40638

Public Disclosure Authorized

SOUCH ASIA REGION

PREM Working Paper Series
Public Disclosure Authorized

A SIMPLE
MACROECONOMIC
AFGHANISTAN’S OPIUM
MODEL
DRUG FOR NEPAL
ECONOMY
Public Disclosure Authorized

Stephane
Christopher WardGuimbert andByrd
and William
Sailesh
December 2004Tiwari
June 2007
Report No. SASPR-5
Report No. SASPR-11
Public Disclosure Authorized

A A
WORLD
WORLD
BANK
BANK
DOCUMENT
DOCUMENT
The findings,
The findings,
interpretations,
interpretations,
and conclusions
and conclusions
expressed AthisWORLD
expressed
in inpaper
this paper areBANK
are entirely
entirely ofDOCUMENT
those those
theofauthor(s)
the author(s)
and should
and should
not benot
attributed
be attributed
in anyinmanner
any manner
to theto the
World Bank,
The findings, World
to Bank,
its affiliated
interpretations, to and
its affiliated
organizations,
organizations,
conclusions or to members
expressed orintothis
members
ofpaper
its Board
of its
are of Board
Executive
entirely of Executive
those DirectorsDirectors
or the countries
of the author(s) or should
and the countries
they
notrepresent.
be they represent.
attributed in any manner to the
World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent.
 About the SASPR Working Paper

The purpose of the SASPR Working Paper kxies is to provide a quick outlet for sharing
more broadly xesearchJanalpsis ofissues dated to development in South Asia. Although the
primaq source of such research analysis is SASPR staff, o&er contributors are most
welcome to use this outlet for rapid publication of thek research that is relevant to South
Asia's developmeat T b e papers rare informal in nature aad basically represent views/andysis
of the concerned author(s1. All papers submitted for publication ate sent for an outside
review to assme quality. 1 provide only a very hght editorial touch. For enquiries about
submission of papers for publication in the series or for copies of published papers, please
coatact Naomi Dass (telephone number 202458-0335).

Sadiq A b e d
Sector Director
South Asia Poverty Reduction and Economic Management
World Baak, Washington D.C.
This papa was prepad as a background paper for the ongoing m a m monitwing work on
Nepal. It is a simple CGE model with basic disb'butional analysis, based on the '123 PRSP'
mdel. It was presented in Nepal in December 2006 and had good fedback. We thank Abhishek
Basnyat, Shanta IDevarajan and the participants of the December 2006 Nepal seminar for their
insights and comments, Emrs are our soIe responsibility.
 Table of Contents

1. Introduction.....................
...................................................................................... 1
2. The Model .................................................................................................................3
2.1. Overall Structure ..............................................................................................
3
2.2. M m m n o m i c Framework ............................................................................... 4
2.3. 1-2-3 Model ......................................................................................................
5
2.4. Lung-Term Growth Projections .......................................................................... 5
2.5. Short-Temr VARs .............................................................................................. 8
2.6. DistributionalImpact ...................... . . . . ...................................................... 9

3.1. Static simulations.............................................................................................

10
3.2. Dynamic simulations ................................ .................................................. 12
4 . Conclusion ............................................................................................................... 15

5. References..............................................................................................................
16

6. Annexes............ .
.... .
...................................................................
17
 A Simple Macroeconomic Model for Nepal

1. Introduction
South Asian economies have grown remarkably jn the past two decades. This has even
raised expectahns that the coming d e d e could truly make a dent on povertp (Devarajan
and Nab; 200q. T h e giowth experience in Nepal - which was mith Sri La& one of the
earliest and quickest liberaher in the region -was in the 1980s and 1990s as encouraging as
in neighboring countries - but these h~$Ievels could not be sustained in the 2000s, wme
growth in the other South Asian economies (notably in India, Figure 1) was fuaher
accelerating.
-
Figum 3: GDP per capita Grwvth. India and Nepal

Note: IOycm ayt~geof G D P p copitta growth (in 2000 US$).S u m : World DrYckpmmt Inliimtors (7006).

At the same time, the incidence of poverty ia Nepal declined from 42 percent in 1995/96 to
31 percent 2003/04 (World Bank, 2006). This strong performance can p a d y be ahbuted
to economic growth, wen though the investment climate gradually deteriorated with the
intensification of the conflict. Another key factor driving the reduction in poverty incidence
has been the rapid increase in remittances to more than 12 percent of GDP:an increasing
numbers of Nepnlis are going abroad to work particularly as low wage workers i n India,
Middle East countries, Malaysia, Korea and others. Other factors dnving the significant
reduction in poverty are an increase in agricultural real wages, increased utbanization and
reducuon in dependenq ratios.

Questions about the sustainability of this trend are however numerous. Is the growth
deceleration of the eady 2000s leadvlg to an increase in poveq? Can hscal poky help
acctIerate the reduction in poverty? Other questions are raised about NepaI" external
position, the role of r e m i ~ c e and
s the fluctuations of the red exchange rate. Fox instance,
are the pressures on t h e exchange rate (with some real appreciation in recent yeats) having
further distributional impact? Are the increasing inflows of remittances (possibly coupled
a with increashg external assistance in the future) having an impact an the exchange rate
CtDutch disease", by which inflows of foreign cuncncy lad to a real appre.dpti;ri of the
nationd m e ncy, which has a negative impact on exports)? These quesuons mo~vatethis
paper and the development of a framework linking the macroeconolnic poky environment
 r

and poverty outcomes. Indeed, whire a wide variety of macro-econome~cand computable

p e r a I equilibrium type models have been developed for Nepal, they have w d y faced
three key limits.

