔𝑎,ℎ
𝑔𝑎)
+ 𝑐𝑚𝑖𝑛𝑔,ℎ
EQ_Y(h).. 𝑌ℎ = ∑ (𝑅𝑔,𝑓𝑘,ℎ × 𝐹𝐷𝑔,𝑓𝑘,ℎ)𝑔,𝑓𝑘
+ ∑ 𝑊𝑍𝑓𝑡𝑧 × 𝐻𝐹𝑆𝑈𝑃𝑓𝑡𝑧,ℎ𝑔,𝑓𝑡𝑧
+ ∑ ∑ 𝑊𝑉𝑓𝑡𝑣,𝑣 × 𝐻𝐹𝑆𝑈𝑃𝑓𝑡𝑣,ℎ𝑣|𝑚𝑎𝑝ℎ𝑣(ℎ,𝑣)𝑓𝑡𝑣
+ ∑ 𝑊𝑍𝑓𝑡𝑤 × 𝐻𝐹𝑆𝑈𝑃𝑓𝑡𝑤,ℎ𝑓𝑡𝑤
EQ_TROUT(h).. 𝑇𝑅𝑂𝑈𝑇ℎ = 𝑡𝑟𝑜𝑢𝑡𝑠ℎℎ × 𝑌ℎ
EQ_EXPROC(h).. 𝐸𝑋𝑃𝑅𝑂𝐶ℎ = 𝑒𝑥𝑝𝑟𝑜𝑐𝑠ℎℎ × 𝑌ℎ
EQ_SAV(h).. 𝑆𝐴𝑉ℎ = 𝑠𝑎𝑣𝑠ℎℎ × 𝑌ℎ
EQ_CPI(h).. 𝐶𝑃𝐼ℎ = ∑ 𝑃𝐻𝑔,ℎ × 𝛼𝑔,ℎ𝑔
EQ_RY(h).. 𝑅𝑌ℎ =
𝑌ℎ
𝐶𝑃𝐼ℎ
2) MARKET CLOSURE:
Market clearing block for commodities
EQ_HMKT(g,h).. 𝐻𝑀𝑆𝑔,ℎ = 𝑄𝑃𝑔,ℎ − 𝑄𝐶𝑔,ℎ − ∑ 𝐼𝐷𝑔,𝑔𝑎,ℎ𝑔𝑎
EQ_VMKT(g,v).. 𝑉𝑀𝑆𝑔,𝑣 = ∑ 𝐻𝑀𝑆𝑔,ℎℎ|𝑚𝑎𝑝ℎ𝑣(ℎ,𝑣)
+ 𝑝𝑎𝑐𝑘𝑠𝑜𝑙𝑑𝑔
EQ_ZMKT(g).. 𝑍𝑀𝑆𝑔,𝑣 = ∑ 𝑉𝑀𝑆𝑔,𝑣𝑣
EQ_VMKTfix(gtv,v).. 𝑉𝑀𝑆𝑔𝑡𝑣,𝑣 = 𝑣𝑚𝑠𝑓𝑖𝑥𝑔𝑡𝑣,𝑣
EQ_ZMKTfix(gtz).. 𝑍𝑀𝑆𝑔𝑡𝑧 = 𝑧𝑚𝑠𝑓𝑖𝑥𝑔𝑡𝑧
Market clearing block for factors
EQ_HFV(f,h).. 𝐻𝐹𝐷𝑓,ℎ = ∑ 𝐹𝐷𝑔,𝑓,ℎ𝑔
EQ_FCSTR(g,fk,h).. 𝐹𝐷𝑔,𝑓𝑘,ℎ = 𝑓𝑖𝑥𝑓𝑎𝑐𝑔,𝑓𝑘,ℎ
EQ_HFMKT(ft,h).. 𝐻𝐹𝑀𝑆𝑓𝑡,ℎ = 𝐻𝐹𝑆𝑈𝑃𝑓𝑡,ℎ − ∑ 𝐹𝐷𝑔,𝑓𝑡,ℎ𝑔
EQ_HFSUP(ft,h).. 𝐻𝐹𝑆𝑈𝑃𝑓𝑡,ℎ
ℎ𝑓𝑠𝑢𝑝𝑓𝑡,ℎ0 + ℎ𝑓𝑠𝑛𝑒𝑤𝑟𝑒𝑓𝑓𝑡,ℎ
= [∑ (𝑊𝐷𝑓𝑡,𝑑)𝜁𝑓𝑡,ℎ
𝑑|𝑚𝑎𝑝ℎ𝑑(ℎ,𝑑)]
𝑓∈𝑓𝑡𝑑
+ [(𝑊𝑍𝑓𝑡,𝑑)𝜁𝑓𝑡,ℎ
]𝑓∈𝑓𝑡𝑧∪𝑓𝑡𝑤
EQ_VFMKT(ft,v).. 𝐷𝐹𝑀𝑆𝑔,𝑑 = ∑ 𝐻𝐹𝑀𝑆𝑔,ℎℎ|𝑚𝑎𝑝ℎ𝑑(ℎ,𝑑)
EQ_ZFMKT(ft).. 𝑍𝐹𝑀𝑆𝑓𝑡 = ∑ 𝑉𝐹𝑀𝑆𝑓𝑡,𝑣𝑣
EQ_VFMKTFIX(ftv,v).. 𝑉𝐹𝑀𝑆𝑓𝑡𝑑,𝑑 = 𝑣𝑓𝑚𝑠𝑓𝑖𝑥𝑓𝑡𝑣,𝑣
EQ_ZFMKTFIX(ftz).. 𝑍𝐹𝑀𝑆𝑓𝑡𝑧 = 𝑧𝑓𝑚𝑠𝑓𝑖𝑥𝑓𝑡𝑧
For simulations with a budget constraint
EQ_FDCOBB(g,f,h)
(only for purchased
factors)
𝐹𝐷𝑔,𝑓,ℎ × 𝑊𝑍𝑓 = 𝑝𝑖𝑏𝑢𝑑𝑔𝑒𝑡𝑔,ℎ
Table S8. Production Function Parameter Estimates and Standard Errors
Production
Activity Parameter
Household Group
Refugee Host
Gihembe Nyabiheke Kigeme Gihembe Nyabiheke Kigeme
Crop
Shift Parameter
NAa
8.23 7.34 5.68
se 1.07 1.09 1.13
Purchased Inputs 0.21 0.18 0.42
se 0.11 0.09 0.11
Land 0.22 0.21 0.12
se 0.07 0.08 0.08
Labor 0.13 0.39 0.10
se 0.10 0.14 0.08
Capital 0.45 0.22 0.36
se b b b
N 116 147 188
R-Squared 0.37 0.22 0.41
Livestockc
Shift Parameter
NAa
6.61 6.61 6.61
se 1.49 1.49 1.49
Land 0.52 0.52 0.52
se 0.19 0.19 0.19
Labor 0.16 0.16 0.16
se 0.18 0.18 0.18
Capital 0.31 0.31 0.31
se 0.04 0.04 0.04
N 296 296 296
RMSE 0.69 0.69 0.69
Retail