Firs&theit usefulness has generally failed to m s c e n d the immediate purpose or time for
which they were initially built, mainly a reflection of their complexity. For example, the
macro-model built by the Energy Commission Sector to forecast energy demand in 1987
was too parochial in its scope and dthough the multi-sector model developed by K h a d
Thapa and Elbers in 1987 had economy-wide coverage, its utility did not outlast the Seventh
Plan. S&yI a fairly comprehensive input-output matrix covering 39 sectors was
developed by the Development Study Consultants in 1390, but the matrix has not been
updated since then. Another notable work is the computable general equilibrium modd that
was developed under the auspices of the Intemationd Development Research Center's
(IDRC) Micro Aspects of Macro Adjustment Policies (MIMAP) to analyze the poverty
impact of trade liberaliza~onin Nepal in 2001. While this is perhaps the ody work that
models and combines the general cquilibriwn effects of macroeconomic policies with
. &striSutional effects, its coverage is h t e d to trade policies and it does not lend itself
conveniently to generalizations over a broader set of macro policies.

Second, their complexity has also often limited their influence in policy dialogue,
Macroeconomic shocks and policies have indeed complicated effects on growth and income
distribution as h e y get transmitted through the economy. Policymakers in Nepal feel that
there is a gIarifig absence of a rigorous framework to analyze the povesty and dis~butional
impact of macroeconomic policies in Nepal and the proposed mcrmconomic model is
intended to be a con&bution towards hUtng this void.

Third, complex models also require a detailed data input. Even i5 such data can be made
avadablt, the intensity of the effort required and the difficulty to update frequently this input
reduce considerably the life expectancy of these complex models.

For that purpose, we use a model proposed by Devarajan and Go (2003) for Zambia aad a
few other PRSP countries. Drawing lessons horn past endeavors to hstitutioaalize such
models into the government's policymaking machinery, the proposed model should be
simple, its data requirements not overly demanding, and its mechanics relatively easy to
understand and use. Under h e s t criteria, the 123-PRSP model seems appropriate for the .
kind of poticy questions that this papex seeks to tackle. Here, w e also hghhght the potential
usefulness of the mode1 by xeviewb~ga few policy questions related to oil prices, remittances,
real exchange rate levels, public expendimes, and growth.

The paper is structured as folIows. The next section reviews the model, its overall structure
and its five main modules. T h e following section o u h e s a number of hypothetical
scenarios and their growth and distributional implications. A final section conc~udesm
 2, The Model
2.1. Overall Structure
The model is based on Devarajan and Go (2003). The objective of this model is to be a
middle ground between existing macroeconomic consistency frameworks and ma-sector
Computable General Equilibrium (CGE) models (with disGbutional analysis). It is a middle
ground in the sense that it strikes a balance between complexity and simplidty, sttiving for
an outcome easy to understand for policy rnakm which capture enough of the multiplicity
of channels by which policies hfluence poverty.

The mode1 works as follows (Figure 2):

- for each period, a simple sbtic CGE model (called 1-2-3) generates a set of prices
(wages, sector-specific profits, nnd relative prices) which is consistent with a set of
macraeconomic polides and @i) external shocks; this block estimates the
dis~butiondeffect of policies and externd shocks, while being esseatially p w t h -
neutral,
- the same set of macroeconomic policies and extemd shocks generate a set of short-
to medium-tern GDP growth rates through two growth mod& (a short-term md a
medium-tern modq; this block focuses only on growth, without distributional
impact;
- both the growth trajectory over time and changes of rebtive prices have an impact
on poverty through the combined effect of across-the-board income growth and
changes in the &sttibution;
- consistency between both aspects is also maintained through a set of
macroeconomic identities.
Figwe 2: Saucture of the Nepal 123PRSF Model

Macroeconomic Policies
and External Shocks

G Distribution

Poverty [=] (Households)

Soum: Devarqkn and Go (2003).

' i
The balance of this section r d e w s the five subcomponents of this model.
2.2, Macroeconomic Framework
T h e macroeconomic framework is based on the W M - X mod4 maintained by the World
B a s s Nepal economic c o m q team. Thrs b e w o r k is consistent with the
macroeconomic scenariQ that underpins the IMF's program with the Government ofNepal
(GoN).'
- Real growth is expected to pick up from 2.5% to 4.5%, ie. fiom around 0.5% to
2.5% in per capita t m s . This is d y based on an analysis of medium-term
growth trends (see also Section 2.4). Growth in this scen&o is notably &en by an
increase in private investment, gradually kadEng to a rebound in exports.
- Inflation would remain under control at around 5%.'
- Domestic revenues would gradually increase, but less than public expenditures,
hence I w b g to a hrgher fiscal deficit. T h e increase in public expenditures would be
&en by a mix of recurrent and capital spending.
- The curxent account would xemain in surplus with fuaher growth in remittances -
but the trade deficit would further increase, the increase in exports being more than
offset by strong demand for imports. The cwrent account surplus would ktgely
translate into an increase in gross xesemes.
Table 1: Macroeconomic Framework '

This baseline scenario is contingent on a number of factors, in parti& improvement in the

political and security situation. This would &ve an improvemant in the investment c h u t e
and better demand prospects, hence leadmg hrms to increase investment Better security
2 .
1
Currently the year 2003104 is used as a bast year. This makes the base year consistent with the most
recent household suney (NLSS 11, cf. World B e 2006). Once estimates of the national accounts are
finalized for subsequent years, this could be updated to make 2004i05 or 2005106 the base year.
The consistency framework allows to review implications of the plicies and shocks on monetary
accounts. However, none of the blocks explicitly deal with monetary issues (they all work in real terms).
t
 would also ease the recent inflationaty pressures. ~t would also reduce the pressure on
recurrent expendims in the s e d t g sector, while dowing fur&= wenue growth and
implementation of capital expen&tures. T h e no& e x h g e rate is assumed to remain
pegged to the Indian Rupee.

2.3. 1-23 Model

T h e 1-2-3 Model is a simple, spreadsheet-based CEE modd (Devatajan, Lewis, Robinson,
1990). Its name refers to its shple structure, covering 1 country, 2 sectors (exports vs. all
other b a l goods produced, c d e d domestic goods), and 3 commodities (domestic, imports,
and exports). It captures the impact of macroeconomic poficies on the relative price of h e
domestic good to either exports or imports, a relative price which can be interpreted as a real
exchange rate (see the equations in annex).

The outcome of the model is a set of relative prices, sector-spec& profits and wages that
are consistent with the chbration of the model, The equilibrium is reached by setting the
prices that zlle driving (i) the productive d o c a ~ o nbetween domestic goods and exports and
(3the consumption allocation between domestic goods and imports. Government and
households are subject to a budget c o n s h t . In addition, there is an external balance
constraint (with the current account being equal to an exogeaeous level of foreign savings).
The model assumes that, behind the goods sectors, the labor market is competitive and the
equilibrium wages bring the matket to full employment. Capital is assumed to be fixed and
sector-specific.

In its static version, the model is calibrated with a s d set of macroeconomic indicators (all
fiom the national accounts). In addition, the calibration requires two elasticities, the
ehsticity of ttansformation between the two production sector (domestic goods and exports)
and the elasticity of substitution in t h e utility fuaction between the two goods used in the
country (domestic goods and imports). For Nepal, both are set a t 0.60 (see Devarajan, Go,
Lz,1999).
In out model based on the 123PRSP framework this static model is run for each year of the
simulation (it still remains a static model as there is no intertemporal m a e a t i o n or
constraint). En other words, the model is calibrated for each yeas with the baseline
macroeconomic h e w o r k (Section 2.2), which, in variant a d y s i s , is adjusted by either
long-term g o d projections OX.short-term shock impacts (see next two sub-sections),

2.4. Long-Term Growth Projections

An altemahe to the base case growth projections is to use a model for medium- to Ioag-
term growth. Out model uses a module developed by Ianchovichina and Kacker (2005). It
is based on cross-country growth regressions (cf. Loayza et al., 2005, using GMM systems
estimators on a 5-year average panel dataset). T h e key drivers are shown in Table 2 (and
data sources are presented in Box 1) and include: initial conditions (level of income per
a
capita and output gap vs. potential output); macroeconomic environment (terms of trade;
real exchange sate; voladlity; trade openness; banking crises; h a n d sector); h'bman and
physical capital endowment (education; telephone); and role of Government (governance
and government consumption).
 A

As shown in Table 2 and Figure 3, the model - which is estimated an a pand of countries,
not on Nepal only - has a good fit for Nepal in the past, with an underestimation of the
growth acceleration in the &st half of the 1970s and an overestimation of growth since the
mid-1990s. T h e Iatter gap is likely to be related to the poEticd and security issues that Nepd
confronted d h g the recent period: assuming a brighter future for the country h r n that
point of view, the model can be expected to generate realistic projections for the future (a
"peace and stability" growth dividend).
Table 2: Long-Term Growth Projections
1480.84 7-80 t87UT4 157579 188W 188580 1mQd 189589 XXXF2M)4
lrvbal GCP Per Capth (USSJ 856 673 683 777 sDe 93s I OE3 7.lR 1233
IntOal Owrpul Gap 22% 11% 0 1% -2 7% 12% -? 3% 4 1% 1.8% -1 2%
E d u a h o n d a r y EnoRinml 142 142 151 149 21.7 284 304 374 47.0
F~nanclal (kplh -to m i k CmdWGDP 13 20 40 50 85 H7 t45 255 14.3
Trade Openness. Trade VdwnerGOP 504 533 S t W.8 595 $82 Ba3 BB6 713
~ovemment~urdank-.wn~eC ~ ~ U ~ ~ ~ V G D P 71 77 83 7.7 83 93 a4 91 9.a
Puale Irifsastructum M m Te- Lhes psr 1WOpeopfe 079 l.t& 14d D62 083 ts5 3!3 748 1307
Govomance (1.26) (1.24) (1 24) (124) (0.93) (0 58) (0 42) (0 44)
Lack d Pncn StablMy lml&tlplRafa 10336 tO5DB 31057 1 ~ ~
11055 ~ 111.13
) 1 1 1 36 TO781 1W48
C y c h l Vchrlty S D M Outpi Cap 2.0% 2.4% 23% 07% 32% 14% 1.1% 08% 17%
Real Exmange Rake 2000-100
Swlernlc Banklw Chm
T ~ r m sU! T Q ~ R
G M h
19d

48% -14%
245 218

-52%
193

00%
170

OW
140
0.20
00%
- 108

0.m -2.W
57 89

04%
per U p b Q W h
Actual 02% 05% 53% 26% 00% 29% 3MC 15% 13%
MWol 07% 0.3% 13% 07% 25% 9.8% 3.2% 2.3%

Sam: fanchotri.hi# and h k w (2005). .