Shift Parameter 6.25 6.22 5.72 6.10 5.94 5.87
se 1.26 1.31 1.34 0.66 0.69 0.70
Labor 0.65 0.65 0.65 0.71 0.71 0.71
se 0.38 0.38 0.38 0.19 0.19 0.19
Capital 0.12 0.12 0.12 0.16 0.16 0.16
se 0.06 0.06 0.06 0.03 0.03 0.03
N 65.00 65.00 65.00 178.00 178.00 178.00
R-Squared 0.29 0.29 0.29 0.26 0.26 0.26
Other
Shift Parameter 7.06 6.71 6.20 7.20 6.61 6.55
se 0.86 0.95 0.93 0.54 0.58 0.57
Labor 0.74 0.74 0.74 0.71 0.71 0.71
se 0.28 0.28 0.28 0.14 0.14 0.14
Capital 0.08 0.08 0.08 0.09 0.09 0.09
se 0.05 0.05 0.05 0.03 0.03 0.03
N 53.00 53.00 53.00 152.00 152.00 152.00
R-Squared 0.33 0.33 0.33 0.35 0.35 0.35
Source: Cobb-Douglas (double-log) production function estimates from household (crop, livestock) and business (retail,
other) survey microdata.
a Land constraints inside camps prevent refugees from carrying out crop and livestock production
b Crop production exhibits constant returns to scale
c Livestock was assumed to have Constant Returns to scale with similar technologies across regions varying on by the
shift parameter. RMSE: Root Mean Squared Error
Table S9. Expenditure Function Parameter Estimates and Standard Errors
Expenditure and
Standard Error
Household Group
Refugee Host
Gihembe Nyabiheke Kigeme Gihembe Nyabiheke Kigeme
Crop 0.75 0.51 0.41 0.22 0.18 0.20
se 0.09 0.03 0.03 0.34 0.02 0.01
R-Squared 0.28 0.72 0.42 0.00 0.40 0.53
Livestock 0.01 0.01 0.01 0.03 0.02 0.03
se 0.00 0.01 0.00 0.00 0.00 0.00
R-Squared 0.01 0.02 0.08 0.15 0.18 0.12
Retail 0.10 0.10 0.13 0.12 0.05 0.08
se 0.00 0.02 0.01 0.02 0.01 0.01
R-Squared 0.72 0.16 0.46 0.17 0.19 0.17
Other 0.07 0.07 0.08 0.20 0.08 0.10
se 0.01 0.01 0.01 0.02 0.07 0.01
R-Squared 0.32 0.31 0.15 0.43 0.01 0.27
Non-local (residual) 0.09 0.31 0.38 0.42 0.67 0.58
N 165 155 199 173 148 236
Source: Linear expenditure function estimates from household survey microdata.