I B m 1: Data Source for Long-Term Projections

T h e key variables used in the leng-term projections are discussed in this box. Ianchovicha and Gcker (2005)
presents more details on these vdnbles and highlight how many of these arc highly correlated with other
potentid proxies (e.g. telephone lines with other public infrastructure indicator).
Pcr Capita GDP: ratio of total GDP (1995 PPP-adjusted US$ from Hcston ct d., 2002) to total population.
Supplemented with information from the hfF for the 2000-2004period.
Output Gap: difference berween the log of actual per capita GDP and the log of potential (trend) per capita
GDP at the starr of the period. 'She treed i s identified using the Baxter-King hlter.
Main Telephone Lines per 1000 People: telephone mainlines are the lines connecting a customer's
equipment to the public mtched telephone network; from WDI (for the 3975-2004 and the GDN dataset for
(1960-1 975).
Governance: &st principal component of four indicatars of the Internation4 Country Risk Guide (ICRG):
prevalence of law and order, quality of bureaucracy, absence OF corruption, and accountability of public
offiaals.
Secondaxy Emohent: ratio of total secondary enrollment, regardless of age, to the population of the age
group that officially corresponds to that level of education. From WDI and the GDN datasct.
Private Domestic CreditJGDP: ratio of the stock of claims on he private sector by deposit money banks
and other hancial instituaons to GDP. X h data sources are Beck et d (2000)and the WDI.
Trade Volume JGDP:residual of a regression of the log of the ratio of trade (exports and imports) to GDP
on the logs of area and popdadon, and dummies for oil exporting and landlocked countries.Based an data
from World Development Netwoxk (2002)and the World Bank (2[#)2).
Terms of Trade Growth: World Development Indicators and the GDN dataset.
Government Consumption/GDP: ratio of government consumption to GDP from the WDI.
100+Inflaiion Rare: InflaGon - an indicator of macroeconomic stability is measured by the consumer pdce -
index from WDI.
Standard Deviation (S.D). ofOutput Gap: by-product of the procedure to construct output gap. L .
Real Exchange Rate, 1995=100 {measure for external imbalances and the risk of balance-of-paymentscrises):
GDN dataset and W3I. An increase in the index indicates an appreciauon.
Systemic Banking Crisis: number of years in which a countxy underwent a systemic banKing cisis relative to
the number of p a r s in the corresponding period. The source for these data is the Wf3 dataset of bmking crises.
Population: World Bank's Human Development Network for the period 2000-2090.
a
 Figute 3: Annual: real GDP pet capita g r h ( A c d s and Estimates from Model)

Sonrce: Bared oa lunchovichrna and Kockef (2GQ5).

Moving fornard, a first set of projections assumes no progess in the underlying drivers of
growth (left panel af Figure 4). This leads to a projection of less than 1%of per capita
growth per am=, with fairly small confidence intewals (population growth is around 2-2
M%, hence total r e d economic growth would be around 3%))
Figure 4: LongTefm Projections (Annual real GDP per capita growth)
N o change in drivers of gtowth Cbntinued progress in f i v e r s of grwvth

Source: Aurl~or~' b u ~ t don Imchovichina and Kacktr

culc;~hft~n~, (2007).

In a b h - c a s e scenario, w e assume the followkg

- Secondary enrollment increases by 5 percentage points a year, which would be short

of Government's objectives, but is in line with recent progress;
- A simrlar tread is assumed for infrastructure;
- Government consumption is somewhat reduced, with lower needs for security
expensltures, a rationalization of some programs, somewhat offset by %her
allocations for operations and maintenance of infrastructure and recurrent costs for
health and education;
- Governance is assumed to gradually improve;
- Inflation remains around 5%, but growth volatility is reduced to levels observed in
the 1980s; and
- The real exchange sate and terns of trade stay constant.

1
5.

'The contideuce intervals are computed following the method in Kraay and MonoIcroussos (1999) which
assumes that there is no uncertainty associated with the forecasted explanatory variables and the parameter
estimates other than the wansjtional convergence effect. The fornula for the 60% confidence interval is
presented in Ianchovichina and Kacker (2005).
 This leads to a se&nd set of projections with growth picking up toward 2% per year and pa
.* capita (right panel of F i e 4). baseline scenario (Section 2.2) is consistent with &is
second set oflong-term growth projections.

2.5. Short-Term VARs

In addition to the centtd scenario discussed above for the short to medium term Vable 1)
and the medium to long-tern projections (Figure 4), the 123PRSP model includes the
simulation of two different short-term shocks on the economy. This simulauon is done
through two t t i ~ aautoregressive models (Vector Auto Regression, or VAR), as
presented in Table 3.

First, a change in the real exchange rate4would have a number of implications. Based on
these estimated VARs, a positive shock (appreciation) has a negative impact on GDP g r d
the &st year (a l0Yo appreciation reduces growth by 0.89 percentage points the h t year),
even though the impact is slightly reversed the second year. This reflects a typical ] - m e
reaction, where the trade deficit i n i d I y improves after the appreciation (because of pice
stickiness, ~olumesof imports and exports are initially unaffected, and the only effect is on
prices, the appreciation making imports cheaper and exports more profirable) before
deteriorating (as volumes of imports, which are now cheaper, increase and volumes of
exports, now less competitive, decreases). In tbis case, the net effect of an appreciation is a
deterioration in the current account balance (which assumes that-the bfarshd-lerner holds
- i.e. the sum of the demand elasticities t.a import and export is above I].~
Table 5: Short-Term Growth Elasticities
Year 1 Year 2
Shock on Real Exchange Rate -0.089 0.038
Shock on Govenunent Expcnditues 0.124 -0.019
m
e r n V A R f i r ~ b c j r two+&
( a ) Impube n r p o n ~ e ~ h the r~ (Z) RER is cdmhted ar lhc tmak mgbted
fibon in i o ~irade)
l m q g c ofbihtgzaI RERr mlyith two tradingparinm~,I& and dc Kcfl ofbe WWU Indm
WPI, US CPI and UJ'Jgmqfir hd-mpntc.t, fyoddpn'cc~and the world txchanffcrats rerptmrPtm~4. iVtpnli C P I

To link this simulation with h e 123 model, it is assumed that the red appreciation would be
caused by a terms of tsade shock that would reduce (in real domestic terms) import prices
(with an elasticity assumed to be 0.6)and increase export prices (withan elasticigr assumed
to be O.Z).~

Second, a temporary increase in government expenditures would have a positive impact 4

on GDP, with a multiplier effect (10% increase in expenditures translating in 12.4 points of
GDP growth) in the &st pear, part of which is offset in the subsequent two yeas. To link

4
The real exchange rate is defined as the nominal exchange rate (foreign currency per unit of NepaLi w e e )
multiplied by the ratio of prices in Nepal o v e ~prices in other countries. Aa increase indicates an
appreciation of the real exchange rate. The effective exchange rate is based on India and the US (the latter
being used as a pTexy for the rest of the world).
',Thapa (2062) indeed concludes that this conditi~nis met in the case of Nepal.
In both cases, it is assumed that firms do not fully pass through the change in exchange rate, with
-_
importers reducing their margins to remain competitive and exporters increasing their margins. Ideally, the
V A R would directly link GDP to terms of trade: however such link was not directly significant with
Nepal's data.
this with 123 model, it also assumed that such shock increases equally capital ancE r e m e n t
expenditures, and that it also leads to an increase in the heal deficit
Simikr simuEati0~1~
were run for a number of other combinations of policy variables such as
agricultural GDP (subject to exogenous dimaac shocks), government consumption - which
would presumably have a greater impact on short term growth - Indian GDP growth rate
and the growth rate of Indian demand (the last two axe key drives ofthe external demand to
Nepal). However, the impulses generated through the combination of these vatiables were
not strong enough to generate enough vadance ia the p a v q outcomes if they were
transmitted h o u g h t h e system.

2.6. Distributional Impact

The h a 1 block of the 123PRSP model is related to the distributional analysis. This module
is based on household data from the Nepal Living Standard Sutlrey (NUS, 2004/05]. To
map this data with results Emm the other blocks, changes in household utiliq is decomposed
in three terms:"
- the relative chaage in wages, weighted by salaries in the economy (average wage
&es labor supply)

- change in profits;
- minus changes in relative prices, weighted by the demand for each commdlity.

To assess changes in utiliv for each decile, tZlis approach requires decomposing wages,
profits in the domestic sector7 profits in the export sector, domestic expenditure fur the
import commodity, and domestic expenditure for the domestic commodity. This data is
summarized in Table 4 below.

In terms of income, profits from the domestic sector, which indudes traditional sectors (and
agriculture), are a bigh proportion of income for the poorest, whle profits from the export
sector is biased toward sicher deciles. Wages are somewhere in the middle, with the highest
proportion toward the 3d decile. With respect to expenditures, the propordm of imported
goods in total consumption increases steady throughout the income distribution, with poor
household having viaudy no access to them.

This pattern has important implications for the model. In parti&, a real appreciation of
the exchange rate will make imported goods cheaper relative to domestic goods, hence
benefit mmaifiIy the less poor; at she same h e , the poor, who get n relatively lugher
proportion oftheir income from prohts in the domestic sector, d be relatively worse off.

' Thl~can bc dtrivcd by looking at househo1d's indirect utility as a h c t i o n of wages, profits, and prices,
differentiating this function, and applying Shephard's Lemma.
1