Table S10. Sensitivity Analysis of Model Assumptions in Simulations of an Additional
Refugee (US$/refugee, US$1=700 RWF)
Model Assumptions: (A) Base
Model
(B) No
Labor
market
impact
(C) Least
constrained
(D) Most
constrained
Elasticity of labor supply 100 100 100 0
Labor Market Impact
Considered Yes No Yes No
Inputs constrained No No No Yes
Fixed capital Yes Yes No Yes
Iterations 1000 100 100 100
Camp
Gihembe (Cash)
Real income (Inflation-
adjusted) 206 199 221 79
To Refugees 164 124 166 113
To Host-country Households 41 74 56 -34
Nyabiheke (Cash)
Real income (Inflation-
adjusted) 253 243 214 80
To Refugees 183 127 173 116
To Host-country Households 69 114 41 -36
Kigeme (in-kind)
Real income (Inflation-
adjusted) 146 144 199 81
To Refugees 159 110 170 101
To Host-country Households -13 34 29 -20
(A) Base model used in the main text (one additional refugee brings WFP transfer, other private transfers, and increases
labor supply)
(B) Base model without labor market impact (wage + business income of a refugee)
(C) Base model + capital holdings increase by the same value as the labor market impact of one refugee (wage +
business income of a refugee)
(D) Base model with dramatically reduced elasticity of labor supply, no labor market impacts, and an added constraint
on input purchases.
Results of sensitivity analysis: Participation in labor markets allows refugees to capture more of
the spillovers they create, but it does not change the overall size of the spillover as long as the labor
supply around the camp is elastic. Local capital investment brings higher multipliers, particularly
in the in-kind camp. Only in the most constrained scenario, where the economy has no ability to
increase local production and capture spillover benefits, do we see significantly reduced overall
spillovers and negative impacts on host economy households.
Fig. S1: Increase in real income for one extra refugee at different values of elasticity
of labor supply (variations on Base Model)
Notes: As far as we know, there are no estimates of wage elasticities of labor supply for
Rwanda. Typical estimates for European countries range from 0.03 to 0.6 (1). Estimates
for African countries vary widely. Abdulai and Delgado (2) estimate a wage elasticity of
0.32 for men and 0.66 for women in Ghana. In rural Malawi, an experimental study
measured this elasticity from the change in probability of working on a given day as
wages change. This yielded an estimated labor supply elasticity of 0.16; however, 74
percent of individuals chose to work at the lowest wage offer (3). None of these studies
involves a change in labor supply due to the presence of refugees or other immigrants.
Given the lack of work opportunities inside the refugee camp, it is likely that local
elasticities of labor supply in our study areas are higher than most national estimates.
High elasticities of labor supply (ε) reduce the labor market benefit of adding an
additional refugee worker to the local workforce. This labor-supply effect overwhelms
any positive impacts, for example, via higher wage income; thus, the relationship
between the elasticity of labor supply and the total real-income impact is decreasing and
converges to those at the bottom of Fig. 2 as ε increases. Fig. S1 illustrates this between ε
= 0 and ε = 4, at which point most of the convergence is complete. The decrease is
sharper at the in-kind camp where, with high ε, wages do not decrease to compensate for
lower output prices resulting from refugee sales of food aid. If ε < 100, actual impacts are
larger than in our simulations using the base model.
References for Supplementary Materials:
1. Bargain O, Orsini K, Peichl A (2014) Comparing Labor Supply Elasticities in
Europe and the United States: New Results. J Human Resources 49(3): 723-838.
2. Abdulai A, Delgado CL (1999) Determinants of Nonfarm Earnings of Farm-based
Husbands and Wives in Northern Ghana. AJAE 81(1): 117-130.
3. Golderg J (2016) Kwacha Gonna Do? Experimental Evidence about Labor Supply
in Rural Malawi. American Economic Journal: Applied Economics 8(1): 129-149.