9
 Table 4: Distribution of Incqmt and Expenditures
Share of Income
Profits
Wages (domestic (export
Decites sector) sector)
1 0.33 0.53 0.1 5
2 0.37 0.45 0.1 8
3 0.40 0.44 0.1 5
4 0.36 0.44 0.20
5 0.33 0.45 0-22
6 0.28 0.49 0.23
7 0.30 0.44 0.26
8 0.26 0,45 0.30
9 0.23 0.39 0.37
10 0.21 0.44 0.34
-
Average
-
0.31 0.45 0.24 1 0.05 1
0.95
m: & t ~ on inramc and mpc~dtrmsuse a mqping flht~fjvm the N U S U. The mqpiirg rn- tbc cbntcaiion
auuibbb in the N U S TI t~ th 2 sdm a ~ 3d mmmoditit~wed in thc I23PRSP modd Income: JVqps inmm~k .as it is
fmm the NLTS-II mabh on w a p ( a ~ d~ W W S S tbr WIW cash and in-kind t m n r i g p y in a p ' m h n and om.&
up8"acI~rs, inrlkding fib, piece mtc andpmanmt emplym~. h d - h r n dcfine~as ir inchtd~~ aj%i 20 p e n t of
fum income (dall &cileI$ and a q ' n g shun fmm 10 parent of the /OWCSI qrcinii/e to 50 pmmtfor the bights4 ofentnprirc
income. Rnm'ttancc immc, lvhirh ir terhnicalb a b q f t e r , is d o ixcbdcd in i n m m f i o m c p r f r . I m e fmm byfits in hme~fr'c
dsjinrd as mmkst iJ q ~ r i s a dpf ths rmaa&aI.
Expendim -. Fecndtum on &mcstic-m& inchdcs spmding on mcb items a r
H ad~~cation, hvrchold nnt, ufilSicr, und
lobam. For the n'cbtrt thne dtuh, 70 pmm diha opsnditun onfood i r i r ~ u m dto k on i q o r t J . Tbt on&pmnnnnntnt sum
Hm a@, a ~ ' n m2h1
mimp~eeditem$ Ijcommer dvmbk~.
~ ~ x p m d i t son g tpkcd urns$ &Jcs with &e bmrt &ciIc~
+ding on4 I0 pmcnt d their mpnditun on &rabbi OR iqwted wmmodrtic~rmd the ticbed psding s h as hkb ar 50
pmr.

3. Scenarios
3.1. Static simulations
We first sun a few simple static simulations (using 2004 as base year, cf. Table 5):'
- First, we increase the nominal exchange rate by 20%. Since h is the nurneraire in
the model it implies that all prices and incomes increase by 20% as well, leaving d +

quantities unchanged

In all these simulations, the user needs to define the "closure" of the model. As a bast case, we always
assume that savings are fxed - hence the adjustment is made on invesbnent h some cases, we suggest the
implication of an adjustment on savings (with hvesbnent fixed), for instance through a change in foreign
savings (current account).
 .- Table 5: Static Scenarios
1 2 3 3 bl$ 1
20% inmaw in 20% i m in Zm in tOO/bin- in
Reduction in
rwmiml w r l d imporl privak transfers foreip
hpTttaritTs
exchanm rate pritx? fmm l o b $ %
Import
Prices 20.0% 20.W 0.W 0.W 4.5%
V01~me 0.0% -14.7% 6.4% 0.9% 0.9%
Exports
hice3 20.m 0.0% 0.0% 0.0% 0.0%
Val# 0.0% 3.3% -4.2% -0.6% E.2%
Domutte G o d s
Price8 20.0% 4.6% 9.1% 1.2% -2.5% '

Volume 0.W 4.8% 1.W 0.1% 4.3%

Real Wnges 0.0% 4.7% TAX 1.0% -2.0%
Real Income 0.0% -5.6% -1.1% -0.2% 0.4%
Exptndltures
Top 20 0.0% -6.6% 3.3% 0.4% -1.1%
Bottom 20 0.0% -8.8% 3.W 0.5% -1.0%
Income
TOD 20 0.0% -4.5% -1.8% -0.346 0.6%
~ o k 20
m 0.094 -6.3% 0.6% 0.1% -0.1%
Poverty 32.0% 35.5% 31.5% 32.0% 32 P/*
Gin3 (incomt) 24.5 24.7 24.3 24.5 24.6
Note: the ~abhtionsoff pow a n d i m ti@ena a i/htmhhw.

- Second, we assume a 20% increase in the overall price of impom (e.g. impact of
large increase in internationa1 oil prices). This leads to a 15% decrease in imports as
- assuming no change in foreip savings (hence in the w e n t account) - the country
can afford less imports (the impact would be dampened had we assumed an
adjustment through higher foreign savings). This is offset by a small (3%) increase iD.
expoas to increase Nepal's capacisg to buy imports. To ensure this redocation of
resources to the export sector, the price of domestic goods has to decrease - its
decrease relative to the price of exports means a depredation of the real exchange
rate. This also leads to a decrease in the demand for domestic goods - but it is very
small (less than 1'/or. The income and expenditure of poorer households suffer most
from the shock. Hence both the o v d reduction in income (about 5%) and the
negative distributional impact add up to an increase in poverty.

- Third, we assume a 20% increase in private transfers (e.g. remittances). With the
current account (or foreign savings) h e d , this allows the txade b h n c e to deteriorate,
hence imports can inctease (6%). But it also allows more consumption, hence a
larger supply of domestic goods, which drives an increase in the price of domestic
goods. This drives away resources fiom the export sector (though an decrease in
the relative price of exports), hence leading to a lower volume of exports (4%). This
has a redistributive impact, as all incorne increases but the reduction in exporn
penalizes dchet households. Overall p o v e q is reduced, but only shghtly because of
the impact of the exchange rate appreciation ("Dutch disease")).

- Given the constfilction of t h e model, the impact of an increase in official grants has
broadly speaking the same impact (but the impact is of a lower magnitude since a
L
 20% increase in o f f i d grants implies, in absolute t-S, n smaller impact than a
.-
20% incrase in private transfers)?

- Findy, impact of the recent mde &&ation

to assess the in Nepal, we assume a
reduc~oaof impoa tariff rates (from the cuttent effe&e rate of 10% to 5'/o). This
makes imports cheaper, leading to additional consumption (at the expense of
investment) and a substitution of imports for domestic goods: imports increase by
1%. This leads to a lower price of domestic goods - a depreciation of the real
exchange rate which benefits the expoa sector (also increasing by IYo), duwing the
w e n t account to remain b&nced.la However, we assume that the reform is
financed by a increase in VAT: this leads to ae increase in domestic prices, which
reduced the impact on poveq. Overall, despite a small increase in income
inequality, poverty remains constant

3.2. Dynamic simulations

In this setion, we review the implications of the medium-term base case (Section 2.2.) and
of a number of scenarios as deviation fiom this base case. The base case assumes that real
economic growth picks kom around 2.5% to 4.5% in the next fout years (around 0.5% to
2.5% in per capita terns), which is consistent with a slow improvement in t h e drivers of
powth pigurc 4).

In this base case, real income increases with economic growth for all deciles (Table 6). T h e
appreciation of the real exchange rate in t h i s base case leads to some distributional shift, with
the income of the richer deciles increasing (more dependent on exports) somewhat less than
the income of poorer deciles (and a smder opposite effect on expenditures, as richer d e d e s
tend to consume more imports which are made reIaheIy cheaper by the real appreckeon).
The effect leads to a small reduction in inequality (contrary to the increase observed between
95/96 and 03/O4]. Illustratively, this could lead to a reduction of poverty incidence from
32% to 24% in 5 years {cf. Figure 5). This would be a reduction of a magnitude similar to
the-previous five years.

9
There is however a different composition in the increase in income, with a higher proportion of
consumption in the case of private transfers and a higher proportion of investment in the case of official
grants, which presumably would have an q a c t over the long run in a dynamic simulation. Although the
= model does not capture this stylized fact, it i s likely that an increase in official grants will have a*lower
import content than an increase in private fransfws.
" D e increase of exports as a result of import iaiff liberahation (i-e. '"import tariffs are a tax on export")
is a standard result o f such model. This largely disappears if the foreign savings can be adjusted: then the
higher demand for imports is fmaneed by more foreign borrowing, with almost no impact on expo*.
 Figwe 5: Change in 1ncomt'~isttiitdon
(Illustrative) r

~ on the &rih una3sif I m ~ s t din !ha ttxt (SaBIe 4). Th m n t p o q

Note: this t b is bard is e d e d to match
the 08;auIerlimatc q a 3 2 % ~ - hcadcvrmt. Thir mn'ian k 7i U u r ~ ' y cand mf m~prrrisc,u h k ~ ~ itg hi v d bc j;d fo
&c info m ~ nth t ~ t i o ~ I ~ m ofa~ nbgmrmoI
u n &b~~tion flinmmcC S o m r Anth'ccthbfim.

Let us now assume that the new Government builds a b u d g e t with a 15% haease in
emendimes (spread equally over recurrent and capital expenditures). This could fox
instance be viewed as the impact of more s e d t y spending or a d d i t i d development
investment. W e assume that total invesment adjusb to available savings (hence more public
i n v e s m e n t lends to less private investmeng cf. footnote 8 on closure). The impact is as
follows:
- GDP growth accelerates in the short-run (cf. VAR model in Section 2.51, but this is
partly offset by subsequent dedines after the temporary budget increase;
- T h e fiscd deficit blooms the titst year. The assumpdon is that the access to foreign
borrowing remains coastant as a share of GDP. Hence the deterioration of the heal
deficit raises issues of sustalkabiliq;
- Given the large savings rate of the private sector, the increase in GDP leads to an
increase in private savings wkch more than offset the reduction in government
savings. As a resdq even though the current account (i.e. for- savings) is fixed,
investment can increase slightly (less than 1%) despite the lower government savings;
- T h e hrst year, there is a small depreciation of the exchange rate, which benefits the
export sector;
- As a resul~although the budget increase benefits the whole distribution through its
short-term Keynesian effect, richer households benefit more horn the accelerated
growth of the export sector. As a result, the poverty headcount is reduced by an
additional l/z point - i.e. the initial growth impact is somewhat offset by the
subsequent slow-down and the dis&ibutional impact of the exchange rate
depredation.
If all h
s increase goes to very effective expendituxes, we could assume that after this
increase in expenditures (and its short-term demand effect) growth potential i s increased,
leading to a sustained increase in the growth rate. This wodd h h e r reinforce the impact
on poverty.

W e now turn to a second scenario, with an a~wrecia~on

of the exchanpe rate of 10% in 2006
and 2007 (resulting from a terns-of-trade shock, see Section 2.5). This can be viewed as
mdyzlrg the impact of &t real appreciation that occurred when inflation pressures
developed in Nepal in 2005 and 2006 (oil shock, supply constraints, etc.) while the m e n 7
remained pegged to the Indian rupee and inflation was steady in India. The impact is as
foUows (again assuming that invesiment adjusts to savings):

- GDP growth deteriorates by around 1% per year (cf. V& Sec~on2.5);

- The appreciation has a negative impact on external competitiveness, hence
constraining exports (despite reallocations to reduce the price of exports in local
currency). lmpoas on the other hand are boosted by heir lower relative price {and
despite o v c d lower demand).
- Lower dative domestic prices Iead to an increase in private consumption {despite
t h e overd decrease k.1 GDP),which has a positive effect throughout the distribution
@ut with a h@er impact on the poorest);
- Lower growth reduces bousehoId income, with the richest suffering most given the
impact on the export sector: hence the overall dismbutional effect is more favorable
for the poorest than for the richest;
- As a result, the povexty headcount wodd decrease by 1 '/z pt less than in the basehe
in this scenario, despite a lower GG,
- k
t US hndy note the opposite s c k o , a depreciation of the exchange rate,
that
would have symmetric effect, boosting growth in the short-tern, with expenditure
 switching to domestic goods md prduckoa switchirlg to exports, leading to a
further reduction in poveap.

Finally, assuming w e move to a lugher-case growth scenario, with an additional 2% growth

every year from 2005 to 2009 (hence above the high-case scenario in Figure 4), the p o ~ e q
impact is even larger:

- By assumption, GDP growth is 2% hzgher werg year ,which reinforces the effect on
p o v e q reduction;
- There is a shght shift from domestic production to exports, which has a
distribuuond impact favoring richer househoIds.

4. Conclusion
T h e simple model discussed in this paper is a good vehicle to hqhlight in a simple way some
of the complex transmission mechanisms of economic policy to poverty. It raises a number
of interesting questions with tentative quantified answers. First, not surpisingly, economic
gowth is a key driver to poverty reduction. From that point of view, the long-term growth
model gives an interesting perspective on the impact of the conflict and political instability
on economic outcomes and the potential prospects for n "peace dividend". Second, through
its implications on economic growth, both its level and composition (distributioaal effect),
the real exchange rate is a key transmission mechanisms and cannot be ignored ia the
andysis of policies. This has important implications for the analysis of increased inflows of
f o r e p currency - be it private remittances or pubIic external assistance.

By developing a simple model, this work is expected not only to enhance t h e capacity to
address selected policy issues, but also to generate a higher demand for such analytical work
to inform, policy choices. The next step with this simple model is therefore to expose it to a
wide range of potential stakeholders to get their views and insights and assess how it can
potentially help informing policy debates.

With respect to the model itself, it i s of course tempting to r e h e in by adding to its

complexity - lookmg at more sectors, refining the concept of poverty, introducing a refined
dynamic in the model (notably about the impact ofpublic investment on growth), etc. But
this would of course be to the detriment to its simplicity. Hence the priority might be to
expaad the library of simulations done with this model, possibIy addmg some sensitivi~
analysis (on some of the key elastidues, the closure used, the quality of the underlying macro
scenario, the assumptions required to calcuIate the dktributio~of income and expenditures
by sector and commo&ty, etc.), and a number of other simdatisns (notabIy the VARs).
Selected rehements can then be considered as long as they presme the simplicity of the
rnodeL
5. References
Devarajan, S., Lewis, J.D., Robinson, S. (19901, "Palicy h s o n s from Trade-focused,
Two-sector Models", Journal of Policy Modeling, 12(4):625-57.

Devarajan, S., Go,D. (20031, 'Yhe 'r23PRSP Model'", Chapter 13 in Bourguignon and
Per& da Silva, Tookt for Evaluating the Poverty and Disfributional Impact of
Economic Policies,
Devarajan, S., Go, D.,Li, H. (1999), "Quantifying the fiscal effects of trade reform: A
general equilibrium model estimated for 60 countries", mimeo, July 1999.

Government of Nepal (200x), NLSS IT.

Ianchovichha, Kacker (2005), "Growth trends in the developing world: country

forecasts and determinants", World Bank Policy Research Working Paper, 3775,
http:I/intresources.wor1dbank~0r~DTEAme~o~r~es/ian~h0vi~hinaka~ker20060213~~
df.
-

Kraay, G and G. Monokroussos (19991, "Growth Forecasts Using Time Series ant3
Growth Models'" World Bank P o l i q Research Working Paper, 2224.

Loayza, N., P. Fajnzylber, C. Calderbn (2005) "Economic Growth in Latin America

and the Caribbean: Stylized Facts, Explanations, and Forecasts," World Bank Study,
World Bank and Central Bank of Chile Working Paper, 265, June 2004.

Thapa, N.B (2002), "An econometric analysis of the impact of red effective exchange
rate an economic activities in Nepal", Nepal Rastra Bank Economic Review: Occasional
Paper, 14, April.

World Bank (20063, 'Xesilience amidst conflict: an assessment of poverty in Nepal,

f 995-96 and 2003-04'" World Bank Report, 36235, June.
6. Annexes
1W Model:
R dF k

(11 ~=%m
a)a"=F F i P P )
13) P = C + ZE+
(4) I d = MI*

(5, = W,F?

E:Eq~w~gwd
M; Impat g m d ~ W d p r i a d i n ~ p d
D*:srppluor-rn p~?~uktpriecdexg&gcd
d ] : ~ ~ b r ~ g o o d f?':Td nb
Q? 5upppraf pd t': wrt w w r d e
Q': Demand h rr*posle g o d F sale5'@m-**
P':Dome~priceuIexpwt~ ??drPCltilx r;*ie
P":Danc5kmnbm~.atgwd h=gowrrment-
F':~**ddmstrg~~d f t : f o r ~ % m 5 m b o ~
P: s a k ~ p - ~ o f c v m p o s r t ~ ~ n:6 w g n mmitlxlcpr la mw
C P md w - a - s A- ravings me
P.Pme of- p d X A r m - W k ar4wt
R E*chnge rate G:Aealg~vwnmentM
T: TDI m e B: Baiance d Wade
$1: G1- rarlngs n:h m d *rn- cM?ly
Y: TOMincwne rr: In?pc+rtstbsEuDOn
&&&j
c:AQimgateml!aqJh
S:Aggeg;r* sni?((s
2 Agg-e red Tivehm

Source: Devarajan and G o (2003).